the american journal of cardiology vol 107 issue 4 february 2011

EDITOR IN CHIEF

William C. Roberts, MDBaylor Heart & Vascular InstituteBaylor University Medical CenterWadley Tower No. 4573600 Gaston AvenueDallas, Texas 75246(214)826-8252Fax: (214)826-2855

ASSOCIATE EDITORS

Vincent E. FriedewaldPaul A. GrayburnASSISTANT EDITORSRobert C. KowalJeffrey M. Schussler

EDITORIAL BOARDCARDIOVASCULARMEDICINEIn AdultsAntonio AbbateJ. Dawn AbbottGeorge S. AbelaJamil AboulhosnJoseph S. AlpertMartin A. AlpertEzra A. AmsterdamJeffrey L. AndersonEvan AppelbaumRichard W. AsingerPablo AvanzasGary John BaladyEric BatesJeroen J. BaxGeorge A. BellerWilliam E. BodenMonty M. BodenheimerRobert O. BonowJeffrey S. BorerMartial G. BourassaEugene BraunwaldJeffrey A. BrinkerDavid L. BrownMichael E. CainRichard O. Cannon IIIBernard R. ChaitmanKanu ChatterjeeJohn S. ChildRobert J. CodyLawrence S. CohenMarc CohenC. Richard ContiMichael H. CrawfordJames P. DaubertGregory J. DehmerJames A. de LemosAnthony N. DeMariaPablo DenesGeorge A. DiamondJohn P. DiMarcoAllen DollarMichael J. DomanskiGerald DorrosUri ElkayamKenneth A. EllenbogenMyrvin H. EllestadStephen G. EllisToby R. EngelAndrew E. EpsteinN. A. Mark Estes, IIIMichael Ezekowitz

Rodney H. FalkJohn A. FarmerDavid P. FaxonTed FeldmanJack FerlinzJerome L. FlegGerald F. FletcherJoseph A. FranciosaGary S. FrancisW. Bruce FyeWilliam H. GaaschJulius M. GardinBernard J. GershS. David GertzMihai GheorghiadeD. Luke GlancyStephen P. GlasserMichael R. GoldSamuel Z. GoldhaberRobert E. GoldsteinSidney GoldsteinSteven A. GoldsteinJ. Anthony GomesAntonio M. Gotto, Jr.K. Lance GouldDonald C. HarrisonPhilip D. HenryL. David HillisDavid R. Holmes, Jr.Mun K. HongWilliam G. HundleyAmi S. IskandrianAllan S. JaffeHani JneidGreg L. KaluzaJoel S. KarlinerJohn A. KastorSanjiv KaulEllen C. KeeleyKenneth M. KentRichard E. KerberDean J. KereiakesMorton J. KernSpencer B. King IIIRobert E. KleigerGeorge J. KleinLloyd W. KleinPaul KligfieldRobert A. KlonerJohn B. KostisCharles LandauRichard L. LangeCarl J. LavieCarl V. LeierB. T. Liang

Joseph Lindsay, Jr.Gregory Y.H. LipFrancisco Lopez-JimenezJoseph LoscalzoG.B. John ManciniFrancis E. MarchlinskiFrank I. MarcusBarry J. MaronMartin S. MaronRandolph P. MartinAttilo MaseriCharles MaynardMichael D. McGoonDarren K. McGuireRaymond G. McKayJawahar L. MehtaBernard MeierFranz H. MesserliEric L. MichelsonRichard V. MilaniAlan B. MillerWayne L. MillerGary S. MintzFred MoradyLori MoscaArthur J. MossJames E. MullerGerald B. NaccarelliNavin C. NandaRobert A. O’RourkeErik Magnus OhmanRichard L. PageSebastian T. PalmeriSeung-Yung ParkEugene R. PassamaniAlan S. PearlmanCarl J. PepineJoseph K. PerloffBertram PittPhilip J. PodridDon PoldermansArshed A. QuyyumiCharles E. RackleyC. Venkata RamNathaniel ReichekRobert RobertsJennifer G. RobinsonLynda E. RosenfeldMelvin M. ScheinmanDavid J. SchneiderJohn S. SchroederPatrick Washington SerruysPravin M. ShahPrediman K. ShahJamshid Shirani

Robert J. SiegelMarc A. SilverRoss J. Simpson, Jr.Steven N. SinghBurton E. SobelJohn C. SombergDavid H. SpodickLynne W. StevensonGregory W. StoneJohn R. StrattonJonathan M. TobisEric J. TopolByron F. VandenbergHector O. VenturaGeorge W. VetrovecRobert A. VogelRon WaksmanDavid D. WatersNanette K. WengerRobert WilenskyJames T. WillersonClyde W. YancyBarry L. ZaretDouglas P. ZipesIn Infants and ChildrenHugh D. AllenBruce S. AlpertStanley J. GoldbergWarren G. GuntherothHoward P. GutgesellJohn D. KuglerJames E. LockJohn W. MooreLowell W. PerryDavid J. SahnRichard M. SchiekenCARDIOVASCULAR SURGERYEugene H. BlackstoneLawrence I. BonchekLawrence H. CohnJohn A. ElefteriadesHartzel V. SchaffRELATED SPECIALISTSL. Maximilian BujaJean-Pierre DespresMichael EmmettGiovanni FilardoBarry A. FranklinCharles B. HigginsJeffrey E. SaffitzRenu Virmani

A5

CONTENTS THE AMERICAN JOURNAL OF CARDIOLOGY�

VOL. 107, NO. 4 FEBRUARY 15, 2011

Coronary Artery DiseaseAn Early and Simple Predictor of Severe Left Mainand/or Three-Vessel Disease in Patients WithNon–ST-Segment Elevation Acute CoronarySyndrome ......................................................495Masami Kosuge, Toshiaki Ebina, Kiyoshi Hibi,Satoshi Morita, Mitsuaki Endo, Nobuhiki Maejima,Noriaki Iwahashi, Kozo Okada, Toshiyuki Ishikawa,Satoshi Umemura, and Kazuo Kimura

Reperfusion by Primary Percutaneous CoronaryIntervention in Patients With ST-Segment ElevationMyocardial Infarction Within 12 to 24 Hours of theOnset of Symptoms (from a Prospective NationalObservational Study [PL-ACS]) .........................501Marek Gierlotka, Mariusz Gasior, Krzysztof Wilczek,Michal Hawranek, Janusz Szkodzinski, Piotr Paczek,Andrzej Lekston, Zbigniew Kalarus, Marian Zembala,and Lech Polonski

ReviewPercutaneous Coronary Intervention for NonST-Elevation Acute Coronary Syndromes: Which,When and How? ............................................509Robert K. Riezebos, Jan G.P. Tijssen,Freek W.A. Verheugt, and Gerrit J. Laarman

Coronary Artery DiseaseLong-Term Follow-Up of Patients With First-TimeChest Pain Having 64-Slice ComputedTomography ...................................................516Fabiola B. Sozzi, Filippo Civaia, Philippe Rossi,Jean-Francois Robillon, Stephane Rusek,Frederic Berthier, Francois Bourlon, Laura Iacuzio,Gilles Dreyfus, and Vincent Dor

Usefulness of Cooling and Coronary Catheterizationto Improve Survival in Out-of-Hospital CardiacArrest ............................................................522Dion Stub, Christopher Hengel, William Chan,Damon Jackson, Karen Sanders, Anthony M. Dart,Andrew Hilton, Vincent Pellegrino, James A. Shaw,Stephen J. Duffy, Stephen Bernard, and David M. Kaye

Two-Year Safety and Effectiveness of Sirolimus-Eluting Stents (from a Prospective Registry) .......528Bimmer E. Claessen, Roxana Mehran, Martin B. Leon,Eric A. Heller, Giora Weisz, George Syros,Gary S. Mintz, Theresa Franklin-Bond,Irene Apostolidou, Jose P.S. Henriques,Gregg W. Stone, Jeffrey W. Moses, andGeorge D. Dangas

Comparison of Morbidity and Mortality in DiabeticsVersus Nondiabetics Having Isolated CoronaryBypass Versus Coronary Bypass plus ValveOperations Versus Isolated Valve Operations ....535Serenella Castelvecchio, Lorenzo Menicanti,Ekaterina Baryshnikova, Carlo de Vincentiis,Alessandro Frigiola, and Marco Ranucci, for theSurgical and Clinical Outcome Research (SCORE)Group

Heart Failure

Relation of Bundle Branch Block to Long-Term(Four-Year) Mortality in Hospitalized Patients WithSystolic Heart Failure ......................................540Alon Barsheshet, Ilan Goldenberg, Moshe Garty,Shmuel Gottlieb, Amir Sandach,Avishag Laish-Farkash, Michael Eldar, andMichael Glikson

Characteristics of Depression Remission and ItsRelation With Cardiovascular Outcome AmongPatients With Chronic Heart Failure (from theSADHART-CHF Study) ......................................545Wei Jiang, Ranga Krishnan, Maragatha Kuchibhatla,Michael S. Cuffe, Carolyn Martsberger,Rebekka M. Arias, and Christopher M. O’Connor, forthe SADHART-CHF Investigators

Warfarin Use and Outcomes in Patients WithAdvanced Chronic Systolic Heart Failure WithoutAtrial Fibrillation, Prior Thromboembolic Events, orProsthetic Valves .............................................552Marjan Mujib, Abu-Ahmed Z. Rahman, Ravi V. Desai,Mustafa I. Ahmed, Margaret A. Feller,Inmaculada Aban, Thomas E. Love, Michel White,Prakash Deedwania, Wilbert S. Aronow,Gregg Fonarow, and Ali Ahmed

Editorial

The Risk of Thromboembolism in Heart Failure:Does It Merit Anticoagulation Therapy? ............558Eduard Shantsila and Gregory Y.H. Lip

Heart Failure

Trials on the Effect of Cardiac Resynchronization onArterial Blood Pressure in Patients With HeartFailure ...........................................................561Sameer Ather, Sripal Bangalore, Srinath Vemuri,Long B. Cao, Biykem Bozkurt, and Franz H. Messerli

A6 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 107 FEBRUARY 15, 2011

Patient Perception Versus Medical Record Entry ofHealth-Related Conditions Among Patients WithHeart Failure ..................................................569Adnan S. Malik, Grigorios Giamouzis,Vasiliki V. Georgiopoulou, Lucy V. Fike,Andreas P. Kalogeropoulos, Catherine R. Norton,Dan Sorescu, Sidra Azim, Sonjoy R. Laskar,Andrew L. Smith, Sandra B. Dunbar, and Javed Butler

Effectiveness of Serial Increases in Amino-TerminalPro–B-Type Natriuretic Peptide Levels to Indicate theNeed for Mechanical Circulatory Support inChildren With Acute Decompensated HeartFailure ...........................................................573Derek T.H. Wong, Kristen George, Judith Wilson,Cedric Manlhiot, Brian W. McCrindle, Khosrow Adeli,and Paul F. Kantor

Arrhythmias and ConductionDisturbances

Relation of Obesity to Recurrence Rate and Burdenof Atrial Fibrillation .........................................579Maya Guglin, Kuldeep Maradia, Ren Chen, andAnne B. Curtis

Roundtable Discussion (CME)

The Editor’s Roundtable: Implantable Cardioverter-Defibrillators in Primary Prevention of SuddenCardiac Death and Disparity-Related Barriers toImplementation ...............................................583Vincent E. Friedewald, Gregg C. Fonarow,Brian Olshansky, Clyde W. Yancy, andWilliam C. Roberts

Valvular Heart Disease

Comparison of the Effectiveness and Safety of Low-Molecular Weight Heparin Versus UnfractionatedHeparin Anticoagulation After Heart ValveSurgery ..........................................................591Claudia Bucci, William H. Geerts, Andrew Sinclair,and Stephen E. Fremes

Congenital Heart Disease

Seeking Optimal Relation Between OxygenSaturation and Hemoglobin Concentration in AdultsWith Cyanosis from Congenital Heart Disease ...595Craig S. Broberg, Ananda R. Jayaweera,Gerhard P. Diller, Sanjay K. Prasad, Swee Lay Thein,Bridget E. Bax, John Burman, andMichael A. Gatzoulis

Cardiomyopathy

Relation of Pulse Pressure to Blood PressureResponse to Exercise in Patients With HypertrophicCardiomyopathy .............................................600Kevin S. Heffernan, Martin S. Maron,Eshan A. Patvardhan, Richard H. Karas, andJeffrey T. Kuvin, the Vascular Function Study Group

Editorial

Clinical Challenges of Genotype Positive(�)–Phenotype Negative (�) Family Members inHypertrophic Cardiomyopathy .........................604Barry J. Maron, Laura Yeates, andChristopher Semsarian

Miscellaneous

Usefulness of Repeated N-Terminal Pro-B-TypeNatriuretic Peptide Measurements as IncrementalPredictor for Long-Term Cardiovascular OutcomeAfter Vascular Surgery ....................................609Dustin Goei, Jan-Peter van Kuijk, Willem-Jan Flu,Sanne E. Hoeks, Michel Chonchol,Hence J.M. Verhagen, Jeroen J. Bax, andDon Poldermans

Usefulness of At Rest and Exercise Hemodynamicsto Detect Subclinical Myocardial Disease in Type 2Diabetes Mellitus .............................................615Christine L. Jellis, Tony Stanton, Rodel Leano,Jennifer Martin, and Thomas H. Marwick

Specific Characteristics of Sudden Death in aMediterranean Spanish Population ...................622M. Teresa Subirana, Josep O. Juan-Babot,Teresa Puig, Joaquın Lucena, Antonio Rico,Manuel Salguero, Juan C. Borondo, Jorge Ordoñez,Josep Arimany, Rafael Vazquez, Lina Badimon,Gaetano Thiene, and Antonio Bayes de Luna

Clinical and Prognostic Relevance ofEchocardiographic Evaluation of Right VentricularGeometry in Patients With Idiopathic PulmonaryArterial Hypertension ......................................628Stefano Ghio, Anna Sara Pazzano, Catherine Klersy,Laura Scelsi, Claudia Raineri, Rita Camporotondo,Andrea D’Armini, and Luigi Oltrona Visconti

Clinically Significant Incidental Findings AmongHuman Immunodeficiency Virus-Infected MenDuring Computed Tomography for Determination ofCoronary Artery Calcium .................................633Nancy Crum-Cianflone, James Stepenosky,Sheila Medina, Dylan Wessman, David Krause, andGilbert Boswell

CONTENTS A11

Case Report

Self-Terminated Ventricular Fibrillation andRecurrent Syncope ..........................................638Yuval Konstantino, Angela Morello,Peter J. Zimetbaum, and Mark E. Josephson

Readers’ Comments

Comparison of 600 Versus 300-mg ClopidogrelLoading Dose in Patients With ST-Segment ElevationMyocardial Infarction Undergoing PrimaryCoronary Angioplasty .....................................641

Long-Term Follow Up of Atrioventricular Block inTranscatheter Aortic Valve Implantation ............641

Instructions to Authors can be found at the AJCwebsite: www.AJConline.org

Classifieds on pages A10, A37

Full Text: www.ajconline.org

A18 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 107 FEBRUARY 15, 2011

icmgpCcrhaWcwAs

An Early and Simple Predictor of Severe Left Main and/orThree-Vessel Disease in Patients With Non–ST-Segment

Elevation Acute Coronary Syndrome

Masami Kosuge, MD*, Toshiaki Ebina, MD, Kiyoshi Hibi, MD, Satoshi Morita, PhD,Mitsuaki Endo, MD, Nobuhiki Maejima, MD, Noriaki Iwahashi, MD, Kozo Okada, MD,

Toshiyuki Ishikawa, MD, Satoshi Umemura, MD, and Kazuo Kimura, MD

Clopidogrel should be initiated as soon as possible in patients with non–ST-segmentelevation acute coronary syndrome (NSTE-ACS) except those who urgently require coro-nary artery bypass grafting (CABG). The present study assessed the ability to predictsevere left main coronary artery and/or 3-vessel disease (LM/3VD) that would most likelyrequire urgent CABG based on only clinical factors on admission in 572 patients withNSTE-ACS undergoing coronary angiography. Severe LM/3VD was defined as >75%stenosis of LM and/or 3VD with >90% stenosis in >2 proximal lesions of the left anteriordescending coronary artery and other major epicardial arteries. Patients were divided intothe 3 groups according to angiographic findings: no LM/3VD (n � 460), LM/3VD but notsevere LM/3VD (n � 57), and severe LM/3VD (n � 55). Severe LM/3VD was associatedwith a higher rate of urgent CABG compared to no LM/3VD and LM/3VD but not severeLM/3VD (46%, 2%, and 2%, p <0.001). On multivariate analysis, degree of ST-segmentelevation in lead aVR was the strongest predictor of severe LM/3VD (odds ratio 29.1,p <0.001), followed by positive troponin T level (odds ratio 1.27, p � 0.044). ST-segmentelevation >1.0 mm in lead aVR best identified severe LM/3VD with 80% sensitivity, 93%specificity, 56% positive predictive value, and 98% negative predictive value. In conclusion,ST-segment elevation >1.0 mm in lead aVR on admission electrocardiogram is highlysuggestive of severe LM/3VD in patients with NSTE-ACS. Selected patients with thisfinding might benefit from promptly undergoing angiography, withholding clopidogrel toallow early CABG. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:
495–500)
pdlaatS(opY

Dual antiplatelet therapy with clopidogrel and aspirinshould be initiated as soon as possible in patients withnon–ST-segment elevation acute coronary syndrome(NSTE-ACS).1,2 However, such combination therapy canncrease perioperative bleeding in patients undergoing earlyoronary artery bypass grafting (CABG).3–7 Therefore, oneight consider with-holding clopidogrel until coronary an-

iography and definition of the coronary anatomy.8 Theroportion of patients with NSTE-ACS who undergoABG during hospitalization is 9% to 21%.4,5,8–12 CABGan often be deferred for several days, and few patientsequire urgent CABG. Ideally, clopidogrel should be with-eld in the minority of patients who urgently require CABGnd should be given to the remaining majority of patients.

e previously examined clinical factors related to left mainoronary artery and/or 3-vessel disease (LM/3VD) thatould most likely lead to CABG in patients with NSTE-CS but did not evaluate severity of coronary lesions in that

tudy.13 In the present study, we assessed the ability topredict “severe” LM/3VD, which would most likely to

Division of Cardiology, Yokohama City University Medical Center,Yokohama, Japan. Manuscript received August 28, 2010; revised manu-script received and accepted October 1, 2010.

*Corresponding author: Tel: 81-45-261-5656; fax: 81-45-261-9162.
E-mail address: [email protected] (M. Kosuge).
0002-9149/11/$ – see front matter © 2011 Elsevier Inc. All rights reserved.doi:10.1016/j.amjcard.2010.10.005

require urgent CABG, using only clinical factors on admis-sion in patients with NSTE-ACS.

Methods

We studied 572 consecutive patients (mean age 67 � 11years, range 30 to 92, 397 men and 175 women) who wereadmitted to Yokohama City University Medical Center(Yokohama, Japan) coronary care unit and fulfilled thefollowing criteria: (1) typical chest discomfort attributed tocardiac ischemia, lasting �5 minutes, occurring within 24hours before hospital admission, and involving an unstablepattern of pain including pain at rest, new onset, severe orfrequent angina, or accelerating angina14; (2) no conditionsrecluding evaluation ST-segment changes on electrocar-iogram (ECG) such as left or right bundle branch block,eft ventricular hypertrophy, or ventricular pacing; (3) fullyssessable ECGs on admission; and (4) fully assessablengiographic data during hospitalization. We excluded pa-ients with nonischemic or atypical pain, persistent newT-segment elevation in leads other than lead aVR, recent�6 months) percutaneous coronary intervention, or previ-us CABG. All patients gave informed consent. The studyrotocol was approved by the internal review board ofokohama City University Medical Center.
Standard 12-lead ECGs were recorded on admission at a
www.ajconline.org

mailto:[email protected]

c

KSPPR

HHPCEB

C

S

496 The American Journal of Cardiology (www.ajconline.org)

paper speed of 25 mm/s and an amplification of 10 mm/mV.All ECGs were examined by a single investigator who wasblinded to all other clinical data. ST-segment shifts were mea-

Table 1Clinical characteristics

Age (years)MenSystolic blood pressure on admission (mm Hg)Heart rate on admission (beats/min)

illip class �II on admissionymptom onset �6 hoursrevious myocardial infarctionrevious percutaneous coronary interventionisk factorsHypertensionDiabetes mellitusSmokingHyperlipidemia*Family history of coronary artery diseaseemoglobin on admission (g/dl)igh-sensitivity C-reactive protein on admission (mg/dl) 0ositive troponin T on admissionreatine kinase-MB on admission (IU/L)stimated glomerular filtration rate on admission (ml/min/1.73 m2)rain natriuretic peptide on admission (pg/ml)†

Cardiac procedures and outcomes at 30 daysDeathMyocardial (re)infarctionDeath/myocardial (re)infarctionUrgent percutaneous coronary interventionUrgent coronary artery bypass surgeryUrgent revascularization (percutaneous coronary intervention or

coronary artery bypass surgery)ardiac proceduresPercutaneous coronary interventionCoronary artery bypass surgeryAny revascularization (percutaneous coronary intervention or

coronary artery bypass surgery)

Data are presented as mean � SD, median (interquartile range), or num* Fasting total cholesterol concentration �220 mg/dl, fasting triglycerid† Available for 360 patients.

Table 2Electrocardiographic findings

Variable

No(n

ST-segment depression �0.5 mm 288Maximal ST-segment depression (mm) 0Sum of ST-segment depressions (mm) 2Number of leads with ST-segment depression �0.5 mm 2ST-segment elevation �0.5 mm in lead aVR 68

T-segment elevation in lead aVR (mm) 0

Data are presented as mean � SD or number of patients (percentage).

sured 80 ms after the J-point for ST-segment depression and 20

ms after this point for ST-segment elevation using the preced-ing TP segment as a baseline.15 ST-segment deviation wasonsidered present if deviation was �0.5 mm in any lead.14

LM/3VD pValue

M/3VD Nonsevere Severe� 460) (n � 57) (n � 55)

6 � 11 69 � 10 68 � 11 0.060%) 39 (68%) 36 (66%) 0.780 � 25 150 � 32 141 � 26 0.076 � 17 81 � 20 89 � 23 �0.001%) 9 (16%) 17 (31%) �0.0018%) 43 (75%) 49 (89%) 0.139%) 18 (32%) 12 (22%) 0.070%) 15 (26%) 5 (9%) 0.06

6%) 42 (74%) 38 (69%) 0.490%) 29 (51%) 30 (55%) �0.0010%) 22 (39%) 23 (42%) 0.180%) 25 (44%) 29 (53%) 0.616%) 13 (23%) 16 (29%) 0.754 � 2 13 � 2 13 � 2 0.033.061–0.323) 0.180 (0.079–0.453) 0.253 (0.099–0.801) 0.0059%) 28 (49%) 33 (60%) �0.0014 � 16 18 � 24 27 � 36 �0.0018 � 25 58 � 28 58 � 26 0.0046–179) 187 (81–429) 230 (67–571) �0.001� 297) (n � 32) (n � 31)

.2%) 1 (2%) 2 (4%) 0.010%) 3 (5%) 5 (9%) 0.23%) 4 (7%) 7 (13%) 0.004%) 7 (12%) 5 (9%) 0.22%) 1 (2%) 25 (46%) �0.001%) 8 (14%) 30 (55%) �0.001

9%) 36 (63%) 14 (25%) �0.001%) 13 (23%) 40 (73%) �0.0013%) 49 (86%) 54 (98%) �0.001

patients (percentage).entration �150 mg/dl, or use of antihyperlipidemic therapy.

LM/3VD p Value

D Nonsevere Severe) (n � 57) (n � 55)

53 (93%) 55 (100%) �0.0010 1.7 � 1.1 2.6 � 1.7 �0.0016 6.7 � 5.1 10.5 � 7.3 �0.0015 5.1 � 2.6 6.1 � 2.2 �0.001

39 (68%) 50 (91%) �0.0013 0.6 � 0.5 1.2 � 0.7 �0.001

No L(n

6322 (7

157

26 (6356 (786 (190 (2

304 (6136 (3229 (5230 (5120 (2

1.131 (0135 (2

16

67 (2(n

1 (014 (315 (329 (67 (2

36 (8

272 (527 (6

291 (6

ber ofe conc

LM/3V� 460

(63%).8 � 1..6 � 3..5 � 2.(15%).1 � 0.

A qualitative assay for cardiac-specific troponin T (de-

tcL

ewcrngthkpr

a

(bioDvuasiip

tta

R

497Coronary Artery Disease/Prediction of Severe LM/3VD in NSTE-ACS

tection limit 0.1 ng/ml of cardiac-specific troponin T; RocheDiagnostics, Tokyo, Japan) was performed on admission.Troponin T �0.1 ng/ml was defined as positive. Bloodsamples for measuring hemoglobin, plasma high-sensitivityC-reactive protein levels, and estimated glomerular filtra-tion rate were also taken on admission. Japanese equationswere used to calculate estimated glomerular filtration ratefrom serum creatinine level.16 Brain natriuretic peptide wassimultaneously measured in 360 patients. Creatine ki-nase-MB levels were measured on admission, at 3-hourintervals during the first 24 hours, and in any patient withsuspected reinfarction.

All patients underwent cardiac catheterization a medianof 3 days after admission. Urgent cardiac catheterizationwas performed in patients with unstable hemodynamicsfrom ischemic attacks or with ischemic attacks that couldnot be controlled by intensive drug treatment. Type andtiming of revascularization were left to the discretion of thetreating physician. All coronary angiograms were evaluatedby a single investigator who was blinded to all other clinicaldata. Stenosis �50% in the diameter of the LM or stenosisof �75% in �1 major epicardial vessel or its main brancheswas considered clinically significant. Severe LM/3VD wasdefined as (1) �75% stenosis of the LM, (2) 3VD with�90% stenosis of the proximal portion of the left anteriordescending coronary artery and �90% stenosis of the rightcoronary artery and/or left circumflex coronary artery, and(3) definitions 1 and 2. Patients were categorize according topresence (n � 112) or absence (n � 460) of LM/3VD, andhe former group was subdivided according to severity oforonary lesions: nonsevere LM/3VD (n � 57) and severeM/3VD (n � 55).

Demographic data, risk factors for coronary artery dis-ase, and data from physical examination on admissionere collected. Major adverse events such as death, myo-

ardial (re)infarction, or urgent revascularization were alsoecorded for all patients. Myocardial infarction was diag-osed according to cardiac enzyme levels or electrocardio-raphic criteria. Enzymatic evidence of myocardial infarc-ion was defined as an increase of creatine kinase-MB toigher than the upper limit of normal if the previous creatineinase-MB level was in the normal range or 50% above therevious level if the previous level was above the normalange.17 Electrocardiographic evidence of myocardial in-

farction was defined as new clinically significant Q waves in�2 contiguous leads distinct from the enrollment myocar-dial infarction.17 Patients were followed for 30 days afterdmission.

Results are expressed as mean � SD or as frequencypercentage), and high-sensitivity C-reactive protein andrain natriuretic peptide levels are expressed as median andnterquartile range. Data were compared by 1-way analysisf variance, Kruskal-Wallis test, and chi-square analysis.ifferences were considered statistically significant at palue �0.05. Multivariate logistic regression analysis wassed to identify clinical predictors of severe LM/3VDmong the variables associated (p �0.05) with this diagno-is on univariate analysis. Odds ratios and 95% confidencentervals were calculated. In addition, sensitivity, specific-ty, positive predictive value, negative predictive value, and
redictive accuracy of predictors of severe LM/3VD iden- w
ified on multivariate analysis were determined. SPSS sta-istical software (SPSS, Inc., Chicago, Illinois) was used forll analyses.

esults

Baseline characteristics are listed in Table 1. Patients

Figure 1. Representative electrocardiogram of a patient with severe leftmain coronary artery and/or 3-vessel disease. Troponin T was positive onadmission. ST-segment elevation in lead aVR was 4.5 mm on admissionelectrocardiogram. Urgent coronary angiography showed 90% stenosis ofthe left main trunk.

ith LM/3VD, especially severe LM/3VD, had a more

lCpnpfcNoE

ua


rapid heart rate, higher prevalences of Killip class �II,diabetes mellitus, positive troponin T, and higher levels ofhigh-sensitivity C-reactive protein, creatine kinase-MB, andbrain natriuretic peptide than did patients without LM/3VD.LM/3VD was associated with lower levels of hemoglobinand estimated glomerular filtration rate. There were no sig-nificant differences in other clinical variables among the 3groups.

Urgent CABG was more frequently done in patients withsevere LM/3VD (46%). In contrast, urgent CABG was donein only 2% of patients with LM/3VD but not severe LM/3VD.

Electrocardiographic findings are presented in Table 2.Compared to patients without LM/3VD, those with LM/3VD, especially severe LM/3VD, had a higher prevalenceand a larger amount of ST-segment depression, a largernumber of leads other than lead aVR with ST-segmentdepression, and a higher prevalence and greater magnitudeof ST-segment elevation in lead aVR. Figure 1 shows arepresentative ECG of a patient with severe LM/3VD.

In multivariate models, degree of ST-segment elevationin lead aVR was the strongest predictor of severe LM/3VD,followed by positive troponin T (Table 3). Sensitivity, spec-

Table 3Univariate and multivariate predictors of severe left main coronary artery

Variable

Systolic blood pressureHeart rateKillip class �IIPrevious percutaneous coronary interventionDiabetes mellitusHigh-sensitivity C-reactive proteinPositive troponin TCreatine kinase-MBEstimated glomerular filtration rateMaximal ST-segment depressionSum of ST-segment depressionsNumber of leads with ST-segment depression �0.5 mmDegree of ST-segment elevation in lead aVR

CI � confidence interval.

Table 4Comparison of ST-segment elevation in lead aVR and positive troponinT for predicting severe left main coronary artery and/or three-vesseldisease

Sensitivity Specificity PPV NPV PredictiveAccuracy

ST-segment elevationin lead aVR

�0.5 mm 91% 79%† 32%† 99% 80%†

�1.0 mm 80% 93% 56% 98% 92%�1.5 mm 27%† 98%† 58% 93%† 91%

Positive troponin T 60%* 69%† 17%† 94%† 68%†

NPV � negative predictive value; PPV � positive predictive value.* p �0.05; † p �0.01 versus ST-segment elevation �1.0 mm in leadaVR.

ificity, positive predictive value, negative predictive value,

and predictive accuracy of ST-segment elevation in leadaVR and positive troponin T for severe LM/3VD are pre-sented in Table 4. ST-segment elevation �1.0 mm in leadaVR best identified severe LM/3VD.

Discussion

Our study showed that ST-segment elevation �1.0 mmin lead aVR and positive troponin T on admission (espe-cially the former) were highly suggestive of severe LM/3VD, and the converse was also true, i.e., absence of thesefindings was rarely associated with severe LM/3VD. To ourknowledge, this is the first study to establish a reliabletechnique for early identification of patients with severeLM/3VD who are most likely to require urgent CABG inpatients with NSTE-ACS. Our findings have important im-plications for identification of high-risk patients and selec-tion of optimal treatment strategy in the setting of NSTE-ACS.

The standard 12-lead ECG, which is an inexpensive,noninvasive, and readily available clinical tool, has a centralrole in diagnosis and immediate triage for NSTE-ACS andprovides important prognostic information. In particular,presence of ST-segment depression on admission ECG hasbeen recognized to be a strong predictor of adverse out-comes in patients with NSTE-ACS.14,17–20 The Global Uti-ization of Strategies to Open Occluded Arteries in Acuteoronary Syndrome IV (GUSTO-IV ACS) trial of 7,800atients with NSTE-ACS has highlighted the striking prog-ostic value of ST-segment depression on admission com-ared to expanded biomarker profiles and traditional riskactors.18 However, most previous studies assessing thelinical significance of admission ECG in patients withSTE-ACS have focused on ST-segment deviation in leadsther than lead aVR; i.e., clinicians have used an “11-lead”CG, neglecting lead aVR.

Several studies have found that analysis of lead aVR isseful for evaluation of NSTE-ACS.13,15,21,22 Gorgels etl21 reported that ST-segment elevation in lead aVR accom-

panied by ST-segment depression in leads I, II, and V to V

three-vessel disease

atio (95% CI) p Value

Univariate Multivariate

0.020 0.07�0.001 0.29�0.001 0.29

0.045 0.800.001 0.08

�0.001 0.30(1.10–2.78) �0.001 0.044

�0.001 0.33�0.001 0.32�0.001 0.053�0.001 0.055�0.001 0.24

(9.54–49.8) �0.001 �0.001

and/or

Odds R

1.27

29.1

4 6during episodes of angina strongly suggests LM/3VD in

trssLipvsSNa

oa

SN

adfl

499Coronary Artery Disease/Prediction of Severe LM/3VD in NSTE-ACS

patients with angina at rest. Barrabés et al15 demonstratedhat presence of ST-segment elevation in lead aVR predictsisk of in-hospital death in patients with a first non–ST-egment elevation acute myocardial infarction. In thattudy, ST-segment elevation in lead aVR was also related toM/3VD; however, coronary angiography was performed

n only 56% of subjects within 6 months after infarction. Wereviously demonstrated that presence of ST-segment ele-ation �0.5 mm in lead aVR on admission ECG stronglyuggested LM/3VD and had a higher prognostic value thanT-segment depression in other leads in patients withSTE-ACS who underwent coronary angiography in the

cute phase.13,22 However, previous studies, including ours,did not consider severity of LM/3VD, which has clinicalimplications for timing of CABG in relation to dual anti-platelet therapy. An increased risk of perioperative bleedingevents due to early clopidogrel administration is clinicallyproblematic in patients with LM/3VD who urgently requireCABG. In such patients, postponing CABG for several daysmight seriously compromise outcomes. Timing of CABGdepends on many factors including severity of coronarylesions, risk of ongoing ischemia, general condition of apatient, bleeding risk associated with upstream antithrom-botic therapies, and local logistic factors such as collocationof cardiac surgical services and surgical waiting lists. Thepresent study examined predictors of patients with severeLM/3VD likely to require urgent CABG, considering thecoronary anatomy. We demonstrated that ST-segment ele-vation �1.0 mm in lead aVR was the most accurate pre-dictor of severe LM/3VD. However, its positive predictivevalue was 56%, which was moderate. More importantly, thenegative predictive value of ST-segment elevation �1.0mm in lead aVR for detection of severe LM/3VD was 98%,which was very high. Absence of this finding was rarelyassociated with severe LM/3VD. If ST-segment elevation�1.0 mm in lead aVR is absent, treatment with upstreamclopidogrel is strongly recommended. Lead aVR has aunique position because the positive pole is oriented towardthe right upper side of the heart and looks into the leftventricular cavity from the right shoulder in the setting ofNSTE-ACS.23 Lead aVR is therefore referred to as a “cavitylead,” and ST-segment elevation in this lead might reflectglobal subendocardial ischemia.24 In patients with LM/3VD, severe extensive ischemia of the subendocardial layerleads to ST-segment elevation in lead aVR and extensiveST-segment depression in leads other than lead aVR. Themagnitude of these changes is thought to reflect severity ofLM/3VD. In the present study, LM/3VD, especially severeLM/3VD, was associated with a greater degree and extent ofST-segment depression and a greater degree of ST-segmentelevation in lead aVR. A meta-analysis of 12,030 patientswith stable coronary artery disease enrolled in 60 studiesdemonstrated that amount of ST-segment depression duringexercise stress testing is strongly associated with criticalcoronary artery disease such as LM/3VD.25 Furthermore, agreater degree and extent of ST-segment depression, notonly its presence or absence, has been shown to correlatewith an increased likelihood of LM/3VD and poor outcomesin patients with NSTE-ACS.17,19,20 The present study dem-nstrated that the value of ST-segment elevation in lead
VR for detection of severe LM/3VD surpassed that of
T-segment depression in other leads in patients withSTE-ACS.Recently approved antiplatelet agents such as prasugrel

nd ticagrelor, a new reversible agent, have been shown toecrease ischemic events compared to clopidogrel, but theormer increased the risk of perioperative bleeding7 and theatter did not decrease the risk of perioperative bleeding.26

Until an antiplatelet agent that decreases ischemic eventsand decreases perioperative bleeding compared to clopi-dogrel becomes available, some patients will be exposed toa risk of urgent CABG-related bleeding caused by upstreamdual antiplatelet therapy.

This study was retrospective, performed at a single cen-ter, and included a small number of patients who underwentcoronary angiography during hospitalization. However, theproportion of patients undergoing CABG during hospital-ization in this study (14%) was similar to that in previousstudies.4,10,12 Moreover, because our subjects underwentcardiac catheterization a median of 3 days after admission,our data on clinical outcomes according to angiographicfindings cannot be generalized to hospitals that provideearly invasive strategies. Further studies in larger numbersof patients are needed to verify our results.

1. Anderson JL, Adams CD, Antman EM, Bridges CR, Califf RM, CaseyDE Jr, Chavey WE II, Fesmire FM, Hochman JS, Levin TN, LincoffAM, Peterson ED, Theroux P, Wenger NK, Wright RS, Smith SC Jr,Jacobs AK, Halperin JL, Hunt SA, Krumholz HM, Kushner FG, LytleBW, Nishimura R, Ornato JP, Page RL, Riegel B. ACC/AHA guide-lines for the management of patients with unstable angina/non–ST-segment elevation myocardial infarction: a report of the AmericanCollege of Cardiology/American Heart Association task force on prac-tice guidelines (writing committee to revise the 2002 guidelines for themanagement of patients with unstable angina/non-st-elevation myo-cardial infarction). Circulation 2007;116:803–877.

2. Bassand JP, Hamm CW, Ardissino D, Boersma E, Budaj A, Fernán-dez-Avilés F, Fox KA, Hasdai D, Ohman EM, Wallentin L, Wijns W.Guidelines for the diagnosis and treatment of non-ST-segment eleva-tion acute coronary syndromes. The task force for the diagnosis andtreatment of non-ST-segment elevation acute coronary syndromes ofthe European Society of Cardiology. Eur Heart J 2007;28:1598–1660.

3. Berger JS, Frye CB, Harshaw Q, Edwards FH, Steinhubl SR, BeckerRC. Impact of clopidogrel in patients with acute coronary syndromesrequiring coronary artery bypass surgery: a multicenter analysis. J AmColl Cardiol 2008;52:1693–1701.

4. Ebrahimi R, Dyke C, Mehran R, Manoukian SV, Feit F, Cox DA,Gersh BJ, Ohman EM, White HD, Moses JW, Ware JH, Lincoff AM,Stone GW. Outcomes following pre-operative clopidogrel administra-tion in patients with acute coronary syndromes undergoing coronaryartery bypass surgery: the ACUITY (Acute Catheterization and UrgentIntervention Triage strategY) trial. J Am Coll Cardiol 2009;53:1965–1972.

5. Mehta RH, Roe MT, Mulgund J, Ohman EM, Cannon CP, Gibler WB,Pollack CV Jr, Smith SC Jr, Ferguson TB, Peterson ED. Acute clopi-dogrel use and outcomes in patients with non–ST-segment elevationacute coronary syndromes undergoing coronary artery bypass surgery.J Am Coll Cardiol 2006;48:281–286.

6. Fox KA, Mehta SR, Peters R, Zhao F, Lakkis N, Gersh BJ, Yusuf S.Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Eventstrial. Benefits and risks of the combination of clopidogrel and aspirinin patients undergoing surgical revascularization for non-ST-elevationacute coronary syndrome: the Clopidogrel in Unstable Angina toPrevent Recurrent Ischemic Events (CURE) trial. Circulation 2004;110:1202–1208.

7. Wiviott SD, Braunwald E, McCabe CH, Montalescot G, Ruzyllo W,Gottlieb S, Neumann FJ, Ardissino D, De Servi S, Murphy SA,Riesmeyer J, Weerakkody G, Gibson CM, Antman EM; TRITON-TIMI 38 Investigators. Prasugrel versus clopidogrel in patients with
acute coronary syndromes. N Engl J Med 2007;357:2001–2015.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2


8. Barker CM, Anderson HV. Acute coronary syndromes: don’t bypassthe clopidogrel. J Am Coll Cardiol 2009;53:1973–1974.

9. Deyell MW, Ghali WA, Ross DB, Zhang J, Hemmelgarn BR; AlbertaProvincial Project for Outcome Assessment in Coronary Heart Disease(APPROACH) Investigators. Timing of nonemergent coronary arterybypass grafting and mortality after non–ST elevation acute coronarysyndrome. Am Heart J 2010;159:490–496.

0. Sadanandan S, Cannon CP, Gibson CM, Murphy SA, DiBattiste PM,Braunwald E; TIMI Study Group. A risk score to estimate the likeli-hood of coronary artery bypass surgery during the index hospitaliza-tion among patients with unstable angina and non–ST-segment eleva-tion myocardial infarction. J Am Coll Cardiol 2004;44:799–803.

1. Chew DP, Mahaffey KW, White HD, Huang Z, Hoekstra JW, Fergu-son JJ, Califf RM, Aylward PE. Coronary artery bypass surgery inpatients with acute coronary syndromes is difficult to predict. AmHeart J 2008;155:841–847.

2. Mehta RH, Chen AY, Pollack CV Jr, Roe MT, Zalenski RJ, ClementsEA, Gibler WB, Ohman EM, Harrington RA, Peterson ED. Challengesin predicting the need for coronary artery bypass grafting at presenta-tion in patients with non–ST-segment elevation acute coronary syn-dromes. Am J Cardiol 2006;98:624–627.

3. Kosuge M, Kimura K, Ishikawa T, Ebina T, Shimizu T, Hibi K, TodaN, Tahara Y, Tsukahara K, Kanna M, Okuda J, Nozawa N, Ozaki H,Yano H, Umemura S. Predictors of left main or three-vessel disease inpatients who have acute coronary syndromes with non-ST-segmentelevation. Am J Cardiol 2005;95:1366–1369.

4. Cannon CP, McCabe CH, Stone PH, Rogers WJ, Schactman M,Thompson BW, Pearce DJ, Diver DJ, Kells C, Feldman T, WilliamsM, Gibson RS, Kronenberg MW, Ganz LI, Anderson HV, BraunwaldE. The electrocardiogram predicts one-year outcome of patients withunstable angina and non-Q wave myocardial infarction: results of theTIMI III registry ECG ancillary study. J Am Coll Cardiol 1997;30:133–140.

5. Barrabés JA, Figueras J, Moure C, Cortadellas J, Soler-Soler J. Prog-nostic value of lead aVR in patients with a first non–ST-segmentelevation acute myocardial infarction. Circulation 2003;108:814–819.

6. Matsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, YamagataK, Tomino Y, Yokoyama H, Hishida A. On behalf of the collaboratorsdeveloping the Japanese equation for estimated GFR. Revised equa-tions for estimated GFR from serum creatinine in Japan. Am J KidneyDis 2009;53:982–992.

7. Savonitto S, Cohen MG, Politi A, Hudson MP, Kong DF, Huang Y,Pieper KS, Mauri F, Wagner GS, Califf RM, Topol EJ, Granger CB.Extent of ST-segment depression and cardiac events in non-ST-seg-
ment elevation acute coronary syndromes. Eur Heart J 2005;26:2106–2113.
8. Westerhout CM, Fu Y, Lauer MS, James S, Armstrong PW, Al-HattabE, Califf RM, Simoons ML, Wallentin L, Boersma E; GUSTO-IVACS Trial Investigators. Short- and long-term risk stratification inacute coronary syndromes: the added value of quantitative ST-segmentdepression and multiple biomarkers. J Am Coll Cardiol 2006;48:939–947.

9. Yan RT, Yan AT, Mahaffey KW, White HD, Pieper K, Sun JL, PepineCJ, Biasucci LM, Gulba DC, Lopez-Sendon J, Goodman SG;SYNERGY Trial Investigators. Prognostic utility of quantifying evo-lutionary ST-segment depression on early follow-up electrocardio-gram in patients with non–ST-segment elevation acute coronary syn-dromes. Eur Heart J 2010;31:958–966.

0. Holmvang L, Clemmensen P, Lindahl B, Lagerqvist B, Venge P,Wagner G, Wallentin L, Grande P. Quantitative analysis of the admis-sion electrocardiogram identifies patients with unstable coronary arterydisease who benefit the most from early invasive treatment. J Am CollCardiol 2003;41:905–915.

1. Gorgels AP, Vos MA, Mulleneers R, de Zwaan C, Bar FW, WellensHJ. Value of the electrocardiogram in diagnosing the number ofseverely narrowed coronary arteries in rest angina pectoris. Am JCardiol 1993;72:999–1003.

2. Kosuge M, Kimura K, Ishikawa T, Ebina T, Hibi K, Tsukahara K,Kanna M, Iwahashi N, Okuda J, Nozawa N, Ozaki H, Yano H,Kusama I, Umemura S. Combined prognostic utility of ST segment inlead aVR and troponin T on admission in non–ST-segment elevationacute coronary syndromes. Am J Cardiol 2006;97:334–339.

3. Yu PN, Stewart JM. Subendocardial myocardial infarction with specialreference to the electrocardiographic changes. Am Heart J 1950;39:862–880.

4. Kligfield P, Gettes LS, Bailey JJ, Childers R, Deal BJ, Hancock EW,van Herpen G, Kors JA, Macfarlane P, Mirvis DM, Pahlm O, Rau-taharju P, Wagner GS. Recommendations for the standardization andinterpretation of the electrocardiogram: part I: The electrocardiogramand its technology: a scientific statement from the American HeartAssociation Electrocardiography and Arrhythmias Committee, Coun-cil on Clinical Cardiology; the American College of Cardiology Foun-dation; and the Heart Rhythm Society. Circulation 2007;115:1306–1324.

5. Detrano R, Gianrossi R, Mulvihill D, Lehmann K, Dubach P, ColomboA, Froelicher V. Exercise-induced ST segment depression in the di-agnosis of multivessel coronary disease: a meta analysis. J Am CollCardiol 1989;14:1501–1508.

6. Wallentin L, Becker RC, Budaj A, Cannon CP, Emanuelsson H, HeldC, Horrow J, Husted S, James S, Katus H, Mahaffey KW, Scirica BM,Skene A, Steg PG, Storey RF, Harrington RA, Freij A, Thorsén M;
PLATO Investigators. Ticagrelor versus clopidogrel in patients withacute coronary syndromes. N Engl J Med 2009;361:1045–1057.

Reperfusion by Primary Percutaneous Coronary Intervention inPatients With ST-Segment Elevation Myocardial Infarction Within

12 to 24 Hours of the Onset of Symptoms (from a ProspectiveNational Observational Study [PL-ACS])

Marek Gierlotka, MD, PhDa,*, Mariusz Gasior, MD, PhDa, Krzysztof Wilczek, MD, PhDa,Michal Hawranek, MD, PhDa, Janusz Szkodzinski, MD, PhDa, Piotr Paczek, MD, PhDd,

Andrzej Lekston, MD, PhDa, Zbigniew Kalarus, MD, PhDb, Marian Zembala, MD, PhDc,and Lech Polonski, MD, PhDa

The aim of the present study was to investigate whether reperfusion by primary percuta-neous coronary intervention (PCI) improves 12-month survival in late presenters withST-segment elevation myocardial infarction (STEMI). We analyzed 2,036 patients withSTEMI presenting 12 to 24 hours from onset of symptoms, without cardiogenic shock orpulmonary edema and not reperfused by thrombolysis, of 23,517 patients with STEMIenrolled in the Polish Registry of Acute Coronary Syndromes from June 2005 to August2006. An invasive approach was chosen in 910 (44.7%) of late presenters and 92% of themunderwent reperfusion by PCI. Patients with an invasive approach had lower mortalityafter 12 months than patients with a conservative approach (9.3% vs 17.9%, p <0.0001).The benefit of an invasive approach was also observed after multivariate adjustment witha relative risk 0.73 for 12-month mortality (95% confidence interval 0.56 to 0.96) and in asubpopulation of patients selected by a propensity-score matching procedure with anadjusted relative risk 0.73 for 12-month mortality (0.58 to 0.99). In conclusion, almost 1/2of late presenters with STEMI were considered eligible for reperfusion by primary PCI.These patients had a lower 12-month mortality rate than they would have had if they hadbeen treated conservatively, which supports the idea of late reperfusion in STEMI. How-ever, whether all late presenters with STEMI should be treated invasively remains unan-swered. Nevertheless, until a randomized trial is undertaken, late presenters with STEMIcould be considered for reperfusion by primary PCI. © 2011 Elsevier Inc. All rights
reserved. (Am J Cardiol 2011;107:501–508)
Optimal management for patients with ST-segment ele-vation myocardial infarction (STEMI) who arrive at a hos-pital late remains uncertain. A few recently published stud-ies have suggested that reperfusion by percutaneouscoronary intervention (PCI) could be beneficial to patients ifperformed just after 12 hours from onset of symptoms.1–7 Inaddition, the most recent European guidelines on STEMIrecommend PCI for late presenters with ongoing ischemia.8

The aim of our analysis was to assess the current use ofinvasive treatment and mechanical reperfusion by primaryPCI applied 12 to 24 hours from onset of symptoms inpatients with STEMI arriving at a hospital 12 to 24 hoursfrom onset of symptoms and to determine the influence of

aThird and bFirst Departments of Cardiology and cDepartment of Car-diac Surgery and Transplantology, Silesian Center for Heart Diseases,Medical University of Silesia, Zabrze, Poland; dCardiology Department,Polish Medical Group, Sosnowiec, Poland. Manuscript received July 12,2010; revised manuscript received and accepted October 7, 2010.

The Polish Registry of Acute Coronary Syndromes (PL-ACS) is sup-ported by an unrestricted grant from the Polish Ministry of Health, Warsaw,Poland.

E-mail address: [email protected] (M. Gierlotka).

this invasive approach on 12-month mortality in clinicalpractice.

Methods

We used data from the Polish Registry of Acute Coro-nary Syndromes (PL-ACS), for which the method and re-sults of the first 100,193 patients have been described.9 Inbrief, the PL-ACS is an ongoing, nationwide, multicenter,prospective, observational study of consecutively hospital-ized patients with the entire spectrum of ACSs in Poland.The pilot phase of the registry commenced in October 2003in the Silesia region. In subsequent months, further regionswere opened and, since June 2005, all Polish regions collectdata for the PL-ACS.

Hospitals were invited to enter the registry if they had acoronary care unit, a cardiology unit, a cardiac surgical unit,an internal medicine unit, or an intensive care unit or if theyhospitalized �10 patients with ACS per year.

A detailed protocol with inclusion and exclusion criteria,methods and logistics, and definitions of all fields in theregistry dataset was prepared before the registry was started.However, it has since been revised 2 times. In May 2004,
the protocol was adapted to be compatible with the Cardi-
www.ajconline.org


bp

mvwWvSsugpa

P

ility


ology Audit and Registration Data Standards (CARDS).10

Nevertheless, the PL-ACS case-report form (CRF) coversonly part of the CARDS dataset. A second revision (May2005) added new fields to the CRF, which included exactdates and times of onset of symptoms, coronary angiogra-phy, and PCI procedures.

According to the protocol, all admitted patients withsuspected ACS were screened for eligibility to enter theregistry but were not enrolled until the ACS was confirmed.Patients were then classified as having unstable angina,non-STEMI, or STEMI. STEMI was defined as the presenceof (1) ST-segment elevation consistent with infarction of�2 mm in contiguous chest leads and/or ST-segment ele-vation of �1 mm in �2 standard leads or new left bundleranch block and (2) positive cardiac necrosis markers. If aatient was hospitalized during the same ACS in �1 hos-

pital (transferred patient), all hospitals were required tocomplete the registry data. These hospitalizations werelinked together during data management and were analyzedas 1 ACS.

Data were collected by skilled physicians who were incharge of each patient and entered directly into an electronicCRF or temporarily printed onto a CRF before being trans-ferred to an electronic CRF. Internal checks for missing orconflicting data and values markedly out of the expectedrange were implemented by the software. In the SilesianCenter for Heart Diseases (Zabrze, Poland) data manage-ment and analysis center, further edit checks were applied ifnecessary.

All-cause mortality data with exact dates of deaths wereobtained from official mortality records from the NationalHealth Fund. Vital status at 12 months after STEMI wasavailable for all patients enrolled up to August 2006.

For the present analysis, we selected patients withSTEMI who were enrolled during a consecutive 15-monthperiod, from the time of the second protocol revision (June2005) to August 2006. In addition, we excluded patientswith severe hemodynamic disturbances on admission such

Table 1Baseline clinical characteristics of patients assigned to invasive and conse

Variable

Age (years), mean � SDWomenSmokersHypertensionDiabetes mellitusObesity (body mass index �30 kg/m2)

revious myocardial infarctionPrevious coronary bypassPrevious percutaneous coronary interventionsCardiac arrest before admissionAnterior wall infarct locationHeart rate on admission (beats/min), mean � SDOther than sinus rhythmSystolic blood pressure (mm Hg), mean � SDKillip class II on admissionSymptom-onset-to-admission time (hours), median (interquartile range)Hospitalization in hospital with percutaneous coronary intervention capab

as pulmonary edema (Killip class III) and cardiogenic shock m

(Killip class IV) for whom the recommended window forinvasive treatment exceeded 12 hours. Furthermore, patientsinitially treated with thrombolysis were excluded.

Symptom-onset-to-admission time was calculated for ev-ery patient as an absolute difference in minutes betweendates and times of hospital admission and onset of symp-toms. Only patients with symptom-onset-to-admissiontimes 12 to 24 hours comprised the study population. Twogroups of patients were then identified based on whethercoronary angiography had been performed and time of theinvasive procedure.

Patients were included in the invasive-approach group ifthey had coronary angiography performed 12 to 24 hoursfrom onset of symptoms. All other patients (treated nonin-vasively or with coronary angiography performed �24hours from symptom onset) were included in the conserva-tive-approach group.

The main outcome measurement was 12-month all-causemortality. In-hospital outcomes were death from any reason,recurrent MI (defined as an ischemic event that met Euro-pean Society of Cardiology/American College of Cardiol-ogy criteria for reinfarction and was evidently clinicallydistinct from the index event at time of admission),11 andstroke (defined as an acute neurologic deficit that lasted�24 hours and affected the ability to perform daily activi-ties or resulted in death).

Continuous variables are reported as mean � SD oredian and interquartile range as appropriate. Categorical

ariables are expressed as percentages. Student’s t test and,hen the assumption of normality was violated, Mann-hitney U test were used for comparison of continuous

ariables. Normality of distribution was checked with thehapiro-Wilk test. To compare categorized variables, chi-quare test was used. Follow-up mortality was analyzedsing the Kaplan-Meier method, and differences betweenroups were compared with log-rank test. Multivariate Coxroportional hazard model regression was performed todjust the influence of the invasive approach on 12-month

approaches

Conservative Approach Invasive Approach p Value(n � 1,126) (n � 910)

67.2 � 13.2 63.3 � 11.8 �0.000144.0% 32.4% �0.000129.1% 42.5% �0.000163.1% 63.3% 0.9423.9% 23.4% 0.8017.9% 18.2% 0.8216.6% 11.2% 0.0005

5.3% 2.9% 0.00581.2% 1.0% 0.591.0% 1.7% 0.18

41.3% 45.2% 0.08081 � 21 78 � 17 0.000110.0% 4.8% �0.0001

139 � 28 136 � 27 0.03224.0% 11.8% �0.0001

16 (14–21) 15 (12–18) �0.000126.1% 100% �0.0001

rvative

ortality, including all parameters listed in Table 1, except

ylpcpmsoostSb

Figure 1. Analysis scheme of ST-segmen

CS

Diuretic 31.4% 19.2% �0.0001

SR

DC361

503Coronary Artery Disease/Late Reperfusion by PCI in STEMI

for place of hospitalization. Hazard ratios and 95% confi-dence intervals were calculated.

Due to differences in baseline characteristics betweengroups, a propensity-score method was used to identifycomparable patients treated with invasive and conserva-tive approaches.12 Multivariate logistic regression anal-

sis, including all baseline characteristic parametersisted in Table 1 except for place of hospitalization, waserformed to calculate the predicted probability of re-eiving an invasive approach (propensity score) for everyatient. Patients from the conservative group were thenatched to patients from the invasive group. Because of

ubstantial differences in baseline parameters, only 68%f patients remained after the matching procedure. Theriginal multivariate Cox proportional hazard model regres-ion was recalculated in the matched groups to further adjusthe influence of an invasive approach on 12-month mortality.ubgroup analysis to identify patients who would potentially

t elevation myocardial infarction.

Table 3In-hospital and 12-month outcomes

Variable ConservativeApproach

InvasiveApproach

p Value

(n � 1,126) (n � 910)

troke in hospital 0.6% 0.7% 0.92ecurrent myocardial infarctionin hospital

5.2% 2.6% 0.0041

eath in hospital 6.8% 2.6% �0.0001ombined in-hospital outcome 12.2% 5.7% �0.00010-day mortality 10.0% 5.1% �0.0001-month mortality 14.2% 7.5% �0.00012-month mortality 17.9% 9.3% �0.0001

Table 2In-hospital treatment procedures and pharmacotherapy

Variable ConservativeApproach

InvasiveApproach

p Value

(n � 1,126) (n � 910)

Coronary angiography 15.0% 100% �0.0001Door-to-angiography time

(minutes)�90 0.6% 86.2%91–120 1.2% 4.6% �0.0001�120 98.2% 9.2%

Primary percutaneous coronaryintervention

12.1% 91.8% �0.0001

Initial Thrombolysis InMyocardial Infarctiongrade 2 or 3 flow

22.6% 21.8% 0.85

Final Thrombolysis InMyocardial Infarctiongrade 2 flow

6.7% 7.2% 0.84

Final Thrombolysis InMyocardial Infarctiongrade 3 flow

90.3% 87.3% 0.33

Stent implantation 93.4% 90.2% 0.23Coronary bypass 0.3% 1.2% 0.011Aspirin 95.7% 97.0% 0.10Thienopyridines 42.0% 87.8% �0.0001Glycoprotein IIb/IIIa receptor

inhibitor3.6% 22.0% �0.0001

Heparin 76.9% 69.8% 0.0003� Blocker 74.7% 79.0% 0.022

alcium antagonist 5.0% 3.7% 0.18tatin 74.9% 86.6% �0.0001

Angiotensin-convertingenzyme inhibitor

71.1% 80.8% �0.0001

Nitrate 53.0% 27.6% �0.0001

enefit from an invasive approach was performed with univar-

pcbyMKcp

dCttaC

C


iate Cox proportional hazard model regression in the matchedpopulation.

A 2-sided p value �0.05 was considered statisticallysignificant. For all calculations, STATISTICA 7.1 (StatSoft,Inc., Tulsa, Oklahoma) was used.

Results

In total 23,517 patients with STEMI were hospitalized in385 hospitals from June 2005 to August 2006. After ex-cluding patients with pulmonary edema and cardiogenicshock on admission and, subsequently, those whose reper-fusion was primarily by thrombolysis, there were 19,453patients, and symptom-onset-to-admission time was calcu-lated. A prehospital delay from 12 to 24 hours was observedin 2,036 patients (10.5%). Criteria for an invasive approachwere fulfilled by 44.7% of them. The remaining patientswere assigned to the conservative-approach group (Figure 1).

Differences in clinical characteristics between groupsresented in Table 1 show that an invasive approach washosen more frequently in patients with a more favorableaseline risk profile. These patients were on average 4 yearsounger, more often men, less frequently had a history ofI and previous bypass surgery, and were less often inillip class II on admission. About 1/4 of patients from the

onservative group were treated in hospitals with PCI ca-ability.

Most from the invasive group had coronary angiographyone within 90 minutes of time of admission (Table 2).oronary angiography was also performed in 15% of pa-

ients assigned to the conservative approach. In almost allhese patients, door-to-angiography time exceeded 2 hours,nd symptom-onset-to-angiography time was �24 hours.

Figure 2. Adjusted relative risk of 12-month mortality for an invasive apABG � coronary artery bypass grafting; CI � confidence interval.

onsequently, PCI was performed in 12% of patients in

conservative group, whereas 92% of patients in the invasivegroup received primary PCI. Significant differences werealso observed in pharmacotherapy during hospitalizationwith the invasive group having greater use of thienopyri-dines, glycoprotein IIb/IIIa inhibitors, � blockers, statins,and angiotensin-converting enzyme-inhibitors and less fre-quent use of nitrates and diuretics.

In-hospital and 12-month outcomes are presented inTable 3. Patients in whom an invasive approach was chosenhad approximately 1/2 the mortality of the conservativegroup after 30 days and 12 months. A similar trend wasobserved for in-hospital reinfarctions. After multivariateadjustment (Figure 2), an invasive approach remained sig-nificantly associated with lower relative risk of 12-monthmortality (hazard ratio 0.73, 95% confidence interval 0.56to 0.96).

The lower 12-month mortality of an invasive approachwas also observed in the subpopulation of patients selectedby the propensity-score matching procedure (Figure 3,Table 4). Of note, however, the guidelines’ recommendedpharmacologic treatment was more widely used in patientswho received an invasive approach. In addition, we did notfind any specific subgroup of patients for whom an invasiveapproach could be potentially harmful (Figure 4).

Discussion

The main finding of our analysis is that current clinicalpractice of invasive treatment of selected patients withSTEMI with late presentation 12 to 24 hours from onset ofsymptoms is not harmful and leads to lower 12-monthmortality compared to a conservative approach. The benefitin mortality, primarily large without adjustment, remained

by multivariate Cox proportional hazard model regression (n � 2,036).
proach
significant after multivariate analysis and propensity-score

month m


matching, which were essential due to significant differ-ences in clinical characteristics that favored patients se-lected for invasive treatment. Therefore, our study stronglysupports the idea of late reperfusion in patients with STEMI,and to our knowledge, this is the first report on mortalitydecrease. Nevertheless, there are several points to be dis-cussed when interpreting and comparing the results to otherstudies.

A comment should be made regarding the design of thestudy. It is a retrospective analysis of a large prospectiveregistry. The definition of an invasive approach as coronaryangiography performed in the period of 12 to 24 hours fromonset of symptoms was, in our opinion, the best compro-mise between accuracy and everyday clinical practice. The

Figure 3. Kaplan-Meier mortality curves for total 12-

intension was to include all patients potentially eligible for

primary PCI, making the analysis as close as possible to arandomized setting. As a result, 15% of patients in theconservative-approach group underwent coronary angiogra-phy and, subsequently, most underwent PCI. Because door-to-angiography time exceeded 2 hours in almost all of theseprocedures, we can probably treat them as planned proce-dures instead of immediate PCI for reperfusion. Further-more, we did not know whether there was ongoing ischemiaon admission, although in those cases the benefit of imme-diate revascularization is much more endorsed. Reasons asto why each patient was treated invasively are also missing.We can assume that, based on the symptoms and signs,these patients were good candidates for late reperfusion forthe attending physician. Consequently, the proportion of

ortality in (A) all patients and (B) matched patients.

patients with ongoing ischemia may have been larger with

blioS

ldse

T

O


an invasive approach, which could have biased the resultsseriously. Interestingly, in current clinical practice in Po-land, as many as 45% of late presenters, in the period of 12to 24 hours, were considered eligible for reperfusion ther-apy. Of note, it was before the most recent recommenda-tions were published.8 Consequently, it is not surprising thataseline characteristics differed between groups, with aower risk profile in the group of patients who received annvasive approach. This skewed risk profile is typical for allbservational studies comparing PCI to medical therapy inTEMI, including other studies of late reperfusion.1,2 These

2 reported analyses showed a significant mortality benefit of

Table 4Comparison of clinical characteristics, treatment, and outcomes of patientpropensity-score method (n � 1,386)

Variable

Baseline clinical characteristicsAge (years), mean � SDWomenSmokersHypertensionDiabetes mellitusObesityPrevious myocardial infarctionPrevious coronary bypassPrevious percutaneous coronary interventionCardiac arrest before admissionAnterior wall infarct locationHeart rate on admission (beats/min), mean � SDOther than sinus rhythmSystolic blood pressure (mm Hg), mean � SDKillip class II on admissionSymptom-onset-to-admission time (hours), median (interquartile range)Hospitalization in hospital with percutaneous coronary intervention capareatmentCoronary angiographyPrimary percutaneous coronary intervention

Initial Thrombolysis In Myocardial Infarction grade 2 or 3 flowFinal Thrombolysis In Myocardial Infarction grade 2 flowFinal Thrombolysis In Myocardial Infarction grade 3 flowStent implantation

Coronary bypassAspirinThienopyridinesGlycoprotein IIb/IIIa receptor inhibitorHeparin� BlockerCalcium antagonistStatinAngiotensin-converting enzyme inhibitorNitrateDiureticutcomesStroke in hospitalRecurrent myocardial infarction in hospitalDeath in hospitalCombined in-hospital outcome30-day mortality6-month mortality12-month mortality

late reperfusion that was not significant after adjustment in

multivariate analysis1 or showed a strong trend towardower mortality after propensity-score matching.2 Also, ran-omized trials performed thus far have not shown anyignificant decrease in the number of major adverse clinicalvents including mortality.3–7 In fact, the Beyond 12-hours

Reperfusion AlternatiVe Evaluation (BRAVE-2) study isthe only randomized study comparing primary PCI to med-ical treatment in a period of potentially successful reperfu-sion therapy of 12 to 48 hours.3,4 The investigators showedsignificantly smaller final infarct size and improvement inmyocardial salvage, which was a primary end point of thestudy. The large Occluded Artery Trial (OAT) compared

nvasive approach and patients with conservative approach matched by

Conservative Approach Invasive Approach p Value(n � 693) (n � 693)

64.6 � 13.2 64.7 � 11.7 0.8936.4% 38.2% 0.4737.2% 37.1% 0.9664.2% 61.6% 0.3222.9% 23.7% 0.7516.7% 16.2% 0.7711.8% 12.3% 0.804.0% 3.8% 0.780.9% 1.0% 0.781.3% 1.3% 1.0

41.9% 41.7% 0.9678 � 17 79 � 18 0.72

5.8% 5.9% 0.90138 � 27 138 � 26 0.85

13.3% 14.3% 0.5915 (13–19) 15 (13–19) 0.49

27.9% 100% �0.0001

16.5% 100% �0.000113.7% 91.3% �0.000119.4% 21.6% 0.624.3% 7.9% 0.21

93.6% 85.8% 0.03694.7% 89.7% 0.120.3% 1.6% 0.012

95.4% 96.7% 0.2243.4% 87.2% �0.00014.5% 21.7% �0.0001

74.9% 69.7% 0.03175.6% 77.3% 0.454.9% 4.0% 0.44

73.0% 86.2% �0.000169.4% 79.4% �0.000151.5% 28.4% �0.000124.4% 20.6% 0.095

0.6% 0.9% 0.534.9% 2.9% 0.0525.1% 2.9% 0.039

10.0% 6.4% 0.0147.7% 5.8% 0.16

11.4% 8.7% 0.08914.3% 10.5% 0.034

s with i

bility

coronary interventions to medical treatment in persistent

orrtrpfisistr

agimfiadmcdapssmi


occluded coronary arteries 3 to 28 days (median 8) after MI,and it reported similar mortality, reinfarction rate, and se-vere heart failure New York Heart Association class IVafter 5 years in the 2 groups.13 However, the median delayf 8 days makes the coronary interventions in this studyather subacute PCI than primary PCI for STEMI. In theirecently published study, Busk et al7 reported that substan-ial myocardial salvage was observed even when the infarct-elated artery was totally occluded at time of primary PCIerformed 12 to 72 hours from onset of STEMI. Althoughnal infarct size after primary PCI was larger in late pre-enters than in early presenters, the 12-hour limit failed todentify which patients had the potential for myocardialalvage. Together with the results of the BRAVE-2 study,hese investigators provide a potential explanation of theesults of our study.

Figure 4. Subgroup analysis of 12-month mortality for conservative andand multivariate adjustment in matched groups of patients (n � 1,38gender, smoking status, hypertension, diabetes, obesity, previous myocneous coronary intervention, cardiac arrest before admission, electrocathan sinus rhythm on admission, systolic blood pressure on admission, Kas in Figure 2.

Other factors that may influence the effect of an invasive m

pproach on mortality are significant differences betweenroups in the prescribed pharmacotherapy during hospital-zation. Patients treated invasively received more completeedical therapy than patients treated conservatively. Thisnding has also been reported previously2 and cannot bettributed to selection bias, although the differences did notisappear in the propensity-score matched subgroups. Aore probable reason is better medical care provided by

ardiologists in hospitals with available primary PCI proce-ures.14 In contrast, this can be an additional potentialdvantage of choosing a strategy of late reperfusion inatients with STEMI admitted after 12 hours from onset ofymptoms. Of note, older age and certain co-morbiditiesuch as diabetes and previous MI were associated with aore evident benefit from an invasive approach. Interest-

ngly, in patients with Killip class II on admission, 12-

ve approaches by univariate Cox proportional hazard model regressionltivariate Cox proportional hazard model regression adjusted for age,infarction, previous coronary artery bypass grafting, previous percuta-phic anterior wall location of infarction, heart rate on admission, otherass on admission, and symptom-onset-to-admission time. Abbreviation

invasi6). Muardial

rdiograillip cl

onth mortality was similar in the invasive and conserva-

1

1

1

1

1

1


tive groups. This result is somewhat surprising becausehemodynamic instability is thought to be 1 of the indica-tions for an invasive approach in this group of patients.15

We do not have an explanation for this finding.There are several limitations to our study, and some were

discussed earlier. The PL-ACS is a prospective observa-tional study that includes data from various regions andhospitals throughout Poland. Because patients were treatedat different hospitals in different regions, there are inherentuncontrolled differences in clinical practice. In addition, theretrospective nature of our study is a potential weakness.The major limitation is some important measurements suchas status of ongoing ischemia on admission and detailedreasons for choosing an invasive or conservative strategy ina particular case, which was discussed earlier. Furthermore,even after propensity-score matching, the groups are likelyto be biased by potentially important parameters that are notavailable in the registry. In addition, all patients from theinvasive-approach group were treated in PCI centers com-pared to only 1/4 of patients from the conservative group.An adjustment for this difference was not possible with thepresent design of our study. Therefore, the reported signif-icant mortality decrease after adjustment should be inter-preted with caution.

Acknowledgment: We thank all the physicians and nurseswho participated in the PL-ACS Registry, members of theexpert committee, regional co-ordinators, and employeesof the National Health Fund of Poland for their logisticsupport.

Appendix

PL-ACS expert committee: Lech Polonski, MD, PhD(chairman), Mariusz Gasior, MD, PhD (cochairman), MarekGierlotka, MD, PhD (cochairman), and Zbigniew Kalarus,MD, PhD (cochairman), Zabrze; Andrzej Cieslinski, MD,PhD, Poznan; Jacek Dubiel, MD, PhD, Cracow; Robert Gil,MD, PhD, Grzegorz Opolski, MD, PhD, and WitoldRuzyllo, MD, PhD, Warsaw; Michal Tendera, MD, PhD,Katowice; and Marian Zembala, MD, PhD, Zabrze.

1. Zahn R, Schiele R, Schneider S, Gitt AK, Wienbergen H, Seidl K,Bossaller C, Hauf GF, Gottwik M, Altmann E, Rosahl W, Senges J.Primary angioplasty versus no reperfusion therapy in patients withacute myocardial infarction and a pre-hospital delay of �12–24 hours:results from the pooled data of the maximal individual therapy in acutemyocardial infarction (MITRA) registry and the myocardial infarctionregistry (MIR). J Invasive Cardiol 2001;13:367–372.

2. Elad Y, French WJ, Shavelle DM, Parsons LS, Sada MJ, Every NR.Primary angioplasty and selection bias in patients presenting late (�12
h) after onset of chest pain and ST elevation myocardial infarction.J Am Coll Cardiol 2002;39:826–833.
3. Schömig A, Mehilli J, Antoniucci D, Ndrepepa G, Markwardt C, DiPede F, Nekolla SG, Schlotterbeck K, Schühlen H, Pache J, SeyfarthM, Martinoff S, Benzer W, Schmitt C, Dirschinger J, Schwaiger M,Kastrati A. Mechanical reperfusion in patients with acute myocardialinfarction presenting more than 12 hours from symptom onset: arandomized controlled trial. JAMA 2005;293:2865–2872.

4. Parodi G, Ndrepepa G, Kastrati A, Conti A, Mehilli J, Sciagra R,Schwaiger M, Antoniucci D, Schømig A. Ability of mechanical reper-fusion to salvage myocardium in patients with acute myocardial in-farction presenting beyond 12 hours after onset of symptoms. AmHeart J 2006;152:1133–1139.

5. Abbate A, Biondi-Zoccai GG, Appleton DL, Erne P, SchoenenbergerAW, Lipinski MJ, Agostoni P, Sheiban I, Vetrovec GW. Survival andcardiac remodeling benefits in patients undergoing late percutaneouscoronary intervention of the infarct-related artery: evidence from ameta-analysis of randomized controlled trials. J Am Coll Cardiol2008;51:956–964.

6. Silva JC, Rochitte CE, Júnior JS, Tsutsui J, Andrade J, Martinez EE,Moffa PJ, Menegheti JC, Kalil-Filho R, Ramires JF, Nicolau JC. Latecoronary artery recanalization effects on left ventricular remodellingand contractility by magnetic resonance imaging. Eur Heart J 2005;26:36–43.

7. Busk M, Kaltoft A, Nielsen SS, Bøttcher M, Rehling M, Thuesen L,Bøtker HE, Lassen JF, Christiansen EH, Krusell LR, Andersen HR,Nielsen TT, Kristensen SD. Infarct size and myocardial salvage afterprimary angioplasty in patients presenting with symptoms for �12 hvs. 12–72 h. Eur Heart J 2009;30:1322–1330.

8. Van de Werf F, Bax J, Betriu A, Blomstrom-Lundqvist C, Crea F, FalkV, Filippatos G, Fox K, Huber K, Kastrati A, Rosengren A, Steg PG,Tubaro M, Verheugt F, Weidinger F, Weis M. Management of acutemyocardial infarction in patients presenting with persistent ST-seg-ment elevation. Eur Heart J 2008;29:2909–2945.

9. Polonski L, Gasior M, Gierlotka M, Kalarus Z, Cieslinski A, DubielJS, Gil RJ, Ruzyllo W, Trusz-Gluza M, Zembala M, Opolski G. PolishRegistry of Acute Coronary Syndromes (PL-ACS). Characteristicstreatments and outcomes of patients with acute coronary syndromes inPoland. Kardiol Pol 2007;65:861–872.

0. Flynn MR, Barrett C, Cosío FG, Gitt AK, Wallentin L, Kearney P,Lonergan M, Shelley E, Simoons ML. The Cardiology Audit andRegistration Data Standards (CARDS), European data standards forclinical cardiology practice. Eur Heart J 2005;26:308–313.

1. The Joint European Society of Cardiology/American College of Car-diology Committee. Myocardial infarction redefined—a consensusdocument of the Joint European Society of Cardiology/American Col-lege of Cardiology committee for the redefinition of myocardial in-farction. Eur Heart J 2000;21:1502–1513.

2. Rosenbaum PR, Rubin DR. The central role of the propensity score inobservational studies for causal effects. Biometrika 1983;70:41–55.

3. Hochman JS, Lamas GA, Buller CE, Dzavik V, Reynolds HR, Abram-sky SJ, Forman S, Ruzyllo W, Maggioni AP, White H, Sadowski Z,Carvalho AC, Rankin JM, Renkin JP, Steg PG, Mascette AM, SopkoG, Pfisterer ME, Leor J, Fridrich V, Mark DB, Knatterud GL. Coro-nary intervention for persistent occlusion after myocardial infarction.N Engl J Med 2006;355:2395–2407.

4. Casale PN, Jones WL, Wolf FE, Pei Y, Eby LM. Patients treated bycardiologists have a lower in-hospital mortality for acute myocardialinfarction. J Am Coll Cardiol 1998;32:885–889.

5. Antman EM, Hand M, Armstrong PW, Bates ER, Green LA, Halasya-mani LK, Hochman JS, Krumholz HM, Lamas GA, Mullany CJ,Pearle DL, Sloan MA, Smith SC Jr, Anbe DT, Kushner FG, Ornato JP,Jacobs AK, Adams CD, Anderson JL, Buller CE, Creager MA, Et-tinger SM, Halperin JL, Hunt SA, Lytle BW, Nishimura R, Page RL,Riegel B, Tarkington LG, Yancy CW. 2007 Focused update of the
ACC/AHA 2004 guidelines for the management of patients with ST-elevation myocardial infarction. Circulation 2008;117:296–329.

Nr

0d

Percutaneous Coronary Intervention for Non ST-Elevation AcuteCoronary Syndromes: Which, When and How?

Robert K. Riezebos, MDa,*, Jan G.P. Tijssen, PhDb, Freek W.A. Verheugt, MD, PhDa, andGerrit J. Laarman, MD, PhDc

The presentation of patients with suspected non ST-elevation acute coronary syndromes isquite diverse. Therefore, the diagnostic workup and choice of treatment may vary accord-ingly. Major issues regarding the evaluation are the likelihood of the diagnosis and the riskfor adverse events. These factors should guide the choice of diagnostic test. Patients withincreased risk for ischemic events and patients with recurrent ischemia are most likely tobenefit from revascularization. In addition, when percutaneous coronary intervention isconsidered, evidence suggests that sufficient time should be allowed for pharmacologicstabilization, reducing the possibility of periprocedurally inflicted myocardial infarction.However, postponement of intervention may lead to an increase of new spontaneous events,and high-risk patients should apply for revascularization soon after pharmacologic stabi-lization. The extent of revascularization performed by percutaneous coronary interventiondepends predominantly on patient characteristics and anatomy but should be limited toflow-obstructive lesions. In conclusion, patients presenting with non–ST elevation acutecoronary syndromes constitute a very diverse population; diagnostic workup, treatment,and the timing of a possible intervention should be tailored individually. © 2011 Elsevier
Inc. All rights reserved. (Am J Cardiol 2011;107:509–515)
Ia

cttitb

Patients with chest pain represent a large and increasingproportion of all acute medical presentations worldwide. Ofall those presenting for evaluation, only a minority haveacute coronary syndromes (ACS). Distinguishing which pa-tients have ACS remains a diagnostic challenge. The prin-cipal pathophysiologic mechanism of ACS is myocardialunderperfusion, which is caused by atherosclerotic plaquerupture or erosion, with different degrees of superimposedthrombus.1,2 Electrocardiography provides the initial clas-sification. Patients are divided into those with persistentST-segment elevation and those without persistent ST-seg-ment elevation or non ST-elevation ACS (NSTEACS). Inthis review, we discusses the diagnostic challenges whenNSTEACS are suspected. In addition, we address the role ofrisk stratification in relation to the choice of treatment strat-egy. When an invasive approach is preferred, an importantissue is the timing of the intervention. The available evi-dence on this topic is discussed in detail. We conclude withthe evidence regarding the type and extent of revasculariza-tion in patients with multivessel disease.

Diagnostics and Risk Assessment

In patients presenting with suspected NSTEACS, 2 ma-jor issues must be addressed. The first challenge is to con-firm the diagnosis. Guidelines recommend the use of ele-mentary tools, such as symptoms, risk profile for coronary

aOnze Lieve Vrouwe Gasthuis; bAcademic Medical Center, Universityof Amsterdam, Amsterdam; and cTweeSteden Ziekenhuis, Tilburg, The

etherlands. Manuscript received August 16, 2010; revised manuscripteceived and accepted October 5, 2010.

*Corresponding author: Tel: 31-20-5993033; fax: 31-20-5994618.
aE-mail address: [email protected] (R.K. Riezebos).
002-9149/11/$ – see front matter © 2011 Elsevier Inc. All rights reserved.oi:10.1016/j.amjcard.2010.10.016

artery disease, electrocardiography, and biomarkers, to es-timate the likelihood of disease. In addition, echocardiog-raphy in the acute phase can be used to clarify the diagno-sis.1,3 However, the diagnosis sometimes remains uncertain.n these cases, the clinical probability of ACS should bessessed.4 Although American College of Cardiology

(ACC) and American Heart Association (AHA) as well asEuropean Society of Cardiology (ESC) guidelines do notprovide guidance on this topic, the next diagnostic test ofchoice should depend on the likelihood of disease. In Figure1, an algorithm is proposed in which the preferred perfor-mance of a diagnostic test is related to the estimated prob-ability of NSTEACS. In case of low clinical probability,patients are to be discharged safely, so a diagnostic testshould be used with high sensitivity and high negativepredictive value. Ischemia testing such as exercise testingwith or without an imaging modality is frequently used inthe subacute setting. However, such tests are most useful inpatients with intermediate probability of ACS. In our opin-ion, poorly performing tests, such as treadmill or bicycleexercise tests, should be restricted to prognostic purposesonly. Despite being not recommended by current ESCguidelines, computed tomographic angiography (CTA) isat present the most accurate noninvasive test to rule outcoronary artery disease.1,5 New sophisticated scan proto-ols, using prospective electrocardiographically gatedriggering, substantially reduce radiation exposure (effec-ive dose value approximately 3 mSv), without reducingmage quality.6 Extracardiac findings such as pulmonaryumors, pulmonary embolism, and aortic dissection can alsoe detected.7 In selected patients with acute chest pain, the

diagnostic accuracy of CTA is excellent.8 In addition, this
pproach is more cost effective and less time-consuming.9
www.ajconline.org


P

wwsNOd

pNTtmAbra

iDt


In case of a high probability of ACS, patients should beadmitted to the hospital for clinical follow-up and treatment.In these patients, false-positive results are more likely tooccur. Accordingly, the diagnosis of ACS should be waivedonly on the basis of tests with high sensitivity and specific-ity, invasive coronary angiography currently being the goldstandard. In this population, coronary angiography is able toexclude coronary artery disease reliably. This should bestrived for, because even in the presence of electrocardio-graphic changes and troponin increase, about a fifth of thepatients suspected of high-risk NSTEACS show no signif-icant lesions on coronary angiography.10,11 These patientsgenerally are at low risk and should be evaluated for alter-native pathologies.

Because of the absence of validated scoring systems toestimate the probability of NSTEACS in patients with chestpain, there is limited information on the distribution of theeventual diagnoses across the various levels of suspicion ofACS. A small trial by Goldstein et al9 evaluated the use ofCTA in about 200 patients with chest pain and low proba-bility of ACS. The mean Thrombolysis In Myocardial In-farction (TIMI) risk score was 1.2. The number of patientsdiagnosed with NSTEACS was about 10%, and the remain-der had noncardiac chest pain. The percentage of patientswho underwent percutaneous coronary intervention (PCI)was 4%, and the percentage requiring coronary artery by-pass grafting (CABG) was 2%.9 In the Optimal Timing of

Figure 1. Algorithm showing the clinical application of diagnostic testsaccording to the probability of NSTEACS. ECG � electrocardiography.

CI in Unstable Angina (OPTIMA) trial, about 250 patients

ith suspected intermediate- to high-risk NSTEACS under-ent acute coronary angiography.11 The mean TIMI risk

core was 3.8. Of these, 78% were diagnosed withSTEACS, and the remainder had noncardiac chest pain.f all patients, 55% were treated with PCI, and 10% un-erwent CABG.

The second issue to be addressed in patients with sus-ected NSTEACS involves risk assessment. Patients withSTEACS represent a prognostically heterogenous group.herefore, risk stratification plays a central role in evalua-

ion and management. For this purpose, multiple scoringodels have been developed, with the Global Registry ofcute Coronary Events (GRACE) and TIMI risk scoreseing the most widely used. The 2 models show a strongelation between indicators of the likelihood of NSTEACSnd prognosis.12,13 The GRACE risk tool was developed on

the basis of data from a large multinational cohort study(GRACE) and validated in subsequent GRACE and GlobalUse of Strategies to Open Occluded Coronary Arteries(GUSTO) IIb cohorts.12,14 Recently, the GRACE score wasprospectively revalidated in a large contemporary cohort.15

The TIMI score was developed using data from the TIMI11B trial13 and prospectively validated in several cohorts,ncluding that of the Treat Angina With Aggrastat andetermine Cost of Therapy With an Invasive or Conserva-

ive Strategy (TACTICS)–TIMI 18 trial.16 The GRACEscore estimates the risk for death up to 6 months, and theTIMI risk score addresses the 14-day risk for death, recur-rent myocardial infarction (MI) or urgent revascularization.This risk estimation, together with individual patient char-acteristics, should further guide treatment strategy.

Indications for Urgent Revascularization

A subset of patients with NSTEACS are considered tohave such an increased mortality risk that immediate revas-cularization is recommended.1,2 These include cardiogenicshock, severe left ventricular dysfunction, suspected leftmain stem disease, recurrent or refractory ischemia at restdespite intensive pharmacologic treatment, mechanicalcomplications such as acute mitral regurgitation, and sus-tained ventricular tachycardia. This recommendation isbased on a single study that suggested better outcomes withrevascularization in patients presenting with cardiogenicshock.17 However, most patients can be medically stabi-lized. These patients should be evaluated for an invasiveapproach.

Routine Invasive Versus Selective Invasive Therapy

In the past 2 decades, multiple trials have evaluateddifferent clinical strategies regarding coronary angiographyand subsequent revascularization of clinically stabilized pa-tients with NSTEACS. Two general approaches haveemerged, the first being the “early invasive” or “routineinvasive” strategy, involving routine early coronary angiog-raphy followed by revascularization when appropriate. Thesecond is the “conservative” or “selective invasive” ap-proach, with initial pharmacologic management and coro-nary angiography followed by revascularization for recur-
rent ischemia only. This new ischemia may either be

aTd2gM

mr

iit1rRttp

tiisaIIpioaa

sgd

le1

thod

olog

icdi

ffer

ence

sam

ong

tria

lsev

alua

ting

rout

ine

vers

usse

lect

ive

inva

sive

appr

oach

inno

nST

-ele

vatio

nac

ute

coro

nary

synd

rom

es

alT

ime

Fram

ePa

tient

Incl

usio

nT

imin

gof

Cat

hete

riza

tion

inth

eR

outin

eIn

vasi

veA

ppro

ach

RI

SIPr

imar

yT

rial

Out

com

e

%R

evas

cula

riza

tion

% PCI

%R

evas

cula

riza

tion

% PCI

MI

Mor

talit

y

RI

SIR

ISI

ISC

II18

1996

–199

8(6

-mon

thFU

)�

7da

ys77

%42

%37

%18

%94

/1,2

07*

(7.8

%)

124/

1,22

6(1

0.1%

)23

/1,2

07(1

.9%

)36

/1,2

26(2

.9%

)

CT

ICS–

IMI

816

1997

–199

9(6

-mon

thFU

)4–

48ho

urs

64%

42%

35%

29%

53/1

,114

†(4

.8%

)76

/1,1

06(6

.9%

)37

/114

(3.3

%)

39/1

,106

(3.5

%)

A31

919

97–2

001

(1-y

ear

FU)

�72

hour

s57

%36

%28

%16

%34

/895

(4%

)44

/915

(5%

)41

/895

(4.5

%)

36/9

15(4

%)

US2

020

01–2

003

(1-y

ear

FU)

24–4

8ho

urs

79%

61%

54%

40%

90/6

04‡

(15%

)59

/596

(10%

)15

/604

(2.5

%)

15/5

96(2

.5%

)

p�

0.03

;†

p�

0.00

2;‡

p�

0.00

5.U

�fo

llow

-up;

RI

�ro

utin

ein

vasi

ve;

SI�

sele

ctiv

ein

vasi

ve.

511Review/Invasive Strategies for NSTEACS

spontaneous or provoked by noninvasive stress testing.1,2

Currently, AHA and ACC as well as ESC guidelines sup-port routine invasive management in intermediate- to high-risk patients with NSTEACS.1,2

Four large randomized controlled trials have dominatedthe debate on the routine performance of invasive diagnos-tics in NSTEACS. The results, unfortunately, were quitediverse (Table 1). In 1999, the Fragmin and Fast Revascu-larisation During Instability in Coronary Artery Disease(FRISC) II trial showed a significant reduction in the com-bined end point of death and MI with the routine invasiveapproach.18 The observed difference was driven mainly byn excess in MI in the selective invasive group. The TACTICS–IMI 18 trial, published in 2001, showed similar results: aecrease in MI but no significant mortality benefit.16 In003, the Randomized Intervention Trial of Unstable An-ina (RITA) 3 trial failed to show any benefit for death orI.19 Ultimately, in 2005, the Invasive Versus Conservative

Treatment in Unstable Coronary Syndromes (ICTUS) trialwas published.20 This study, with optimal medical treat-

ent in both arms, showed an increased MI risk in theoutine invasive arm, with no difference in mortality.

Interpretation of the study results is difficult because ofmportant differences in method. Foremost, when the stud-es are compared, there appears to be a marked variation inhe intensity of revascularization between study arms (Table). The conservative arm of the ICTUS trial20 showed aevascularization rate similar to the routine invasive arm inITA 3.19 Also, the definition of MI differed between the

rials. The low biomarker threshold used in the ICTUSrial20 may partly explain the higher number of MIs inatients requiring PCI.

The improved use of anticoagulants, dual-antiplateletherapy, statins, and angiotensin-converting enzyme inhib-tors may also be part of the assumed demise of the routinenvasive treatment benefit. This is most clear for the use oftatins. In the FRISC II18 and TACTICS–TIMI 1816 trials,pproximately 1/2 the patients received statins at discharge.n RITA 3,19 this had already increased to 70%, whereas theCTUS trial20 provided high-dose statin treatment to 92% ofatients. Although less sharp, the use angiotensin-convert-ng enzyme inhibitors showed similar patterns. On the basisf the ICTUS trial, the current AHA and ACC guidelinescknowledge the option of a selective invasive strategy withggressive medical treatment.2

It may not be surprising that long-term (5-year) fol-low-up of the aforementioned trials showed discordant re-sults.21,22 Remarkably, the initial negative RITA 3 trialuggested a marked 5-year benefit in the routine invasiveroup regarding death and MI (odds ratio 0.78, 95% confi-ence interval 0.61 to 0.99, p � 0.04).23 Indeed, a recent

meta-analysis based on individual 5-year follow-up patientdata from FRISC II, RITA 3, and ICTUS showed a reduc-tion in MI using the routine invasive strategy.24

The relation between treatment effect and patient risk hasbeen evaluated in subanalyses of several trials. Regardlessof the risk score used, there appeared to be a consistenttreatment benefit for the invasive approach in high-riskpatients compared to low-risk patients. The FRISC II andTACTICS–TIMI 18 trials as well as the 5-year follow-up of
RITA 3 showed the greatest benefit of the routine invasiveT
abM

e

Tri

FR TA T 1

RIT

ICT * F

alaRtt

NottapI

attt

ttpile

2th

odol

ogic

diff

eren

ces

amon

gtr

ials

eval

uatin

gth

etim

ing

ofin

vasi

veap

proa

chin

non

ST-e

leva

tion

acut

eco

rona

rysy

ndro

mes

alT

ime

Fram

ePa

tient

Incl

usio

n

Ear

lyst

rate

gyD

elay

edst

rate

gy

Med

ian

Tim

eto

Cat

hete

riza

tion

(hou

rs)

% PCI

Med

ian

Tim

eto

PCI

(hou

rs)

Med

ian

Tim

eto

cath

eter

izat

ion

(hou

rs)

% PCI

Med

ian

Tim

eto

PCI

(hou

rs)

Tri

alO

utco

me

at1

Mon

th

MI

Mor

talit

y

Ear

lyD

elay

edE

arly

Del

ayed

RC

OO

L26

2000

–200

22.

464

%N

A86

70%

NA

12/2

03(5

.9%

)21

/207

(10.

1%)

3/20

7(1

.4%

)0/

203

(0.2

%)

AC

S27

2003

–200

814

60%

1650

55%

5257

/1,5

93(3

.6%

)59

/1,4

38(4

.1%

)46

/1,5

93(2

.9%

)47

/1,4

38(3

.3%

)T

IMA

11

2004

–200

72

100%

2.5

1.8

99%

2744

/73*

(60.

3%)

26/6

9(3

7.7%

)0/

73(0

%)

0/69

(0%

)O

AR

D28

2006

–200

81.

180

%N

A20

.570

%N

A16

/175

(9.1

%)

8/17

7(4

.5%

)5/

175

(2.9

%)

2/17

7(1

.1%

)

p�

0.00

5.A

�no

tav

aila

ble.


approach in high-risk patients.16,18,23 This resulted in a widecceptance of the routine invasive approach in this subpopu-ation. The clinical application of the aforementioned TIMInd GRACE risk scores has been evaluated extensively.emarkably, recent data from the GRACE registry suggest

he presence of an inverse relation between patient risk andhe rate of PCI.25 In daily practice, angiographic findings

and referral practice may more substantially influence thedecision to proceed to PCI than patients’ risk status.

In conclusion, the different outcomes in the large trialsevaluating the invasive approach in NSTEACS mainly re-flect the changes in study protocols and in pharmacologictreatment. For clinical practice, it seems reasonable to con-sider a liberal selective invasive approach equivalent to atemperate routine invasive approach. The patients with thehighest risks for adverse outcomes are thought to derive thegreatest benefit from invasive evaluation and revasculariza-tion. However, because clinical judgment on risk estimationappears to be challenging, the use of systematic and accu-rate risk stratification methods seems important.

Timing of Percutaneous Coronary Intervention

In the past few years, several studies have evaluated theinfluence of the timing of intervention in patients withNSTEACS. Once again, comparison of data and interpreta-tion of the results are difficult, mainly because of method-ologic differences among the studies (Table 2). CurrentAHA and ACC as well as ESC guidelines do not givespecific recommendations on this topic.1,2

The first study published evaluating the timing of theroutine invasive approach was the Intracoronary StentingWith Antithrombotic Regimen Cooling-Off (ISAR-COOL)trial.26 This trial randomized patients with suspected

STEACS to an early (�24 hours after anginal complaints)r a 3- to 5-day deferred invasive diagnostic strategy. Al-hough there was no difference between groups regardinghe individual end points, the combined end point of deathnd MI occurred significantly less in the early arm com-ared to the deferred strategy.26 Recently, the Timing ofntervention in Acute Coronary Syndrome (TIMACS)27 and

Angioplasty to Blunt the Rise of Troponin in Acute Coro-nary Syndromes Randomized for an Immediate or DelayedIntervention (ABOARD)28 trials provided important infor-mation on the feasibility of a very early invasive diagnosticroutine. TIMACS is clearly the largest study performed,with approximately 1,500 patients in its 2 arms. This study’sresults were negative with regard to its end points. How-ever, a subanalysis of a high-risk population, defined ashaving GRACE risk scores �140, suggested a benefit in theearly arm. The ABOARD trial28 evaluated a primary PCIpproach for NSTEACS compared to elective catheteriza-ion on the next day. The trial failed to show any benefit forhis approach. In addition, there appeared to be a trendoward more MIs in the early group.

The influence of timing of PCI remains difficult to de-ermine because the aforementioned trials randomized to theiming of coronary angiography, and only a portion ofatients were treated with PCI (Table 2). It is likely that thenfluence of timing of coronary angiography is less pro-

nounced in patients who are treated conservatively or whoTab

Me

Tri

ISA

TIM

OP

AB * N

aphwoMcmmbl

R

asgtN

aro

ctblehe

pNlltr


eventually undergo CABG. It is clear that a fast invasivediagnostic approach has diagnostic benefits and facilitatesthe logistics of further treatment planning. However, thequestion remains: should early angiography always be fol-lowed by prompt intervention? The proper answer can beobtained only from a randomized study in which patientsare randomized between immediate and delayed PCI, as wasdone in the OPTIMA trial.11 Although the trial was termi-nated early because of slow patient recruitment, it suggestedthe presence of an early hazard. After acute coronary an-giography in 251 patients admitted with NSTEACS, thistrial randomized 142 acute patients eligible for PCI to im-mediate (0.5 hours) or deferred (24 hours) PCI. Moreover,OPTIMA used only 1 infarct definition and included all MIsin its end point, including evolving MI at randomization.This was done because with very early PCI, periproceduralMI is hard to distinguish from a spontaneously evolving MIthat started before PCI. OPTIMA showed that MI wassignificantly more common in patients receiving immediatePCI (Table 2).11 This difference was most likely due to anexcess of periprocedural infarctions in the immediatelytreated group. This seems to contradict with a recentlypublished post hoc analysis of the Acute Catheterization andUrgent Intervention Triage Strategy (ACUITY) trial, whichsuggested a better outcome with urgent revascularization.29

Although this study included a large patient sample, the designof the ACUITY trial was not suited to detect the influenceof timing of PCI. These observational studies are extremelyliable to indication bias and should therefore be interpretedwith the utmost caution.

Are there any clues regarding the optimal timing ofintervention that can be distilled from the data provided bythe 4 trials on this topic? When it is suggested that theinfluence of timing of invasive therapy is the most pro-nounced just after an acute event, it is likely that the timingof initiation of therapy in the early invasive group will bethe most important variable. In this case, a time-event rela-tion can be estimated using the relative risk for MI at 30days for each trial and plotted against the time of admittanceto diagnostic catheterization, the latter being at least re-

Figure 2. The relation between timing of the early intervention and therelative risk (RR) for MI against a delayed strategy at 30-day follow-up.

motely related to timing of intervention. Figure 2 shows this

interpretation, which suggests a U-shaped curve for time-event relation.

Using potent antiplatelet and anticoagulation therapy, theearly hazard is not as pronounced as in the past.30 In thecute setting, PCI is still most likely to counteract plaqueassification by intracoronary manipulation, leading to aigher rate of periprocedural MI. It seems reasonable toant to treat patients with PCI after pharmacokinetic onsetf the initiated medication to reduce periprocedural inflictedI. Therefore, sufficient time is needed to allow pharma-

ologic stabilization, but the postponement of interventionay lead to an increase of new spontaneous events. Oneay expect that patients at high risk for recurrent events

enefit most from revascularization soon after pharmaco-ogic stabilization.

evascularization Methods

Although numerous clinical trials have compared PCInd CABG, few trials have compared PCI and CABG in aelected population of patients with NSTEACS.31 The An-ina With Extremely Serious Operative Mortality Evalua-ion (AWESOME) trial randomized patients with aSTEACS to PCI using bare-metal stents or CABG.32

Short- and long-term mortality rates were similar, but PCIwas associated with an increase in recurrent ischemia andrepeat revascularization. The current guidelines recommendCABG for patients with disease of the left main coronaryartery, multivessel disease, and impaired left ventricularfunction.1,2 Contemporary trials show nevertheless that PCIprovides an alternative in patients with fewer complex cor-onary artery disease.33 Although the use of modern stentsnd scoring systems aids in the feasibility of PCI in high-isk patient groups, it remains associated with a higher ratef repeat procedures.

Patients who present with ACS often show multipleoronary lesions, of which �1 is responsible for the symp-oms. These so-called culprit lesions can be identified eithery angiographic characteristics or by coronary territory. Theatter requires the determination of the localization of isch-mia. In patients with ST-elevation MIs, multivessel PCIas been associated with an increased rate of ischemicvents compared to PCI of the culprit lesion alone.34 In

contrast, in stable patients, no differences in events wereobserved.35 Although there is a lack of prospective data inatients with NSTEACS, a large registry of patients withSTEACS treated with PCI showed multivessel revascu-

arization to be equivalent compared with PCI of the culpritesion alone regarding death or MI. In this registry, mul-ivessel PCI was associated with a lower rate of repeatevascularization.36 In case of multivessel approach, frac-

tional flow reserve guidance should be considered whileselective intervention limited to flow obstructive lesionsresults in a decrease in adverse events.37

Future Directions

In the past 20 years, the rise of invasive coronary diag-nostics, interventions, and pharmacotherapies has revolu-tionized modern cardiology. Strategies based on different
pathophysiologic assumptions such as plaque sealing38 and

nrtcet

Wept


primary PCI11,28 have been considered. Undeniably, coro-ary revascularization has played a dynamic role. Futureesearch should focus on better identification of those pa-ients with high risk for recurrent unstable disease. Plaqueomposition and morphology using CTA or optical coher-nce tomography are being evaluated as promising newechniques.39–41 There is increasing evidence that the use of

CTA in patients with suspected NSTEACS can provideimportant information on the pathophysiology of the acuteevent by recognizing the vulnerable plaque40 (Figure 3).

hen an early invasive strategy is preferred, optical coher-nce tomography is able to identify underlying plaque mor-hology and detect thrombi of different stages of organiza-ion.41 How this new insight will influence clinical decision

making and whether this will alter the choice of therapy willbe the subject of debate in the coming years.

1. ESC guidelines on the diagnosis and treatment of non ST-segmentelevation acute coronary syndromes. Eur Heart J 2007;28:1598–1660.

2. ACC/AHA 2007 guidelines for the management of patients with un-stable angina/non ST-elevation myocardial infarction: a report of theAmerican College of Cardiology/American Heart Association TaskForce on Practice Guidelines. Circulation 2007;116:e148–e304.

3. ACC/AHA/ASE 2003 guideline update for the clinical application ofechocardiography: summary article: a report of the American Collegeof Cardiology/American Heart Association Task Force on PracticeGuidelines (ACC/AHA/ASE Committee to Update the 1997 Guide-lines for the Clinical Application of Echocardiography). Circulation2003;108:1146–1162.

4. Scirica BM. Acute coronary syndrome: emerging tools for diagnosisand risk assessment. J Am Coll Cardiol 2010;55:1403–1415.

5. Schuetz GM, Zacharopoulou NM, Schlattmann P, Dewey M. Meta-analysis: Noninvasive coronary angiography using computed tomog-raphy versus magnetic resonance imaging. Ann Intern Med 2010;152:167–177.

6. Consensus document on coronary computed tomographic angiographyACCF/ACR/AHA/NASCI/SAIP/SCAI/SCCT 2010 expert consensus.

Figure 3. The use of CTA in the initial evaluation of NSTEACS. CTA wasperformed in a 48-year-old man presenting with chest pain. There was anintermediate probability of ACS. Multiplanar reconstruction of the leftanterior descending coronary artery (LAD) showed a moderately severemixed stenosis in the proximal LAD with evidence of superimposedthrombus (white arrow).

J Am Coll Cardiol 2010;55:2663–2699.

7. Johnson KM. Extracardiac findings on cardiac computed tomography.J Am Coll Cardiol 2010;55:1566–1568.

8. Meijboom WB, van Mieghem CA, Mollet NR, Pugliese F, WeustinkAC, van Pelt N, Cademartiri F, Nieman K, Boersma E, de Jaegere P,Krestin GP, de Feyter PJ. 64-slice computed tomography coronaryangiography in patients with non ST-elevation acute coronary syn-drome. Heart 2007;93:1386e92.

9. Goldstein JA, Gallagher MJ, O’Neill WW, Ross MA, O’Neil BJ, RaffGL. A randomized controlled trial of multi-slice coronary computedtomography for evaluation of acute chest pain. J Am Coll Cardiol2007;49:863–871.

10. Roe MT, Harrington RA, Prosper DM, Pieper KS, Bhatt DL, LincoffAM, Simoons ML, Akkerhuis M, Ohman EM, Kitt MM, Vahanian A,Ruzyllo W, Karsch K, Califf RM, Topol EJ; The Platelet GlycoproteinIIb/IIIa in Unstable Angina: Receptor Suppression Using IntegrilinTherapy (PURSUIT) Trial Investigators. Clinical and therapeutic pro-file of patients presenting with acute coronary syndromes who do nothave significant coronary artery disease. Circulation 2000;102:1101–1106.

11. Riezebos RK, Ronner E, Ter Bals E, Slagboom T, Smits PC, ten BergJM, Kiemeneij F, Amoroso G, Patterson MS, Suttorp MJ, Tijssen JG,Laarman GJ; OPTIMA trial. Immediate versus deferred coronary an-gioplasty in non-ST-elevation acute coronary syndromes. Heart 2009;95:807–812.

12. Fox KA, Dabbous OH, Goldberg RJ, Pieper KS, Eagle KA, Van deWerf F, Avezum A, Goodman SG, Flather MD, Anderson FA Jr,Granger CB. Prediction of risk of death and myocardial infarction inthe six months after presentation with acute coronary syndrome: pro-spective multinational observational study (GRACE). BMJ 2006;333:1091.

13. Antman EM, Cohen M, Bernink PJ, McCabe CH, Horacek T, PapuchisG, Mautner B, Corbalan R, Radley D, Braunwald E. The TIMI riskscore for unstable angina/non-ST elevation MI: a method for prognos-tication and therapeutic decision making. JAMA 2000;284:835–842.

14. Granger CB, Goldberg RJ, Dabbous O, Pieper KS, Eagle KA, CannonCP, Van De Werf F, Avezum A, Goodman SG, Flather MD, Fox KA;Global Registry of Acute Coronary Events Investigators. Predictors ofhospital mortality in the Global Registry of Acute Coronary Events.Arch Intern Med 2003;163:2345–2353.

15. Pieper KS, Gore JM, FitzGerald G, Granger CB, Goldberg RJ, Steg G,Eagle KA, Anderson FA, Budaj A, Fox KA; Global Registry of AcuteCoronary Events (GRACE) Investigators. Validity of a risk-predictiontool for hospital mortality: the Global Registry of Acute CoronaryEvents. Am Heart J 2009;157:1097–1105.

16. Cannon CP, Weintraub WS, Demopoulos LA, Robertson DH, Gorm-ley GJ, Braunwald E. Comparison of early invasive and conservativestrategies in patients with unstable coronary syndromes treated withthe glycoprotein IIb/IIIa inhibitor tirofiban. N Engl J Med 2001;344:1879–1887.

17. Hochman JS, Boland J, Sleeper LA, Porway M, Brinker J, Col J,Jacobs A, Slater J, Miller D, Wasserman H. Current spectrum ofcardiogenic shock and effect of early revascularisation on mortality:results of an international registry. Circulation 1995;91:873–881.

18. Fragmin and Fast Revascularisation During Instability in CoronaryArtery Disease Investigators. Invasive compared with non-invasivetreatment in unstable coronary-artery disease: FRISC II prospectiverandomised multicentre study. Lancet 1999;354;708–715.

19. Fox KA, Poole-Wilson PA, Henderson RA, Clayton TC, ChamberlainDA, Shaw TR, Wheatley DJ, Pocock SJ; Randomized InterventionTrial of Unstable Angina Investigators. Interventional versus conser-vative treatment for patients with unstable angina or non-ST-elevationmyocardial infarction: the British Heart Foundation RITA 3 random-ised trial. Lancet 2002;360:743–751.

20. de Winter RJ, Windhausen F, Cornel JH, Dunselman PH, Janus CL,Bendermacher PE, Michels HR, Sanders GT, Tijssen JG, VerheugtFW; Invasive Versus Conservative Treatment in Unstable CoronarySyndromes (ICTUS) Investigators. Early invasive versus selectivelyinvasive management for acute coronary syndromes. N Engl J Med2005;353:1095–1104.

21. Hirsch A, Windhausen F, Tijssen JG, Verheugt FW, Cornel JH, deWinter RJ; Invasive Versus Conservative Treatment in Unstable Cor-onary Syndromes (ICTUS) Investigators. Long-term outcome after an
early invasive versus selective invasive treatment strategy in patients with


non-ST-elevation acute coronary syndrome and elevated cardiac troponinT (the ICTUS trial): a follow-up study. Lancet 2007;369:827–835.

22. Damman P, Hirsch A, Windhausen F, Tijssen JG, de Winter RJ;ICTUS Investigators. A randomised comparison of an early invasiveversus selective invasive management in patients with non-ST-eleva-tion acute coronary syndrome. J Am Coll Cardiol 2010;55:858–864.

23. Fox KAA, Poole-Wilson P, Clayton TC. 5-year outcome of an inter-ventional strategy in non-ST elevation acute coronary syndrome: theBritish Heart Foundation RITA 3 randomised trial. Lancet 2005;366:914–920.

24. Fox KA, Clayton TC, Damman P, Pocock SJ, de Winter RJ, TijssenJG, Lagerqvist B, Wallentin L; FIR Collaboration. Long-term outcomeof a routine versus selective invasive strategy in patients with non-ST-segment elevation acute coronary syndrome a meta-analysis ofindividual patient data. J Am Coll Cardiol 2010;55:2435–2445.

25. Fox KA, Anderson FA Jr, Dabbous OH, Steg PG, López-Sendón J,Van de Werf F, Budaj A, Gurfinkel EP, Goodman SG, Brieger D;GRACE Investigators. Intervention in acute coronary syndromes: dopatients undergo intervention on the basis of their risk characteristics?The Global Registry of Acute Coronary Events (GRACE). Heart2007;93:177–182.

26. Neumann FJ, Kastrati A, Pogatsa-Murray G, Mehilli J, Bollwein H,Bestehorn HP, Schmitt C, Seyfarth M, Dirschinger J, Schömig A.Evaluation of prolonged antithrombotic pretreatment (“cooling-off”strategy) before intervention in patients with unstable coronary syn-dromes: a randomized controlled trial. JAMA 2003;290:1593–1599.

27. Mehta SR, Granger CB, Boden WE, Steg PG, Bassand JP, Faxon DP,Afzal R, Chrolavicius S, Jolly SS, Widimsky P, Avezum A, RupprechtHJ, Zhu J, Col J, Natarajan MK, Horsman C, Fox KA, Yusuf S;TIMACS Investigators. Early versus delayed intervention in acutecoronary syndromes. N Engl J Med 2009;360:2165–2175.

28. Montalescot G, Cayla G, Collet JP, Elhadad S, Beygui F, Le Breton H,Choussat R, Leclercq F, Silvain J, Duclos F, Aout M, Dubois-RandéJL, Barthélémy O, Ducrocq G, Bellemain-Appaix A, Payot L, StegPG, Henry P, Spaulding C, Vicaut E; ABOARD Investigators. Imme-diate vs delayed intervention for acute coronary syndromes: a random-ised clinical trial. JAMA 2009;302:947–954.

29. Sorajja P, Gersh BJ, Cox DA, McLaughlin MG, Zimetbaum P, Costan-tini C, Stuckey T, Tcheng JE, Mehran R, Lansky AJ, Grines CL, StoneGW. Impact of delay to angioplasty in patients with acute coronarysyndromes undergoing invasive management. J Am Coll Cardiol 2010;55:1416–1424.

30. de Feyter PJ, Suryapranata H, Serruys PW, Beatt K, van Domburg R,van den Brand M, Tijssen JJ, Azar AJ, Hugenholtz PG. Coronaryangioplasty for unstable angina: immediate and late results in 200consecutive patients with identification of risk factors for unfavorableearly and late outcome. J Am Coll Cardiol 1988;12:324–333.

31. Bravata DM, Gienger AL, McDonald KM, Sundaram V, Perez MV,
Varghese R, Kapoor JR, Ardehali R, Owens DK, Hlatky MA. Sys-tematic review: the comparative effectiveness of percutaneous coro-
nary interventions and coronary artery bypass graft surgery. Ann InternMed 2007;147:703–716.

32. Morrison DA, Sethi G, Sacks J, Henderson W, Grover F, Sedlis S,Esposito R, Ramanathan K, Weiman D, Sucedo J, Antakli T, ParameshV, Pett S, Vernon S, Birjiniuk V, Welt F, Krucoff M, Wolfe W, LuckeJ, Mediratta S, Booth D, Barbiere C, Lewis D. Percutaneous coronaryintervention versus coronary artery bypass graft surgery for patientswith medically refractory myocardial ischaemia and risk factors foradverse outcomes with bypass: a multicenter, randomised trial. Inves-tigators of the Department of Veterans Affairs Cooperative Study#385, the Angina With Extremely Serious Operative Mortality Eval-uation (AWESOME). J Am Coll Cardiol 2001;38:143–149.

33. Serruys PW, Morice MC, Kappetein AP, Colombo A, Holmes DR,Mack MJ, Ståhle E, Feldman TE, van den Brand M, Bass EJ, VanDyck N, Leadley K, Dawkins KD, Mohr FW. Percutaneous coronaryintervention versus coronary-artery bypass grafting for severe coro-nary artery disease. N Engl J Med 2009;360:961–972.

34. Corpus RA, House JA, Marso SP, Grantham JA, Huber KC Jr, LasterSB, Johnson WL, Daniels WC, Barth CW, Giorgi LV, Rutherford BD.Multivessel percutaneous intervention in patients with multivesseldisease and acute myocardial infarction. Am Heart J 2004;148:493–500.

35. Ijsselmuiden AJ, Ezechiels J, Westendorp IC, Tijssen JG, Kiemeneij F,Slagboom T, van der Wieken R, Tangelder G, Serruys PW, LaarmanG. Complete versus culprit vessel percutaneous coronary interventionin multivessel disease: a randomised comparison. Am Heart J 2004;148:467–474.

36. Shishehbor MH, Lauer MS, Singh IM, Chew DP, Karha J, Brener SJ,Moliterno DJ, Ellis SG, Topol EJ, Bhatt DL. In unstable angina ornon-ST-segment acute coronary syndrome, should patients with mul-tivessel coronary artery disease undergo multivessel or culprit-onlystenting? J Am Coll Cardiol 2007;49:849–854.

37. Tonino PA, De Bruyne B, Pijls NH, Siebert U, Ikeno F, van’ t Veer M,Klauss V, Manoharan G, Engstrøm T, Oldroyd KG, Ver Lee PN,MacCarthy PA, Fearon WF; FAME Study Investigators. Fractionalflow reserve versus angiography for guiding percutaneous coronaryintervention. N Engl J Med 2009;360:213–224.

38. Meier B. Plaque sealing by coronary angioplasty. Heart 2004;90:1395–1398.

39. van Velzen JE, Schuijf JD, de Graaf FR, Nucifora G, Pundziute G,Jukema JW, Schalij MJ, Kroft LJ, de Roos A, Reiber JH, van der WallEE, Bax JJ. Plaque type and composition as evaluated non-invasivelyby MSCT angiography and invasively by VH IVUS in relation to thedegree of stenosis. Heart 2009;95:1990–1996.

40. Russo V, Zavalloni A, Bacchi Reggiani ML, Buttazzi K, Gostoli V,Bartolini S, Fattori R. Incremental prognostic value of coronary CTangiography in patients with suspected coronary artery disease. CircCardiovasc Imaging 2010;3:351–359.

41. Toutouzas K, Karanasos A, Stefanadis C. Multiple plaque morpholo-
gies assessed by optical coherence tomography in a patient with acutecoronary syndrome. Heart 2010;96:1335–1336.

ep

Ps

0d

Long-Term Follow-Up of Patients With First-Time Chest PainHaving 64-Slice Computed Tomography

Fabiola B. Sozzi, MD, PhDa,b,*, Filippo Civaia, MDa, Philippe Rossi, MDa,Jean-Francois Robillon, MDa, Stephane Rusek, MSca, Frederic Berthier, MSca,

Francois Bourlon, MDa, Laura Iacuzio, MDa, Gilles Dreyfus, MDa, and Vincent Dor, MDa

A paucity of data on outcome of coronary multislice computed tomography (CT) isavailable. The aim of this study was to assess the long-term follow-up of 64-slice CT in ahomogenous patient group. In total 222 patients (136 men, 61%, 59 � 11 years of age) withchest pain at intermediate risk of coronary artery disease (CAD) and no previous CADunderwent 64-slice CT. Coronary lesions were considered significant or not based on athreshold of 50% luminal narrowing. Plaques were classified as calcified, noncalcified, andmixed based on type. End point during follow-up was major adverse cardiac events(nonfatal myocardial infarction, unstable angina requiring hospitalization, myocardialrevascularization). Coronary plaques were detected in 162 patients (73%). Coronary arterystenosis was significant in 62 patients. Normal arteries were found in 59 patients (27%).During a mean follow-up of 5 � 0.5 years, 30 cardiac events occurred. Annualized eventrates were 0% in patients with normal coronary arteries, 1.2% in patients with nonsignif-icant stenosis, and 4.2% in patients with significant stenosis (p <0.01). Predictors of cardiacevents were presence of significant stenosis, proximal stenosis, and multivessel disease.Noncalcified and mixed plaques had the worse prognosis (p <0.05). In conclusion, 64-CTprovides long-term incremental value in patients at intermediate risk of CAD. © 2011
Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:516–521)
pmnbaarfipdmn(hupttt

al

Coronary multislice computed tomography (CT) is increas-ingly being used as a tool for noninvasive visualization ofcoronary arteries.1 The technique provides information on ath-rosclerotic plaque burden and to some extent on plaque com-osition.2–4 Accuracy of coronary CT needs to be assessed in

management outcome studies in which diagnostic and thera-peutic strategies would be decided based on CT alone, withoutreference to any coronary angiographic results. Therefore, thepurpose of the present study was to explore extent, degree, andstructure/function of coronary atherosclerosis by 64-CT in ahomogenous patient population with chest pain, intermediateprobability of coronary artery disease (CAD), and no previouscardiac events and to analyze the impact of these variables onlong-term follow-up.

Methods

The study group was composed of 222 consecutive pa-tients (136 men, 61%, mean age 59 � 11 years) whounderwent 64-CT at the Cardiothoracic Centre of Monaco,Monte Carlo from January to October 2005. All patientsincluded in the study presented with chest pain suspiciousfor angina. A subgroup of patients had previous equivocalstress test results. Eligibility criteria for this study weresuspected but no previously known CAD and intermediatepretest likelihood of CAD (score 9 to 15 points) according

aMonaco Cardiothoracic Centre, Monte Carlo, Monaco; bIRCCS Fondoliclinico, Milan, Italy. Manuscript received July 26, 2010; revised manu-cript received and accepted October 5, 2010.

tE-mail address: [email protected] (F. Sozzi).

to a modification of a method by Diamond and Forrester5 asublished by Morise et al.6,7 Exclusion criteria for recruit-ent were previous documented CAD, history of percuta-

eous transluminal coronary angioplasty or coronary arteryypass grafting, nonchest pain indication for CT, and lownd high pretest likelihoods of CAD. A structured interviewnd clinical history were obtained, and the following cardiacisk factors were assessed before CT: (1) hypertension (de-ned as blood pressure �140/90 mm Hg or use of antihy-ertensive agents), (2) hyperlipidemia as defined by low-ensity lipoprotein cholesterol �140 mg/dl, (3) diabetesellitus (defined as fasting glucose level �120 mg/dl or

eed for insulin or oral antidiabetic medicines), (4) smokingcurrent or previous habit), (5) body mass index, (6) familyistory of CAD in first-degree relatives, and (7) medicationse. All patients gave written informed consent to the studyrotocol, which was approved by the local ethics commit-ee. All scans were performed with a 64-slice computedomographic scanner that features a temporal resolutionime of 165 ms and a spatial resolution of 0.4 mm3 (Siemens

Somatom Sensation 64 Cardiac, Siemens, Forchheim, Ger-many). If heart rate was �65 beats/min, additional � block-ers (metoprolol 5 mg intravenously to a maximum dose of10 mg) were provided. Nitroglycerin (0.4 mg sublingually)was used in all studies unless contraindicated. Patients withknown allergy to iodine, significant arrhythmia (atrial fibril-lation and frequent premature beats) or rapid heart rate (90beats/min), impaired renal function (serum creatinine �1.3mg/dl), and contraindications to � blockade were not im-ged. The following parameters were applied for CT: col-imation of 64 � 0.6 mm, tube rotation time of 330 ms, and
ube current of 450 mA at 120 kV. Nonionic contrast ma-
www.ajconline.org


llo

A

CCP

P

PPN

TTCMN

517Coronary Artery Disease/Risk Stratification With 64-Computed Tomography

terial was administered in the antecubital vein at 80 to 105ml, depending on total scan time, and a flow rate of 5 ml/s(Iodixanol 320, GE Healthcare SA, Velizy-Villacoublay,France and Iomeron 400, Bracco Imaging, Courcouronnes,Evry cedex, France) followed by a saline flush of 50 ml ata flow rate of 5 ml/s. The bolus-tracking technique was usedto synchronize the arrival of contrast in coronary arteriesand initiation of the scan. Automated detection of peakincrease in the aortic root was used for timing of the scan.All images were acquired during an inspiratory breath-holdof approximately 10 seconds, with simultaneous registrationof a patient’s electrocardiogram. To evaluate presence ofcoronary artery plaques, reconstructions in diastole (typi-cally 75% of cardiac cycle) were generated with a slicethickness of 0.75 mm at an increment of 0.6 mm. If motionartifacts were present, additional reconstructions were madeat different time points of the RR interval. Axial datasetswere transferred to a remote workstation (Syngo, Siemens,Berlin, Germany) for postprocessing and subsequent eval-uation. Radiation dose-decreasing techniques, i.e., dosemodulation along the z-axis and pulsing algorithm along theRR interval, were employed for all scans. All studies wereanalyzed by 2 experienced readers blinded to all clinicalvariables, history, and patient demographics. Coronary ar-teries were divided into 17 segments according to the mod-ified American Heart Association classification.8 Only seg-ments with a diameter �1.5 mm were included. Eachsegment was classified as interpretable or not. Patients wereexcluded from analysis for (1) an uninterpretable proximalor midsegment or (2) �3 uninterpretable segments in gen-eral. Then, interpretable segments were evaluated for pres-ence of any atherosclerotic plaque using axial images andmultiplanar reconstructions. After visual inspection of vol-ume-rendered images, which depicted the gross coronaryartery luminal configuration, coronary artery plaques werecarefully inspected on axial images, curved multiplanarreformatted images, and cross-sectional multiplanar refor-matted images. Coronary plaques were defined as structures�1 mm2 within and/or adjacent to the coronary arteryumen, which could be clearly distinguished from the vesselumen and the surrounding pericardial tissue, epicardial fat,r the vessel lumen itself.9 One coronary plaque was as-

signed per coronary segment. An assessment of plaquecomposition was also allowed. Differentiation was madeamong noncalcified plaques (composed exclusively of ma-terial having density �130 HU), calcified plaques (com-

Table 1Patient characteristics (n � 222 patients)

Variable Total Normal Arteri(n � 59)

ge (years) 59.2 � 10.8 53.1 � 11.2Men 136 (61%) 33 (55%)Hypertension 91 (41%) 16 (27%)Diabetes mellitus 38 (17%) 6 (10%)Hyperlipidemia* 78 (35%) 18 (30%)Smoker 49 (22%) 6 (10%)Body mass index �30 kg/m2 51 (23%) 9 (15%)

* Low-density lipoprotein cholesterol �140 mg/dl.

posed exclusively of high-density material �130 HU), and

mixed plaques (having components of noncalcified and cal-cified material). Atherosclerotic lesion was deemed signif-icant if diameter stenosis was �50%. Lesions below thisthreshold were considered nonsignificant or mild. For eachpatient, number of diseased coronary segments, number ofsegments with significant stenosis, and number of each typeof plaque was calculated. Computed tomograms withoutcoronary lesions were considered normal; computed tomo-grams showing coronary wall irregularities or �1 coronaryplaque (significant or not) were defined as abnormal. Sig-nificant coronary stenosis were further classified as local-ized in 1 epicardial artery or 2 or 3 epicardial arteries (leftanterior descending, left circumflex, right coronary arteries)and significant plaques in the left main and/or proximal leftanterior descending coronary artery. Follow-up informationwas obtained by clinical visits, telephone contact, or question-naires sent by mail. All reported events were verified by hos-pital records or direct contacts with the attending physician.The following clinical events were recorded: (1) cardiac death(including death without definitive cause), (2) nonfatal acutemyocardial infarction (AMI), (3) unstable angina pectoris re-quiring hospitalization, and (4) coronary revascularization (bycoronary angioplasty or bypass). Coronary revascularizationoccurring soon after CT was performed as a consequence of itsresult. Therefore, patients undergoing coronary revasculariza-

Nonsignificant Stenosis Significant Stenosis p Value(n � 101) (n � 62)

60.6 � 9.8 62.7 � 9.7 �0.00162 (61%) 41 (67%) 0.3942 (42%) 33 (53%) 0.01619 (19%) 13 (21%) 0.2736 (36%) 24 (39%) 0.5827 (27%) 16 (26%) 0.05625 (25%) 17 (27%) 0.32

Table 2Computed tomographic findings of study population (n � 222)

Normal coronary circulation 59 (27%)oronary stenosis �50% 101 (45%)oronary stenosis �50% 62 (28%)laques in left main/proximal left anterior

descending coronary artery67 (9%)

laques in all left anterior descendingcoronary artery

280 (39%)

laques in left circumflex coronary artery 172 (24%)laques in right coronary artery 199 (28%)umber of coronary arteries narrowed �50%1 43 (19.4%)2 21 (9.5%)3 21 (9.5%)otal number of segments with plaques 718 (19.2%)otal plaque score 3.2alcified plaque (significant or not) 374 (52%)ixed plaque (significant or not) 273 (33%)oncalcified plaque (significant or not) 109 (15%)

es

tion procedures sooner than 90 days after CT were excluded

pAy

CmceSdwCopctrr

R

ctp3o

tsstvtt3ppbsso

iac even


from further analysis. The end point of this study was majoradverse cardiac events including nonfatal AMI, unstableangina pectoris requiring hospitalization, and revascular-ization. Definitions of AMI and unstable angina werepreviously described.10,11 For multiple events in a given

atient, the first was the event included in the analysis.ll patients were followed for a mean period of 5.0 � 0.5ears (maximum follow-up period 5.5 years).

Analyses were performed with SAS 9.1.3 (SAS Institute,ary, North Carolina). All continuous data are presented asean � SD, and all categorical data are reported as per-

entage or absolute number. A p value �0.05 was consid-red statistically significant. In univariate analysis, unpairedtudent’s t tests or chi-square tests were used to assessifferences between groups. Kaplan-Meier survival curvesere constructed for computed tomographically diagnosedAD and were compared with log-rank test. Effect of CADn hard cardiac events at CT was determined using Coxroportional hazard model. After adjustment for all baselinelinical characteristics, a forward stepwise model was usedo determine independent predictors of coronary atheroscle-otic variables on computed tomogram (p �0.05). Hazardatios and confidence intervals were calculated.

esults

Of the 246 patients included in the study, 13 were ex-luded from analysis because of poor image quality relatedo cardiac motion artifact and respiratory motion artifact; 11atients declined to participate in the follow-up study and0 patients were lost during 5-year follow-up. Replies were

Figure 1. Kaplan-Meier curves for survival free of hard card

btained from 192 patients (follow-up rate 87%). All pa-

ients presented with chest pain suspicious for angina. Amaller subset of patients (83, 37%) initially underwenttress testing with equivocal findings and continued symp-oms that warranted further evaluation. A complete over-iew of baseline characteristics of the entire study popula-ion is presented in Table 1. Based on coronary computedomographic results, the study population was divided into

subgroups: patients with normal arteries (59, 27%),atients with nonsignificant stenosis (101, 45%), andatients with significant stenosis (62, 28%). A correlationetween clinical risk factors and occurrence and grade oftenosis is presented in Table 1. As presented, significanttenosis occurred more frequently in patients who werelder, had hypertension, and a smoking habit (p �0.05).

Computed tomographic characteristics are listed in Table2. After exclusion of 45 inaccessible segments (1.2%)because of motion artifacts, plaque burden was evaluatedin 3,729 segments. According to plaque texture, 28 (45%)were from patients with obstructive calcified plaques, 16(26%) were from those with obstructive mixed plaques,and 14 (23%) were from those with obstructive noncal-cified plaques. A combination of different significantplaques (calcified and/or noncalcified and/or mixedplaques) was found in 4 patients (6%). Patients withsignificant calcified plaques were on average significantlyolder then patients with significant noncalcified andmixed plaques (63 � 8.2 vs 61 � 7.1, p �0.01) and morefrequently affected by hypertension (18 vs 10, p �0.05).

Prognostic analysis was performed in 192 patients. Inthese patients, 30 major cardiac events occurred (total event

ts in patients with normal arteries and any type of stenosis.

rate 15.6%). Nonfatal AMI affected 2 patients (1%) and

A

CI � confidence interval; HR � hazard ratio.


unstable angina 14 patients (7.3%); coronary revasculariza-tion was performed in 14 patients (7.3%, 11 with angio-plasty and 3 with bypass). No cases of cardiac deaths wererecorded. Annual cardiac event rates were 4.2% in patientswith significant stenosis, 1.2% in patients with nonsignifi-cant stenosis, and 0% in patients with normal arteries on

with nonsignificant stenosis and significant stenosis in 1 vessel and 2 and

Table 4Predictors of events by multivariate Cox proportional hazards regressionanalysis*

Coronary Risk Factor HR 95% CI p Value

Obstructive plaque 8.28 2.48–27.64 0.0006Obstructive plaque in left anterior

descending coronary artery5.98 1.29–27.61 0.022

Obstructive plaque in circumflexcoronary artery

4.84 1.50–15.63 0.008

Obstructive plaque in right coronaryartery

1.66 0.51–5.36 0.40

Proximal obstructive plaque 3.73 0.96–14.49 0.05Obstructive plaque (per segment) 1.69 1.13–2.54 0.0111 segment with significant stenosis 3.29 0.16–69.71 0.442 segments with significant stenosis 17.44 1.17–261.1 0.038�2 segments with significant stenosis 14.13 1.23–162.9 0.034Mixed plaques 6.04 1.32–18.7 0.01Noncalcified plaque 5.1 1.05–16.8 0.049

Abbreviations as in Table 3.* Adjusted for age, gender, hypertension, diabetes mellitus, hyperlipid-

emia, smoking, family history of coronary disease, and obesity.

Figure 2. Kaplan-Meier curves for survival free of hard cardiac events in patients3 vessels.

Table 3Predictors of cardiac events by univariate Cox proportional hazardsregression analysis

Variables HR 95% CI p Value

ge (per year) 1.028 0.98–1.17 0.05Men 1.36 0.52–3.53 0.53Hypertension 1.87 0.73–4.73 0.19Diabetes mellitus 1.38 0.45–4.18 0.58Hyperlipidemia 1.16 0.45–3.00 0.76Smoker 1.81 0.68–4.81 0.24Obesity �30 kg/m2 1.77 0.65–4.78 0.26Proximal versus distal plaques 3.5 0.92–13.2 0.005Obstructive plaques in left

anterior descending artery5.74 1.67–19.72 0.006

Obstructive plaques in leftcircumflex artery

3.52 1.43–8.66 0.006

Obstructive plaques in rightcoronary artery

1.73 0.68–4.40 0.25

Obstructive plaques (persegment)

1.46 1.10–1.94 0.009

1 segments with significantstenosis

3.72 0.34–41.07 0.28

2 segments with significantstenosis

8.43 0.94–75.42 0.057

�2 segments with significantstenosis

10.47 1.26–87.02 0.03

Noncalcified plaque 4.185 1.2–13.9 0.04Calcified plaque 2.699 0.8–9.06 0.10Mixed plaques 5.04 1.62–15.7 0.005

64-slice computed tomogram (Figures 1 and 2). Patients

scospsApo“

onwmatapgP

ac even


with cardiac events were significantly older. Univariateanalysis is presented in Table 3. Predictors of all cardiacevents are presented in Table 4. Any obstructive plaque,multivessel distribution, proximal stenosis, and noncalcifiedand mixed plaques had the worse prognosis (Figures 1 to 3).

Discussion

To our knowledge, this is the longest follow-up study inthe largest homogenous cohort of patients at intermediaterisk with no previous CAD who underwent CT to investi-gate chest pain. Patients with cardiac events had more ex-tensive atherosclerosis on computed tomogram as reflectedby a large number of segments showing significant plaques.Cardiac event rate of patients with normal arteries on 64-slice computed tomogram was 0%, highlighting an excel-lent negative predictive value of normal arteries on 64-slicecomputed tomogram on long-term analysis. These patientsmay indeed be reassured without need for further testing.Higher event risk was associated with proximal diseaseseverity, particularly within the proximal portion of the leftanterior descending and left circumflex coronary arteries.Risk of events was considerably higher in patients withsignificant stenosis, although patients with nonsignificantstenosis still showed higher risk compared to patients with-out CAD on 64-slice computed tomogram. Figure 2 showsthat after 3-year follow-up patients with nonsignificant ste-nosis had a similar cardiac event rate compared to patientswith significant 1-vessel stenosis (overlapping of the 2 sur-vival curves after 3-year follow-up). This preliminary ob-servation needs further investigation with larger population

Figure 3. Kaplan-Meier curves for survival free of hard cardi

studies. A main finding of our study is that plaque compo-

sition represents a long-term predictor of cardiac events. Asshown in Figure 3, noncalcified and mixed plaques carrieda worse prognosis compared to calcified plaques. Referringto the plaque vulnerability concept, Mann et al12 studied 31ubjects who died suddenly of CAD. They found that lipidore size and minimal cup thickness, 2 major determinantsf plaque vulnerability, were not related to absolute plaqueize or degree of stenosis. Accordingly, atheroscleroticlaque growth and destabilization are highly variable. Manyerial angiographic studies have demonstrated that mostMIs occur from occlusion of coronary arteries that did notreviously contain significant stenosis; furthermore, the cor-nary artery with the most severe stenosis is usually not theculprit” artery.13,14 Thus, plaque progression and clinical

outcome are not always closely related, and each is poorlypredicted on clinical and angiographic grounds becausemost plaques that underlie an AMI are �70% stenosed.15 Inur study patients with any type of plaque (significant andonsignificant) had a worse prognosis compared to patientsith normal vessels (p �0.05). In addition, noncalcified andixed plaques on 64-slice computed tomogram represented

n independent predictor of cardiac events. Of interest,hese 2 types of plaque composition may represent lessdvanced and possibly less stabilized atherosclerosis com-ared to dense calcified lesions. However, further investi-ations are clearly needed to support these observations.undziute et al16 in 2007 published the first follow-up study

of CT in 100 patients with suspected or known CAD, 55 ofwhom were studied with 16-slice CT. They demonstrated asignificant prognostic value for cardiac events and a very

ts in patients with noncalcified, calcified, and mixed plaques.

good prognosis for patients without obstructive CAD. Car-

eaaifihdCtTp


rigan et al17 analyzed a group of 227 patients with variousstimated pretest probabilities of CAD (low, intermediate,nd high). Absence of obstructive CAD was associated with99% freedom from cardiac death, AMI, and revascular-

zation during an average of 2.3 years of follow-up. Similarndings were reported by Hadamitzky et al18 and van Werk-oven et al19 in follow-up studies of 18 months and 621ays, respectively. To date, the prognostic performance ofT has mostly been tested in symptomatic mixed popula-

ions of different estimated pretest probabilities of CAD.he prognostic value of CT is strongly dependent on theretest risk profile. Meijboom et al20 showed that the clin-

ical value of CT is higher in patients with an intermediatepretest probability of CAD. In this study we selected thiscategory of patients.

The present study is based on an analysis of a relativelysmall population. In general, CT was associated with in-creased radiation exposure, although radiation doses arerapidly decreasing with newer acquisition protocols. Also,no dedicated algorithms that allow quantification of plaquestenosis or volume are available for CT. Diabetes, hyper-lipidemia, and hypertension were simply categorized aspresent or not. Classifications could be based on diseaseseverity in future studies. Incidence of incomplete fol-low-up was rather high. The number of events in this pop-ulation was quite small compared to the number of variablesused for adjustment in the multivariable model. In addition,soft end points were emerging compared to the others. Thiscan be explained by the fact that the population was selectedfor not having had any previous CAD; in additiona, theeffect of medical therapy can change the prognosis.

1. Kuettner A, Kopp AF, Schroeder S, Rieger T, Brunn J, Meisner C,Heuschmid M, Trabold T, Burgstahler C, Martensen J, Schoebel W,Selbmann HK, Claussen CD. Diagnostic accuracy of multidetectorcomputed tomography coronary angiography in patients with angio-graphically proven coronary artery disease. J Am Coll Cardiol 2004;43:831–839.

2. Achenbach S, Moselewski F, Ropers D, Ferencik M, Hoffmann U,MacNeill B, Pohle K, Baum U, Anders K, Jang IK, Daniel WG, BradyTJ. Detection of calcified and non-calcified coronary atheroscleroticplaque by contrast-enhanced, submillimeter multidetector spiral com-puted tomography: a segment-based comparison with intravascularultrasound. Circulation 2004;109:14–17.

3. Leber AW, Becker A, Knez A, von Ziegler F, Sirol M, Nikolaou K,Ohnesorge B, Fayad ZA, Becker CR, Reiser M, Steinbeck G, Boek-stegers P. Accuracy of 64-slice computed tomography to classify andquantify plaque volumes in the proximal coronary system: a compar-ative study using intravascular ultrasound. J Am Coll Cardiol 2006;47:672–677.

4. Leber AW, Knez A, Becker A, Becker C, von Ziegler F, Nikolaou K,Rist C, Reiser M, White C, Steinbeck G, Boekstegers P. Accuracy ofmultidetector spiral computed tomography in identifying and differ-entiating the composition of coronary atherosclerotic plaques: a com-
parative study with intracoronary ultrasound. J Am Coll Cardiol 2004;43:1241–1247.
5. Diamond GA, Forrester JS. Analysis of probability as an aid in theclinical diagnosis of coronary artery disease. N Engl J Med 1979;300:1350–1358.

6. Morise AP, Haddad WJ, Beckner D. Development and validation of aclinical score to estimate the probability of coronary artery disease inmen and women presenting with suspected coronary disease. Am JMed 1997;102:350–356.

7. Morise AP, Jalisi F. Evaluation of pretest and exercise test scores toassess all-cause mortality in unselected patients presenting for exercisetesting with symptoms of suspected coronary artery disease. J Am CollCardiol 2003;42:842–850.

8. Austen WG, Edwards JE, Frye RL, Gensini GG, Gott VL, Griffith LS,McGoon DC, Murphy ML, Roe BB. A reporting system on patientsevaluated for coronary artery disease. Report of the Ad Hoc Commit-tee for Grading of Coronary Artery Disease, Council on Cardiovascu-lar Surgery, American Heart Association. Circulation 1975;51:5–40.

9. Min JK, Shaw LJ, Devereux RB, Okin PM, Weinsaft JW, Russo DJ,Lippolis NJ, Berman DS, Callister TQ. Prognostic value of multide-tector coronary computed tomographic angiography for prediction ofall-cause mortality. J Am Coll Cardiol 2007;50:1161–1170.

10. Myocardial infarction redefined—a consensus document of the JointEuropean Society of Cardiology/American College of Cardiologycommittee for the redefinition of myocardial infarction. Eur Heart J2000;21:1502–1513.

11. Braunwald F. Unstable angina. A classification. Circulation 1989;80:410–414.

12. Mann JM, Davies MJ. Vulnerable plaque. Relation of characteristics todegree of stenosis in human coronary arteries. Circulation 1996;94:928–931.

13. Ambrose JA, Tannenbaum MA, Alexopoulos D, Hjemdahl-MonsenCE, Leavy J, Weiss M, Borrico S, Gorlin R, Fuster V. Angiographicprogression of coronary artery disease and the development of myo-cardial infarction. J Am Coll Cardiol 1988;12:56–62.

14. Little WC, Constantinescu M, Applegate RJ, Kutcher MA, BurrowsMT, Kahl FR, Santamore WP. Can coronary angiography predict thesite of a subsequent myocardial infarction in patients with mild-to-moderate coronary artery disease? Circulation 1988;78:1157–1566.

15. Kolodgie FD, Burke AP, Farb A, Gold HK, Yuan J, Narula J, FinnAV, Virmani R. The thin-cap fibroatheroma: a type of vulnerableplaque: the major precursor lesion to acute coronary syndromes. CurrOpin Cardiol 2001;16:285–292.

16. Pundziute G, Schuijf JD, Jukema JW, Boersma E, de Roos A, van derWall EE, Bax JJ. Prognostic value of multislice computed tomographycoronary angiography in patients with known or suspected coronaryartery disease. J Am Coll Cardiol 2007;49:62–70.

17. Carrigan TP, Nair D, Schoenhagen P, Curtin RJ, Popovic ZB, Halli-burton S, Kuzmiak S, White RD, Flamm SD, Desai MY. Prognosticutility of 64-slice computed tomography in patients with suspected butno documented coronary artery disease. Eur Heart J 2009;30:362–371.

18. Hadamitzky M, Freissmuth B, Meyer T, Hein F, Kastrati A, MartinoffS, Schömig A, Hausleiter J. Prognostic value of coronary computedtomographic angiography for prediction of cardiac events in patientswith suspected coronary artery disease. JACC Cardiovasc Imaging2009;2:404–411.

19. van Werkhoven JM, Gaemperli O, Schuijf JD, Jukema JW, Kroft LJ,Leschka S, Alkadhi H, Valenta I, Pundziute G, de Roos A, van derWall EE, Kaufmann PA, Bax JJ. Multislice computed tomographycoronary angiography for risk stratification in patients with an inter-mediate pretest likelihood. Heart 2009;95:1607–1611.

20. Meijboom WB, van Mieghem CA, Mollet NR, Pugliese F, WeustinkAC, van Pelt N, Cademartiri F, Nieman K, Boersma E, de Jaegere P,Krestin GP, de Feyter PJ. 64-slice computed tomography coronaryangiography in patients with high, intermediate, or low pretest prob-
ability of significant coronary artery disease. J Am Coll Cardiol 2007;50:1469–1475.

pc5

AH

0d

Usefulness of Cooling and Coronary Catheterization to ImproveSurvival in Out-of-Hospital Cardiac Arrest

Dion Stub, MBBSa,b,*, Christopher Hengel, MBBSa, William Chan, MBBSa,b,Damon Jackson, MBBSa, Karen Sanders, RN, GradDipEda,

Anthony M. Dart, BA, BM, BCh, DPhila,b, Andrew Hilton, MBBSc, Vincent Pellegrino, MBBSc,James A. Shaw, MBBS, PhDa,b, Stephen J. Duffy, MBBS, PhDa, Stephen Bernard, MBBS, MDc, and

David M. Kaye, MBBS, PhDa,b

Survival rates after out-of-hospital cardiac arrest (OHCA) continue to be poor. Recentevidence suggests that a more aggressive approach to postresuscitation care, in particularcombining therapeutic hypothermia with early coronary intervention, can improve prog-nosis. We performed a single-center review of 125 patients who were resuscitated fromOHCA in 2 distinct treatment periods, from 2002 to 2003 (control group) and from 2007 to2009 (contemporary group). Patients in the contemporary group had a higher prevalenceof cardiovascular risk factors but similar cardiac arrest duration and prehospital treatment(adrenaline administration and direct cardioversion). Rates of cardiogenic shock (48% vs41%, p � 0.2) and decreased conscious state on arrival (77% vs 86%, p � 0.2) were similarin the 2 cohorts, as was the incidence of ST-elevation myocardial infarction (33% vs 43%,p � 0.1). The contemporary cohort was more likely to receive therapeutic hypothermia(75% vs 0%, p <0.01), coronary angiography (77% vs 45%, p <0.01), and percutaneouscoronary intervention (38% vs 23%, p � 0.03). This contemporary therapeutic strategy wasassociated with better survival to discharge (64% vs 39%, p <0.01) and improved neuro-logic recovery (57% vs 29%, p <0.01) and was the only independent predictor of survival(odds ratio 5.5, 95% confidence interval 1.2 to 26.2, p � 0.03). Longer resuscitation time,presence of cardiogenic shock, and decreased conscious state were independent predictorsof poor outcomes. In conclusion, modern management of OHCA, including therapeutichypothermia and early coronary angiography is associated with significant improvement insurvival to hospital discharge and neurologic recovery. © 2011 Elsevier Inc. All rights
fthf

ahouHmgwrAosodRtet

In a previous study conducted in Australia before the wide-spread adoption of postresuscitation strategies such as cooling,survival after admission to hospital for out-of-hospital cardiacarrest (OHCA) was 25%, comparable to most registry data.1

Given the rapid uptake of such approaches since that time, wehypothesized that advances in basic life support and postresus-citation hospital care have improved outcomes. Accordingly,we performed a single-center retrospective review of all pa-tients with OHCA admitted to our hospital from 2002 to 2003and from 2007 to 2009.

Methods

Melbourne has approximately 3.9 million inhabitants,which is served by a comprehensive centrally co-ordinated

aAlfred Hospital Heart Centre, Melbourne, Victoria, Australia; bBakerIDI Heart Diabetes Institute, Melbourne, Victoria, Australia; cAlfred Hos-ital Intensive Care Unit, Melbourne, Victoria, Australia. Manuscript re-eived August 25, 2010; revised manuscript received and accepted October, 2010.

Dr. Stub is supported by a scholarship from the Cardiac Society ofustralia & New Zealand, Sydney, NSW and an award from Baker IDIeart and Diabetes Institute, Melbourne, Victoria, Australia.

oE-mail address: [email protected] (D. Stub).

ambulance system, which is described elsewhere.1 The Al-red Hospital (Melbourne, Victoria, Australia) is a largeertiary-/quaternary-care referral center that provides 24-our emergency coronary and cardiac surgical interventionsor patients with acute coronary syndromes.

In this retrospective analysis we evaluated clinical char-cteristics and outcomes of all patients who had an out-of-ospital ventricular fibrillation arrest with sustained returnf spontaneous circulation (ROSC), defined as �20 min-tes, and who were subsequently hospitalized at the Alfredospital. Analysis was performed for 2 treatment periods: aodern treatment paradigm, 2007 to 2009 (contemporary

roup), and a historical control group, 2002 to 2003. Dataere obtained from Ambulance Victoria and hospital

ecords. The study was performed in accordance with thelfred Hospital ethics committee guidelines. Interrogationf the hospital database identified 326 patients with pre-umed OHCA. Seventy-seven patients were excluded sec-ndary to noncardiac causes such as trauma, stroke, andrug overdose. Excluded were 4 patients who did not haveOSC on arrival, 11 patients transferred from other insti-

utions, and 109 patients because of asystole or pulselesslectrical activity as their initial rhythm. Our study popula-ion, therefore, consisted of 125 patients with OHCA sec-
ndary to ventricular arrhythmia.
www.ajconline.org


fCt

Om

spiFstsyns

R

srnicmbdvr

523Coronary Artery Disease/Cooling and Catheterization in Cardiac Arrest

Ambulance Victoria uses a 2-tier system of ambulanceparamedics, most of whom have advanced life supportskills, and intensive care paramedics who are authorized toperform endotracheal intubation and administer a range ofcardiac drugs. Melbourne also uses a medical emergencyresponse program in which ambulance and fire brigadeservices respond to cardiac arrests.2 Cardiac arrest protocolsollow the recommendations of the Australian Resuscitationouncil.3 After hemodynamic stabilization, patients are

ransported urgently to the nearest hospital.In the 2 treatment periods, hospital care for patients with

HCA was modeled on relevant International Liaison Com-ittee on Resuscitation guidelines at the time.4,5 Decision

regarding need for cardiac catheterization was made by thetreating cardiologist. Intensive care treatment including tar-get hemodynamic and metabolic parameters and choice ofinotropic agents were decided by the treating physicianaccording to general critical care guidelines. Therapeutichypothermia was induced and maintained through a com-bination of ice-cold intravenous fluids, simple ice packs,and surface cooling blankets. The 2 significant changes topostresuscitative care in patients with OHCA during the

Table 1Baseline characteristics

Characteristic Control Contemporary p Value(n � 44) (n � 81)

Age (years) 64 � 17 61 � 16 NSMen 34 (77%) 68 (84%) NSCurrent/ex-smoker 10 (23%) 45 (56%) �0.01Diabetes mellitus 3 (7%) 9 (11%) NSHypertension 13 (30%) 47 (58%) �0.01Hyperlipidemia 5 (11%) 37 (46%) �0.01History of coronary disease 14 (32%) 28 (35%) NSInitial rhythm

Ventricular fibrillation 40 (91%) 77 (95%) NSVentricular tachycardia 4 (9%) 4 (5%) NS

Basic life supportWitnessed arrest 41 (93%) 75 (93%) NSBystander resuscitation 33 (75%) 57 (70%) NS

Ambulance responseCall to arrival (minutes) 6 (5–9) 7 (6–10) NSTotal time until return of

circulation (minutes)26 (15–35) 23 (14–30) NS

Number of shocks 3.9 � 3.5 3.8 � 4.3 NSAdrenaline administration 35 (80%) 58 (72%) NS

CauseAcute coronary syndrome 30 (68%) 50 (62%) NSST-segment elevation

myocardial infarct19 (43%) 27 (33%) NS

Condition on arrival to hospitalUnconscious 38 (86%) 62 (77%) NSCardiogenic shock 18 (41%) 39 (48%) NS

InterventionsTherapeutic hypothermia 0 (0%) 61 (75%) �0.01Coronary angiography 20 (45%) 62 (77%) �0.01Emergent angiography 11 (25%) 49 (61%) �0.01Percutaneous coronary

intervention10 (23%) 31 (38%) 0.03

Coronary bypass graft surgery 0 (0%) 4 (5%) NS

Values are presented as mean � SD, number of patients (percentage), ormedian (interquartile range).

study period were use of mild therapeutic hypothermia for 7

unconscious patients (to preserve neurologic function) andincreasing use of emergency coronary angiography to assessand treat underlying coronary artery disease as the causefor OHCA.

The primary outcome was survival to hospital discharge.Secondary outcome was “good” neurologic recovery, de-fined as cerebral performance categories (CPCs) 1 and 2.The CPC is a simple-to-use widely used cerebral perfor-mance measurement.6

Statistical analyses were performed with SPSS 16(SPSS, Inc., Chicago, Illinois). Numerical normally dis-tributed data were analyzed using Student’s t test (pre-ented as mean � SD) and non-normal data were com-ared by Mann-Whitney test (presented as median withnterquartile range). Proportions were analyzed withisher’s exact test. A p value �0.05 was regarded astatistically significant. Prognostic factors that were foundo be significant (p �0.10) in preliminary univariate analy-es were entered into a multivariate logistic regression anal-sis. All variables were entered into the equation simulta-eously to control for effects of confounding (a subsequenttepwise analysis provided similar results).

esults

Baseline characteristics of the study population are pre-ented in Table 1. Important prehospital factors includingates of witnessed cardiac arrest, bystander cardiopulmo-ary resuscitation, time until ROSC, and adrenaline admin-stration by paramedics were similar in the control andontemporary cohorts. On arrival to the emergency depart-ent the incidence of cardiogenic shock, defined as systolic

lood pressure �90 mm Hg or requiring inotropic support,id not differ significantly between treatment periods (41%s 48%, p � NS) and rates of decreased conscious stateequiring intubation did not differ significantly (86% vs

Table 2Outcome of coronary angiography in study population

Coronary Angiographic Variable Control Contemporary p Value(n � 20) (n � 62)

Normal 6 (32%) 13 (21%) 0.36Single-vessel disease 12 (63%) 17 (27%) 0.02Multivessel disease 2 (11%) 31 (50%) 0.01Coronary arteries �50%

stenosis, mean � SD1.5 � 0.8 1.9 � 0.8 NS

Infarct-related arteryLeft anterior descending

coronary artery10 (53%) 17 (27%) 0.02

Left circumflex coronary artery 0 (0%) 6 (12%) NSRight coronary artery 1 (5%) 14 (23%) NSGrafts 0 (0%) 1 (2%) NSLeft main coronary artery 1 (5%) 3 (2%) NSMultivessel with no clear

culprit2 (10%) 8 (16%) NS

Preintervention ThrombolysisIn Myocardial Infarctiongrade 0–2 flow

8 (42%) 23 (47%) NS

Thrombus-containing lesion 4 (21%) 11 (22%) NS

7%, p � NS).

dos(cw

c

DMPIG

A


In the contemporary treatment group 75% of all patientsreceived therapeutic hypothermia, representing 98% of co-matose patients after ventricular fibrillation (Table 1). Nopatients in the control group received therapeutic hypother-mia. Rates of coronary angiography and percutaneous cor-onary intervention (PCI) were significantly increased in thecontemporary treatment group (77% vs 45%, p �0.01; 38%vs 23%, p � 0.03, respectively; Table 1). Of patients un-ergoing coronary angiography, 50% of patients did not gon to PCI for various clinical reasons (Table 2). PCI wasuccessful in �90% of patients in the 2 treatment groupssee Table 3). Incidence of the left anterior descendingoronary artery (LAD) as the culprit infarct-related arteryas higher in the control group (53% vs 27%, p � 0.02).

Figure 1. Outcomes based on contemporary (black bars) and historicalontrol (gray bars) treatment periods (*p �0.01).

Table 3Outcome of percutaneous coronary intervention in study population

Variable HistoricalControl PCI

(n � 10)

Modern PCI(n � 31)

pValue

Procedural success 9 (90%) 29 (94%) NSFinal Thrombolysis In

Myocardial Infarctiongrade 3 flow

8 (80%) 28 (90%) NS

Door-to-balloon time(minutes)

145 (112 to 345) 120 (105 to 167) NS

Stents per patient 1.2 � 0.4 1.3 � 1.1 0.03Mean stent length (mm) 17.9 � 5 19.8 � 10 NSMean stent diameter (mm) 3.1 � 0.8 3.7 � 3 NS

rug-eluting stents 0 (0%) 7 (23%) NSultivessel intervention 0 (0%) 6 (19%) NS

eak troponin mean (range) 92 (0–186) 53 (0–179) NSntra-aortic balloon pump 4 (40%) 8 (30%) NSlycoprotein IIb/IIIainhibitor

4 (40%) 23 (56%) NS

spiration catheter 0 5 (16%) NS

Values are presented as mean � SD, number of patients (percentage), ormedian (interquartile range).

Multivessel disease (defined as multiple coronary lesions

with �50% stenosis) was more prevalent in the contempo-rary treatment group (50% vs 11%, p �0.01).

Survival to hospital discharge in the contemporary treat-ment group was 64% compared to 39% in the historicalcontrol (p �0.01). Discharge with favorable neurologicoutcome (CPC 1 or 2) was also significantly improved (57%vs 30%, p � 0.01; Figure 1). Of survivors in the contem-porary treatment group, 89% made a good neurologic re-covery. Cause of death was similar in the 2 periods with70% of patients dying due to poor neurologic outcome, 25%due to persistent cardiac dysfunction, and 5% due to mul-tiorgan failure. Unadjusted predictors associated with sur-vival are presented in Table 4. Survivors were significantlymore likely to be managed by the contemporary treatmentparadigm and undergo coronary angiography and suc-cessful PCI. In unconscious patients, there was a signif-icant increase in survival (61% vs 37%, p � 0.03) andgood neurologic outcome (54% vs 27%, p � 0.01) inthose patients receiving therapeutic hypothermia. Whenadjusting for key prehospital and postresuscitative factors(Figure 2, Table 5), negative predictors of survival in-cluded cardiogenic shock, resuscitation times �20 min-utes, and decreased conscious state. The contemporarytreatment regimen was a significant independent predic-tor of survival (odds ratio 5.5, 95% confidence interval1.2 to 26.2, p � 0.03).

Discussion

This study has demonstrated that a contemporary treat-ment paradigm with focused co-ordinated postresuscitativecare combining therapeutic hypothermia with coronary an-giography is associated with significant improvements inshort-term clinical outcomes. The 64% survival to dischargerate is significantly better than other registry data of patientswith OHCA and favorably compares to other institutionswith similar treatment protocols.7,8

In the contemporary treatment group, 98% of comatosepatients received therapeutic hypothermia. Of these pa-

Table 4Significant univariate predictors of survival

Characteristic OddsRatio

95%Confidence

Interval

pValue

Age 0.24 0.11–0.5 0.01Basic life support

Bystander resuscitation 3.3 1.5–7.5 0.01Ambulance response

Return of circulation �20 minutes 0.08 0.03–0.21 0.01Number of shocks 0.83 0.73–0.94 0.01

Condition on arrivalUnconscious 0.29 0.10–0.84 0.02Cardiogenic shock 0.11 0.05–0.26 0.01

InterventionsCooling* 2.7 1.1–6.4 0.02Coronary angiography 7.6 3.2–17.5 0.01Successful coronary intervention 2.1 0.95–4.4 0.07Contemporary management 2.9 1.3–6.1 0.01

* For unconscious patients only.

tients, 60% survived, and of the survivors, 89% made a

oo

CCCRU


favorable neurologic recovery. This observation is impor-tant because uncertainty about neurologic recovery of pa-tients with OHCA has previously cast doubt about themerits (and futility) of early invasive strategies such ascoronary intervention. With the introduction of therapeutichypothermia and the recognized difficulty in predicting neu-rologic outcomes when patients first arrive, there is a smallrole for early neurologic prognostication as a basis forfurther treatment decisions.9,10 Therapeutic hypothermiawas a significant unadjusted predictor of survival despite alarge proportion of patients with cardiogenic shock whohave previously been excluded from randomized studies.Use of mild therapeutic hypothermia is supported by 2 largerandomized controlled trials11,12 and is a recommended partf a standardized treatment strategy for comatose survivorsf cardiac arrest.13

ST-segment elevation on initial electrocardiogram wassimilar in the 2 groups in approximately 1/3 of patients.There have been several observational studies and system-atic reviews highlighting the importance of emergency PCIin patients with ST-segment elevation myocardial infarctionand OHCA.14–16 Recent reports have illustrated the furtherbenefits of combining therapeutic hypothermia with early

Figure 2. Independent predictors of surviv

Table 5Multivariate predictors of survival

Characteristic Odd Ratio 95%Confidence

Interval

p Value

ontemporary management 5.5 1.2–26.2 0.03oronary angiography 4.3 0.97–19 0.06ardiogenic shock 0.12 0.02–0.54 0.01eturn of circulation �20 minutes 0.12 0.03–0.55 0.01nconscious on arrival 0.13 0.01–0.9 0.05

coronary intervention.17–19 This has led to the recommen-

dation by the American Heart Association that all patientswith ST-segment elevation myocardial infarction andOHCA be managed at centers capable of 24-hour coronaryintervention.20

Absence of ST-segment elevation in the setting of car-diac arrest has been shown to occur in up to 40% of OHCAscaused by unstable coronary plaques and coronary throm-bosis.21,22 Likewise, in our study 42% of patients under-going emergency PCI did not have ST-segment elevationon electrocardiogram. This has led to the increasingadoption of emergency coronary angiography for all pa-tients with OHCA of suspected cardiac origin and todeveloping appropriate systems of care to cater to suchtreatment protocols.23–26

The caveat to this approach is the significant number ofpatients with OHCA who undergo coronary angiographyand do not go on to emergency revascularization. In ourstudy 50% of patients undergoing angiography did not goon to PCI and 23% had angiographically normal coronaryarteries. As part of an early cardiac catheterization protocolthey received antiplatelet and antithrombotic agents, whichhave been hypothesized to have their own positive effects inthe setting of cardiac arrest associated with coagulationdisruption.27,28 It was interesting to note that significantlymore patients from the control group had the LAD as theinfarct-related artery compared to the contemporary cohort(53% vs 27%, p � 0.02). This is possibly explained by therelatively small numbers of patients in the control groupundergoing coronary angiography (n � 20). Most LADinfarcts are generally larger and associated with greaterhemodynamic disturbance than non-LAD infarcts and morethan likely influenced the decision to proceed to coronaryangiography in the control group.

There was a trend to shorter resuscitation times in the

� confidence interval; OR � odds ratio.
al. CI
contemporary treatment group, which may have contributed

1

1

1

1

2


to improved outcomes. However, this did not reach statis-tical significance (p � 0.09). During the study period sev-eral key changes to prehospital care included adopting achest compression-to-ventilation ratio of 30:2, in line withInternational Liaison Committee on Resuscitation guide-lines, and focusing on uninterrupted chest compressions.4

As in other recent studies on optimizing prehospital care, itis reasonable to assume that these measures also contributedto improved patient survival.29,30

There are several limitations to our study. This is asingle-center retrospective review and thus subject to po-tential confounders and selection bias contributing to re-sults. The patients comprise a selected group who achievedROSC before transport to hospital. Hospital and intensivecare paramedics have extensive experience in managingpatients with OHCA, and therefore the result’s applicabilityto other health care networks is uncertain. In the assessmentof neurologic recovery, the CPC score was chosen becauseof its ease of use and wide reporting in the literature.Although simple to use, this scoring system has not beenwell validated and was retrospectively assigned based onpatient follow-up and clinical notes. With regard to thera-peutic hypothermia, time spent at target temperature rangewas not recorded, making the quality of hypothermia diffi-cult to ascertain. The study, however, does indicate that ina contemporary treatment era, an aggressive approach topatients with OHCA of suspected cardiac origin is associ-ated with significantly improved survival to hospital dis-charge and neurologic recovery. Further study and random-ized trials with particular focus on establishing systems ofprehospital care and role of early coronary intervention arerequired.

1. Jennings PA, Cameron P, Walker T, Bernard S, Smith K. Out-of-hospital cardiac arrest in Victoria: rural and urban outcomes. Med JAust 2006;185:135–139.

2. Smith KL, McNeil JJ. Cardiac arrests treated by ambulance paramed-ics and fire fighters. Med J Aust 2002;177:305–309.

3. Morley PT, Walker T. Australian Resuscitation Council: adult ad-vanced life support (ALS) guidelines 2006. Crit Care Resuscitation2006;8:129–131.

4. Timerman S, Gonzalez MM, Mesquita ET, Marques FR, Ramires JA,Quilici AP, Timerman A. The International Liaison Committee onResuscitation (ILCOR). Roll in guidelines 2005–2010 for cardiopul-monary resuscitation and emergency cardiovascular care. Arq BrasCardiol 2006;87(suppl):e201–e208.

5. American Heart Association guidelines for cardiopulmonary resus-citation and emergency cardiovascular care. Circulation 2005;112(suppl):IV1–IV203.

6. Jacobs I, Nadkarni V, Bahr J, Berg RA, Billi JE, Bossaert L, CassanP, Coovadia A, D’Este K, Finn J, Halperin H, Handley A, Herlitz J,Hickey R, Idris A, Kloeck W, Larkin GL, Mancini ME, Mason P,Mears G, Monsieurs K, Montgomery W, Morley P, Nichol G, NolanJ, Okada K, Perlman J, Shuster M, Steen PA, Sterz F, Tibballs J,Timerman S, Truitt T, Zideman D. Cardiac arrest and cardiopulmonaryresuscitation outcome reports: update and simplification of the Utsteintemplates for resuscitation registries. A statement for healthcare pro-fessionals from a task force of the international liaison committee onresuscitation (American Heart Association, European ResuscitationCouncil, Australian Resuscitation Council, New Zealand ResuscitationCouncil, Heart and Stroke Foundation of Canada, InterAmerican HeartFoundation, Resuscitation Council of Southern Africa). Resuscitation2004;63:233–249.

7. Sunde K, Pytte M, Jacobsen D, Mangschau A, Jensen LP, Smedsrud C,
Draegni T, Steen PA. Implementation of a standardised treatment
protocol for post resuscitation care after out-of-hospital cardiac arrest.Resuscitation 2007;73:29–39.

8. Werling M, Thoren AB, Axelsson C, Herlitz J. Treatment and outcomein post-resuscitation care after out-of-hospital cardiac arrest when amodern therapeutic approach was introduced. Resuscitation 2007;73:40–45.

9. Young GB. Clinical practice. Neurologic prognosis after cardiac ar-rest. N Engl J Med 2009;361:605–611.

0. Wijdicks EF, Hijdra A, Young GB, Bassetti CL, Wiebe S. Practiceparameter: prediction of outcome in comatose survivors after cardio-pulmonary resuscitation (an evidence-based review): report of theQuality Standards Subcommittee of the American Academy of Neu-rology. Neurology 2006;67:203–210.

11. Bernard SA, Gray TW, Buist MD, Jones BM, Silvester W, GutteridgeG, Smith K. Treatment of comatose survivors of out-of-hospital car-diac arrest with induced hypothermia. N Engl J Med 2002;346:557–563.

12. The Hypothermia after Cardiac Arrest Study Group. Mild therapeutichypothermia to improve the neurologic outcome after cardiac arrest.N Engl J Med 2002;346:549–556.

13. Neumar RW, Nolan JP, Adrie C, Aibiki M, Berg RA, Bottiger BW,Callaway C, Clark RS, Geocadin RG, Jauch EC, Kern KB, Laurent I,Longstreth WT Jr, Merchant RM, Morley P, Morrison LJ, Nadkarni V,Peberdy MA, Rivers EP, Rodriguez-Nunez A, Sellke FW, SpauldingC, Sunde K, Vanden Hoek T. Post-cardiac arrest syndrome: epidemi-ology, pathophysiology, treatment, and prognostication. A consensusstatement from the International Liaison Committee on Resuscitation(American Heart Association, Australian and New Zealand Council onResuscitation, European Resuscitation Council, Heart and StrokeFoundation of Canada, InterAmerican Heart Foundation, ResuscitationCouncil of Asia, and the Resuscitation Council of Southern Africa);the American Heart Association Emergency Cardiovascular CareCommittee; the Council on Cardiovascular Surgery and Anesthesia;the Council on Cardiopulmonary, Perioperative, and Critical Care; theCouncil on Clinical Cardiology; and the Stroke Council. Circulation2008;118:2452–2483.

14. Noc M, Radsel P. Urgent invasive coronary strategy in patients withsudden cardiac arrest. Curr Opin Crit Care 2008;14:287–291.

15. Garot P, Lefevre T, Eltchaninoff H, Morice MC, Tamion F, Abry B,Lesault PF, Le Tarnec JY, Pouges C, Margenet A, Monchi M, LaurentI, Dumas P, Garot J, Louvard Y. Six-month outcome of emergencypercutaneous coronary intervention in resuscitated patients after car-diac arrest complicating ST-elevation myocardial infarction. Circula-tion 2007;115:1354–1362.

16. Hosmane VR, Mustafa NG, Reddy VK, Reese CL, DiSabatino A,Kolm P, Hopkins JT, Weintraub WS, Rahman E. Survival and neuro-logic recovery in patients with ST-segment elevation myocardial in-farction resuscitated from cardiac arrest. J Am Coll Cardiol 2009;53:409–415.

7. Hovdenes J, Laake JH, Aaberge L, Haugaa H, Bugge JF. Therapeutichypothermia after out-of-hospital cardiac arrest: experiences with pa-tients treated with percutaneous coronary intervention and cardiogenicshock. Acta Anaesthesiol Scand 2007;51:137–142.

8. Knafelj R, Radsel P, Ploj T, Noc M. Primary percutaneous coronaryintervention and mild induced hypothermia in comatose survivors ofventricular fibrillation with ST-elevation acute myocardial infarction.Resuscitation 2007;74:227–234.

9. Kern KB, Rahman O. Emergent percutaneous coronary interventionfor resuscitated victims of out-of-hospital cardiac arrest. CatheterCardiovasc Interv 2010;75:616–624.

0. Nichol G, Aufderheide TP, Eigel B, Neumar RW, Lurie KG, BufalinoVJ, Callaway CW, Menon V, Bass RR, Abella BS, Sayre M, Dough-erty CM, Racht EM, Kleinman ME, O’Connor RE, Reilly JP, Oss-mann EW, Peterson E. Regional systems of care for out-of-hospitalcardiac arrest: A policy statement from the American Heart Associa-tion. Circulation 2010;121:709–729.

21. Spaulding CM, Joly LM, Rosenberg A, Monchi M, Weber SN, Dhain-aut JF, Carli P. Immediate coronary angiography in survivors ofout-of-hospital cardiac arrest. N Engl J Med 1997;336:1629–1633.

22. Dumas F, Cariou A, Manzo-Silberman S, Grimaldi D, Vivien B,Rosencher J, Empana JP, Carli P, Mira JP, Jouven X, Spaulding C.Immediate percutaneous coronary intervention is associated with bet-
ter survival after out-of-hospital cardiac arrest: insights from the


PROCAT (Parisian Region Out of Hospital Cardiac Arrest) Registry.Circ Cardiovasc Interv 2010;3:200–207.

23. Peberdy MA, Ornato JP. Post-resuscitation care: is it the missing linkin the chain of survival? Resuscitation 2005;64:135–137.

24. Ewy GA, Kern KB. Recent advances in cardiopulmonary resuscita-tion: cardiocerebral resuscitation. J Am Coll Cardiol 2009;53:149–157.

25. Merchant RM, Abella BS, Khan M, Huang KN, Beiser DG, NeumarRW, Carr BG, Becker LB, Vanden Hoek TL. Cardiac catheterizationis underutilized after in-hospital cardiac arrest. Resuscitation 2008;79:398–403.

26. Callaway CW, Schmicker R, Kampmeyer M, Powell J, Rea TD, DayaMR, Aufderheide TP, Davis DP, Rittenberger JC, Idris AH, Nichol G.
Receiving hospital characteristics associated with survival after out-of-hospital cardiac arrest. Resuscitation 2010;81:524–529.
27. Bottiger BW, Motsch J, Bohrer H, Boker T, Aulmann M, Nawroth PP,Martin E. Activation of blood coagulation after cardiac arrest is notbalanced adequately by activation of endogenous fibrinolysis. Circu-lation 1995;92:2572–2578.

28. Schneider A, Bottiger BW, Popp E. Cerebral resuscitation after card-iocirculatory arrest. Anesth Analg 2009;108:971–979.

29. Garza AG, Gratton MC, Salomone JA, Lindholm D, McElroy J,Archer R. Improved patient survival using a modified resuscitationprotocol for out-of-hospital cardiac arrest. Circulation 2009;119:2597–2605.

30. Lund-Kordahl I, Olasveengen TM, Lorem T, Samdal M, Wik L,Sunde K. Improving outcome after out-of-hospital cardiac arrest bystrengthening weak links of the local chain of survival; quality of
advanced life support and post-resuscitation care. Resuscitation2010;81:422– 426.

heDelpSrtt

US2

JF

0d

Two-Year Safety and Effectiveness of Sirolimus-Eluting Stents(from a Prospective Registry)

Bimmer E. Claessen, MDa,b, Roxana Mehran, MDa,c, Martin B. Leon, MDa,d, Eric A. Heller, MDa,d,Giora Weisz, MDa,d, George Syros, MDa,d, Gary S. Mintz, MDa,d, Theresa Franklin-Bond, MSa,d,

Irene Apostolidou, MDa,d, Jose P.S. Henriques, MD, PhDb, Gregg W. Stone, MDa,d,Jeffrey W. Moses, MDa,d, and George D. Dangas, MD, PhDa,c,*

Uncertainty exists about the long-term safety and efficacy outcomes of sirolimus-elutingstents (SESs) in unselected patients. The present study was performed to evaluate thesafety and efficacy of the SES in treatment of patients with coronary artery disease in anunselected population. Over a 2-year period, 1,504 consecutive patients undergoing per-cutaneous coronary intervention with >1 SES were enrolled. The primary end point wasthe occurrence of target vessel failure (TVF; a composite of cardiac death, myocardialinfarction, or clinically driven target vessel revascularization). An independent clinicalevent committee adjudicated all adverse events up to 2-year follow-up. Dual antiplatelettherapy was recommended for >1 year throughout the study period. Mean age was 65 �11 years; 75% were men, and 34% were diabetics. SESs were implanted for off-labelindications in 86% of cases. TVF rates were 3.3%, 6.9%, 11.5%, and 15.5% at 30-day,6-month, 1-year, and 2-year follow-ups, respectively. The 2-year cumulative rate of defi-nite/probable stent thrombosis was 0.9%; 0.2% was very late thrombosis, occurring from1 year to 2 years. Patients off dual antiplatelet therapy at 6 months had a significantlyincreased rate of subsequent death from noncardiac causes. Patients off dual antiplatelettherapy at 1 year had a significantly decreased rate of subsequent clinically driven targetlesion revascularization. In conclusion, use of SESs in unselected patients with coronaryartery disease was associated with a low TVF rate at 2 years with an acceptable incidenceof stent thrombosis. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:
528–534)
Tto

M

asccDJIsi(ntipahntoA

The United States Food and Drug Administration (FDA)approved the use of drug-eluting stents (DESs) for treatmentof coronary artery disease based on results of randomizedcontrolled trials showing a significant decrease in restenosisand need for repeat revascularization with DESs comparedto bare metal stents.1–3 Typically, randomized pivotal trialsave excluded patients with complex coronary artery dis-ase with high risk for cardiac events. In clinical practice,ESs have also been used for off-label indications.4,5 How-

ver, expanded use of DES in everyday clinical practice isess well studied, and the possibility of unrecognized com-lications may exist.6–8 The Comprehensive Assessment ofirolimus-Eluting Stents in Complex Lesions (MATRIX)egistry was designed to evaluate the safety and efficacy ofhe sirolimus-eluting stent (SES) in an unselected popula-ion of patients with obstructive coronary artery disease.

aCardiovascular Research Foundation, New York, New York; bAca-demic Medical Center, University of Amsterdam, Amsterdam, The Neth-erlands; cMount Sinai Medical Center, New York, New York; dColumbia

niversity Medical Center, New York, New York. Manuscript receivedeptember 22, 2010; revised manuscript received and accepted October 5,010.

The MATRIX registry was funded by a research grant from Cordis/ohnson and Johnson, Warren, New Jersey, to the Cardiovascular Researchoundation, New York, New York.

*Corresponding author: Tel: 212-241-7014; fax 212-241-0273
wE-mail address: [email protected] (G.D. Dangas).

he SES was the only DES approved by the FDA at theime of study initiation. This report focuses on clinicalutcomes in the MATRIX registry up to 2-year follow-up.

ethods

The MATRIX registry was conducted under an FDA-pproved investigative device exemption and was a pro-pective, open-label, nonrandomized registry of 1,504onsecutive patients �18 years undergoing percutaneousoronary intervention (PCI) requiring the placement of �1ES (Cypher, Cordis, Johnson and Johnson, Warren, New

ersey) for single- or multivessel coronary artery disease.nclusion criteria were (1) �1 lesion with �50% diametertenosis in a native coronary artery or a bypass graft requir-ng PCI with stenting not to exceed 108 mm of stent length,2) de novo and restenotic lesions including in stent reste-osis and radiation failure, (3) reference diameter from 2.5o 3.5 mm, and (4) ability to understand and grant writtennformed consent. Exclusion criteria were (1) confirmedregnancy at time of index PCI, (2) known allergies tospirin, clopidogrel, or ticlopidine, (3) known allergies toeparin and bivalirudin, (4) known allergy to any compo-ent of a SES, and (5) a significant medical condition that inhe investigators’ opinion might interfere with a patient’sptimal participation in this study. From March 2004 tougust 2006, consecutive patients (n � 1,504) who under-
ent PCI with the approved SES at the 2 participating sites
www.ajconline.org


la

SPPPI

N

M

529Coronary Artery Disease/Two-Year Outcomes of Sirolimus-Eluting Stents

(Lenox Hill Hospital, New York, New York, and ColumbiaUniversity Medical Center, New York, New York) wereconsidered for enrollment in this study. The respective in-stitutional review boards approved the protocol and all pa-tients granted written informed consent.

Table 1Baseline characteristics (n � 1,504)

Age (years) 64.8 � 11.4Body mass index (kg/m2) 29 � 5.6Man 74.5%Diabetes mellitus 33.7%

Insulin-dependent diabetes mellitus 7.2%Hypertension* 82.5%Hyperlipidemia† 84.7%

moker 10.9%revious myocardial infarction 33.2%revious percutaneous coronary intervention 44.1%revious coronary artery bypass grafting 20.9%ndication for index procedureStable angina with abnormal stress test 63.7%Unstable angina pectoris 33.2%Acute myocardial infarction 3.4%umber of coronary arteries narrowed 2.0 � 0.91 35.8%2 33.2%�3 31.0%oderate/severe left ventricular dysfunction 8.2%

Data are presented as mean � SD or percentage.* Defined as a documented history of hypertension diagnosed and/or

treated by a physician.† Defines as a documented history of hyperlipidemia diagnosed and/or

treated by a physician.

Table 2Procedural characteristics (n � 1,504)

Number of lesions treated 2,879Number of lesions treated per patient 1.9 � 1.0Number of vessels treated per patient 1.3 � 0.5Treated vessel

Right coronary artery 32.0%Left anterior descending coronary artery 44.5%Left circumflex coronary artery 35.4%Ramus intermedius 4.4%Left main coronary artery 3.3%Saphenous vein graft 4.5%Arterial bypass graft 0.6%

Lesion length (mm) 18.0 � 9.8Reference vessel diameter (mm) 3.00 � 0.46Preprocedural minimum lumen diameter (mm) 0.68 � 0.39Postprocedural minimum lumen diameter (mm) 2.34 � 0.43Acute gain (mm) 1.67 � 0.48American Heart Association/American College

of Cardiology lesion type B2/C66.6%

Number of stents per patient 2.1 � 1.2Number of stents per lesion 1.1 � 0.5Stent length per lesion (mm) 24.1 � 12.7Stent diameter (mm) 3.03 � 0.42Predilatation performed 64.7%Postdilatatation performed 37.6%Maximum inflation pressure (atm) 15.5 � 2.6

Data are presented as mean � SD or percentage.

PCI and stent implantation were performed in the stan-

dard manner. Heparin was administered to maintain anactivated clotting time �250 seconds, and bivalirudin wasused as an alternative anticoagulant in most cases (85%)according to standard clinical practice in the 2 clinical sites.After intracoronary injection of nitroglycerin, pre- and post-procedural angiographies of the involved vessel(s) wereperformed in �2 near orthogonal views showing the targetesion free of foreshortening or vessel overlap to allow forccurate quantitative coronary angiographic measurements.9

Pre- and postdilatation were performed at the operator’s dis-cretion. In the event of an additional stent requirement, an SESwas used. The following SES sizes were used in the MATRIXregistry: 8, 18, 23, and 33 mm in length, with diameters of 2.5,3.0, and 3.5 mm. Use of glycoprotein IIb/IIIa inhibitors wasleft to the discretion of the operator. Successful stent implan-tation was defined as the achievement of a final diameter

Figure 1. Proportions of patients treated with a thienopyridine (mediumblue), aspirin (light blue), thienopyridine and aspirin (dark blue), none(black), or unknown (blue-gray), at 30 days, 180 days, 1 year, and 2 years.

Figure 2. Kaplan-Meier event rates of clinical end points cardiac death(light blue), myocardial infarction (teal), target vessel revascularization(dark blue), and target vessel failure (deep blue) at 30 days, 180 days, 1year, and 2 years. Kaplan-Meier event rates of Q-wave and non–Q-wavemyocardial infarction were 0.2% and 2.3% at 30 days, 0.3% and 2.5% at180 days, 0.4% and 2.8% at 1 year, and 0.4% and 3.5% at 2 years.

stenosis �50% by quantitative coronary angiography after

e at 2 y

L

L


stent implantation with normal flow. According to protocol,physicians prescribed aspirin 325 mg/day for 1 month, 81mg/day thereafter plus clopidogrel 75 mg/day for �1 year afterthe procedure. Ticlopidine was an option for possible clopi-dogrel allergy. A patient was provided with the prescriptionand a written instruction sheet. General practitioners and gen-eral cardiologists were informed by written transmission of thepreliminary and final procedure reports that included informa-tion on dose and duration of dual antiplatelet therapy.

Follow-up was planned at 30 days, 6 months, 1 year, and2 years after the index procedure; at these respective timepoints follow-up was available for 99.1%, 97.8%, 95.5%,and 85.3% of patients. At each follow-up time point, infor-mation was collected on antiplatelet medication adherenceand occurrence of clinical events. All clinical end points(see below) were adjudicated by an independent clinicalevents committee.

The primary end point was the occurrence of target

Figure 3. Target vessel failur

Figure 4. Incidence and timing of definite/probable stent thrombosis (ST)(crosses) in patients using thienopyridines (gray bars) and those withunknown thienopyridine status (white bars); patients off thienopyridinescould not be confirmed. No stent thrombosis event occurred when patientswere confirmed to be off thienopyridines.

vessel failure (TVF), a composite of cardiac death, Q-wave-

and non–Q-wave myocardial infarctions, or clinicallydriven target vessel revascularization. Q-wave myocardialinfarction was defined as the development of new patho-logic Q waves �0.04 second in duration in �2 contiguousleads as assessed by the electrocardiographic core labora-tory with creatine kinase or creatine kinase-MB levels in-creased above normal. Non–Q-wave myocardial infarctionwas defined as an increase of creatine kinase levels to �2times the upper normal limit with increased creatine ki-nase-MB in the absence of new pathologic Q waves. Targetvessel revascularization was considered clinically driven inpatients with a positive functional study result, ischemicchanges on electrocardiogram consistent with the targetvessel, an in-lesion diameter stenosis �50% by quantitativecoronary angiography if the patient has ischemic symptoms,or an in-lesion diameter stenosis �70% by quantitativecoronary angiography in the absence of ischemic symp-toms. Secondary clinical endpoints included rates of indi-

ears in high-risk subgroups.

Table 3Two-year Kaplan-Meier event rates of landmark analysis of patients onand off dual antiplatelet therapy

On Off p Value

andmark set at 6 monthsSubjects 1,199 45Death 2.5% 8.0% 0.047

From cardiac causes 1.0% 0.0% 0.539From noncardiac causes 1.6% 8.0% �0.001

Myocardial infarction 1.1% 0.0% 0.519Target vessel

revascularization8.7% 0.0% 0.056

Target vessel failure 9.9% 0.0% 0.041andmark set at 1 yrSubjects 1,106 180Death 1.5% 2.9% 0.166

From cardiac causes 0.6% 1.2% 0.383From noncardiac causes 0.9% 1.8% 0.152

Myocardial infarction 0.7% 1.2% 0.490Target vessel

revascularization4.7% 0.6% 0.0279

Target vessel failure 5.5% 3.0% 0.179

vidual clinical events and stent thrombosis. Stent thrombo-


sis was categorized according to definitions proposed by theAcademic Research Consortium as definite or probablestent thrombosis.9 Timing of stent thrombosis was classifiedas acute (�24 hours), subacute (24 hours to 30 days), late (1month to 1 year), and very late (�1 year).

Continuous variables were summarized using mean �SD. Categorical variables were summarized using fre-quencies. Survival curves using all available follow-updata were constructed for time-to-event variables usingthe Kaplan-Meier method and compared by log-rank test.Data on patients who were lost to follow-up were cen-sored at the time of the last contact. To investigate theimpact of cessation of dual antiplatelet therapy on sub-sequent clinical events, we performed landmark analysescomparing event rates (death, myocardial infarction, andclinically driven target vessel revascularization) betweenpatients on and off dual antiplatelet therapy. Two land-mark time points were considered, 6 months and 1 year.Statistical analyses were performed using SAS 9.1 (SASInstitute, Cary, North Carolina).

Results

In total 1,504 patients were enrolled in the MATRIXregistry; mean age was 65 � 11 years, 75% were men, and34% had diabetes mellitus. Additional baseline characteris-tics are listed in Table 1. SESs were successfully implantedin 98.6% of lesions and a mean of 2.1 � 1.2 SESs (perpatient) was implanted during the index procedure. Table 2lists procedural characteristics for the study cohort. Most

Table 4Two-year clinical event rates of patients treated with sirolimus-eluting ste

Year ofPublication

Number ofSES-Treated

Patients

Randomized trialsSIRIUS, E-SIRIUS, and C-SIRIUS

pooled analysis2006 758

RAVEL 2003 120ENDEAVOR III 2007 112SIRTAX 2008 503ISAR-TEST-3 2009 202LEADERS 2009 850SORT OUT III* 2010 1,170

Observational studiesMATRIX present 1,504Roy et al. 2008 2,099Ong et al. 2006 508Kaltoft et al. 2009 2,202Kimura et al. 2009 10,778

Direct cross-trial comparisons are discouraged.*Eighteen-month event rates. This trial excluded all postprocedural eveC-SIRIUS � Canadian study of sirolimus-coated stent in treatment of patie

trial of the Medtronic endeavor drug [ABT-578] eluting coronary stent systemartery lesions; E-SIRIUS � European study of sirolimus-coated stent in treateluting stents with different polymer coating to reduce restenosis; LEADERinfarction; NA � not available; RAVEL � randomized comparison of a sirsirolimus-coated stent in treatment of patients with de novo coronary arteryrevascularization; SORT OUT III � Danish organization on randomizedrevascularization; TVR � target vessel revascularization.

patients (86%) underwent stenting for off-label indications,

including multivessel stenting (n � 462, 30.7%), bifurca-tion lesions (n � 295, 19.6%), saphenous vein grafts (n �67, 4.5%), long-term total occlusions (n � 58, 3.9%), andacute myocardial infarction (n � 49, 3.3%). At time ofhospital discharge, 99.6% of patients were being treatedwith aspirin and 99.6% of patients were treated withclopidogrel or ticlopidine. Figure 1 shows the proportionof patients on aspirin and clopidogrel or ticlopidine dur-ing follow-up; high levels of adherence were docu-mented.

Rates of cardiac death, myocardial infarction, and isch-emia-driven target vessel revascularization and the compos-ite of TVF at 30-day, 6-month, 1-year, and 2-year follow-upare shown in Figure 2.

Figure 3 shows TVF rates at 2 years for selectedhigh-risk subgroups (multivessel stenting, diabetes mel-litus, bifurcation lesions, saphenous vein grafts, long-term total occlusions, acute myocardial infarction, andprevious brachytherapy). The highest 2-year TVF rate(39.0%) was observed in the brachytherapy group due tohigh rates of clinically driven target vessel revasculariza-tion (35.8%) and stent thrombosis (6.9%). In the diabeticsubgroup we observed a trend toward a higher incidenceof TVF at 2 years in insulin-treated diabetics compared tononinsulin-treated diabetics (24.9% vs 18.7%, p � 0.11).

Figure 4 shows the timing of definite/probable stentthrombosis events and whether a patient was using dualantiplatelet therapy at time of stent thrombosis. Rates ofacute, subacute, late, and very late definite/probable stent

andomized trials and registries

F Death/MI Death MI TVR TLR ST

% NA 2.1% 4.7% NA 5.7% 0.9%

NA 5.0% 2.5% NA 2.5% 0.0%% NA 4.5% 3.6% NA 4.5% NA% NA 5.0% 3.6% 9.3% 7.8% 2.4%

6.4% 5.0% 2.0% NA 10.4% 1.0%% 9.1% 5.1% 5.8% 8.8% 7.3% 2.5%% NA 2.7% 0.9% 3.3% 1.7% 0.5%

% 7.4% 3.9% 3.9% 12.1% 10.0% 0.9%% NA NA NA 13.2% NA 1.8%% 9.7% 5.8% NA 8.2% NA 0.4%

NA 5.9% 4.1% NA NA 1.7%NA 7.2% 1.5% NA 10.2% 0.9%

a 5-day period.de novo coronary artery lesions; ENDEAVOR III � randomized controlled

the cypher sirolimus-eluting coronary stent system in de novo native coronarypatients with de novo coronary artery lesions; ISAR-TEST-3 � rapamycin-us eluted from a durable versus erodable stent coating; MI � myocardial

luting stent with a standard stent for coronary revascularization; SIRIUS �; SIRTAX � sirolimus-eluting versus paclitaxel-eluting stents for coronarywith clinical outcome III; ST � stent thrombosis; TLR � target lesion

nts in r

TV

10.9

NA11.612.3

NA15.4

4.5

15.422.615.4

NANA

nts fornts withversus

ment ofS � lim

olimus-elesionstrials

thrombosis were 0%, 0.3%, 0.3%, and 0.2%, respectively.

elRippsr0t

vwtd

ucl

(t(SS

val


Of the 13 patents with stent thrombosis, 10 patients (77%)had the event while on confirmed dual antiplatelet therapyand 3 patients (23%) while at an unknown antiplatelettherapy status.

Antiplatelet therapy follow-up information was availablefor 88.8% and 89.5% of patients at 6 months and 1 year,respectively. Table 3 presents results of landmark analysescomparing event rates between patients on and off dualantiplatelet therapy. Patients off dual antiplatelet therapy at6 months had a significantly increased rate of subsequentdeath from noncardiac causes. Patients off dual antiplatelettherapy at 1 year had a significantly decreased rate ofsubsequent clinically driven target lesion revascularization.

Discussion

Results from this first registry under an FDA investiga-tive device exemption showed that use of SESs in treatmentof coronary artery disease in a high-risk unselected popu-lation (86% of patients received SESs for off-label indica-tions) was associated with low TVF (15.5%) and acceptablestent thrombosis (0.9%) rates up to 2 years.

The 1-year TVF rate in the present report was 11.5%,which is slightly higher than those reported in 2 recentlypublished all-comer trials comparing next-generation DESs(6% to 9% TVF at 1 year).10,11 However, the present trialnrolled a larger proportion of patients with diabetes mel-itus (34%) compared to patients in COMPARE (18%) andESOLUTE All Comers (23%). Moreover, stent design

mprovements in second-generation DESs might also ex-lain the favorable results in these recent trials, althoughrospective randomized data to confirm this hypothesis istill pending. One-year definite/probable stent thrombosisates in COMPARE and RESOLUTE All Comers were.7% to 3.0%, which is comparable to the rate in the presentrial, which was 0.7%.

Table 4 presents how 2-year clinical outcomes from thepresent study compare to those from randomized studiesand nonrandomized registries evaluating the SES for which2-year results have been published.6,12–21 Of note, amongregistries 2-year TVF rate was the lowest in the presentstudy, on par with the TVF rate reported by Ong et al,16

although diabetes mellitus was considerably more frequentin our cohort (34% vs 18%). Therefore, our data add to thegrowing body of evidence suggesting use of SESs for off-label indications is safe and effective.

Two-year safety and efficacy data have been reported forthe 3 other FDA-approved DES types. The most publisheddata are available for the paclitaxel-eluting stents; 2 yearTVF rates range from 13.1% to 21.1%, and stent thrombosisrates range from 1.3% to 2.4%.17,22–24 Two-year TVF andstent thrombosis rates for the zotarolimus-eluting stent were11.1% and 1.9%, respectively, in a randomized trial com-paring paclitaxel-eluting stents to zotarolimus-elutingstents.14 Two-year TVF and stent thrombosis rates for theeverolimus-eluting stent were 10.4% and 1.2%, respec-tively, in a pooled analysis of 2 randomized trials comparingeverolimus-eluting stents to paclitaxel-eluting stents.24

Analysis of various high-risk subgroups showed a high2-year TVF rate (30.8%) in patients treated for lesions in a
saphenous vein graft. The increased rate of adverse clinical s
events after treatment of saphenous vein graft lesions is wellestablished and is due to the nature of diffusely diseaseddegenerated vein grafts with an increased risk of periproce-dural complications due to potential distal embolization and anincreased need for repeat intervention.25–27 We observed aery high 2-year TVF rate (39.0%) in the subgroup of patientsith a history of intracoronary brachytherapy; high stent

hrombosis rates after intracoronary brachytherapy are wellocumented and have led to a decrease of its use over time.28,29

In the subgroup of 58 patients treated for long-term totalocclusions, TVF rate was 19.1% at 2 years, which is rela-tively similar to the 15.5% overall TVF rate in MATRIX.This is consistent with observations from a randomizedcontrolled trial of SESs in a lower-risk patient populationwith (sub)total coronary occlusions, which reported a favor-able event rate of 10.0% at 3 years.30 Our results confirm these of SESs as an effective treatment strategy for theseomplex lesions, typically treated with multiple (and mostlyong) stents.

The 2-year TVF rate of diabetic patients in MATRIX20.1%) compares well to those from diabetic patients inhe Registro Regionale Angioplastiche Emilia-RomagnaREAL) registry (23.3%) and the Rapamycin-Elutingtent Evaluated at Rotterdam Cardiology Hospital (RE-EARCH) registry (18.2%).22,31

TVF rates were relatively lower in the remaining high-risk in subgroups of acute myocardial infarction (15.0%),multiple vessels (19.6%), and bifurcation lesions (16.9%).In 2 small randomized clinical trials evaluating the use ofSESs in acute myocardial infarction, 2-year TVF rates were24.2% and 21%.32,33 To our knowledge, no previous studieshave reported 2-year clinical event rates for treatment ofmultiple vessels or bifurcation lesions with SESs. However,3-year death, myocardial infarction, and revascularizationrates in the Arterial Revascularization Strategies-II (ARTS-II) trial, which evaluated SESs in patients with multivesseldisease, were 3.0%, 2.8%, and 11.0%, respectively, com-pared to 4.1%, 4.6%, and 13.2% at 2 years in the presentstudy.34

Patients off dual antiplatelet therapy at 6 months had asignificantly increased rate of subsequent death from non-cardiac causes, possibly due to noncardiac diagnoses lead-ing to antiplatelet discontinuation. Although information onmedical co-morbidities (e.g., cancer, lung disease, etc.) wasnot routinely collected in the MATRIX registry, these re-sults suggest that patients who discontinue dual antiplatelettherapy are typically patients with severe co-morbidity whoare more likely to have a fatality from noncardiac causesand less likely to undergo repeat intervention. A secondanalysis showed no increased risk of developing a clinicalevent if dual antiplatelet therapy was discontinued after 12months. Interestingly, none of patients who sustained a stentthrombosis were confirmed to be off dual antiplatelet ther-apy before the event. Previous studies have shown an asso-ciation between clopidogrel discontinuation and subsequentstent thrombosis; this association was not found in theMATRIX registry.6–8 Possible explanations could be theery low overall incidence of stent thrombosis in MATRIXnd the high percent long-term adherence to dual antiplate-et therapy. These 2 factors may have limited the power to
how an association between clopidogrel discontinuation


and stent thrombosis in this 1,504-patient registry. How-ever, our present observation is in accordance with a recentinvestigation in an Asian population suggesting no clinicalbenefit of continuing dual antiplatelet therapy �1 year afterstenting.35

All clinical end points up to 2 years were adjudicated byan independent clinical event committee. However, longer-term follow-up is needed to confirm if the low TVF andstent thrombosis rates are maintained. Furthermore, al-though MATRIX is a large 1,504-patient registry, this studywas underpowered to evaluate a definitive association be-tween premature discontinuation of dual antiplatelet therapyand incidence of stent thrombosis.

Acknowledgment: We thank the clinical event adjudicationcommittee: Allen Jeremias, MD, Steven O. Marx, MD, andStanley Schneller, MD.

1. Morice MC, Serruys PW, Sousa JE, Fajadet J, Ban HE, Perin M,Colombo A, Schuler G, Barragan P, Guagliumi G, Molnar F, FaloticoR. A randomized comparison of a sirolimus-eluting stent with a stan-dard stent for coronary revascularization. N Engl J Med 2002;346:1773–1780.

2. Moses JW, Leon MB, Popma JJ, Fitzgerald PJ, Holmes DR,O’Shaughnessy C, Caputo RP, Kereiakes DJ, Williams DO, TeirsteinPS, Jaeger JL, Kuntz RE. Sirolimus-eluting stents versus standardstents in patients with stenosis in a native coronary artery. N EnglJ Med 2003;349:1315–1323.

3. Caixeta A, Leon MB, Lansky AJ, Nikolsky E, Aoki J, Moses JW,Schofer J, Morice MC, Schampaert E, Kirtane AJ, Popma JJ, Parise H,Fahy M, Mehran R. 5-Year clinical outcomes after sirolimus-elutingstent implantation insights from a patient-level pooled analysis of 4randomized trials comparing sirolimus-eluting stents with bare-metalstents. J Am Coll Cardiol 2009;54:894–902.

4. Marroquin OC, Selzer F, Mulukutla SR, Williams DO, Vlachos HA,Wilensky RL, Tanguay JF, Holper EM, Abbott JD, Lee JS, Smith C,Anderson WD, Kelsey SF, Kip KE. A comparison of bare-metal anddrug-eluting stents for off-label indications. N Engl J Med 2008;358:342–352.

5. Win HK, Caldera AE, Maresh K, Lopez J, Rihal CS, Parikh MA,Granada JF, Marulkar S, Nassif D, Cohen DJ, Kleiman NS. Clinicaloutcomes and stent thrombosis following off-label use of drug-elutingstents. JAMA 2007;297:2001–2009.

6. Kimura T, Morimoto T, Nakagawa Y, Tamura T, Kadota K, YasumotoH, Nishikawa H, Hiasa Y, Muramatsu T, Meguro T, Inoue N, HondaH, Hayashi Y, Miyazaki S, Oshima S, Honda T, Shiode N, Namura M,Sone T, Nobuyoshi M, Kita T, Mitsudo K. Antiplatelet therapy andstent thrombosis after sirolimus-eluting stent implantation. Circulation2009;119:987–995.

7. Lasala JM, Cox DA, Dobies D, Baran K, Bachinsky WB, Rogers EW,Breall JA, Lewis DH, Song A, Starzyk RM, Mascioli SR, DawkinsKD, Baim DS. Drug-eluting stent thrombosis in routine clinical prac-tice: two-year outcomes and predictors from the TAXUS ARRIVEregistries. Circ Cardiovasc Interv 2009;2:285–293.

8. van Werkum JW, Heestermans AA, Zomer AC, Kelder JC, SuttorpMJ, Rensing BJ, Koolen JJ, Brueren BR, Dambrink JH, Hautvast RW,Verheugt FW, ten Berg JM. Predictors of coronary stent thrombosis:the Dutch Stent Thrombosis Registry. J Am Coll Cardiol 2009;53:1399–1409.

9. Cutlip DE, Windecker S, Mehran R, Boam A, Cohen DJ, van Es GA,Steg PG, Morel MA, Mauri L, Vranckx P, McFadden E, Lansky A,Hamon M, Krucoff MW, Serruys PW. Clinical end points in coronarystent trials: a case for standardized definitions. Circulation 2007;115:2344–2351.

10. Serruys PW, Silber S, Garg S, van Geuns RJ, Richardt G, BuszmanPE, Kelbaek H, van Boven AJ, Hofma SH, Linke A, Klauss V, WijnsW, Macaya C, Garot P, Dimario C, Manoharan G, Kornowski R,
Ischinger T, Bartorelli A, Ronden J, Bressers M, Gobbens P, Negoita
M, Van LF, Windecker S. Comparison of zotarolimus-eluting andeverolimus-eluting coronary stents. N Engl J Med 2010;363:136–146.

11. Kedhi E, Joesoef KS, McFadden E, Wassing J, Van MC, Goedhart D,Smits PC. Second-generation everolimus-eluting and paclitaxel-elut-ing stents in real-life practice (COMPARE): a randomised trial. Lancet2010;375:201–209.

12. Kaltoft A, Jensen LO, Maeng M, Tilsted HH, Thayssen P, Bottcher M,Lassen JF, Krusell LR, Rasmussen K, Hansen KN, Pedersen L,Johnsen SP, Sorensen HT, Thuesen L. 2-Year clinical outcomes afterimplantation of sirolimus-eluting, paclitaxel-eluting, and bare-metalcoronary stents: results from the WDHR (Western Denmark HeartRegistry). J Am Coll Cardiol 2009;53:658–664.

13. Klauss V, Serruys PW, van Es GA, Buszman P, Ischinger T, Eberli F,Corti R, Wijns W, Morice MC, Di Mario C, Schuler G, Linke A, JuniP, Windecker S. Biolimus-eluting stent with biodegradable polymerversus sirolimus-eluting stent with durable polymer for coronary re-vascularisation (leaders): a randomised non-inferiority trial: 24 monthsfollow-up. Am J Cardiol 2009;104:XVI.

14. Leon MB, Kandzari DE. Two-year outcomes from the ENDEAVORIII trial: a randomized trial of the ENDEAVOR zotarolimus-elutingstent compared with the cypher sirolimus-eluting stent. J Am CollCardiol 2007;49(suppl):45B.

15. Morice MC, Serruys P, Costantini C, Wuelfert E, Wijns W, Fajadet J,Columbo A, Guagliumi G, Molnar F, Hayashi EB, Sousa JEM, PerinM. Two-year follow-up of the RAVEL study: A randomized studywith the sirolimus-eluting Bx VELOCITY stent in the treatment ofpatients with de-novo native coronary artery lesions. J Am Coll Car-diol 2003;41(suppl):32A.

16. Ong AT, van Domburg RT, Aoki J, Sonnenschein K, Lemos PA,Serruys PW. Sirolimus-eluting stents remain superior to bare-metalstents at two years: medium-term results from the Rapamycin-ElutingStent Evaluated at Rotterdam Cardiology Hospital (RESEARCH) reg-istry. J Am Coll Cardiol 2006;47:1356–1360.

17. Roy P, de la Bonello L, De LA, Okabe T, Pinto Slottow TL, SteinbergDH, Torguson R, Smith K, Xue Z, Satler LF, Kent KM, Suddath WO,Pichard AD, Waksman R. Two-year outcome of patients treated withsirolimus- versus paclitaxel-eluting stents in an unselected populationwith coronary artery disease (from the REWARDS Registry). Am JCardiol 2008;102:292–297.

18. Schampaert E, Moses JW, Schofer J, Schluter M, Gershlick AH,Cohen EA, Palisaitis DA, Breithardt G, Donohoe DJ, Wang H, PopmaJJ, Kuntz RE, Leon MB. Sirolimus-eluting stents at two years: apooled analysis of SIRIUS, E-SIRIUS, and C–SIRIUS with emphasison late revascularizations and stent thromboses. Am J Cardiol 2006;98:36–41.

19. Rasmussen K, Maeng M, Kaltoft A, Thayssen P, Kelbaek H, TilstedHH, Abildgaard U, Christiansen EH, Engstrom T, Krusell LR,Ravkilde J, Hansen PR, Hansen KN, Abildstrom SZ, Aaroe J, JensenJS, Kristensen SD, Botker HE, Madsen M, Johnsen SP, Jensen LO,Sorensen HT, Thuesen L, Lassen JF. Efficacy and safety of zotaroli-mus-eluting and sirolimus-eluting coronary stents in routine clinicalcare (SORT OUT III): a randomised controlled superiority trial. Lan-cet 2010;375:1090–1099.

20. Billinger M, Beutler J, Taghetchian KR, Remondino A, Wenaweser P,Cook S, Togni M, Seiler C, Stettler C, Eberli FR, Luscher TF, WandelS, Juni P, Meier B, Windecker S. Two-year clinical outcome afterimplantation of sirolimus-eluting and paclitaxel-eluting stents in dia-betic patients. Eur Heart J 2008;29:718–725.

21. Byrne RA, Kufner S, Tiroch K, Massberg S, Laugwitz KL, BirkmeierA, Schulz S, Mehilli J. Randomised trial of three rapamycin-elutingstents with different coating strategies for the reduction of coronaryrestenosis: 2-year follow-up results. Heart 2009;95:1489–1494.

22. Balducelli M, Ortolani P, Marzaroli P, Piovaccari G, Menozzi A,Manari A, Sangiorgio P, Tarantino F, Rossi R, Maresta A, Tondi S,Passerini F, Guastaroba P, Grilli R, Marzocchi A. Comparison of2-year clinical outcomes with sirolimus and paclitaxel-elutingstents for patients with diabetes: results of the Registro RegionaleAngiopLastiche Emilia-Romagna Registry. Catheter Cardiovasc In-terv 2009;75:327–334.

23. Leon MB, Kandzari DE, Eisenstein EL, Anstrom KJ, Mauri L, CutlipDE, Nikolsky E, O’Shaughnessy C, Overlie PA, Kirtane AJ, McLaurinBT, Solomon SL, Douglas JS Jr, Popma JJ. Late safety, efficacy, andcost-effectiveness of a zotarolimus-eluting stent compared with a pa-
clitaxel-eluting stent in patients with de novo coronary lesions: 2-year

2

2

2

2

2

2

3

3

3

3

3

3


follow-up from the ENDEAVOR IV trial (Randomized, ControlledTrial of the Medtronic Endeavor Drug [ABT-578] Eluting CoronaryStent System Versus the Taxus Paclitaxel-Eluting Coronary StentSystem in De Novo Native Coronary Artery Lesions). JACCCardiovasc Interv 2009;2:1208–1218.

4. Onuma Y, Serruys PW, Kukreja N, Veldhof S, Doostzadeh J, Cao S,Stone GW. Randomized comparison of everolimus- and paclitaxel-eluting stents: pooled analysis of the 2-year clinical follow-up from theSPIRIT II and III trials. Eur Heart J 2010;31:1071–1078.

5. Hoffmann R, Hamm C, Nienaber CA, Levenson B, Bonzel T, Sabin G,Senges J, Zahn R, Tebbe U, Pfannebecker T, Richardt HG, SchneiderS, Kelm M. Implantation of sirolimus-eluting stents in saphenous veingrafts is associated with high clinical follow-up event rates comparedwith treatment of native vessels. Coron Artery Dis 2007;18:559–564.

6. Okabe T, Lindsay J, Buch AN, Steinberg DH, Roy P, Slottow TL,Smith K, Torguson R, Xue Z, Satler LF, Kent KM, Pichard AD,Weissman NJ, Waksman R. Drug-eluting stents versus bare metalstents for narrowing in saphenous vein grafts. Am J Cardiol 2008;102:530–534.

7. van Twisk PH, Daemen J, Kukreja N, van Domburg RT, Serruys PW.Four-year safety and efficacy of the unrestricted use of sirolimus- andpaclitaxel-eluting stents in coronary artery bypass grafts. EuroInter-vention 2008;4:311–317.

8. Costa MA, Sabate M, van der Giessen WJ, Kay IP, Cervinka P,Ligthart JM, Serrano P, Coen VL, Levendag PC, Serruys PW. Latecoronary occlusion after intracoronary brachytherapy. Circulation1999;100:789–792.

9. Waksman R. Late thrombosis after radiation. Sitting on a time bomb.Circulation 1999;100:780–782.

0. Rahel BM, Laarman GJ, Kelder JC, ten Berg JM, Suttorp MJ. Three-year clinical outcome after primary stenting of totally occluded native

coronary arteries: a randomized comparison of bare-metal stentimplantation with sirolimus-eluting stent implantation for the treat-ment of total coronary occlusions (Primary Stenting of Totally Oc-cluded Native Coronary Arteries [PRISON] II study). Am Heart J2009;157:149–155.

1. Daemen J, Garcia-Garcia HM, Kukreja N, Imani F, de Jaegere PP,Sianos G, van Domburg RT, Serruys PW. The long-term value ofsirolimus- and paclitaxel-eluting stents over bare metal stents in pa-tients with diabetes mellitus. Eur Heart J 2007;28:26–32.

2. Menichelli M, Parma A, Giulivi A, Violini R. Randomized trial ofSirolimus-Eluting Stent Versus Bare-Metal Stent in Acute MyocardialInfarction (SESAMI 2 years results) (abstract). Am J Cardiol 2007;100(suppl):210L.

3. Valgimigli M, Campo G, Arcozzi C, Malagutti P, Carletti R, Ferrari F,Barbieri D, Parrinello G, Percoco G, Ferrari R. Two-year clinicalfollow-up after sirolimus-eluting versus bare-metal stent implantationassisted by systematic glycoprotein IIb/IIIa inhibitor infusion in pa-tients with myocardial infarction: results from the STRATEGY study.J Am Coll Cardiol 2007;50:138–145.

4. Serruys PW, Daemen J, Morice MC, De BB, Colombo A, Macaya C,Richardt G, Fajadet J, Hamm C, Dawkins KD, Vranckx P, Bressers M,Van DR, Schuijer M, Wittebols K, Pieters M, Stoll HP. Three-yearfollow-up of the ARTS-II#—sirolimus-eluting stents for the treatmentof patients with multivessel coronary artery disease. EuroIntervention2008;3:450–459.

5. Park SJ, Park DW, Kim YH, Kang SJ, Lee SW, Lee CW, Han KH,Park SW, Yun SC, Lee SG, Rha SW, Seong IW, Jeong MH, Hur SH,Lee NH, Yoon J, Yang JY, Lee BK, Choi YJ, Chung WS, Lim DS,Cheong SS, Kim KS, Chae JK, Nah DY, Jeon DS, Seung KB, Jang JS,
Park HS, Lee K. Duration of dual antiplatelet therapy after implanta-tion of drug-eluting stents. N Engl J Med 2010;362:1374–1382.

puriodfis

ahha

AsO

c

0d

Comparison of Morbidity and Mortality in Diabetics VersusNondiabetics Having Isolated Coronary Bypass Versus CoronaryBypass plus Valve Operations Versus Isolated Valve Operations

Serenella Castelvecchio, MDa,b,*, Lorenzo Menicanti, MDa, Ekaterina Baryshnikova, PhDb,Carlo de Vincentiis, MDa, Alessandro Frigiola, MDa, and Marco Ranucci, MDb, for the Surgical and

Clinical Outcome Research (SCORE) Group

The impact of diabetes mellitus (DM) on the outcome of patients requiring cardiac surgeryhas been investigated in previous decades. However, the profile of cardiac surgical practiceis changing in addition to changes in patients’ risk profile, making the results inconclusive.In this study we sought to investigate the impact of DM on operative mortality andmorbidity in patients undergoing cardiac surgery and adjust for patient and diseasecharacteristics. In total 10,709 patients (9,229 nondiabetics and 1,480 diabetics) wereadmitted to the study; 5,557 patients (1,012 diabetics) underwent an isolated coronaryoperation, 1,775 patients (278 diabetics) underwent coronary plus valve operations, and3,337 patients (209 diabetics) underwent valve operations. To control for differences inpatient and disease characteristics, a propensity score (for DM) was performed. DMincreased crude morbidity and this difference was maintained after risk adjustment forpropensity score; conversely, the crude operative mortality risk was higher in diabetics butnot significantly after adjustment for propensity score. Thereafter, DM remained indepen-dently associated to operative mortality risk in the valve population only (odds ratio 2.53,95% confidence interval 1.45 to 4.4, p � 0.001). In conclusion, DM has a significant impacton operative mortality of patients undergoing heart valve surgery. Although diabeticpatients undergoing coronary operations are not at increased risk of operative mortality,morbidity is significantly affected in the overall population. © 2011 Elsevier Inc. All
rights reserved. (Am J Cardiol 2011;107:535–539)
The global burden of diabetes mellitus (DM) is rapidlyincreasing and approximately 8% of adults in developedcountries have DM.1 Because of the proportion, it is ex-ected that an increasing number of patients with DM willndergo cardiac surgery in the future, making diabetes-elated operative risk assessment an important tool. Thempact of DM on the outcome of patients requiring coronaryperations has been systematically investigated in previousecades. However, results have been changing over timerom series where DM was found to be associated withncreased early and 30-day mortalities2,3 to more recenttudies where this finding was not confirmed.4,5 More re-

cently, because the profile of cardiac surgical practice ischanging (coronary surgery is decreasing, whereas the num-ber of patients requiring valve surgery is increasing)6 inddition to changes in patients’ risk profile, more attentionas been placed on risk assessment of patients undergoingeart valve surgery. However, currently few risk modelsddressing this important issue have been proposed7–10 and,

most important, these still have strict limitations that can

Departments of aCardiac Surgery and bCardiothoracic and Vascularnesthesia and ICU, IRCCS Policlinico San Donato, Milan, Italy. Manu-

cript received July 22, 2010; revised manuscript received and acceptedctober 5, 2010.

*Corresponding author: Tel: 39-02-5277-4842; fax: 39-02-5277-4615.E-mail address: [email protected] (S. Castelvec-

hio).


lead to an underestimation of the weight of DM in riskstratification. In this study we sought to investigate theimpact of DM on operative mortality and morbidity in asingle large institutional series of patients undergoing car-diac surgery and consider patient and disease characteristicswith propensity-score adjustment.

Methods

This is a retrospective study based on our prospectiveinstitutional database of cardiac surgical patients. The localethics committee approved the study design and waived theneed for an informed consent of patients. All patients pro-vided written consent to the scientific treatment of their datain an anonymous form at time of hospitalization.

All patients operated from April 2000 to April 2009 wereadmitted to this study. Patients were assigned to the diabeticor nondiabetic group according to their condition at hospitaladmission. Exclusion criteria were age �18 years and con-genital heart operations.

The study population included 10,709 patients; 5,557patients underwent an isolated coronary operation, 1,775patients underwent coronary plus valve operations, and3,337 underwent valve operations (single or multiple, in-cluding valve plus ascending aorta operations).

Patients with DM were identified as those receiving oralantidiabetic treatment and/or insulin at time of surgery.
Patients receiving nutritional modifications as the sole treat-
www.ajconline.org



ment for hyperglycemia were not considered diabetic; pa-tients in whom DM was discovered and treated duringhospitalization were considered diabetics. Insulin treatmentof transient postoperative hyperglycemia was not consid-ered a criterion for being included in the diabetic group.During and after the operation, blood glucose levels werecontrolled by insulin infusion to maintain a level �180mg/dl. At the time of surgery, all patients were on state-of-the-art optimized medical therapy. Mean duration of DMwas not available, and glucose tolerance, insulin resistance,and glycosylated hemoglobin levels were not evaluated.

Demographic (age, gender, and weight) data were col-lected. Preoperative laboratory assays included serum cre-atinine value (milligrams per deciliter) and hematocrit (per-centage). Cardiac function was assessed by left ventricularejection fraction measured before the operation with echo-cardiographic assessment. For repeated different measure-ments, the lowest value was used; other cardiac-relatedfactors collected were recent (within 30 days) myocardialinfarction, unstable angina, congestive heart failure (HF),preoperative use of intra-aortic balloon pump, and presence

Table 1Demographics and preoperative and operative details in diabetic andnondiabetic patients

Variable Nondiabetics(n � 9,229)

Diabetics(n � 1,480)

pValue

Age (years) 65.5 � 11.7 67.6 � 8.5 0.001Men 6,406 (69%) 1,011 (68%) 0.394Weight (kg) 73.1 � 14.7 75.5 � 14.4 0.001Hematocrit (%) 39.4 � 4.5 38.1 � 4.7 0.001Serum creatinine (mg/dl) 1.19 � 0.8 1.35 � 0.9 0.001Long-term dialytic

treatment61 (0.7%) 26 (1.8%) 0.001

Left ventricular ejectionfraction

0.52 � 0.11 0.49 � 0.12 0.001

Recent myocardialinfarction

1,352 (14.6%) 367 (24.8%) 0.001

Unstable angina pectoris 560 (6.1%) 110 (7.4%) 0.044Congestive heart failure 414 (4.5%) 127 (8.6%) 0.001Preoperative intra-aortic

balloon pump50 (0.5%) 15 (1.0%) 0.030

Active endocarditis 60 (0.7%) 8 (0.5%) 0.622Previous vascular

surgery364 (7.9%) 107 (7.2%) 0.001

Previous cardiac surgery 490 (5.3%) 57 (3.9%) 0.018Chronic obstructive

pulmonary disease617 (6.7%) 152 (10.3%) 0.001

Previous cerebrovascularaccident

402 (4.4%) 110 (7.4%) 0.001

Isolated coronary arterybypass grafting

4,545 (49.2%) 1,012 (68.4%) 0.001

Coronary artery bypassgrafting plus valveoperation

1,497 (16.2%) 278 (18.7%) 0.015

Valve operation 3,128 (33.9%) 209 (14.1%) 0.001Urgent operation 369 (4.0%) 72 (4.9%) 0.119Cardiopulmonary bypass

duration (minutes)75.6 � 37 72.2 � 34 0.001

Cross-clamp time(minutes)

50 � 26 45.7 � 23 0.001

Data expressed as mean � SD or number of patients (percentage).

of active endocarditis. The following co-morbid conditions

were recorded: chronic obstructive pulmonary diseasetreated with medication at time of surgery, previous cere-brovascular accident, chronic renal failure on dialytic treat-ment, previous vascular surgery, and previous cardiac surgery.

Operative data recorded were isolated coronary arterybypass graft (CABG) operation, CABG plus valve opera-tion, valve operation, urgent operation, and cardiopulmo-nary bypass (CPB) duration (minutes).

Postoperative outcome data included mechanical venti-lation time (hours), intensive care unit stay (days), postop-erative hospital stay (days), acute renal failure (peak post-operative serum creatinine level �2.0 mg/dl and �2 timesthe preoperative value), sepsis, stroke, and surgical re-ex-ploration. Major morbidity was defined according to theSTS National Database Risk Stratification Subcommittee as1 of the following: mechanical ventilation time �48 hours,sepsis or mediastinitis, acute renal failure, stroke, or surgicalre-exploration. Myocardial infarction and acute respiratoryinsufficiency were separately considered additional out-come variables. Operative mortality was defined as in-hos-pital mortality or mortality within 30 days from the operation.

Data in text and tables are expressed as mean � SD ofthe mean for continuous variables and as number and per-centage for categorical variables. Differences betweenmeans were assessed using unpaired Student’s t test andPearson chi-square test for differences between percentages.A propensity-score analysis determining the propensity ofbelonging to the diabetic group was performed with thefollowing steps: (1) factors significantly different betweendiabetics and nondiabetics were established as previouslystated, (2) factors associated to diabetic conditions (p �0.1)were entered into a multivariable stepwise forward logisticregression analysis, (3) the logistic equation provided apropensity score for likelihood of being diabetic, and (4)this score was used as an adjustment factor for all subse-quent analyses. Accuracy of the propensity score waschecked using receiver operating curve analysis. Calibrationof the propensity score was checked with Hosmer-Leme-show statistics.

Morbidity and mortality rates for each factor in thediabetic and nondiabetic groups were assessed using uni-

Table 2Factors determining propensity for being a diabetic patient atmultivariable stepwise forward logistic regression analysis

Factor With Diabetes RegressionCoefficient

p Value forAssociation

Age (years) 0.009 0.002Weight (kg) 0.013 0.001Hematocrit (%) �0.053 0.001Serum creatinine (mg/dl) 0.090 0.002Left ventricular ejection fraction �0.012 0.001Recent myocardial infarction 0.262 0.001Congestive heart failure 0.518 0.001Previous vascular surgery 0.350 0.004Chronic obstructive pulmonary disease 0.282 0.005Previous cerebrovascular accident 0.372 0.002Combined operation 0.606 0.001Isolated coronary artery bypass grafting 1.140 0.001Constant �1.836

variate and multivariable logistic regression analyses, re-

tS

R

1gtahltc

cp

wlo(

537Coronary Artery Disease/Diabetes and Cardiac Surgery

spectively, obtaining crude and adjusted odds ratios (ORs)with 95% confidence intervals (CIs).

Sensitivity analysis for subgroups of isolated coronary,coronary plus valve, and valve operations was performed.

Statistical significance was settled at a p value �0.05; allests were 2-sided; statistical analyses were performed withPSS 13.0 (SPSS, Inc., Chicago, Illinois).

esults

The number of diabetic patients in our population was,480 (13.8%), with 9,229 nondiabetic patients. The 2roups differed significantly in demographics, co-morbidi-ies, and operative details (Table 1). The diabetic group had

more severe risk profile due to older age, lower baselineematocrit and ejection fraction, higher serum creatinineevel, higher rate of many co-morbidities (long-term dialyticreatment, recent myocardial infarction, unstable angina,

Figure 1. Crude and adjusted values for propensity-score morbidity.

Figure 2. Crude and adjusted values for propensity-score mortality inoverall population and in subgroups.

ongestive HF, previous vascular and cardiac operations,

chronic obstructive pulmonary disease, previous cerebro-vascular accident), and a longer CPB duration. Patients inthe nondiabetic group had a slightly lower rate of CABGplus valve operations, a lower rate of isolated CABG oper-ations, and a higher rate of valve operations.

Twelve factors (Table 2) were independently associ-ated with diabetes and were used for developing thepropensity score.

According to this score, the patient population was strat-ified for the propensity of belonging to the diabetic group,with a likelihood of 1.6% to 85.4%. Accuracy of the pro-pensity score was confirmed by receiver operating curveanalysis, which revealed an area under the curve of 0.72(95% CI 0.706 to 0.734, p � 0.001) and the calibration wasonfirmed by Hosmer-Lemeshow statistics (chi-square 11.5,� 0.175).At univariate analysis, diabetic patients demonstrated a

orse outcome, with longer mechanical ventilation time,onger intensive care unit and hospital stay, and higher ratef major morbidity, acute renal failure, stroke, and mortalityFigures 1 and 2, Table 3). Rates of myocardial infarction

were 2% in nondiabetic patients and 1.6% in diabetic pa-tients (p � 0.356). Conversely, diabetic patients showed ahigher rate of acute respiratory insufficiency (3.7% vs 2.5%in nondiabetic patients, p � 0.005). When adjusted forpropensity score, diabetic patients still demonstrated alonger hospital stay and a higher risk of major morbidityand acute renal failure but not a different mortality risk. TheOR for major morbidity was 1.19 (95% CI 1.02 to 1.39, p �0.024) and the OR for acute renal failure was 1.36 (95% CI1.08 to 1.32, p � 0.01).

Sensitivity analysis on mortality was performed for iso-lated coronary operations, coronary plus valve operations,and valve operations. There were 5,557 patients (1,012diabetics, 18.2%) who underwent an isolated coronary op-eration, 1,775 (278 diabetics, 15.6%) who underwent acombined operation, and 3,337 (209 diabetics, 6.3%) whounderwent a valve operation. Within groups of patientsundergoing coronary operations, DM was a significant riskfactor for operative mortality at univariate analysis. Afteradjustment for propensity score, DM remained indepen-dently associated to operative mortality risk in the valvepopulation only (OR 2.53, 95% CI 1.45 to 4.4, p � 0.001).

Operative mortality rate in diabetic patients undergoing avalve operation was 9.4% (18 patients). Thirteen patientshad a cardiac cause for mortality, with low cardiac output

Table 3Outcome data in diabetic and nondiabetic patients

Variable Nondiabetics(n � 9,229)

Diabetics(n � 1,480)

p Value

Crude Adjusted

Mechanicalventilation(hours)

26 � 86 34 � 108 0.002 0.232

Intensive care unitstay (days)

3.1 � 5.0 3.8 � 6.7 0.001 0.036

Hospital stay(days)

6.5 � 7.1 7.6 � 9.9 0.001 0.001

Data expressed as mean � SD.

and multiorgan failure, 3 patients had refractory liver fail-

ntdws

tfaOfc

tCosisiuiamsppe

ca

id

ddmhatdrfhtd


ure, 1 had severe sepsis, and 1 died intraoperatively due toinability to wean from CPB.

Discussion

There are major findings of this study. (1) DM increasescrude morbidity and this difference is maintained after riskadjustment for propensity score. (2) Conversely, the crudeoperative mortality risk is higher in diabetics but not sig-nificantly after adjustment for propensity score. (3) How-ever, when cardiac operations were split into coronary,coronary combined with valve, and valve operations, oper-ative mortality risk was significantly higher in diabetic pa-tients undergoing valve operations even after adjustment forthe propensity score.

Results of the present study partly agree with those ofother investigators. This may be related to different defini-tions of morbidity or different subsets of patients. More-over, operative conduct may have been different accordingto different institutional protocols. Thus far, we found anassociation between DM and postoperative stroke in agree-ment with Szabó et al11 and Herlitz et al12 who found theincidence of stroke to be significantly increased in diabeticpatients undergoing CABG. Conversely, Kubal et al4 foundo association between DM and stroke; they speculated thathe results could be explained with the higher percentage ofiabetic patients undergoing off-pump CABG in agreementith previous reports showing a significant decrease in

troke when CPB was avoided.13 However, more recently,the Randomized On/Off Bypass (ROOBY) trial showed nodifference between on-pump and off-pump procedures inoccurrence of the primary short-term end point includingstroke.14

Previous studies have reported an association betweenDM and postoperative renal insufficiency in coronary11 orvalve15 surgery. More specifically, Grayson et al15 foundhat insulin-dependent DM was an independent risk factoror developing acute renal failure in all cardiac surgery withn adjusted OR of 3.31 (95% CI 1.75 to 6.26, p �0.001).ur findings partly concur with that; however, lack of in-

ormation regarding type of DM in our dataset makes theomparison inappropriate.

Previous studies have reported controversial results onhe impact of DM on early mortality of patients undergoingABG surgery. Thourani et al3 in the late 1990s reported anbserved mortality of 3.9% in diabetic patients after CABGurgery and found DM to be an independent predictor ofn-hospital mortality. Other studies from previous series hadimilar findings.2,16 More recent studies have reported noncreased risk of in-hospital mortality for diabetic patientsndergoing CABG surgery,4,5 suggesting different reasonsn interpreting the results as advances in surgical, perfusion,nd anesthetic techniques and/or strict perioperative glucoseanagement. Our results obtained in a large, more recent

eries confirm that operative mortality risk, adjusted forropensity score, is not affected by DM in the overallopulation or in patients undergoing isolated coronary op-rations or coronary plus valve operations.

However, although coronary artery disease is the mostommon cardiac manifestation in diabetic patients, DM
ppears to be strongly linked to HF and it has a greater
mpact on the prognosis of HF than of coronary arteryisease.17 DM and congestive HF commonly coexist; each

condition increases the likelihood of developing the other,and when they occur together in the same patient risk ofmortality increases markedly.18–20 Our data confirm therelation between DM and HF with the percentage of patientswith congestive HF being higher in diabetic patients com-pared to nondiabetics. However, such differences did notsimilarly affect operative mortality risk when cardiac oper-ations were split into coronary, coronary combined withvalve, and valve operations. It is well known that valvesurgery produces higher operative mortality, from 3% to 7%in symptomatic patients.21 This finding partly supports ourata showing an even higher operative mortality rate iniabetic patients undergoing a valve operation (9.4%);oreover, operative mortality risk remained significantly

igher in diabetic patients undergoing valve operations evenfter adjustment for propensity score. At least 2 explana-ions may account for this finding. Patients affected by valveisease without coronary artery involvement are usuallyeferred to surgery when symptoms of congestive HF occur,urther increasing operative mortality risk in patients with aigher percentage of congestive HF due to DM. Alterna-ively, the existence of a primary myocardial disease iniabetic patients22 leading to left ventricular dysfunction,

which in turn could increase operative risk, cannot be ex-cluded. Therefore, it could be hypothesized that an earlypostoperative prognosis is more affected by the underlyingpathophysiologic disease rather than by the surgery itself.

This study has some limitations. First, DM was notsystematically assessed using standardized diagnostic crite-ria. Mean duration of DM was not available, and glucosetolerance, insulin resistance, and glycosylated hemoglobinlevels were not evaluated, limiting the accuracy of ouranalysis. However, all patients were followed by the refer-ring clinician before being scheduled for cardiac surgery,and the diagnosis of DM was known before hospital admit-tance. Second, this is a retrospective observational studyand cannot account for all variables affecting outcome orvariables included in this analysis might be overestimated.However, retrospective comparisons with propensity-scoreadjustment are recognized as highly robust and may in somecases be acceptable as randomized control trials.

1. Harris MI, Flegal KM, Cowie CC, Eberhardt MS, Goldstein DE, LittleRR, Wiedmeyer HM, Byrd-Holt DD. Prevalence of diabetes, impairedfasting glucose, and impaired glucose tolerance in U.S. adults. TheThird National Health and Nutrition Examination Survey, 1988–1994.Diabetes Care 1998;21:518–524.

2. Cohen Y, Raz I, Merin G, Mozes B. Comparison of factors associatedwith 30-day mortality after coronary artery bypass grafting in patientswith versus without diabetes mellitus. Israeli Coronary Artery Bypass(ISCAB) Study Consortium. Am J Cardiol 1998;81:7–11.

3. Thourani VH, Weintraub WS, Stein B, Gebhart SS, Craver JM, JonesEL, Guyton RA. Influence of diabetes mellitus on early and lateoutcome after coronary artery bypass grafting. Ann Thorac Surg 1999;67:1045–1052.

4. Kubal C, Srinivasan AK, Grayson AD, Fabri BM, Chalmers JA. Effectof risk-adjusted diabetes on mortality and morbidity after coronaryartery bypass surgery. Ann Thorac Surg 2005;79:1570–1576.

5. Rajakaruna C, Rogers CA, Suranimala C, Angelini GD, Ascione R.The effect of diabetes mellitus on patients undergoing coronary sur-gery: a risk-adjusted analysis. J Thorac Cardiovasc Surg 2006;132:
802–810.

1

1

1

1

1

1

1

1

1

2

2

539Coronary Artery Disease/Diabetes and Cardiac Surgery

6. Rabkin E, Schoen FJ. Cardiovascular tissue engineering. CardiovascPathol 2002;11:305–317.

7. Edwards FH, Peterson ED, Coombs LP, DeLong ER, Jamieson WR,Shroyer ALW, Grover FL. Prediction of operative mortality after valvereplacement surgery. J Am Coll Cardiol 2001;37:885–892.

8. Nowicki ER, Birkmeyer NJ, Weintraub RW, Leavitt BJ, Sanders JH,Dacey LJ, Clough RA, Quinn RD, Charlesworth DC, Sisto DA, UhligPN, Olmstead EM, O’Connor GT; Northern New England Cardiovas-cular Disease Study Group and the Center for Evaluative ClinicalSciences, Dartmouth Medical School. Multivariable prediction of in-hospital mortality associated with aortic and mitral valve surgery inNorthern New England. Ann Thorac Surg 2004;77:1966–1977.

9. Florath I, Rosendahl UP, Mortasawi A, Bauer SF, Dalladaku F, EnnkerIC, Ennker JC. Current determinants of operative mortality in 1400patients requiring aortic valve replacement. Ann Thorac Surg 2003;76:75–83.

10. Ambler G, Omar RZ, Royston P, Kinsman R, Keogh BE, Taylor KM.Generic, simple risk stratification model for heart valve surgery. Cir-culation 2005;112:224–231.

1. Szabó Z, Håkanson E, Svedjeholm R. Early postoperative outcomeand medium-term survival in 540 diabetic and 2239 nondiabetic pa-tients undergoing coronary artery bypass grafting. Ann Thorac Surg2002;74:712–719.

2. Herlitz J, Wognsen GB, Emanuelsson H, Haglid M, Karlson BW,Karlsson T, Albertsson P, Westberg S. Mortality and morbidity indiabetic and nondiabetic patients during a 2-year period after coronaryartery bypass grafting. Diabetes Care 1996;19:698–703.

3. Patel NC, Deodhar AP, Grayson AD, Pullan DM, Keenan DJ, HasanR, Fabri BM. Neurological outcomes in coronary surgery: independenteffect of avoiding cardiopulmonary bypass. Ann Thorac Surg 2002;74:400–406.

4. Shroyer AL, Grover FL, Hattler B, Collins JF, McDonald GO, KozoraE, Lucke JC, Baltz JH, Novitzky D. Veterans Affairs Randomized

2

On/Off Bypass (ROOBY) Study Group. On-pump versus off-pumpcoronary-artery bypass surgery. N Engl J Med 2009;361:1827–1837.

5. Grayson AD, Khater M, Jackson M, Fox MA. Valvular heart operationis an independent risk factor for acute renal failure. Ann Thorac Surg2003;75:1829–1835.

6. Morris JJ, Smith LR, Jones RH, Glower DD, Morris PB, MuhlbaierLH, Reves JG, Rankin JS. Influence of diabetes and mammary arterygrafting on survival after coronary bypass. Circulation 1991;84(suppl):III275–III284.

7. De Groote P, Lamblin N, Mouquet F, Plichon D, McFadden E, VanBelle E, Bauters C. Impact of diabetes mellitus on long-term survivalin patients with congestive heart failure. Eur Heart J 2004;25:656–662.

8. MacDonald MR, Petrie MC, Hawkins NM, Petrie JR, Fisher M,McKelvie R, Aguilar D, Krum H, McMurray JJ. Diabetes, left ven-tricular systolic dysfunction, and chronic heart failure. Eur Heart J2008;29:1224–1240.

9. Gustafsson I, Brendorp B, Seibaek M, Burchardt H, Hildebrandt P,Køber L, Torp-Pedersen C. Danish Investigators of Arrhythmia andMortality on Dofetilide Study Group. Influence of diabetes and dia-betes-gender interaction on the risk of death in patients hospitalizedwith congestive heart failure. J Am Coll Cardiol 2004;43:771–777.

0. Pocock SJ, Wang D, Pfeffer MA, Yusuf S, McMurray JJ, SwedbergKB, Ostergren J, Michelson EL, Pieper KS, Granger CB. Predictors ofmortality and morbidity in patients with chronic heart failure. EurHeart J 2006;27:65–75.

1. Vahanian A, Baumgartner H, Bax J, Butchart E, Dion R, Filippatos G,Flachskampf F, Hall R, Iung B, Kasprzak J, Nataf P, Tornos P,Torracca L, Wenink A. Guidelines on the management of valvularheart disease: the Task Force on the Management of Valvular HeartDisease of the European Society of Cardiology. Eur Heart J 2007;28:230–268.

2. Boudina S, Abel ED. Diabetic cardiomyopathy revisited. Circulation2007;115:3213–3223.

oo

g

Ia

CcINN

0d

Relation of Bundle Branch Block to Long-Term (Four-Year)Mortality in Hospitalized Patients With Systolic Heart Failure

Alon Barsheshet, MDa,*, Ilan Goldenberg, MDa, Moshe Garty, MD, MScb, Shmuel Gottlieb, MDc,Amir Sandach, PhDd, Avishag Laish-Farkash, MDc, Michael Eldar, MDc, and Michael Glikson, MDc

There is controversy regarding type of bundle branch block (BBB) that is associated withincreased mortality risk in patients with heart failure (HF). The present study was designedto explore the association between BBB pattern and long-term mortality in hospitalizedpatients with systolic HF. Risk of 4-year all-cause mortality was assessed in 1,888 hospi-talized patients with systolic HF (left ventricular ejection function <50%) without apacemaker in a prospective national survey. Cox proportional hazards regression modelingwas used to compare mortality risk in patients with right BBB (RBBB; 10%), left BBB(LBBB; 14%), and no BBB (76%) on admission electrocardiogram. At 4 years of follow up,mortality rates were highest in patients with RBBB (69%), intermediate in those withLBBB (63%), and lowest in those without BBB (50%, p <0.001). Multivariate analysisdemonstrated a significant 36% increased mortality risk in patients with RBBB versus noBBB (p � 0.002) but no significant difference in mortality risk for patients with LBBBversus no BBB (hazard ratio 1.04, p � 0.66). RBBB versus LBBB was associated with a 29%(p � 0.035) increased risk for 4-year mortality in the total population and with a 58% (p �0.015) increased risk in patients with ejection fraction <30%. In conclusion, RBBB but notLBBB on admission electrocardiogram is associated with a significant increased long-termmortality risk in hospitalized patients with systolic HF. Deleterious effects of RBBB comparedto LBBB appear to be more pronounced in patients with more advanced left ventricular
dysfunction. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:540–544)
psmpl

M

e2h

Prolongation of QRS interval (�120 ms) in patients withheart failure (HF) is common (14% to 47%)1 and is asso-ciated with higher all-cause mortality, cardiovascular death,or hospitalization for HF compared to patients with HF andnormal QRS interval.2,3 There is controversy regarding typef bundle branch block (BBB) that is associated with poorerutcome in patients with HF,4–8 with most studies showing

that left BBB (LBBB) is an independent prognostic marker,whereas right BBB (RBBB) is a weaker marker or notassociated with worse prognosis. Conversely, we previouslyshowed in hospitalized patients with HF that RBBB, but notLBBB, is associated with increased 1-year mortality risk, anassociation that was stronger for patients with systolic HF,particularly for patients with severe left ventricular (LV)dysfunction.9 However, currently there are limited data re-arding the effect of BBB pattern on long-term mortality in

aCardiology Division, University of Rochester Medical Center, Roch-ester, New York; bRecanati Center for Internal Medicine and Research,Rabin Medical Center, Petah Tiqva, Israel; cHeart Institute, Sheba MedicalCenter and Sackler School of Medicine, Tel Aviv University, Tel Aviv,Israel; dDepartment of Mathematics, Bar-Ilan University, Ramat Gan,srael. Manuscript received August 14, 2010; revised manuscript receivednd accepted October 1, 2010.

The Heart Failure Survey in Israel 2003 was supported by the Israelenter for Disease Control, Ramat Gan, Israel; The Israel Medical Asso-iation, Ramat Gan, Israel; Teva, Petah Tiqva, Israel; Levant, Herzelia,srael; Neopharm, Petah Tiqva, Israel; Pfizer, Herzelia, Israel; Aventis,etania, Israel; Dexxon, Or Akiva, Israel; Medisson, Petah Tiqva, Israel;ovartis, Petah Tiqva, Israel; and Schering-Plough, Petah Tiqva, Israel.

E-mail address: [email protected] (A. Barsheshet).

atients with LV dysfunction. Accordingly, the presenttudy aimed to investigate the association between QRSorphology and long-term mortality in 1,888 patients hos-

italized with systolic HF who were prospectively fol-owed-up over an extended 4-year period.

ethods

Baseline and admission characteristics of patients werextracted from the Heart Failure Survey in Israel (HFSIS;003) database. Design and methods of the HFSIS registryave been described previously.10 Briefly, the survey, con-

ducted in March and April 2003, included 4,102 patientsadmitted with a diagnosis of HF. Criteria used for diagnosisof HF were symptoms of HF (at rest or during exercise) andobjective evidence of cardiac dysfunction at rest.11 Therewere 3 subgroups of diagnoses for hospitalized patients: (1)acute de novo HF, (2) worsening of chronic HF, and (3)chronic stable HF with hospitalization unrelated to HF ex-acerbation. There were 2,090 patients with HF and LVejection fraction (LVEF) �50% as demonstrated by echo-cardiography. We excluded from the study 188 patients whohad a permanent pacemaker including a biventricular pace-maker and 14 patients who lacked electrocardiographicdata. Thus, the final analysis included 1,888 patients. Theend point of the study was all-cause mortality, which wasassessed for all patients by matching their identificationnumbers with the Israeli National Population Registry. Mor-tality data were obtained for all study patients at a 4-yearperiod from hospitalization, providing an extended fol-
low-up to the previously reported 1-year outcome study.9
www.ajconline.org


d

a

mv

H

A

L

541Heart Failure/BBB and Mortality in Heart Failure Patients

LBBB was defined as QRS duration �120 ms, uprightcomplexes with notched R waves in leads I, V5, and V6, andQS or rS pattern in lead V1. RBBB was defined as QRSuration �120 ms, a monophasic R wave in lead V1 or rSR

in leads V1 and V2, and deep slurred S waves in leads I, V5,nd V6. LVEF classes determined by echocardiography

with visual assessment were classified as normal (�50%),mildly impaired (40% to 49%), moderately impaired (30%to 39%), and severely impaired (�30%). Median and inter-quartile range timing of echocardiography were 0 monthand 0 month to 6 months.

Characteristics of patients categorized by BBB type werecompared by nonparametric Kruskal-Wallis test or chi-square test. Cumulative probability of survival by BBB typewas graphically displayed according to the Kaplan-Meiermethod with comparison by log-rank test. To examine therelation between RBBB, LBBB, and no BBB and mortality,several models were applied. First, potential variables (iden-tified in previous published studies as risk factors for mor-

Table 1Clinical characteristics of study patients by bundle branch block pattern

Variable Total(n � 1,888)

istoryAge (years) 73 (63–80)Women 33%Hypertension 65%Diabetes mellitus 53%Smoker 35%Coronary heart disease 82%New York Heart Association functional

class III to IV40%

Previous myocardial infarction 62%Previous stroke 13%Chronic obstructive pulmonary disease 18%Atrial fibrillation 25%Acute heart failure 61%Left ventricular ejection fraction estimated

by echocardiography (%)40–49 30%30–39 36%�30 34%

Admission systolic blood pressure (mm Hg) 135 (118–157)Admission heart rate (beats/min) 81 (70–98)dmission laboratory valuesCreatinine (mg/dl)† 1.2 (0.9–1.6)Sodium (mmol/L) 138 (136–141)Hemoglobin (g/dl) 12.5 (11.0–13.8)ong-term medications� Blockers 72%Angiotensin-converting enzyme inhibitors

or angiotensin receptor blockers81%

Furosemide 75%Spironolactone 25%Digoxin 17%Statins 53%

Data are presented as median (interquartile range) or percentage of pat* For overall difference among the 3 subgroups.† To convert creatinine to micromoles per liter, multiply by 88.4.

tality or clinical variables that were associated with mortal-

ity) were evaluated by univariate analysis and selected basedon clinical and statistical significance. Second, multivariateanalysis was carried out using Cox proportional hazards re-gression modeling adjusted for age (continuous), gender, NewYork Association (NYHA) functional classes III to IV versusI to II, previous myocardial infarction, atrial fibrillation, pre-vious stroke, diabetes, chronic obstructive pulmonary disease,cirrhosis, malignant tumor, LVEF class, admission creatininelevels (continuous), systolic blood pressure �115 versus �115

m Hg, sodium �136 versus �136 mEq/L, hemoglobin �10ersus �10 g/dl and long-term use of statins, � blockers, and

angiotensin-converting enzyme inhibitors or angiotensin re-ceptor blockers. Analyses were conducted with SAS 9.2 (SASInstitute, Cary, North Carolina).

Results

Of the 1,888 patients with systolic HF, 306 (14%) hadLBBB on admission electrocardiogram and 193 (10%) had

Bundle Branch Block p Value*

Left306, 14%)

Right(n � 193, 10%)

None(n � 1,389, 76%)

68–81) 74 (67–81) 71 (62–79) �0.00133% 25% 35% 0.02665% 69% 64% 0.39255% 55% 53% 0.55431% 38% 36% 0.14482% 82% 82% 0.99551% 45% 35% �0.001

55% 67% 62% 0.02113% 11% 14% 0.56317% 17% 18% 0.85529% 28% 24% 0.12060% 57% 61% 0.588

15% 20% 35% �0.00134% 40% 36%51% 40% 29%118–153) 131 (116–156) 136 (119–158) 0.36269–95) 80 (68–92) 82 (70–100) 0.061

1.0–1.8) 1.4 (1.1–1.8) 1.2 (0.9–1.6) 0.160135–141) 138 (135–141) 139 (136–141) 0.90311.2–13.7) 12.5 (11.3–13.8) 12.4 (10.9–13.8) 0.563

69% 71% 73% 0.33283% 82% 80% 0.510

89% 81% 71% �0.00134% 32% 22% �0.00128% 23% 14% �0.00152% 43% 55% 0.005

(n �

76 (

135 (80 (

1.3 (138 (12.6 (

ients.

RBBB. Table 1 presents baseline clinical characteristics of

L

L

�oem issing h


patients indicating that patients with LBBB were older thanthose with RBBB or no BBB. Prevalence of NYHA func-tional classes III to IV and low EF was highest in LBBB,lower in RBBB, and lowest in no BBB. Long-term therapywith furosemide, spironolactone, and digoxin was morefrequent in patients with LBBB compared to those withRBBB or no BBB. In the RBBB subgroup the proportion ofwomen was smaller than in the LBBB or no-BBB sub-groups. Systolic pulmonary arterial pressure data wereavailable for 841 patients (45%). There was a trend towarda higher systolic pulmonary arterial pressure in patients withRBBB compared to those with LBBB or no BBB (mean �SD 46 � 14, 43 � 15, and 43 � 16 mm Hg, respectively,p � 0.107 for overall comparison among the 3 subgroups).

For the total study population, Kaplan-Meier survival

Figure 1. Kaplan-Meier survival curves for patients with heart failure and lpatterns.

Table 2Multivariate analysis: bundle branch block pattern as a predictor of four-y

Variable Number of Patients

eft ventricular ejection fraction �50%All patients 1,800Right versus no bundle branch block 182 vs 1,322Left versus no bundle branch block 296 vs 1,322Right versus left bundle branch block 182 vs 296eft ventricular ejection fraction �30%All patients 609Right versus no bundle branch block 70 vs 389Left versus no bundle branch block 150 vs 389Right versus left bundle branch block 70 vs 150

Adjusted for age, gender, New York Heart Association functional classe10 g/dl, systolic blood pressure �115 mm Hg, left ventricular ejection f

f stroke, diabetes mellitus, chronic obstructive pulmonary disease, malignnzyme inhibitors/angiotensin receptor blockers, and statins. Thirty-sevenissing sodium levels, 10 were missing creatinine levels, and 10 were m

curves (Figure 1) demonstrated that during the first 8

months of follow-up mortality rates were higher in patientswith RBBB and LBBB compared to those who had no BBB.After 8 months and until the end of the extended 4-yearfollow-up period, curves of RBBB and LBBB separated,showing graded decrements in survival in patients withRBBB, LBBB, and no BBB, respectively. Mortality rates at4 years were 69%, 63%, and 50% in the RBBB, LBBB, andno-BBB groups, respectively (p �0.001; Figure 1). Consis-tent with these findings, multivariate analysis (Table 2)showed a significant increase in mortality risk in patientswith RBBB compared to those with no BBB, whereas nostatistically significant difference in mortality risk was ob-served between patients with LBBB and those with no BBB.Comparison of the outcome between patients with RBBBand those with LBBB showed a significant 29% (p � 0.035)

ricular ejection fraction �50% analyzed according to bundle branch block

rtality

Crude Mortality Hazard Ratio(95% confidence interval)

p Value

(54%) — —(68%) vs 665 (50%) 1.36 (1.12–1.65) 0.002(63%) vs 665 (50%) 1.04 (0.88–1.23) 0.655(68%) vs 187 (63%) 1.29 (1.02–1.64) 0.035

(65%) — —(77%) vs 242 (62%) 1.55 (1.14–2.11) 0.005(67%) vs 242 (62%) 1.03 (0.81–1.31) 0.823(77%) vs 100 (67%) 1.58 (1.09–2.28) 0.015

IV, admission creatinine levels, sodium level �136 mmol/L, hemoglobinclass, previous myocardial infarction, history of atrial fibrillation, historyepatic cirrhosis, and long-term use of � blockers, angiotensin-convertingwere missing New York Heart Association functional class data, 23 wereemoglobin levels.

eft vent

ear mo

976124187124

39654

10054

s III andractionancy, h

patients

increase in long-term mortality risk in the former subgroup

A

543Heart Failure/BBB and Mortality in Heart Failure Patients

compared to the latter subgroup. Similar results were foundfor the 2 subgroups of acute and chronic HF. Notably,RBBB was associated with a similar magnitude of riskincrease as demonstrated for other known predictors ofmortality risk in this population including age, diabetesmellitus, advanced LV dysfunction, or renal dysfunction(Table 3).

Long-term mortality risk associated with RBBB com-pared to no BBB or LBBB on admission electrocardiogramwas even more pronounced in patients with more advancedLV dysfunction (LVEF �30%). Kaplan-Meier survivalcurves showed a significant increase in 4-year mortality inpatients with RBBB compared to those with LBBB or no

Table 3Independent predictors of four-year mortality for patients with heart failu

Variable

Hazard Ratio(95% confidence in

Right versus no bundle branch block 1.36 (1.12–1.6Left versus no bundle branch block 1.04 (0.88–1.2

ge per decade 1.42 (1.33–1.5New York Heart Association class III to IV 1.62 (1.42–1.8Left ventricular ejection fraction �30% 1.47 (1.24–1.7Diabetes mellitus 1.47 (1.29–1.6Chronic obstructive pulmonary disease 1.40 (1.19–1.6Creatinine per 1 mg/dl* 1.22 (1.16–1.2Sodium �136 mmol/L 1.20 (1.04–1.3Systolic blood pressure �115 mm Hg 1.44 (1.24–1.6Hepatic cirrhosis 1.63 (1.16–2.2Malignancy 1.63 (1.30–2.0

* To convert creatinine to micromoles per liter, multiply by 88.4.

Figure 2. Kaplan-Meier survival curves for patients with heart failure and lpatterns.

BBB throughout follow-up (77%, 66%, and 62%, respec-

tively, p � 0.008; Figure 2). Multivariate analysis (Table 2)comparing long-term outcome of patients with BBB showeda 58% increase in mortality risk (p � 0.015) in patients withRBBB compared to those who had LBBB.

Discussion

The present study is the first to assess the long-termeffect of BBB pattern on mortality risk in hospitalizedpatients with systolic HF. The findings of the present studyextend our previous observation regarding the short-term(1-year) outcome of patients with HF and RBBB.9 We haveshown that in patients with systolic HF RBBB is associated

eft ventricular ejection fraction �50% and �30%

Left Ventricular Ejection Fraction

�30%

p Value Hazard Ratio(95% confidence interval)

p Value

0.002 1.55 (1.14–2.11) 0.0050.655 1.03 (0.81–1.31) 0.823

�0.001 1.31 (1.18–1.44) �0.001�0.001 1.45 (1.18–1.78) �0.001�0.001 — —�0.001 1.30 (1.05–1.61) 0.016�0.001 1.59 (1.24–2.03) �0.001�0.001 1.15 (1.05–1.26) 0.004

0.012 1.26 (1.01–1.57) 0.043�0.001 1.38 (1.12–1.72) 0.003

0.004 2.07 (1.23–3.46) 0.006�0.001 1.62 (1.08–2.41) 0.019

ricular ejection fraction �30% analyzed according to bundle branch block

re and l

�50%

terval)

5)3)2)5)4)8)4)9)8)8)7)4)

eft vent

with a significant increased long-term (4-years) mortality

ttrccathpL

1

1

1

1

1

1

1

1


risk compared to a no-BBB or LBBB pattern on baselineelectrocardiogram. Furthermore, our data suggest that theassociation between RBBB and increased mortality risk ismore pronounced in patients with a lower LVEF.

Prolonged QRS interval was shown to be an independentpredictor of high postdischarge morbidity and mortality inpatients hospitalized due to HF.2,3 Several studies investi-gating the predictive value of QRS morphology in patientswith HF yielded conflicting results regarding mortality riskassociated with BBB pattern.4–8 Some studies have shownhat in patients with HF the prevalence of LBBB is higherhan in those with RBBB (13% to 25% vs 6% to 14%,espectively),4,5,7 LBBB is associated with more severe HFharacterized by advanced NYHA functional class and de-reased LVEF, whereas RBBB is more prevalent in mennd is not associated with advanced HF symptoms or ven-ricular dysfunction.5,6,12 In contrast to previous studies, weave shown that RBBB but not LBBB is an independentredictor of 4-year mortality. Notably, RBBB compared toBBB was associated with a 29% increase (p � 0.035) in

long-term mortality in patients with LVEF �50% and a58% increase (p � 0.015) in long-term mortality in patientswith LVEF �30%.

Contrasting results among the different studies may berelated to important differences in characteristics among thepopulations studied. Notably, the present study consistedmostly of elderly men with coronary heart disease, which isa relatively unique population.

Acquired RBBB is often associated with pulmonary hy-pertension13,14 and right-sided HF,15,16 whereas presence ofacquired LBBB in patients with HF is more closely corre-lated with LV structure and function.17 Consistently in thepresent study we observed a trend toward higher systolicpulmonary arterial pressure in patients with RBBB com-pared to those with LBBB or no BBB. We did not collectinformation regarding clinical or echocardiographic pres-ence of right-sided HF. However, patients with left systolicHF and RBBB may have higher systolic pulmonary arterypressure and worse right ventricular function, leading toincreased mortality. Furthermore, in the present study therisk associated with RBBB was more pronounced in pa-tients with more advanced LV dysfunction. Thus, in thispopulation coexistence of RBBB with advanced LV dys-function may represent biventricular failure. In contrast,presence of LBBB may not provide incremental prognosticinformation in patients who already have more advancedLV dysfunction.7

The present study has several limitations; we could notdetermine the type of death in the study population andaccordingly did not determine the mechanism related to theobserved increase in long-term mortality risk. Furthermore,because we did not collect QRS duration data, except forcriteria used for RBBB and LBBB (which included QRSduration �120 ms), we were unable to determine whetherQRS duration or type of BBB is a better predictor of
mortality.
1. Kashani A, Barold SS. Significance of QRS complex duration inpatients with heart failure. J Am Coll Cardiol 2005;46:2183–2192.

2. Wang NC, Maggioni AP, Konstam MA, Zannad F, Krasa HB, BurnettJC Jr, Grinfeld L, Swedberg K, Udelson JE, Cook T, Traver B,Zimmer C, Orlandi C, Gheorghiade M. Clinical implications of QRSduration in patients hospitalized with worsening heart failure andreduced left ventricular ejection fraction. JAMA 2008;299:2656–2566.

3. Iuliano S, Fisher SG, Karasik PE, Fletcher RD, Singh SN. QRSduration and mortality in patients with congestive heart failure. AmHeart J 2002;143:1085–1091.

4. McCullough PA, Hassan SA, Pallekonda V, Sandberg KR, Nori DB,Soman SS, Bhatt S, Hudson MP, Weaver WD. Bundle branch blockpatterns, age, renal dysfunction, and heart failure mortality. Int J Car-diol 2005;102:303–308.

5. Baldasseroni S, Gentile A, Gorini M, Marchionni N, Marini M, Ma-sotti G, Porcu M, Maggioni AP. Intraventricular conduction defects inpatients with congestive heart failure: left but not right bundle branchblock is an independent predictor of prognosis. A report from theItalian Network on Congestive Heart Failure (IN-CHF database). ItalHeart J 2003;4:607–613.

6. Baldasseroni S, Opasich C, Gorini M, Lucci D, Marchionni N, MariniM, Campana C, Perini G, Deorsola A, Masotti G, Tavazzi L, MaggioniAP. Left bundle-branch block is associated with increased 1-yearsudden and total mortality rate in 5517 outpatients with congestiveheart failure: a report from the Italian network on congestive heartfailure. Am Heart J 2002;143:398–405.

7. Mueller C, Laule-Kilian K, Klima T, Breidthardt T, Hochholzer W,Perruchoud AP, Christ M. Right bundle branch block and long-termmortality in patients with acute congestive heart failure. J Intern Med2006;260:421–428.

8. Zimetbaum PJ, Buxton AE, Batsford W, Fisher JD, Hafley GE, LeeKL, O’Toole MF, Page RL, Reynolds M, Josephson ME. Electrocar-diographic predictors of arrhythmic death and total mortality in themulticenter unsustained tachycardia trial. Circulation 2004;110:766–769.

9. Barsheshet A, Leor J, Goldbourt U, Garty M, Schwartz R, Behar S,Luria D, Eldar M, Glikson M. Effect of bundle branch block patternson mortality in hospitalized patients with heart failure. Am J Cardiol2008;101:1303–1308.

0. Barsheshet A, Garty M, Grossman E, Sandach A, Lewis BS, GottliebS, Shotan A, Behar S, Caspi A, Schwartz R, Tenenbaum A, Leor J.Admission blood glucose level and mortality among hospitalized non-diabetic patients with heart failure. Arch Intern Med 2006;166:1613–1619.

1. Remme WJ, Swedberg K. Comprehensive guidelines for the diagnosisand treatment of chronic heart failure. Task force for the diagnosis andtreatment of chronic heart failure of the European Society of Cardiol-ogy. Eur J Heart Fail 2002;4:11–22.

2. Abdel-Qadir HM, Tu JV, Austin PC, Wang JT, Lee DS. Bundle branchblock patterns and long-term outcomes in heart failure. Int J Cardiol2010. [Epub ahead of print]. DOI:10.1016/j.ijcard.2010.01.012.

3. Petrov DB. Appearance of right bundle branch block in electrocardio-grams of patients with pulmonary embolism as a marker for obstruc-tion of the main pulmonary trunk. J Electrocardiol 2001;34:185–188.

4. Ocal A, Yildirim N, Ozbakir C, Saricam E, Ozdogan OU, Arslan S,Tufekcioglu O, Sabah I. Right bundle branch block: a new parameterrevealing the progression rate of mitral stenosis. Cardiology 2006;105:219–222.

5. Abd El Rahman MY, Abdul-Khaliq H, Vogel M, Alexi-Meskishvili V,Gutberlet M, Lange PE. Relation between right ventricular enlarge-ment, QRS duration, and right ventricular function in patients withtetralogy of Fallot and pulmonary regurgitation after surgical repair.Heart 2000;84:416–420.

6. Robalino BD, Whitlow PL, Underwood DA, Salcedo EE. Electrocar-diographic manifestations of right ventricular infarction. Am Heart J1989;118:138–144.

7. Grines CL, Bashore TM, Boudoulas H, Olson S, Shafer P, Wooley CF.
Functional abnormalities in isolated left bundle branch block. Theeffect of interventricular asynchrony. Circulation 1989;79:845–853.

CMa

M

0d

Characteristics of Depression Remission and Its Relation WithCardiovascular Outcome Among Patients With Chronic Heart

Failure (from the SADHART-CHF Study)

Wei Jiang, MDa,b,*, Ranga Krishnan, MDb, Maragatha Kuchibhatla, PhDc, Michael S. Cuffe, MDa,Carolyn Martsberger, PhDb, Rebekka M. Arias, BSd, and Christopher M. O’Connor, MDa, for the

SADHART-CHF Investigators

Depression is prevalent in patients with heart failure and is associated with a significantincrease in hospitalizations and death. Primary results of the Sertraline Against Depressionand Heart Disease in Chronic Heart Failure (SADHART-CHF) trial revealed that sertra-line and placebo had comparable effects on depression and cardiovascular outcomes. Inthis study, we explored whether remission from depression was associated with bettersurvival and aimed to characterize participants who remitted during the trial. Based ondepression response during the 12-week treatment phase, SADHART-CHF participantswere divided into 2 groups: (1) remission, defined as participants whose last measuredHamilton Depression Rating Scale (HDRS) score was <8, and (2) nonremission, defined asparticipants whose last measured HDRS score was >8. Patients who dropped out beforehaving any repeat HDRS were not included. Baseline characteristics and survival differ-ences up to 5 years were evaluated between the remission and nonremission groups. Of the469 SADHART-CHF participants, 208 (44.3%) achieved remission, 194 (41.4%) remaineddepressed, and 67 (14.3%) dropped out or died without any repeat HDRS assessment.Patients in the remission group had significantly fewer cardiovascular events than those inthe nonremission group (1.34 � 1.86 vs 1.93 � 2.71, adjusted p � 0.01). Men patients weremore likely to remit than women patients (56.5 vs 44.8%, p � 0.02). The remission grouphad milder depressive symptoms at baseline compared to the nonremission group (HDRS17.0 � 5.4 vs 19.6 � 5.5, Beck Depression Inventory scale 17.9 � 6.5 vs 20.3 � 7.2,p <0.001 for the 2 comparisons). In conclusion, this study indicates that remission fromdepression may improve the cardiovascular outcome of patients with heart
failure. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:545–551)
dpssta(sla

Depression is a common and well-documented co-mor-bidity in patients with heart failure (HF).1,2 It is associatedwith substantial morbidity and mortality and with lowerquality of life and functional status.2–9 The adverse relationof depression to HF is independent of HF cause and otherconventional risk factors. Sertraline appeared to be safe indepressed patients with HF10; however, it did not demon-strate any superiority over placebo for depression and HFsurvival in the Sertraline Against Depression and HeartDisease in Chronic Heart Failure (SADHART-CHF) trial.10

Recently, Carney et al11 and Glassman et al12 reported thatpatients with ischemic heart disease whose depression sig-nificantly decreased during the study periods that was notnecessarily related to the trial intervention had better sur-

Departments of aMedicine and bPsychiatry and Behavioral Sciencesand cCenter for Aging, Duke University Medical Center, Durham, North

arolina; dDuke Clinical Research Institute, Durham, North Carolina.anuscript received August 10, 2010; revised manuscript received and

ccepted October 1, 2010.The SADHART-CHF study was funded by the National Institute of

ental Health, Bethesda, Maryland.*Corresponding author: Tel: 919-668-0762; fax: 919-668-5271.

cE-mail address: [email protected] (W. Jiang).


vival than those patients whose depression persisted.Whether this phenomenon applies to patients with HF isunknown. This study therefore aimed to explore if remis-sion is associated with better cardiovascular outcome ofpatients with HF and major depressive disorder and toexamine characteristics that may differentiate patientswhose depression remitted from ones whose depressionremained during the trial.

Methods

The SADHART-CHF database was used for this analy-sis. A detailed method of the SADHART-CHF trial hasbeen previously published.10,13 SADHART-CHF was a ran-omized double-blind study of sertraline versus matchinglacebo in patients with HF and co-morbid major depres-ion. In addition, all participants received nurse-facilitatedupport (NFS). Primary end points of the SADHART-CHFrial were change across time in severity of depressions measured by the Hamilton Depression Rating ScaleHDRS) total score and change in composite cardiovasculartatus.10,13 The protocol was reviewed and approved by theocal institutional review board at each participating center,nd all participants provided written, voluntary, informed
onsent before enrollment.
www.ajconline.org


N

I�

PHHDCPPIIB

B

D

B

D

AH


SADHART-CHF trial patients were randomized 1:1 tosertraline 50 mg/day or matched placebo for a 12-weekacute treatment phase. Study drug dose was titrated up in50-mg/day increments every 2 weeks with a maximum doseof 200 mg/day, depending on severity of depressive symp-toms and tolerability of participants to the study drug. Par-


Characteristics Remis

Sertraline(n � 106)

Age (years), mean � SD 63.0 � 9.9Men 62.3%Race

White 52.8%Black 40.6%Other 6.6%

Left ventricular ejection fraction, mean � SD 31.5 � 8.9ew York Heart Association classII 32.8%III 51.9%IV 16.0%

schemic cause 68.9%1 heart failure hospitalization in previous 12 months 57.6%

revious myocardial infarction 52.8%istory of hypertension 93.4%yperlipidemia 83.0%iabetes mellitus 55.7%urrent smoking 32.1%revious coronary bypass 36.8%acemaker 7.6%mplantable cardioverter–defibrillator 14.2%npatient at enrollment 70.8%aseline medicationsAngiotensin-converting enzyme inhibitor 68.9%Angiotensin receptor blocker 9.4%� Blocker 84.9%Aspirin 86.8%Digoxin 14.2%Statin 72.6%Loop diuretic 60.4%Calcium channel blocker 8.5%aseline Hamilton Depression Rating Scale, mean �

SD (range)16.9 � 5.6 (6–30

epression severity (Hamilton Depression RatingScale score)

�17 31.6%18–22 24.8%�23 18.5%aseline Beck Depression Inventory, mean � SD

(range)18.3 � 6.9 (9–40

epressive symptom dimension (Beck DepressionInventory)

Cognitive/affective 4.4 � 3.7 (0–19Somatic/affective 11.8 � 3.9 (5–26Appetitive 2.1 � 1.7 (0–7)ntidepressant use before enrollment 6.6%istory of major depressive disorder 15.1%

* The p values apply to differences among the 4 groups.† These p values demonstrate that depression severity was significantly

(p � 0.96).

ticipants who were unable to tolerate the 50-mg/day dose

were allowed to remain in the study provided a minimum25-mg/day dose was tolerated. The study medication wastapered off after the 12-week acute treatment phase. Allparticipants, regardless of acute-phase completion, enteredthe long-term follow-up phase and were contacted at 6months 12 months, and annually thereafter to evaluate clin-

roup Nonremission Group p Value*

Placebo(n � 102)

Sertraline(n � 90)

Placebo(n � 104)

2 � 9.8 61.9 � 10.5 62.0 � 11.3 0.5266.7% 50.0% 55.8% 0.09

0.7155.9% 56.7% 60.6%35.3% 40.0% 34.6%

8.9% 3.3% 4.8%7 � 10.2 32.2 � 9.7 29.8 � 10.1 0.20

0.5627.5% 32.2% 27.9%51.1% 43.3% 45.2%21.6% 24.4% 26.9%67.7% 68.9% 69.2% 0.9962.4% 52.2% 72.1% 0.0344.1% 51.1% 43.3% 0.4183.3% 87.8% 83.7% 0.1073.5% 71.1% 82.7% 0.0945.1% 51.1% 47.1% 0.4324.5% 24.4% 28.9% 0.8231.4% 28.9% 32.7% 0.693.9% 8.9% 5.8% 0.52

13.7% 22.2% 22.1% 0.2079.4% 72.2% 79.8% 0.29

76.5% 72.2% 68.3% 0.549.8% 10.0% 5.7% 0.67

84.3% 91.1% 83.7% 0.4380.4% 81.1% 84.6% 0.5718.6% 15.6% 18.3% 0.7965.7% 71.1% 69.2% 0.7359.8% 61.1% 61.5% 0.994.9% 13.3% 10.6% 0.22

1 � 5.2 (7–30) 19.6 � 5.8 (5–35) 19.5 � 5.2 (7–33) 0.0002

�0.0001/0.96†

30.5% 16.4% 21.5%24.1% 25.6% 25.6%17.4% 29.4% 34.8%

6 � 6.2 (8–34) 21.3 � 7.2 (9–41) 19.5 � 5.3 (7–40) 0.0009

0 � 3.4 (0–21) 5.5 � 4.0 (0–16) 4.6 � 4.1 (0–19) 0.075 � 3.7 (3–21) 13.7 � 4.5 (4–28) 12.4 � 4.0 (4–24) 0.0030 � 1.8 (0–6) 2.1 � 1.6 (0–6) 2.4 � 1.9 (0–7) 0.28

4.9% 8.9% 8.0% 0.738.8% 22.2% 16.4% 0.08

ted with remission status (p �0.0001) but not with treatment assignment

sion G

61.

29.

) 17.

) 17.

) 4.) 11.

2.

associa

ical events and vital status. The long-term follow-up phase

cssbtt

547Heart Failure/Depression Remission and Heart Failure Prognosis

continued until the last enrolled participant completed a6-month follow-up.

All participants received NFS as a mechanism to ascer-tain safety and study compliance. Primary goals for theresearch personnel applying supportive measures were toincrease recruitment and assessment of participants, to as-certain safety of participants, and to increase participantcompliance and retention. Such support was provided bynurses and other study personnel with experience or trainingin clinical psychiatry and supervised by the study psychia-trist. Supportive measures included �10 hours of activeinteraction between research personnel and study partici-pants during the 12-week acute phase. The interaction con-sisted of 3 face-to-face visits (1 during recruitment/baselineassessment followed by 2 visits conducted primarily inparticipants’ homes) and 4 follow-up telephone contacts.Research personnel aimed to provide psychological supportand conduct medical and psychiatric health evaluations.Personnel were instructed to provide active and empatheticlistening and validation skills and soothing and other emo-tional support strategies. Research personnel were asked notto push for dialog with participants who were more reservedand aimed to establish an individual interpersonal relation-ship with each participant.

For the present study, SADHART-CHF participantswere divided into 2 groups, irrespective of treatment assign-ment: (1) remission, defined as participants whose last mea-sured HDRS score was �8, and (2) nonremission, definedas participants whose last measured HDRS score was �8.Patients who dropped out before having any repeat HDRS

Table 2Performance of each group during 12-week acute treatment phase

Characteristics Remission Grou

(n � 208)

Sertraline

Treatment assignment 26.4%Days in 12-week intervention 77.4 � 19.5 (19–119) 78.8 �Dosing 64.1 � 30.7 (50–150) 65.8 �Last Hamilton Depression Rating Scale

score, mean � SD (range)3.8 � 2.1 (0–7) 3.5 �

Last Beck Depression Inventory score,mean � SD (range)

4.8 � 3.2 (0–16) 4.9 �

Table 3Reasons for termination during 12-week treatment phase

Reasons forTermination DuringAcute Phase

Remission Group(n � 208, 22.5%)

NonremissionGroup

(n � 194, 33.5%)

pValue

Patient withdrew 1 (0.5%) 4 (2.1%) 0.15Patient lost to follow-up 4 (1.9%) 7 (3.6%) 0.30Side effect 13 (6.3%) 16 (8.3%) 0.44Withdraw by study

physician5 (2.4%) 14 (7.2%) �0.02

Noncompliance 19 (9.1%) 13 (6.7%) 0.37Death 3 (1.44%) 5 (2.58%) 0.42Other 2 (0.9%) 6 (3.1%) 0.13

were not included in this study. w

Primary end points of this analysis were survival or timeto death and rate of recurrent cardiovascular events and/ordeath until last follow-up. Cardiovascular events were ad-judicated by a blinded clinical events committee as a com-ponent of the primary trial.10 Clinical characteristics exam-ined among these groups included effects of randomization,dosing of study medication, and depressive symptomatol-ogy measured by the HDRS and the Beck Depression In-ventory (BDI) scale. Baseline depressive symptomatologywas examined in 3 ways (1) severity of depression based onHDRS scores (mild �17, moderate 18 to 22, severe �23),(2) overall HDRS and BDI scores, and (3) depressive symp-tom dimension, i.e., cognitive/affective (items 2, 3, 5, 6, 7,8, 9, 12, and 14), somatic/affective (items 1, 4, 10, 11, 13,15, 16, 17, 21, and 22), and appetitive (items 18 and 19/20)based on the BDI scale.14 Comparative analysis of baselineharacteristics was performed on remission status with re-pect to treatment assignment. Change in HDRS by remis-ion status over the 12-week treatment period was analyzedy random coefficient models. The final model includedreatment, remission status, time in weeks, and the square ofime and the interaction of each of these 2 time variables

Figure 1. Survival by patients in the depression remission group assignedto placebo (blue line), patients with remission assigned to sertraline (redline), patients in the nonremission group who were assigned to placebo(green line), and patients in the nonremission group assigned to sertraline(black line). Unadjusted hazard ratio for the nonremission versus remissiongroups was 1.23 (95% confidence interval 0.95 to 1.59).

Nonremission Group p Value

(n � 194)

o Sertraline Placebo

22.4% 25.9%1–110) 73.5 � 21.8 (13–104) 71.8 � 24.1 (11–108) 0.06150–200) 112.2 � 57.2 (50–200) 108.7 � 55.5 (50–200) �0.0001–7) 13.9 � 5.4 (8–31) 13.5 � 4.7 (8–27) 0.0002

–19) 13.6 � 7.8 (3–41) 12.5 � 7.2 (2–36) 0.0009

p

Placeb

25.4%18.9 (29.1 (2.1 (0

3.6 (0

ith remission status and study site. The model also in-

salU

R

CSTo2�e(1ttpH

rl

tepeT

(t

e remis


cluded the random effects of patient, patient-by-time inter-action, and square of patient-by-time interaction. All partic-ipants (remission and nonremission groups) were includedin the analysis. Cox proportional regression modeling wasused to evaluate survival differences between participantsclassified in the remission and nonremission groups. Datawere censored at time to death. Logistic regression analysiswas used to test differences of cardiovascular events be-tween groups. Baseline HDRS scores, treatment assignment(sertraline or placebo), age, gender, baseline left ventricularejection fraction, New York Heart Association (NYHA)class, ischemic cause, and study site were included in theregression model. Assumptions of the model were assessedusing standard techniques. Logistic regression was alsoused to determine which subsets of patients were likely toremit over the 12-week treatment phase. Investigatedcharacteristics were gender, baseline HDRS scores (�17vs �17), BDI scores (�16 vs �16), and somatic/affectivecores (lower vs higher than or equal to mean). Statisticalnalyses using SAS 9.1 (SAS Institute, Cary, North Caro-ina) were performed by statistical personnel within Dukeniversity Medical Center (Durham, North Carolina).

esults

In total 469 participants were enrolled in SADHART-HF at Duke University Medical Center and 3 Duke Healthystem affiliates from August 13, 2003 to March 3, 2008.he primary analysis of SADHART-CHF has been previ-usly reported.13 Of the 469 SADHART-CHF participants,08 patients (44.3%) had HDRS scores that improved to8, whereas 194 (41.4%) patients had a HDRS �8 at the

nd of 12-week intervention. There were 67 subjects14.3%) who did not have any repeat HDRS during the2-week acute phase intervention and were excluded fromhis study. These participants had higher NYHA class HFhan the other 2 groups (NYHA class II 17.9 vs 29.9,

�0.05). Ten patients (14.9%) died before any repeatDRS assessment.Most baseline characteristics were similar between the

emission and nonremission groups with respect to sertra-

Table 4Number of cardiovascular events per participant among groups

Cardiovascular Events Remis

(n

Sertraline

Any nonfatal cardiovascular event* 1.04 � 1.73Acute myocardial infarction 0.04 � 0.19Exacerbation of heart failure 0.42 � 1.05Unstable angina pectoris 0.11 � 0.48Arrhythmia 0.05 � 0.25Syncope 0.04 � 0.19Stroke 0.05 � 0.25Transient ischemic attack 0Other 0.36 � 0.81Any cardiovascular event and/or death† 1.27 � 1.81

* Adjusted p � 0.02; † adjusted p � 0.01 for comparisons between th

ine or placebo assignment (Table 1). More men remitted

han women (64.4 vs 53.1%, p � 0.02), although differ-nces between groups were not statistically significant. Hos-italization due to HF exacerbation within 1 year beforenrollment was higher in the nonremission group (p � 0.03;able 1).

The remission group had longer treatment duration78 � 19 days) during the 12-week acute treatment phasehan the nonremission group (73 � 23, p �0.001; Table 2).

The interaction of remission by time was significant (p �0.001) but not remission by square of time (p � 0.19).Change in HDRS total score over the course of the 12-weekacute phase (i.e., time) intervention and square of time weresignificant (time p � 0.001, square of time p � 0.001). Thedifference in the change of HDRS scores between the re-mission and nonremission groups was statistically signifi-cant (mean � SE �4.8 � 0.49, p � 0.001). Although theyremained depressed, symptoms of patients in the nonremis-sion group improved notably from baseline (mean � SE�4.1 � 0.33, p �0.001). In the nonremission group, 40remained significantly depressed (HDRS total score �17).There were no differences in these measurements betweentreatment assignments. Doses of sertraline and placebo weresignificantly different between the remission and nonremis-sion groups, but there was no difference within treatmentassignment (Table 2). One hundred twelve of the 402 par-ticipants (27.9%) dropped out during the 12-week treatmentphase after providing �1 repeat measurement of depressivesymptoms. Reasons for dropping out during the acute treat-ment phase are listed in Table 3.

Average length of follow-up was 798 � 493 days(range 1 to 1,832) for the entire study population. Pa-tients whose depression remitted had longer survival thanthose whose depression remained (866 � 479 vs 793 �483 days); however, the result of the Cox proportionalregression analysis revealed that the difference of sur-vival between the 2 groups was not statistically signifi-cant (hazard ratio 1.23 for nonremission vs remissiongroup, 95% confidence interval 0.95 to 1.59). Kaplan-Meier survival curves for the remission and nonremissiongroups with respect to treatment assignment are shown in

roup Nonremission Group

) (n � 194)

Placebo Sertraline Placebo

1.18 � 1.86 1.82 � 3.08 1.52 � 2.280.07 � 0.29 0.07 � 0.29 0.02 � 0.140.57 � 1.40 0.88 � 2.02 0.70 � 1.340.07 � 0.39 0.20 � 0.62 0.18 � 0.630.11 � 0.37 0.20 � 0.71 0.14 � 0.53

0 0.01 � 0.11 0.05 � 0.210.03 � 0.22 0.06 � 0.23 0.04 � 0.21

0 0.02 � 0.15 00.32 � 0.64 0.40 � 0.96 0.42 � 0.831.40 � 1.90 2.1 � 3.13 1.79 � 2.3

sion and nonremission groups.

sion G

� 208

Figure 1.

ev

tt2awswrhrii

D

aanhse

SrCsvtdpa

tadoci(

pw

pOscrcrtmppdcppadsftm


There were 606 cardiovascular events in the entire studypopulation including deaths, with 230 in the remissiongroup and 323 in the nonremission group. Table 4 present asummary of differences of various cardiovascular eventsamong the 4 groups. Patients whose depression remitted hada smaller number of overall nonfatal cardiovascular eventsper participant and fatal and nonfatal combined cardiovas-cular events (mean � SD 1.11 � 1.79 and 1.34 � 1.86)compared to those whose depression remained (mean � SD1.66 � 2.68 and 1.92 � 2.71). Differences were statisticallysignificant and remained so after covariate analysis withage, gender, baseline ejection fraction, NYHA classes, isch-emic cause of HF, baseline HDRS scores, and treatmentassignment (p �0.05 for all comparisons). There was noassociation of baseline depressive symptoms and cardiovas-cular events (p � 0.82 for overall nonfatal cardiovascularvents, p � 0.88 for fatal and nonfatal combined cardio-ascular events).

Baseline HDRS and self-rated BDI scores were lower inhe remission group (HDRS 17.0 � 5.4, BDI 17.9 � 6.5)han in the nonremission group (HDRS 19.6 � 5.5, BDI0.3 � 7.2, p �0.001 for the 2 comparisons). Furthernalysis of depressive symptom dimension revealed that itas the dimension of the somatic/affective symptoms mea-

ured by BDI at baseline (Table 1) that separated patientshose depression remitted from patients whose depression

emained (p � 0.003). Although the nonremission groupad higher baseline cognitive/affective symptoms than theemission group, the difference was not statistically signif-cant (p � 0.07; Table 1). Odds ratios for the 4 character-stics of remission are presented in Table 5.

iscussion

This analysis of the SADHART-CHF trial demonstratedn association of depression remission during the 12-weekctive intervention and significantly decreased fatal andonfatal cardiovascular events. Such findings support theypothesis that successful depression treatment may be as-ociated with a cardiac benefit in patients with HF. Carneyt al11 performed a post hoc analysis in the Enhancing

Recovery in Coronary Heart Disease (ENRICHD) study to

Table 5Logistic analysis assessing baseline characteristics that best predictremission

Characteristics OddsRatio

95% ConfidenceInterval

Male gender 1.57 1.055–2.345Male gender � Hamilton Depression

Rating Scale score �172.48 1.544–3.973

Male gender � Beck Depression Inventoryscore �16

2.27 1.441–3.554

Male gender � somatic/affective score nohigher than mean

2.01 1.324–3.042

Male gender � Hamilton DepressionRating Scale score �17 �somatic/affective score no higher thanmean

2.68 1.583–4.524

examine whether depression remission at 6 months from

baseline was associated with survival. They found that pa-tients who were randomized into the intervention arm andremained depressed at the 6-month follow-up had signifi-cantly worse survival than those patients whose depressionremitted, although there was no survival difference betweenthe intervention and control groups. Furthermore, Glassmanet al12 examined a 7-year survival difference between the

ADHART trial participants whose depression significantlyemitted and those whose depression remained based on thelinical Global Impression-Improvement (CGI) subscale

core during the 24-week sertraline or placebo trial inter-ention. The investigators demonstrated that depressed pa-ients with ischemic heart disease whose CGI score wasecreased to 1 or 2 had significant higher survival than thoseatients whose CGI score was �2 irrespective of treatmentssignment.

The SADHART-CHF study design followed the tradi-ional phase II to III clinical trial model of randomizationnd placebo control, and it focused primarily on examiningifferences between active treatment assignments, i.e., drugr psychotherapy versus placebo or usual care. Sertralineompared to placebo did not result in a statistically signif-cant higher rate of remission in the SADHART-CHF study54.1% vs 49.5%, p � 0.36).10 Several factors may have

contributed to this finding including the placebo effect; thetherapeutic impact of NFS, or the uniqueness of co-morbiddepression in patients with HF. Therapeutic response toplacebo has been recognized in research and in clinicalpractice.15,16 Several meta-analyses have demonstrated alacebo response �50% in depression trials, a finding thatas particularly evident among nonpublished trials.17–19 In

recent years, an increase in scientific attention to the placeboeffect has yielded evidence that the effect may have aneurobiological foundation.20,21 However, the potential im-act of NFS on treatment response cannot be overlooked.ne major predisposing factor for depression is a weak

ocial support network. Persistent and negative life stressorsoupled with limited supportive structures are believed toesult in a decrease of mental function in patients withhronic illnesses. The NFS-fostered relationship may haveeplaced ineffective or insufficient social networks and es-ablished an alliance that contributed to functional improve-ent.22–24 Specific evidence to support the therapeutic im-

act of NFS includes maintenance of remission in thesearticipants after 4 to 6 weeks when placebo effects tend toecrease.25 Most of the study population was naive to psy-hological and psychiatric interventions; therefore, thesearticipants may have been innately more responsive tosychosocial supportive measures. In addition, the fact thatlarge proportion of study participants had a mild baselineepression in severity (i.e., HDRS total score 8 to 17) attudy entry may have contributed to the high remission rateor those participants. Other studies of this type have failedo demonstrate survival or prognostic benefit of active treat-ent compared to controls.11,26,27 Trials that failed to show

treatment benefit of pharmacologic agents hypothesized thatthe lack of a statistically significant difference in outcomeswas due to insufficient power.27–30

SADHART-CHF had a similar design to other studies inpatients with heart disease and co-morbid depression. In
SADHART-CHF, the sample size was believed to be ade-

paiapEd

hpwltdsftrarsnwpcsTdwiutt

ctrtovgadmm

1

1


quate to evaluate differences in depression and clinicaloutcome; however, there was no benefit observed with an-tidepressant treatment over placebo.10 In contrast, in theENRICHD trial, open-label treatment of selective serotoninreuptake inhibitors showed a statistically significant 40%decrease in death or nonfatal myocardial infarction, with acrude hazard of 0.61 (95% confidence interval 0.41 to 0.90)and an adjusted hazard ratio of 0.53 (95% confidence inter-val 0.38 to 0.84).28 In this study, because antidepressantrescription was at the discretion of the study physiciansnd not randomized or controlled, the impact of the findings controversial. Whether or not a survival benefit wasssociated with depression remission was not reported in therimary analysis. However, in the subgroup analysis forNRICHD, successful treatment of depression appeared toecrease the risk of cardiovascular events.11

The present findings of the SADHART-CHF studiesraise several important questions including the need to de-termine whose depression is easier to remit versus thosewho are more resistant to interventions, to examine whethersubsets of depressed patients with HF may be at higher riskfor cardiovascular events, and to identify effective antide-pressant(s) for this particular population. Our research in-dicated that men with HF and mild depression (HDRS �17)ave nearly a 2.5-fold chance for depression remission com-ared to the rest of the study population. In contrast, patientsho had greater somatic affective symptoms were less

ikely to have remission. These data suggest that future trialsesting therapeutics may need to focus on patients whoseepression is more severe and do not respond to generalupportive measures. This information also may be usefulor clinical practice. Nevertheless, future studies are neededo confirm the observed association between depressionemission and cardiovascular outcomes. NFS appears to bepowerful therapeutic technique that results in depression

emission in certain patients with HF. A more sophisticatedtudy design and refinement of support measures will beeeded to fully evaluate the effects of an NFS in patientsith HF or other medical conditions. Cause of depression inatients with cardiac disease or other long-term medicalonditions may be more heterogenous than in populationstudied in traditional trials that test antidepressant efficacy.herefore, outcome studies testing the association betweenepression treatment and prognosis may need to be alignedith real-world clinical practice techniques such as employ-

ng a naturalistic approach in the acclimatization of individ-al differences and identifying characteristics of popula-ions who respond differently to various antidepressivereatments.

Results of this analysis should be interpreted in theontext of limitations. This was a secondary analysis of arial, which was originally designed to assess treatmentesponse to sertraline versus placebo, examining subgroupshat are defined by changes in measurements over the coursef a trial. Therefore, any changes in depression and cardio-ascular outcomes between the remission and nonremissionroups may be considered due to pre-existing baseline char-cteristic differences between those who do and those whoo not respond to sertraline or other interventional measure-ents. Whether these participants received depression treat-
ent after the 12-week acute phase of trial was not further
evaluated, similar to the studies of Carney et al11 and Glass-man et al.12 Although the analysis was adjusted for differ-ences in baseline variables and baseline depression severitywas not associated with the prognosis, the fact that partic-ipants with remission remitted had a baseline lower depres-sive score, especially by the self-administered test, mighthave reflected a subpopulation who had perceived theirillness milder and/or were more receptive to psycho-sup-portive intervention. In addition, this study had limitedstatistical power to evaluate survival.

Acknowledgment: We thank Pfizer, Inc., New York, NewYork, for supplying the study drug.

1. Pelle AJM, Gidron YY, Szabo BM, Denollet J. Psychological predic-tors of prognosis in chronic heart failure. J Card Fail 2008;14:341–350.

2. Rutledge T, Reis VA, Linke SE, Greenberg BH, Mills PJ. Depressionin heart failure- A meta-analytic review of prevalence, interventioneffects, and associations with clinical outcomes. J Am Coll Cardiol2006;48:1527–1537.

3. Jiang W, Kuchibhatla M, Clary GL, Cuffe MS, Christopher EJ, Alex-ander JD, Califf RM, Krishnan RR, O’Connor CM. Congestive heartfailure—relationship between depressive symptoms and long-termmortality in patients with heart failure. Am Heart J 2007;154:102–108.

4. Johansson P, Dahlstrom U, Alehagen U. Depressive symptoms andsix-year cardiovascular mortality in elderly patients with and withoutheart failure. Scand Cardiovasc J 2007;41:299–307.

5. Muller-Tasch T, Peters-Klimm F, Schellberg D, Holzapfel N, Barth A,Junger J, Szecsenyi J, Herzog W. Depression is a major determinant ofquality of life in patients with chronic systolic heart failure in generalpractice. J Card Fail 2007;13:818–824.

6. Penninx B, Beekman ATF, Honig A, Deeg DJH, Schoevers RA, vanEijk JTM, van Tilburg W. Depression and cardiac mortality—resultsfrom a community-based longitudinal study. Arch Gen Psychiatry2001;58:221–227.

7. Rozzini R, Sabatini T, Frisoni GB, Trabucchi M. Depression andmajor outcomes in older patients with heart failure. Arch Intern Med2002;162:362–363.

8. Rumsfeld JS, Havranek E, Masoudi FA, Peterson ED, Jones P, TooleyJF, Krumholz HM, Spertus JA, Cardiovascular Outcomes Res C.Depressive symptoms are the strongest predictors of short-term de-clines in health status in patients with heart failure. J Am Coll Cardiol2003;42:1811–1817.

9. Vaccarino V, Kasl SV, Abramson J, Krumholz HR. Depressive symp-toms and risk of functional decline and death in patients with heartfailure. J Am Coll Cardiol 2001;38:199–205.

0. O’Connor CM, Jiang W, Kuchibhatla M, Silva S, Cuffe M, CallwoodD, Zakhary B, Stough W, Arias R, Rivelli S, Krishnan R, for theSADHART-CHF Investigators. Safety and efficacy of sertraline fordepression in patients with heart failure: results of the SADHART-CHF trial. J Am Coll Cardiol 2010;56:692–699.

1. Carney RM, Blumenthal JA, Freedland KE, Youngblood M, Veith RC,Burg MM, Cornell C, Saab PG, Kaufmann PG, Czajkowski SM, JaffeAS, for the ENRICHD Investigators. Depression and late mortalityafter myocardial infarction in the Enhancing Recovery in CoronaryHeart Disease (ENRICHD) study. Psychosom Med 2004;66:466–474.

12. Glassman AH, Bigger JT, Gaffney M. Psychiatric characteristics as-sociated with Long-term mortality among 361 patients having an acutecoronary syndrome and Major depression seven-year follow-up ofSADHART participants. Arch Gen Psychiatry 2009;66:1022–1029.

13. Jiang W, O’Connor C, Silva SG, Kuchibhatla M, Cuffe MS, CallwoodDD, Zakhary B, Henke E, Arias RM, Krishnan R, for the SADHART-CHF Investigators. Safety and efficacy of sertraline for depression inpatients with CHF (SADHART-CHF): a randomized, double-blind,placebo-controlled trial of sertraline for major depression with con-gestive heart failure. Am Heart J 2008;156:437–444.

14. Linke SE, Rutledge T, Johnson BD, Vaccarino V, Bittner V, Cor-nell CE, Eteiba W, Sheps DS, Krantz DS, Parashar S, Merz CNB.
Depressive symptom dimensions and cardiovascular prognosis

1

1

2

2

2

2

2

2

2

2

2

2

3


among women with suspected myocardial ischemia. A report fromthe National Heart, Lung, and Blood Institute-sponsored women’sischemia syndrome evaluation. Arch Gen Psychiatry 2009;66:499 –507.

15. Katz J. The Silent World of Doctor and Patient. New York: Free Press,1984.

16. Shapiro A, Shapiro E. The placebo: is it much ado about nothing? In:Harrington A, ed. The Placebo Effect: an Interdisciplinary Explora-tion, 1st Ed. Cambridge, MA: Harvard University Press, 1997:12–36.

17. Kirsch I. Listening to Prozac but hearing placebo: a meta-analysis ofantidepressant medication. Prev Treat 1998;1:0002a.

8. Kirsch I, Scoboria A, Nicholls S. The emperor’s new drugs: an anal-ysis of antidepressant medication data submitted to the U.S. Food andDrug Administration. Prev Treat 2002;5:23.

9. Khan A, Detke M, Khan SRF, Mallinckrodt C. Placebo response andantidepressant clinical trial outcome. J Nerv Ment Dis 2003;191:211–218.

0. Leuchter AF, Cook IA, Witte EA, Morgan M, Abrams M. Changes inbrain function of depressed subjects during treatment with placebo.Am J Psychiatry 2002;159:122–129.

1. Mayberg HS, Silva JA, Brannan SK, Tekell JL, Mahurin RK,McGinnis S, Jerabek PA. The functional neuroanatomy of the placeboeffect. Am J Psychiatry 2002;159:728–737.

2. Ong LML, Dehaes J, Hoos AM, Lammes FB. Doctor-patient commu-nication—a review of the literature. Soc Sci Med Soc Sci Med 1995;40:903–918.

3. Stewart MA. Effective physician-patient communication and healthoutcomes—a review. CMAJ 1995;152:1423–1433.

4. Wolfaardt UB, Reddon JR, Joyce AS. Assessing the efficacy of anti-depressants: the transactional paradigm. Med Hypotheses 2005;64:1229–1236.

5. Walsh BT, Seidman SN, Sysko R, Gould M. Placebo response instudies of major depression—variable, substantial, and growing.JAMA 2002;287:1840–1847.

6. Frasure-Smith N. The Montreal Heart Attack Readjustment Trial.J Cardiopulm Rehabil 1995;15:103–106.

7. Glassman AH, O’Connor CM, Califf RM, Swedberg K, Schwartz P,Bigger JT, Krishnan KRR, van Zyl LT, Swenson JR, Finkel MS,Landau C, Shapiro PA, Pepine CJ, Mardekian J, Harrison WM. Grp S.Sertraline treatment of major depression in patients with acute MI orunstable angina. JAMA 2002;288:701–709.

8. Berkman LF, Blumenthal J, Burg M, Carney RM, Catellier D, CowanMJ, Czajkowski SM, DeBusk R, Hosking J, Jaffe A, Kaufmann PG,Mitchell P, Norman J, Powell LH, Raczynski JM, Schneiderman N,ENRICHD Investigators. Effects of treating depression and low-perceived social support on clinical events after myocardial infarc-tion—the Enhancing Recovery in Coronary Heart Disease patients(ENRICHD) randomized trial. JAMA 2003;289:3106 –3116.

9. Taylor CB, Youngblood ME, Catellier D, Veith RC, Carney RM, BurgMM, Kaufmann PG, Shuster J, Mellman T, Blumenthal JA, KrishnanR, Jaffe AS. Investigators E. Effects of antidepressant medication ofmorbidity and mortality in depressed patients after myocardial infarc-tion. Arch Gen Psychiatry 2005;62:792–798.

0. Van Melle JP, De Jonge P, Honig A, Schene AH, Kuyper AMG, CrijnsH, Schins A, Tulner D, Van den Berg MP, Ormel J, Investigators M-I.
Effects of antidepressant treatment following myocardial infarction.Br J Psychiatry 2007;190:460–466.

lli

F

l

CLMa

fHo(B

0d

Warfarin Use and Outcomes in Patients With Advanced ChronicSystolic Heart Failure Without Atrial Fibrillation, Prior

Thromboembolic Events, or Prosthetic Valves

Marjan Mujib, MD, MPHa, Abu-Ahmed Z. Rahman, MDb, Ravi V. Desai, MDc,Mustafa I. Ahmed, MDa, Margaret A. Feller, MPHa, Inmaculada Aban, PhDa,

Thomas E. Love, PhDd, Michel White, MDe, Prakash Deedwania, MDf, Wilbert S. Aronow, MDg,Gregg Fonarow, MDh, and Ali Ahmed, MD, MPHa,i,*

Warfarin is often used in patients with systolic heart failure (HF) to prevent adverseoutcomes. However, its long-term effect remains controversial. The objective of this studywas to determine the association of warfarin use and outcomes in patients with advancedchronic systolic HF without atrial fibrillation (AF), previous thromboembolic events, orprosthetic valves. Of the 2,708 BEST patients, 1,642 were free of AF without a history ofthromboembolic events and without prosthetic valves at baseline. Of these, 471 patients(29%) were receiving warfarin. Propensity scores for warfarin use were estimated for eachpatient and were used to assemble a matched cohort of 354 pairs of patients with andwithout warfarin use who were balanced on 62 baseline characteristics. Kaplan-Meier and Coxregression analyses were used to estimate the association between warfarin use and outcomesduring 4.5 years of follow-up. Matched participants had a mean age � SD of 57 � 13 yearswith 24% women and 24% African-Americans. All-cause mortality occurred in 30% ofmatched patients in the 2 groups receiving and not receiving warfarin (hazard ratio 0.86,95% confidence interval 0.62 to 1.19, p � 0.361). Warfarin use was not associated withcardiovascular mortality (hazard ratio 0.97, 95% confidence interval 0.68 to 1.38, p �0.855), or HF hospitalization (hazard ratio 1.09, 95% confidence interval 0.82 to 1.44, p �0.568). In conclusion, in patients with chronic advanced systolic HF without AF or otherrecommended indications for anticoagulation, prevalence of warfarin use was high. How-ever, despite a therapeutic international normalized ratio in those receiving warfarin, itsuse had no significant intrinsic association with mortality and hospitalization. Published
by Elsevier Inc. (Am J Cardiol 2011;107:552–557)
iefottwce

M

odr

bae1phL

Heart failure (HF) is a hypercoagulable state, and pa-tients with HF and low left ventricular ejection fraction(LVEF) may be at increased risk of LV thrombus formationand thromboembolic events.1–3 Although use of anticoagu-ants is recommended in patients with HF and atrial fibril-ation (AF) and/or a previous thromboembolic event,4 theres conflicting evidence of the benefit of anticoagulation use

aUniversity of Alabama at Birmingham, Birmingham, Alabama; bCapeear Valley Hospital, Fayetteville, North Carolina; cLehigh Valley Hospi-

tal, Allentown, Pennsylvania; dCase Western Reserve University, Cleve-and, Ohio; eMontreal Heart Institute, Montreal, Quebec, Canada; fUniver-

sity of California at San Francisco, Fresno, California; gNew York Medicalollege, Valhalla, New York; hUniversity of California at Los Angeles,os Angeles, California; iVA Medical Center, Birmingham, Alabama.anuscript received September 21, 2010; revised manuscript received and

ccepted October 5, 2010.Dr. Ahmed is supported by Grants R01-HL085561 and R01-HL097047

rom the National Heart, Lung, and Blood Institute/National Institutes ofealth, Bethesda, Maryland, and a generous gift from Ms. Jean B. Morrisf Birmingham, Alabama. The Beta-Blocker Evaluation of Survival TrialBEST) is conducted and supported by the National Heart, Lung, andlood Institute in collaboration with the BEST study investigators.

tE-mail address: [email protected] (A. Ahmed).
002-9149/11/$ – see front matter Published by Elsevier Inc.oi:10.1016/j.amjcard.2010.10.012

n patients with HF without AF and/or previous thrombo-mbolic events.5–9 However, because risk of LV thrombusormation increases with decreasing LVEF, clinicians areften concerned about the risk of LV thrombus formation inheir patients with HF and markedly low LVEF. The objec-ive of the present study was to determine the association ofarfarin use and outcomes in patients with advanced

hronic systolic HF without AF and/or previous thrombo-mbolic events.

ethods

We conducted a post hoc analysis of the public-use copyf the Beta-Blocker Evaluation of Survival Trial (BEST)ata for the present study. The BEST was a multicenterandomized placebo-controlled clinical trial of bucindolol, a

� blocker, in HF, the methods and results of which haveeen previously published.10 Briefly, 2,708 patients withdvanced chronic systolic HF were enrolled from 90 differ-nt sites across the United States and Canada from May995 to December 1998. All but 1 patient consented to beart of the public-use copy of the data. At baseline, patientsad a mean duration of 49 months of HF and had a meanVEF of 23%. All patients had New York Heart Associa-

ion (NYHA) class III to IV symptoms and �90% of all

www.ajconline.org


553Warfarin and Outcomes in Chronic Systolic Heart Failure

patients were receiving angiotensin-converting enzyme in-hibitors, diuretics, and digitalis.

Data on use of warfarin at baseline were available in all2,707 participants. For the present analysis, we excluded692 patients with AF, 343 patients with a history of throm-boembolic diseases, and 30 patients with prosthetic valvesat baseline. Thus, our final sample was 1,642, of which 471

Table 1Baseline patient characteristics by use of warfarin before and after propen

Variable Before Prop

No warfarin(n � 1,171)

W(

Age (years) 60 � 12Women 283 (24%) 1African-American 302 (26%) 1Current smoking 213 (18%)Body mass index (kg/m2) 37 � 9New York Heart Association class III 1,086 (93%) 4Medical history

Heart failure duration (months) 46 � 48Coronary artery disease 670 (57%) 2Angina pectoris 624 (53%) 2Hypertension 715 (61%) 2Diabetes mellitus 452 (39%) 1Hyperlipidemia 527 (45%) 2Ventricular fibrillation 79 (7%)Peripheral vascular disease 164 (14%)

MedicationsBucindolol 600 (51%) 2Angiotensin-converting enzyme inhibitors/

angiotensin receptor blocker1,137 (97%) 4

Digitalis 1,056 (90%) 4Diuretics 1,086 (93%) 4Vasodilators 504 (43%) 2Aspirin 738 (63%)Statins 273 (23%) 1

Physical examinationPulse (beats/min) 82 � 13Systolic blood pressure (mm Hg) 119 � 19 1Diastolic blood pressure (mm Hg) 72 � 11Jugular venous distention 399 (42%) 8S3 gallop 477 (41%) 2Pulmonary rales 162 (14%)Hepatomegaly 115 (10%)Edema 288 (25%)

Laboratory dataHemoglobin (g/dl) 13.9 � 1.6 14Serum creatinine (mg/dl) 1.2 � 0.4 1Serum potassium (mEq/L) 4.34 � 0.46 4.Plasma norepinephrine (pg/ml) 484 � 272 5Partial thromboplastin time (seconds) 28 � 8International normalized ratio 1.1 � 0.3 2

Left bundle branch block byelectrocardiogram

303 (26%) 1

Cardiothoracic ratio by chest x-ray 54.8 � 7.2 55Pulmonary edema by chest x-ray 114 (10%) 5Left ventricular ejection fraction by nuclear

scan (%)23.5 � 7.2 21

Right ventricular ejection fraction bynuclear scan (%)

35.5 � 11.7 33

Values presented as number of patients (percentages) or mean � SD.

patients (29%) were receiving warfarin at baseline. Consid-

ering the significant imbalances in baseline characteristicsbetween the 2 groups (Table 1), we used propensity scoresto assemble a matched cohort of 354 pairs of patients whowere well balanced on 62 baseline characteristics.11–17 Pro-pensity scores for warfarin use were estimated for each ofthe 1,642 patients using a nonparsimonious multivariablelogistic regression model.18,19 Absolute standardized differ-

tching

atching After Propensity Matching

n1)

p Value No warfarin(n � 354)

Warfarin(n � 354)

p Value

�0.001 57 � 14 57 � 12 0.353) 0.532 87 (25%) 83 (23%) 0.781) 0.226 84 (24%) 87 (25%) 0.857) 0.406 73 (21%) 68 (19%) 0.709

0.643 37 � 9 37 � 8 0.446) 0.255 322 (91%) 327 (92%) 0.583

0.505 45 � 47 45 � 45 0.939) 0.724 190 (54%) 192 (54%) 0.937) 0.312 172 (49%) 178 (50%) 0.708) �0.001 191 (54%) 191 (54%) 1.000) 0.038 124 (35%) 118 (33%) 0.693) 0.756 145 (41%) 149 (42%) 0.825) 0.002 31 (9%) 33 (9%) 0.896) 0.186 53 (15%) 52 (15%) 1.000

) 0.337 168 (48%) 172 (49%) 0.821) 0.875 340 (96%) 345 (98%) 0.425

) 0.001 333 (94%) 334 (94%) 1.000) 0.573 326 (92%) 327 (92%) 1.000) 0.920 146 (41%) 150 (42%) 0.818) �0.001 91 (26%) 96 (27%) 0.640) 0.511 87 (25%) 83 (23%) 0.794

0.100 83 � 13 83 � 13 0.979�0.001 115 � 16 115 � 16 0.708

0.114 72 � 11 71 � 11 0.738) 0.003 134 (38%) 147 (42%) 0.356) �0.001 165 (47%) 173 (49%) 0.582

) 0.001 36 (10%) 35 (10%) 1.000) 0.973 36 (10%) 39 (11%) 0.807) 0.068 76 (22%) 74 (21%) 0.924

6 0.003 14.0 � 1.6 14.1 � 1.6 0.5224 0.722 1.2 � 0.4 1.2 � 0.4 0.80847 0.046 4.31 � 0.47 4.31 � 0.46 0.8071 0.011 511 � 325 509 � 302 0.917

�0.001 29 � 13 34 � 7 �0.0010 �0.001 1.1 � 0.5 2.2 � 0.9 �0.001) 0.850 98 (28%) 94 (27%) 0.796

9 0.101 55.1 � 7.3 55.4 � 6.9 0.590) 0.505 37 (11%) 36 (10%) 1.0003 �0.001 22.4 � 7.5 22.1 � 7.2 0.520

.5 0.003 33.7 � 11.9 34.1 � 12.4 0.660

sity ma

ensity M

arfarin � 47

56 � 1207 (23%08 (23%94 (20%37 � 829 (91%

44 � 4465 (56%38 (51%39 (51%56 (33%08 (44%54 (12%78 (17%

29 (49%58 (97%

49 (95%33 (92%04 (43%98 (21%17 (25%

83 � 1314 � 1671 � 1136 (48%40 (51%38 (8%46 (10%96 (20%

.2 � 1.

.2 � 0.29 � 0.24 � 3234 � 8.2 � 1.24 (26%

.5 � 6.1 (11%.4 � 7.

.6 � 12

ences were estimated to evaluate the prematch imbalance

m

Awrd

jias�b


and postmatch balance and presented as a Love plot. Anabsolute standardized difference of 0% indicates no residualbias and differences �10% are considered inconsequential.

BEST participants were followed for a minimum of 18onths and a maximum of 4.5 years.10 Primary outcome for

the present analysis was all-cause mortality during 4.1 yearsof follow-up (mean 2, range 10 days to 4.14 years). Sec-ondary outcomes were cardiovascular and HF mortalitiesand all-cause and HF hospitalizations. Kaplan-Meier andCox regression analyses were used to determine associa-tions between warfarin use and outcomes during 4.1 yearsof follow-up. Log-minus-log scale survival plots were usedto check proportional hazards assumptions. Subgroup anal-yses were conducted to determine the homogeneity of as-sociation between use of warfarin and all-cause mortality.All statistical tests were 2-tailed with a p value �0.05considered statistically significant. All data analyses wereperformed using SPSS 18 for Windows (SPSS, Inc., Chi-cago, Illinois).

Results

Matched patients had a mean age � SD of 57 � 13years, 170 � 24% were women, and 171 � 24% were

frican-Americans. Before matching, patients receivingarfarin were younger, had a lower mean of LVEF and

ight ventricular EF, a lower prevalence of hypertension and

Figure 1. Absolute standardized differences comparing covariate valuesmatching.

iabetes mellitus, but had a greater symptom burden such as i

ugular venous distention and S3 gallop. These and othermbalances in baseline characteristics were well balancedfter matching (Figure 1, Table 1). After matching, absolutetandardized differences for all measured covariates were10% (most were �5%), suggesting substantial covariate

alance across groups (Figure 1). Median international normal-

tients with and without warfarin use before and after propensity score

Figure 2. Kaplan-Meier plots for all-cause mortality by use of warfarin.CI � confidence interval; HR � hazard ratio.

for pa

zed ratios (INRs; interquartile range) were 2.0 (1.1) and 1.0

2tpusb

0paach

D

tp

B

A

r

ll-caus


(0.1) for matched patients receiving and not receiving warfarin,respectively.

Overall, 212 matched patients (30%) died from all causesduring a 2.1-year median follow-up. All-cause mortalityoccurred in 30% and 30% of matched patients receiving andnot receiving warfarin, respectively (hazard ratio 0.86, 95%confidence interval 0.62 to 1.19, p � 0.361; Figure 2, Table). The association between warfarin use and all-cause mor-ality was homogenous across a wide spectrum of partici-ants, including those with LVEFs �20% and �20% (Fig-re 3). When we used LVEF as a continuous variable, wetill did not observe any statistically significant interaction

Table 2Effects of warfarin on all-cause mortality in BEST trial

Outcomes Events (%)

No Warfarin Warfarin

efore matchingPatients 1,171 471Unadjusted 321 (27%) 159 (34%)Multivariable-adjusted† — —Propensity-adjusted — —fter matchingPatients 354 354All-cause mortality 106 (30%) 106 (30%)

* Absolute rate increase was calculated by subtracting rates of eventsounded).

† Multivariable model includes all covariates displayed in Figure 1.

Figure 3. Association of use of warfarin and a

etween use of warfarin and LVEF (p for interaction � u

.815). Prematch-unadjusted, multivariable-adjusted, andropensity score–adjusted hazard ratios between warfarinnd no-warfarin use in the 1,642 patients before matchingre listed in Table 2. Warfarin had no association withardiovascular and HF mortalities and all-cause and HFospitalizations after matching (Table 3).

iscussion

Findings from the present study demonstrate that in pa-ients with advanced chronic systolic HF without AF and/orrevious thromboembolic events, prevalence of warfarin

Absolute RiskIncrease*

Hazard Ratio(95% confidence interval)

p Value

�7% 1.20 (0.99–1.45) 0.062— 1.14 (0.91–1.43) 0.253— 1.08 (0.87–1.35) 0.477

0% 0.86 (0.62–1.19) 0.361

group receiving from those not receiving warfarin (before values were

e mortality in subgroups of matched patients.

in the

se was relatively high. Nevertheless, our data suggest that

wr d

Nmcfiwr

ceaiw

B

A

r


despite achieving a mean therapeutic INR, warfarin use didnot provide any intrinsic survival benefit in patients withadvanced chronic systolic HF who had no other establishedindications for anticoagulation. These findings are impor-tant because many practicing physicians perceive advancedchronic systolic HF as a prethrombotic stage and prescribewarfarin, although current American College of Cardiology/American Heart Association guidelines have no clear rec-ommendation on this.4 Thus, warfarin might be prescribed

ithout any proved benefit and with a potential increasedisk of bleeding and other adverse effects.20

The unadjusted association of warfarin use with in-creased risk of cardiovascular mortality is rather surprisingbecause pre-match patients receiving warfarin were youngerand had lower or similar baseline prevalences of cardiovas-cular co-morbidities (Table 1). These suggest strong con-founding by a history of ventricular fibrillation, greatersymptom burden, and lower mean LVEF and right ventric-ular EF, which are known to increase risk of cardiovasculardeath. The near significant unadjusted association betweenwarfarin use and sudden cardiac death is likely due toincreased prevalence of ventricular fibrillation in warfa-rin users. Despite a greater burden of HF symptoms inwarfarin users, lack of significant unadjusted associationsof warfarin use with HF mortality and HF hospitalizationis intriguing but may suggest that LVEF and HF symp-toms were rather weak confounders.

Lack of significant associations of warfarin use withmortality and hospitalization in matched patients suggestslack of an intrinsic effect of warfarin on outcomes in pa-tients with advanced systolic HF without AF and/or throm-boembolic disorders. Although a low LVEF is often con-sidered an indication for warfarin use in these patients,findings from our subgroup analysis suggest that the asso-ciation between warfarin use and all-cause mortality wassimilar regardless of LVEF categories. Lack of an intrinsiceffect of warfarin in patients with advanced systolic HFwithout AF and/or thromboembolic events, despite a ther-apeutic INR, suggests that thromboembolic events may not

Table 3Effects of warfarin on other outcomes in BEST trial

Outcomes Events (%)

No Warfarin Warfarin

efore matchingPatients 1,171 471Cardiovascular mortality 268 (23%) 143 (30%)Sudden cardiac death 151 (13%) 81 (17%)Heart failure mortality 91 (8%) 45 (10%)All-cause hospitalization 702 (60%) 291 (62%)Heart failure hospitalization 412 (35%) 186 (40%)fter matchingPatients 354 354Cardiovascular mortality 84 (24%) 95 (27%)Heart failure mortality 27 (8%) 31 (9%)All-cause hospitalization 217 (61%) 225 (64%)Heart failure hospitalization 131 (37%) 141 (40%)

* Absolute rate increase was calculated by subtracting rates of eventsounded).

underlie mechanisms of death or hospitalization in these c

patients. Findings from our study provide further evidencesupporting current guideline recommendations that use ofwarfarin in patients with HF should be restricted to thosewith AF and/or previous thromboembolic events.4

There is conflicting evidence in the literature regardingthe role of warfarin in patients with advanced systolic HFwithout AF or other indications for anticoagulation.5,21 In 1study warfarin use was associated with lower mortality andmorbidity in patients with mild to moderate (2/3 had NYHAclass I to II symptoms) chronic systolic HF.6 Our study isistinguished by patients with more advanced HF (all withYHA class III to IV symptoms) and use of a propensity-atched design that allowed the assembly of a balanced

ohort. Findings from our study are consistent with thoserom the largest randomized clinical trial of anticoagulationn patients with HF and normal sinus rhythm to date inhich there was no difference in outcomes between patients

eceiving warfarin (open label), aspirin, or clopidogrel.8

However, this study was not considered definitive because itwas terminated prematurely because of slow enrollmentresulting in an estimated power of only 40% to detect a 20%difference. In the absence of another large clinical trial withadequate power, observational studies such as ours addfurther evidence of lack of benefit for therapeutic anticoag-ulation in these patients.

As in all observational studies, a key limitation of ourstudy is potential confounding by an unmeasured covariate.Sensitivity analysis would normally help quantify the de-gree of a hidden bias that would need to be present toinvalidate conclusions based on significant associations inan observational study. However, sensitivity analyses canbe performed only if the observed association is statisticallysignificant.22 Another limitation is lack of data on otherardiovascular events including stroke and adversevents such as bleeding. In conclusion, in patients withdvanced chronic systolic HF without AF and/or otherndications for anticoagulation, despite a mean INR thatas therapeutic, use of warfarin was not associated with

Absolute RiskIncrease*

Hazard Ratio(95% confidence interval)

p Value

�7% 1.29 (1.06–1.58) 0.013�4% 1.31 (1.00–1.71) 0.052�2% 1.19 (0.83–1.70) 0.336�2% 0.99 (0.86–1.14) 0.883�5% 1.13 (0.95–1.34) 0.174

�3% 0.97 (0.68–1.38) 0.855�1% 0.73 (0.38–1.39) 0.332�3% 0.96 (0.76–1.22) 0.763�3% 1.09 (0.82–1.44) 0.568

group receiving from those not receiving warfarin (before values were
in the
linical outcomes.

1

1


1. Gibbs CR, Blann AD, Watson RD, Lip GY. Abnormalities of hemo-rheological, endothelial, and platelet function in patients with chronicheart failure in sinus rhythm: effects of angiotensin-converting enzymeinhibitor and beta-blocker therapy. Circulation 2001;103:1746–1751.

2. Jafri SM, Ozawa T, Mammen E, Levine TB, Johnson C, Goldstein S.Platelet function, thrombin and fibrinolytic activity in patients withheart failure. Eur Heart J 1993;14:205–212.

3. Lip GY, Gibbs CR. Does heart failure confer a hypercoagulable state?Virchow’s triad revisited. J Am Coll Cardiol 1999;33:1424–1426.

4. Hunt SA, Abraham WT, Chin MH, Feldman AM, Francis GS, GaniatsTG, Jessup M, Konstam MA, Mancini DM, Michl K, Oates JA, RahkoPS, Silver MA, Stevenson LW, Yancy CW, Antman EM, Smith SC Jr,Adams CD, Anderson JL, Faxon DP, Fuster V, Halperin JL, HiratzkaLF, Jacobs AK, Nishimura R, Ornato JP, Page RL, Riegel B. ACC/AHA 2005 guideline update for the diagnosis and management ofchronic heart failure in the adult: a report of the American College ofCardiology/American Heart Association Task Force on PracticeGuidelines (Writing Committee to Update the 2001 Guidelines for theEvaluation and Management of Heart Failure): developed in collabo-ration with the American College of Chest Physicians and the Inter-national Society for Heart and Lung Transplantation: endorsed by theHeart Rhythm Society. Circulation 2005;112(suppl)e154–e235.

5. Lip GY, Gibbs CR. Anticoagulation for heart failure in sinus rhythm:a Cochrane systematic review. QJM 2002;95:451–459.

6. Al-Khadra AS, Salem DN, Rand WM, Udelson JE, Smith JJ, KonstamMA. Warfarin anticoagulation and survival: a cohort analysis from theStudies of Left Ventricular Dysfunction. J Am Coll Cardiol 1998;31:749–753.

7. Cleland JG, Findlay I, Jafri S, Sutton G, Falk R, Bulpitt C, Prentice C,Ford I, Trainer A, Poole-Wilson PA. The Warfarin/Aspirin Study inHeart failure (WASH): a randomized trial comparing antithromboticstrategies for patients with heart failure. Am Heart J 2004;148:157–164.

8. Massie BM, Collins JF, Ammon SE, Armstrong PW, Cleland JG,Ezekowitz M, Jafri SM, Krol WF, O’Connor CM, Schulman KA, TeoK, Warren SR. Randomized trial of warfarin, aspirin, and clopidogrelin patients with chronic heart failure: the Warfarin and AntiplateletTherapy in Chronic Heart Failure (WATCH) trial. Circulation 2009;119:1616–1624.

9. Ripley TL, Nutescu E. Anticoagulation in patients with heart failureand normal sinus rhythm. Am J Health Syst Pharm 2009;66:134–141.

10. Beta-Blocker Evaluation of Survival Trial Investigators. The BEST
investigators. A trial of the beta-blocker bucindolol in patients withadvanced chronic heart failure. N Engl J Med 2001;344:1659–1667.
11. Wahle C, Adamopoulos C, Ekundayo OJ, Mujib M, Aronow WS,Ahmed A. A propensity-matched study of outcomes of chronic heartfailure (HF) in younger and older adults. Arch Gerontol Geriatr2009;49:165–171.

12. Meyer P, Ekundayo OJ, Adamopoulos C, Mujib M, Aban I, White M,Aronow WS, Ahmed A. A propensity-matched study of elevatedjugular venous pressure and outcomes in chronic heart failure. Am JCardiol 2009;103:839–844.

13. Ekundayo OJ, Dell’Italia LJ, Sanders PW, Arnett D, Aban I, Love TE,Filippatos G, Anker SD, Lloyd-Jones DM, Bakris G, Mujib M, AhmedA. Association between hyperuricemia and incident heart failureamong older adults: a propensity-matched study. Int J Cardiol 2010;142:279–287.

14. Desai RV, Banach M, Ahmed MI, Mujib M, Aban I, Love TE, WhiteM, Fonarow G, Deedwania P, Aronow WS, Ahmed A. Impact ofbaseline systolic blood pressure on long-term outcomes in patientswith advanced chronic systolic heart failure (insights from the BESTtrial). Am J Cardiol 2010;106:221–227.

15. Bowling CB, Pitt B, Ahmed MI, Aban IB, Sanders PW, Mujib M,Campbell RC, Love TE, Aronow WS, Allman RM, Bakris GL, AhmedA. Hypokalemia and outcomes in patients with chronic heart failureand chronic kidney disease: findings from propensity-matched studies.Circ Heart Fail 2010;3:253–260.

16. Alper AB, Campbell RC, Anker SD, Bakris G, Wahle C, Love TE,Hamm LL, Mujib M, Ahmed A. A propensity-matched study of lowserum potassium and mortality in older adults with chronic heartfailure. Int J Cardiol 2009;137:1–8.

17. Ahmed MI, Ekundayo OJ, Mujib M, Campbell RC, Sanders PW, PittB, Perry GJ, Bakris G, Aban I, Love TE, Aronow WS, Ahmed A. Mildhyperkalemia and outcomes in chronic heart failure: A propensitymatched study. Int J Cardiol 2009;144:383–388.

8. Rubin DB. Using propensity score to help design observational stud-ies: Application to the tobacco litigation. Health Serv Outcomes ResMethodol 2001;2:169–188.

9. Rosenbaum PR, Rubin DB. The central role of propensity score inobservational studies for causal effects. Biometrika 1983;70:41–55.

20. Fihn SD, Callahan CM, Martin DC, McDonell MB, Henikoff JG,White RH. The risk for and severity of bleeding complications inelderly patients treated with warfarin. The National Consortium ofAnticoagulation Clinics. Ann Intern Med 1996;124:970–979.

21. Konstam MA. Antithrombotic therapy in heart failure: WATCHfulwondering. Circulation 2009;119:1559–1561.

22. Rosenbaum PR. Sensitivity to hidden bias. In: Rosenbaum PR, ed.
Observational Studies, Vol 1. New York: Springer-Verlag; 2002:105–170.

ts

The Risk of Thromboembolism in Heart Failure: Does It MeritAnticoagulation Therapy?

Eduard Shantsila, MD, and Gregory Y.H. Lip, MD*

atbptcte

i

Sfo340

tFw

The development of thromboembolism in patients withsystolic heart failure (HF) is well described. Thus, one maynaturally presume that anticoagulant therapy could be ben-eficial in patients with HF. However, although the benefitsof adequate anticoagulation are of no doubt in those withconcomitant atrial fibrillation (AF), is there enough evi-dence to advocate routine anticoagulation therapy for pa-tients with HF in sinus rhythm?1,2

Prothrombotic Factors in Heart Failure: Virchow’sTriad Revisited

Various abnormalities seen in patients with severe leftventricular (LV) systolic dysfunction, including endothelialdamage and dysfunction, abnormal blood stasis, and a hy-percoagulable state, provide a milieu contributing to throm-bogenesis and thromboembolism. These 3 components arerepresentative of Virchow’s triad of thrombogenesis, origi-nally proposed �150 years ago.

Indeed, endothelial damage and dysfunction are a hall-mark of HF, and their presence adversely affects prognosisin this condition. Endothelial dysfunction is accompaniedby the imbalance of pro- and anticoagulant systems, withclear shift toward a prothrombotic direction. Plasma mark-ers of endothelial damage and dysfunction, such as vonWillebrand factor, soluble thrombomodulin, and solubleE-selectin, are significantly increased in patients with acuteand chronic HF.3,4 A dysfunctional endothelium also ac-ively produces inflammatory cytokines (e.g., tumor necro-is factor–� and interleukin-1), which are significantly up-

regulated in HF and further promote thrombogenesis.5

In addition, the presence of dilated and/or dysfunctionalcardiac chambers creates areas of blood stasis, particularlyin patients with dilated cardiomyopathy, large anterior myo-cardial infarctions, and LV aneurysms. Stasis acceleratesthe activation of the coagulation system and fibrin forma-tion. Also, abnormal blood constituents leading to a pro-thrombotic or hypercoagulable state have been reported inHF patients, as reflected by high circulating biomarker lev-els, including fibrinogen, antithrombin III, fibrinopeptide A,and fibrin D-dimer.6,7 Activation of the neuroendocrine sys-tem, especially upregulation of angiotensin and endothelinproduction, further enhances the prothrombotic state in HF.8

Thrombosis in Chronic Heart Failure

Evidence of an increased risk for thromboembolic events(stroke, pulmonary and peripheral thromboembolism) in

University of Birmingham Centre for Cardiovascular Sciences, CityHospital, Birmingham, United Kingdom. Manuscript received October 18,2010; revised manuscript received and accepted October 19, 2010.

dE-mail address: [email protected] (G.Y.H. Lip).

patients with chronic HF free of AF in large prospectivecohorts is relatively limited, although retrospective analysesof HF treatment trials and data from some epidemiologiccohort studies are available. Many older cohort studies haveincluded proportions of patients with AF, which may wellbe asymptomatic and/or paroxysmal in nature, which isitself a strong independent risk factor for thromboembolism.Without associated AF, the risk for thromboembolism maybe small; for example, 1 analysis of patients with HF in NewYork Heart Association class II and III without AF foundonly a 1% annual risk for thromboembolism.9 However,lmost half of sudden cardiac deaths in HF have been showno be due to acute myocardial infarction or coronary throm-osis, and 27% of deaths in HF originally classified asrogressive congestive HF are actually caused by coronaryhrombosis.10 Given that sudden cardiac death is a majorontributor to mortality in HF populations, it is probablehat the impact of thromboembolism in HF might be under-stimated.

The risk for thromboembolism may be particularly highn patients with severely depressed cardiac contractility.11,12

In the Survival and Ventricular Enlargement (SAVE) trial,for example, the risk for stroke was twofold higher inpatients with ejection fractions (EFs) �28% compared tothose with EFs �28%, and every 5% reduction in the EFwas associated with an 18% increase in stroke risk.12 In the

udden Cardiac Death in Heart Failure Trial (SCD-HeFT),rom which patients with AF were excluded, the 4-year ratef thromboembolic events was 3.5% in those with EFs of0% to 35%, 3.6% in those with EFs of 20% to 30%, and.6% in those with EFs �20%, equivalent to annual rates of.9%, 0.9%, and 1.2%, respectively.13

Oral Anticoagulation in Heart Failure

Retrospective analyses of large HF trials have producedcontroversial results on the role of oral anticoagulation. Forexample, the analyses from the Studies of Left VentricularDysfunction (SOLVD) and SAVE trials found that warfarinseemed to be beneficial in patients with HF, being associ-ated with a significant 24% relative risk reduction in all-cause mortality as well as a lower risk for HF hospitaliza-tion, but not with a reduction of thromboembolic events.However, no benefits from warfarin for thromboembolismprevention were evident in the Veterans Affairs Vasodila-tor–Heart Failure Trial (V-HeFT)14 and SCD-HeFT.13

Recent controlled clinical trials of oral anticoagulationfor HF have had poor recruitment and small numbers andwere underpowered.15–17 The largest published study washe Warfarin and Antiplatelet Therapy in Chronic Heartailure (WATCH) trial, which randomized patients with HFith EFs �30% to receive warfarin or aspirin or clopi-
ogrel, but the study was terminated early because of poor
www.ajconline.org


M

ahrBllya

1

559Editorial/Anticoagulation Therapy in Heart Failure

recruitment. Despite being underpowered, the WATCHstudy did show a strong trend favoring warfarin over aspirinfor the reduction of nonfatal stroke (0.7% vs 2.1%), as wellas fewer hospitalizations in the warfarin group (16.1%)compared to aspirin (22.2%) and clopidogrel (18.3%).However, the grim side of warfarin therapy was a significantincrease in the bleeding rate (5.5%) compared to aspirin(3.6%) and clopidogrel (2.5%).16

In this issue of The American Journal of Cardiology,ujib et al18 provide further evidence that the universal

administration of oral anticoagulation in patients with HFmay not be beneficial. The study has the advantage ofstudying �1,600 patients with severe LV dysfunction, andthe investigators conclude that despite therapeutic interna-tional normalized ratios in those receiving warfarin, its usehad no significant effect on mortality and hospitalization.However, the study represents a post hoc analysis of apreviously published randomized clinical trial, and admin-istration of anticoagulants was at the discretion of the treat-ing physician and not randomized. Second, only 471 pa-tients received warfarin, with a median internationalnormalized ratio of 2.0, which perhaps suggests that abouthalf the patients treated with warfarin received suboptimalanticoagulation. Indeed, good anticoagulation control (ex-pressed as time in therapeutic range) is crucial for the bestoutcomes when warfarin is used. Third, detailed data onstroke and thromboembolism incidence were not recorded.Indeed, given the very high mortality of patients with HF(30% in this study), relatively small changes in thrombo-embolism-related mortality could pass unnoticed.

Given that the secondary end point of stroke was signif-icantly reduced in WATCH, attention to this outcome infuture analyses would be needed in any future clinical trials,rather than the focus on overall mortality, which may not besignificantly affected. Although Mujib et al18 tried to ex-clude patients with AF from their study, many patients withHF could develop AF in paroxysms and asymptomatically;thus, it is essentially how hard one looks to exclude under-lying AF from being a major confounder.

Current Approach and Future Directions

Current guidelines from the American Heart Associationand American College of Cardiology, the American Collegeof Chest Physicians, and the European Society of Cardiol-ogy do not support the routine use of warfarin in cardiomy-opathy in sinus rhythm.19–21 In many patients with HF, whore often elderly with multiple co-morbidities and polyp-armacy, assessment of bleeding risk would also be useful;ecent guidelines for AF have recommended the HAS-LED bleeding risk score (hypertension, abnormal renal or

iver function, stroke, bleeding history or predisposition,abile international normalized ratio, elderly [aged �65ears], and drugs or alcohol) as a simple, practical score tossess bleeding risk,22 but this has yet to be validated in a

large cohort of patients with HF.The ongoing double-blind, multicenter Warfarin Aspirin

Reduced Cardiac Ejection Fraction (WARCEF) trial aims toadequately address the utility of oral anticoagulation com-pared to aspirin in patients with HF with EFs �35%, and the
results are awaited. Finally, the development of novel oral
anticoagulants that overcome the limitations and disutilityof warfarin may also change the place of oral anticoagulanttherapy in the management of HF. Time will tell.

1. Rietbrock S, Plumb JM, Gallagher AM, van Staa TP. How effectiveare dose-adjusted warfarin and aspirin for the prevention of stroke inpatients with chronic atrial fibrillation? An analysis of the UK GeneralPractice Research Database. Thromb Haemost 2009;101:527–534.

2. Hughes M, Lip GY; Guideline Development Group, National ClinicalGuideline for Management of Atrial Fibrillation in Primary and Sec-ondary Care, National Institute for Health and Clinical Excellence.Stroke and thromboembolism in atrial fibrillation: a systematic reviewof stroke risk factors, risk stratification schema and cost effectivenessdata. Thromb Haemost 2008;99:295–304.

3. Shantsila E, Lip GY. The endothelium and thrombotic risk in heartfailure. Thromb Haemost 2009;102:185–187.

4. Chong AY, Lip GY, Freestone B, Blann AD. Increased circulatingendothelial cells in acute heart failure: comparison with von Wille-brand factor and soluble E-selectin. Eur J Heart Fail 2006;8:167–172.

5. Lip GY, Gibbs CR. Does heart failure confer a hypercoagulable state?Virchow’s triad revisited. J Am Coll Cardiol 1999;33:1424–1426.

6. Jug B, Vene N, Salobir BG, Sebestjen M, Sabovic M, Keber I.Prognostic impact of haemostatic derangements in chronic heart fail-ure. Thromb Haemost 2009;102:314–320.

7. Gombos T, Makó V, Cervenak L, Papassotiriou J, Kunde J, HársfalviJ, Förhécz Z, Pozsonyi Z, Borgulya G, Jánoskuti L, Prohászka Z.Levels of von Willebrand factor antigen and von Willebrand factorcleaving protease (ADAMTS13) activity predict clinical events inchronic heart failure. Thromb Haemost 2009;102:573–580.

8. Sbarouni E, Bradshaw A, Andreotti F, Tuddenham E, Oakley CM,Cleland JG. Relationship between hemostatic abnormalities and neu-roendocrine activity in heart failure. Am Heart J 1994;127:607–612.

9. Freudenberger RS, Wilson AC, Kostis JB; AFFIRM Investigators andCommittees. Comparison of rate versus rhythm control for atrial fi-brillation in patients with left ventricular dysfunction (from the AF-FIRM study). Am J Cardiol 2007;100:247–252.

0. Uretsky BF, Thygesen K, Armstrong PW, Cleland JG, Horowitz JD,Massie BM, Packer M, Poole-Wilson PA, Ryden L. Acute coronaryfindings at autopsy in heart failure patients with sudden death: resultsfrom the assessment of treatment with lisinopril and survival (ATLAS)trial. Circulation 2000;102:611–616.

11. Dries DL, Rosenberg YD, Waclawiw MA, Domanski MJ. Ejectionfraction and risk of thromboembolic events in patients with systolicdysfunction and sinus rhythm: evidence for gender differences in thestudies of left ventricular dysfunction trials. J Am Coll Cardiol 1997;29:1074–1080.

12. Loh E, Sutton MS, Wun CC, Rouleau JL, Flaker GC, Gottlieb SS,Lamas GA, Moyé LA, Goldhaber SZ, Pfeffer MA. Ventricular dys-function and the risk of stroke after myocardial infarction. N EnglJ Med 1997;336:251–257.

13. Freudenberger RS, Hellkamp AS, Halperin JL, Poole J, Anderson J,Johnson G, Mark DB, Lee KL, Bardy GH; SCD-HeFT Investigators.Risk of thromboembolism in heart failure: an analysis from the SuddenCardiac Death in Heart Failure Trial (SCD-HeFT). Circulation 2007;115:2637–2641.

14. Dunkman WB, Johnson GR, Carson PE, Bhat G, Farrell L, Cohn JN;The V-HeFT VA Cooperative Studies Group. Incidence of thrombo-embolic events in congestive heart failure. Circulation 1993;87:VI94–VI101.

15. Cleland JG, Findlay I, Jafri S, Sutton G, Falk R, Bulpitt C, Prentice C,Ford I, Trainer A, Poole-Wilson PA. The Warfarin/Aspirin Study inHeart Failure (WASH): a randomized trial comparing antithromboticstrategies for patients with heart failure. Am Heart J 2004;148:157–164.

16. Massie BM, Collins JF, Ammon SE, Armstrong PW, Cleland JG,Ezekowitz M, Jafri SM, Krol WF, O’Connor CM, Schulman KA, TeoK, Warren SR; WATCH Trial Investigators. Randomized trial ofwarfarin, aspirin, and clopidogrel in patients with chronic heart failure:the Warfarin and Antiplatelet Therapy in Chronic Heart Failure(WATCH) trial. Circulation 2009;119:1616–1624.

17. Cokkinos DV, Haralabopoulos GC, Kostis JB, Toutouzas PK; HELASinvestigators. Efficacy of antithrombotic therapy in chronic heart fail-
ure: the HELAS study. Eur J Heart Fail 2006;8:428–432.


18. Mujib M, Rahman AAZ, Desai RV, Ahmed MI, Feller MA, Aban I,Love TE, White M, Deedwania P, Aronow WS, Fonarow G, AhmedA. Warfarin use and outcomes in patients with advanced chronicsystolic heart failure without atrial fibrillation, prior thromboembolicevents, or prosthetic valves. Am J Cardiol 2010;107:552–557.

19. Hirsh J, Fuster V, Ansell J, Halperin JL; American Heart Association/American College of Cardiology Foundation. American Heart Asso-ciation/American College of Cardiology Foundation guide to warfarintherapy. J Am Coll Cardiol 2003;41:1633–1652.

20. Albers GW, Amarenco P, Easton JD, Sacco RL, Teal P. Antithrom-botic and thrombolytic therapy for ischemic stroke: the Seventh ACCP
Conference on Antithrombotic and Thrombolytic Therapy. Chest2004;126:483S–512S.
21. Swedberg K, Cleland J, Dargie H, Drexler H, Follath F, Komajda M,Tavazzi L,Smiseth OA, Gavazzi A, Haverich A, Hoes A, Jaarsma T,Korewicki J, Lévy S, Linde C, Lopez-Sendon JL, Nieminen MS,Piérard L, Remme WJ; Task Force for the Diagnosis and Treatment ofChronic Heart Failure of the European Society of Cardiology. Guide-lines for the diagnosis and treatment of chronic heart failure: executivesummary (update 2005): the Task Force for the Diagnosis and Treat-ment of Chronic Heart Failure of the European Society of Cardiology.Eur Heart J 2005;26:1115–1140.

22. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. Anovel user-friendly score (HAS-BLED) to assess one-year risk of
major bleeding in atrial fibrillation patients: the Euro Heart Survey.Chest. 2010;138:1093–1100.

lptsipp

(t

Cl

NskaS

aNr

0d

Trials on the Effect of Cardiac Resynchronization on ArterialBlood Pressure in Patients With Heart Failure

Sameer Ather, MDa, Sripal Bangalore, MD, MHAb, Srinath Vemuri, MDa, Long B. Cao, MDa,Biykem Bozkurt, MD, PhDa, and Franz H. Messerli, MDc,*

Cardiac resynchronization therapy (CRT) increases cardiac performance in patients withheart failure, but its effect on arterial pressure is not well established. To determine theeffect of CRT on systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulsepressure (PP) a systematic review using standard nomenclatures for CRT was done inScopus (MEDLINE and Embase), Cochrane Controlled Trials Register, National Institutesof Health http://www.ClinicalTrials.gov database, and bibliography of select meta-analysesfor studies evaluating CRT in patients with dilated cardiomyopathy. Two independentinvestigators extracted the articles based on predefined criteria. The primary outcome wasdifference in arterial pressure parameters from baseline to after CRT in nonrandomizedcohort trials. This was then validated by comparing the change in arterial pressure betweenCRT and medical therapy groups in randomized controlled trials. A random-effects modelwas used for analyses. Analyses of 15 nonrandomized studies showed that CRT resulted inan increase (from baseline) in SBP by 4.4 mm Hg (95% confidence interval [CI] 0.8 to 8.0,p � 0.02), no change in DBP (p � 0.21), and an increase in PP by 2.8 mm Hg (95% CI 1.0to 4.6, p � 0.003). Results from the 3 randomized controlled trials were concordant with anincrease in SBP by 3.9 mm Hg (95% CI 1.1 to 6.8, p � 0.007), no effect on DBP (p � 0.40),and an increase in PP by 4.3 mm Hg (95% CI 4.1 to 4.5, p <0.001) compared to medicaltherapy. In conclusion, CRT is associated with a modest increase in SBP and PP in patientswith heart failure. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:
561–568)
More than 90% of patients with heart failure (HF) have ahistory of hypertension.1 In contrast, in severe HF decreasingeft ventricular function is unable to sustain the high bloodressure (BP) despite compensatory mechanisms (salt and wa-er retention, vasoconstriction, sympathetic stimulation and de-ensitization, cardiac hypertrophy, and cellular changes includ-ng appearance of slow myosin, prolongation of actionotential, post-translational modifications in calcium handlingroteins, and increase in collagen).2,3 Thus, cardiac output

decreases in parallel with systolic BP (SBP) and pulse pressure(PP).4 In patients with HF, cardiac resynchronization therapyCRT) improves left ventricular systolic function,5 HF symp-oms,6 quality of life,7 exercise tolerance,8 maladaptive remod-

eling,9 morbidity (HF admissions), and mortality.10 Americanollege of Cardiology/American Heart Association guide-

ines11 recommend (class I) CRT in patients with ejectionfraction �35%, QRS duration �120 ms, sinus rhythm, and

ew York Heart Association class III/ambulatory class IV HFymptoms on optimal medical therapy. However, it is notnown if this improvement in systolic function translates inton increase in BP. Limited data seem to suggest an increase inBP but this has not been consistently reported or studied.8,10

aBaylor College of Medicine, Houston, Texas; bBrigham and Women’sHospital, Boston, Massachusetts; cSt. Luke’s–Roosevelt Hospital Centernd Columbia University, College of Physicians and Surgeons, New York,ew York. Manuscript received September 22, 2010; revised manuscript

eceived and accepted October 7, 2010.*Corresponding author: Tel: 212-523-7373; fax: 212-523-7765.
E-mail address: [email protected] (F.H. Messerli).

We hypothesized that CRT with its associated improvement incardiac function would result in an increase in BP profile.Thus, our primary objective was to evaluate the effect of CRTon SBP, diastolic BP (DBP), and PP.

Methods

Eligible studies were prospective nonrandomized cohortstudies that reported BP profile at baseline and after CRT withfirst follow-up within 6 months of CRT or randomized con-trolled trials of CRT (with or without implantable cardio-verter–defibrillator) that reported BP profile in the CRT andmedical therapy groups. To avoid major medication changesover follow-up period and additional cardiovascular insults, alimit of 6 months of follow-up was used for nonrandomizedcohort studies. Studies were identified by searching electronicdatabases, including Scopus (MEDLINE 1966 to October2009, Embase 1980 to October 2009), Cochrane Trials Reg-ister, and National Institutes of Health http://www.ClinicalTri-als.gov database (closed studies only) using the terms “cardiacresynchronization” or “biventricular pacing” or “biventricularpacemaker” or “multisite pacing” or “multisite pacemaker” or“dual-site pacing” or “dual-site pacemaker” or “left ventricularpacing” or “left ventricular pacemaker.” In addition, referencelists of select meta-analyses were searched for reports of rele-vant studies.12–17 Studies in which the intervention includedrevascularization (coronary artery bypass grafting or percuta-neous coronary intervention) at the time of CRT were excludedas revascularization could have potentially confounded the
effect of CRT on BP outcomes (Figure 1).
www.ajconline.org

http://www.ClinicalTrials.gov




fbsiustcoectac

SfTbvwiw

fssD

pwspn

amidco3

AIsOmmaiv(v

pdV�

ewED

(t

D


Eligible studies had to fulfill the following inclusion crite-ria: (1) randomized or nonrandomized cohort studies of CRTin patients with dilated cardiomyopathy, (2) studies reportingoutcomes of interest (SBP, DBP, or PP; before and after CRTin nonrandomized cohort studies and CRT vs medical therapyin randomized controlled trials), and (3) follow-up �6 monthsor nonrandomized cohort studies. There were no restrictionsased on language or year of publication. Studies were re-tricted to published data. Studies that had duplicated data,ncluding same group of patients or for whom there werepdated results available, were excluded. We included onlytudies that did not exclude nonresponders from their analyseso prevent bias towards a positive result. Further, studies in-luding patients with ischemic and/or nonischemic cardiomy-pathy were included in this meta-analysis, whereas studiesvaluating the effect of CRT, specifically in pediatric patients,ongenital heart disease, hypertrophic cardiomyopathy, restric-ive cardiomyopathy, chemotherapy-induced cardiomyopathy,nd infectious cardiomyopathy, e.g., Chagas disease, were ex-luded.

Primary analyses were changes in BP parameters, i.e.,BP, DBP, and PP from baseline (before CRT) to that atollow-up (after CRT) in nonrandomized cohort studies.his was validated by comparing changes in BP parametersetween a CRT group (with/without implantable cardio-erter–defibrillator) and a medical therapy group (with/ithout implantable cardioverter–defibrillator) in random-

zed controlled trials. The 2 analyses were done separatelyithout pooling the data.Eligibility assessment and data abstraction were per-

ormed independently by 2 authors (S.V. and L.B.C) andupervised by S.A. We extracted inclusion criteria, exclu-ion criteria, baseline data, outcomes, and report quality.isagreements were resolved by consensus.

Figure 1. Schematic representation of data search and acquisition.CMP � dilated cardiomyopathy; F/U � follow-up.

To assess risk of bias in nonrandomized cohort studies,

resence of single or double blinding and documentation ofithdrawal were ascertained. For nonrandomized cohort

tudies, intermediate risk of bias was defined as a lowossibility of bias in the 2 domains. As the studies wereonrandomized, none were considered at low risk.

For randomized controlled trials, methodologic quality wasssessed by reported allocation generation, allocation conceal-ent, blinding, documentation of withdrawal, selective report-

ng, and intention-to-treat analysis in line with the recommen-ation by the Cochrane Collaboration.18 For randomizedontrolled trials, high risk of bias was defined as a possibilityf bias in �4 domains, moderate as a possibility of bias in 2 todomains, and low risk as a possibility of bias in �1 domain.Statistical analyses were done using Comprehensive Meta-

nalysis 2.2.046 in accordance with the Preferred Reportingtems for Systematic reviews and Meta-Analyses (PRISMA)tatement (for randomized controlled trials) and Meta-analysisf Observational Studies in Epidemiology (MOOSE) state-ent for others. Mean difference was chosen as the principaleasurement of effect as the unit of measurement was same

cross all studies. Studies included reported mean or differencen means and standard deviation (SD) or p value for theariables. If the SD was not available for 1 of the 3 variablesSBP, DBP, PP), then the SD was calculated using a before-ersus-after correlation of 0.7 from the other 2 SDs available.

Data were analyzed for heterogeneity by I2 statistic pro-osed by Higgins and Thompson19 separately for nonran-omized cohort studies and randomized controlled trials.alues �30% indicated mild heterogeneity and those50% substantial heterogeneity.19 In the presence of het-

erogeneity, a random effects model (DerSimonian–Lairdapproach)20 was used to pool the data; otherwise, a fixed-ffects model (inverse variance) was used. Publication biasas assessed and quantified using the regression intercept ofgger et al21 and corrected by the trim-and-fill method ofuval and Tweedie.22

Results

Fifteen nonrandomized cohort studies and 3 randomizedcontrolled trials met our inclusion criteria for analyses (Fig-ure 1). Of the 18 studies included in the meta-analyses, 15(nonrandomized controlled trials) compared variables be-fore and after CRT23–37 in 492 patients, whereas 3 studiesrandomized controlled trials) compared CRT (n � 1,637)o optimal medical management (n � 727).8,10,38 Baseline

characteristics and inclusion and exclusion criteria are pre-sented in Tables 1 and 2. There were 2 studies on the PacingTherapies in Congestive Heart Failure (PATH-CHF) trialbut separate variables were extracted from each study.23,24

Two trials with 3 published studies included patients withepicardial pacing,23,24,34 whereas the other studies usedtransvenous biventricular pacing.

All nonrandomized cohort studies reported withdrawalsor crossovers or had no withdrawal. In 1 study, patientswere unaware of their treatment 27 and 1 study documentedthat it was single blinded but did not specify who wasblinded.37 Thus, these 2 studies were considered at inter-mediate risk of bias, whereas the rest of the nonrandomizedcohort studies were at high risk of bias. Among the ran-
domized controlled trials, based on the 6 parameters sug-

tHaiS0a

t�r(t(

tHSnCt

SlYmt

cfgoi

riii9smt0

ttg

ejectio

563Heart Failure/Cardiac Resynchronization and Blood Pressure

gested by the Cochrane Review, the Cardiac Resynchroni-zation–Heart Failure (CARE-HF)10 and Comparison ofMedical Therapy, Pacing, and Defibrillation in Heart Fail-ure (COMPANION)8 trials were at low risk of bias with lowrisk in 4 of 6 categories, whereas the third study by Piccir-illo et al38 was at high risk of bias across all categories.

In nonrandomized cohort studies, compared to baselinehere was a significant increase in SBP (difference �4.4 mmg, 95% confidence interval [CI] �0.8 to �8.0, p � 0.02)

fter CRT (Figure 2). This was concordant in the random-zed controlled trials, where there was a significantly higherBP (difference �3.9 mm Hg, 95% CI �1.1 to �6.8, p �.007) in the CRT group compared to the medical therapylone group (Figure 3).

In nonrandomized cohort studies, compared to baselinehere was no change in DBP (difference �1.0 mm Hg, 95% CI

0.6 to �2.6, p � 0.20) after CRT (Figure 4). Similarly, in theandomized controlled trials, there was no change in DBPdifference �0.5 mm Hg, 95% CI �0.7 to �1.7, p � 0.40) inhe CRT group compared to the medical therapy alone groupFigure 5).

In nonrandomized cohort studies, compared to baselinehere was a significant increase in PP (difference �2.8 mmg, 95% CI �1.0 to �4.6, p � 0.003) after CRT (Figure 6).imilarly, in randomized controlled trials, there was a sig-ificant increase in PP (difference �4.3 mm Hg, 95%I �4.1 to �4.5, p �0.001) in the CRT group compared to

he medical therapy alone group (Figure 7).There was significant heterogeneity for analyses of

BP, DBP, and PP. Sensitivity analyses based on fol-ow-up period, year of publication, size of study, Nework Heart Association class, mean ejection fraction,ean QRS duration, presence of ischemic cardiomyopa-

Table 1Baseline characteristics of included trials

Study Subjects Age(years)

NYHAClass

BasEF

Randomized controlled trials*Piccirillo et al,38 2006 15/16 65/65 3.7/3.7 0.22Bristow et al,8 2004 308/1,212 68/67 3.2/3.1 0.22Cleland et al,10 2005 404/409 66/67 3.1/3.1 0.25

Nonrandomized cohort studiesAuricchio et al,24 1999 25 62 3.2 0Auricchio et al,23 2003 85 60 NA 0Bakker et al,25 2000 12 64 3.7 0Braunschweig et al,26 2000 16 64 3.1 0Flevari et al,27 2006 25 66 3.2 0Fung et al,28 2008 97 64 3.2 0Inage et al,29 2008 17 63 3.2 0Knaapen et al,30 2004 14 58 3.1 0Kubanek et al,31 2006 43 62 3.2 0Madaric et al,32 2007 28 67 3 0Mullens et al,33 2008 19 66 3.4 0Nelson et al,34 2000 22 59 NA 0Sogaard et al,35 2001 22 61 3.4 0Vanderheyden et al,36 2008 10 64 NA 0Waggoner et al,38 2006 57 61 3.2 0

* For randomized controlled trials data are presented as medical therapNA � absence of data; NYHA � New York Heart Association; EF �

hy, and industry funding were performed but did not

hange the assessed heterogeneity. Thus, a random-ef-ects model was used to minimize the effect of hetero-eneity. There was no significant publication bias basedn the Egger regression intercept in any of the compar-sons (funnel plots not included).

There were 2 studies each with low and intermediateisk of bias. On redoing the analysis with only low-/ntermediate-risk studies, there was a trend toward anncrease in SBP compared to baseline in 2 nonrandom-zed cohort studies after CRT (difference �4.8 mm Hg,5% CI �0.6 to �10.1, p � 0.08), whereas SBP wasignificantly higher in the CRT group compared to the opti-al medical therapy group in the 2 randomized controlled

rials (difference �3.9 mm Hg, 95% CI �0.6 to �7.3, p �.02).8,10,27,37 Similarly, on analysis of low-/intermediate-risk

studies, the 2 nonrandomized cohort trials showed a trendtowards an increase in PP (difference �2.5 mm Hg, 95% CI�0.5 to �5.6, p � 0.099) compared to baseline, whereashe CARE-HF trial showed a significant increase in PP inhe CRT group compared to the optimal medical therapyroup (difference �4.3 mm Hg, 95% CI �4.1 to �4.5, p

�0.001).10,27,37 There was no effect of CRT on DBP inlow-/intermediate-risk studies in randomized controlled tri-als or nonrandomized cohort trials.

In addition, to rule out the effect of change in medicationon BP profile, we analyzed 11 nonrandomized cohort stud-ies in which none of the BP medications were changedduring follow-up. In this subanalysis there was no changein the results with an increase in SBP (difference �5.6mm Hg, 95% CI 1.0 to �10.2, p � 0.02),24 –26,29,30,32–36

an increase in PP (difference �2.8 mm Hg, 95% CI 0.5to �5.1, p � 0.02),23,25,26,30,32,34 and no effect on DBP(difference �0.4 mm Hg, 95% CI �1.7 to �2.6, p �

QRS Width(ms)

SinusRhythm (%)

IschemicCause (%)

Men (%) EF AfterCRT

159/160 1/1 1/1 0.80/0.81 0.28/0.22158/160 1/1 0.59/0.54 0.69/0.67 NA160/160 1/1 0.36/0.40 0.73/0.74 NA

168 1 0.33 0.56 NA155 1 0.38 0.66 NA194 1 0.33 0.42 0.21181 0.56 0.69 0.94 NA195 1 0.44 0.76 0.34NA 1 0.42 0.74 0.34161 NA 0.12 0.65 0.33173 1 0.43 0.57 0.37195 0.79 0.42 0.86 0.24171 1 0.5 0.82 0.28177 NA 0.55 0.8 0.34175 1 0.23 NA NA184 1 0.56 0.88 0.23179 1 0.4 0.8 0.34180 1 0.25 0.76 0.34

c resynchronization arms.n fraction.

eline(%)

/0.23/0.21/0.25

.21

.23

.15

.22

.27

.26

.25

.25

.22

.25

.21

.2

.23

.19

.25

y/cardia

0.4).24 –26,30,32,34

R

N

2

consumption. Other abbreviations as in Table 1.


Figure 2. Effect of cardiac resynchronization therapy on systolic blood pressure shown by paired comparison of baseline data to data after cardiac

Table 2Inclusion criteria and baseline blood pressure profile in included trials

Study Inclusion Criteria SBP (mm Hg) DBP (mm Hg)

andomized controlled trials*Piccirillo et al,38 2006 Patients with HF with NYHA class �III, EF �0.35, QRS duration �120 ms, sinus

rhythm, and ischemic cause109/112 69/68

Bristow et al,8 2004 Patients with HF with NYHA class �III, EF �0.35, QRS duration �120 ms, sinusrhythm, PR �150 ms, HF hospitalization in previous year

112/111 64/68

Cleland et al,10 2005 Patients with HF �18 years old with NYHA class �III, EF �0.35, QRS duration�120 ms, sinus rhythm, LVEDD �30 mm

110/110 70/70

onrandomized cohort studiesAuricchio et al,24 1999 Patients with HF with NYHA class �III, QRS duration �120 ms, sinus rhythm, PR

�150 ms90 57

Auricchio et al,23 2003 Patients with HF 18–75 years old with NYHA class �II, EF �0.30, QRS duration�120 ms, sinus rhythm, peak VO2 �18 ml/min/kg on maximal exercise

113 NA

Bakker et al,25 2000 Patients with HF 18–75 years old with NYHA class �III, LBBB with QRS duration�120 ms, sinus rhythm

NA NA

Braunschweig et al,26 2000 Patients with HF with NYHA class �III, EF �0.40, QRS duration �150 ms 109 70Flevari et al,27 2006 Patients with HF with NYHA class �III, EF �0.35, QRS duration �120 ms, sinus

rhythm110 70

Fung et al,28 2008 Standard CRT indication: NYHA class �III, EF �0.35, QRS duration �120 ms 105 NAInage et al,29 2008 Patients with HF with NYHA class �III, QRS duration �120 ms 98 NAKnaapen et al,30 2004 Patients with HF with NYHA class �III, EF �0.35, QRS duration �120 ms, sinus

rhythm, LVEDD �55 mm114 71

Kubanek et al,31 2006 Patients with HF with NYHA class �III, EF �0.35, QRS duration �140 ms,LVEDD �60 mm

120 76

Madaric et al,32 2007 Standard indication for CRT: NYHA class �III, EF �0.35, QRS duration �120ms, sinus rhythm

112 71

Mullens et al,33 2008 Patients with HF with NYHA class �III, EF �0.35, LBBB with QRS duration�120 ms, HR �70 beats/min, sum of dyssynchrony �102 ms, preserved AVconduction

111 NA

Nelson et al,34 2000 Patients with HF with NYHA class �III, EF �0.35, QRS duration �140 ms, sinusrhythm

110 72

Sogaard et al,35 2001 Patients with HF with NYHA class �III, QRS duration �120 ms, sinus rhythm 103 NAVanderheyden et al,36 2008 Patients with HF with NYHA class �III, EF �0.25, LBBB with QRS duration

�140 ms, sinus rhythm, total sum of dyssynchrony � 102 ms, HR �70 beats/min

113 NA

Waggoner et al,37 2006 Patients with HF with NYHA class �III, EF �0.35, QRS duration �150 ms, sinusrhythm, LVEDD �60 mm

113 67

* Randomized controlled trials data are presented as medical therapy/cardiac resynchronization arms.AV � atrioventricular; HR � heart rate; LBBB � left bundle branch block; LVEDD � left ventricular end-diastolic dimension; VO � oxygen

resynchronization therapy from nonrandomized cohort studies. *p �0.05; **p �0.01; ***p �0.001. � � difference.

.001. A

**p �0

.001. A

0.001.


Discussion

Our results indicate that in patients with HF withstandard indications for CRT, there was a moderate in-

Figure 3. Effect of cardiac resynchronization therapy on systolic blood presstherapy from randomized controlled trials. *p �0.05; **p �0.01; ***p �0

Figure 4. Effect of cardiac resynchronization therapy on diastolic bloodresynchronization therapy from nonrandomized cohort studies. *p �0.05;

Figure 5. Effect of cardiac resynchronization therapy on diastolic blood ptherapy from randomized controlled trials. *p �0.05; **p �0.01; ***p �0

Figure 6. Effect of cardiac resynchronization therapy on pulse pressure shotherapy from nonrandomized cohort studies. *p �0.05; **p �0.01; ***p �

crease in SBP and PP compared to baseline. This was

confirmed in analyses of patients from randomizedcontrolled trials, where there was a modest increase in BPparameters compared to medical therapy alone. However,we did not find any difference in DBP between

wn by comparison of cardiac resynchronization therapy to optimal medicalbbreviation as in Figure 2.

re shown by paired comparison of baseline data to data after cardiac.01; ***p �0.001. Abbreviation as in Figure 2.

by comparison of cardiac resynchronization therapy to optimal medicalbbreviation as in Figure 2.

paired comparison of baseline data to data after cardiac resynchronizationAbbreviation as in Figure 2.

ure sho

pressu

ressure

wn by

groups.

fccuiCi

d

aeHme

sa

wcm

caneuae

breviat


There is a close correlation between HF and hyperten-sion. Hypertension ultimately leads to HF and patientswith HF have a high prevalence of hypertension. How-ever, in patients with advanced HF, a low SBP is usuallyseen even in patients who were previously hypertensive.This is termed “decapitated hypertension” in which pa-tients who are hypertensive to begin with develop nor-mal/low BP as HF progresses. This results from decreas-ing pump function and cardiac output despite thepresence of compensatory mechanisms such as peripheralvasoconstriction. This inability to generate higher SBP isaccepted as an indicator of poor pump function. Patientswith decapitated hypertension are difficult to managebecause of their inability to tolerate HF medications thatcan potentially lower BP such as angiotensin-convertingenzyme inhibitors/angiotensin receptor blockers, diuret-ics, and � blockers. Although CRT increases ejectionraction in patients with HF, increase in BP has not beenonclusively shown. Our results show that CRT is asso-iated with a modest increase in BP in this patient pop-lation, which can potentially lead to reversal of decap-tated hypertension. This potential increase in SBP usingRT can provide a window to the treating cardiologist to

ntroduce/continue HF medications.This interplay among high BP, hypertensive HF, and

ilated cardiomyopathy was lucidly discussed by Oakley39

about 3 decades ago: “The development of left ventricularfailure because of hypertension determines a decrease of thepreviously raised BP to normal levels and, since the failureusually persists the BP remains normal. If the patient re-covers from HF, then the BP rises and the diagnosis is likelyto be ‘hypertension.’ In other words, the ‘diagnosis” variesbetween dilated cardiomyopathy and hypertension accord-ing to left ventricular function and only if a patient withdilated cardiomyopathy, HF and normal BP actually recov-ers and develops high BP can the causal or conditioningrole of high BP be proved.”

Although hypertension is associated with developmentof incident HF,40 higher SBP has a protective survivaleffect in patients with established HF.41,42 In our analy-sis, CRT increased SBP in patients with advanced HFwho had normal to low-normal SBP. Whether this im-provement in BP improves survival is not clearly defined.In a recent study by Tanaka et al,43 an increase in SBPfter CRT was associated with a decrease in the combinednd point of HF hospitalization and all-cause mortality.owever, further studies are needed to prove if improve-ent in arterial pressure improves survival independent of

Figure 7. Effect of cardiac resynchronization therapy on pulse pressure showfrom randomized controlled trials. *p �0.05; **p �0.01; ***p �0.001. Ab

jection fraction.

In the setting of decreased cardiac output, there isympathetic activation and parasympathetic withdraw-l.44 CRT improves cardiac function by reverting asyn-

chronous mechanical events in patients with HF, espe-cially in patients with a wide QRS complex or withechocardiographic dyssynchrony. Moreover, this im-provement in myocardial function is associated with res-titution of the defunct autonomic function in patientswith HF.45 Thus, improvement in myocardial function

ithout a concomitant increase in adrenergic activityould be a possible mechanism behind improvement inortality using CRT.Given the paucity of data, we used both randomized

ontrolled trials and nonrandomized cohort studies for ournalysis. The effect of CRT on BP profile was heteroge-eous. Because extensive sensitivity analyses could notxplain this heterogeneity, a random-effects model wassed to mitigate the effect of this heterogeneity. As a result,lthough we could pool the studies, we were not able toxplain the possible cause of this heterogeneity.

1. Levy D, Larson MG, Vasan RS, Kannel WB, Ho KK. The progressionfrom hypertension to congestive heart failure. JAMA 1996;275:1557–1562.

2. Kushnir A, Shan J, Betzenhauser MJ, Reiken S, Marks AR. Role ofCaMKIIdelta phosphorylation of the cardiac ryanodine receptor inthe force frequency relationship and heart failure. Proc Natl Acad SciU S A 2010;107:10274–10279.

3. Katz AM. Cardiomyopathy of overload. A major determinant of prog-nosis in congestive heart failure. N Engl J Med 1990;322:100–110.

4. Nohria A, Lewis E, Stevenson LW. Medical management of advancedheart failure. JAMA 2002;287:628–640.

5. Kawaguchi M, Murabayashi T, Fetics BJ, Nelson GS, Samejima H,Nevo E, Kass DA. Quantitation of basal dyssynchrony and acuteresynchronization from left or biventricular pacing by novel echo-contrast variability imaging. J Am Coll Cardiol 2002;39:2052–2058.

6. Auricchio A, Stellbrink C, Sack S, Block M, Vogt J, Bakker P,Huth C, Schondube F, Wolfhard U, Bocker D, Krahnefeld O,Kirkels H. Pacing Therapies in Congestive Heart Failure (PATH-CHF) Study Group. Long-term clinical effect of hemodynamicallyoptimized cardiac resynchronization therapy in patients with heartfailure and ventricular conduction delay. J Am Coll Cardiol 2002;39:2026 –2033.

7. Werling C, Weisse U, Siemon G, Kiessling AH, Rameken M,Schwacke H, Saggau W, Senges J, Seidl K. Biventricular pacing inpatients with ICD: how many patients are possible candidates? J Tho-rac Cardiovasc Surg 2002;50:67–70.

8. Bristow MR, Saxon LA, Boehmer J, Krueger S, Kass DA, DeMarco T, Carson P, DiCarlo L, DeMets D, White BG, DeVries DW,Feldman AM. Comparison of Medical Therapy, Pacing, and Defi-brillation in Heart Failure (COMPANION) Investigators. Cardiac-resynchronization therapy with or without an implantable defibril-

omparison of cardiac resynchronization therapy to optimal medical therapyion as in Figure 2.

n by c

lator in advanced chronic heart failure. N Engl J Med 2004;350:2140 –2150.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4


9. St John Sutton MG, Plappert T, Abraham WT, Smith AL, DeLurgioDB, Leon AR, Loh E, Kocovic DZ, Fisher WG, Ellestad M, Messen-ger J, Kruger K, Hilpisch KE, Hill MR, Multicenter InSync Random-ized Clinical Evaluation (MIRACLE) Study Group. Effect of cardiacresynchronization therapy on left ventricular size and function inchronic heart failure. Circulation 2003;107:1985–1990.

0. Cleland JG, Daubert JC, Erdmann E, Freemantle N, Gras D, Kappen-berger L, Tavazzi L. Cardiac Resynchronization–Heart Failure(CARE-HF) Study Investigators. The effect of cardiac resynchroniza-tion on morbidity and mortality in heart failure. N Engl J Med 2005;352:1539–1549.

1. Epstein AE, DiMarco JP, Ellenbogen KA, Estes NA III, FreedmanRA, Gettes LS, Gillinov AM, Gregoratos G, Hammill SC, Hayes DL,Hlatky MA, Newby LK, Page RL, Schoenfeld MH, Silka MJ, Steven-son LW, Sweeney MO, Smith SC Jr, Jacobs AK, Adams CD, Ander-son JL, Buller CE, Creager MA, Ettinger SM, Faxon DP, Halperin JL,Hiratzka LF, Hunt SA, Krumholz HM, Kushner FG, Lytle BW, Nish-imura RA, Ornato JP, Page RL, Riegel B, Tarkington LG, Yancy CW,American College of Cardiology/American Heart Association TaskForce on Practice Guidelines (Writing Committee to Revise the ACC/AHA/NASPE 2002 Guideline Update for Implantation of CardiacPacemakers and Antiarrhythmia Devices), American Association forThoracic Surgery, Society of Thoracic Surgeons. ACC/AHA/HRS2008 guidelines for device-based therapy of cardiac rhythm abnormal-ities: a report of the American College of Cardiology/American HeartAssociation Task Force on Practice Guidelines (Writing Committee toRevise the ACC/AHA/NASPE 2002 Guideline Update for Implanta-tion of Cardiac Pacemakers and Antiarrhythmia Devices): developedin collaboration with the American Association for Thoracic Surgeryand Society of Thoracic Surgeons. Circulation 2008;117(suppl):e350–e408.

2. Bradley DJ, Bradley EA, Baughman KL, Berger RD, Calkins H,Goodman SN, Kass DA, Powe NR. Cardiac resynchronization anddeath from progressive heart failure: a meta-analysis of randomizedcontrolled trials. JAMA 2003;289:730–740.

3. Rivero-Ayerza M, Theuns DA, Garcia-Garcia HM, Boersma E, Simo-ons M, Jordaens LJ. Effects of cardiac resynchronization therapy onoverall mortality and mode of death: a meta-analysis of randomizedcontrolled trials. Eur Heart J 2006;27:2682–2688.

4. Lam SK, Owen A. Combined resynchronisation and implantable de-fibrillator therapy in left ventricular dysfunction: Bayesian networkmeta-analysis of randomised controlled trials. BMJ 2007;335:925.

5. McAlister FA, Ezekowitz JA, Wiebe N, Rowe B, Spooner C, CrumleyE, Hartling L, Klassen T, Abraham W. Systematic review: cardiacresynchronization in patients with symptomatic heart failure. AnnIntern Med 2004;141:381–390.

6. McAlister FA, Ezekowitz J, Hooton N, Vandermeer B, Spooner C,Dryden DM, Page RL, Hlatky MA, Rowe BH. Cardiac resynchroni-zation therapy for patients with left ventricular systolic dysfunction: asystematic review. JAMA 2007;297:2502–2514.

7. Abdulla J, Haarbo J, Kober L, Torp-Pedersen C. Impact of implantabledefibrillators and resynchronization therapy on outcome in patientswith left ventricular dysfunction—a meta-analysis. Cardiology 2006;106:249–255.

8. Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews ofInterventions Version 5.0.2 (updated September 2009). The CochraneCollaboration, 2009. Available at: www.cochrane-handbook.org.

9. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–1558.

0. DerSimonian R, Laird N. Meta-analysis in clinical trials. ControlledClin Trials 1986;7:177–188.

1. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629–634.

2. Duval S, Tweedie R. Trim and fill: A simple funnel-plot-based methodof testing and adjusting for publication bias in meta-analysis. Biomet-rics 2000;56:455–463.

3. Auricchio A, Stellbrink C, Butter C, Sack S, Vogt J, Misier AR, BockerD, Block M, Kirkels JH, Kramer A, Huvelle E. Pacing therapies inCongestive Heart Failure II Study Group, Guidant Heart Failure ResearchGroup. Clinical efficacy of cardiac resynchronization therapy using leftventricular pacing in heart failure patients stratified by severity of ven-tricular conduction delay. J Am Coll Cardiol 2003;42:2109–2116.

4. Auricchio A, Stellbrink C, Block M, Sack S, Vogt J, Bakker P, Klein
H, Kramer A, Ding J, Salo R, Tockman B, Pochet T, Spinelli J. The
Pacing Therapies for Congestive Heart Failure Study Group. TheGuidant Congestive Heart Failure Research Group. Effect of pacingchamber and atrioventricular delay on acute systolic function of pacedpatients with congestive heart failure. Circulation 1999;99:2993–3001.

5. Bakker PF, Meijburg HW, de Vries JW, Mower MM, Thomas AC,Hull ML, Robles De Medina EO, Bredee JJ. Biventricular pacing inend-stage heart failure improves functional capacity and left ventric-ular function. J Interv Card Electrophysiol 2000;4:395–404.

6. Braunschweig F, Linde C, Gadler F, Ryden L. Reduction of hospitaldays by biventricular pacing. Eur J Heart Fail 2000;2:399–406.

7. Flevari P, Theodorakis G, Paraskevaidis I, Kolokathis F, KostopoulouA, Leftheriotis D, Kroupis C, Livanis E, Kremastinos DT. Coronaryand peripheral blood flow changes following biventricular pacing andtheir relation to heart failure improvement. Europace 2006;8:44–50.

8. Fung JW, Yip GW, Zhang Q, Fang F, Chan JY, Li CM, Wu LW, ChanGC, Chan HC, Yu CM. Improvement of left atrial function is associ-ated with lower incidence of atrial fibrillation and mortality aftercardiac resynchronization therapy. Heart Rhythm 2008;5:780–786.

9. Inage T, Yoshida T, Hiraki T, Ohe M, Takeuchi T, Nagamoto Y,Fukuda Y, Gondo T, Imaizumi T. Chronic cardiac resynchronizationtherapy reverses cardiac remodelling and improves invasive haemo-dynamics of patients with severe heart failure on optimal medicaltreatment. Europace 2008;10:379–383.

0. Knaapen P, van Campen LM, de Cock CC, Gotte MJ, Visser CA,Lammertsma AA, Visser FC. Effects of cardiac resynchronizationtherapy on myocardial perfusion reserve. Circulation 2004;110:646–651.

1. Kubanek M, Malek I, Bytesnik J, Fridl P, Riedlbauchova L, KarasovaL, Lanska V, Kautzner J. Decrease in plasma B-type natriuretic pep-tide early after initiation of cardiac resynchronization therapy predictsclinical improvement at 12 months. Eur J Heart Fail 2006;8:832–840.

2. Madaric J, Vanderheyden M, Van Laethem C, Verhamme K, Feys A,Goethals M, Verstreken S, Geelen P, Penicka M, De Bruyne B,Bartunek J. Early and late effects of cardiac resynchronization therapyon exercise-induced mitral regurgitation: relationship with left ventric-ular dyssynchrony, remodelling and cardiopulmonary performance.Eur Heart J 2007;28:2134–2141.

3. Mullens W, Bartunek J, Wilson Tang WH, Delrue L, Herbots L,Willems R, De Bruyne B, Goethals M, Verstreken S, VanderheydenM. Early and late effects of cardiac resynchronization therapy onforce-frequency relation and contractility regulating gene expression inheart failure patients. Heart Rhythm 2008;5:52–59.

4. Nelson GS, Curry CW, Wyman BT, Kramer A, Declerck J, Talbot M,Douglas MR, Berger RD, McVeigh ER, Kass DA. Predictors ofsystolic augmentation from left ventricular preexcitation in patientswith dilated cardiomyopathy and intraventricular conduction delay.Circulation 2000;101:2703–2709.

5. Sogaard P, Kim WY, Jensen HK, Mortensen P, Pedersen AK, Kris-tensen BO, Egeblad H. Impact of acute biventricular pacing on leftventricular performance and volumes in patients with severe heartfailure. A tissue Doppler and three-dimensional echocardiographicstudy. Cardiology 2001;95:173–182.

6. Vanderheyden M, Mullens W, Delrue L, Goethals M, Verstreken S,Wijns W, de Bruyne B, Bartunek J. Endomyocardial upregulation ofbeta1 adrenoreceptor gene expression and myocardial contractile re-serve following cardiac resynchronization therapy. J Card Fail 2008;14:172–178.

7. Waggoner AD, Rovner A, de las Fuentes L, Faddis MN, Gleva MJ,Sawhney N, Davila-Roman VG. Clinical outcomes after cardiac re-synchronization therapy: importance of left ventricular diastolic func-tion and origin of heart failure. J Am Soc Echocardiogr 2006;19:307–313.

8. Piccirillo G, Magri D, di Carlo S, De Laurentis T, Torrini A, MateraS, Magnanti M, Bernardi L, Barilla F, Quaglione R, Ettorre E, Mari-gliano V. Influence of cardiac-resynchronization therapy on heart rateand blood pressure variability: 1-year follow-up. Eur J Heart Fail2006;8:716–722.

9. Oakley C. Diagnosis and natural history of congested (dilated) cardio-myopathies. Postgrad Med J 1978;54:440–450.

0. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA,Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr,Roccella EJ, National Heart, Lung, and Blood Institute, Joint Na-
tional Committee on Prevention. Detection, evaluation, and treat-
http://www.cochrane-handbook.org

4

4


ment of high blood pressure. National High Blood PressureEducation Program Coordinating Committee. The Seventh Reportof the Joint National Committee on Prevention, Detection,Evaluation, and Treatment of High Blood Pressure: the JNC 7report. JAMA 2003;289:2560 –2572.

41. Cheng RK, Horwich TB, Fonarow GC. Relation of systolic bloodpressure to survival in both ischemic and nonischemic systolic heartfailure. Am J Cardiol 2008;102:1698–1705.

42. Lee TT, Chen J, Cohen DJ, Tsao L. The association between blood
pressure and mortality in patients with heart failure. Am Heart J2006;151:76–83.
43. Tanaka Y, Tada H, Yamashita E, Sato C, Irie T, Hori Y, Goto K,Iwamoto J, Manni H, Yokokawa M, Naito S, Oshima S, Taniguchi K.Change in blood pressure just after initiation of cardiac resynchroni-zation therapy predicts long-term clinical outcome in patients withadvanced heart failure. Circ J 2009;73:288–294.

4. Cohn JN. The management of chronic heart failure. N Engl J Med1996;335:490–498.

5. Marijon E, Boveda S, Chevalier P, Bulava A, Winter JB, Lambiez M, DefayeP, on behalf of the Mona Lisa Study Group. Monitoring of heart rate vari-
ability in heart failure patients with cardiac resynchronisation therapy: interestof continuous and didactic algorithm. Int J Cardiol 2010;144:166–169.

wb

G

Patient Perception Versus Medical Record Entry of Health-RelatedConditions Among Patients With Heart Failure

Adnan S. Malik, MDa, Grigorios Giamouzis, MD, PhDb, Vasiliki V. Georgiopoulou, MDa,Lucy V. Fike, MPHa, Andreas P. Kalogeropoulos, MDa, Catherine R. Norton, MDa,

Dan Sorescu, MDa, Sidra Azim, MDa, Sonjoy R. Laskar, MDa, Andrew L. Smith, MDa,Sandra B. Dunbar, RN, DSNc, and Javed Butler, MD, MPHa,*

A shared understanding of medical conditions between patients and their health careproviders may improve self-care and outcomes. In this study, the concordance betweenresponses to a medical history self-report (MHSR) form and the corresponding providerdocumentation in electronic health records (EHRs) of 19 select co-morbidities and habits in230 patients with heart failure were evaluated. Overall concordance was assessed using the� statistic, and crude, positive, and negative agreement were determined for each condi-tion. Concordance between MHSR and EHR varied widely for cardiovascular conditions(� � 0.37 to 0.96), noncardiovascular conditions (� � 0.06 to 1.00), and habits (� � 0.26 to0.69). Less than 80% crude agreement was seen for history of arrhythmias (72%), dyslip-idemia (74%), and hypertension (79%) among cardiovascular conditions and lung disease(70%) and peripheral arterial disease (78%) for noncardiovascular conditions. Perfectagreement was observed for only 1 of the 19 conditions (human immunodeficiency virusstatus). Negative agreement >80% was more frequent than >80% positive agreement fora condition (15 of 19 [79%] vs 8 of 19 [42%], respectively, p � 0.02). Only 20% of patientshad concordant MSHRs and EHRs for all 7 cardiovascular conditions; in 40% of patients,concordance was observed for <5 conditions. For noncardiovascular conditions, only 28%of MSHR-EHR pairs agreed for all 9 conditions; 37% agreed for <7 conditions. Cumula-tively, 39% of the pairs matched for <15 of 19 conditions. In conclusion, there is significantvariation in the perceptions of patients with heart failure compared to providers’ recordsof co-morbidities and habits. The root causes of this variation and its impact on outcomesneed further study. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:
569–572)
mstpn(wokpci

M

rpEMG

Heart failure (HF) prevalence is growing and primarilyaffects the elderly.1 The complex array of physiologic, psy-chological, social, and health care delivery issues that ac-company HF make it a difficult chronic disease to manage.2

Optimal self-care behavior is important for achieving thebest outcomes for chronic diseases such as HF. For patientsto actively participate in their care, however, it is importantfor them to have a clear understanding of their health-related problems.3 This is particularly critical for patients

ith HF, as they tend to be older, have a higher co-mor-idity burden, and often require complex treatment plans.4

From a provider perspective, medical record documentation

aCardiology Division, Emory University School of Medicine, Atlanta,Georgia; bDepartment of Cardiology, Larissa University Hospital, Larissa,

reece; and cEmory University School of Nursing, Atlanta, Georgia.Manuscript received August 13, 2010; revised manuscript received andaccepted October 11, 2010.

Funding: This project was funded by the Emory University Heart andVascular Board grant titled “The Atlanta Cardiomyopathy Consortium,”and supported in part by PHS grant (UL1 RR025008, KL2 RR025009 orTL1 RR025010) from the Clinical and Translational Science Award pro-gram, National Institutes of Health, National Center for Research Re-sources.

vE-mail address: [email protected] (J. Butler).

of disease states is an essential part of care provision.5 Thisis especially true in the current era of increasing use ofelectronic health records (EHRs), as many providers com-municate information exclusively through this medium.6 It

ay be assumed that what is documented in EHRs is theame as patients’ understanding and reporting. However, ifhis is not true, this discordance may lend itself to pooratient self-care behavior (related to not understanding orot reporting their conditions) or to insufficient medical caredue to misunderstanding by providers). In the current era,hether EHR entries are congruent with patients’ reportingf health-related conditions, and to what extent, is notnown. In this study, we sought to assess and compareatient self-report versus EHR documentation of cardiovas-ular and noncardiovascular conditions and behavioral hab-ts in patients with HF.

ethods

The data for this study were derived from patients en-olled in the Atlanta Cardiomyopathy Consortium. Thisrospective cohort study is enrolling patients from themory University Hospital, Emory University Hospitalidtown, and the Grady Memorial Hospital in Atlanta,eorgia. All patients undergo detailed medical history sur-
eys, electrocardiography, 6-minute walk tests, standard-
www.ajconline.org


nafrcttmts

1m

mdwbadedpTdhorT“m

AMWELIMILSDHCSHB

A

D

(


ized questionnaires, and collection of blood and urine sam-ples at baseline. Every 6 months, patients are contacted toassess outcomes, including interim medication changes,procedures, new disease diagnoses, and hospitalizations.Mortality data are collected through medical record review,information obtained from family members, and Social Se-curity Death Index query. The institutional review board hasapproved the study. At the time of this analysis, a total of238 patients were enrolled; we included 230 of these pa-tients (96.6%), excluding 8 patients who did not completemedical history surveys.

Research nurses abstracted data from the EHRs indepen-dently without discussion with the patients or their surveydocumentation. The main source of EHR data (n � 222[96.5%]) was Emory Healthcare’s electronic medical recordsystem, which is based on the Cerner Millennium platform(Cerner Corporation, Kansas City, Missouri). The systemprovides a comprehensive view of clinical data collectedacross hospitals and clinics. Data on 8 patients (3.5%) werecollected from the EHR system at the Grady MemorialHospital, which is based on the Siemens Medical Solutions(Malvern, Pennsylvania) Health Services platform.

All patients completed a medical history self-report(MHSR) form, which included questions regarding cardio-vascular conditions (history of heart attack or myocardialinfarction, high blood pressure or hypertension, high cho-lesterol, heart rhythm problems or arrhythmias, coronaryartery bypass graft surgery, coronary stent placement, andimplantable cardioverter defibrillator or pacemaker implan-tation), and noncardiovascular conditions (diabetes mellitus,peripheral arterial disease, pulmonary disease, liver disease,peptic ulcer disease, thyroid disease, cancer, osteoarthritis,and human immunodeficiency virus (HIV) infection). Thepulmonary disease question was open ended, allowing pa-tients to manually enter specific diagnoses. Data on historyof tobacco, alcohol, and cocaine use were also obtained.

To assess the reliability of EHR data abstraction, dataon 10% of the total charts, selected using a randomnumber generator (http://www.random.org), were inde-pendently abstracted. Cumulative agreement between the2 independent EHR data abstractions for all study vari-ables was 93.1%.

EHR data for each condition (yes or no) were comparedwith the data from MHSR forms (yes or no), and concor-dance was assessed using the � statistic. Crude, positive,and negative agreement were calculated to facilitate inter-pretation of � values.7 Crude agreement is equal to theumber of pairs that agree divided by the number of pairsvailable for analysis. The number of pairs available dif-ered for each condition because of missing values in patientesponses. Positive and negative agreement measures werealculated. The positive agreement measure is the ratio ofotal concordant positive responses over the average posi-ive responses of patients and EHRs. The negative agree-ent is the ratio of total concordant negative responses over

he average negative responses of patients and EHRs. Kappatatistics were interpreted as follows8: values of 0.93 to 1.00

denote almost perfect agreement, 0.81 to 0.92 very goodagreement, 0.61 to 0.80 substantial agreement, 0.41 to 0.60moderate agreement, 0.21 to 0.40 fair agreement, 0.01 to
0.20 slight agreement, and 0 no agreement. Finally, to f
summarize agreement by patient, the sum of the number ofconcordant conditions per participant was calculated. Therewere 7 cardiovascular and 9 noncardiovascular conditionsand 3 habits included in the summary measure. McNemar’sstatistic was calculated for paired comparisons. Finally,patients’ responses as “don’t know” to select conditionswere captured and compared with EHR data. All analyseswere performed using SAS version 9.2 (SAS Institute Inc.,Cary, North Carolina).

Results

The baseline patient characteristics and treatment patternare listed in Table 1. The mean age of patients was 56.6 �1.9 years; 64.5% were men, and 55.2% were white. Theean left ventricular ejection fraction was 39.3 � 14.6%.Table 2 lists the agreement data. There was fair agree-

ent for arrhythmia history and moderate agreement foryslipidemia and hypertension. The strongest agreementas noted for procedural care, including coronary arteryypass grafting and implantable cardioverter-defibrillatornd/or pacemaker implantation. For noncardiovascular con-itions, there was only fair agreement for pulmonary dis-ase; of the 47 of 82 patients (57%) who entered specificiagnoses, there was poor agreement for chronic obstructiveulmonary disease, asthma, and sleep apnea (not listed inable 2). There was very good agreement for cancer andiabetes mellitus and perfect agreement for HIV. For alco-ol use, there was fair agreement. In 80% of the patients (12f 15) in whom there was disagreement, the patients did noteport alcohol use when the EHRs suggested histories.here was moderate agreement for cocaine and tobacco use.Don’t know” responses were uncommon, including 13 foryocardial infarction (12 had no EHR documentation), 6

Table 1Baseline characteristics (n � 230)

Characteristic Value

ge (years) 56.6 � 11.9ale 149 (64.8%)hite 127 (55.2%)

ducation (years) 14.1 � 3.1iving alone 41 (17.9%)

nsured 212 (92.2%)arried 143 (62.2%)

schemic cause of HF 69 (31.3%)eft ventricular ejection fraction (%) 39.3 � 14.6ystolic blood pressure (mm Hg) 112 � 18iastolic blood pressure (mm Hg) 71 � 11eart rate (beats/min) 72 � 11reatinine (mg/dl) 1.4 � 1.1odium (mEq/L) 138 � 3emoglobin (g/dl) 13.3 � 1.8rain natriuretic peptide (ng/L) 202 (73–664)

�-blocker use 219 (94.8%)ngiotensin-converting enzyme inhibitor orangiotensin receptor blocker use

197 (85.6%)

efibrillator/pacemaker 145 (64.5%)

Data are expressed as mean � SD, number (percentage), or medianinterquartile range).

or stents and 1 for coronary bypass surgery (all with no

http://www.random.org

dc3

fvt(tdmifmcq[4

D

pcobcpacwd

N

B

ol Educ

571Heart Failure/Patient Perception Versus Medical Record Entry

EHR entries), and 2 for diabetes mellitus (1 had EHRdocumentation).

A summary measure of patient versus EHR agreementis shown in Figure 1. Of the 7 cardiovascular conditions,only 20% of patients had concordant MHSR responses and

Figure 1. Summary measures of patient versus EHR concordance forco-morbidities. Suboptimal proportional concordance was noted for car-diovascular and noncardiovascular co-morbidities and for the behavioralhabits assessed.

Table 2Agreement for cardiovascular and noncardiovascular conditions

Co-Morbidity Number Yes/Yes No/No Yes/No N

CardiovascularArrhythmia 191 103 35 36Dyslipidemia* 223 94 71 34Hypertension 224 127 49 26Stent 221 15 185 21Myocardial infarction 196 50 123 16Coronary bypass surgery 225 39 183 3Defibrillator/pacemaker 221 140 77 3oncardiovascularPeptic ulcer disease 218 1 198 4Peripheral arterial disease 230 4 176 0Lung disease 203 29 113 39Osteoarthritis 215 10 174 23Liver disease 225 3 212 2Thyroid 214 26 174 7Cancer 229 34 185 8Diabetes mellitus 228 69 144 6HIV infection 182 1 181 0ehavioralExcess alcohol use 218 3 200 3Cocaine abuse 224 11 198 14Tobacco 203 64 108 31

Kappa p values �0.001 for all conditions except peripheral arterial dare by patient report first, followed by EHR documentation.

* Use of lipid-lowering medications or fulfilling the National Cholester

EHR entries for all conditions, and 40% agreed for �5 con- o

itions. For noncardiovascular conditions, 28% agreed for all 9onditions, and 37% agreed for �7 conditions. Cumulatively,9% of the pairs matched for �15 of 19 co-morbidities.

For cardiovascular conditions, positive agreement rangedrom 59% (stent placement) to 99% (implantable cardio-erter-defibrillator and/or pacemaker implantation). Nega-ive agreement ranged from 57% (arrhythmias) to 99%coronary bypass surgery). For noncardiovascular condi-ions, positive agreement ranged from 9.5% (peptic ulcerisease) to 100% (HIV infection), whereas negative agree-ent ranged from 79% (pulmonary disease) to 100% (HIV

nfection). Positive agreement for habits ranged from 29%or alcohol use to 81% for tobacco use, and negative agree-ent ranged from 87% for tobacco to 96% for alcohol and

ocaine use. Negative agreement of �80% was more fre-uent than positive agreement (15 of 19 [79%] vs 8 of 1942%] conditions, respectively, p � 0.02). Approximately0% of pairs were discordant for �4 conditions.

iscussion

In this study, we observed considerable variability inatient report versus EHR entry of a range of medicalonditions and habits, including many conditions for whichptimization of care and outcomes requires participation onehalf of patients. Agreement was expectedly better foronditions involving interventions (e.g., defibrillator im-lantation, coronary bypass surgery). We found better neg-tive agreement between MHSR and EHRs (i.e., when theondition was absent) than positive (i.e., when the conditionas present). These results provide insights into an un-erstudied area of health care delivery that may influence

Crude Agreement Positive Agreement Negative Agreement �

72% 80% 57% 0.3774% 76% 71% 0.4879% 84% 67% 0.5190% 59% 95% 0.5488% 81% 91% 0.7399% 97% 99% 0.9598% 99% 98% 0.96

91% 10% 95% 0.0678% 14% 88% 0.1170% 49% 79% 0.2886% 39% 92% 0.3296% 38% 98% 0.3693% 79% 96% 0.7596% 87% 97% 0.8593% 90% 95% 0.85

100% 100% 100% 1.00

93% 29% 96% 0.2693% 59% 96% 0.5685% 81% 87% 0.69

(p � 0.006) and peptic ulcer disease (p � 0.27). All yes and no citations

ation Program Adult Treatment Panel III criteria.

o/Yes

1724220701

155022887290

1210

isease

utcomes. The accurate capture of patient information by

capi

ihsntfassteEscctsApcrmf


EHRs depends on patient awareness and the documenta-tion practices of providers; however, the effectivenesswith which patient information is captured is unknown.O’Malley et al9 surveyed physicians with EHR experience,hief medical officers of EHR vendors, and thought leadersnd showed a significant gap between policy makers’ ex-ectations and clinicians’ assessments of EHR as a tool tomprove care coordination.

A large portion of health care costs and hospitalizationsn HF are related not only to worsening HF but also to theigh burden of co-morbidities seen in these patients. Braun-tein et al10 showed that 39% of patients with HF had �5oncardiac co-morbidities, and only 4% had none. Impor-antly, patients with HF with �5 co-morbidities accountedor 81% of total inpatient days. Given the burgeoning costnd poor outcomes for patients with HF, increasing empha-is is being placed on nonpharmacologic care, includingelf-care, as a mechanism for improving outcomes.11 Pa-ients’ recognition of their medical conditions is critical toffective self-care. Our study showed that 37% of MHSR-HR pairs exhibited less than moderate agreement, under-coring a problem as well as an opportunity for improvingare. Many conditions had agreement of �80%. Thereould be multiple explanations for these results (e.g., pa-ients may not fully understand the terminology or theignificance of a disease or meaning of their symptoms).lternatively, providers may not be documenting or askingatients pertinent questions. We also observed a high dis-ordance for history of lipid abnormalities, peripheral arte-ial disease, and hypertension; these co-morbidities com-only accompany HF, and patient participation is important

or optimal treatment.12,13 Similarly, coexisting pulmonarydisease may exacerbate or be confused with HF symptomsand affects HF prognosis and treatment options.14 Manyinstances of cocaine and tobacco use were not mentioned inthe medical records. The inability to identify these be-haviors naturally leads to inadequate patient counselingon the importance of cessation. Also, a significant seg-ment of patients denied alcohol use that was neverthelessdocumented in their EHRs. Interestingly, we noted betternegative (absence of disease) as opposed to positive(presence of disease) agreement. Whether this is relatedto perceptual challenges on behalf of patients, lack ofdocumentation by providers, incomplete co-morbidityclassification, or the variable prevalence of different dis-ease states needs further study.

This study was limited by its size and by the fact that data
were collected at a single academic medical center.
1. Lloyd-Jones D, Adams R, Brown T, Carnethon M, Dai S, DeSimone G, Ferguson T, Ford E, Furie K, Gillespie C, Go A,Greenlund K, Haase N, Hailpern S, Ho P, Howard V, Kissela B,Kittner S, Lackland D, Lisabeth L, Marelli A, McDermott M, MeigsJ, Mozaffarian D, Mussolino M, Nichol G, Roger V, Rosamond W,Sacco R, Sorlie P, Stafford R, Thom T, Wasserthiel-Smoller S,Wong N, Wylie-Rosett J. Heart disease and stroke statistics—2010update: a report from the American Heart Association. Circulation2010;121:e46 – e215.

2. Liu L. Changes in cardiovascular hospitalization and comorbidity ofheart failure in the United States: findings from the National HospitalDischarge Surveys 1980–2006. Int J Cardiol In press.

3. Epstein R, Alper B, Quill T. Communicating evidence for participa-tory decision making. JAMA 2004;291:2359–2366.

4. Baker D, Asch S, Keesey J, Brown J, Chan K, Joyce G, Keeler E.Differences in education, knowledge, self-management activities, andhealth outcomes for patients with heart failure cared for under thechronic disease model: the improving chronic illness care evaluation.J Card Fail 2005;11:405–413.

5. Bayliss EA, Ellis JL, Steiner JF. Subjective assessments of comorbid-ity correlate with quality of life health outcomes: initial validation ofa comorbidity assessment instrument. Health Qual Life Outcomes2005;3:51.

6. Smith P, Araya-Guerra R, Bublitz C, Parnes B, Dickinson L, VanVorst R, Westfall J, Pace W. Missing clinical information duringprimary care visits. JAMA 2005;293:565–571.

7. Cicchetti DV, Feinstein AR. High agreement but low kappa: II. Re-solving the paradoxes. J Clin Epidemiol 1990;43:551–558.

8. Landis JR, Koch GG. The measurement of observer agreement forcategorical data. Biometrics 1977;33:159–174.

9. O’Malley AS, Grossman JM, Cohen GR, Kemper NM, Pham HH. Areelectronic medical records helpful for care coordination? Experiencesof physician practices. J Gen Intern Med 2010;25:177–185.

10. Braunstein JB, Anderson GF, Gerstenblith G, Weller W, Niefeld M,Herbert R, Wu AW. Noncardiac comorbidity increases preventablehospitalizations and mortality among Medicare beneficiaries withchronic heart failure. J Am Coll Cardiol 2003;42:1226 –1233.

11. Riegel B, Moser DK, Anker SD, Appel LJ, Dunbar SB, Grady KL,Gurvitz MZ, Havranek EP, Lee CS, Lindenfeld J, Peterson PN,Pressler SJ, Schocken DD, Whellan DJ. State of the science: pro-moting self-care in persons with heart failure: a scientific statementfrom the American Heart Association. Circulation 2009;120:1141–1163.

12. Velagaleti RS, Massaro J, Vasan RS, Robins SJ, Kannel WB, Levy D.Relations of lipid concentrations to heart failure incidence: the Fra-mingham Heart Study. Circulation 2009;120:2345–2351.

13. Kostis JB, Davis BR, Cutler J, Grimm RH Jr, Berge KG, Cohen JD,Lacy CR, Perry HM Jr, Blaufox MD, Wassertheil-Smoller S, BlackHR, Schron E, Berkson DM, Curb JD, Smith WM, McDonald R,Applegate WB; SHEP Cooperative Research Group. Prevention ofheart failure by antihypertensive drug treatment in older persons withisolated systolic hypertension. JAMA 1997;278:212–216.

14. Iversen KK, Kjaergaard J, Akkan D, Kober L, Torp-Pedersen C,Hassager C, Vestbo J, Kjoller E. The prognostic importance of lungfunction in patients admitted with heart failure. Eur J Heart Fail
2010;12:685–691.

tdaBrw

Pv2

0d

Effectiveness of Serial Increases in Amino-Terminal Pro–B-TypeNatriuretic Peptide Levels to Indicate the Need for Mechanical

Circulatory Support in Children With AcuteDecompensated Heart Failure

Derek T.H. Wong, MDa, Kristen George, MScNb, Judith Wilson, MScNb, Cedric Manlhiot, BScb,Brian W. McCrindle, MDb, Khosrow Adeli, MD, PhDb, and Paul F. Kantor, MBBChb,*

We sought to determine prospectively whether serial assessment of the natriureticpeptide prohormone, amino-terminal pro–B-type natriuretic peptide (NT–pro-BNP),correlated with clinical severity and outcomes in children hospitalized for acute de-compensated heart failure (ADHF). Patients (>1 month of age) admitted from 2005 to2007 with ADHF requiring intravenous vasoactive/diuretic therapy for ADHF wereeligible. Serum NT–pro-BNP levels were obtained within 24 hours of admission and atprespecified intervals, and clinical caregivers were blinded to these levels. End pointsincluded hospital discharge, death or cardiac transplantation, and care escalationincluding the need for mechanical circulatory support (MCS) was noted. Twenty-fourpatients were enrolled: 22 survived to hospital discharge and 2 died. Ten required MCS(of which 6 underwent cardiac transplantation). Two patients underwent transplanta-tion without MCS. For the entire cohort, NT–pro-BNP levels peaked at days 2 to 3 afteradmission, with a subsequent gradual decrease until discharge. However, for those whodid require MCS, NT–pro-BNP failed to decrease consistently until after MCS initia-tion. At discharge, NT–pro-BNP levels were significantly decreased from admissionlevels but remained well above normal for all patients. Single-point NT–pro-BNP levelson admission did not correlate with independently assessed clinical scores of heartfailure severity or predict the need for MCS in this cohort. In conclusion, serialNT–pro-BNP levels demonstrated an incremental trend after 48 hours in patients whowent on to require MCS but decreased in all other patients and may therefore assist thedecision to initiate or avoid MCS after admission for pediatric ADHF. © 2011
fo

M

wrmammoaas(yiPecvg

In children, acute decompensated heart failure (ADHF)due to cardiomyopathy or failed congenital heart diseaserepair constitutes a common indication for cardiac trans-plantation.1 Admission with severe HF frequently raisesconcerns that a patient has declared the need for transplan-tation assessment.2 In this regard, recent advances havemade use of mechanical circulatory support (MCS) in pe-diatrics a more popular and feasible method of bridging totransplantation or to recovery.3 However, the appropriateiming of MCS device placement is sometimes difficult toetermine. We investigated whether initial assessment ofmino-terminal pro–B-type natriuretic peptide (NT–pro-NP) in children admitted for management of ADHF cor-

elated with formal assessment of clinical status, andhether serial assessment of NT–pro-BNP might be a use-

aDepartment of Pediatrics, Division of Pediatric Cardiology, Children’sHospital of Eastern Ontario, Ottawa, Ontario, Canada; bDepartment of

ediatrics, Labatt Family Heart Centre, Hospital for Sick Children, Uni-ersity of Toronto, Toronto, Ontario, Canada. Manuscript received January5, 2010; revised manuscript received and accepted October 11, 2010.

oE-mail address: [email protected] (P.F. Kantor).

ul tool to predict the need for MCS in children with ADHFf diverse causes.

ethods

This was a prospective observational study conductedith the approval of the Hospital for Sick Children (To-

onto, Ontario, Canada) research ethics board. Patients ad-itted to this institution from May 2005 to July 2007 withdiagnosis of ADHF requiring escalation of HF manage-ent involving a need for intravenous diuretics, inotropicedication, or MCS were eligible for enrollment. Diagnosis

f HF was made by the clinician responsible for admissionnd not the research team. Two of the investigators (D.W.nd P.K.) adjudicated each case for eligibility based onpecific criteria before enrollment. Inclusion criteria were1) a clinical diagnosis of ADHF, (2) age �1 month to �18ears at admission, and (3) escalation of HF therapy withntravenous diuretic therapy, inotropic medication, or MCS.atients with congenital or acquired heart disease werequally eligible. Exclusion criteria included (1) plannedorrective cardiovascular surgery or catheter-based inter-ention on the same admission (to avoid the effect of sur-ical or catheter intervention as a confounder of patient
utcome) and (2) patients �1 month of age because their
www.ajconline.org


fest

aaad�thtitp

c1vaBtctac

mNtcla(tKafmvtCwpoReoicapow(p

ng outc


NT–pro-BNP levels are known to be increased as part of thenormal postnatal circulatory adaptation.4–6 All patients whomet the criteria for enrollment were approached and in-formed consent was obtained.

On admission, demographic information, underlying car-diac diagnosis, laboratory investigations (i.e., biochemistry,radiologic studies, electrocardiograms), and echocardio-graphic data were recorded by the research team. A briefsymptomatic history and physical examination were per-formed at time of admission and at time of discharge bymembers of the research team (D.W., J.W., and K.G.). Apatient’s clinical course in hospital—length of stay, dis-charge status and medications—and need for inotropic sup-port, diuretics, vasoactive medications, antiarrhythmics, andanticoagulation were recorded. We noted whether the out-come of each admission was 1 of the predefined endpoints of discharge, death, or heart transplantation. Otherclinically relevant events were also defined and notedprospectively, including CCCU admission, mechanicalventilation, and inotropic and vasoactive medication us-age. At time of admission and at time of attaining adefined end point, the previously validated New YorkUniversity Pediatric Heart Failure Index (NYU-PHFI)score, Ross classification, and where age appropriate theNew York Heart Association (NYHA) score were deter-mined.7,8 To isolate the effect of the NT–pro-BNP levelrom clinical severity assessment and the decision toscalate hemodynamic support, clinicians and the re-earch team were blinded to the NT–pro-BNP data untilhe study was completed.

NT–pro-BNP levels were obtained within 24 hours ofdmission, at intervals of 2 to 3, 6 to 8, and 13 to 15 daysfter admission, and weekly thereafter. Patients who weredmitted to the critical care unit had NT–pro-BNP levelsrawn daily. If patients were on stable long-term MCS for1 week, NT–pro-BNP levels were drawn weekly while in

he critical care unit. A final sample was drawn within 48ours of planned discharge or immediately before heartransplantation. Samples were stored in clotted blood spec-men containers at �80°C and analyzed after completion ofhe entire study by our clinical laboratory as a research

Figure 1. Flow diagram depicti

rotocol. Assays were performed on an Elecsys electro- C

hemiluminescent immunoassay system using Elecsys010/2010 immunoassay analyzers (Roche Diagnostics, La-al, Quebec, Quebec, Canada) with appropriate calibrationnd control methods.9 Because neither NT–pro-BNP norNP levels were routinely measured in our institution at the

ime of this study, no patient received any clinically indi-ated BNP or NT–pro-BNP assay during admission. Inves-igators were also blinded to results of the NT–pro-BNPssays until after the clinical component of the study wasompleted.

Data are presented as mean � SD, medians with mini-um and maximum values, and frequencies as appropriate.T–pro-BNP levels over time are represented as box-plots

o express the severely skewed level distribution. To ac-ount for the skewed distribution of NT–pro-BNP, a naturalogarithmic transformation was applied to NT–pro-BNP inll analyses. Basic comparisons between end point groupsMCS vs no MCS) were obtained through Fisher’s exactest, Student’s t test with Satterthwaite correction, andruskal-Wallis analysis of variance for continuous vari-

bles with skewed distribution. Factors associated with needor MCS were sought in univariable logistic regressionodels using need for MCS before discharge as a binary

ariable. No multivariable modeling was attempted due tohe limited number of patients enrolled in the study.hanges in NT–pro-BNP over time and associated factorsere assessed in linear regression models adjusted for re-eated measurements over time to an autoregressive (firstrder with model-based estimation of covariance structure).egression parameters were estimated using generalizedstimating equations. Mean values were used for imputationf missing variables when necessary. Time since admissionn days was treated as a continuous variable. An interactionriterion was created between study groups (MCS vs not)nd time to estimate differences in rate of change in NT–ro-BNP for each group. Because of the small numberf patients who died or required heart transplantation weere not able to search for predictors of negative outcomes

even as a combined end point). All statistical analyses wereerformed using SAS 9.1 (SAS Institute, Cary, North

omes for entire patient cohort.

arolina).

CLA

egated

575Heart Failure/Serial NT–pro-BNP in Pediatric Heart Failure

Results

The clinical pathway overview for the cohort is dia-grammed in Figure 1.

In all, 24 consecutive patients consented to enroll-ment. Median age (range) was 8.7 years (0 to 17.7) andmedian weight was 29 kg (3 to 80). Cardiomyopathy wasthe underlying cause of HF in 17 cases, including idio-pathic (8), myocarditis (2), ischemic (1), arrhythmia-induced (2), and unclassified (4) phenotypes. Structuralcongenital heart disease was present in 4 patients, with 3other patients including 1 with post-transplant rejectionand hemodynamic failure.

Eighteen patients were in Ross/NYHA class III or IV atadmission, with 5 in class II. One patient (an infant with amitochondrial cardiomyopathy) appeared clinically asymp-tomatic to the research team but later deteriorated precipi-tously, requiring MCS. Most patients showed a change insymptomatic status between admission and the end point,with all those dying or requiring transplantation reachingstage NHYA/Ross class IV before that end point. Medianejection and shortening fractions, as measured by echocar-diography on admission for the entire cohort, were 23%(range 6 to 64) and 14% (range 3 to 50), respectively. Only7 patients were treated with diuretic therapy alone, followedby initiation of oral angiotensin-converting enzyme inhib-iter and/or � blocker. Most patients in this cohort (17 of 24)required inotropic or inodilator medications—17 receivedmilrinone, 9 received epinephrine, 6 received vasopressin, 4received norepinephrine, and 3 received dobutamine.Eleven patients required a combination of multiple inotropicand/or vasoactive medications. No patients received nesirit-ide, which was unavailable in Canada at the time of thisstudy. Three patients received digoxin during their hospi-talization.

Twelve patients (50%) were admitted to the critical careunit, and all of these required mechanical ventilation. Tenpatients required MCS, of which 7 were managed withextracorporeal membranous oxygenation and 3 patients re-

Table 1Clinical and laboratory data at presentation and serial median B-type natrcirculatory support at prespecified intervals

Patient Characteristics

New York University Pediatric Heart Failure Index 13Ross/New York Heart Association classification 3.5Lactate (mmol/L) 7.8pH 7.25

reatinine (�mol/L) 89eft ventricular ejection fraction (%) 17mino-terminal pro–B-type natriuretic peptide

(pg/ml), median (range)On admission 30,4062–3 days 39,6536–8 days 26,51013–15 days 16,245At discharge/exit 3,984Maximum 40,713

Note that patients requiring mechanical circulatory support are not segr

quired the Berlin Heart EXCOR (Berlin Heart, Berlin, Ger-

many) ventricular assist device. Differences in clinical pre-sentation characteristics between those patients who wenton to require MCS and those who did not are presented inTable 1. Patients who eventually required MCS had a sig-nificantly lower pH and a higher lactate and creatinine onadmission than those patients who did not require MCS.In the group requiring MCS median time between admis-sion and initiation of support was 13 days (range 0 to 36).Mean duration on MCS was 26 � 33 days (median 17,range 2 to 107). Eight patients underwent heart transplan-tation, 6 of whom underwent bridging by MCS before theirheart transplantation. Two patients died—1 patient imme-diately after heart transplantation (after needing MCS forfailed single-ventricle palliation) and the other patient hav-ing care withdrawn after a large intracranial hemorrhagewhile on MCS for HF due to dilated cardiomyopathy. Over-all survival to hospital discharge was 92% for this cohort,

Figure 2. Trend of amino-terminal pro–B-type natriuretic peptide (pi-cograms per milliliter) over time for entire cohort. ADMIT � admis-sion; D2–3 � days 2 to 3 of admission; D7–10 � days 7 to 10 ofadmission; D13–15 � days 13 to 15 of admission; DC � time ofdischarge from study.

peptide levels in patients who did or did not require mechanical

S No MCS p Value10) (n � 14)

) 13 (9–21) NS3 (1–4) 0.06

2.6) 2.0 (1.2–5.8) 0.027.41) 7.41 (7.28–7.46) 0.01

50) 46 (4–83) 0.02) 30 (9–64) 0.05

0–63,862) 10,743 (3,760–62,221) 0.033–87,899) 9,262 (3,094–70,580) 0.01–39,252) 8,058 (568–63,847) 0.03–43,740) 8,594 (1,070–32,775) 0.12

48,397) 8,933 (3,612–38,270) 0.564–87,899) 15,680 (7,111–70,580) 0.001

into before or after mechanical circulatory support in this tabulation.

iuretic

MC(n �

(1–24(2–4)(3.3–1(7.03–(31–1(9–41

(11,40(20,79(3,529(3,204(766–(25,20

with a transplant-free survival to discharge of 63%.

a

cidrw1rnotal(pd

(picogra


Serum NT–pro-BNP levels were available for all pa-tients. Duration of hospitalization varied significantly de-pending on the clinical course of each patient (range 2 to125 days), and as a result the number of NT–pro-BNPmeasurements per patient also varied. Sixteen patients hadmeasurements that spanned all 5 specified intervals. Thetrend of NT–pro-BNP over time for the entire cohort isdisplayed in Figure 2. It can be appreciated that the level ofNT–pro-BNP was 100 to 1,000 times the upper limit ofnormal (normal range �200 ng/ml). Despite a carefully

Figure 3. Plots of amino-terminal pro–B-type natriuretic peptide

Figure 4. Trend of amino-terminal pro–B-type natriuretic peptide (pico-grams per milliliter) over time for the first 14 days of admission for patientssegregated according to final outcome of requiring mechanical circulatorysupport. On admission those who were later to require mechanical circu-latory support form a single cohort that then diverges into 2 curves overtime as patients are placed onto mechanical circulatory support. Over time,as patients require mechanical circulatory support, their amino-terminalpro–B-type natriuretic peptide levels no longer contribute to the “NeedMCS—Prior to Support” curve and contribute to the “Needed MCS—OnSupport” curve as patients move from 1 curve to the other (p �0.01 for the2 groups before mechanical circulatory support and for those who requiredmechanical circulatory support comparing levels before and after initiationof mechanical circulatory support).

conducted clinical severity assessment, we found that the

correlation coefficient between HF symptom scores at ad-mission and NT–pro-BNP levels was weak (Figure 3). Thisapplied to the NYHA/Ross class status (r � 0.15, p � 0.52)nd the NYU-PHFI (r � 0.17, p � 0.46).

After admission we observed that NT–pro-BNP levelsontinued to increase for the overall cohort initially, peak-ng at days 2 to 3 of hospitalization, with an overall gradualecrease over time thereafter. Peak level of NT–pro-BNPecorded in patients who required MCS during admissionas 40,713 pg/ml (range 25,204 to 87,899) compared to5,680 pg/ml (range 7,111 to 70,580) for those who did notequire MCS (p � 0.001). Although this difference is sig-ificant, absolute peak level was not found to be predictivef the outcome of MCS, in part because the time at whichhe peak occurred was not consistent. In regression modelsdjusted for repeated measurements, serial NT–pro-BNPevels decreased in those patients who did not require MCS�953 pg/ml/day), but not for those who were eventuallylaced on MCS, before the initiation of MCS (�601 pg/ml/ay, p � 0.04, vs no MCS; Figure 4). For those patients who

required MCS, NT–pro-BNP decreased only after beingplaced on MCS, with a significantly different rate of changeover time compared to previously being on MCS (�1,354pg/ml/day, p �0.001, vs before MCS).

At time of discharge, levels of NT–pro-BNP were sig-nificantly lower in all surviving patients compared to ad-mission. However, these levels were still well above thenormal reference range, even in patients with a normaliza-tion of ejection fraction. At time of discharge NT–pro-BNPlevels did not correlate with Ross classification score,NYHA class, or NYU-PHFI score (Figure 3). Due to thesmall numbers of deaths/cardiac transplantations, it was notpossible to determine in this cohort whether NT–pro-BNPlevels were predictive of death or need for transplantation asa composite end point.

Discussion

In this study, we investigated whether initial assessmentof NT–pro-BNP levels correlated well with HF severity at

ms per milliliter) versus clinical scores of heart failure severity.

admission and whether serial assessment of NT–pro-BNP

ailNpimerasp

vMelfmcwt

wwa

ppd

ppssocnla

erwtocBMdpoMlit

NliosecNi

dwhnmwp

ccstdsptpcsumodi

Ar

577Heart Failure/Serial NT–pro-BNP in Pediatric Heart Failure

levels indicated the important clinical outcome of MCSrequirement. Although we were able to demonstrate a mar-ginal difference in the independently administered NYHA/Ross score (p � 0.06) on admission between those who didnd those who did not go on to require MCS, the differencen these scores (3.0 vs 3.5) was not very marked and wouldikely not be clinically discernable. The more detailedYU-PHFI, which incorporates medical therapies andhysical signs and symptoms, was also not discriminatoryn this setting for the primary outcome of MCS when ad-inistered independently of the clinical team. This finding

mphasizes the limitation of subjective and patient-/parent-eported functional status (for the Ross/NYHA score) butlso indicates that a single-point assessment incorporatingeveral objective parameters may not be very useful inredicting the course of ADHF in children.

Serial assessment of NT–pro-BNP levels, however, re-ealed important differences between patients who requiredCS and those who did not. We also noted significant differ-

nces in single-point (at admission) assessments of arterial pH,actate, creatinine and to a lesser extent left ventricular ejectionraction in children who later required MCS. These “traditionaleasurements” may have played a significant role in the de-

ision-making process for MCS by the clinical caregivers, ande cannot comment on their independent predictive value for

his outcome given the small sample.Although the levels of NT–pro-BNP we encountered

ere 100 to 1,000 times the established normal ranges andere extremely increased compared to published values of

dults with congestive HF,10,11 other pediatric studies (us-ing the same assay method) have shown similar extremeincreases in NT–pro-BNP.12–14 Single time-point natriureticeptide levels in the outpatient context have shown goodrognostic value for later hospitalization, transplantation, oreath.15 In contrast to Ratnasamy et al16 and other investi-

gators,6,17 we found that single-point assessment of NT–ro-BNP on admission did not correlate closely with inde-endently administered clinical scores of congestive HFeverity. This loss of discriminatory power may reflect thekewed increase in acuity and severity of congestive HF inur cohort of patients: all required hospitalization for theirongestive HF and 50% required mechanical ventilation. Ofote, no patients in our study previously had NT–pro-BNPevels measured and 75% of patients had not been followedt all in an HF clinic setting.

Our data suggest that interpretation of NT–pro-BNP lev-ls becomes more complex when admission for ADHF isequired. Although maximal attained levels of NT–pro-BNPere higher in patients who required MCS during admission

han those who did not, absolute levels were not predictivef this outcome in a regression model. We found a statisti-ally significant difference in rate of change of NT–pro-NP between those who did and those who did not requireCS. NT–pro-BNP levels decreased in those patients who

id not require MCS, but not for those who were eventuallylaced on MCS, before initiating MCS. After being placedn MCS the rate of decrease was faster after being placed onCS than that noted before being on MCS (p �0.001). The

ack of decrease of NT–pro-BNP before MCS may be anndicator of failure of medical management to successfully
reat the HF. We speculate that the rate of decrease of
T–pro-BNP may be related to the degree of volume un-oading of the left ventricle, as has been described by Milt-ng et al18 in adult subjects. We found that for those patientsn MCS a more rapid decrease in NT–pro-BNP was ob-erved after MCS than before, possibly representing morefficient ventricular decompression. As a result, we con-lude that after admission for ADHF, serial assessments ofT–pro-BNP, rather than its absolute level, may be a useful

ndicator of patient risk in pediatric ADHF.This is supported by recent data of Heise et al14 who

showed a dramatic and sustained decrease in NT–pro-BNP levels within 1 week of EXCOR implantation.Moreover, Sodian et al19 demonstrated that the rate of

ecrease of BNP was more rapid in those who wereeaned off MCS compared to those who died or requiredeart transplantation. Due to our small numbers, we wereot able to determine whether use of extracorporealembranous oxygenation versus the EXCOR correlatedith a greater or lesser decrease after MCS in NT–ro-BNP.

Limitations: Severe ADHF in children remains an un-ommon scenario beyond the newborn period, and thisohort does not reflect the variety of symptom severitieseen in a more stable outpatient population already onherapy. Therefore, the ability of NT–pro-BNP levels toiscriminate between symptom severity groups in a broaderetting is not refuted by our data. Also, this study was notowered to assess the predictive value of NT–pro-BNP forhe outcome of death or transplantation because this endoint was too infrequent. Larger patient numbers or thereation of a pediatric ADHF registry may allow for greatertatistical power that will allow clinicians to determine thetility of NT–pro-BNP in the various phases of pediatric HFanagement. Our population was predominantly made up

f children with cardiomyopathies. Application of theseata to children with complex congenital heart lesions andn particular Fontan circulation is uncertain.

cknowledgment: The researchers thank Roche Laborato-ies for donating the NT–pro-BNP assays.

1. Kirk R, Edwards LB, Aurora P, Taylor DO, Christie JD, Dobbels F,Kucheryavaya AY, Rahmel AO, Stehlik J, Hertz MI. Registry of theinternational society for heart and lung transplantation: twelfth officialpediatric heart transplantation report-2009. J Heart Lung Transplant2009;28:993–1006.

2. Kantor PF, Mertens LL. Clinical practice: heart failure in children. PartI: clinical evaluation, diagnostic testing, and initial medical manage-ment. Eur J Pediatr 2009;169:269–279.

3. Kirklin JK. Mechanical circulatory support as a bridge to pediatriccardiac transplantation. Semin Thorac Cardiovasc Surg Pediatr CardSurg Annu 2008;80–85.

4. Mir TS, Marohn S, Laer S, Eiselt M, Grollmus O, Weil J. Plasmaconcentrations of N-terminal pro-brain natriuretic peptide in controlchildren from the neonatal to adolescent period and in children withcongestive heart failure. Pediatrics 2002;110(suppl):e76.

5. Koch A, Singer H. Normal values of B type natriuretic peptide ininfants, children, and adolescents. Heart 2003;89:875–878.

6. Mir TS, Laux R, Hellwege HH, Liedke B, Heinze C, von Buelow H,Laer S, Weil J. Plasma concentrations of aminoterminal pro atrialnatriuretic peptide and aminoterminal pro brain natriuretic peptide inhealthy neonates: marked and rapid increase after birth. Pediatrics
2003;112:896–899.

1

1

1


7. Ross RD, Daniels SR, Schwartz DC, Hannon DW, Shukla R, KaplanS. Plasma norepinephrine levels in infants and children with conges-tive heart failure. Am J Cardiol 1987;59:911–914.

8. Connolly D, Rutkowski M, Auslender M, Artman M. The New YorkUniversity Pediatric Heart Failure Index: a new method of quanti-fying chronic heart failure severity in children. J Pediatr 2001;138:644 – 648.

9. Yeo KT, Dumont KE, Brough T. Elecsys NT-ProBNP and BNPassays: are there analytically and clinically relevant differences?J Card Fail 2005;11(suppl):S84–S88.

10. James SK, Lindahl B, Siegbahn A, Stridsberg M, Venge P, ArmstrongP, Barnathan ES, Califf R, Topol EJ, Simoons ML, Wallentin L.N-terminal pro-brain natriuretic peptide and other risk markers for theseparate prediction of mortality and subsequent myocardial infarctionin patients with unstable coronary artery disease: a Global Utilizationof Strategies To Open occluded arteries (GUSTO)-IV substudy. Cir-culation 2003;108:275–281.

11. Westerhout CM, Fu Y, Lauer MS, James S, Armstrong PW, Al-HattabE, Califf RM, Simoons ML, Wallentin L, Boersma E. Short- andlong-term risk stratification in acute coronary syndromes: the addedvalue of quantitative ST-segment depression and multiple biomarkers.J Am Coll Cardiol 2006;48:939–947.

12. Cohen S, Springer C, Avital A, Perles Z, Rein AJ, Argaman Z, Nir A.Amino-terminal pro-brain-type natriuretic peptide: heart or lung disease inpediatric respiratory distress? Pediatrics 2005;115:1347–1350.

13. Fried I, Bar-Oz B, Perles Z, Rein AJ, Zonis Z, Nir A. N-terminal
pro–B-type natriuretic peptide levels in acute versus chronic left ven-tricular dysfunction. J Pediatr 2006;149:28–31.
14. Heise G, Lemmer J, Weng Y, Hubler M, Alexi-Meskishvili V,Bottcher W, Hetzer R, Berger F, Stiller B. Biomarker responses duringmid-term mechanical cardiac support in children. J Heart Lung Trans-plant 2008;27:150–157.

15. Price JF, Thomas AK, Grenier M, Eidem BW, O’Brian Smith E,Denfield SW, Towbin JA, Dreyer WJ. B-type natriuretic peptide pre-dicts adverse cardiovascular events in pediatric outpatients withchronic left ventricular systolic dysfunction. Circulation 2006;114:1063–1069.

16. Ratnasamy C, Kinnamon DD, Lipshultz SE, Rusconi P. Associationsbetween neurohormonal and inflammatory activation and heart failurein children. Am Heart J 2008;155:527–533.

7. Ohuchi H, Takasugi H, Ohashi H, Okada Y, Yamada O, Ono Y,Yagihara T, Echigo S. Stratification of pediatric heart failure on thebasis of neurohormonal and cardiac autonomic nervous activities inpatients with congenital heart disease. Circulation 2003;108:2368 –2376.

8. Milting H, Ellinghaus P, Seewald M, Cakar H, Bohms B, Kassner A,Korfer R, Klein M, Krahn T, Kruska L, El Banayosy A, Kramer F.Plasma biomarkers of myocardial fibrosis and remodeling in terminalheart failure patients supported by mechanical circulatory supportdevices. J Heart Lung Transplant 2008;27:589–596.

9. Sodian R, Loebe M, Schmitt C, Potapov EV, Siniawski H, Muller J,Hausmann H, Zurbruegg HR, Weng Y, Hetzer R. Decreased plasmaconcentration of brain natriuretic peptide as a potential indicator of
cardiac recovery in patients supported by mechanical circulatory assistsystems. J Am Coll Cardiol 2001;38:1942–1949.

t4pds

Relation of Obesity to Recurrence Rate andBurden of Atrial Fibrillation

Maya Guglin, MD*, Kuldeep Maradia, MD, Ren Chen, MD, MPH, and Anne B. Curtis, MD

Obesity is associated with new-onset atrial fibrillation (AF). However, the effect of obesityon AF recurrence or burden has not been studied. The aim of this study was to investigatethe relation between AF recurrence, AF burden, and body mass index (BMI). A limited-access data set from the Atrial Fibrillation Follow-Up Investigation of Rhythm Manage-ment (AFFIRM) trial provided by the National Heart, Lung, and Blood Institute was used.Statistical analysis was done with a generalized linear mixed model. In 2,518 patients whohad BMIs recorded, higher BMI was associated with a higher number of cardioversions(odds ratio [OR] 1.017, 95% confidence interval [CI] 1.005 to 1.029 for a BMI increase of1 kg/m2; OR 1.088, 95% CI 1.024 to 1.155 for a BMI increase of 5 kg/m2; OR 1.183, 95%CI 1.049 to 1.334 for a BMI increase of 10 kg/m2; p � 0.006 for each). Increased BMI wasalso associated with a higher likelihood of being in AF on follow-up (OR 1.020, 95% CI1.002 to 1.038 per 1 kg/m2 increased BMI, p � 0.0283; OR 1.104, 95% CI 1.011 to 1.205 per5 kg/m2 increased BMI, p � 0.0283; OR 1.218, 95% CI 1.021 to 1.452 per 10 kg/m2

increased BMI, p � 0.0283). In a multivariate analysis, left atrial size but not BMI was anindependent predictor of AF recurrence and AF burden. Because left atrial size wascorrelated with BMI, the effect of BMI on AF can be likely explained by greater left atrialsize in subjects with higher BMIs. In conclusion, obesity is associated with a higherincidence of recurrence of AF and greater AF burden. © 2011 Elsevier Inc. All rights
Obesity is a risk factor for the development of new-onset atrial fibrillation (AF). Multiple studies have doc-umented a strong and independent association betweenbody mass index (BMI) and the incidence of AF.1– 4 Inhe Framingham Heart Study, obese participants had a5% to 50% increased risk for incident AF compared toarticipants with normal BMI, independent of other car-iovascular risk factors.1 In a Danish study, overweightubjects were also at increased risk for incident AF.2 In

addition to increasing the susceptibility of developingAF, a recent longitudinal cohort study over 21 years5

suggested that obesity was an independent predictor ofprogression from paroxysmal to permanent AF. How-ever, the association between obesity and total AF burdenor recurrence rate has not been studied.

Methods

To evaluate the relation of obesity with recurrence of AFor burden of AF, we used a limited access data set from theAtrial Fibrillation Follow-Up Investigation of Rhythm Man-agement (AFFIRM) trial, provided by the National Heart,Lung, and Blood Institute (Bethesda, Maryland). Detailedselection criteria for the study population, their baselinecharacteristics, and randomization into rate-control versusrhythm-control arms was previously explained.6

Department of Medicine, University of South Florida, Tampa, Florida.Manuscript received August 17, 2010; revised manuscript received andaccepted October 7, 2010.

E-mail address: [email protected] (M. Guglin).

Our main independent variable was BMI. We used BMI(calculated as weight in kilograms divided by height inmeters squared) entered in the data set by the AFFIRMinvestigators as a surrogate measure of obesity. It was an-alyzed as a continuous and a categorical variable.

Two outcome measures were AF recurrence and AFburden. We used the number of cardioversions donethroughout the follow-up period (electrical as well as phar-macological) as a surrogate marker of AF recurrence, andthe number of follow-up visits when patients were in AF asa surrogate marker of AF burden. All the data were analyzedwith a generalized linear mixed model using SAS (SASInstitute Inc., Cary, North Carolina). A univariate analysiswas done first to identify variables linked to the recurrencerate and total burden of AF. The connection between BMIas our main variable of interest and the 2 outcomes was thenexamined in detail, for the whole AFFIRM population andfor the rate- and rhythm-control arms separately.

The following variables were also checked for associa-tion with AF recurrence rate or AF burden: age, use ofangiotensinogen-converting enzyme inhibitors, use of �blockers, systolic blood pressure, history of hypertension,history of coronary artery disease, history of coronary arterybypass surgery, history of congestive heart failure, historyof diabetes, history of cardiomyopathy, history of myocar-dial infarction, New York Heart Association class at base-line, the left ventricular ejection fraction, and left atrial size.Variables found to be significantly associated with the out-comes were then put in a multivariate model. Because leftventricular ejection fractions were missing in �50% of thecases, we calculated fractional shortening on the basis of left
ventricular systolic and diastolic dimensions (fractional
www.ajconline.org


t1i

ok

1a


shortening � [left ventricular diastolic dimension � leftventricular systolic dimension]/left ventricular diastolic di-mension) and used it for the final analysis. A p value of�0.05 was considered statistically significant.

Results

In the AFFIRM study, 4,060 patients were enrolled atbaseline. We excluded 1,542 patients who did not havebaseline BMI information. Of the remaining 2,518 patients,1,255 were assigned to the rate-control arm and 1,263 to therhythm-control arm; the mean BMIs were 29.0 and 28.8kg/m2, respectively. These 2,518 patients had 22,753 fol-low-up visits and a total of 1,094 cardioversions, eitherpharmacologic or electrical: 888 in the rhythm-control armand 206 in the rate-control arm.

In a univariate analysis, BMI, left atrial size, age, andhistory of hypertension were independently associated witha higher AF recurrence rate (Table 1).

In the study population as a whole (n � 2,518), higherBMI was associated with a greater number of cardiover-sions. The odds ratios (OR) of receiving cardioversion were1.017 (95% confidence interval [CI] 1.005 to 1.029, p �0.006) for a BMI increase of 1 kg/m2, 1.088 (95% CI 1.024o 1.155, p � 0.006) for a BMI increases of 5 kg/m2, and.183 (95% CI 1.049 to 1.334, p � 0.006) for a BMIncrease of 10 kg/m2.

In the rhythm-control arm (n � 1,263), in which ahigher rate of cardioversions was expected, the ORs forcardioversion were 1.015 (95% CI 1.003 to 1.028), 1.079(95% CI 1.013 to 1.148), and 1.164 (95% CI 1.027 to1.319) for BMI increases of 1, 5, and 10 kg/m2, respec-tively (p � 0.0178 for each). In the rate-control arm, inwhich there was a lower rate of cardioversions, the as-sociation between BMI and number of cardioversions

Table 1Variables linked to number of cardioversions and burden of atrial fibrillat

Variable Number of Cardioversions

OR 95% CI

BMI 1.016 1.004–1.029Age 0.99 0.981–0.999Hypertension* 1.236 1.064–1.50Left atrial size 1.345 1.158–1.561

* By history.

Table 2Effect of body mass index as a continuous variable on number ofcardioversions

BMI Increase(kg/m2)

Study Arm Cardioversions

OR 95% CI p Value

1 Rate control 1.023 0.99–1.058 0.1788Rhythm control 1.015 1.003–1.028 0.0178



was not significant (Table 2).

When patients were classified into underweight (BMI�18.5 kg/m2), normal weight (BMI 18.5 to 24.9 kg/m2),verweight (BMI 25 to 29.9 kg/m2), and obese (BMI �30g/m2), obese patients were more likely to undergo cardio-

version (OR 1.268, p � 0.0194; Table 3), with normal-weight patients used as a reference. We did not find asignificant association between obesity and the number ofcardioversions in the rate-control arm. However, in therhythm-control arm, the OR of requiring cardioversion inobese subjects was 1.291 (p � 0.0173), with normal weightused as a reference.

During each follow-up visit, the current rhythm wasrecorded as AF versus no AF (presumed sinus rhythm). Of22,374 follow-up visits, patients were found to be in AF oratrial flutter on 8,686 visits: 6,289 in the rate-control groupand 2,397 in the rhythm-control arm. Using a linear mixedmodel, the ORs of a patient being in AF or atrial flutter were1.020 (95% CI 1.002 to 1.038) per 1 kg/m2 BMI increase,.104 (95% CI 1.011 to 1.205) per 5 kg/m2 BMI increase,nd 1.218 (95% CI 1.021 to 1.452) per 10 kg/m2 BMI

increase (p � 0.0283 for each).In the rate-control arm, obese (BMI �30 kg/m2) patients

had an OR of 1.55 (p � 0.0484) of being in AF on afollow-up visit, when normal weight (BMI 18.5 to 24.9kg/m2) was used as a reference (Table 4). No significantassociation between BMI and the likelihood of being in AFwas found in the rhythm-control arm.

Of the possible confounding variables that could influ-ence AF recurrence rate and AF burden, a history of hyper-tension, left ventricular fractional shortening, and left atrialsize were found to be significantly associated with AFrecurrence rate (p � 0.025, p � 0.004, and p �0.001,respectively).

After adjusting for age, history of hypertension, frac-tional shortening, and left atrial size, the latter appeared tobe the only determinant of both outcomes in a multivariateanalysis. At the same time, BMI was correlated significantlywith left atrial size (Spearman’s correlation coefficient 0.22,p �0.0001).

Discussion

In this analysis of a limited-access data set from theAFFIRM trial, we have demonstrated for the first time thatobesity is associated with a higher recurrence rate andgreater burden of AF compared to nonobese patients. Be-cause in the rate-control arm, the strategy was not to restoresinus rhythm, patients spent more time in AF than patientsin the rhythm-control arm. The difference between AF bur-

univariate analysis

Number of Follow-Up Visits in AF

alue OR 95% CI p Value

086 1.01 0.99–1.02 0.4233244 1.02 0.82–1.26 0.857075 0.99 0.98–1.00 0.1403001 1.32 1.14–1.54 0.0003

ion in a

p V

0.00.00.00.0

den in obese versus nonobese patients was significant in the

a

T

)

R

R

581Arrhythmias and Conduction Disturbances/Obesity and Atrial Fibrillation

rate-control arm as well as in the whole data set, but not inthe rhythm-control arm. In contrast, more cardioversions,pharmacologic or electrical, were performed in the rhythm-control arm. More cardioversions in obese versus nonobesepatients were demonstrated in this arm and in the whole dataset, but not in the rate-control arm.

Obesity was first reported as an important, potentiallymodifiable risk factor for new-onset AF by the Framinghaminvestigators. A 4% increase in AF risk per 1 kg/m2 increasein BMI was observed, with adjusted hazard ratios for AFassociated with obesity of 1.52 (95% CI 1.09 to 2.13, p �0.02) and 1.46 (95% CI 1.03 to 2.07, p � 0.03) for men andwomen, respectively, compared to subjects with normalBMIs.1

Subsequently, it was shown that the association of obe-sity with sustained AF is stronger than for transitory orintermittent AF. On average, AF risk is 3% higher per unitincrease in BMI. The risk is higher by 7% per BMI unitincrease for sustained AF, by 4% for intermittent AF, andby 1% for transitory AF. The obesity-AF association ap-pears to be partially mediated by diabetes mellitus butminimally through other cardiovascular risk factors.3

In the longitudinal cohort study from Olmsted County,Minnesota, BMI independently predicted progression topermanent AF. Compared to normal BMI, obesity (BMI 30to 34.9 kg/m2) and severe obesity (BMI �35 kg/m2) were

Table 3Effect of body mass index as a categorical variable on the number of car

Study Arm BMI (kg/m2) Number of Cardio

otal �18.5 10 (6.2%18.5–24.9 231 (4.2%

25–29.9 401 (4.5%�30 452 (5.4%

Rate control �18.5 1 (2.1%18.5–24.9 39 (1.5%

25–29.9 87 (1.9%�30 79 (2.0%

Rhythm control �18.5 9 (7.8%18.5–24.9 192 (6.8%

25–29.9 314 (7.5%�30 373 (8.7%

Table 4Body mass index effect on atrial fibrillation burden

Study Arm BMI (kg/m2) Number of Visit

Total �18.5 30 (19.718.5–24.9 1,952 (36.3

25–29.9 3,451 (39.8�30 3,253 (39.8

ate control �18.5 12 (31.618.5–24.9 1,366 (52.5

25–29.9 2,594 (57.0�30 2,317 (58.9

hythm control �18.5 18 (15.818.5–24.9 586 (21.1

25–29.9 857 (20.8�30 936 (22.1

* p �0.05.

ssociated with increased risk for progression to permanent

AF. This relation was not weakened by left atrial volume,which was independent of and incremental to BMI for theprediction of progression to permanent AF.5 Similarly, inthe Swedish Primary Prevention Study, body surface area atage 20 years (calculated from recalled weight and measuredheight) was strongly related to subsequent AF (p �0.0001),as were midlife BMI and weight gain from age 20 years tomidlife (p �0.0001).7

A meta-analysis of 16 studies enrolling a total of 123,249subjects found that obese subjects have an associated 49%increased risk for developing AF compared to nonobesesubjects. In postoperative AF, however, BMI did not appearto play an important role8 and was even associated with alower incidence of AF.9

In a recently published study by Tedrow et al,4 it wasdemonstrated for the first time that the risk for incident AFis especially high in subjects who gained weight rapidly.Even more important, they proved that this risk decreasesafter normalization of BMI. Obesity therefore appears to bea reversible risk factor for AF.

The association between obesity, left atrial size, and AFis well established.10 Obesity is identified as the most im-portant determinant of left atrial enlargement. In our study,left atrial size was independently correlated with BMI.Therefore, obesity may increase the rate of new-onset AF,the recurrence rate, the transition from paroxysmal to per-

ons

s OR 95% CI p Value

1.518 0.68–3.391 0.3086Reference

1.056 0.863–1.291 0.59871.268 1.039–1.548 0.01941.511 0.138–16.513 0.7353

Reference1.164 0.733–1.849 0.51941.236 0.773–1.977 0.37551.176 0.533–2.594 0.6875

Reference1.107 0.892–1.372 0.35621.291 1.046–1.593 0.0173

OR 95% CI p Value

0.426 0.123–1.473 0.1776Reference

1.268 0.944–1.704 0.1151.333 0.989–1.797 0.05870.39 0.034–4.54 0.4521

Reference1.37 0.896–2.096 0.14611.55 1.003–2.395 0.0484*0.895 0.267–3.004 0.858

Reference1.001 0.705–1.421 0.99651.156 0.816–1.637 0.4155

dioversi

version

)))))))))))

s in AF

%)%)%)%)%)%)%)%)%)%)%)%)

manent AF, and total AF burden, not directly but through

so

tmcamA

atssvsaw

1

1

1


increased left atrial size. It is noteworthy that in the studyfrom the Framingham cohort establishing the link betweenobesity and new-onset AF, after adjustment for left atrialdiameter, BMI was no longer associated with AF risk.1 Theinvestigators concluded that effect of obesity was mediatedby left atrial dilatation.

In contrast, left atrial remodeling and enlargement is awell-known effect of AF itself, and obesity contributing toAF recurrence and burden and therefore promoting leftatrial dilatation is another possible course of the events.

Two other studies used the same data set to analyze effectsof obesity in AF, and both concluded that extra weight isassociated with lower cardiovascular mortality.11,12 Neithertudy, however, addressed the issues of recurrence and burdenf AF in obese versus nonobese patients.

Our findings are in concurrence with other studies inves-igating the connection between BMI and AF. Although theechanism of obesity-related increased risk for AF is un-

lear, a consistent pattern of increased AF incidence, prev-lence, recurrence, and overall burden suggests that lifestyleodifications directed toward a healthier weight may reduceF and all the risks and complications associated with it.We analyzed only the data available from the limited-

ccess data set of the AFFIRM trial and did not have accesso the complete study data. In addition, this was a retro-pective analysis of the main trial; therefore, the resultshould be interpreted with caution. The number of clinicisits at which patients appeared to be in AF was used as aurrogate of AF burden, and the number of cardioversionss a surrogate of AF recurrence rate. Data for left atrial sizeere missing in 24% of patients with known BMIs.

1. Wang TJ, Parise H, Levy D, D’Agostino RB Sr, Wolf PA, Vasan RS,
Benjamin EJ. Obesity and the risk of new-onset atrial fibrillation.JAMA 2004;292:2471–2477.
2. Frost L, Hune LJ, Vestergaard P. Overweight and obesity as riskfactors for atrial fibrillation or flutter: the Danish Diet, Cancer, andHealth Study. Am J Med 2005;118:489–495.

3. Dublin S, French B, Glazer NL, Wiggins KL, Lumley T, Psaty BM,Smith NL, Heckbert SR. Risk of new-onset atrial fibrillation in relationto body mass index. Arch Intern Med 2006;166:2322–2328.

4. Tedrow UB, Conen D, Ridker PM, Cook NR, Koplan BA, Manson JE,Buring JE, Albert CM. The long- and short-term impact of elevatedbody mass index on the risk of new atrial fibrillation: the WHS(Women’s Health Study). J Am Coll Cardiol 2010;55:2319–2327.

5. Tsang TS, Barnes ME, Miyasaka Y, Cha SS, Bailey KR, Verzosa GC,Seward JB, Gersh BJ. Obesity as a risk factor for the progression ofparoxysmal to permanent atrial fibrillation: a longitudinal cohort studyof 21 years. Eur Heart J 2008;29:2227–2233.

6. Wyse DG, Waldo AL, DiMarco JP, Domanski MJ, Rosenberg Y,Schron EB, Kellen JC, Greene HL, Mickel MC, Dalquist JE, CorleySD. A comparison of rate control and rhythm control in patients withatrial fibrillation. N Engl J Med 2002;347:1825–1833.

7. Rosengren A, Hauptman PJ, Lappas G, Olsson L, Wilhelmsen L,Swedberg K. Big men and atrial fibrillation: effects of body size andweight gain on risk of atrial fibrillation in men. Eur Heart J 2009;30:1113–1120.

8. Wanahita N, Messerli FH, Bangalore S, Gami AS, Somers VK, Stein-berg JS. Atrial fibrillation and obesity—results of a meta-analysis. AmHeart J 2008;155:310–315.

9. Banach M, Goch A, Misztal M, Rysz J, Jaszewski R, Goch JH.Predictors of paroxysmal atrial fibrillation in patients undergoing aor-tic valve replacement. J Thorac Cardiovasc Surg 2007;134:1569–1576.

0. Stritzke J, Markus MR, Duderstadt S, Lieb W, Luchner A, Doring A,Keil U, Hense HW, Schunkert H, Investigators MK. The aging processof the heart: obesity is the main risk factor for left atrial enlargementduring aging the MONICA/KORA (Monitoring of Trends and Deter-minations in Cardiovascular Disease/Cooperative Research in the Re-gion of Augsburg) study. J Am Coll Cardiol 2009;54:1982–1989.

1. Ardestani A, Hoffman HJ, Cooper HA. Obesity and outcomes amongpatients with established atrial fibrillation. Am J Cardiol 2010;106:369–373.

2. Badheka AO, Rathod A, Kizilbash MA, Garg N, Mohamad T, Afonso
L, Jacob S. Influence of obesity on outcomes in atrial fibrillation: yetanother obesity paradox. Am J Med 2010;123:646–651.

JEsM

T

The Editor’s Roundtable: Implantable Cardioverter-Defibrillatorsin Primary Prevention of Sudden Cardiac Death and

Disparity-Related Barriers to Implementation

Vincent E. Friedewald, MDa,*, Gregg C. Fonarow, MDb, Brian Olshansky, MDc,
Clyde W. Yancy, MDd, and William C. Roberts, MDe
t2dStivof

Acknowledgment

This CME activity is supported by an educational grantfrom Medtronic, Inc., Minneapolis, Minnesota.

Disclosures

Dr. Friedewald has received honoraria for speaking fromNovartis, East Hanover, New Jersey. Dr. Fonarow has re-ceived honoraria for speaking and consulting and researchgrants from Medtronic; and GlaxoSmithKline, ResearchTriangle Park, North Carolina. Dr. Olshansky has receivedhonoraria for speaking and consulting from Medtronic; andBoston Scientific Corporation, Natick, Massachusetts. Dr.Olshansky has received honoraria for consulting and is amember of the advisory board for Novartis. Dr. Yancy hasno relevant financial relationships to disclose. Dr. Robertshas received honoraria for speaking from Merck, White-house Station, New Jersey; AstraZeneca, Wilmington, Del-aware; and Novartis.

Objectives

Upon completion of the activity, the physician should beable to:

1. Diagnose patients with congestive heart failure whoare candidates for implantable cardioverter-defibrilla-tor (ICD) therapy.

2. Explain the risks and benefits of ICD therapy to pa-tients.

3. Decrease gender and ethnic disparities in treatmentwith ICD therapy.

aAssociate Editor, The American Journal of Cardiology, Clinical Pro-fessor, Department of Internal Medicine, The University of Texas MedicalSchool at Houston, Houston, Texas, and Research Professor, University ofNotre Dame, Notre Dame, Indiana; bThe Eliot Corday Professor in Car-diovascular Medicine and Science, UCLA Division of Cardiology, Uni-versity of California, Los Angeles, Los Angeles, California; cProfessor ofMedicine, University of Iowa College of Medicine, Iowa City, Iowa;dMedical Director, Baylor Heart and Vascular Institute of Baylor Univer-sity Medical Centerand Chief, Cardiothoracic Transplantation, Baylor Uni-versity Medical Center, Dallas, Texas; and eEditor-in-Chief, The Americanournal of Cardiology and Baylor University Medical Center Proceedings,xecutive Director, Baylor Heart and Vascular Institute of Baylor Univer-ity Medical Centerand Dean, A. Webb Roberts Center for Continuingedical Education of Baylor Health Care System, Dallas, Texas.

This discussion took place at Baylor University Medical Center, Dallas,exas, on October 9, 2008.

mE-mail address: [email protected] (V.E. Friedewald).
Am J Cardiol 2011;107:583–5900002-9149/11/$ – see front matter © 2011 Elsevier Inc. All rights reserved.doi:10.1016/j.amjcard.2010.10.003

Target Audience: This activity is designed for cardiol-ogists and all other health care specialists caring for patientswith acute and chronic coronary heart disease.

CME Credit: The A. Webb Roberts Center for Continu-ing Medical Education of Baylor Health Care System, Dal-las, Texas, designates this educational activity for a maxi-mum of 1 AMA PRA Category 1 Credit.™ Physiciansshould only claim credit commensurate with the extent oftheir participation in the activity.

The A. Webb Roberts Center for Continuing MedicalEducation of Baylor Health Care System, Dallas, Texas, isaccredited by the Accreditation Council for ContinuingMedical Education to provide continuing medical educationfor physicians.

CME Provider Privacy Policy and Contact Informa-tion: The A. Webb Roberts Center for Continuing MedicalEducation of Baylor Health Care System (214-820-2317)observes the privacy and confidentiality of CME informa-tion and the personal information of CME participants.Third parties receive only aggregated data about CME ac-tivities that are relevant to their interests and/or the activitiesthey support.

CME Instructions: After reading this article, go onlineat www.AJConline.org to register, complete a post-test witha minimum score of 80%, complete an evaluation, and printa certificate.

Combination of Media: Print and InternetComputer Requirements: Windows 2000, Pentium 3

or greater, 512 ram, 80 gigabytes storageEstimated Time to Complete: 1 hourRelease Date: February 2010Termination Date: February 2011

Introduction

The ICD was first placed into human subjects in 1980 byMirowski, after several years of nonhuman animal test-ing.1,2 In the 30 years since the introduction of ICDs, ICDherapy in the United States has become commonplace, with

broad categories of use for preventing sudden cardiaceath (SCD): primary prevention involves the prevention ofCD in patients without histories of cardiac arrest or sus-

ained ventricular tachycardia, and secondary preventionnvolves the prevention of SCD in patients who have sur-ived prior cardiac arrest, sustained ventricular tachycardia,r other major cardiac events. This Editor’s Roundtableocuses on ICD therapy for primary prevention, which
ainly involves patients with ischemic and nonischemic
www.ajconline.org

http://www.AJConline.org


wt1asodbfiabtmbmcui

tp1lmrrtf

w

d

p

troiaAIa

diafos

p


heart failure “who are receiving optimal medical therapyand have a reasonable expectation of survival with goodfunctional status for �1 year.”3

Discussion

Dr. Friedewald: When did ICD therapy appear?Dr. Olshansky: ICD therapy was developed and first

tested in nonhuman animal models by Michel Mirowski inthe 1970s.2 Although the concept of an ICD was not initially

ell received, Mirowski paved the way for clinical accep-ance when he performed the first human implantation in980.1 The first ICD was large, weighing about 9 oz, withlarge battery and a capacitor to shock the heart. It had few

ettings, its sensing ability was rudimentary, and it lastednly about 18 months. It required major surgery because theevice was implanted in the abdomen and the chest had toe opened, requiring at least 1 week of hospitalization. Therst ICD devices were handmade, so the supply was limited,nd patients sometimes had to wait for weeks before 1 coulde obtained. The technology progressed dramatically overhe next 30 years as the devices became multiprogram-able, used smaller batteries with longer battery life, had

etter capabilities to defibrillate with biphasic shocks, wereade programmable for cardiac pacing, had better leads that

ould be placed intravenously, and were implanted into thepper chest, an easy procedure to perform. Today, ICDmplantation is a low-risk procedure carried out worldwide.

Dr. Yancy: What are the current risks of ICD use?Dr. Olshansky: There are several. Although the focus in

he lay press has been placed on device recalls and leadroblems, improper working devices are rare, maybe 1 in0,000 implants. Other complications including myocardialead perforation, infection, pneumothorax, lead dislodge-ent, and inappropriate shocks (a shock delivered for a

eason other than a life-threatening ventricular tachyar-hythmia). The ICD is designed, however, to protect life athe expense of an occasional inappropriate shock. The riskor inappropriate shock is about 25%.

Dr. Yancy: What is the frequency of serious problemsith ICDs?Dr. Olshansky: Serious problems such as infection and

evice failure occur in about 1% of ICD implants.Dr. Yancy: What is the current role of the ICD in

reventing SCD?Dr. Olshansky: There has been a significant movement

o ICD use for primary prevention of SCD. At 1 time, beforeeceiving an ICD, patients had to experience 2 separateut-of-hospital cardiac arrests, so few patients used to qual-fy for an ICD. We no longer require a prior event, only that

patient is “likely to have cardiac arrest” (Appendix3).lthough there is some controversy about the criteria, most

CD implantations are for primary prevention, not second-ry prevention.

Dr. Fonarow: Much of the growth of ICD utilization isue to the recognition that most antiarrhythmic drugs areneffective for both primary and secondary prevention andre sometimes proarrhythmic, thereby increasing the riskor SCD. Two decades ago, flecainide and encainide were 2f the top 10 cardiac medications prescribed, but they sub-
equently were found to increase all-cause mortality and are
roarrhythmic.4 Other antiarrhythmic drugs, such as amio-darone, also fail to protect against SCD.

Dr. Friedewald: What is the relation between left ven-tricular (LV) dysfunction and SCD?

Dr. Fonarow: Patients with significant LV dysfunc-tion—even in the absence of a prior cardiac event, ventric-ular ectopic beats on ambulatory monitoring, or induciblearrhythmia on electrophysiologic study—are at increasedrisk for SCD. Because up to 1/2 of deaths in patients withLV dysfunction are sudden, prophylactic ICD placement inthis patient population is often indicated. Prospective ran-domized clinical trials in patients receiving optimal heartfailure (HF) treatment with subsequent placement of theICD demonstrated that they aborted SCD when compared todrug treatment alone, in patients with both ischemic andnonischemic forms of cardiomyopathy.

Dr. Friedewald: Do drugs that are not directly antiar-rhythmic, but proven beneficial in treating patients with HF(i.e., � blockers and renin-angiotensin aldosterone inhibi-tors) reduce the risk of SCD in patients with HF?

Dr. Fonarow: Beta blockers reduce death from progres-sive HF as well as SCD in patients with LV dysfunction.Patients on � blockers, however, have a greater relativereduction in death from progressive HF, resulting in in-creased incidence of SCD in this population. The predom-inant effect of angiotensin-converting enzyme inhibitors ison death from progressive HF with possibly a slight reduc-tion in the frequency of SCD. Aldosterone antagonists alsodecrease the risk of death from progressive HF, with pos-sibly a slight reduction in the frequency of SCD. Thus,patients on optimal medical therapy for LV dysfunction andHF have enough residual risk for SCD that usually justifiesprimary ICD placement.

Dr. Yancy: Is there a role for antiarrhythmic drugs inpatients with an ICD?

Dr. Olshansky: There may be a role for the use ofantiarrhythmic drugs in addition to ICD therapy in patientswho receive multiple ICD shocks for ventricular and atrialtachyarrhythmias. As primary therapy to reduce total mor-tality or arrhythmic death, however, antiarrhythmic drugshave no role. The important point is that ICD therapyreduces the incidence of both SCD and overall mortality.

Dr. Fonarow: It is important to separate absolute riskand proportional risk. The proportion of SCD relative todeath from progressive left ventricular dysfunction is higherin patients with less severe HF symptoms—New YorkHeart Association class I or II—compared to patients inclass III or IV HF, in which a greater proportion of deathsare due to progressive HF. Thus, although SCD occurs inpatients in class III and IV HF, the absolute risk for deathsboth from progressive HF death as well as SCD rises withincreasing severity of HF. There is little benefit in prevent-ing SCD in a patient who shortly thereafter dies fromprogressive HF. Thus, identifying patients who derive thegreatest absolute benefit from the therapy and in whom thebenefit outweighs the potential ICD risks is essential. Inpatients with class I to III HF treated with an ICD, thebenefit outweighs the risk and prolongs survival. In class IVpatients, however, because of the ICD impairment on qual-ity of life and functional status, HF not amenable to optimal
medical therapy precludes ICD therapy. Class II and III HF

h

tfctmudwwt

p

atgpafpt

dbt

cLctn

L

Lb

iah

aptpiaainpo8fm

Itmdsvwmotuett

dgh

wcfacc

aguamtdWpeonmad

wlt

585Roundtable Discussion/ICDs in Primary Prevention of SCD

patients frequently die suddenly. Many of them also diefrom progressive HF, so they need protection from bothmajor modes of death.

Dr. Yancy: Thus, patients with class IV HF may not beelped by ICD implantment.

Dr. Fonarow: That is true, except in class IV HF pa-ients, who improve to a better functional class with otherorms of treatment. Thus, class IV patients who are notandidates for cardiac resynchronization therapy, cardiacransplantation, or ventricular assist devices, and who re-ain persistently in class IV are not candidates for ICD

nder current guidelines because there is insufficient evi-ence that they benefit. Ambulatory class IV HF patientsho receive cardiac resynchronization therapy combinedith ICD, however, may have improved survival compared

o patients on optimal medical treatment alone.Dr. Yancy: Is SCD risk related to functional capacity in

atients with HF?Dr. Olshansky: Patients in functional class I and II HF

re less likely to die, and when they die it is more often dueo an arrhythmia. They are the patients that derive thereatest potential benefit from an ICD. Functional class IIIatients also have a high risk of arrhythmic death, but theylso have greater risk of mortality from progressive HF thanunctional class II patients. Thus, the functional class IIIatient can benefit from an ICD. Functional class IV pa-ients may not benefit from ICD therapy.

Dr. Roberts: Are you saying that the greater the LVilatation, the greater the chance that death will be causedy pump failure, and the less the LV dilatation, the greaterhe chance that death will be due to an arrhythmia?

Dr. Olshansky: It is probably true that the larger the LVavity, the greater chance of death from LV failure. ManyV parameters have been studied, however, and only 2 arelinically useful: (1) New York Heart Association func-ional class and (2) LV ejection fraction. Although they areot perfect, they are the best we have now.

Dr. Roberts: The LV ejection fraction is proportional toV cavity size?

Dr. Olshansky: The possibility that LV end-diastolic orV end-systolic volume is an independent predictor has noteen studied. Thus, we use LV ejection fraction.

Dr. Roberts: The same is true for angina pectoris. Withncreasing LV cavity dilatation, there is less frequency ofngina, and patients with grade 4 angina pectoris generallyave normal LV cavity size.

Dr. Olshansky: One of the most challenging issues is riskssessment. Although our discussion is focused on primaryrevention in patients with HF, there are many different patientypes at increased risk for SCD whose LV function is wellreserved. Patients with LV diastolic dysfunction alone, whichs difficult to define, also are at increased risk of SCD. Therelso are patients at increased risk of SCD with certain geneticbnormalities, such as the Brugada syndrome, the long–QTnterval syndrome, and hypertrophic cardiomyopathy, who doot have HF and have normal LV ejection fraction. For exam-le, consider a 17-year-old girl with syncope and a QT intervalf 560 ms on her electrocardiogram. Compare that patient to an5-year-old man with an LV ejection fraction of 10% andunctional class II HF on optimal medical therapy, who has a
uch greater risk of SCD, and a much greater chance of an
CD stimulus than the 17-year-old girl. Thus, clinical charac-eristics other than LV ejection fraction and functional classust be considered. The younger individual might use the

efibrillator less, but her dying from cardiac disease almosturely would be from a cardiac arrhythmia. In the older indi-idual, however, progressive HF and co-morbidities associatedith age (renal dysfunction, atrial arrhythmias, and diabetesellitus) are additional independent risks for mortality. An-

ther group to consider for ICD therapy is comprised of pa-ients with ischemic cardiomyopathy and nonsustained ventric-lar tachycardia (VT) who have relatively well-preserved LVjection fraction and are in New York Heart Association func-ional class I. They also have sufficient risk of VT and ven-ricular fibrillation that may warrant ICD placement.

Dr. Roberts: Let us assume that a 60-year-old manevelops substernal chest pain, goes to the hospital emer-ency department, and 3 hours after the onset of chest painas cardiac arrest. Is that a form of SCD?

Dr. Olshansky: By some definitions, yes.Dr. Roberts: By your definition?Dr. Olshansky: Probably not.Dr. Roberts: The World Health Organization uses death

ithin 24 hours after a change in health status in theirriteria for SCD, which is ridiculous! If a 24-hour definitionrom sudden change of previous health to an event is used,

lot of patients with acute myocardial infarction are in-luded, and I believe they should not be included in theategory of SCD.

Dr. Fonarow: The definition of SCD is important. Therere, however, many definitions of SCD, including a strin-ent 15- or 30-minute definition or when it is completelynexpected and not explained by something else such ascute myocardial infarction. At one time, it was thought thatany episodes of SCD in HF patients were not caused by

achyarrhythmias, rather by other mechanisms such as bra-yarrhythmias, which would not be prevented by an ICD.e now know that a substantial proportion of out-of-hos-

ital SCD events are caused by tachyarrhythmias and can beffectively detected and aborted with the ICD. The numberf deaths that are bradyarrhythmic or due to other mecha-isms such as myocardial infarction or are hyperkalemiaediated and cannot be aborted by the ICD constitute onlysmall proportion of SCD in patients with significant LV

ysfunction after myocardial infarction or ambulatory HF.Dr. Roberts: I use a 6-hour definition for SCD because

ithin 6 hours after onset of chest pain, there is no histo-ogical evidence of acute myocardial infarction. Thus, pa-ients with occluding thrombus are not included.

Dr. Olshansky: That is reasonable, but there is a lot we donot understand about SCD. Many cases of SCD, for example,are due to asystole rather than a ventricular arrhythmia.

Dr. Friedewald: According to current guidelines, whoshould receive an ICD?

Dr. Fonarow: The newest guidelines from the AmericanCollege of Cardiology, American Heart Association, andHeart Rhythm Society are based on randomized clinicaltrials That establish ICD efficacy.3 For primary prevention,placement of an ICD is indicated for (1) patients with LVejection fraction �35% who are at least 40 days after onsetof the infarction with New York Heart Association class II
or III HF; (2) patients with nonischemic dilated cardiomy-

bab

mebcm


opathy and LV ejection fraction �35% who are functionalclass II or III, do not have contraindications for an ICD, andhave a reasonable expectation for good functional status forat least 1 year; and (3) patients at least 40 days post myo-cardial infarction who do not have class II or III HF buthave class I failure with LV ejection fraction �30%. Thesecategories comprise a large number of eligible patients whodo not have any prior arrhythmias but are candidates forICD by virtue of their clinical state, symptom class, anddegree of LV dysfunction determined by LV ejection fraction.

Dr. Friedewald: What is the basis for waiting 40 daysafter an acute myocardial infarction?

Dr. Fonarow: Placement of an ICD soon after acutemyocardial infarction has not shown a net benefit becausesome post–myocardial infarction patients have substantialimprovement in LV function with reperfusion therapy andoptimal medical treatment.4

Dr. Yancy: The guidelines strongly advise that patientse on “reasonable” medical therapy. What is “reasonable,”nd for how long should medical treatment be employedefore ICD placement is considered?

Dr. Fonarow: Many patients with nonischemic cardio-yopathy have spontaneous improvement in LV function,

ven when untreated. Thus, they need a chance for recoveryefore receiving an ICD. Prior guidelines, which have beenontroversial, left this time frame up to the clinician. Inost patients we can start � blockers and angiotensin-

converting enzyme inhibitors at the same time in patientswith HF and up-titrate them within 2 to 3 months to estab-lish optimal medical therapy and then reassess LV function.

Dr. Yancy: It would seem difficult to withhold an ICDfrom a patient with significant LV dysfunction at risk forSCD for such a relatively long period of time.

Dr. Olshansky: Yes, especially in the common scenarioof patients who are not receiving optimal medical therapyfor nonischemic and ischemic cardiomyopathy, as they areat risk for SCD. Other patients of special concern are thosewith a recent myocardial infarction who have poor LVfunction, are being discharged from the hospital, and face aperiod of time before they are seen again. That time periodencompasses an increased risk of SCD. Because patientswith poor LV ejection fractions or impaired LV functionhave a high risk of nonarrhythmic death early, it is not clearthat the ICD offsets that risk. The concern whether we aredoing enough for those patients, however, remains. Someclinicians use “bridging” approaches, such as life vests andthe automatic external defibrillator, but the efficacy data forsuch devices are poor.

Dr. Roberts: What percent of patients with HF die in thehospital?

Dr. Yancy: With acute cardiac decompensation, the av-erage mortality is about 4%, but the range is 2% to 20%.

Dr. Roberts: When a patient with severe HF dies sud-denly, do you attribute the death to sudden ventricularfibrillation or asystole, rather than to pump failure?

Dr. Yancy: This is a dilemma of clinical trials. When wetry to adjudicate such deaths, we use arbitrary definitions.How do you set the time clock, and what does that mean?SCD is an unexpected or unanticipated event at some pointafter a patient is deemed “stable and comfortable.” The
definition of “sudden” is a problem for many clinical trials.
Dr. Roberts: In-hospital death should be separated fromout-of-hospital death.

Dr. Olshansky: I agree. There are some in-hospitalsudden deaths that occur while patients are on the ICD, andthese are reported as SCD. The mechanism can be either aventricular arrhythmia or pulseless electrical activity.

Dr. Roberts: But patients receiving ICD therapy alsodie of HF.

Dr. Friedewald: Is patient proximity to medical careconsidered among the indications for ICD implantation?

Dr. Fonarow: For individuals having out-of-hospitalSCD, even those who next door to the hospital, the chanceof survival with intact neurological function is incrediblylow, about 7%. Thus, geography is not a factor for ICDplacement. Every second without adequate defibrillationmakes survival after cardiac arrest less likely.

Dr. Friedewald: Does a low likelihood of patient adher-ence to cardiac medications affect ICD usage?

Dr. Fonarow: Patient adherence to HF medications isnot a factor in the decision whether to implant an ICD.Patients with ICDs need to take medications. The device isnot a replacement for medications. Patients with ICDs alsorequire optimal medical treatment. The device does nothingto prevent progression of underlying LV failure. It is criti-cally important to continue optimal medical treatment alongwith the ICD.

Dr. Yancy: What is the experience with the ICD inclinical settings in the context of guideline indications?

Dr. Fonarow: Among all patients hospitalized for HF inthe American Heart Association Get With the Guidelines–Heart Failure program, only about 25% of eligible patientsreceive an ICD. Some eligible patients, however, refuse theICD, and some do not meet additional criteria such as HFtime of onset and the presence of co-morbid conditions.When these patients are excluded, about 33% of eligiblepatients are treated with an ICD. In the outpatient settingutilizing a registry like Improve the Use of Evidence-BasedHeart Failure Therapies in the Outpatient Setting5 (IM-PROVE-HF), the utilization rate is higher. About 50% ofHF patients who are candidates under the guidelines receiveICDs. This means, however, that 50% of HF eligible out-patients in cardiology practices do not receive ICDs. Inprimary care settings the rate of ICD placement is evenlower. Thus, despite excellent clinical trial evidence of theirefficacy, a substantial proportion of patients who are eligiblefor ICD treatment do not receive it.

Dr Friedewald: Why is ICD therapy so underutilized?Dr. Fonarow: It is very difficult to understand why

evidence-based therapies, using either medications or de-vices, are not better adhered to, especially in view of all theinformation disseminated about guidelines. One factor withmany other analogous therapies is a substantial lag from thetime the therapy becomes evidence-based and recom-mended in the guidelines to when it is widely adopted. ICDtreatment is relatively new, so the customary time lag is afactor in its underutilization. Another factor is that thistreatment requires much more than a simple written pre-scription. ICD placement involves referral to an electro-physiologist, an invasive procedure, and close follow-up.Thus, some patients, even when fully informed about the
benefits of ICD, decline implantation.

oa

i

lTdor

yfw

fwe

pprTocbit

u

ddpt

m

att


Dr. Olshansky: I am surprised that the ICD utilization isas high as 50% in outpatients. It has been my experience—not supported by any published data—that many physicianseither do not know the data supporting ICD use, do notbelieve the data, or do not want to believe the data, so theysimply do not even think about ICD therapy. There alsohave been negative stories published in the public media.Here is a quote from a major newspaper, which is repre-sentative of some of the public’s perception: “The numberof patients receiving ICDs has actually declined as moredoctors and patients decide the risks and uncertainties thedevice has posed may outweigh the potential benefits. In-dustry estimates and medical studies show that ICDs havesaved the lives of only 10% of the 600,000 people who willreceive them at most. Nine of 10 people who receive ICDsreceive no medical benefit.” This is a misunderstanding ofthe science. When the number of persons needed to treat toget a benefit of saving 1 life is only 10 people, that is atremendously effective therapy. We do not expect that everypatient who receives an ICD will have his or her life savedby the ICD. We only expect risk reduction.

Dr. Yancy: We have excellent data demonstrating utilityof the ICD in the appropriate patient with reduction in totalmortality and sudden death. Some patients who receive anICD never have a firing and when they do, it is an inappro-priate shock. Others are concerned about costs and relatedrisks. What do you think is required: more data? More time?Different perspectives? How do we optimize use of the ICD?

Dr. Olshansky: The issue comes down to the risk/ben-efit ratio and what is expected in The extension of reason-ably functional life, which is difficult to define. If the life ofan 85-year-old person is extended by a few months, is that ofvalue to society or only to the individual? Is it worth the cost?

Dr. Yancy: Is this an ethical argument or a scientificargument?

Dr. Roberts: It is both. Are we going to have rationingf medical care? We are talking about expensive devices incountry that is broke.Dr. Friedewald: What is the average increase in longev-

ty in the group with the ICD?Dr. Olshansky: An exact number has not been calcu-

ated, at least in the Sudden Cardiac Death in Heart Failurerial (SCD-HeFT).6 In that trial, which is probably our bestatabase of patients who have functional class II and III HFn the best medical therapy, there is a 23% relative riskeduction with a mean follow-up of about 45 months.

Dr. Roberts: What is the absolute risk reduction?Dr. Fonarow: The absolute risk reduction over the 4

ears of the study is 7%, so 70 lives would have been lostor each 1,000 patients not receiving an ICD. The mean ageas 65 years, and they were primarily functional class II to III.Dr. Roberts: The study patients had an LV ejection

raction of about 35%. What do we know about patientsho are 40 days post myocardial infarction with greater LV

jection fractions?Dr. Fonarow: An interesting paradox is that patients

with better-preserved LV ejection fractions derive less ben-efit. It is the group of patients with LV ejection fraction�35% and especially �30% that derive much greater ben-efit. Thus, a cutoff of the LV ejection fraction of about 35%
is reasonable, because among patients with an LV ejection s
fraction �35%, the number needed to treat to show a benefitis substantially higher.

Dr. Roberts: What is the range of error for the measure-ment of LV ejection fraction?

Dr. Fonarow: For most studies, the range is �5%.Dr. Yancy: Do you believe there is a greater physician

bias against ICD placement than patient bias?Dr. Olshansky: There is a significant physician bias, but

I cannot quantitate it. Maybe we overemphasize the poten-tial benefits because some trials do not show as muchbenefit as others. There are other post hoc analytic dataabout specific patient populations suggesting that certaintypes of patients do not achieve the same benefit as others.Patients with atrial fibrillation, patients with functional classIII HF, and patients with renal dysfunction may derive lessbenefit. Older patients also may not have as much improve-ment of functional quality of life with the ICD as youngerpersons.

Dr. Fonarow: Such data can be misleading becausetrials are powered to look at the overall population. We canbe misguided when we retrospectively study underpoweredsubgroups that appear to have less benefit with the ICD.

Dr. Yancy: There are data showing no statistical differ-ence in outcomes with ICD therapy as a function of genderor race. There are, however, gender and racial differencesamong patients who receive ICD therapy, given the sameindications. How does this apply to the American HeartAssociation’s Get With the Guidelines–Heart Failure?

Dr. Fonarow: Get With the Guidelines–Heart Failure isa hospital-based registry and performance improvementprogram from the American Heart Association. There arecurrently 424 US hospitals from all regions of the USAparticipating. The registry is comprised of �200,000 pa-tients and includes data on demographics, treatment, LVfunction, previous device placement, hospitalization, andprimary discharge diagnosis of HF or HF as the predomi-nant reason for hospitalization. A study by Hernandez andcolleagues7 looked at the use of the ICD in appropriateatients to see if there were disparities on the basis ofatient age, gender, or race. One third of eligible patientseceived an ICD or planned ICD placement post discharge.here were substantial disparities by race and gender: 44%f eligible white men and 33% of eligible black men re-eived an ICD; 30% of white women and 28% of eligiblelack women received an ICD. Adjustment for co-morbid-ties, insurance coverage, and other factors did not changehese findings.

Dr. Yancy: The same disparities have been found in these of cardiac resynchronization therapy.

Dr. Fonarow: Many published reports have focused onisparities of other cardiovascular procedures, surgery, andevice implants. Among newer or more expensive thera-ies, a certain proportion of patients are less likely to bereated, and race and gender seem to be significant influences.

Dr. Olshansky: Do such disparities also occur withedical treatment?Dr. Yancy: There is variance in the use of � blockers

fter acute myocardial infarction. For other evidence-basedreatments, there is no demonstrable evidence of disparity inhe medical care of inpatients. For activities such as coun-
eling for smoking cessation or appropriate discharge or-

v

fp

bIi


ders, disparities emerge. In mammography for breast cancerscreening, revascularization compared to amputation forperipheral vascular disease, hemodialysis compared to renaltransplantation, and thrombolytic therapy compared to acutepercutaneous coronary intervention, there is clear evidenceof disparate health care.

Dr. Olshansky: Some disparities are patient related andothers are due to physician ignorance or physician bias. Iwonder if they depend on the physician-patient relationship.There are differences, for example, in the care of whitewomen who see white female physicians, white male phy-sicians, African American male physicians, and AfricanAmerican female physicians.

Dr. Yancy: Physicians who are culturally sensitive pro-ide care in a more equitable manner.

Dr. Roberts: Is cigarette smoking a factor in ICD use?Dr. Fonarow: Cigarette smoking is not a relative or

absolute contraindication for ICD therapy.Dr. Roberts: How do you advise patients who dec-

line an ICD?Dr. Fonarow: All of the other appropriate aspects of HF

therapy are important, including medications, lifestyle mod-ifications, and appropriate exercise in a cardiac rehabilita-tion program.

Dr. Olshansky: Some health care providers have astrong bias against persons who smoke cigarettes. Such anattitude toward cigarette smokers is a different type of biasfrom that of gender and race.

Dr. Yancy: Bias occurs in many different modalities andrepresentations, which is why I believe we should consideran “across the board” performance improvement strategyfor ICD therapy.

Dr. Olshansky: I am not an advocate of electronic med-ical records, but perhaps electronic medical records can beused to enforce proper medical management. For example,these guidelines could be built into electronic medicalrecords in a way that a red flag would appear when a patientsatisfied the indications for an ICD. If the physician did notprescribe it, or the patient refused it, the reasons for notusing it would be recorded. This approach also might en-courage device implantation in a manner that is blind to raceand gender.

Dr. Fonarow: Whether systems such as electronic med-ical records can be successful in improving treatment withmodalities as complex as placement of ICDs and cardiacresynchronization devices requires further study. With largegaps in treatment and the disparities that have been identi-fied, we need to do something different from the presentconventional “hit or miss” treatment approach. A morereliable, safe, effective, and unbiased delivery system con-sistently delivering appropriate therapies in all settingsshould be the goal.

Dr. Friedewald: Are there signs of disparity reductionsor ICDs in the Get With the Guidelines–Heart Failurerogram?

Dr. Fonarow: Preliminary data suggest that there haseen some improvement, but it is too early to be certain.n other areas, such as use of renin-angiotensin systemnhibitors and � blockers, there is profound evidence that

these types of programs improve the quality of care and
treatment rates.
Dr. Friedewald: Perhaps guidelines themselves shouldplace greater emphasis on race and gender issues.

Dr. Fonarow: I agree. The 2005 HF guidelines on indi-cations for ICD implantation included a section on specialpopulations indicating that the recommendations were ap-plicable to both genders, all races and ethnicities, and toolder patients. There is also tremendous variation in ICDuse in the outpatient setting compared to the hospital setting.There are some hospitals that have very high rates of qual-ified patients being treated appropriately with an ICD, andothers where the use is nearly zero. Sharing best practiceinformation can decrease these variations, indicating that itis possible to achieve high treatment rates of appropriateguideline-recommended therapies.

Dr. Roberts: It seems to be much easier to set upstandards and diagnostic and therapeutic criteria in institu-tions compared to the private practice setting.

Dr. Fonarow: I agree. In many hospital settings, wherethere are standardized practice protocols and multidisci-plinary teams, quality improvement initiatives are morelikely to succeed. This is true regardless of the hospital type:teaching or nonteaching, rural or urban, large or small. Theoutpatient setting, however, is a far greater challenge, butthere have been successes in this area as well.

Dr. Friedewald: What is the role of electrophysiologictesting in assessing people for possible ICD implantation?

Dr. Olshansky: The electrophysiology test has a poten-tial role in patients who have nonsustained VT, are not postmyocardial infarction, have had no recent acute interven-tion, are clinically relatively stable, and have a LV ejectionfraction �40% with New York Heart Association func-tional class I HF. This subset may benefit from electrophys-iology testing to assess the risk for primary prevention withthe ICD. T-wave alternans has been mentioned as a possiblenoninvasive predictor, but it will never be helpful in deter-mining who should receive an ICD. T-wave alternans willultimately fall the way of other noninvasive tests. We areleft with only a hope for a good noninvasive predictor, sowe must rely mainly on New York Heart Association func-tional class and LV ejection fraction.

Dr. Friedewald: Does persistent atrial fibrillation affectthe indications for ICD placement?

Dr. Yancy: The patient with atrial fibrillation is atgreater risk for HF and SCD.

Dr. Olshansky: For ICD indications, atrial fibrillation isa “2-edged sword.” A post hoc analysis of atrial fibrillationpatients or syncope patients in the SCD-HeFT trial foundthat those patients do not benefit from an ICD but that theyare at higher risk of both total mortality and fatal arrhyth-mia.6 Because those findings are post hoc analyses, it is hardto use those predictors to determine who would benefit themost from ICD placement. We do not have good prospec-tive data to determine the highest risk patients. Some of thehighest risk patients are also the oldest patients and the mostlikely to have other co-morbidities so that they are going todie anyway and therefore may not benefit much by receiv-ing an ICD. Our challenge is to find a middle ground. Thepatient with an LV ejection fraction that is not too abnormalis not going to benefit much from an ICD, but when the LVejection fraction is too low the patient also is not going to
benefit from an ICD. Thus, there is an “in-between” popu-

o

wh

od

b

eagTm

ppteeawa

nlpwpmmwnrabb


lation—with a modestly poor LV ejection fraction and mod-estly reduced functional class—who seem to derive themost benefit from an ICD.

Dr. Roberts: How much does the ICD cost?Dr. Yancy: The usual metric for HF is $50,000 per

quality-adjusted life-year. For an ICD, the cost is about$37,500. Thus, by accepted definitions, it seems reasonablycost effective.

Dr. Olshansky: It is at least as cost effective as many ofur other standard medical therapies.

Dr. Fonarow: I agree.Dr. Roberts: The most common mode of death in patients

ith angina pectoris is cardiac arrest. Would you personallyave an ICD inserted if you had chronic stable angina?

Dr. Yancy: No, I would want surgical revascularization.Dr. Fonarow: There is no evidence that ICD placement

ffers substantial value in chronic angina pectoris, but weo have great evidence in favor of antiplatelet drugs, �

blockers, renin-angiotensin system inhibitors, and aggres-sive statin therapy.

Dr. Roberts: If you had an acute myocardial infarction 50days ago, an ejection fraction of 40%, and class I HF, wouldyou like to receive an ICD if money were not a factor?

Dr. Fonarow: No, because I would believe that withappropriate medical therapy, my risk would be low enough.

Dr. Yancy: What if your LV ejection fraction was 35%?Dr. Fonarow: Then sign me up for an ICD!Dr. Roberts: But the ejection fraction margin of error, at

est, is �5%.Dr. Fonarow: Yes, so there is “wiggle room” for LV

jection fraction criteria. If I am measured at 35%, there ispossibility that I will be at 30%, and that places me in theroup that derives greatest benefit from ICD placement.here is a little built-in error ratio around guideline recom-endations.Dr. Friedewald: What is the future of ICD therapy?Dr. Olshansky: The future of ICD therapy depends in

art on what the health care system can afford. While therice of defibrillators might be driven down by improvedechnology, the ICD will remain expensive. I assume, how-ver, that technology will reduce device size and risk. Forxample, leadless defibrillators are being developed. Theyre big, but it is possible for these devices to be put inithout leads, so there would be no leads to break. There

lso is progress in extending battery life.Dr. Roberts: What is the current average battery life?Dr. Olshansky: About 5 to 6 years, but if a patient only

eeds 1 device and if the device could be put in without aead, the implantation procedure would be much easier anderhaps less expensive. Another issue is to better determinehich patient populations to target, because we want to treatatients who are at the highest risk and are going to get theost benefit from the device. We have to define what iseant by absolute risk that is required for the ICD andhich patients truly benefit from implantation. We alsoeed better markers. While there should not be any bias withegard to who gets an ICD, there may be genetic differencesmong populations who benefit. This area needs researchecause genetic testing may help identify patients who will
enefit the most.
Dr. Friedewald: What more should we do about ethnicdisparities in treatment?

Dr. Yancy: I favor efforts that improve across-the-boardadherence to evidence-based medical and device therapies.My observations over the last decade has demonstrated thatthere are few substantive differences between groups ofindividuals, especially on the basis of race, and that most ofthe differences in outcomes are due to the extent to whichpeople receive evidence-based medical or device therapies.Performance improvement strategies that are deployed in arace-blind, gender-blind, age-blind manner are best. Onemistake we made several years ago when we launchedevidence-based therapy on the basis of race was that weforced the medical community to emphasize race, whichwas polarizing. The uptake in therapy that should haveoccurred did not occur, and many people suffered unneces-sarily. In an era in which we have limited resources the bestway to utilize them is to make certain that every patientaffected with disease has the best possible chance for thebest possible outcome, the lowest utilization of hospitalservices, and the greatest improvement in survival. Theright approach is not a focus targeted toward a single group,because that does not work. Rather, the right approach is toraise the bar for all individuals with disease, recognizingthat some will have a greater gain than others. Thus, widedeployment of quality-driven performance improvementstrategies makes the most sense and is doable.

Dr. Friedewald: Thank you.

Appendix: Recommendations for Implantable Cardio-Defibrillators3

Class I

1. ICD therapy is indicated in patients who are survivorsof cardiac arrest due to VF or hemodynamically un-stable sustained VT after evaluation to define thecause of the event and to exclude any completelyreversible causes. (Level of Evidence: A)

2. ICD therapy is indicated in patients with structural heartdisease and spontaneous sustained VT, whether hemo-dynamically stable or unstable. (Level of Evidence: B)

3. ICD therapy is indicated in patients with syncope ofundetermined origin with clinically relevant, hemo-dynamically significant sustained VT or VF inducedat electrophysiological study. (Level of Evidence: B)

4. ICD therapy is indicated in patients with LV ejectionfraction �35% due to prior myocardial infarctionwho are at least 40 days post-myocardial infarctionand are in NYHA functional Class II or III. (Level ofEvidence: A)

5. ICD therapy is indicated in patients with nonischemicdilated cardiomyopathy who have an LV ejectionfraction �35% and who are in NYHA functionalClass II or III. (Level of Evidence: B)

6. ICD therapy is indicated in patients with LV dysfunc-tion due to prior myocardial infarction who are atleast 40 days post-myocardial infarction, have an LVejection fraction �30%, and are in NYHA functionalClass I. (Level of Evidence: A)

7. ICD therapy is indicated in patients with nonsustained
VT due to prior myocardial infarction, LV ejection


fraction �40%, and inducible VF or sustained VT atelectrophysiological study. (Level of Evidence: B)

Class IIa

1. ICD implantation is reasonable for patients with un-explained syncope, significant LV dysfunction, andnonischemic dilated cardiomyopathy. (Level of Evi-dence: C)

2. ICD implantation is reasonable for patients with sus-tained VT and normal or near-normal ventricularfunction. (Level of Evidence: C)

3. ICD implantation is reasonable for patients with hy-pertrophic cardiomyopathy who have �1 major riskfactor for SCD. (Level of Evidence: C)

4. ICD implantation is reasonable for the prevention ofSCD in patients with arrhythmogenic right ventriculardysplasia/cardiomyopathy who have �1 risk factorfor SCD. (Level of Evidence: C)

5. ICD implantation is reasonable to reduce SCD inpatients with long-QT syndrome who are experienc-ing syncope and/or VT while receiving beta-blockers.(Level of Evidence: B)

6. ICD implantation is reasonable for nonhospitalizedpatients awaiting transplantation. (Level of Evidence: C)

7. ICD implantation is reasonable for patients with theBrugada syndrome who have had syncope. (Level ofEvidence: C)

8. ICD implantation is reasonable for patients with theBrugada syndrome who have documented VT that hasnot resulted in cardiac arrest. (Level of Evidence: C)

9. ICD implantation is reasonable for patients with cat-echolaminergic polymorphic VT who have syncopeand/or documented sustained VT while receiving betablockers. (Level of Evidence: C)

10. ICD implantation is reasonable for patients withcardiac sarcoidosis, giant cell myocarditis, or Cha-gas disease. (Level of Evidence: C)

Class IIb

1. ICD therapy may be considered in patients with non-ischemic heart disease who have an LV ejection frac-tion �35% and who are in NYHA functional Class I.(Level of Evidence: C)

2. ICD therapy may be considered for patients with thelong-QT syndrome and risk factors for SCD. (Levelof Evidence: B)

3. ICD therapy may be considered in patients with syn-cope and advanced structural heart disease in whomthorough invasive and noninvasive investigationshave failed to define a cause. (Level of Evidence: C)

4. ICD therapy may be considered in patients with afamilial cardiomyopathy associated with suddendeath. (Level of Evidence: C)

5. ICD therapy may be considered in patients with LVnoncompaction. (Level of Evidence: C)

Class III

1. ICD therapy is not indicated for patients who do nothave a reasonable expectation of survival with an
acceptable functional status for at least 1 year, even if
they meet ICD implantation criteria specified in theClass I, IIa, and IIb recommendations above. (Levelof Evidence: C)

2. ICD therapy is not indicated for patients with inces-sant VT or VF. (Level of Evidence: C)

3. ICD therapy is not indicated in patients with signifi-cant psychiatric illnesses that may be aggravated bydevice implantation or that may preclude systematicfollow-up. (Level of Evidence: C)

4. ICD therapy is not indicated for NYHA Class IVpatients with drug-refractory HF who are not candi-dates for cardiac transplantation or implantation of acardiac resynchronization therapy device that incor-porates both pacing and defibrillation capabilities.(Level of Evidence: C)

5. ICD therapy is not indicated for syncope of undeter-mined cause in a patient without inducible ventriculartachyarrhythmias and without structural heart disease.(Level of Evidence: C)

6. ICD therapy is not indicated when VF or VT isamenable to surgical or catheter ablation (e.g., atrialarrhythmias associated with the Wolff-Parkinson-White syndrome, right ventricular or LV outflow tractVT, idiopathic VT, or fascicular VT in the absence ofstructural heart disease). (Level of Evidence: C)

7. ICD therapy is not indicated for patients with ventric-ular tachyarrhythmias due to a completely reversibledisorder in the absence of structural heart disease(e.g., electrolyte imbalance, drugs, or trauma). (Levelof Evidence: B)

1. Mirowski M, Reid PR, Mower MM, Watkins L, Gott VL, Schauble JF,Langer A, Heilman MS, Kolenik SA, Frischell RD, Weisfeldt ML. Ter-mination of malignant ventricular arrhythmias with an implantable auto-matic defibrillator in human beings. N Engl J Med 1980;303:322–324.

2. Mirowski M, Mower MM, Langer A, Heilman MS, Schreibman L. Achronically implanted system for automatic defibrillation in active con-scious dogs: experimental model for treatment of sudden death fromventricular fibrillation. Circulation 1978;58:90–94.

3. Epstein AE, DiMarco JP, Ellenbogen KA, Estes NAM, Freedman RA,Gettes LS, Gillinov AM, Gregoratos G, Hammill SC, Hayes DL, HlatkyMA, Newby LK, Page RL, Schonfeld MH, Sitka MJ, Stevenson LW,Sweeny MO. ACC/AHA/HRS 2008 guidelines for device-based ther-apy of cardiac rhythm abnormalities: executive summary. Circulation2008;117:2820–2840.

4. Hohnloser SH, Kuck KH, Dorian P, Roberts RS, Hampton JR, Hatala R,Fain E, Gent M, Connolly SJ, for the DINAMIT Investigators. Prophy-lactic use of an implantable cardioverter-defibrillator after acute myo-cardial infarction. N Engl J Med 2004;351:2481–2488.

5. Fonarow GC, Yancy CW, Albert NM, Curtis AB, Gattis Stough W,Gheorghiade M, Heywood JT, Mehra M, O’Connor CM, Reynolds D,Walsh NM. Improving the use of evidence-based heart failure therapiesin the outpatient setting: the IMPROVE HF performance improvementregistry. Am Heart J 2007;154:12–38.

6. Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R,Domanski M, Troutman C, Anderson J, Johnson G, McNulty SE,Clapp-Channing N, Davidson LD, Fraulo ES, Fishbein DP, Luceri RM,Ip JH, for the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT)Investigators. Amiodarone or an implantable cardioverter-defibrillatorfor congestive heart failure. N Engl J Med 2005;352:225–237.

7. Hernandez AF, Fonarow GC, Liang L, AI-Khatib SM, Curtis LH,LaBresh KA, Yancy CW, Albert NM, Peterson ED. Sex and racialdifferences in the use of implantable cardioverter-defibrillatorsamong patients hospitalized with heart failure. JAMA 2007;298:
1525–1532.

T

Comparison of the Effectiveness and Safety of Low-MolecularWeight Heparin Versus Unfractionated Heparin Anticoagulation

After Heart Valve Surgery

Claudia Bucci, PharmDa,b,*, William H. Geerts, MDc, Andrew Sinclair BScPhmb, andStephen E. Fremes, MDd

Although unfractionated heparin (UFH) is used routinely after heart valve surgery at manyinstitutions, cardiovascular surgery patients have a particularly high risk for developingheparin-induced thrombocytopenia (HIT). The aim of this study was to compare theefficacy and safety of low-molecular-weight heparin (LMWH) or UFH after heart valvesurgery by conducting a retrospective evaluation of consecutive cardiovascular surgerypatients in whom the LMWH dalteparin (n � 100) was used as the postoperative antico-agulant. This group was compared to an earlier group of patients who received UFH (n �103). The main outcomes included the efficacy of the anticoagulant regimens (determinedby the incidence of valve thrombosis, arterial thromboembolic events, and venous throm-boembolic events) and the safety (determined by major bleeding, HIT, thrombotic events inHIT-positive cases, and death). Overall, there were for fewer thrombotic events in theLMWH-treated group (4% vs 11%, p � 0.11). There was a higher rate of bleeding eventsin the UFH-treated group (10% vs 3%, p � 0.08). Six patients in the UFH-treated groupdeveloped HIT, 4 of whom had thrombotic events (HIT with thrombosis). In the LMWH-treated group, 3 patients developed HIT, 1 of whom had HIT with thrombosis. In conclu-sion, in this study, an LMWH regimen after heart valve surgery was effective and safe, withfewer thrombotic, bleeding, HIT, and HIT with thrombosis events. © 2011 Elsevier Inc.
All rights reserved. (Am J Cardiol 2011;107:591–594)
In the past, intravenous unfractionated heparin (UFH)was used routinely at our institution after heart valve re-placement surgery to prevent thrombotic complications(Appendix A). LMWH is associated with a substantially

aDepartment of Pharmacy and bFaculty of Pharmacy, University oforonto; and cThromboembolism Program, Department of Medicine, and

dDivision of Cardiovascular Surgery, Sunnybrook Health Sciences Centre,Toronto, Ontario, Canada. Manuscript received July 15, 2010; revisedmanuscript received and accepted October 11, 2010.

Dr. Bucci has received research grant support from AstraZeneca, Wil-mington, Delaware. Dr. Bucci is a consultant for Sanofi-Aventis, Paris,France; Bristol-Myers Squibb, New York, New York; Bayer Healthcare,Munich, Germany; Boehringer Ingelheim, Ingelheim, Germany; and EliLilly & Company, Indianapolis, Indiana. Dr. Geerts has received researchgrant support from Bayer Healthcare; Pfizer, Inc., New York, New York;and Sanofi-Aventis. Dr. Geerts is a consultant for Bayer Healthcare; Boehr-inger Ingelheim; GlaxoSmithKline, London, United Kingdom; LEOPharma A/S, Ballerup, Denmark; Pfizer, Inc.; and Sanofi-Aventis. Dr.Geerts has received honoraria for presentations from Bayer Healthcare,Boehringer Ingelheim, Pfizer, Inc., and Sanofi-Aventis. Dr. Fremes hasreceived honoraria from Sanofi-Aventis; Bayer Healthcare; AstellasPharma US, Inc., Deerfield, Illinois; Novo Nordisk A/S, Bagsværd, Den-mark; Novadaq, Bonita Springs, Florida; Medtronic, Inc., Minneapolis,Minnesota; Edwards Lifesciences, Irvine, California; Sorin Group USA,Inc., Arvada, Colorado; and Merck & Company, Whitehouse Station, NewJersey. Dr. Fremes has received research support from St. Jude Medical,Inc., St. Paul, Minnesota; Aventis; Proctor & Gamble, Cincinnati, Ohio;Medicure, Winnipeg, Manitoba, Canada; and Merck & Company.

E-mail address: [email protected] (C. Bucci).

lower rate of heparin-induced thrombocytopenia (HIT) andHIT with thrombosis than UFH and may be a safer alter-native after heart valve surgery.1 We replaced intravenousand subcutaneous UFH with subcutaneous low-molecular-weight heparin (LMWH) in prophylactic or therapeuticdoses for early anticoagulation after heart valve replacementsurgery (Appendix B). The objective of the study was toassess the efficacy and safety of anticoagulation withLMWH after heart valve surgery compared to UFH.

Methods

In March 2006, we implemented an “avoid-heparin pol-icy” after cardiovascular surgery (Appendixes A and B).Intraoperative UFH was used in all cases. We conducted aretrospective evaluation of consecutive patients in whomthe LMWH dalteparin was used, and we compared thisgroup to an earlier group of patients who received UFH.This study was approved by the ethics review board ofSunnybrook Health Sciences Centre. The main outcomemeasures included the efficacy (determined by the incidenceof valve thrombosis, arterial thromboembolic events, andvenous thromboembolic events) and the safety (determinedby major bleeding, HIT, thrombotic events in HIT-positivecases, and death) of the 2 anticoagulant regimens. All out-comes collected occurred during the operative hospital ad-mission.

Confirmed HIT was defined by 1 of the following: pos-itive serotonin release assay, positive HIT enzyme-linked
immunosorbent assay plus high clinical probability for HIT
www.ajconline.org


Tu

R

ttpggp

gwTsUpp

LtUvngb

AAML


or strongly positive HIT enzyme-linked immunosorbent as-say (optical density �1.0). HIT was ruled out in patientswith negative results on HIT enzyme-linked immunosorbentassay or serotonin release assay.

Major bleeding was defined as any overt bleeding meet-ing �1 of the following criteria: proved fatal bleeding,intracranial hemorrhage (computed tomography or mag-netic resonance imaging required), retroperitoneal bleeding(ultrasound, computed tomography, or magnetic resonanceimaging required), bleeding requiring an intervention (peri-cardial bleeding requiring reoperation or catheter drainageof blood, pleural bleeding requiring thoracotomy or chesttube, gastrointestinal bleeding requiring surgery or endo-scopic treatment, wound bleeding requiring reoperation),other life-threatening bleeding at a critical site, bleedingrequiring transfusion of �2 U of red blood cells, or bleedingthat resulted in chronic sequelae or prolongation of thehospital stay. Bleeding requiring pericardiocentesis, thora-centesis, or diagnostic endoscopy alone was not consideredmajor. Nonmajor bleeding was defined as any of the fol-lowing: epistaxis requiring nasal packing, airway bleeding,hematuria, hematemesis (but not just coffee grounds), orgastrointestinal bleeding (frank blood or melena stools) notrequiring an intervention.

All analyses were done using InStat version 3 (GraphPadSoftware, San Diego, California). All statistical tests were 2 sidedand used a p value of 0.05 as the threshold for statistical signifi-cance. Baseline discrete variables are presented as frequen-cies or percentages, while continuous variables are pre-


Characteristic Heparin(n � 103

ge (years) 65.1 � 12.6 (31ge �75 years 23 (22%en/women 72/31

eft ventricular function1 57 (55%2 22 (21%3 16 (16%4 8 (8%)Atrial fibrillation* 66 (64%Renal dysfunction† 13 (13%

Valve replacementSingle 95 (92%

Mechanical aortic 27Tissue aortic 21Mechanical mitral 17Tissue mitral 14Mitral annuloplasty 15Tricuspid repair 1

Double 8 (8%)Coronary bypass 43 (42%Length of surgery (hours) 4.9 � 1.7 (2.3Length of stay after surgery (days) 15.9 � 9.9 (5–6

Data are expressed as mean � SD (range) or as number (percentage).* Includes transient and chronic episodes. Transient atrial fibrillation oc

0.57). Chronic atrial fibrillation was present preoperatively and/or was persgroups, respectively (p � 0.17).

† Creatinine clearance �30 ml/min.

sented as mean � SD or as median (interquartile range). t

he frequencies of the clinical end points were comparedsing a chi-square or Fisher’s exact tests.

esults

The control group consisted of 103 consecutive patientsreated with UFH after heart valve surgery from April 2004o May 2006. These patients received only UFH in thera-eutic (83%) or prophylactic (17%) doses. The controlroup patients were compared to 100 heart valve patientsiven therapeutic (73%) or prophylactic (27%) dalteparinostoperatively from March 2006 to August 2007.

The 2 groups were similar for a large number of demo-raphic and clinical characteristics (Table 1). The mean ageas approximately 65 years, and 68% of patients were men.hrombotic and bleeding risk factors in the 2 groups wereimilar apart from greater postoperative aspirin use in theFH patients (Table 2). Approximately 60% of the studyopulation had �1 risk factor for thrombosis, and �80% ofatients had �1 risk factor for bleeding.

Overall, there were fewer thrombotic events in theMWH-treated group, although the difference was not sta-

istically significant (4% vs 11%, p � 0.11; Table 3). In theFH group, there were 11 thrombotic events (5 strokes, 1alve thrombosis, 1 ischemic bowel, 2 transient global am-esia, 1 foot embolus, 1 kidney infarction). In the dalteparinroup, there were 4 thrombotic events (3 strokes, 1 ischemicowel). The thromboembolic events are listed in Table 4 . In

Dalteparin(n � 100)

pValue

66.1 � 12.1 (43–84) 0.5724 (24%) 0.87

67/33 0.760.65

52 (52%)28 (28%)15 (15%)5 (5%)

68 (68%) 0.6716 (16%) 0.55

0.6390 (90%)

2319218

181

10 (10%) 0.6332 (32%) 0.19

5.3 � 1.9 (2.7–12.9) 0.0916.1 � 9.5 (6–64) 0.87

n 49 and 43 patients in the heparin and LMWH groups, respectively (p �fter surgery and occurred in 17 and 25 patients in the heparin and LMWH

)

–87))

)))

))

)

)–10.2)1)

curred iistent a

he UFH-treated group, 4 thrombotic events occurred in

1capgtp

btwi14twg

D

rgph

Vth

LU

d

PPD

593Valvular Heart Disease/

HIT-positive patients, while in the LMWH-treated group, 1thrombotic event occurred in an HIT-positive patient.

There was a higher rate of bleeding events in the UFH-treated group (10% vs 3%, p � 0.08; Table 3). There were0 major bleeding events in the UFH-treated group (5 peri-ardial bleeds, 4 retroperitoneal bleeds, and 1 hemothorax;ll these patients received concomitant aspirin and/or clo-idogrel) and 3 major bleeding events in the LMWH-treatedroup (2 pericardial bleeds, 1 hemothorax; 2 of these pa-ients received concomitant aspirin and/or clopidogrel). One

Table 2Risk factors for thrombosis and bleeding

Risk Factor Heparin(n � 103)


ThromboticAtrial fibrillation* 17 (17%) 26 (26%)Grade 4 left ventricle 8 (8%) 5 (5%)Previous thromboembolism 6 (6%) 12 (12%)Left atrial enlargement 17 (17%) 12 (12%)Previous myocardial infarction 13 (13%) 10 (10%)

BleedingPostoperative aspirin 93 (90%) 32 (32%)Renal dysfunction 13 (13%) 16 (16%)Clopidogrel 3 (3%) 1 (1%)Nonsteroidal anti-

inflammatory drug12 (12%) 27 (27%)

Intraoperative stroke 1 (1%) —International normalized ratio

�5 (postoperatively)4 (4%) 2 (2%)

Coagulation disorder† — 3 (3%)Therapeutic anticoagulation

(postoperative)86 (83%) 73 (73%)

Warfarin (postoperative) 100 (97%) 97 (97%)

* Represents chronic atrial fibrillation (i.e., present preoperatively or atischarge).

† Includes factor VII deficiency (n � 1) and sickle-cell disease (n � 2).

Table 3Comparison of efficacy and safety outcomes

Outcome Heparin(n � 103)


p Value

Thrombotic events 11 (11%) 4 (4%) .11Stroke 5 3Valve thrombosis 1 —Ischemic bowel 1 1Transient global 2AmnesiaFoot embolus 1 —Infarction kidney 1 —

Bleeding events 10 (10%) 3 (3%) 0.08Pericardial 5 2Retroperitoneal 4 —Hemothorax 1 1atients with HIT 6 (6%) 3 (3%) 0.50atients with HIT with thrombosis 4 (4%) 1 (1%) 0.37eath due to HIT 1 (1%) 1 (1%) 1.00

atient in each group had a nonmajor lower gastrointestinal

leed. Most of the bleeds occurred in patients receivingherapeutic anticoagulation with UFH or LMWH or witharfarin. Concomitant aspirin increased the risk for bleed-

ng events (odds ratio 7.43, 95% confidence interval 1.85 to26). Six patients in the UFH-treated group developed HIT,of whom had thrombotic events (HIT with thrombosis). In

he LMWH-treated group, 3 patients developed HIT, 1 ofhom had HIT with thrombosis. There was 1 death in eachroup during hospitalization, both related to HIT (Table 3).

iscussion

Patients who have undergone cardiac surgery routinelyeceive UFH during and after surgery. The product mono-raphs for LMWHs warn against the use of LMWH forrevention of thromboembolism in patients with prostheticeart valves, including those who are pregnant.2 This is

based on 2 cases of valve thrombosis in pregnant womenreceiving enoxaparin.3 The Anticoagulation in Prosthetic

alves and Pregnancy Consensus Report Panel concludedhat the level of anticoagulation with enoxaparin may notave been optimal in these cases.4

LMWH may be a safer alternative to UFH in cardiacsurgery patients because of the lower risk for HIT. How-ever, to date, the safety and efficacy of LMWH after me-chanical heart valve surgery has been poorly evaluated. In anonrandomized case series of 208 patients who underwentheart valve replacement, therapeutic anticoagulation wasmore rapidly and predictably achieved with LMWH thanwith UFH.5 In a larger study with no control group, the useof the LMWH enoxaparin as bridging to therapeutic anti-coagulation with warfarin after mechanical valve replace-ment appeared to be safe and effective.6 In a small study,

MWH patients were matched to patients who receivedFH after mechanical heart valve implantation.7 Although

bridging with LMWH was as safe and effective as bridgingwith UFH, LMWH was associated with reduced length ofhospital stay and costs.

In this study, we found that an LMWH regimen afterheart valve surgery was effective and safe. In addition, therewas a lower risk for thrombosis and bleeding in the LMWH-treated group. No cases of valve thrombosis occurred in theLMWH group of this study. There were also fewer HIT andHIT with thrombosis events in patients receiving LMWH.Our study provides data supporting the use of LMWH inpatients with newly implanted heart valves. We believe thatthis is the first report assessing the development of HIT inthis patient population.

Our study was retrospective in nature and consisted of arelatively small sample. Although the baseline characteris-tics were similar, confounding may be present. Confound-ing from unmeasured factors such as the motivation of theclinical staff and increased attention to postoperative anti-coagulation may have been present. We believe that this isthe first intervention study to systematically attempt to re-duce HIT in cardiac surgery. The results of this preliminarystudy are important as part of ongoing quality assurance of
the protocol implemented at our institution.

M

T

q

M

T


Appendix A: Unfractionated Heparin Anticoagulation Protocol After Heart Valve Replacement

Valve Position Additional RiskFactors*

Heparin Therapy Aspirin Use Target INR

echanical valvesAortic No 5,000 U SC q12h until INR �2.

If INR �2 on POD 4, start IV heparin.None 2.0–3.0

Aortic Yes 5,000 U SC q12h until INR �2.If INR �2 on POD 4, start IV heparin.

ECASA 81 mg/day 2.0–3.0

Mitral No 500 U/hour IV starting 12 hourspostoperatively for 96 hours.

If INR �2 on POD 4, start IV heparin.

None 2.5–3.5

Mitral Yes 500 U/hour IV starting 12 hourspostoperatively for 96 hours.

If INR �2 on POD 4, start IV heparin.

ECASA 81 mg/day 2.5–3.5

issue valvesAortic No None (except for DVT prophylaxis). ECASA 325 mg/day No warfarinAortic Yes 5,000 U SC q12h until INR �2.

If INR �2 on POD 4, start IV heparin.None 2.0–3.0

Mitral and mitral annuloplasty No 5,000 U SC q12h until INR �2.If INR �2 on POD 4, start IV heparin.

ECASA 325 mg/day; start after3 months of warfarin

2.0–3.0; warfarin onlyfor 3 months

Mitral and mitral annuloplasty Yes 5,000 U SC q12h until INR �2.If INR �2 on POD 4, start IV heparin.

None 2.0–3.0

*Atrial fibrillation, large left atrium, left atrial thrombus, previous thromboembolism.DVT � deep venous thrombosis; ECASA � enteric-coated aspirin; INR � international normalized ratio; IV � intravenous; POD � postoperative day;

12h � every 12 hours; SC � subcutaneous.

Appendix B: Low-Molecular-Weight Heparin Anticoagulation Protocol After Heart Valve Replacement

Valve Position LMWH Therapy Aspirin Target INR

echanical valvesAortic Dalteparin 2,500 U SC qhs.

If INR �2 at 96 hours postoperatively,increase dalteparin dose.*

None 2.0–3.0

Mitral Dalteparin 5,000 U SC qhs.If INR �2 at 96 hours postoperatively,

increase dalteparin dose.*

None 2.5–3.5

issue valvesAortic without risk factors None ECASA 325 mg/day NoneAortic with risk factors* Dalteparin 2,500 U SC qhs.

If INR �2 at 96 hours postoperatively,increase dalteparin dose.*

None 2.0–3.0

Mitral and mitral annuloplasty Dalteparin 2,500 U SC qhs. None 2.0–3.0If INR �2 at 96 hours postoperatively,

increase dalteparin dose.*Warfarin for 3 months (unless risk

factors are present), thenECASA 325 mg/day is added

*Sustained or intermittent atrial fibrillation lasting �48 hours, maze procedure.ECASA � enteric-coated aspirin; INR � international normalized ratio; qhs � at bedtime; SC � subcutaneous.

1. Martel N, Lee J, Wells PS. Risk for heparin-induced thrombocytopeniawith unfractionated heparin and low-molecular-weight heparin throm-boprophylaxis: a meta-analysis. Blood 2005;106:2710–2715.

2. Canadian Pharmacists Association. Product monograph heparins: lowmolecular weight. In: e-CPS: Compendium of Pharmaceuticals andSpecialties. Available at: http://www.pharmacists.ca/content/products/ecps_english.cfm. Accessed February 4, 2010.

3. Shapira Y, Sagie A, Battler A. Low-molecular-weight heparin for the treatmentof patients with mechanical heart valves. Clin Cardiol 2002;25:323–327.

4. Anticoagulation and enoxaparin use in patients with prosthetic heart
vavles and/or pregnancy. Clin Cardiol Consensus Rep 2002;3:1–20.
5. Montalescot G, Polle V, Collet JP, Leprince P, Bellanger A, Gandjba-khch I. Low molecular weight heparin after mechanical heart valvereplacement. Circulation 2000;101:1083–1086.

6. Meurin P, Tabet JY, Weber H, Renaud N, Ben Driss A. Low-molecular-weight heparin as a bridging anticoagulant early after mechanical heartvalve replacement. Circulation 2006;113:564–569.

7. Fanikos J, Tsilimingras K, Kucher N, Rosen AB, Hieblinger MD,Goldhaber SZ. Comparison of efficacy, safety and cost of low-molec-ular weight heparin with continuous-infusion unfractionated heparin forinitiation of anticoagulation after mechanical prosthetic valve implan-
tation. Am J Cardiol 2004;93:247–250.
http://www.pharmacists.ca/content/products/ecps_english.cfm

http://www.pharmacists.ca/content/products/ecps_english.cfm

HL

s2

BTaSgdA

0d

Seeking Optimal Relation Between Oxygen Saturation andHemoglobin Concentration in Adults With Cyanosis from

Congenital Heart Disease

Craig S. Broberg, MDa,b,*, Ananda R. Jayaweera, PhDa, Gerhard P. Diller, MDb,Sanjay K. Prasad, MDb, Swee Lay Thein, MDc, Bridget E. Bax, PhDd, John Burman, MDb, and

Michael A. Gatzoulis, MD, PhDb

In patients with cyanosis from congenital heart disease, erythropoiesis is governed by manyfactors that can alter the expected relation between the oxygen saturation (O2sat) andhemoglobin concentration. We sought to define the relation between the O2sat and hemo-globin in such patients and to predict an ideal hemoglobin concentration for a given O2sat.Adults with congenital heart defects and cyanosis were studied prospectively using bloodtests and exercise testing. Nonoptimal hemoglobin was defined as any evidence of inade-quate erythropoiesis (i.e., iron, folate, or vitamin B12 deficiency, increased erythropoietin,reticulocytosis, or a right-shifted oxygen-hemoglobin curve). For patients without thesefactors, a linear regression equation of hemoglobin versus O2sat was used to predict theoptimal hemoglobin for all patients. Of the 65 patients studied, 21 met all the prestudycriteria for an optimal hemoglobin. For all patients, no correlation was found between O2satand hemoglobin (r � �0.22). However, a strong linear correlation was found for thosemeeting the criteria for optimal hemoglobin (r � �0.865, p <0.001). The optimal hemo-globin regression equation was as follows: predicted hemoglobin � 57.5 � (0.444 � O2sat).A negative correlation was found between the hemoglobin difference (predicted minusmeasured) and exercise duration on cardiopulmonary exercise testing (r � �0.396, p �0.005) and the 6-minute walk distance (r � �0.468, p <0.001). In conclusion, a strongrelation between O2sat and hemoglobin concentration can be shown in stable cyanoticpatients and used to predict an optimal hemoglobin. This relation might be useful indefining functional anemia in this group. © 2011 Published by Elsevier Inc. (Am J
Cardiol 2011;107:595–599)
tcWc

pn

In the clinical care of cyanotic patients with congenitalheart disease, it is necessary to assess the appropriateness ofa measured hemoglobin level for a given oxygen saturation(O2sat). Although multiple factors can influence both hemo-globin and O2sat, a tool to predict the optimal relationbetween these variables would be valuable, particularly forsituations in which the hemoglobin might be significantly

aAdult Congenital Heart Disease Program, Oregon Health and SciencesUniversity, Portland, Oregon; bRoyal Brompton and Harefield National

ealth Service Trust, National Heart and Lung Institute, Imperial Collegeondon, London, United Kingdom; cDepartment of Haematological Med-

icine, King’s College London School of Medicine and King’s CollegeHospital, London, United Kingdom; dChild Health, St. George’s, Univer-ity of London, London, United Kingdom. Manuscript received July 30,010; manuscript received and accepted October 1, 2010.

The study was funded by the Clinical Research Committee, Royalrompton Hospital. Dr. Broberg has received support from the Waringrust (London, United Kingdom) through the Royal Brompton Hospitalnd the Tartar Trust (Portland, Oregon) through the Oregon Health andciences University. Dr. Gatzoulis and the Royal Brompton Adult Con-enital Heart Centre have received support from the British Heart Foun-ation, London, United Kingdom and unrestricted research funds fromctelion UK (London, United Kingdom).

*Corresponding author: Tel: (503) 494-8750; fax: (503) 494-8550.
E-mail address: [email protected] (C. Broberg).
002-9149/11/$ – see front matter © 2011 Published by Elsevier Inc.oi:10.1016/j.amjcard.2010.10.019

less than expected, such as postoperatively or after severehemoptysis. We hypothesized that by controlling for factorsthat might alter this relation, particularly those that mightlimit erythropoiesis, such as iron deficiency, an ideal linearrelation could be found that would define an “optimal”hemoglobin level for a given O2sat. We also hypothesizedhat patients with an optimal hemoglobin might have alinical advantage, as measured by the exertional capacity.e therefore prospectively measured the variables that

ould potentially alter the hemoglobin–O2sat association todetermine their optimal relation.

Methods

We prospectively enrolled consecutive adults with con-genital heart disease in a descriptive cross-sectional study.Patients gave consent, and institutional ethics review ap-proved the protocol. Patients were included if they had aknown congenital defect with a right-to-left shunt. We in-cluded patients with a wide range of O2sat, including someatients who had undergone previous repair and had aormal O2sat at the study. All tests were obtained within a

24-hour period. Other data from the present study have beenpreviously reported.1,2

The patients were recruited and seen at the Royal Bro-
mpton Hospital. Additional blood testing was done at
www.ajconline.org


tbnrptvecaK(

pet

ic

lt6p

TRVSR


King’s College Hospital and St. George’s Hospital (Lon-don, United Kingdom). The analyses were performed atthe Oregon Health and Sciences University (Portland,Oregon).

O2sat was measured using transcutaneous spectrometry inhe finger after 5 minutes of rest in the sitting position. Thelood was drawn in the morning with the patient in aonfasting state using a venous cannula in the antecubitalegion. The hemoglobin concentration, packed cell volume,latelet count, basic serum chemistry panels, liver functionests, iron, ferritin, transferrin saturation, red blood cellitamin B12, folate, thyroid stimulating hormone, and serumrythropoietin3 were measured. The percentage of hypo-hromic cells and reticulocyte count were measured usingn automated Coulter counter (Advia 120, Bayer, Unitedingdom). The partial oxygen pressure at half saturation

P50) of the O2–hemoglobin dissociation curve was alsomeasured (Hem-O-Scan, American Instrument Company,Silver Springs, Maryland). The whole blood viscosity overa range of shear was measured using a rotational viscome-ter. The viscosity was then remeasured after the hematocrithad been diluted to 45% using autologous serum.1 Theatients also performed a 6-minute walk test and treadmillxercise, with the measured oxygen consumption and ven-ilatory efficiency recorded, as previously described.2

After the collection of all data, we identified those pa-tients with any evidence of potentially inadequate or exces-sive erythropoiesis, according to the presence of �1 of thefollowing a priori criteria: evidence of iron deficiency, vi-tamin B12 or folate deficiency, elevated serum erythropoie-tin, reticulocytosis, hypochromia, or a significant rightwardshift of the O2–hemoglobin dissociation curve (Table 1).We also excluded patients using various clinical criteria,including acute hospitalization, therapeutic phlebotomywithin the previous 6 months, and recent significant hemop-tysis (requiring hospitalization). Patients with a patent duc-tus arteriosus and differential cyanosis were also excludedfrom the optimal category because of the uncertainty ofwhat the mean O2sat would be. Patients using supplementaloxygen regularly were excluded because their O2sat at roomair might not have accurately reflected their average dailysaturation.

After exclusion of any patient who had met these criteria,a plot of the O and measured hemoglobin was made. A

Table 1Prestudy criteria for determination of adequate erythropoiesis

Variable Cutoff NormalRange

ExcludedPatients (n)

ransferrin saturation �20% 20–45 26ed blood cell folate (�g/L) �200 164–900 0itamin B12 (ng/L) �180 180–900 2erum erythropoietin (IU/L) �25 variable 9eticulocyte count (%) �2 �2% 8

Hypochromic cell count (%) �6 �6% 7P50 of oxygen–hemoglobin

dissociation curve (mm Hg)�29 25–29 10

Patients who met all these criteria were considered to have optimalhemoglobin. Additional clinical exclusions listed in text.

2satlinear regression equation was defined, together with con-

fidence intervals, around this line. Using the regressionequation, the values for the predicted hemoglobin weremade, and the difference between the predicted and mea-sured hemoglobin was obtained (hemoglobin difference) foreach patient. The clinical variables between the patientswith and without an optimal hemoglobin level were com-pared using Student’s t test and correlations using Pearson’scoefficient. The data are expressed as the mean � SD, andp �0.05 was considered statistically significant. No adjust-ment was made for multiple comparisons.

Results

A total of 65 patients were studied (mean age 36 � 12years, 67% women). For the whole group, the O2sat at restwas 81 � 8%, hemoglobin was 19.6 � 2.9 g/dl, and he-matocrit was 60 � 8%. The anatomic diagnoses are listed inTable 2, as well as their group designation. All but 1 patienthad pulmonary vascular disease. Of the 65 patients, 20 hadlikely been cyanotic at birth, with 45 (largely with simpleshunts) having developed right to left shunting over time.No patient was found to have significant renal, liver, orthyroid dysfunction.

Of the 65 patients studied, 44 met �1 criteria for exclu-sion, and most of them met at least 2 criteria (Table 1). Theexclusions on clinical grounds included 16 using supple-mental oxygen, 8 with differential cyanosis, 10 with recentphlebotomy, and 2 with recent hemoptysis. The most com-mon exclusion criterion was iron deficiency, and many ofthose patients met additional criteria associated with irondeficiency (e.g., phlebotomy, hemoptysis, P50 shift, orerythropoietin elevation). After the exclusions, 21 patientshad met all the criteria for adequate erythropoiesis.

For the entire cohort, no significant relation was foundbetween O2sat and hemoglobin (Figure 1). In contrast,when patients with evidence of inadequate erythropoiesiswere excluded, a strong linear relation was found. Theslope and intercept for the regression line defined a pre-dicted optimal hemoglobin as follows: predicted hemo-globin � – 0.444(O2sat) � 57.5.

From this, the predicted hemoglobin for a given O2sat,ncluding the upper and lower confidence intervals, werealculated (Table 3 and Figure 2).

To establish the clinical relevance of our predictedine, we sought correlations with the functional parame-ers. Those with an optimal hemoglobin level had a better-minute walk test distance (415 � 119 vs 333 � 112 m,� 0.011) and treadmill exercise duration (7.48 � 2.70

vs 5.13 � 2.01 minutes, p � 0.001). The correlationcoefficient values between the hemoglobin difference andthe outcome variables are listed in Table 4. A significantinverse correlation was found with the 6-minute walkdistance (Figure 3) and exercise duration, such that agreater difference (i.e., measured hemoglobin less thanpredicted) was associated with poorer function. A similarrelation was seen with the ventilatory efficiency slopeand heart rate reserve. These same variables did notcorrelate with the measured hemoglobin (Figure 3 andTable 4). The correlation with the peak oxygen consump-tion and percentage of the maximum predicted oxygen
consumption did not reach statistical significance. Blood

b

s

p

POT

ar resis

9

597Congenital Heart Disease/Optimal Hemoglobin in Cyanosis

viscosity was not different between the optimal and non-optimal groups, even after adjustment for hematocrit(50 � 10 vs 49 � 10 mPa.s at high shear, p � 0.82, and4.29 � 0.5 vs 4.45 � 1.1 mPa.s at low shear, p � 0.63,

Table 2Study population by anatomic lesion

Diagnosis Patients (n)

Atrial septal defect 4Ventricular septal defect 27Atrioventricular septal defect 8Patent ductus arteriosus 8Truncus arteriosus 6Transposition of great arteries 5

ulmonary atresia with ventricular septal defect 2ther complex 5otal 65

All transposition patients also had ventricular septal defect.* Developed right-to-left shunt in setting of elevated pulmonary vascul

Figure 1. Scatter plot for hemoglobin concentration versus at rest O2sat. FoO2sat (dotted line). For patients meeting the criteria for adequate erythropoptimal patients and for all patients shown.

Table 3Prediction of optimal hemoglobin for a given oxygen saturation

OxygenSaturation (%)

PredictedHemoglobin (g/dl)

95% CI (g/dl)

3 16.1 14.4–17.990 17.5 16.0–19.087 18.8 17.5–20.185 19.7 18.4–21.083 20.6 19.2–21.980 21.9 20.4–23.477 23.2 21.4–25.075 24.1 22.1–26.173 25.0 22.8–27.3

CI, confidence interval.

respectively). p

Discussion

The concept of determining an ideal set point for eryth-ropoiesis in congenital heart disease with cyanosis is notnew. An “optimal hematocrit” between oxygen delivery andhyperviscosity was studied decades ago,4 although limitedy the use of ex vivo models.5 Few clinical studies have

addressed this relation and largely only in children or ado-lescents.5,6 A linear relation has been shown, although lessteep than ours.5,7,8 A right shift of the oxyhemoglobin

dissociation curve has been seen in iron-deficient children.9

We also previously reported a less steep relation in iron-replete adults with patients with Eisenmenger syndrome.6

Our present study, in contrast, was more fastidious with theexclusions, accounting for the steeper slope and narrowerconfidence interval found. Hence, the results of the presentstudy have shown a strong linear association between thehemoglobin level and O2sat that could clinically distinguishatients according to their exercise capacity. For the ex-

oticBirth

CyanosisDeveloped*

Excluded(Nonoptimal)

Included (Optimal)

4 2 225 11 168 7 18 8 00 6 00 4 10 2 00 4 1

45 44 21

tance (Eisenmenger reaction).

tire population, no significant relation was found between hemoglobin anda strong linear relation was found (solid line). Regression equations for

CyanSince

02006525

20

r the enoiesis,

ected range of O2sat, we estimated the relation could be

la

6

6


simplified more conveniently as follows: predicted hemo-globin � 61 – (O2sat/2).

Because all patients with an O2sat �75% met the criteriafor having a nonoptimal hemoglobin, it was impossible topredict the optimal hemoglobin for such patients. Extrapo-lation of our data would predict a very high hemoglobin,which might not be achievable without serious hyperviscos-ity.10 However, any patient with an O2sat �70% is arguablynot in a state of balanced erythropoiesis, because this mustreflect increased tissue oxygen extraction. The greatest he-moglobin in our optimal group was 25 g/dl (packed cellvolume 73%) in a stable patient. We previously reported noadverse effects of viscosity on exercise capacity in thispopulation.1 Hyperviscosity and its related symptoms areikely far more complex than ex vivo methods can measurend very different at the capillary level, in particular.11

Daily activity will vary from patient to patient and will alsoaffect the drive to erythropoiesis.

Figure 2. Predicted O2sat–hemoglobin relation. Relation based on regres-sion slope obtained for optimal patients. Upper and lower confidenceintervals shown. Raw values are listed in Table 3.

Table 4Correlations between exercise parameters and hemoglobin

Functional Variable MeasuredHemoglobin

HemoglobinDifference

(Predicted �Measured)

r p Value r p Value

-Minute walk distance (m) �0.036 0.787 �0.468 �0.001*Exercise duration (minutes) �0.130 0.377 �0.396 0.005*Peak oxygen consumption

(ml/kg/min)�0.157 0.298 �0.261 0.080

Ventilatory efficiency slope 0.107 0.477 0.328* 0.026*Heart rate reserve (beats/min) �0.054 0.717 �0.388* 0.007*

Correlations with measured hemoglobin were not significant; whereascorrelations with hemoglobin difference was significant. Negative relationindicates patients with greater difference (i.e., hemoglobin less than ex-pected) had poorer exercise capacity.

* Statistically significant.

The presence of an optimal hemoglobin does not mean

the patient is asymptomatic, because many factors willcontribute to symptoms in this group. We did not addresswhether the manipulation of hemoglobin levels to an“optimal” level as we have defined would have any effecton symptoms, exercise capacity, or prognosis, althoughwe, and others, have shown improvement after treatmentof iron deficiency in patients with chronic cyanosis.12,13

Exercise capacity also has multiple determinants. Thepurpose of comparing the exercise data in the presentstudy was solely to establish whether the predictive for-mula had any functional relevance. Because at least acomponent of exertional capacity correlated with hemo-globin difference but not hemoglobin concentration itself(Figure 3), the relation we defined seems to have clinicalsignificance. We know of no other method of validatingour results.

Other limitations deserve comment. As an initial explo-ration of this association, we had no guidance on whichfactors would be most important, and not all were impor-tant. Gender differences were not considered, because ourstudy did not have a large enough sample size to justify aseparate analysis of men versus women. We did not studypatients with Fontan physiology, although often such pa-

Figure 3. Correlations between hemoglobin and measured walk distance.(A) Relation between 6-minute walk distance and predicted–measuredhemoglobin was significant. Negative hemoglobin difference indicatedhemoglobin was greater than predicted. (B) In contrast, relation between-minute walk distance and measured hemoglobin was poor.

tients are cyanotic. We have no reason to suspect this

599Congenital Heart Disease/Optimal Hemoglobin in Cyanosis

relation would not be relevant to this group also, and thisdeserves additional investigation. However, we do not thinkour prediction formula should be applied to other cyanoticconditions, such as lung disease, nor to children with con-genital heart defects.

1. Broberg CS, Bax BE, Okonko DO, Rampling MW, Bayne S, HarriesC, Davidson SJ, Uebing A, Khan AA, Thein S, Gibbs JS, Burman J,Gatzoulis MA. Blood viscosity and its relation to iron deficiency,symptoms, and exercise capacity in adults with cyanotic congenitalheart disease. J Am Coll Cardiol 2006;48:356–365.

2. Broberg CS, Ujita M, Prasad S, Li W, Rubens M, Bax BE, DavidsonSJ, Bouzas B, Gibbs JS, Burman J, Gatzoulis MA. Pulmonary arterialthrombosis in Eisenmenger syndrome is associated with biventriculardysfunction and decreased pulmonary flow velocity. J Am Coll Car-diol 2007;50:634–642.

3. Tyndall MR, Teitel DF, Lutin WA, Clemons GK, Dallman PR. Serumerythropoietin levels in patients with congenital heart disease. J Pedi-atr 1987;110:538–544.

4. Crowell JW, Smith EE. Determinant of the optimal hematocrit. J ApplPhysiol 1967;22:501–504.

5. Berman W Jr, Wood SC, Yabek SM, Dillon T, Fripp RR, Burstein R.Systemic oxygen transport in patients with congenital heart disease.Circulation 1987;75:360–368.

6. Diller GP, Dimopoulos K, Broberg CS, Kaya MG, Naghotra US,Uebing A, Harries C, Goktekin O, Gibbs JS, Gatzoulis MA. Presen-

tation, survival prospects, and predictors of death in Eisenmengersyndrome: a combined retrospective and case-control study. EurHeart J 2006;27:1737–1742.

7. Gidding SS, Stockman JA III. Erythropoietin in cyanotic heart disease.Am Heart J 1988;116:128–132.

8. Gidding SS, Bessel M, Liao YL. Determinants of hemoglobin con-centration in cyanotic heart disease. Pediatr Cardiol 1990;11:121–125.

9. Gidding SS, Stockman JA III. Effect of iron deficiency on tissueoxygen delivery in cyanotic congenital heart disease. Am J Cardiol1988;61:605–607.

10. Rosove MH, Perloff JK, Hocking WG, Child JS, Canobbio MM,Skorton DJ. Chronic hypoxaemia and decompensated erythrocytosis incyanotic congenital heart disease. Lancet 1986;2:313–315.

11. Kontras SB, Bodenbender JG, Craenen J, Hosier DM. Hyperviscosityin congenital heart disease. J Pediatr 1970;76:214–220.

12. Gaiha M, Sethi HP, Sudha R, Arora R, Acharya NR. A clinico-hematological study of iron deficiency anemia and its correlation withhyperviscosity symptoms in cyanotic congenital heart disease. IndianHeart J 1993;45:53–55.

13. Tay EL, Peset A, Papaphylactou M, Inuzuka R, Alonso-Gonzalez R,Giannakoulas G, Tzifa A, Goletto S, Broberg C, Dimopoulos K,Gatzoulis MA. Replacement therapy for iron deficiency improvesexercise capacity and quality of life in patients with cyanotic congen-ital heart disease and/or the Eisenmenger syndrome. Int J Cardiol 2010
[Epub ahead of print].

cpupsce

Relation of Pulse Pressure to Blood Pressure Response to Exercisein Patients With Hypertrophic Cardiomyopathy

Kevin S. Heffernan, PhD*, Martin S. Maron, MD, Eshan A. Patvardhan, MBBS,Richard H. Karas, MD, PhD, and Jeffrey T. Kuvin, MD; the Vascular Function Study Group

Almost one third of patients with hypertrophic cardiomyopathy (HC) will have an abnor-mal blood pressure response (ABPR) to exercise, and this has been associated with a greaterrisk of sudden cardiac death. In the present study, we examined the association between thesteady (mean arterial pressure) and pulsatile (pulse pressure) blood pressure components asthey relate to ABPR in patients with HC (n � 70). All patients completed a standard Bruceprotocol during symptom-limited stress testing with concurrent hemodynamic measure-ments. Pulse pressure (PP) was significantly greater in patients with HC with an ABPR(n � 19) than in the patients with HC without an ABPR to exercise (p <0.05). Accordingto binary logistic regression analysis, PP at rest was a significant predictor of ABPR inpatients with HC (p <0.05). Mean arterial pressure was not significantly different betweenthe 2 groups, nor was it a predictor of an ABPR in the presence of HC. Those within thegreatest tertile of PP at rest were 4.8 times more likely to have an ABPR than those withinthe lowest PP tertile (95% confidence interval 1.24 to 18.2, p <0.05). In conclusion,elevations in PP at rest might identify patients with HC at a greater risk of having an ABPRduring exercise. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:
600–603)
Approximately 30% of patients with hypertrophic car-diomyopathy (HC) will have an abnormal blood pressureresponse (ABPR) to exercise, categorized as a failure toincrease (or a potential decrease) in systolic blood pressurewith an increase in exercise intensity. This ABPR has beenassociated with sudden cardiac death in patients withHC.1–4 Blood pressure (BP) has both pulsatile and steadyomponents. The pulsatile component of BP, estimated byulse pressure (PP), reflects the integration of left ventric-lar (LV) systolic function, large artery stiffness, forwardulse wave genesis, and pulse wave reflection. Arterialtiffness is an important determinant of the net cardiovas-ular performance and cardiac energetics at rest and duringxercise.5 As such, increased PP, a manifestation of altered

ventricular–vascular coupling and increased pulsatile af-terload, might be related to the ABPR in patients withHC, but this has yet to be examined. The purpose of thepresent investigation was to test the hypothesis that PP atrest would be associated with an ABPR during exercisein patients with HC.

Methods

A total of 70 patients with HC were recruited from theHC Center at Tufts Medical Center. The diagnosis of HCwas determined using the criteria put forth by the AmericanCollege of Cardiology/European Society of Cardiology

Hypertrophic Cardiomyopathy Center, Division of Cardiology, andMolecular Cardiology Research Institute, Tufts Medical Center, Boston,Massachusetts. Manuscript received August 13, 2010; manuscript receivedand accepted October 7, 2010.

E-mail address: [email protected] (K.S. Heffernan).

clinical expert consensus document on HC. All patients hadLV hypertrophy (wall thickness �15 mm according toechocardiographic demonstration) associated with a nondi-lated cavity in the absence of another cardiac or systemicdisease that could produce this magnitude of hypertrophy.The exclusion criteria included severe valvular disease, re-cent myocardial infarction or unstable cardiac symptoms,peripheral vascular disease, heart failure, end-stage diseasewith systolic dysfunction or LV ejection fraction �40%,severe arrhythmia, chronic obstructive pulmonary disease,recent exertional syncope, previous septal myectomy, alco-hol septal ablation, and coexistent aortic stenosis. Coronaryartery disease was defined as the presence of ischemia orinfarction on single-photon emission computed tomo-graphic nuclear myocardial perfusion imaging or �50%stenosis of an epicardial coronary artery by angiography.The presence or absence of hypertension (systolic BP/diastolic BP �140/90 mm Hg or taking antihypertensivemedication) and clinical symptoms were obtained foreach patient from a questionnaire or the medical records.All subjects gave written informed consent, and the in-stitutional review board at Tufts Medical Center ap-proved the present study.

The patients underwent standard 2-dimensional transtho-racic echocardiography for assessment of the cardiac dimen-sions, followed by a symptom-limited exercise test using astandard Bruce protocol with a concurrent hemodynamic as-sessment. BP was measured using auscultation and sphygmo-manometry. The measures at rest were made with the patientswith HC in the supine position after a brief quiet rest period.BP was measured thereafter at the end of the exercise stage. AnABPR was defined as a reduction in systolic BP during exer-cise relative to systolic BP at rest or an inability to increase
systolic BP �20 mm Hg during exercise. PP was calculated as
www.ajconline.org


w�r

aci(

ciHn0

pofL0lNv

PpttAtAjtntditA

601Cardiomyopathy/Blood Pressure and Exercise in HC

systolic BP minus diastolic BP. The patients were instructed towithhold all cardiovascular medications for 24 to 72 hoursbefore exercise testing.

The presence of LV outflow tract obstruction was as-sessed as previously described at rest, with Valsalva ma-neuver, and during exercise.6 LV outflow tract obstruction

as defined as a peak instantaneous outflow gradient of30 mm Hg using continuous-wave Doppler echocardiog-

aphy.6 Systolic anterior motion and mitral regurgitationwere assessed semiquantitatively (scale 0 to 4), as previ-ously described.6

All data are reported as the mean � SEM. A priori

Table 1Patient characteristics according to blood pressure (BP) response

Variable All ABPR

Yes No

Age (years) 45 � 2 54 � 5 41 � 2*Women 23 (33%) 7 (37%) 16 (31%)Maximum left ventricular

thickness (mm)19.6 � 0.6 20.0 � 1.1 19.5 � 0.6

Left ventricular end-diastolic diameter (mm)

43.2 � 0.8 41.3 � 1.3 43.9 � 1.0

Left ventricular end-systolic diameter (mm)

25.0 � 0.7 25.2 � 1.4 25.0 � 0.9

Left atrial size (mm) 39.8 � 0.9 40.5 � 1.5 39.6 � 1.0Systolic anterior motion

(scale 0–4)1 1 1

Mitral regurgitation (scale0–4)

1 1 1

Left ventricular outflowtract obstruction

33 (47%) 10 (52%) 23 (45%)

New York HeartAssociation class

I 41 (59%) 10 (52%) 31 (61%)II 16 (23%) 3 (16%) 13 (25%)III 13 (19%) 6 (32%) 7 (14%)

Medications� Blocker 41 (59%) 14 (74%) 27 (53%)Calcium channel blocker 26 (37%) 7 (37%) 19 (37%)Diuretic 7 (10%) 5 (26%) 2 (4%)*Angiotensin-converting

enzyme inhibitor12 (17%) 3 (16%) 9 (18%)

Antiarrhythmic 7 (10%) 2 (11%) 5 (10%)Family history

hypertrophiccardiomyopathy

26 (37%) 9 (47%) 17 (33%)

History of chest pain 22 (31%) 6 (32%) 16 (31%)History of syncope 12 (17%) 4 (21%) 8 (16%)Coronary artery disease 17 (24%) 6 (32%) 11 (22%)Heart rate at rest (beats/

min)74 � 2 72 � 4 75 � 2

Systolic blood pressure atrest (mm Hg)

126 � 2 131 � 5 125 � 2

Diastolic blood pressureat rest (mm Hg)

70 � 2 67 � 2 71 � 2

Mean arterial pressure atrest (mm Hg)

88 � 2 89.0 � 3 88 � 2

Pulse pressure at rest (mmHg)

55 � 2 64 � 5 52 � 2*

Data are presented as mean � SEM or n (%).* Significant group difference (p �0.05).

significance was set at p �0.05. The normality of distri-

bution was assessed using the Kolmogorov-Smirnof andShapiro-Wilk tests. Chi-square tests were used to com-pare the categorical variables. Patients with and withoutan ABPR were compared using analysis of variance fornormally distributed variables and the Mann-Whitney Utest for non-normally distributed variables. If the demo-graphic variables differed between the 2 groups, analysisof covariance was used to adjust for the group differ-ences. The patients were then separated into tertiles ac-cording to PP, and binary logistic regression analysis wasused to examine the predictors of ABPR (entered as adiscrete variable).

Results

The baseline demographics are listed in Table 1. Ofthe 70 patients, 19 had an ABPR (�27%). Significantgroup differences were found in age (p � 0.004) anddiuretic use (p � 0.014) between those with and withoutn ABPR to exercise (Table 1). PP at rest was signifi-antly greater in the patients with HC and an ABPR thann those patients with a normal BP response to exercisep � 0.007; Table 1). The differences in PP remained

after adjusting for age and diuretic use with analysis ofcovariance (adjusted mean 63 mm Hg vs 52 mm Hg, p �0.028). The differences in PP also remained after adjust-ing for medication use (adjusted for � blockers, calciumhannel blockers, angiotensin-converting enzyme inhib-tors, Norpace, and other antiarrhythmic agents 64 mmg vs 52 mm Hg, p � 0.008). Mean arterial pressure wasot significantly different between the 2 groups (p �.738; Table 1).

The prevalence of ABPR was not different in theatients with HC with versus without LV outflow tractbstruction (30% vs 24%, p � 0.601). PP was not dif-erent in the patients with HC who did and did not haveV outflow tract obstruction (56 vs 54 mm Hg, p �.612). No gender differences were found in the preva-ence of ABPR (men, 26% vs women, 30%, p � 0.776).o gender differences were found in PP (men, 56 � 3 mm Hgs women, 51 � 3, p � 0.264).

When separating the patients into tertiles according toP, the prevalence of ABPR was significantly greater for theatients with HC with the greatest PP compared to those inhe first (reference group) and second tertile (p � 0.008). Ofhe patients in the greatest tertile (�60 mm Hg), 50% had anBPR compared to 13% and 17% in the second (range 45

o 60 mm Hg) and third (�45 mm Hg) tertiles, respectively.ccording to binary logistic regression analysis, after ad-

usting for potential confounders (age, gender, LV wallhickness, left atrial size, anterior basal septal wall thick-ess, history of chest pain, history of syncope, family his-ory of HC, LV outflow tract obstruction, coronary arteryisease), PP at rest was a significant predictor of the ABPRn patients with HC (p � 0.016). The patients with HC andhe greatest PP at rest were 4.8 times more likely to have anBPR than those with the lowest PP (� � 1.6, Wald � 5.2,

95% confidence interval 1.24 to 18.2, p � 0.023). SystolicBP, diastolic BP, and mean arterial pressure were not sig-
nificant predictors of ABPR.

ttph

qbtt

aAddme

c

cmt

wvdv

oHA

hT

tf

isHtmts


Discussion

The novel findings of the present study were as follows.First, patients with HC and an ABPR to exercise had asignificantly greater PP at rest compared to those with anormal BP response to exercise. Second, PP at rest was asignificant predictor of ABPR during exercise in patientswith HC. Our findings suggest that PP at rest, but not thetraditionally assessed BP components of cardiovascular riskassessment (i.e., systolic BP and diastolic BP) or the steadycomponent of BP (i.e., mean arterial pressure), is a predictor ofthe hemodynamic response to exercise in patients with HC.

Exercise testing is an integral part of the algorithm in riskstratification and the delivery of prophylactic therapy forHC.7 An ABPR to exercise is predictive of sudden cardiacdeath in HC, and sudden cardiac death is the most commoncause of death for patients with HC.8 In the present study,he patients with HC with a high PP at rest were almost 5imes more likely to have an ABPR to exercise. In otheropulations, an elevated PP has been associated with LVypertrophy,9 impaired ventricular relaxation,10 ischemia

during exercise,11 heart failure,12 increased left atrial size,13

and atrial fibrillation,14 all clinically relevant facets of HCpathologic features. Therefore, patients with HC and anelevated PP at rest might be a particularly high-risk cohort.Given the correlation between ABPR and the risk of suddencardiac death in patients with HC, an elevated PP could helpidentify high-risk patients within the HC cohort.

The underlying pathophysiology that causes an ABPR inpatients with HC appears to be multifaceted. The proposedmechanisms include LV systolic dysfunction15 with subse-uent subendocardial myocardial ischemia,16,17 alteredaroreflex-mediated modulation of autonomic outflow tohe heart and vasculature (possibly related to the aforemen-ioned subendocardial myocardial ischemia),18 and subse-

quent altered vascular response to exercise (i.e., excessivedecrease in peripheral/systemic vascular resistance).19 It haslso been suggested that a lower exercise capacity and anBPR in patients with HC might be related to diastolicysfunction20 and a blunted augmentation of stroke volumeuring exercise.21 Although seemingly paradoxical, theseechanisms might have a single underlying and unifying

tiology related to arterial stiffness.Patients with HC will have greater arterial stiffness than

ontrols,22 and arterial stiffness has been shown to be re-lated to reduced exercise capacity in patients with HC.23

Our findings have expanded on previous work by notingthat PP, a manifestation of the stiffening of the conduitvessels coupled with augmented pressure from wave reflec-tions, is associated with an ABPR in those with HC. LVejection of the stroke volume into a stiff aorta, coupled withan early return of reflected pressure waves, increases PP andcardiac energetic demand, reduces myocardial oxygen sup-ply/consumption, reduces subendocardial perfusion,24 im-pairs cardiac systolic and diastolic function, and bluntsstroke volume genesis.25 Thus, increased arterial stiffnessand pulsatile afterload might offer insight into the findingsof LV systolic/diastolic dysfunction, reduced stroke vol-ume/cardiac output, reduced myocardial oxygen consump-tion, and subsequent subendocardial ischemia, contributing
to an ABPR in those with HC.
Finally, a strong relation exists between arterial stiffnessand integrated neural control of BP within the baroreflexarc. Stiffening of the vessels housing the barosensory re-gions might depress mechanotransduction, resulting in areduction of baroreceptor afferent firing per given unit ofarterial pressure change, less inhibition of sympathetic out-flow (altering peripheral vascular tone), and lessened amplifi-cation of cardiac vagal tone (altering LV contractility).26 Thepatients with HC with hemodynamic instability duringlower body negative pressure, a HC cohort with a greaterprevalence of an ABPR, have a lower resting baroreflexsensitivity and exaggerated changes in baroreflex sensi-tivity during lower body negative pressure.19 Thus, in-reased arterial stiffness in HC could also alter baroreflexodulation of the cardiac and vascular autonomic con-

rol, contributing to an ABPR.Similar to previous observations, the prevalence of ABPR

as not influenced by LV outflow tract obstruction.1,2 Pre-ious studies have noted that LV outflow tract obstructionoes not affect peripheral vascular endothelial function orascular stiffness in those with HC.22,27 As such, LV out-

flow tract obstruction did not influence PP in the presentstudy. We noted no gender difference in the ABPR in thosewith HC, and this too was consistent with previous reports.1

The lack of effect of gender on the ABPR might have beenbecause men and women within the age range studied tendto have a similar brachial PP.28 Thus, LV outflow tractbstruction and gender did not modulate PP in those withC and, as such, might not be significant determinants of anBPR during exercise in those with HC.Few therapeutic options are available to attenuate the

ypotensive BP response to exercise in patients with HC.haman et al19 previously reported that neither propranolol

nor clonidine significantly modulated the hypotensive BPresponse to exercise in those with HC. However, the selec-tive serotonin reuptake inhibitor paroxetine was able tonormalize the BP response to exercise.19 Although themechanism for this interaction remains unknown, parox-etine might have a favorable effect on vascular func-tion.29,30 This raises the intriguing possibility that therapieshat improve vascular function in the presence of HC mightavorably affect the BP response to exercise.

The limitations to the present study included the lack ofnformation on clinical end points and examination of aingle surrogate marker (i.e., ABPR). Whether patients withC and an elevated PP are truly a high-risk cohort remains

o be elucidated empirically. We used PP as an indirecteasure of arterial stiffness. Additional research is needed

o corroborate the present findings using more valid bioas-ays of arterial stiffness, such as pulse wave velocity.

1. Frenneaux MP, Counihan PJ, Caforio AL, Chikamori T, McKenna WJ.Abnormal blood pressure response during exercise in hypertrophic car-diomyopathy. Circulation 1990;82:1995–2002.

2. Olivotto I, Maron BJ, Montereggi A, Mazzuoli F, Dolara A, Cecchi F.Prognostic value of systemic blood pressure response during exercisein a community-based patient population with hypertrophic cardiomy-opathy. J Am Coll Cardiol 1999;33:2044–2051.

3. Sadoul N, Prasad K, Elliott PM, Bannerjee S, Frenneaux MP,McKenna WJ. Prospective prognostic assessment of blood pressureresponse during exercise in patients with hypertrophic cardiomy-
opathy. Circulation 1997;96:2987–2991.

603Cardiomyopathy/Blood Pressure and Exercise in HC

4. Ciampi Q, Betocchi S, Lombardi R, Manganelli F, Storto G, Losi MA,Pezzella E, Finizio F, Cuocolo A, Chiariello M. Hemodynamic deter-minants of exercise-induced abnormal blood pressure response inhypertrophic cardiomyopathy. J Am Coll Cardiol 2002;40:278–284.

5. Kelly RP, Tunin R, Kass DA. Effect of reduced aortic compliance oncardiac efficiency and contractile function of in situ canine left ven-tricle. Circ Res 1992;71:490–502.

6. Maron MS, Olivotto I, Zenovich AG, Link MS, Pandian NG, KuvinJT, Nistri S, Cecchi F, Udelson JE, Maron BJ. Hypertrophic cardio-myopathy is predominantly a disease of left ventricular outflow tractobstruction. Circulation 2006;114:2232–2239.

7. Sharma S, Firoozi S, McKenna WJ. Value of exercise testing inassessing clinical state and prognosis in hypertrophic cardiomyopathy.Cardiol Rev 2001;9:70–76.

8. Maron BJ. Hypertrophic cardiomyopathy: a systematic review. JAMA2002;287:1308–1320.

9. Gardin JM, Arnold A, Gottdiener JS, Wong ND, Fried LP, Klopfen-stein HS, O’Leary DH, Tracy R, Kronmal R. Left ventricular mass inthe elderly: the Cardiovascular Health Study. Hypertension 1997;29:1095–1103.

10. Leite-Moreira AF, Correia-Pinto J, Gillebert TC. Afterload inducedchanges in myocardial relaxation: a mechanism for diastolic dysfunc-tion. Cardiovasc Res 1999;43:344–353.

11. Kingwell BA, Waddell TK, Medley TL, Cameron JD, Dart AM. Largeartery stiffness predicts ischemic threshold in patients with coronaryartery disease. J Am Coll Cardiol 2002;40:773–779.

12. Haider AW, Larson MG, Franklin SS, Levy D. Systolic blood pres-sure, diastolic blood pressure, and pulse pressure as predictors of riskfor congestive heart failure in the Framingham Heart Study. Ann InternMed 2003;138:10–16.

13. Vaziri SM, Larson MG, Lauer MS, Benjamin EJ, Levy D. Influence ofblood pressure on left atrial size: the Framingham Heart Study. Hy-pertension 1995;25:1155–1160.

14. Mitchell GF, Vasan RS, Keyes MJ, Parise H, Wang TJ, Larson MG,D’Agostino RB Sr, Kannel WB, Levy D, Benjamin EJ. Pulse pressureand risk of new-onset atrial fibrillation. JAMA 2007;297:709–715.

15. Okeie K, Shimizu M, Yoshio H, Ino H, Yamaguchi M, Matsuyama T,Yasuda T, Taki J, Mabuchi H. Left ventricular systolic dysfunctionduring exercise and dobutamine stress in patients with hypertrophiccardiomyopathy. J Am Coll Cardiol 2000;36:856–863.

16. Ciampi Q, Betocchi S, Losi MA, Ferro A, Cuocolo A, Lombardi R,Villari B, Chiariello M. Abnormal blood-pressure response to exerciseand oxygen consumption in patients with hypertrophic cardiomyopa-thy. J Nucl Cardiol 2007;14:869–875.

17. Yoshida N, Ikeda H, Wada T, Matsumoto A, Maki S, Muro A, ShibataA, Imaizumi T. Exercise-induced abnormal blood pressure responsesare related to subendocardial ischemia in hypertrophic cardiomyopa-thy. J Am Coll Cardiol 1998;32:1938–1942.

18. Kawasaki T, Azuma A, Kuribayashi T, Akakabe Y, Yamano M, Miki S,Sawada T, Kamitani T, Matsubara H, Sugihara H. Vagal enhancement

due to subendocardial ischemia as a cause of abnormal blood pressureresponse in hypertrophic cardiomyopathy. Int J Cardiol 2008;129:59–64.

19. Thaman R, Elliott PM, Shah JS, Mist B, Williams L, Murphy RT,McKenna WJ, Frenneaux MP. Reversal of inappropriate peripheralvascular responses in hypertrophic cardiomyopathy. J Am Coll Cardiol2005;46:883–892.

20. Matsumura Y, Elliott PM, Virdee MS, Sorajja P, Doi Y, McKenna WJ.Left ventricular diastolic function assessed using Doppler tissue im-aging in patients with hypertrophic cardiomyopathy: relation to symp-toms and exercise capacity. Heart 2002;87:247–251.

21. Nagata M, Shimizu M, Ino H, Yamaguchi M, Hayashi K, Taki J,Mabuchi H. Hemodynamic changes and prognosis in patients withhypertrophic cardiomyopathy and abnormal blood pressure responsesduring exercise. Clin Cardiol 2003;26:71–76.

22. Boonyasirinant T, Rajiah P, Setser RM, Lieber ML, Lever HM, DesaiMY, Flamm SD. Aortic stiffness is increased in hypertrophic cardio-myopathy with myocardial fibrosis: novel insights in vascular functionfrom magnetic resonance imaging. J Am Coll Cardiol 2009;54:255–262.

23. Austin BA, Popovic ZB, Kwon DH, Thamilarasan M, Boonyasirinant T,Flamm SD, Lever HM, Desai MY. Aortic stiffness independently predictsexercise capacity in hypertrophic cardiomyopathy: a multimodality imag-ing study. Heart 2010;96:1303–1310.

24. Saeki A, Recchia F, Kass DA. Systolic flow augmentation in heartsejecting into a model of stiff aging vasculature: influence on myocar-dial perfusion-demand balance. Circ Res 1995;76:132–141.

25. Borlaug BA, Melenovsky V, Redfield MM, Kessler K, Chang HJ,Abraham TP, Kass DA. Impact of arterial load and loading sequenceon left ventricular tissue velocities in humans. J Am Coll Cardiol2007;50:1570–1577.

26. Casadei B, Meyer TE, Coats AJ, Conway J, Sleight P. Baroreflexcontrol of stroke volume in man: an effect mediated by the vagus.J Physiol 1992;448:539–550.

27. Heffernan KS, Napolitano CA, Maron MS, Patvardhan EA, Patel AR,Pandian NG, Karas RH, Kuvin JT. Peripheral vascular endothelialfunction in patients with hypertrophic cardiomyopathy. Am J Cardiol2010 1;105:112–115.

28. Mitchell GF, Parise H, Benjamin EJ, Larson MG, Keyes MJ, Vita JA,Vasan RS, Levy D. Changes in arterial stiffness and wave reflectionwith advancing age in healthy men and women: the Framingham HeartStudy. Hypertension 2004;43:1239–1245.

29. Lara N, Archer SL, Baker GB, Le Melledo JM. Paroxetine-inducedincrease in metabolic end products of nitric oxide. J Clin Psychophar-macol 2003;23:641–645.

30. Chrapko W, Jurasz P, Radomski MW, Archer SL, Newman SC, BakerG, Lara N, Le Melledo JM. Alteration of decreased plasma NO
metabolites and platelet NO synthase activity by paroxetine in de-pressed patients. Neuropsychopharmacology 2006;31:1286–1293.

ip�soHwec(efa

DN

tdanhd

CS

Clinical Challenges of Genotype Positive (�)–PhenotypeNegative (�) Family Members in Hypertrophic Cardiomyopathy

Barry J. Maron, MDa, Laura Yeates, BScb, and Christopher Semsarian, MB, BS, PhDb,c,d,*

P

ssadptara

iwes

Hypertrophic cardiomyopathy (HC), a common geneticheart disorder associated with substantial clinical and ge-netic heterogeneity, is the most frequent cause of suddendeath in the young (including competitive athletes).1–3 HCis generally characterized by unexplained left ventricular(LV) hypertrophy,4–6 although with the aid of moleculardiagnosis, it has become evident that disease-causing mu-tations can be associated with virtually any LV wall thick-ness.7–9 Indeed, many genetically affected children (andeven some adults) in HC families do not demonstrate LVhypertrophy at some point in their clinical courses, withincomplete penetrance of the phenotype.8–12 Laboratorynvestigations over the past 2 decades have defined HC as arimary myocardial disease caused by �1,000 mutations in13 genes encoding proteins within and associated with the

arcomere.1,2,8,9,13 This has led to an increasing recognitionf a novel patient subset within the vast and ever expandingC disease spectrum: genetically affected family membersithout clinical or morphologic evidence of the dis-

ase.10,14–16 Such patients are usually referred to as “pre-linical” or “genotype-positive (�)–phenotype-negative�)” (G� P�), and they present the paradox of a rapidlyvolving new patient subgroup that requires a long period ofollow-up to develop clear guidelines with regard to man-gement.

efinition of Genotype-Positive (�)–Phenotype-egative (�)

G� P� patients carry mutations in genes encoding pro-eins of the cardiac sarcomere, judged (or known) to beisease causing for HC.5,10,14,15 Such patients are usuallysymptomatic, often with 12-lead electrocardiographic ab-ormalities but no evidence of the HC phenotype (i.e., LVypertrophy) on 2-dimensional echocardiography and car-iovascular magnetic resonance (CMR) imaging.5,10,14,15

CMR is emerging as a highly relevant imaging modalityfor the identification of the HC phenotype, because of itstomographic high–spatial resolution characteristics. CMR isnot encumbered by certain well-recognized limitations of

aHypertrophic Cardiomyopathy Center, Minneapolis Heart InstituteFoundation, Minneapolis, Minnesota; bAgnes Ginges Centre for Molecular

ardiology, Centenary Institute, Newtown, Australia; cSydney Medicalchool, University of Sydney, Sydney, Australia; and dDepartment of

Cardiology, Royal Prince Alfred Hospital, Camperdown, Australia. Manu-script received August 23, 2010; revised manuscript received and acceptedOctober 1, 2010.

Dr. Semsarian is the recipient of a Practitioner Fellowship from theNational Health and Medical Research Council, Canberra, Australia. Thisstudy was also supported in part by the Hearst Foundations, San Francisco,California (Dr. Maron).

flE-mail address: [email protected] (C. Semsarian).

echocardiography with respect to measurements of LV wallthickness, justifying its inclusion in the assessment of G�� patients.4,17 For example, CMR can provide LV wall

thickness measurements with greater precision, particularlyrelevant to hypertrophy in the borderline zone of 12 to 15mm. Also, in selected patients, CMR may identify segmen-tal regions of LV hypertrophy in the anterolateral LV freewall (or apex), not reliably detected or often underestimatedin magnitude by echocardiography.4,16,17

It has not been our practice to define the HC phenotypesolely by abnormalities on 12-lead electrocardiogra-phy.4,6,8–10 This consideration is due to certain predictablelimitations of electrocardiography as a screening test for theclinical HC spectrum18,19: (1) difficulty in establishing ab-olute and strict partitions for normality at all ages and bodyizes (particularly in children), (2) potential confusion cre-ted by nonspecific alterations unrelated to cardiovascularisease, (3) unpredictable variability in electrocardiographicatterns over time, (4) the documented weak relation be-ween electrocardiographic voltages and LV wall thickness,nd (5) the occurrence of normal results on electrocardiog-aphy in up to 25% of phenotypically expressed HC and inbout 50% of G� P� family members.12,16–18

In contrast, 2-dimensional echocardiography and CMRprovide reproducible, quantitative measures of LV wallthickness for comparison to established normal val-ues.4,6,9,17 On the basis of these considerations, it appearsmost appropriate to define the HC phenotype with respect toLV hypertrophy (as identified directly by contemporaryimaging), while an indirect measure with 12-lead electro-cardiography would likely be associated with a large num-ber of false-positive test results.

Family Studies

In Figure 1, we present 4 families demonstrating signif-cant challenges that arise in G� P� family members,hich are cornerstones of HC clinical decision making:

ligibility versus disqualification from intense competitiveports20–23 and the prevention of sudden death with prophy-

lactic implantable cardioverter-defibrillators (ICDs).24–27

Family A highlights the issue of sports eligibility in G�P� children. Two young female children aged 10 and 12years (III:3 and III:4) are both elite gymnasts. They showedno evidence of HC on electrocardiography and echocardi-ography, but both carry the causative gene mutation,Arg810His in the MYBPC3 gene, also identified in otherclinically affected individuals in the family (II:1 and II:3).Should these children (III:3 and III:4) be removed fromcompetitive sports?

Family B targets the issue of whether decisions regardingsports eligibility in G� P� patients should be further in-

uenced by a family history of a sudden death event. This

www.ajconline.org


Lh� ransplan

605Editorial/Genotype-Positive–Phenotype-Negative HC Patients

family presented for evaluation after a resuscitated cardiacarrest occurred in an 11-year-old female patient (III:2). Withsubsequent clinical screening and genetic testing, the caus-ative gene mutation, Arg495Gln in the MYBPC3 gene, wasidentified in each clinically affected family member, as wellas the 9-year-old sibling (III:3), with no evidence of thedisease phenotype. Should this 9-year-old G� P� brotherbe excluded from future involvement in competitive sports,and furthermore, should an ICD for primary preventioneven be considered in a patient of this age?

Family C explores and extends the potential role forICDs in young G� P� patients with family histories ofsudden death. In the first generation, a 30-year-old man (I:1)died suddenly while jogging, but without a confirmed HCdiagnosis. All 3 adult children are clinically affected, and 2have elected prophylactic ICDs (II:1 and II:4). Geneticscreening identified the Gly733Glu mutation in the MYH7gene in each of the 3 siblings, as well as the 14-year-oldgrandson of the proband, with no evidence of LV hypertro-phy but extensive involvement in competitive athletics (III:2). Should this 14-year-old be withdrawn from sports and/orconsidered for an ICD on the basis of the sudden death

Figure 1. Pedigrees of 4 HC families. In family D, IV:1 had an appropriate Igray symbols with arrows � G� P� subjects; squares � men; circles � w

V hypertrophy); clear symbols � without cardiac evaluation; symbols traneart failure; N � normal on clinical screening with electrocardiography a

resuscitated cardiac arrest; SCD � sudden cardiac death; Tx � heart t

event occurring decades earlier in his grandfather?

Family D illustrates the dilemma for a G� P� relative(III:2), who carries the Lys97Asn mutation in the TNNT2gene. At 38 years of age, she is the sole survivor affected byHC in a family with malignant outcomes, including 4 sud-den death events (III:3, III:4, III:5, and IV:1) and 2 deathsdue to end-stage progression (II:1 and II:3). Should thisadult relative elect a prophylactic ICD, despite absence ofthe HC phenotype?

Commentary

These HC families underscore emerging dilemmas inclinical HC practice, that is, whether the management ofsudden death risk in this new subset of genetically affectedrelatives without clinical evidence of disease (i.e., G� P�)should be similar to more typical patients with HC with LVhypertrophy. As seen in families B and C, the issue ofdisqualification from competitive sports participation andthe advisability of prophylactic ICDs for G� P� membersof HC families are often interwoven, making many of theseclinical decisions particularly complex and challenging.

Sports Eligibility: Family A raises the question of

ck at 16 years of age. Numbers in parentheses indicate age in years. Shadedsolid black symbols � clinically affected subjects with HC phenotype (i.e.,by lines � HC-related cardiac arrest and death. ES � end-stage HC-relatedging studies; �/� � heterozygote for mutation (genotype positive); RCAtation.

CD shoomen;sectednd ima

whether all G� P� patients with HC should be disqualified

t

oic

saHpds

epctTec

Ct

kmpiprtntptlrav

tcfd


from intense competitive sports, an obvious considerationgiven that HC is the most common cause of sudden death inyoung athletes.3,20 Also, intense competitive sports increasehe likelihood of these catastrophic events,20–23 and with-

drawal from this vigorous lifestyle may well reduce thisrisk.22,23,27

Nevertheless, this issue remains unsettled. Notably, the 2available consensus expert panel documents present diamet-rically opposed recommendations, with both unavoidablybased on clinical inferences with little or no hard evi-dence.21–23 For example, the European Society of Cardiol-gy guidelines23 are particularly conservative, recommend-ng disqualification from competitive sports for all genearriers.21–23 In contrast, the United States–based Bethesda

Conference 3622 permits G� P� patients to participate inall competitive sports until LV hypertrophy appears.21–23

These divergent recommendations regarding the same clin-ical scenario cannot be resolved without longitudinal fol-low-up studies in this select subset.

Families B and C raise the consideration of whethersports disqualification for G� P� athletes is even morerelevant when there is a family history of HC-related suddendeath in a close relative. The question of sports eligibility ordisqualification in such subjects is increasing in frequencyas more of these families pursue genetic testing in responseto catastrophic events occurring in relatives. However, thisparticular clinical issue is not specifically addressed in theaforementioned recommendations of either the UnitedStates22 or European23 consensus panel.

A family history of sudden death in a close relative is anacknowledged risk factor in HC,8,9,24–26 although availabledata relate only to phenotype-positive patients with LVhypertrophy and clinically defined disease. While it is un-resolved as to whether this risk marker can (or should) beextrapolated to genetically affected subjects without LVhypertrophy, it nevertheless seems most prudent to discour-age young G� P� relatives with family histories of HCsudden death from engaging in intense competitive sports atan early age.

ICDs For Primary Prevention: The question of whetherG� P� patients should be considered for primary preventionICDs because of a family history of sudden death arises mostfrequently in adults who are part of malignant families withmultiple sudden deaths25,26 (such as family D). As demon-trated by families B and C, this treatment consideration maylso arise with respect to G� P� children and adolescents.owever, the considerable frequency of device-related com-lications in young patients over long follow-up periods ofecades is often a mitigating factor for prophylactic implants inuch G� P� relatives.24,25,27

Nevertheless, ICDs have proved effective in terminatinglife-threatening ventricular tachyarrhythmias in high-riskpatients with HC with overt disease expression.24–27 How-ver, it is largely unresolved as to whether non-hypertro-hied LV muscle in patients with HC-causing mutations canonstitute an electrically unstable substrate capable of po-entially lethal sustained ventricular tachyarrhythmias.herefore, prophylactic ICDs are most likely to be consid-red after LV hypertrophy has appeared, thereby justifying
lose clinical surveillance with echocardiography (and
MR imaging, if available), probably at 12-month intervals,o identify changes in LV wall thickness.

The overwhelming difficulty surrounding this (and other)ey questions related to management of G� P� HC familyembers is the paucity of available outcome data. The

revailing perception has been that sudden death risk in HCs virtually always linked to the presence of LV hypertro-hy.8,9,24,25 Notably, however, 2 cases have been reportedecently in 37-year-old and 43-year-old (nonathlete) pa-ients with MHY7 mutations (but without clinical or phe-otypic evidence of HC), who survived ventricular fibrilla-ion.28 In addition, a few family members reported in there-genotyping era may represent similar sudden deaths inhe absence of LV hypertrophy.11 Although rare, such iso-ated cases suggest the possibility that susceptible G� P�elatives can harbor arrhythmogenic substrates at a cellularnd molecular level capable of triggering life-threateningentricular tachyarrhythmias.

This observation is consistent with other clinical findingshat support the notion that the nonhypertrophied LV myo-ardium in some G� P� relatives may be electrically orunctionally abnormal, that is, with evidence of diastolicysfunction,10,15 or abnormal 12-lead electrocardiographic

patterns,29 as well as risk markers such as nonsustainedventricular tachycardia on ambulatory (Holter) electrocar-diography, abnormal blood pressure response to exercise,30

delayed gadolinium enhancement.31

In contrast, we found no evidence of important arrhyth-mias on ambulatory (Holter) electrocardiographic monitor-ing in our G� P� patients, including in the 4 familiesreported here. Whether or not the occurrence of arrhythmicevents would be enhanced by intense physical activity (suchas competitive sports) is unresolved. Supportive evidencethat lethal events are probably exceedingly rare in G� P�patients can be derived from those HC patients with onlymild phenotypic expression who have generally favorableprognosis and low sudden death risk.8,9 However, it iscertainly possible that the true prevalence of lethal HC-related events in G� P� patients has been underestimated,given that at autopsy, these patients would have structurallynormal hearts and probably not be assigned postmortemcardiac diagnoses.

Conclusions and Clinical Decision Making

Genetic diagnosis in HC has provided many answersbut has also raised a number of important questions,including the emergence of G� P� family members,adding to the complexity of management in this highlyheterogenous disease. Specifically, given the paucity ofclinical information, the relatively small numbers of G�P� family members identified, and the expected low HCevent rate, it is likely that many years of follow-up willbe necessary to acquire and formulate evidence-basedinsights for this HC clinical subset. For example, it isunknown what proportion of G� P� relatives will de-velop LV hypertrophy (or when), nor whether some genecarriers may achieve normal longevity without ever ex-pressing the disease phenotype. Although morphologicconversion of the HC phenotype is most common during
adolescence, it has also been documented occasionally in

au(f(

pptpmrobasivi

Paevdto

tssfpttat

AEs

607Editorial/Genotype-Positive–Phenotype-Negative HC Patients

midlife and beyond.12,32 The explanation for geneticallyffected patients remaining phenotype negative remainsnclear but could be related to a number of factors, whichin addition to age) likely include secondary geneticactors that may influence expression of the mutant genee.g., modifier genes and epigenetic changes).1,2,12,32

Most important in this regard are the management con-siderations raised with respect to the G� P� subgroup. Isdisqualification from intense competitive sports justified?When (or ever) should prophylactic ICDs be considered?These decisions cannot be deferred indefinitely, becausethey are of immediate concern to patients and thereforemust be resolved with the currently available information.Unavoidably, this process will often reflect cultural differ-ences and societal priorities, and reliance on the experiencesand perceptions of the individual physicians and patientsinvolved.

At present, however, it is not possible to reach explicitand definitive judgments addressing these clinical questionswith absolute authority in each G� P� patient. It is ourractice that until clarity is achieved in this area, the mostrudent strategy is to engage young G� P� patients andheir families in a fundamental process predicated on therinciples of full transparency, informed consent, and ulti-ately patient autonomy. Patients are provided with all

elevant and pertinent information (or lack thereof). Specificptions regarding competitive sports or prophylactic defi-rillator implantation are discussed in a detailed and bal-nced fashion, taking into account patient autonomy con-iderations i.e., the comfort level and desires of the fullynformed patient and family in making decisions that in-olve the substantial ambiguity implicit when there arensufficient evidence-based data.

We followed this approach in managing each of the G�� family members shown in Figure 1, and this approach islso consistent with the process used with phenotypicallyxpressed HC patients under consideration for primary pre-ention ICDs in the ambiguous “gray zone” in which sud-en death risk level cannot be assessed with precision usinghe conventional markers, and individual clinical judgmentsn a case-by-case basis are necessary.

In our 4 families, the G� P– subjects in family A choseo reduce their participation in high-level competitiveports, while those in families B to D have continued theirports participation. In addition, the G� P� individual inamily D elected to have an ICD implanted for primaryrevention. Finally, it is advisable to provide G� P� pa-ients with imaging surveillance at regular intervals to de-ermine when or if the HC phenotype develops and withmbulatory electrocardiographic monitoring for the poten-ial detection of ventricular tachyarrhythmias.

cknowledgment: We acknowledge the contributions ofdwin Kirk and Anne Ronan from the state regional geneticervices in New South Wales, Australia.

1. Alcalai R, Seidman JG, Seidman CE. Genetic basis of hypertrophiccardiomyopathy: from bench to the clinics. J Cardiovasc Electro-
physiol 2008;19:104–110.
2. Seidman JG, Seidman C. The genetic basis for cardiomyopathy: frommutation identification to mechanistic paradigms. Cell 2001;104:557–567.

3. Maron BJ, Doerer JJ, Haas TS, Tierney DM, Mueller FO. Suddendeaths in young competitive athletes: analysis of 1866 deaths in theUnited States, 1980-2006. Circulation 2009;119:1085–1092.

4. Maron MS, Maron BJ, Harrigan C, Buros J, Gibson CM, Olivotto I,Biller L, Lesser JR, Udelson JE, Manning WJ, Appelbaum E. Hyper-trophic cardiomyopathy phenotype revisited after 50 years with car-diovascular magnetic resonance. J Am Coll Cardiol 2009;54:220–228.

5. Maron BJ, Seidman CE, Ackerman MJ, Towbin JA, Maron MS,Ommen SR, Nishimura RA, Gersh BJ. What’s in a name? Dilemmasin nomenclature characterizing hypertrophic cardiomyopathy and leftventricular hypertrophy. Circ Cardiovasc Genet 2009;2:81–86.

6. Klues HG, Schiffers A, Maron BJ. Phenotypic spectrum and patternsof left ventricular hypertrophy in hypertrophic cardiomyopathy: mor-phologic observations and significance as assessed by two-dimen-sional echocardiography in 600 patients. J Am Coll Cardiol 1995;26:1699–1708.

7. Ingles J, Doolan A, Chiu C, Seidman J, Seidman C, Semsarian C.Compound and double mutations in patients with hypertrophic cardio-myopathy: implications for genetic testing and counselling. J MedGenet 2005;42:e59.

8. Maron BJ. Hypertrophic cardiomyopathy: a systematic review. JAMA2002;287:1308–1320.

9. Maron BJ, McKenna WJ, Danielson GK, Kappenberger LJ, Kuhn HJ,Seidman CE, Shah PM, Spencer WH III, Spirito P, Ten Cate FJ, WigleED. American College of Cardiology/European Society of Cardiologyclinical expert consensus document on hypertrophic cardiomyopathy.A report of the American College of Cardiology Foundation TaskForce on Clinical Expert Consensus Documents and the EuropeanSociety of Cardiology Committee for Practice Guidelines. Eur Heart J2003;24:1965–1991.

10. Maron BJ, Ho CY. Hypertrophic cardiomyopathy without hypertro-phy: an emerging pre-clinical subgroup composed of genetically af-fected family members. JACC Cardiovasc Imaging 2009;2:65–68.

11. McKenna WJ, Stewart JT, Nihoyannopoulos P, McGinty F, DaviesMJ. Hypertrophic cardiomyopathy without hypertrophy: two familieswith myocardial disarray in the absence of increased myocardial mass.Br Heart J 1990;63:287–290.

12. Maron BJ, Niimura H, Casey SA, Soper MK, Wright GB, Seidman JG,Seidman CE. Development of left ventricular hypertrophy in adultswith hypertrophic cardiomyopathy caused by cardiac myosin-bindingprotein C mutations. J Am Coll Cardiol 2001;38:315–321.

13. Lind JM, Chiu C, Semsarian C. Genetic basis of hypertrophic cardio-myopathy. Expert Rev Cardiovasc Ther 2006;4:927–934.

14. Kelly M, Semsarian C. Multiple mutations in genetic cardiovasculardisease: a marker of disease severity? Circ Cardiovasc Genet 2009;2:182–190.

15. Ho CY, Sweitzer NK, McDonough B, Maron BJ, Casey SA, SeidmanJG, Seidman CE, Solomon SD. Assessment of diastolic function withDoppler tissue imaging to predict genotype in preclinical hypertrophiccardiomyopathy. Circulation 2002;105:2992–2997.

16. Germans T, Wilde AAM, Dijkmans PA, Chai W, Kamp O, Pinto YM,van Rossum AC. Structural abnormalities of the inferoseptal left ven-tricular wall detected by cardiac magnetic resonance imaging in car-riers of hypertrophic cardiomyopathy mutations. J Am Coll Cardiol2006;48:2518–2523.

17. Maron MS, Lesser RJ, Maron BJ. Massive left ventricular hypertrophyin hypertrophic cardiomyopathy significantly underestimated by echo-cardiography but identified by cardiovascular magnetic resonance:implications for management strategies. Am J Cardiol. In press.

18. Montgomery JV, Harris KM, Casey SA, Zenovich AG, Maron BJ.Relation of electrocardiographic patterns to phenotypic expression andclinical outcome in hypertrophic cardiomyopathy. Am J Cardiol 2005;96:270–275.

19. McLeod CJ, Ackerman MJ, Nishimura RA, Tajik AJ, Gersh BJ,Ommen SR. Outcome of patients with hypertrophic cardiomyopathyand a normal electrocardiogram. J Am Coll Cardiol 2009;54:229–233.

20. Maron BJ. Sudden death in young athletes. N Engl J Med 2003;349:1064–1075.

21. Pelliccia A, Zipes DP, Maron BJ. Bethesda Conference #36 and theEuropean Society of Cardiology Consensus Recommendations revis-
ited a comparison of U.S. and European criteria for eligibility and

2

3

3

3


disqualification of competitive athletes with cardiovascular abnormalities.J Am Coll Cardiol 2008;52:1990–1996.

22. Maron BJ, Zipes DP. 36th Bethesda Conference: eligibility recom-mendations for competitive athletes with cardiovascular abnormalities.J Am Coll Cardiol 2005;45:1312–1375.

23. Pelliccia A, Fagard R, Bjornstad HH, Anastassakis A, Arbustini E,Assanelli D, Biffi A, Borjesson M, Carrè F, Corrado D, Delise P,Dorwarth U, Hirth A, Heidbuchel H, Hoffmann E, Mellwig KP,Panhuyzen-Goedkoop N, Pisani A, Solberg EE, van-Buuren F, Van-hees L, Blomstrom-Lundqvist C, Deligiannis A, Dugmore D, GliksonM, Hoff PI, Hoffmann A, Hoffmann E, Horstkotte D, Nordrehaug JE,Oudhof J, McKenna WJ, Penco M, Priori S, Reybrouck T, Senden J,Spataro A, Thiene G. Recommendations for competitive sports par-ticipation in athletes with cardiovascular disease: a consensus docu-ment from the Study Group of Sports Cardiology of the WorkingGroup of Cardiac Rehabilitation and Exercise Physiology and theWorking Group of Myocardial and Pericardial Diseases of the Euro-pean Society of Cardiology. Eur Heart J 2005;26:1422–1445.

24. Maron BJ, Spirito P, Shen W-K, Haas TS, Formisano F, Link MS,Epstein AE, Almquist AK, Daubert JP, Lawrenz T, Boriani G, EstesNA III, Favale S, Piccininno M, Winters SL, Santini M, Betocchi S,Arribas F, Sherrid MV, Buja G, Semsarian C, Bruzzi P. Implantablecardioverter-defibrillators and prevention of sudden cardiac death inhypertrophic cardiomyopathy. JAMA 2007;298:405–412.

25. Maron BJ. Contemporary insights and strategies for risk stratification
and prevention of sudden death in hypertrophic cardiomyopathy. Cir-culation 2010;121:445–456.
26. Bos JM, Maron BJ, Ackerman MJ, Haas TS, Sorajja P. Nishimura RA,Gersh BJ, Ommen SR. Role of family history of sudden death in riskstratification and prevention of sudden death with implantable defi-brillators in hypertrophic cardiomyopathy. Am J Cardiol 2010;106:1481–1486.

27. Maron BJ, Spirito P. Implantable defibrillators and prevention ofsudden death in hypertrophic cardiomyopathy. J Cardiovasc Electro-physiol 2008;19:1118–1126.

28. Christiaans I, Lekanne dit Deprez RH, van Langen IM, Wilde AA.Ventricular fibrillation in MYH7-related hypertrophic cardiomyopathybefore onset of ventricular hypertrophy. Heart Rhythm 2009;6:1366–1369.

9. Panza JA, Maron BJ. Relation of electrocardiographic abnormalities toevolving left ventricular hypertrophy in hypertrophic cardiomyopathyduring childhood. Am J Cardiol 1989;63:1258–1265.

0. Michels M, Soliman OI, Phefferkorn J, Hoedemaekers YM, KofflardMJ, Dooijes D, Majoor-Krakauer D, Ten Cate FJ. Disease penetranceand risk stratification for sudden cardiac death in asymptomatic hy-pertrophic cardiomyopathy mutation carriers. Eur Heart J 2009;30:2593–2598.

1. Strijack B, Ariyarajah V, Soni R, Jassal DS, Greenberg CR, McGregorR, Morris M. Late gadolinium enhancement cardiovascular magneticresonance in genotyped hypertrophic cardiomyopathy with normalphenotype. J Cardiovasc Magn Reson 2008;10:58.

2. Maron BJ, Seidman JG, Seidman CE. Proposal for contemporary
screening strategies in families with hypertrophic cardiomyopathy.J Am Coll Cardiol 2004;44:2125–2132.

aAM2

wR

0d

Usefulness of Repeated N-Terminal Pro-B-Type Natriuretic PeptideMeasurements as Incremental Predictor for Long-Term

Cardiovascular Outcome After Vascular Surgery

Dustin Goei, MDa, Jan-Peter van Kuijk, MDa, Willem-Jan Flu, MDa, Sanne E. Hoeks, PhDb,Michel Chonchol, MDc, Hence J.M. Verhagen, MDa, Jeroen J. Bax, MDd, and

Don Poldermans, MDa,*

Plasma N-terminal pro–B-type natriuretic peptide (NT–pro-BNP) levels improve preop-erative cardiac risk stratification in vascular surgery patients. However, single preoperativemeasurements of NT–pro-BNP cannot take into account the hemodynamic stress causedby anesthesia and surgery. Therefore, the aim of the present study was to assess theincremental predictive value of changes in NT–pro-BNP during the perioperative periodfor long-term cardiac mortality. Detailed cardiac histories, rest left ventricular echocardi-ography, and NT–pro-BNP levels were obtained in 144 patients before vascular surgeryand before discharge. The study end point was the occurrence of cardiovascular deathduring a median follow-up period of 13 months (interquartile range 5 to 20). Preopera-tively, the median NT–pro-BNP level in the study population was 314 pg/ml (interquartilerange 136 to 1,351), which increased to a median level of 1,505 pg/ml (interquartile range404 to 6,453) before discharge. During the follow-up period, 29 patients (20%) died, 27(93%) from cardiovascular causes. The median difference in NT–pro-BNP in the survivorswas 665 pg/ml, compared to 5,336 pg/ml in the patients who died (p � 0.01). MultivariateCox regression analyses, adjusted for cardiac history and cardiovascular risk factors (age,angina pectoris, myocardial infarction, stroke, diabetes mellitus, renal dysfunction, bodymass index, type of surgery and the left ventricular ejection fraction), demonstrated thatthe difference in NT–pro-BNP level between pre- and postoperative measurement was thestrongest independent predictor of cardiac outcome (hazard ratio 3.06, 95% confidenceinterval 1.36 to 6.91). In conclusion, the change in NT–pro-BNP, indicated by repeatedmeasurements before surgery and before discharge is the strongest predictor of cardiacoutcomes in patients who undergo vascular surgery. © 2011 Elsevier Inc. All rights
rvscfw

M

wttit

Patients who undergo vascular surgery are at high riskfor peri- and postoperative cardiac events due to under-lying coronary artery disease.1 N-terminal pro–B-typenatriuretic peptide (NT–pro-BNP) levels improve preop-erative cardiac risk stratification for surgical patients.2– 4

NT–pro-BNP is a cardiac neurohormone that is synthe-sized in the ventricular myocardium and is released inresponse to ventricular wall stretching and myocardialischemia.5,6 However, a single preoperative measurementof NT–pro-BNP cannot reflect the hemodynamic changescaused by anesthesia and surgical stress. However, theymight in fact be the consequence hemodynamic instabil-

Departments of aVascular Surgery and bAnaesthesiology, ErasmusMedical Center, Rotterdam, The Netherlands; cDivision of Renal Diseasesnd Hypertension, University of Colorado Denver Health Sciences Center,urora, Colorado; and dDepartment of Cardiology, Leiden Universityedical Center, Leiden, The Netherlands. Manuscript received August 9,

010; revised manuscript received and accepted October 5, 2010.Dustin Goei, Jan-Peter van Kuijk, Willem-Jan Flu and Sanne E. Hoeks

ere supported by an unrestricted research grant from “Lijf en Leven”otterdam, the Netherlands.

tE-mail address: [email protected] (D. Poldermans).

ities during the perioperative period, with subsequentepisodes of prolonged subclinical myocardial ischemia,associated with adverse long-term cardiac outcomes.7

Previous studies have demonstrated that changes in NT–pro-BNP over brief periods are related to adverse out-comes in acute coronary syndromes and acute decompen-sated heart failure.8,9 However, data on the use ofepeated perioperative NT–pro-BNP measurements inascular surgery patients are lacking. In the presenttudy, we evaluated the incremental predictive value ofhanges in NT–pro-BNP during the perioperative periodor long-term cardiac mortality in patients who under-ent vascular surgery.

ethods

The study population consisted of patients who under-ent elective vascular surgery at the Erasmus Medical Cen-

er (Rotterdam, The Netherlands) from 2007 to 2010. Pa-ients were identified in a prospectively maintained databasencluding all patients who underwent vascular surgery athis institution. The medical ethics committee of the hospi-
al was informed about the study, and all procedures of this
www.ajconline.org


artd

L

DA

CA


retrospective study met the approval of the medical ethicscommittee of the Erasmus Medical Center.

Before surgery, a detailed cardiac history was obtainedfrom each patient, including angina pectoris, myocardialinfarction, percutaneous coronary intervention or coronaryartery bypass grafting, heart failure (defined as the presenceof heart failure symptoms according the New York Heart

Table 1Baseline characteristics of the source population (n � 144)

Variable Value

DemographicsAge (years) 68 � 10Men 99 (69%)Angina pectoris 32 (22%)Myocardial infarction 59 (41%)Coronary revascularization 41 (29%)Heart failure 24 (17%)Stroke 38 (26%)Smokers 52 (26%)Hypertension* 100 (69%)Diabetes mellitus 40 (28%)Hypercholesterolemia† 75 (52%)Renal dysfunction 35 (24%)Chronic obstructive pulmonary disease 26 (18%)

Site and type of surgeryAbdominal aortic 74 (51%)Lower extremity 56 (39%)Carotid 14 (10%)Open surgery‡ 106 (74%)

MeasurementsBody mass index (kg/m2) 26 � 3.8LVEF (%)

�30 10 (7%)30–40 30 (21%)40–50 59 (41%)�50 45 (31%)

Serum creatinine (�mol/L) 90 (72–113)NT–pro-BNP (pg/ml)

Preoperative 314 (136–1,351)Postoperative 1,505 (404–6,453)

Data are expressed as mean � SD, number (percentage), or median (IQR).* Blood pressure �140/90 mm Hg or medical therapy to control hyperten-

sion.† Plasma cholesterol �5.5 mmol/L or treatment with lipid-lowering drugs‡ Endovascular or open vascular procedures.

Table 2Medications at screening and discharge (n � 144)

Medication At Screening At Discharge p Value

Aspirin 98 (68%) 95 (66%) NSStatins 110 (77%) 115 (80%) NS�-blocking agents 137 (95%) 141 (98%) NS

iuretics 43 (30%) 43 (30%) NSngiotensin-convertingenzyme inhibitors

47 (33%) 48 (33%) NS

alcium antagonists 29 (20%) 27 (19%) NSngiotensin receptorblockers

17 (12%) 35 (24%) 0.002

Association classification or previous hospital admission for

decompensated heart failure), and stroke or transient isch-emic attack. Furthermore, cardiovascular risk factors wererecorded, and included age, smoking history, hypertension(blood pressure �140/90 mm Hg or medical therapy tocontrol hypertension), diabetes mellitus (fasting glucoselevel �7.0 mmol/L or medication to control diabetes), hy-percholesterolemia (plasma cholesterol level �5.5 mmol/Lor treatment with lipid-lowering drugs), and renal dysfunc-tion (defined as serum creatinine �2 mg/dl). Other datacollected included history of chronic obstructive pulmonarydisease (defined as a forced expiratory volume in 1 second�70% of age- and gender-predicted value), site of surgery(abdominal aortic, lower extremity, or carotid), and type ofprocedure (endovascular or open).

Finally, the use of the following medications was re-corded at baseline and at the time of discharge: aspirin,statins, �-blocking agents, calcium antagonists, diuretics,ngiotensin-converting enzyme inhibitors, and angiotensineceptor blockers. Treatment goals were defined accordingo current guidelines for patients with peripheral arterialisease and included low-dose aspirin (80 mg/day), statins,

�-blocking agents (titrated to a perioperative heart rate of 50to 70 beats/min) in patients with or at risk for ischemic heartdisease, and angiotensin-converting enzyme inhibitors inpatients with left ventricular (LV) ejection fractions(LVEFs) �40%.10

Peripheral venous blood samples were obtained for mea-surement of NT–pro-BNP levels in all patients during thepreoperative outpatient clinic or at hospital admission andbefore discharge. NT–pro-BNP concentration was deter-mined using an electrochemiluminescence assay on an Elec-sys (Hoffman-La Roche, Basel, Switzerland). The methodis a “sandwich”-type quantitative immunoassay, based onpolyclonal antibodies against epitopes in the N-terminal partof pro-BNP. The lower detection limit was 5 pg/ml. Intra-assay coefficients of variance at 271 and 6,436 pg/ml were1.9% and 0.9%, respectively. Assays were performed by alaboratory technician blinded to the patients’ clinical data.Importantly, pre- and postoperative NT–pro-BNP levelswere unknown for the treating physician and were not usedfor clinical management.

Preoperatively, patients underwent 2-dimensional trans-thoracic echocardiography during rest. Cardiac evaluationwas performed using a portable Acuson Cypress ultrasoundsystem (Siemens Medical Solutions USA, Inc., MountainView, California) with a 3V2C transducer (3.0, 3.5, 2.5, and2.0 MHz). The LVEF was assessed in the apical, 4-cham-ber, or 2-chamber view with the patient in left lateral decu-bitus position. Quantification of LV volumes was performedusing the modified Simpson’s rule, with inter- and intraob-server variability of 9% to 12% and 6%, respectively.11 The

VEF was calculated as (LV end-systolic volume � LVend-diastolic volume) � 100/LV end-diastolic volume. LVdysfunction was defined as impaired LV systolic functionwith a LVEF �40%. Of note, all echocardiographic studiesperformed were for research purposes and were not used forclinical management.

The end point of this study was the occurrence of car-diovascular death, defined as any death with a cardiovascu-lar cause, including those deaths after cardiac procedures,
cardiac arrest, myocardial infarction, pulmonary embolus,

AdwHqa

Ssaohdt�ttdavtt

aa

611Miscellaneous/Repeated NT–pro-BNP Measurements and Long-Term Prognosis

stroke, or sudden deaths not ascribed to other causes. Mor-tality was considered cardiovascular unless explicit proof ofa noncardiac cause could be delivered. Long-term mortalitywas assessed by approaching the municipal civil registries.

Dichotomous data are described as numbers and per-centages. The continuous variables age and body massindex are described as mean � SD. Continuous data witha significant skewed distribution were compared usingthe Mann-Whitney U test and are expressed as median(interquartile range [IQR]). Receiver-operating charac-teristic curve analysis was used to assess the optimalcut-off value of NT–pro-BNP for the prediction of long-term cardiac mortality. The optimal value of preoperativeNT–pro-BNP for predicting long-term cardiac mortalitywas defined as the concentration with the largest sum ofsensitivity plus specificity. Changes in plasma NT–pro-

Figure 1. Receiver-operating characteristic curve of NT–pro-BNP levels topredict long-term cardiovascular mortality. Sensitivity and 1 � specificityre plotted for various levels. The ideal cut-off value is indicated by therrow.

Figure 2. The changes in NT–pro-BNP levels between the pre- and post-operative periods were compared between survivors and patients who died.

BNP levels from preoperatively until the first 30 postop-

erative days were calculated as median differences withtheir interquartile range. We applied multivariate Coxregression analyses to evaluate the relation betweenplasma levels of NT–pro-BNP and the subsequentchanges in relation to the study end point. In the regres-sion analyses, NT–pro-BNP was entered as a continuousdependent variable and was log-transformed to obtainnormality. Multivariate regression analyses were ad-justed for cardiac risk factors12 and factors recognized toinfluence NT–pro-BNP levels13–15: age, angina pectoris,myocardial infarction, stroke, diabetes mellitus, renaldysfunction, body mass index, type of surgery, and theLVEF. Interactions between renal dysfunction, intraop-erative fluids administered, medication at discharge (di-uretics, angiotensin-converting enzyme inhibitors, andangiotensin receptor blockers) and the change in NT–pro-BNP levels with the study end point were evaluated byforcing these interaction terms in the multivariate regres-sion model. Interaction terms were included in the mul-tivariate regression model only when significant. Wereport crude and adjusted hazard ratios and their 95%confidence intervals. For all tests, p values �0.05 (2sided) were considered significant. All analyses wereperformed using SPSS version 15.0 (SPSS, Inc., Chicago,Illinois).

Results

The study population consisted of 144 patients withperipheral arterial disease referred for elective noncardiacvascular surgery. Most of the patients underwent abdominalaortic surgery (n � 74 [51%]). Baseline characteristics ofthe total study population are listed in Table 1. The meanage of the patients was 68 � 10 years, and 69% were men.

lmost half of the patients (41%) had histories of myocar-ial infarction. Diabetes mellitus and renal dysfunctionere present in 28% and 24% of patients, respectively.ypertension and hypercholesterolemia were the most fre-uent cardiovascular risk factors and were observed in 69%nd 52% of the patients, respectively.

At baseline, 99 patients (69%) had LVEFs �40%.ymptomatic heart failure was based on the presence ofigns and symptoms according to New York Heart Associ-tion classification and was diagnosed in 24 patients (17%)f the total study population. Of these patients, 17 (71%)ad systolic or combined heart failure, while 7 (29%) wereiagnosed with heart failure with preserved ejection frac-ions. In the remaining 75 patients (52%) with LVEFs

40%, but without clinical symptoms of heart failure, sys-olic or combined LV dysfunction was observed in 40 pa-ients (53%), and 35 patients (47%) had isolated diastolicysfunction. The high prevalence of LV dysfunction waslso reflected by the use of diuretics and angiotensin-con-erting enzyme inhibitors, which were prescribed in 1/3 ofhese patients. Furthermore, cardioprotective medications inhe population, such as aspirin, statins, and � blockers, were

prescribed in �2 of 3 patients. With the exception of an-giotensin receptor blockers (p � 0.02), no significant dif-ferences were found between medications at screening and
at discharge (Table 2).

lpdbdipccpw(pcpah(iaNUfpprtppd

D

Ba

prioc

veatsps

bevqs


The median preoperative NT–pro-BNP level in our high-risk vascular surgery population was 314 pg/ml (IQR 136 to1,351). During the first 30 postoperative days, repeatedmeasurements of NT–pro-BNP were performed, and themedian postoperative NT–pro-BNP level increased to 1,505pg/ml (IQR 404 to 6,453). The median difference for thetotal population between pre- and postoperative NT–pro-BNP levels was 969 pg/ml (IQR 139 to 4,337). In allpatients, the first postoperative NT–pro-BNP measurementwas performed before hospital discharge. The mean lengthof hospital stay was 6.8 � 3.1 days. After a median fol-ow-up period of 13 months (IQR 5 to 20), the mortality endoint was reached in 29 patients (20%), 27 (93%) of whomied secondary to cardiovascular causes. The associationetween preoperative NT–pro-BNP level and long-term car-iovascular mortality was assessed using a receiver-operat-ng characteristic curve (Figure 1). For preoperative NT–ro-BNP, the area under the curve was 0.668 (95%onfidence interval 0.619 to 0.716), and the optimum dis-riminate threshold was 350 pg/ml. The changes in NT–ro-BNP levels between the pre- and postoperative periodsere compared between survivors and patients who died

Figure 2). The median preoperative NT–pro-BNP level inatients who died during the follow-up period was signifi-antly higher compared to the survivors (795 vs 269 pg/ml,� 0.002). In addition, the median difference between pre-

nd postoperative NT–pro-BNP level was significantlyigher in patients who died compared to the survivors5,336 vs 665 pg/ml, p � 0.010). We found no significantnteraction between renal dysfunction, intraoperative fluidsdministered, medications at discharge, and the change inT–pro-BNP levels with respect to the study end point.sing multivariate Cox regression analyses with adjustment

or demographics and cardiac risk factors, preoperative NT–ro-BNP level as a log-transformed variable was an inde-endent predictor of long-term cardiac mortality (hazardatio 2.57 95% confidence interval 1.16 to 5.70). Impor-antly, the change in NT–pro-BNP level between pre- andostoperative measurements was the strongest independentredictor of cardiac outcome (hazard ratio 3.06 95% confi-ence interval 1.36 to 6.91; Table 3).

iscussion

The present study demonstrates that a change in NT–pro-NP, indicated by repeated measurements before surgery

Table 3Long-term mortality

Variable Univariate Multivariate*

HR 95% CI HR 95% CI

Preoperative NT–pro-BNP†

2.83 1.63–4.93 2.57 1.16–5.70

NT–pro-BNP change† 3.31 1.76–6.23 3.06 1.36–6.91

* Adjusted for age, angina pectoris, myocardial infarction, stroke, dia-betes mellitus, renal dysfunction, body mass index, type of surgery, and theLVEF.

† NT–pro-BNP was entered as a log-transformed variable.

nd before discharge, is an incremental and independent

redictor of an increased long-term cardiovascular mortalityisk on top of clinical risk factors. Importantly, the changen NT–pro-BNP level was the strongest predictor of cardiacutcome and yielded a threefold increased risk for the oc-urrence of long-term cardiac events.

Postoperative cardiac events in patients who undergoascular surgery are more common in patients with preop-rative myocardial ischemia, LV dysfunction, and valvebnormalities compared to patients without these condi-ions.12,16 There has been considerable evidence demon-trating that a single determination of NT–pro-BNP is aromising marker in the setting of preoperative cardiac risktratification.2,4,17 In addition to cardiac risk factors only,

NT–pro-BNP above the threshold of 350 pg/ml was anexcellent tool for further risk stratification (C-statistic �0.86) in patients who undergo elective noncardiac vascularsurgery.17 Rodseth et al18 demonstrated that NT–pro-BNPabove the optimal discriminatory threshold of 280 pg/ml,determined by receiver-operating characteristic curve anal-ysis, was associated with 30-day and intermediate-term car-diac outcomes. We observed an optimal discriminatorythreshold of 350 pg/ml for long-term outcomes, which wasin fact close to the median concentration at baseline (314pg/ml). However, it seems unlikely that there is a dichoto-mous threshold that defines a normal or abnormal NT–pro-BNP value. In 2 large meta-analyses of mixed cohorts ofnon–cardiac surgery patients, the decision threshold forNT–pro-BNP varied widely from 201 to 791 pg/ml.19,20 Onthe basis of these data, it is more likely that the perioperativecardiovascular risk increases as NT–pro-BNP concentra-tions increase. Furthermore, noncardiac factors such as re-nal dysfunction, pulmonary hypertension, chronic obstruc-tive pulmonary disease, and body mass index mightinfluence NT–pro-BNP levels.21,22

The role of postoperative NT–pro-BNP determination isless clear. Mahla et al23 hypothesized that the differentiationetween preoperative and postoperative NT–pro-BNP lev-ls is important, because restriction to a single preoperativealue does not reflect the variable hemodynamic conse-uences of anesthesia and risks associated with type ofurgery. In the study by Mahla et al,23 218 patients sched-

uled for vascular surgery were enrolled, and the optimaldiscriminate threshold for postoperative NT–pro-BNP wascalculated at 860 pg/ml. They concluded that a single post-operative NT–pro-BNP determination provides importantadditional prognostic information to preoperative levels.However, several important limitations of that study shouldbe acknowledged. Pre- and postoperative NT–pro-BNP lev-els were analyzed as 2 separate predictors of long-termoutcomes, but no attention was given to the change inNT–pro-BNP levels during the perioperative period. Fur-thermore, using multivariate regression analyses, no adjust-ments were performed for conventional cardiovascular riskfactors. Most important, however, no adjustments were per-formed for preoperative NT–pro-BNP level, which is themost evident confounder of increased postoperative NT–pro-BNP level.

The present study is the first to demonstrate that thechange in NT–pro-BNP level between pre- and postopera-tive measurement is the strongest independent predictor of
cardiac outcomes, after adjustment for conventional risk

vefpisllig

ovephpofafh

1

1

1

1

613Miscellaneous/Repeated NT–pro-BNP Measurements and Long-Term Prognosis

factors and preoperative NT–pro-BNP levels. In our popu-lation the median difference between pre- and postoperativeNT–pro-BNP levels was 969 pg/ml. In line with other stud-ies, the present findings demonstrated high NT–pro-BNPlevels after major surgery. Importantly, the reasons for theseelevations remain largely unknown but could be explainedby several pathophysiologic pathways. Natriuretic peptiderelease is an index of activation of neurohumoral axis in thesetting of LV overload and myocardial ischemia24 to reduceentricular wall stress. It could be speculated that a postop-rative increase of NT–pro-BNP reflects impaired cardiacunction with prognostic power beyond that of a singlereoperative determination because it incorporates the phys-ologic consequences of anesthesia, fluid shifts, surgicaltress, duration of the procedure, and intraoperative bloodoss. Notably, a correlation between pre- and poststressevels of NT–pro-BNP and the risk and extent of inducibleschemia has been demonstrated previously, in the nonsur-ical setting25 and in noninvasive risk assessment before

noncardiac surgery.4,26 As such, increases in plasma levelsf NT–pro-BNP may have a role in identifying high-riskascular patients who may require more extensive postop-rative cardiovascular follow-up. Furthermore, it offersrospect of several applications, including the selection ofigher risk subjects for recruitment to therapeutic trials andotentially providing a reliable surrogate index of efficacyf new treatments. In contrast, in patients with chronic heartailure, previous clinical trials have examined the value ofdding measurements of NT–pro-BNP to standard heartailure treatment with the effort to improve outcomes andave returned mixed results.27–30

Potential limitations of these data merit consideration.First, the study population consisted of patients referredto a tertiary referral center and may not fully represent ageneral population scheduled for elective vascular sur-gery. Second, unknown factors that influence NT–pro-BNP levels and their interaction with identifying risk foradverse events need to be accounted. Although currently,no consensus exists regarding the reference range ofNT–pro-BNP values, we used the change in NT–pro-BNP level between pre- and postoperative measurement.Our findings suggest the incorporation of NT–pro-BNPdeterminations in the diagnostic procedure before andafter surgery in vascular surgery patients. Changes be-tween pre- and postoperative NT–pro-BNP levels provideimportant additional prognostic information and shouldprovoke clinicians to find the cause responsible for theelevation of plasma NT–pro-BNP levels. Investigatorsundertaking clinical trials or cohort studies should beencouraged to incorporate serial neurohormonal mea-surements in their study designs. Furthermore, studies inlarger patient populations are required to clarify the op-timal predictive cut-off value for the prediction of long-term cardiac events in addition to known cardiac riskfactors.

1. Mangano DT. Perioperative cardiac morbidity. Anesthesiology 1990;72:153–184.

2. Yeh HM, Lau HP, Lin JM, Sun WZ, Wang MJ, Lai LP. Preoperative
plasma N-terminal pro-brain natriuretic peptide as a marker of cardiac
risk in patients undergoing elective non-cardiac surgery. Br J Surg2005;92:1041–1045.

3. Dernellis J, Panaretou M. Assessment of cardiac risk before non-cardiac surgery: brain natriuretic peptide in 1590 patients. Heart 2006;92:1645–1650.

4. Feringa HH, Bax JJ, Elhendy A, de Jonge R, Lindemans J, SchoutenO, van den Meiracker AH, Boersma E, Schinkel AF, Kertai MD, vanSambeek MR, Poldermans D. Association of plasma N-terminal pro-B-type natriuretic peptide with postoperative cardiac events in patientsundergoing surgery for abdominal aortic aneurysm or leg bypass. Am JCardiol 2006;98:111–115.

5. Schnabel R, Rupprecht HJ, Lackner KJ, Lubos E, Bickel C, Meyer J,Munzel T, Cambien F, Tiret L, Blankenberg S. Analysis of N-termi-nal-pro-brain natriuretic peptide and C-reactive protein for risk strat-ification in stable and unstable coronary artery disease: results from theAtheroGene study. Eur Heart J 2005;26:241–249.

6. Kragelund C, Gronning B, Kober L, Hildebrandt P, Steffensen R.N-terminal pro-B-type natriuretic peptide and long-term mortality instable coronary heart disease. N Engl J Med 2005;352:666–675.

7. Landesberg G, Shatz V, Akopnik I, Wolf YG, Mayer M, Berlatzky Y,Weissman C, Mosseri M. Association of cardiac troponin, CK-MB,and postoperative myocardial ischemia with long-term survival aftermajor vascular surgery. J Am Coll Cardiol 2003;42:1547–1554.

8. Heeschen C, Hamm CW, Mitrovic V, Lantelme NH, White HD.N-terminal pro-B-type natriuretic peptide levels for dynamic risk strat-ification of patients with acute coronary syndromes. Circulation 2004;110:3206–3212.

9. Bettencourt P, Azevedo A, Pimenta J, Frioes F, Ferreira S, Ferreira A.N-terminal-pro-brain natriuretic peptide predicts outcome after hospi-tal discharge in heart failure patients. Circulation 2004;110:2168–2174.

0. Poldermans D, Bax JJ, Boersma E, De Hert S, Eeckhout E, Fowkes G,Gorenek B, Hennerici MG, Iung B, Kelm M, Kjeldsen KP, KristensenSD, Lopez-Sendon J, Pelosi P, Philippe F, Pierard L, Ponikowski P,Schmid JP, Sellevold OF, Sicari R, Van den Berghe G, Vermassen F,Hoeks SE, Vanhorebeek I, Vahanian A, Auricchio A, Ceconi C, DeanV, Filippatos G, Funck-Brentano C, Hobbs R, Kearney P, McDonaghT, McGregor K, Popescu BA, Reiner Z, Sechtem U, Sirnes PA,Tendera M, Vardas P, Widimsky P, De Caterina R, Agewall S, AlAttar N, Andreotti F, Anker SD, Baron-Esquivias G, Berkenboom G,Chapouotot L, Cifkova R, Faggiano P, Gibbs S, Hansen HS, Iserin L,Israel CW, Kornowski R, Eizagaechevarria NM, Pepi M, Piepoli M,Priebe HJ, Scherer M, Stepinska J, Taggart D, Tubaro M, Guidelinesfor pre-operative cardiac risk assessment and perioperative cardiacmanagement in non-cardiac surgery: the Task Force for PreoperativeCardiac Risk Assessment and Perioperative Cardiac Management inNon-Cardiac Surgery of the European Society of Cardiology (ESC)and endorsed by the European Society of Anaesthesiology (ESA). EurHeart J 2009;30:2769–2812.

1. McGowan JH Cleland JG. Reliability of reporting left ventricularsystolic function by echocardiography: a systematic review of 3 meth-ods. Am Heart J 2003;146:388–397.

2. Boersma E, Poldermans D, Bax JJ, Steyerberg EW, Thomson IR,Banga JD, van De Ven LL, van Urk H, Roelandt JR. Predictors ofcardiac events after major vascular surgery: Role of clinical charac-teristics, dobutamine echocardiography, and beta-blocker therapy.JAMA 2001;285:1865–1873.

3. Goei D, Schouten O, Boersma E, Welten GMJM, Dunkelgrun M,Lindemans J, van Gestel YRBM, Hoeks SE, Bax JJ, Poldermans D.Influence of renal function on the usefulness of N-terminal pro-B-typenatriuretic peptide as a prognostic cardiac risk marker in patientsundergoing noncardiac vascular surgery. Am J Cardiol 2008;101:122–126.

14. Redfield MM, Rodeheffer RJ, Jacobsen SJ, Mahoney DW, Bailey KR,Burnett JC Jr. Plasma brain natriuretic peptide concentration: impactof age and gender. J Am Coll Cardiol 2002;40:976–982.

15. Daniels LB, Clopton P, Bhalla V, Krishnaswamy P, Nowak RM,McCord J, Hollander JE, Duc P, Omland T, Storrow AB, AbrahamWT, Wu AH, Steg PG, Westheim A, Knudsen CW, Perez A, Kaza-negra R, Herrmann HC, McCullough PA, Maisel AS. How obesityaffects the cut-points for B-type natriuretic peptide in the diagnosis ofacute heart failure. Results from the Breathing Not Properly Multina-
tional Study. Am Heart J 2006;151:999–1005.

2

2

2

3


16. Lee TH, Marcantonio ER, Mangione CM, Thomas EJ, Polanczyk CA,Cook EF, Sugarbaker DJ, Donaldson MC, Poss R, Ho KK, LudwigLE, Pedan A, Goldman L. Derivation and prospective validation of asimple index for prediction of cardiac risk of major noncardiac sur-gery. Circulation 1999;100:1043–1049.

17. Schouten O, Hoeks SE, Goei D, Bax JJ, Verhagen HJ, Poldermans D.Plasma N-terminal pro-B-type natriuretic peptide as a predictor ofperioperative and long-term outcome after vascular surgery. J VascSurg 2009;49:435–441.

18. Rodseth RN, Padayachee L, Biccard BM. A meta-analysis of theutility of pre-operative brain natriuretic peptide in predicting early andintermediate-term mortality and major adverse cardiac events in vas-cular surgical patients. Anaesthesia 2008;63:1226–1233.

19. Karthikeyan G, Moncur RA, Levine O, Heels-Ansdell D, Chan MT,Alonso-Coello P, Yusuf S, Sessler D, Villar JC, Berwanger O, Mc-Queen M, Mathew A, Hill S, Gibson S, Berry C, Yeh HM, DevereauxPJ. Is a pre-operative brain natriuretic peptide or N-terminal pro-B-type natriuretic peptide measurement an independent predictor of ad-verse cardiovascular outcomes within 30 days of noncardiac surgery?A systematic review and meta-analysis of observational studies. J AmColl Cardiol 2009;54:1599–1606.

20. Ryding AD, Kumar S, Worthington AM, Burgess D. Prognostic valueof brain natriuretic peptide in noncardiac surgery: a meta-analysis.Anesthesiology 2009;111:311–319.

21. DeFilippi C, van Kimmenade RR, Pinto YM. Amino-terminal pro-B-type natriuretic peptide testing in renal disease. Am J Cardiol 2008;101:82–88.

22. de Lemos JA, Hildebrandt P. Amino-terminal pro-B-type natriuretic pep-tides: testing in general populations. Am J Cardiol 2008;101:16–20.

23. Mahla E, Baumann A, Rehak P, Watzinger N, Vicenzi MN, Maier R,Tiesenhausen K, Metzler H, Toller W. N-terminal pro-brain natriureticpeptide identifies patients at high risk for adverse cardiac outcomeafter vascular surgery. Anesthesiology 2007;106:1088–1095.

24. Hama N, Itoh H, Shirakami G, Nakagawa O, Suga S, Ogawa Y,Masuda I, Nakanishi K, Yoshimasa T, Hashimoto Y, Yamaguchi M,

Hori R, Yasue H, Nakao K. Rapid ventricular induction of brainnatriuretic peptide gene expression in experimental acute myocardialinfarction. Circulation 1995;92:1558–1564.

5. Sabatine MS, Morrow DA, de Lemos JA, Omland T, Desai MY,Tanasijevic M, Hall C, McCabe CH, Braunwald E. Acute changes incirculating natriuretic peptide levels in relation to myocardial isch-emia. J Am Coll Cardiol 2004;44:1988–1995.

6. Feringa HH, Schouten O, Dunkelgrun M, Bax JJ, Boersma E, ElhendyA, de Jonge R, Karagiannis SE, Vidakovic R, Poldermans D. PlasmaN-terminal pro-B-type natriuretic peptide as long-term prognosticmarker after major vascular surgery. Heart 2007;93:226–231.

27. Lainchbury JG, Troughton RW, Frampton CM, Yandle TG, Hamid A,Nicholls MG, Richards AM. NTproBNP-guided drug treatment forchronic heart failure: design and methods in the “BATTLESCARRED”trial. Eur J Heart Fail 2006;8:532–538.

28. Pfisterer M, Buser P, Rickli H, Gutmann M, Erne P, Rickenbacher P,Vuillomenet A, Jeker U, Dubach P, Beer H, Yoon SI, Suter T, Oster-hues HH, Schieber MM, Hilti P, Schindler R, Brunner-La Rocca HP.BNP-guided vs symptom-guided heart failure therapy: the Trial ofIntensified vs Standard Medical Therapy in Elderly Patients WithCongestive Heart Failure (TIME-CHF) randomized trial. JAMA 2009;301:383–392.

9. Jourdain P, Jondeau G, Funck F, Gueffet P, Le Helloco A, Donal E,Aupetit JF, Aumont MC, Galinier M, Eicher JC, Cohen-Solal A,Juilliere Y. Plasma brain natriuretic peptide-guided therapy to improveoutcome in heart failure: the STARS-BNP multicenter study. J AmColl Cardiol 2007;49:1733–1739.

0. Schou M, Gustafsson F, Videbaek L, Markenvard J, Ulriksen H, RydeH, Jensen JC, Nielsen T, Knudsen AS, Tuxen CD, Handberg J, So-rensen PJ, Espersen G, Lind-Rasmussen S, Keller N, Egstrup K,Nielsen OW, Abdulla J, Nyvad O, Toft J, Hildebrandt PR. Design andmethodology of the NorthStar Study: NT-proBNP stratified follow-up
in outpatient heart failure clinics—a randomized Danish multicenterstudy. Am Heart J 2008;156:649–655.

reattdvivn

Usefulness of At Rest and Exercise Hemodynamics to DetectSubclinical Myocardial Disease in Type 2 Diabetes Mellitus

Christine L. Jellis, MDa, Tony Stanton, MD, PhDa, Rodel Leano, BSa, Jennifer Martin, MD, PhDa,and Thomas H. Marwick, MD, PhDa,b,*

Patients with type 2 diabetes mellitus (T2DM) might have subclinical myocardial dysfunc-tion identified at rest or unmasked during exercise. We examined the correlates of themyocardial exercise response in patients with T2DM. Myocardial dysfunction was soughtduring at rest and exercise echocardiography in 167 healthy patients with T2DM (97 men,55 � 10 years). Myocardial ischemia was excluded using stress echocardiography. Stan-dard echocardiography and color tissue Doppler imaging measures (early diastolic tissuevelocity [Em], strain, and strain rate) were acquired at baseline and peak stress. Thecalibrated integrated backscatter was calculated from the at rest parasternal long-axis view.The longitudinal diastolic functional reserve index after exercise was defined as �Em [1 �(1/Embase)]. The clinical, anthropometric, and metabolic data were collected at rest andstress. Subclinical myocardial dysfunction at baseline (n � 24) was independently associ-ated with weight (odds ratio [OR] 1.02, p � 0.04) and hemoglobin A1c (OR 1.36, p � 0.03).This group displayed an impaired exercise response that was independently associated witha reduced exercise capacity (OR 0.84, p � 0.034) and longitudinal diastolic functionalreserve index (OR 0.69, p � 0.001). Inducible myocardial dysfunction (stress Em <�9.9cm/s) was identified after exercise in 70 of the remaining 143 subjects. This finding wasassociated with calibrated integrated backscatter (OR 1.08, p � 0.04) and lower peak heartrate (OR 0.97, p � 0.002) but not metabolic control. The intensity of the metabolicderangement in patients with T2DM was associated with subclinical at rest myocardialdysfunction, but not with the myocardial exercise response. In conclusion, the associationof an abnormal stress response with nonmetabolic factors, including backscatter andblunted peak heart rate, suggests potential roles for myocardial fibrosis and cardiacautonomic neuropathy in patients with nonischemic diabetic heart disease. © 2011
In addition to hastening atherosclerosis, type 2 diabetesmellitus (T2DM) and the metabolic syndrome1 have beenlinked to myocardial disease in the absence of ischemicheart disease and hypertension.2 This is likely multifacto-ial, secondary to the accumulation of advanced glycatednd products, myocardial fibrosis, microvascular disease,nd autonomic neuropathy. Diabetic heart disease is ini-ially asymptomatic; however, nonspecific symptoms of fa-igue, dyspnea, or reduced exercise tolerance will graduallyevelop. Early detection might facilitate measures to pre-ent disease progression. Tissue velocity and deformationmaging can detect myocardial dysfunction when the con-entional 2-dimensional echocardiographic parameters areormal.3 In early diabetic heart disease, myocardial function

aThe University of Queensland, Brisbane, Australia; and bClevelandClinic, Cleveland, Ohio. Manuscript received September 25, 2010; manu-script received and accepted October 5, 2010.

This study was supported in part by a Centres for Clinical ResearchExcellence award (455832) from the National Health and Medical Re-search Council, Canberra, Australia. Dr. Jellis was supported by a ResearchEntry Scholarship from the Vincent Fairfax Family Foundation, Sydney,Australia; and the Royal Australasian College of Physicians, Sydney,Australia.

E-mail address: [email protected] (T.H. Marwick).

might be preserved at rest, with exercise unmasking a blunt-ing of contraction and relaxation, indicative of an abnormalfunctional reserve.4 Longitudinal function is typically re-duced initially, reflective of the early involvement of thesubendocardial fibers.5 Impairment in the at rest and peakexercise systolic tissue velocity has been associated withcommon metabolic risk factors in asymptomatic patients.6

We sought to identify whether early diastolic tissue velocity(Em), deformation imaging, and tissue characterizationcould identify diabetic heart disease not apparent at rest andexamined the correlates of myocardial dysfunction withexercise.

Methods

A total of 167 apparently healthy subjects with T2DM(97 men, 55 � 10 years) and no macro- or microvascularcomplications of T2DM or history of hypertension or val-vular, congenital, or ischemic heart disease were recruitedfrom the hospital clinics of the Princess Alexandra Hospitaland its local community. Sinus rhythm and normal renalfunction were required for inclusion. Antihypertensive med-ications were withheld for �12 hours before testing. Thehuman research ethics committees of Princess AlexandraHospital and the University of Queensland (Brisbane, Aus-
tralia) approved the present study.
www.ajconline.org


h

tco[

cwst

h


Clinical data were collected regarding subject age, gen-der, weight, height, waist and hip circumference, smokingstatus, and duration of T2DM. Venesection was performedbefore exercise after the subjects had fasted for �8 hours.The tested parameters included fasting glucose, hemoglobinA1c (HbA1c), creatinine, and lipid profile. Microalbumin-uria, a marker of microvascular disease, was quantifiedusing a random urinary albumin/creatinine ratio and definedas �2.5 g/mol for men and �3.5 g/mol for women. The

Table 1Characteristics of myocardial dysfunction at rest and unmasked by exerci

Variable

Normal(n � 143)

Age (years) 55 � 10Type 2 diabetes mellitus duration (years) 5.3 � 5.5Weight (kg) 89.5 � 17.0Body mass index (kg/m2) 31.3 � 5.4Fasting glucose (mmol/L) 8.1 � 2.8Hemoglobin A1c (%) 7.4 � 1.4Total cholesterol

mmol/L 4.8 � 0.9mg/dl 186 � 35

Low-density lipoprotein cholesterolmmol/L 2.7 � 0.8mg/dl 104 � 31

Creatinine (mmol/L) 79 � 18Microalbuminuria (%) [albumin/creatinine (g/mol)] 16 (11%); 0.7*;

IQR 0.9Statin therapy 55 (38%)Angiotensin-converting enzyme inhibitors and/or

angiotensin-receptor blockers58 (41%)

� Blockers 9 (6%)At rest heart rate (beats/min) 85 � 13At rest systolic blood pressure (mm Hg) 133 � 17At rest diastolic blood pressure (mm Hg) 81 � 10Peak heart rate (beats/min) 163 � 20Exercise capacity (METs) 9.2 � 3.2At rest end-systolic volume (ml) 25 � 10At rest end-diastolic volume (ml) 73 � 19At rest ejection fraction (%) 65 � 7Stress end-systolic volume (ml) 18 � 7Stress end-diastolic volume (ml) 69 � 19Stress ejection fraction (%) 74 � 5Change in ejection fraction (%) 9 � 6At rest early diastolic tissue velocity (cm/s) �5.7 � 1.5At rest systolic tissue velocity (cm/s) 4.9 � 1.1Stress early diastolic tissue velocity (cm/s) �9.9 � 2.5Stress systolic tissue velocity (cm/s) 8.0 � 2.0Change in systolic tissue velocity (cm/s) 3.1 � 1.9Left ventricular longitudinal functional reserve index 3.5 � 2.5At rest strain (%) �20.8 � 3.1Stress strain (%) �21.2 � 3.5At rest strain rate (s�1) �1.3 � 0.3Stress strain rate (s�1) �1.9 � 0.5Calibrated integrated backscatter (dB) �17.1 � 5.4Height-indexed left ventricular mass (g/m2.7) 51.0 � 16.7

No statistically significant difference noted between at rest or Stressigh-density lipoprotein cholesterol, hypoglycemic therapy, or peak systo* Median value given as nonparametric distribution.IQR � interquartile range.

eart rate and blood pressure were measured at baseline, B

hroughout exercise, and during recovery. The peak exerciseapacity was estimated in METs according to the durationf exercise using the equation: METs � [speed � (0.1 �grade � 1.8]) � 3.5]/3.5.

Standard commercially available cardiac ultrasound ma-hines (Vivid 7, General Electric Medical Systems, Mil-aukee, WI) were used to perform M-mode and 2-dimen-

ional echocardiography to assess the chamber wallhickness, valvular morphology, and chamber volumes.

st Em Stress Em

bnormaln � 24)

p Value Normal(n � 73)

Abnormal(n � 70)

p Value

53 � 11 0.389 53 � 9 58 � 10 0.0079.7 � 8.9 0.024 5.1 � 5.4 6.1 � 6.3 0.3108.5 � 28.5 0.035 89.7 � 17.4 90.2 � 16.9 0.8483.7 � 7.2 0.054 31.5 � 5.6 31.2 � 5.3 0.7689.6 � 3.8 0.030 8.2 � 3.0 8.2 � 2.8 0.8738.1 � 1.6 0.024 7.5 � 1.4 7.4 � 1.4 0.458

0.105 0.0194.5 � 0.8 5.0 � 0.8 4.7 � 1.074 � 31 193 � 31 182 � 39

0.035 0.0842.3 � 0.8 2.8 � 0.8 2.6 � 0.989 � 31 108 � 31 101 � 3577 � 24 0.624 78 � 17 80 � 19 0.4595%); 1.1*;IQR 2.2

0.064 7 (10%); 0.8*;IQR 0.7

8 (11%); 0.7*;IQR 0.9

0.720

6 (67%) 0.010 24 (33%) 32 (46%) 0.1168 (75%) 0.005 28 (39%) 30 (43%) 0.474

1 (4%) 0.684 1 (1%) 8 (11%) 0.01388 � 16 0.330 88 � 12 81 � 13 0.00138 � 17 0.259 132 � 16 134 � 17 0.57286 � 8 0.033 81 � 10 81 � 9 0.94959 � 19 0.363 170 � 17 155 � 20 �0.0017.5 � 2.5 0.013 9.4 � 3.3 9.0 � 3.0 0.43929 � 12 0.137 26 � 10 25 � 10 0.31478 � 26 0.210 72 � 20 73 � 19 0.75364 � 8 0.241 64 � 6 67 � 7 0.01710 � 8 0.224 18 � 7 18 � 7 0.90671 � 23 0.590 69 � 18 70 � 21 0.71873 � 6 0.553 74 � 5 74 � 6 0.8529 � 7 0.825 10 � 6 7 � 7 0.014

3.1 � 1.0 �0.001 �5.8 � 1.4 �5.5 � 1.6 0.3652.9 � 1.4 �0.001 4.9 � 1.1 4.8 � 1.2 0.6146.3 � 4.5 �0.001 �11.5 � 1.9 �8.1 � 1.7 �0.0015.4 � 2.6 �0.001 8.3 � 1.9 7.5 � 2.2 0.0122.5 � 1.6 0.184 3.5 � 1.8 2.6 � 1.9 0.0061.3 � 2.2 �0.001 5.3 � 1.7 1.6 � 1.7 �0.0019.6 � 3.7 0.09 �20.6 � 3.0 �20.8 � 3.2 0.6051.2 � 3.6 0.978 �20.6 � 3.7 �22.0 � 3.2 0.0201.4 � 0.3 0.160 �1.3 � 0.3 �1.3 � 0.3 0.7282.0 � 0.5 0.181 �1.9 � 0.5 �1.9 � 0.4 0.6926.3 � 5.7 0.539 �18.0 � 4.9 �15.9 � 5.8 0.0195.2 � 16.6 0.125 52.4 � 18.1 49.6 � 15.0 0.360

roups for gender, smoking status, height, waist/hip ratio, triglycerides,iastolic blood pressure.

se

At Re

A(

93

1

6 (2

11

1

1

�

�

�1�2��

�14

Em glic or d

aseline parasternal and apical images were acquired in

abwt

m

ww

617Miscellaneous/Subclinical Diabetic Myocardial Disease

gray scale and color tissue Doppler imaging (TDI) formats.These were used as the at rest reference images for com-parison of the stress echocardiographic images in the sameviews and to enable off-line TDI parameter measurement.Stress echocardiography was performed to exclude induciblesegmental myocardial dysfunction indicative of underlying he-modynamically significant epicardial coronary artery disease.The subjects underwent treadmill exercise using the Bruceprotocol, and the peak images in the parasternal and apicalviews using the same formats were acquired. The images weredigitally saved for off-line analysis.

The left ventricular end-systolic and end-diastolic vol-umes were quantified at rest and peak to calculate theejection fraction using the modified Simpson biplanemethod. Conventional apical views (4-chamber, 2-chamber,and long-axis) in color TDI format at rest and peak wereused to obtain the tissue velocity, strain, and strain ratecurves from the 6 basal segments using standard commer-cial software (Echopac, GE Vingmed). The peak systolictissue velocity (Sm) and peak Em were calculated from thetissue velocity curves by placing a sample volume at theannulus of the mitral valve. An average value from 3 con-secutive tissue velocity curves was established. On somestress images, the transmitral and TDI diastolic velocitycurve peaks were fused. In these instances, we waited untilthe heart rate had decreased and they had separated.

Abnormal diastolic function at rest was defined as an atrest septal Em �2 SD of normal for age.7 Normal age-djusted values for peak exercise stress Em have not yeteen defined. Hence, abnormal exercise diastolic functionas identified when the stress septal Em was ��9.9 cm/s,

he mean value in a nondiabetic control population.4 Lon-gitudinal left ventricular diastolic functional reserve index,a measure of augmentation in diastolic relaxation duringexercise, was calculated using the previously defined equa-tion: longitudinal left ventricular diastolic function re-serve index � �Em [1 � (1/Embase)], where �Em wasthe change in Em from baseline to peak exercise andEmbase was the early diastolic tissue velocity at baseline

easured at the septal mitral annulus.4 This normalizationequation was used to express the �Em with respect to thebaseline Em at rest. According to previous studies, thepreserved contractile reserve was defined as augmentationof the ejection fraction �4%8 and/or an increase in thestress Sm from the baseline at rest Sm (�Sm) �2.4 cm/s.4

The strain and strain rate curves were derived from theapical views in color TDI format by placing sample vol-umes in the mid-myocardial layer of the 6 basal segmentsand tracking the position of the sample volume throughoutthe cardiac cycle. An average peak value from 3 consecutivecurves was used to calculate the strain and strain rate. Theangle of incidence between the transducer and the wall ofinterest was maintained at �20° to not underestimate thedegree of deformation. The calibrated integrated backscatter(cIB) was calculated by measuring the tissue intensity of thepericardium, posterior wall, and anterior septum in theparasternal long-axis view. Automated tissue tracking en-abled maintenance of the sample volume of interest withinthe designated myocardial segment throughout the cardiaccycle. An integrated backscatter curve was derived using
commercial software (Echopac, GE Vingmed). The mean
cIB was calculated by subtracting the pericardial integratedbackscatter intensity at end-diastole from the integratedbackscatter intensity of the posterior wall and the anteriorseptum, which were then averaged.

Interobserver and intraobserver variability were assessedby repeat measurement of the tissue velocity, strain, strainrate, and cIB on the at rest and peak stress images from 10randomly selected subjects. The original observer (C.J.) wasunaware of the previous measurements performed �4weeks earlier to evaluate the reproducibility. A second ob-server (R.L.) was unaware of the first observer’s results toassess the variability. In addition to the absolute differencebetween the measurements, the intraclass correlation coef-ficient was used to determine overall variability. This anal-ysis of reliability was performed using a 2-way randomeffects model to assess absolute agreement. The results areexpressed as the mean � SD. The analysis between definedcategorical groups (normal vs abnormal) was performed usingStudent’s independent t test for continuous variables and thechi-square test for categorical variables. Independent associa-tions between the echocardiographic and metabolic parameterswere sought with a stepwise selection method to build logisticor linear regression models of the independent variables. Can-didate variables for the models were selected from the unad-justed correlates listed in Table 1 that were significant atp �0.10 and not co-linear. Statistical analysis was per-formed using standard software (Statistical Package for SocialSciences, version 16, SPSS, Chicago, Illinois). p Values �0.05 were statistically significant.

Results

All 167 subjects had an ejection fraction at rest of �50%and no evidence of inducible ischemia on the exercise stressechocardiogram. Sinus rhythm was maintained throughouttesting in all participants. At baseline, 24 subjects had sub-clinical dysfunction as shown by a reduced at rest Em(septal Em �2 SD of normal for age). Differences werenoted between the metabolic parameters of those with ab-normal findings and the 143 subjects with normal at restmyocardial function (Table 1). The subjects with abnormalfindings weighed more, had a greater fasting blood glucoseand HbA1c level, and had had a longer duration of T2DM.Probably because of the greater use of statin therapy, thelow-density lipoprotein cholesterol level was lower. Nodifference in serum creatinine was noted between the 2groups, and the prevalence of microalbuminuria was low.The height-indexed left ventricular mass demonstrated nodisparity between the 2 groups. In a logistic regressionmodel, weight (odds ratio [OR] 1.02, p � 0.04) and HbA1c(OR 1.36, p � 0.027), but not the duration of T2DM, wereassociated with dysfunction at rest. On linear regressionanalysis, the duration of T2DM (� � �0.214, p � 0.006)

as an independent correlate of the Em at rest and weightas of borderline significance (� � �0.148, p � 0.055).The analysis of the exercise responses of the same

groups showed those with abnormal myocardial dysfunctionat rest displayed reduced exercise capacity. In addition to areduced Em, this group had a low Sm at rest. The Em, Sm,and longitudinal diastolic functional reserve index were all
significantly reduced with exercise, although no differences

0ggtc

rEs

0

d with

TM

ASASA

C


were found in deformation or tissue intensity (Table 1).Logistic regression modeling of these baseline groups andthe exercise response demonstrated that exercise capacity(OR 0.838, p � 0.034) and longitudinal diastolic functionalreserve index (OR 0.689, p � 0.001) were inversely asso-ciated with myocardial dysfunction at rest. Contractile re-serve, as measured by the change in ejection fraction or�Sm with exercise, was similarly preserved in both groups,irrespective of their baseline Em magnitude (Table 1).

Of the 143 patients with normal function at rest, 70 hadinducible myocardial dysfunction after exercise (stress Em��9.9 cm/s). No difference was seen between the meta-bolic and physiologic markers of those with abnormal stressEm compared to those with a normal stress Em. However,patients with an abnormal stress Em were older and hadlower at rest and peak heart rates. No difference in mi-croalbuminuria or the height-indexed left ventricular masswas noted between the 2 groups. The antihypertensive reg-imens were similar between the 2 groups with respect toangiotensin-converting enzyme inhibitors and angiotensin-receptor blockers. �-Blocker use was very low overall butmarginally greater in the abnormal stress response group(Table 1). Linear regression modeling showed that the peakheart rate was an independent associate of stress Em (� �0.413, p �0.001).

Patients with an abnormal stress response to exercise hada significantly smaller longitudinal diastolic functional re-serve index than those without inducible myocardial dys-function (Table 1). When corrected for total exercise ca-pacity, the association between a reduced longitudinal

Figure 1. Abnormal stress Em response associate

diastolic functional reserve index and abnormal Em at s

rest and abnormal stress Em was maintained. No relationwas displayed between reduction of longitudinal diastolicfunctional reserve index and the derangement of meta-bolic parameters. However, a reduced longitudinal dia-stolic functional reserve index was independently pre-dicted by a reduced peak heart rate (� � 0.27, p �

.001). The contractile reserve was preserved in bothroups after exercise; however, this was of a significantlyreater magnitude in the normal stress Em group than inhe abnormal stress Em group for both �Sm and thehange in ejection fraction (Table 1).

Neither the strain or strain rate at rest nor the stress strainate were associated with an abnormal at rest Em or stressm. The independent metabolic associations of impairedtrain at rest were weight (� � �0.16, p � 0.035), HbA1c

(� � �0.16, p � 0.03), and triglycerides (� � �0.17, p �.03). No relation between the metabolic parameters and the

significantly greater cIB than normal stress Em.

able 2ean difference in measurements between same and different observers

SameObserver

BetweenObservers

t rest early diastolic tissue velocity (cm/s) 0.1 � 0.4 0.6 � 1.3tress early diastolic tissue velocity (cm/s) �0.5 � 1.3 �1.7 � 3.0t rest strain (%) �1.7 � 1.8 �0.7 � 3.5tress strain (%) �0.3 � 1.9 �0.5 � 4.2t rest strain rate (s�1) �0.1 � 0.2 �0.1 � 0.3

Stress strain rate (s�1) 0.02 � 0.3 �0.1 � 0.4alibrated integrated backscatter (dB) �0.02 � 4.7 �0.2 � 5.7

train rate at rest was identified. Peak strain was not strongly

0wppwtcusatw

sat

fsr(s0

D

Tfwspdacarmlbi

a


associated with the metabolic parameters, with only renalfunction independently related (creatinine, � � �0.20, p �.01). The peak strain rate was independently associatedith age (� � �0.18, p � 0.03), peak heart rate (� � 0.21,� 0.01), and systolic blood pressure at rest (� � �0.18,� 0.02). No difference was seen in the cIB between thoseith a normal versus an abnormal Em at rest. However,

hose with an abnormal stress Em demonstrated a signifi-antly greater cIB than those with a normal stress Em onnivariate analysis (Figure 1). Differences in the cIB per-isted when corrected for the use of potential antifibroticgents (angiotensin-converting enzyme inhibitors or angio-ensin-receptor blockers), although statistical significanceas lost (�16 � 6 dB vs �18 � 5 dB, p � 0.081), likely

related to the large proportion of subjects excluded. Thepeak heart rate (OR 0.97, p � 0.002) and cIB (OR 1.1, p �0.04), but not age or heart rate at rest, were independentlypredictive of myocardial dysfunction with exercise. Agreater cIB was independently associated with a lowerEm at rest (� � �0.16, p � 0.04) and greater waist/hipratio (� � 0.29).

The interobserver and intraobserver measurementshowed good concordance, with the results reproduciblend without significant variability from original observa-

Figure 2. Comparison between baseline at rest and peak stress color TDIB) Em responses to exercise stress. White arrows indicate Em peak; abnoge.

ions. Compared to the original results, no significant dif- c

erence was seen in the at rest Em, stress Em, at rest strain,tress strain, at rest strain rate, stress strain rate, or cIB whenemeasured by the original observer or second observerTable 2). The overall reliability of the inter- and intraob-erver measures was high, with an intraclass correlation of.986 (95% confidence interval 0.981 to 0.991).

iscussion

The intensity of metabolic disturbances in those with2DM has been associated with subclinical myocardial dys-

unction for both diastolic and systolic parameters at rest butas unrelated to the stress response. Instead, abnormal

tress responses appear to be associated with myocardialroperties consistent with structural change. Diabetic heartisease is characterized by myocardial collagen depositionnd myofibrillar hypertrophy in the absence of valvular,ongenital, hypertensive, or ischemic heart disease. Unlikefocal scar in ischemic cardiomyopathy, this fibrosis is a

eactive and labile process governed by extrinsic, primarilyetabolic, factors such as blood glucose control, accumu-

ation of advanced glycation end products, lipid profile,lood pressure, and obesity. Many of these variables resultn stimulation of inflammatory pathways and activation of

elocity curves for patients with normal (patient A) and abnormal (patientress Em defined as ��9.9 cm/s; both patients had normal Em at rest for

tissue vrmal st

ytokines and the renin-angiotensin-aldosterone system,

swfieg

mptaopda

lt

ppltc

tfdpmedpEcd

spriBptaa


leading to interstitial and perivascular fibrosis. Over time,the increasing myocardial collagen burden leads to in-creased ventricular stiffness and diastolic dysfunction. Inaddition to fibrosis, microvascular ischemia and cardiacautonomic neuropathy likely play important roles in theconstellation of pathologic processes that combine to causemyocardial dysfunction in the setting of T2DM.

The correlates of the imaging findings at rest are sup-portive of previous findings relating impaired Sm to meta-bolic derangement.6 Independent metabolic correlates ofubclinical myocardial dysfunction have included HbA1c,eight, triglycerides, and the duration of diabetes. Thesendings agree with the previously documented deleteriousffects of chronic hyperglycemia, obesity,9 and hypertri-lyceridemia10 on myocardial function. The closer relation

between weight and Em at rest, rather than body massindex, likely reflects the increasingly recognized detrimen-tal effect of visceral adiposity in T2DM that has beenrecently linked to increased cardiovascular risk and mortal-ity.11 The association between a reduction in the sensitive

yocardial parameters at rest such as Em and Sm andoorer metabolic control in the present study might reflecthe increased stiffness and impaired myocardial relaxationssociated with increased collagen deposition in the settingf myocardial fibrosis. Previous histologic studies have sup-orted this finding by demonstrating exaggerated myocar-ial collagen deposition in the settings of hyperglycemiand arterial hypertension.12 Our findings appear unrelated to

both systolic blood pressure and left ventricular mass. Al-though no difference was found in the cIB between thedesignated at rest Em groups, a relation between the meta-bolic parameters, including abdominal obesity, and cIB wasnoted. This also supports a link between diabetic metabolicderangement and myocardial fibrosis.

Our results lend to the speculation that increased myo-cardial stiffness secondary to fibrosis might play a moreimportant role during exercise stress than at rest and, con-versely, that the at rest metabolic disturbances are lessrelevant to the stress response. Because early diastole isprimarily dependent on myocardial relaxation to maximizeventricular filling, it can be expected that diastolic param-eters such as Em will be affected earlier in the diseaseprocess of fibrosis than the systolic markers. Hence, usingEm as a measure of myocardial relaxation might be a moresensitive parameter of very early myocardial dysfunction,particularly in the asymptomatic population. The results ofthe present study have demonstrated that myocardial dys-function unmasked by exercise (defined by an impairedstress Em) could be identified in approximately 50% ofasymptomatic patients with T2DM and normal resting func-tion (Figure 2). As expected, a significant proportion ofthose with an impaired diastolic stress response also clearlydemonstrated a reduced myocardial systolic response toexercise with a reduction in the stress Sm and impairedcontractile reserve.

In addition to the TDI parameters, myocardial reflectivitymeasured using cIB can be employed noninvasively to char-acterize the myocardial tissue for evidence of collagen dep-osition, with a greater cIB (less negative) score indicative ofincreased myocardial reflectivity as a measure of greater
fibrosis.13 The correlation between backscatter and histo-
ogically quantified collagen has been previously valida-ed.14 Within our T2DM population, an inverse association

was found between cIB and stress Em, with a greater cIBassociated with exercise-induced myocardial dysfunctiondespite normal function at rest. This supports the hypothesisthat fibrosis is responsible for the impaired relaxation seenin T2DM—even early in the disease process. The cIB atrest might therefore be a useful tool in the prediction of anabnormal exercise response.

Exercise capacity is well recognized as a significantpredictor of cardiovascular and all-cause mortality.15 Im-aired exercise capacity has previously been shown even inatients with uncomplicated T2DM and appears to be re-ated to reduced peak oxygen consumption, perhaps owingo factors limiting oxygen delivery, rather than glycemicontrol.16 This reduction in oxygen delivery might result in

failure to adequately achieve the metabolic demands ofmany tissues during stress, including the myocardium andskeletal muscle, thereby reducing the exercise capacity. Thefailure of oxygen delivery during periods of increased met-abolic demand might be related to impaired compensatorycirculatory regulation in the setting of cardiac autonomicneuropathy, an underlying disease of the microvascular cir-culation impeding oxygen delivery to the muscle bed, andalso failure to upregulate myocardial uptake of oxygenowing to myocyte replacement with fibrosis.

Impaired exercise capacity has previously been associ-ated with reduced left ventricular diastolic functional re-serve in a heterogeneous population with impaired myocar-dial relaxation at rest.17 Our results support this relation inhe T2DM population, with subjects with myocardial dys-unction at rest demonstrating both reduced longitudinaliastolic functional reserve index and reduced exercise ca-acity. However, a unique finding was that those with nor-al myocardial function at rest but an abnormal response to

xercise stress (abnormal stress Em) had significantly re-uced longitudinal diastolic functional reserve index com-ared to those with a normal stress response (normal stressm). These findings highlight the important role of myo-ardial diastolic relaxation in maintaining normal myocar-ial function and exercise capacity.

The strongest associations with blunted contractile re-ponse to exercise on the TDI parameters were increasedatient age and a diminished peak heart rate response. Aeduction of the peak heart rate was also associated withmpaired longitudinal diastolic functional reserve index.lunting of the increase in heart rate with exercise hasreviously been documented in patients with T2DM, withhis finding most prominent in association with cardiacutonomic neuropathy involving both the parasympatheticnd the sympathetic nervous systems.18 Our findings sug-

gest that these patients might have early cardiac autonomicneuropathy that is only revealed with maximum stimulationof the sympathetic nervous system at peak exercise. How-ever, this could also be attributable to the concomitant useof rate-controlling medications. This reduction in the peakheart rate might result in reduced maximal cardiac output,which would further impair oxygen delivery during periodsof increased metabolic demand.

Although metabolic factors correlated with myocardial
dysfunction evident at rest, they correlated poorly with


myocardial dysfunction unmasked with exercise. The im-plication is that reliance purely on the metabolic and at restvariables might fail to detect subclinical diabetic heart dis-ease in a proportion of patients. This could result in thesepatients being incorrectly classified as having normal myo-cardial function, resulting in them not receiving therapiesthat could improve their cardiac function or prevent furtherdeterioration, such as exercise, neurohormonal antagonists,and more stringent blood pressure or blood glucose control.Using exercise to unmask patients with subclinical myocar-dial dysfunction in T2DM, the clinician might have a moresensitive method to detect a potentially “at risk” population.

The present study was primarily limited by its observa-tional nature. Thus, although associations have been found,direct causal relations cannot be attributed. The continua-tion of prescribed hypoglycemic, antihypertensive, and an-tilipid therapy during the present study was unavoidable.However, these agents were not more prevalent in thegroups without myocardial dysfunction. The use of noninva-sive functional stress echocardiography to exclude macrovas-cular ischemic heart disease means that false-negative findingswere possible. However, our center has both high sensitivityand specificity in stress echocardiogram interpretation com-pared to coronary angiography,19 making it unlikely that he-modynamically significant coronary lesions were underappre-ciated. Invasive coronary angiography of this asymptomaticpopulation for the purposes of the present observational studycould not be justified on ethical grounds.

1. Alberti KG, Zimmet P, Shaw J. The metabolic syndrome—a newworldwide definition. Lancet 2005;366:1059–1062.

2. Rubler S, Dlugash J, Yuceoglu YZ, Kumral T, Branwood AW, Grish-man A. New type of cardiomyopathy associated with diabetic glomer-ulosclerosis. Am J Cardiol 1972;30:595–602.

3. Fang ZY, Yuda S, Anderson V, Short L, Case C, Marwick TH.Echocardiographic detection of early diabetic myocardial disease.J Am Coll Cardiol 2003;41:611–617.

4. Ha JW, Lee HC, Kang ES, Ahn CM, Kim JM, Ahn JA, Lee SW, ChoiEY, Rim SJ, Oh JK, Chung N. Abnormal left ventricular longitudinalfunctional reserve in patients with diabetes mellitus: implication fordetecting subclinical myocardial dysfunction using exercise tissueDoppler echocardiography. Heart 2007;93:1571–1576.

5. Fang ZY, Leano R, Marwick TH. Relationship between longitudinaland radial contractility in subclinical diabetic heart disease. Clin SciLond 2004;106:53–60.

6. Vinereanu D, Nicolaides E, Tweddel AC, Madler CF, Holst B, Boden
LE, Cinteza M, Rees AE, Fraser AG. Subclinical left ventriculardysfunction in asymptomatic patients with type II diabetes mellitus,
related to serum lipids and glycated hemoglobin. Clin Sci Lond2003;105:591–599.

7. Sun JP, Popovic ZB, Greenberg NL, Xu XF, Asher CR, Stewart WJ,Thomas JD. Designation of tissue Doppler normal range. In: MarwickTH, Yu CM, Sun JP, eds. Myocardial Imaging—Tissue Doppler andSpeckle Tracking. Massachusetts: Blackwell Publishing; 2007:36–51.

8. Leung DY, Griffin BP, Stewart WJ, Cosgrove DM III, Thomas JD,Marwick TH. Left ventricular function after valve repair for chronicmitral regurgitation: predictive value of preoperative assessment ofcontractile reserve by exercise echocardiography. J Am Coll Cardiol1996;28:1198–1205.

9. Wong CY, O’Moore-Sullivan T, Leano R, Byrne N, Beller E, Mar-wick TH. Alterations of left ventricular myocardial characteristicsassociated with obesity. Circulation 2004;110:3081–3087.

10. de las Fuentes L, Waggoner AD, Brown AL, Davila-Roman VG.Plasma triglyceride level is an independent predictor of altered leftventricular relaxation. J Am Soc Echocardiogr 2005;18:1285–1291.

11. Czernichow S, Kengne AP, Huxley RR, Batty GD, de Galan B,Grobbee D, Pillai A, Zoungas S, Marre M, Woodward M, Neal B,Chalmers J. Comparison of waist-to-hip ratio and other obesity indicesas predictors of cardiovascular disease risk in people with type-2diabetes: a prospective cohort study from ADVANCE. Eur J Cardio-vasc Prev Rehabil Epub 2010 Jul 12.

12. Di Bello V, Talarico L, Picano E, Di Muro C, Landini L, Paterni M,Matteucci E, Giusti C, Giampietro O. Increased echodensity of myo-cardial wall in the diabetic heart: an ultrasound tissue characterizationstudy. J Am Coll Cardiol 1995;25:1408–1415.

13. Pardo Mindan FJ, Panizo A. Alterations in the extracellular matrixof the myocardium in essential hypertension. Eur Heart J 1993;14(Suppl J):12–14.

14. Picano E, Pelosi G, Marzilli M, Lattanzi F, Benassi A, Landini L,L’Abbate A. In vivo quantitative ultrasonic evaluation of myocardialfibrosis in humans. Circulation 1990;81:58–64.

15. Snader CE, Marwick TH, Pashkow FJ, Harvey SA, Thomas JD, LauerMS. Importance of estimated functional capacity as a predictor ofall-cause mortality among patients referred for exercise thallium sin-gle-photon emission computed tomography: report of 3,400 patientsfrom a single center. J Am Coll Cardiol 1997;30:641–648.

16. Regensteiner JG, Sippel J, McFarling ET, Wolfel EE, Hiatt WR.Effects of non-insulin-dependent diabetes on oxygen consumptionduring treadmill exercise. Med Sci Sports Exerc 1995;27:875–881.

17. Ha JW, Choi D, Park S, Choi EY, Shim CY, Kim JM, Ahn JA, LeeSW, Oh JK, Chung N. Left ventricular diastolic functional reserveduring exercise in patients with impaired myocardial relaxation at rest.Heart 2009;95:399–404.

18. Bottini P, Tantucci C, Scionti L, Dottorini ML, Puxeddu E, Reboldi G,Bolli GB, Casucci G, Santeusanio F, Sorbini CA. Cardiovascularresponse to exercise in diabetic patients: influence of autonomic neu-ropathy of different severity. Diabetologia 1995;38:244–250.

19. Ingul CB, Stoylen A, Slordahl SA, Wiseth R, Burgess M, MarwickTH. Automated analysis of myocardial deformation at dobutamine
stress echocardiography: an angiographic validation. J Am Coll Car-diol 2007;49:1651–1659.

icSB

Specific Characteristics of Sudden Death in a MediterraneanSpanish Population

M. Teresa Subirana, MDa,*, Josep O. Juan-Babot, MD, PhDb, Teresa Puig, MD, PhDc,Joaquín Lucena, MD, PhDd, Antonio Rico, MD, PhDd, Manuel Salguero, MD, PhDe,

Juan C. Borondo, MDf, Jorge Ordóñez, MD, PhDg, Josep Arimany, MD, PhDh,Rafael Vázquez, MD, PhDi, Lina Badimon, MD, PhDb, Gaetano Thiene, MDj, and

Antonio Bayés de Luna, MD, PhDb

Most of the data reported on sudden cardiac death has been from studies of Anglo-Saxonpatients. We conducted a study to ascertain the relation between sudden death (SD) andsome epidemiologic, clinical, and biochemical parameters and to assess the coronaryhistopathologic aspects of subjects in a Spanish population who had died suddenly. A totalof 204 subjects (86% men), aged 12 to 80 years (mean 54 � 15), who had died fromout-of-hospital natural SD were evaluated. Only 15% of subjects had been previouslydiagnosed with heart disease. Pathologic evidence of underlying cardiovascular disease wasfound in 90% of cases, with coronary heart disease (CHD) the most frequent (58%). TheCHD was acute coronary thrombosis in 41% and a stable plaque with luminal narrowingof >75% in 59%. An old myocardial infarction was found in 31% of the SD victims.Cardiac hypertrophy was found in 48%, with no relation between the presence of cardiachypertrophy and CHD. Patients with stable plaques had a greater heart weight than didthose with acute coronary thrombosis (p � 0.02). Male gender, older age, smoking, andlow-density lipoprotein cholesterol/high-density lipoprotein cholesterol ratio of >3 wereassociated with CHD. A greater percentage of patients with an eroded and/or rupturedplaque than patients with a stable plaque were smokers. Only smoking and a low-densitylipoprotein/high-density lipoprotein cholesterol ratio of >3 were associated with an erodedand/or ruptured plaque. In conclusion, compared with the findings from studies of Anglo-Saxon patients, a lower incidence of CHD and acute coronary thrombosis and a greaterincidence of cardiac hypertrophy were found in SD victims of a Mediterranean Spanish
population. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:622–627)
tdaaoscihUi

Sudden death (SD) represents 12% to 13% of overallnatural mortality when the temporal definition is restrictedto death occurring �2 hours after the onset of symptoms,1–4

with approximately 50% of deaths in patients with cardio-vascular disease.5–7 Sudden cardiac death (SCD) representsabout 80% to 90% of all SDs.2,3,8 Therefore, SCD consti-

aDepartment of Cardiology, Hospital de la Santa Creu i Sant Pau,Barcelona, and Universitat Autònoma de Barcelona, Barcelona, Spain;bCardiovascular Research Center, CSIC-ICCC, Hospital de la Santa Creui Sant Pau, Barcelona, Spain; cDepartment of Epidemiology, Hospital de laSanta Creu i Sant Pau, Barcelona, and Universitat Autònoma de Barcelona,Spain; dForensic Pathology Service, Institute of Legal Medicine, Seville,Spain; eNational Institute of Toxicology and Forensic Sciences, Sev-lle, Spain; fNational Institute of Toxicology and Forensic Sciences, Bar-elona, Spain; gDepartment of Biochemistry, Hospital de la Santa Creu iant Pau, Barcelona, Spain; hInstitute of Legal Medicine of Catalonia,arcelona, Spain; iDepartment of Cardiology, Hospital Universitario Nues-

tra Señora de Valme, Seville, Spain; and jDepartment of Medico-Diagnos-tic Sciences and Special Therapies, University of Padua Medical School,Padua, Italy. Manuscript received July 16, 2010; manuscript received andaccepted October 5, 2010.

This work was supported by a grant from “Redes temáticas de inves-tigación cooperativa. Instituto de Salud Carlos III” (G03-078).

*Corresponding author: Tel: (34) 93-556-5945; fax: (34) 93-556-5603.
E-mail address: [email protected] (M.T. Subirana).

utes one of the most important challenges of modern car-iology. Our understanding of the pathophysiologic mech-nisms of SCD, as well as the correlation between SCD andssociated diseases and risk factors, has mainly been basedn data from studies of white Anglo-Saxon patients thathowed coronary heart disease (CHD) to be a fundamentalause of SCD.9,10 The Seven Countries Study demonstratedn a 25-year follow-up period that Southern European co-orts had a lower risk of fatal CHD than other European ornited States cohorts.11 However, only partial data on the

ncidence of SCD in Spain are available,8 and it is unknownwhether its clinical and pathologic characteristics differfrom those reported from Anglo-Saxon countries.12 Thepresent study was designed to investigate the relation be-tween SCD and different epidemiologic, clinical, and bio-chemical parameters in a Spanish population and to char-acterize the coronary histopathologic features of thesubjects who died suddenly.

Methods

A total of 204 victims of out-of-hospital natural SDaged 12 to 80 years from 2 Spanish autonomous commu-nities, Catalonia and Andalusia, were included. Thosewho had died long after an aborted SD were excluded.
Autopsy was performed within 18 hours after death. A
www.ajconline.org


bod

liwqlccdahoh

dwt

laotgtAttoV�Srw

tc3is

(smc

EA

ttaaast

s

isc(oCM

tdstt

apL

iwa

623Miscellaneous/Sudden Death in a Mediterranean Population

forensic autopsy protocol13,14 was applied. Femorallood and vesical or pelvic kidney urine samples werebtained by puncture. The coronary arteries and myocar-ium were specifically studied.

Information on the sociodemographic data, cardiovascu-ar risk factors, and present and previous disorders suggest-ng cardiovascular disease in the victims and their relativesas obtained from the close family or friends by telephoneuestionnaire. The cardiovascular risk factors were col-ected as continuous (i.e., height, weight, body mass index),ategorical (i.e., yes, no, exsmoker for smoking), and di-hotomous (i.e., regular alcohol intake, physical activity,iabetes, hypertension, dyslipemia, and the use of psycho-ctive drugs) variables. In the case of previously recognizedeart disease, an attempt was made to compare the databtained from the relatives with the data retrieved from theospital records.

SD was defined as a natural, nonviolent, unexpectedeath occurring within 1 hour of the onset of symptoms orithin 24 hours of a previously stable medical condition, if

he event had not been witnessed.Death from CHD was diagnosed when cross-sectional

uminal narrowing of a major coronary artery of �75% orn acute thrombosis related to rupture or erosion of a cor-nary plaque was found. The culprit plaque was defined ashat with an acute thrombus or, in its absence, that with thereatest degree of cross-sectional luminal narrowing rela-ive to the internal elastic lamina at the narrowest segment.n acute ruptured plaque consisted of a continuous luminal

hrombus with an underlying lipid-rich core. When thehrombus was in direct contact with the intimal layer, with-ut rupture of a lipid pool, the plaque was defined as eroded.ulnerable plaques were defined as having a fibrous cap65 �m with macrophage and T-lymphocyte infiltration.

table plaques were defined as those causing luminal nar-owing of �75% in the absence of luminal thrombosis andere considered vulnerable or nonvulnerable.12

A heart weight �450 g in men and �400 g in womenwas considered cardiac hypertrophy.15,16 We used the heartweight, instead of left ventricular thickness, because theheart rate might provide more information about the cardiacmass. In dilated hearts with an abnormal cardiac mass, theleft ventricular thickness can sometimes be normal.

The smoking status of each patient was classified ascurrent daily smoker,17 nonsmoker, or exsmoker.

Hypertension was defined according to the guidelines ofthe European Society of Hypertension/European Society ofCardiology.18

The heart tissue was fixed in formaldehyde by retro-grade perfusion at systemic pressure. The coronary arter-ies were dissected and embedded in paraffin, and 5-�m-hick sections were stained with hematoxylin-eosin. Theoronary arteries were studied by serial sectioning at-mm intervals after decalcification. Any segment show-ng cross-sectional luminal narrowing of �50% wastudied histologically.

Histologic images were studied in a Leica MZ-9.5Leica Microsistemas, Barcelona, Spain) stereomicro-cope to quantify the stenotic area. Image capture andorphometric study were performed using a Sony 3CCD

olor video camera and processed using Visilog (Sony

SPAC, Barcelona, Spain), version 4.1.5 (Noesis, Saintubin, France), software.Labeled blocks from a representative transverse slice of

he anterior, lateral, and posterior free wall of the left ven-ricle, posterior free wall of the right ventricle, and anteriornd posterior interventricular septum and 1 block from eachtrium were taken for study of the myocardium. In addition,ny area with a significant macroscopic abnormality wasampled and analyzed using hematoxylin-eosin, Masson’srichrome, and Van Gieson stains.

Two investigators, using a double-headed light micro-cope, performed the analysis simultaneously.

From the blood samples, the total cholesterol, triglycer-des, high-density lipoprotein (HDL) cholesterol, low-den-ity lipoprotein (LDL) cholesterol, very-LDL cholesterol,hylomicrons, apolipoproteins B and CIII, and lipoproteina) were measured. The cotinine and glucose levels werebtained from the urine samples (Roche Diagnostics, Sanugat del Vallès, Barcelona, Spain and DRG Diagnostics,arburg, Germany).Descriptive analyses were initially performed. The quan-

itative variables are reported using the mean and standardeviation. Relations between categorical variables weretudied using the chi-square test. A comparison of the quan-itative variables between the 2 groups was performed usinghe t test and of ordinal variables using the Mann-Whitney

nonparametric U test.The variables analyzed univariately by logistic regres-

sion analysis to predict coronary artery disease and type ofatherosclerotic plaques included gender, age, body massindex (�24.9 vs 25 to 29.9 vs �30 kg/m2), presence orbsence of smoking, hypertension, diabetes, alcohol intake,hysical activity (yes vs no), diabetes (yes vs no) andDL/HDL cholesterol ratio (�3 vs �3).

Variables showing statistical significance (p �0.10) werencluded in the multivariate regression model to determinehich were independently related to the prognosis. In all

nalyses, contrasts were made bilaterally with an � of 5%(Statistical Package for Social Sciences, version 15.0, SPSS,

Table 1Pathologic findings associated with sudden death (SD) (n � 204)

Cardiovascular disease 183Heart disease 161

Coronary heart disease 119 (58%)Hypertensive left ventricular hypertrophy 20 (10%)Valvular heart disease 5 (2%)Idiopathic left ventricular hypertrophy 4 (2%)Dilated cardiomyopathy 4 (2%)Hypertrophic cardiomyopathy 3 (2%)Arrhythmogenic right ventricular

dysplasia/cardiomyopathy3 (2%)

Myocarditis 1 (1%)Congenital heart disease 1 (1%)Amyloidosis 1 (1%)

Vascular disease 22Pulmonary embolism 8 (4%)Aortic dissection 9 (4%)Cerebral hemorrhage 5 (2%)

Noncardiovascular disease 7

Chicago, Illinois).

6

8

t

sts

oanw

2i

dvw

uia(p(j

ihoshtic

o(p

herwise


Results

A total of 204 subjects (86% males), who had died fromout-of-hospital natural SD, were evaluated. Of the 204 sub-jects, 175 were males (86%) and 29 were females, with amean age of 54 � 15 years (males 53 � 15 years; females1 � 13 years; p � 0.014). Only 21 subjects were �35

years (range 12 to 34).The mean body mass index was 30 � 8 kg/m2 and was

greater for the females than for the males (34 � 10 vs 29 �kg/m2; p � 0.024). Of the 204 subjects, 58% (62% of

males and 27% of females) were smokers, 52% had a historyof regular alcohol intake, 39% had a history of hypertension,35% a history of dyslipemia, and 18% had a history of diabe-tes. The smokers were younger than the nonsmokers (52 � 13vs 57 � 18 years, respectively; p �0.001). Those with hyper-tension were older than those without (60 � 13 vs 49 � 16years, respectively; p �0.0001). The mean age of those with anLDL/HDL cholesterol ratio of �3 was 54 � 16 years and was55 � 13 years for those with an LDL/HDL cholesterol ratio�3 (p � 0.59). Of the subjects with diabetes mellitus, themean age was 61 � 12 years compared to 52 � 16 years forhose without diabetes mellitus (p �0.001).

The urine cotinine level was measured in 75% of theubjects. A high association (87%) was found between co-inine present in the urine and data on positive smokingtatus.

A history of cardiovascular symptoms was found in 33%f the subjects (60% females), including dyspnea, angina,nd/or syncope, but only 15% had been previously diag-osed with heart disease (15% males and 13% females),ith documented myocardial infarction in 10%.Of the 204 subjects, 14% had a family history of SD and

4% had a first-degree relative who had had myocardialnfarction.

Most deaths occurred while the subject was resting oroing mild exercise (71%), and, from the information pro-ided by the family and relatives, 20% of those whose death

Table 2Biologic risk factors for sudden death (SD) cases stratified by heart disea

Variable CHD and Non-CH

CHD(n � 119)

Non-CHD(n � 85)

Cholesterol (mmol/L) 6.5 � 2.1 5.2 � 1.8Triglycerides (mmol/L) 2.5 � 1.5 2.0 � 1.2High-density lipoprotein cholesterol (mmol/L) 1.0 � 0.4 1.2 � 0.5Low-density lipoprotein cholesterol (mmol/L) 3.9 � 1.4 3.2 � 1.3Low-density lipoprotein/high-density

lipoprotein cholesterol ratio4.3 � 2.2 3.1 � 1.8

Very-low-density lipoprotein cholesterol(mmol/L)

0.9 � 0.6 0.7 � 0.6

Chylomicrons (mmol/L) 1.7 � 1.1 1.1 � 0.5Lipoprotein (a) (g/L) 0.7 � 0.6 0.6 � 0.5Apolipoprotein B (g/L) 1.2 � 0.4 0.9 � 0.4Apolipoprotein CIII (g/L) 0.2 � 0.1 0.1 � 0.1Urine glucose (mmol/L) 24.1 � 64.9 10.7 � 44.5Urine cotinine (positive) 52% 43%

All parameters were measured from blood samples, unless indicated ot

as witnessed had complained of chest pain. d

Of the 204 cases, 183 (90%) could have been related tonderlying cardiovascular disease. Heart disease was foundn 161 subjects (79%), with CHD the most frequent (58%)nd significantly different (p � 0.016) between the males62%) and females (38%). Hypertensive cardiac hypertro-hy was found in 20 subjects (9.9%), aortic dissection in 94.4%), and pulmonary embolism in 8 (3.9%). In 14 sub-ects, the cause of death could not be ascertained (Table 1).

The mean heart weight was 498 � 123 g (454 � 115 gn females and 506 � 123 g in males; p � 0.038). Cardiacypertrophy was diagnosed in 41% of the females and 49%f the males (48% of those with SD). In the males, aignificant relation (p �0.0001) was found between cardiacypertrophy and a history of hypertension, with 75% ofhose with hypertension versus 34% of those without, hav-ng a hypertrophic heart. No relation was found betweenardiac hypertrophy and CHD.

On histologic study of the myocardium, a scar from anld, healed, myocardial infarction was found in 64 subjects31% of all those with SD) and was more frequent in maleatients �60 years old (47% vs 29%; p � 0.015). No

Figure 1. Culprit plaques. (A) Stable plaque. (B) Vulnerable plaque.(C) Eroded plaque. (D) Ruptured plaque. Macroscopic and microscopicimages. Hematoxylin-eosin stain, original magnification �60.

s and culprit plaque type

Culprit Plaque Type

ue Eroded and/or Ruptured(n � 49)

Stable (�75%) and/or Vulnerable(n � 70)

p Value

1 6.9 � 2.2 6.1 � 2.0 0.052.6 � 1.5 2.3 � 1.5 0.231.0 � 0.4 1.1 � 0.4 0.40

1 4.4 � 1.4 3.6 � 1.3 0.011 4.9 � 2.3 3.9 � 2.0 0.008

1.0 � 0.7 0.8 � 0.6 0.08

1.6 � 1.3 1.8 � 1.0 0.290.7 � 0.6 0.7 � 0.5 0.88

1 1.3 � 0.4 1.1 � 0.3 0.011 0.2 � 0.1 0.1 � 0.1 0.33

30.0 � 80.5 18.3 � 45.5 0.1471% 37% 0.001

.

se statu

D

p Val

�0.000.050.05

�0.00�0.00

0.01

0.270.61

�0.00�0.00

0.900.31

ifferences related to age were found in the female patients.

c

Acavava

nlasrnwdoc

ph9htIw4

Ci

na

MA


Finally, 46% of these cases did not meet the criteria (eroded orruptured plaque/luminal cross-sectional narrowing �75%) toonsider SD related to CHD.

The coronary arteries were evaluated in all SD subjects.single vessel (left anterior descending coronary artery,

ircumflex coronary artery, or right coronary artery) wasffected in 43 subjects (36%); 2 vessels in 48 (40%), and 3essels in 28 (24%). No females had the main left coronaryrtery affected, and in only 15 cases (14% of males) was thisessel affected. In 74% of the subjects with CHD, the leftnterior descending coronary artery was affected.

The biologic risk factors for SCD according to a diag-osis of CHD or non-CHD and the type of culprit plaque areisted in Table 2. Regarding coronary plaque morphology,n eroded (53%) and/or ruptured plaque (47%) was ob-erved as the culprit plaque in 49 subjects (41%). In theemaining (59%), a stable plaque with cross-sectional lumi-al narrowing of �75% was considered the culprit plaque,ith a vulnerable anatomy in 4 (Figure 1). An old myocar-ial infarction was present in 45% of those with an erodedr ruptured culprit plaque and in 60% of those with a stable

Table 3Uni- and multivariate logistic regression analysis of significant coronary h

Variable CHD(n � 119)

ale gender 91%ge (years) 57 � 12

Hypertension 45%Cigarette smokers 63%Body mass index (kg/m2)

�24.9 22%25–29.9 43%�30 35%

Low-density lipoprotein/high-density lipoproteincholesterol ratio �3

74%

Diabetes mellitus 20%Physical activity 44%Alcohol consumption 69%

Data are presented as % or mean � standard deviation.

Table 4Uni- and multivariate logistic regression analysis of culprit plaque type

Variable Eroded and/or Rupture(n � 49)

Male gender 94%Age (years) 54 � 12Hypertension 33%Cigarette smokers 76%Body mass index (kg/m2)

�24.9 21%25–29.9 45%�30 33%

Low-density lipoprotein/high-density lipoproteincholesterol ratio �3

87%

Diabetes mellitus 16%Physical activity 37%Alcohol consumption 72%

Data are presented as % or mean � standard deviation.

ulprit plaque. l

A relation was found between the type of culpritlaque and heart weight, with patients with stable plaquesaving a greater cardiac weight (528 � 119 g vs 479 �9 g; p � 0.025). The same relation was found when aealed myocardial infarction was present in either ofhese groups (564 � 122 g vs 491 � 107 g; p � 0.017).n hearts with acute thrombosis, the mean heart weightas 511 � 90 g for those with a ruptured plaque and51 � 100 g for those with an eroded plaque (p � 0.023).

An eroded or ruptured plaque was found in 33% of thosewith a history of hypertension versus 67% of those with-out hypertension (p � 0.04) and in only 13% of thegroup with cardiac hypertrophy and a history of hyper-tension.

The patients with CHD were significantly older thanthose without (57 � 12 vs 50 � 18 years; p �0.01). The

HD prevalence was significantly greater in the males thann the females (91% vs 9%; p �0.01).

Regarding biologic cardiovascular risk factors, no sig-ificant differences were found between SCD with CHDnd SCD without CHD, except for apolipoprotein B, apo-

sease (CHD)

Non-CHD(n � 85)

p Value


79% 0.018 0.01750 � 18 0.001 �0.001

32% 0.11 —50% 0.08 0.038

0.3330%28%42%44% �0.001 �0.001

14% 0.37 —40% 0.79 —64% 0.33 —

e Stable (�75%) and/or Vulnerable Plaque(n � 70)

p Value


89% 0.33 —59 � 12 0.04 —

53% 0.06 —54% 0.03 0.04

0.89 —23%42%36%63% 0.004 0.005

23% 0.38 —50% 0.23 —67% 0.37 —

eart di

d Plaqu

ipoprotein CIII, and the lipid profile. HDL cholesterol,

mtsHao(r

saawr

rct

amIcl

tN

aweww

ioWtf(Hgc

tpiplaoSp


LDL cholesterol, and very-LDL cholesterol were signifi-cantly greater in those with CHD (Table 2).

From the multivariate regression analysis, only male gen-der, older age, smoking, and an LDL/HDL cholesterol ratio of�3 were significantly associated with CHD (Table 3).

Those with an eroded and/or ruptured culprit plaquewere younger (54 � 12 vs 59 � 12 years; p � 0.04) and

ore likely to be smokers (76% vs 54%; p � 0.03) thanhose with a stable culprit plaque. Using univariate regres-ion analysis, younger age, smoking history, and an LDL/DL cholesterol ratio of �3 were associated with an eroded

nd/or ruptured culprit plaque (acute thrombosis). However,n multivariate regression analysis, only smoking historyodds ratio 2.5; p � 0.04) and an LDL/HDL cholesterolatio of �3 (odds ratio 4.2; p � 0.005) were associated with

these types of plaque (Table 4). Furthermore, using multi-variate regression analysis, no significant difference be-tween a ruptured or an eroded plaque and the cardiovascularrisk factors studied was found.

Discussion

In Anglo-Saxon countries, CHD has been the underlyingcause of SD in 80% to 90% of cases.19,20 The incidence ofSCD in Spain has been estimated to be one of the lowest inthe industrialized countries.8,21 However, the prevalence ofcardiovascular risk factors in the Mediterranean area22,23 isnot as low as one might expect. The results of the presentstudy have provided epidemiologic and, in particular, anato-mopathologic information on SD in Spain that might ex-plain the differences in SD between Anglo-Saxon and someMediterranean countries; these differences are probably re-lated to lifestyle and environment.24

As reported in other studies,25,26 a significant observa-tion was the high percentage of those who died from SDwho had a history of SD (14%) or myocardial infarction(24%) in first-degree relatives. Logically, this would sup-port the idea of a genetic factor involved in SD and, inparticular, CHD, as recently reported.27 A case-controltudy by Friedlander et al25 revealed that a family history ofcute myocardial infarction or SD was more commonmong those who had died from SD than in control subjects,ith this association mostly independent of other common

isk factors with familial aggregation.According to the information provided by the family and

elatives, only 20% of those who had died from SD hadomplained of chest pain, lower than the 37% reported inhe Maastrich study.28 This is consistent with the lower

incidence of underlying CHD in our study population. Itmight indicate that SD could be the first manifestation ofcardiovascular disease, as has been reported by other stud-ies,9,29 making it difficult to establish methods of preventingSD in the general population.

In the present study, 90% of cases were associated withcardiovascular disease, with CHD the most frequent. Com-pared to the findings from studies of Anglo-Saxon patients,we found a clearly lower incidence of CHD (58% vs 80% to90%9) and acute coronary thrombosis (41% vs 52%12).These findings support the lower incidence of acute coro-nary syndrome reported in the Mediterranean area, long
considered a consequence of diet30 and/or, in a broader
spect, the “Mediterranean culture.” However, the involve-ent of factors such as genetics should also be considered.

t is well known that when examining the same levels ofholesterol, the incidence of myocardial infarction has beenower in Spain than in Anglo-Saxon countries.22,31

In contrast to what has been reported in Anglo-Saxonpopulations, we found a greater percentage of cardiac hy-pertrophy without significant disarray (48% vs 13 to15%).32,33 In our male subjects, this was related to hyper-ension (p �0.001). The Massa Ventricolare Sinistraell’ipertensione Arteriosa study34 showed the strong, con-

tinuous, and independent relation between the left ventric-ular mass and subsequent cardiovascular morbidity, includ-ing SD.

Pathologic signs of CHD were found predominantly inour male subjects (62% vs 38%); however, a low number offemale subjects were included in the present study. In 59%of those with CHD, a stable plaque was considered theculprit plaque responsible for SD and a healed myocardialinfarction was found in 60% of these cases. This mightsuggest that if acute myocardial ischemia is a cause of SD,arrhythmia in the setting of myocardial scars could also bea very important component.

A relation was found between stable plaque and cardiachypertrophy, with patients with stable plaques having agreater cardiac mass (528 � 19 vs 479 � 99 g; p � 0.025).As with the findings from Burke et al,12,35 our study showed

lower frequency of acute coronary thrombosis in patientsith cardiac hypertrophy and a history of hypertension. An

roded or ruptured plaque was found in only 33% of thoseith a history of hypertension compared to 67% of thoseithout hypertension.Regarding the remaining coronary risk factors, a signif-

cant relation was found between CHD and male gender,lder age, smoking, and LDL/HDL cholesterol ratio of �3.hen we attempted to analyze its influence on the plaque

ype, patients with eroded and/or ruptured plaques wereound to be younger (p � 0.04), more likely to be smokersp � 0.03), and to have a greater probability of an LDL/DL cholesterol ratio of � 3 (p � 0.004). In contrast,ender, diabetes, regular alcohol intake, and hypertensionould not be related to the type of plaque.

The possible limitations of the present study includedhat, although it was performed prospectively, it was notossible to perform autopsy studies every day, rendering itmpossible to obtain exact information on the incidence andrevalence of SD in this population. Nevertheless, we be-ieve our findings can provide important epidemiologic andnatomopathologic information and offer us the possibilityf establishing comparative data with the data from Anglo-axon countries. Also, in accordance with previous re-orts,36 we found a lower prevalence of CHD in females

with SD (Table 3). Nevertheless, we must emphasize thelow number of females subjects included in our study.Finally, although we have previously justified the use ofheart weight instead of the left ventricular wall thickness asa variable, this could be considered a possible limitation.Therefore, we conducted a supplementary analysis to studythe relation between the thickness of the left ventricle andthe presence of coronary disease, without finding a relation
between these 2 variables. Moreover, a significant linear

a

AacaB

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3


correlation (r2 � 0.144) was seen between the heart weightnd ventricular wall thickness.

In 14 cases, no abnormal pathologic findings were found.t least some of these cases might correspond to an undi-

gnosed channelopathy. Even though minor structural myo-ardial abnormalities had been reported in some symptom-tic or asymptomatic patients with a channelopathy, such asrugada syndrome,37 usually no structural alterations will

be demonstrated by routine invasive and noninvasive exam-inations. We believe that 1 of our patients might havepresented with Brugada syndrome because in a pre-employ-ment medical examination, an atypical right bundle branchblock on the electrocardiogram was reported.

Acknowledgment: We thank A. Bartomeu, J. Guindo, andJ. Medallo for their excellent collaboration in this study.

1. Engelstein ED, Zipes DP. Sudden cardiac death. In: Alexander RW,Schlant RC, Fuster V, eds. The Heart, Arteries and Veins. 9th ed. NewYork: McGraw-Hill; 1998:1081–1112.

2. Myerburg RJ, Castellanos A. Cardiac arrest and sudden death. In:Braunwald E, ed. Heart Disease: A Textbook of Cardiovascular Med-icine. 5th ed. Philadelphia: WB Saunders; 1997:742–779.

3. Kuller L, Lilienfeld A, Fisher R. An epidemiological study of suddenand unexpected deaths in adults. Medicine (Baltimore) 1967;46:341–361.

4. Schatzkin A, Cupples LA, Heeren T, Morelock S, Kannel WB. Suddendeath in the Framingham Heart Study: differences in the incidence andrisk factors by sex and coronary disease status. Am J Epidemiol1984;120:888–899.

5. Demirovic J, Myerburg RJ. Epidemiology of sudden coronary death:an overview. Prog Cardiovasc Dis 1994;37:39–48.

6. Kannel WB, Schatzkin A. Sudden death: lessons from subsets inpopulation studies. J Am Coll Cardiol 1985;5:141B–149B.

7. Thiene G, Basso C, Corrado D. Cardiovascular causes of sudden death.In: Silver MD, Gotlieb AI, Schoen FJ, eds. Cardiovascular Pathology.3rd ed. Philadelphia: Churchill Livingstone; 2001:326–374.

8. Marrugat J, Elosua R, Gil M. Epidemiology of sudden cardiac death inSpain. Rev Esp Cardiol 1999;52:717–725.

9. Priori SG, Aliot E, Blomstrom-Lundqvist C, Bossaert L, Breithardt G,Brugada P, Camm AJ, Cappato R, Cobbe SM, Di Mario C, Maron BJ,McKenna WJ, Pedersen AK, Ravens U, Schwartz PJ, Trusz-Gluza M,Vardas P, Wellens HJ, Zipes DP. Task Force on Sudden Cardiac Deathof the European Society of Cardiology. Eur Heart J 2001;22:1374–1450.

0. Huikuri HV, Castellanos A, Myerburg RJ. Sudden death due to cardiacarrhythmias. N Engl J Med 2001;345:1473–1482.

1. Keys A. Coronary heart disease in seven countries. Circulation 1970;41(Suppl 1):1–211.

2. Burke AP, Farb A, Malcom GT, Liang Y-H, Smialek J, Virmani R.Coronary risk factors and plaque morphology in men with coronarydisease who died suddenly. N Engl J Med 1997;336:1276–1282.

3. Brinkmann B. Harmonisation of medico-legal autopsy rules. Int J LegMed 1999;113:1–14.

4. Basso C, Burke M, Fornes P, Gallagher PJ, de Gouveia RH, SheppardM, Thiene G, van der Wal A. Guidelines for autopsy investigation ofsudden cardiac death. Virchows Arch 2008;452:11–18.

5. Linzbach A. Heart failure from the point of view of quantitativeanatomy. Am J Cardiol 1960;5:370–382.

6. Grant RP. Aspects of cardiac hypertrophy. Am Heart J 1953;46:154–158.

7. Dobson A, Kuulasmaa K, Moltchanov V, Evans A, Fortmann SP,Jamrozik K, Sans S, Tuomilehto J. Changes in cigarette smokingamong adults in 35 populations in the mid 1980s: Who MONICAProject. Tob Control 1998;7:14–21.

8. Guidelines Committee. European Society of Hypertension-EuropeanSociety of Cardiology guidelines for the management of arterial hy-pertension. J Hypertens 2003;21:1011–1053.

9. Kuller LH. Sudden death definition and epidemiologic considerations.Prog Cardiovasc Dis 1980;23:1–12.

0. Myerburg RJ, Interian A Jr, Mitrani RM, Kessler KM, Castellanos A.Frequency of sudden cardiac death and profiles of risk. Am J Cardiol1997;80:10F–19F.

1. Grupo valenciano de Estudios sobre la Muerte Súbita. Muerte súbitaen la ciudad de Valencia. Rev Esp Cardiol 1987;40(Suppl):85–94.

2. Masiá R, Pena A, Marrugat J, Sala J, Vila J, Pavesi M, Covas M, AubóC, Elosua R. High prevalence of cardiovascular risk factors in Gerona,Spain, a province with low myocardial infarction incidence: REGICORInvestigators. J Epidemiol Commun Health 1998;52:707–715.

3. WHO Scientific Group. Technical Report Series 726. Sudden CardiacDeath. Geneve: World Health Organization: 1985:5–25.

4. Keys A, KeysM. In: How to Eat Well and Stay Well. The Mediterra-nean Way. Garden City, NY: Doubleday; 1975.

5. Friedlander Y, Siscovick DS, Weinmann S, Austin MA, Psaty BM,Lemaitre RN, Arbogast P, Raghunathan TE, Cobb LA. Family historyas a risk factor for primary cardiac arrest. Circulation 1998;97:155–160.

6. Jouven X, Desnos M, Guerot C, Ducimetiere P. Predicting suddendeath in the population: the Paris Prospective Study I. Circulation1999;99:1978–1983.

7. Myocardial Infarction Genetics Consortium. Genome-wide associationof early onset myocardial infarction with single nucleotide polymor-phisms and copy number variants. Nat Genet 2009;41:334–341.

8. de Vreede-Swagemakers JJ, Gorgels AP, Dubois-Arbouw WI, van ReeJW, Daemen MJ, Houben LG, Wellens HJ. Out-of-hospital cardiacarrests in the 1990s: a population based study in the Maastricht area onincidence, characteristics and survival. J Am Coll Cardiol 1997;30:1500–1505.

9. Lown B. Sudden cardiac death: the major challenge confronting con-temporary cardiology. Am J Cardiol 1979;43:313–328.

0. de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N.Mediterranean diet, traditional risk factors, and the rate of cardiovas-cular complications after myocardial infarction: final report of theLyon Diet Heart Study. Circulation 1999;99:779–785.

31. Kromhout D. On the waves of the Seven Countries Study: a publichealth perspective on cholesterol. Eur Heart J 1999;20:796–802.

32. Virmani R, Burke AP, Farb A. Sudden cardiac death. CardiovascPathol 2001;10:211–218.

33. Doolan A, Langlois N, Semsarian C. Causes of sudden cardiac deathin young Australians. Med J Aust 2004;180:110–112.

34. Verdecchia P, Carini G, Circo A, Dovellini E, Giovannini E, Lom-bardo M, Solinas P, Gorini M, Maggioni AP. Left ventricular mass andcardiovascular morbidity in essential hypertension: the MAVI study.J Am Coll Cardiol 2001;38:1829–1835.

35. Burke AP, Farb A, Liang Y, Smialek J, Virmani R. Effect of hyper-tension and cardiac hypertrophy on coronary artery morphology insudden cardiac death. Circulation 1996;94:3138–3145.

36. Albert CM, McGovern BA, Newell JB, Ruskin JN. Sex differences incardiac arrest survivors. Circulation 1996;93:1170–1176.

37. Frustaci A, Russo MA, Climenti C. Structural myocardial abnormal-
ities in asymptomatic family members with Brugada syndrome andSCN5A gene mutation. Eur Heart J 2009;30:11763.

Clinical and Prognostic Relevance of Echocardiographic Evaluationof Right Ventricular Geometry in Patients With Idiopathic

Pulmonary Arterial Hypertension

Stefano Ghio, MDa,*, Anna Sara Pazzano, MDa, Catherine Klersy, MDb, Laura Scelsi, MDa,Claudia Raineri, MDa, Rita Camporotondo, MDa, Andrea D’Armini, MDa, and

Luigi Oltrona Visconti, MDa

The aim of the present study was to assess the clinical and prognostic significance of rightventricular (RV) dilation and RV hypertrophy at echocardiography in patients withidiopathic pulmonary arterial hypertension. Echocardiography and right heart catheter-ization were performed in 72 consecutive patients with idiopathic pulmonary arterialhypertension admitted to our institution. The median follow-up period was 38 months. Thepatients were grouped according to the median value of RV wall thickness (6.6 mm) andthe median value of the RV diameter (36.5 mm). On multivariate analysis, the meanpulmonary artery pressure (p � 0.018) was the only independent predictor of RV wallthickness, and age (p � 0.011) and moderate to severe tricuspid regurgitation (p � 0.027)were the independent predictors of RV diameter. During follow-up, 22 patients died. Thedeath rate was greater in the patients with a RV diameter >36.5 mm than in patients witha RV diameter <36.5 mm: 15.9 (95% confidence interval 9.4 to 26.8) vs 6.6 (95% confidenceinterval 3.3 to 13.2) events per 100-person years (p � 0.0442). In contrast, the death ratewas similar in patients with RV wall thickness above or below the median value. However,among the patients with a RV wall thickness >6.6 mm, a RV diameter >36 mm was notassociated with a poorer prognosis (p � 0.6837). In conclusion, in patients with idiopathicpulmonary arterial hypertension, a larger RV diameter is a marker of a poor prognosis buta greater RV wall thickness reduces the risk of death associated with a dilated right
ventricle. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;107:628–632)
fanv

The aim of the present study was to assess the determi-nants and prognostic relevance of right ventricular (RV)dilation and RV hypertrophy at echocardiography in pa-tients with idiopathic pulmonary arterial hypertension(IPAH) and to test the hypothesis that a greater RV wallthickness is associated with better circulatory function and abetter prognosis in such patients.

Methods

From July 1996 to March 2009, 72 patients were con-secutively admitted to our institution for the evaluation ofchronic pulmonary hypertension and were diagnosed withIPAH. The diagnosis was made after having ruling out theknown causes of pulmonary hypertension.1 Patients withdifferent etiologies of pulmonary hypertension were ex-cluded. All patients underwent right heart catheterizationand ultrasound examination during the hospitalization pe-riod. The patients were followed up for a median of 38months (interquartiles range 14 to 71). During follow-up,

aCardiac, Thoracic, and Vascular Department and bBiometry and Clin-ical Epidemiology, Fondazione IRCCS Policlinico S. Matteo, Pavia, Italy.Manuscript received July 30, 2010; manuscript received and acceptedOctober 5, 2010.

*Corresponding author: Tel: (0039) 382-50-3713; fax: (0039) 382-50-3159.

E-mail address: [email protected] (S. Ghio).


the patients were treated according to international guide-lines. The enrollment of patients expanded over severalyears; thus, the treatments varied over time. However, theclinical decisions were never made on the basis of theechocardiographic parameters of RV function because thiswas never recommended by the guidelines.

The echocardiographic examinations were performed inthe same laboratory using commercially available ultra-sound equipments. The complete echocardiographic proto-col and intra- and interobserver agreement for the mostimportant echocardiographic parameters have been previ-ously reported.2 All echocardiographic data were averagedor 3 beats. The 2 parameters of interest in the presentnalysis (i.e., the RV end-diastolic diameter and the thick-ess of the RV free wall) were determined in the parasternaliew (Figure 1).3

A Swan Ganz thermodilution catheter (American Ed-wards Laboratories, Irvine, California) was inserted trans-cutaneously by way of the right internal jugular vein. Thethermistor was connected to a dedicated computer to displaythe cardiac output on-line. The following hemodynamicparameters were measured or calculated: systemic bloodpressure (arm cuff sphygmomanometer), right atrial pres-sure, systolic, diastolic, and mean pulmonary artery pres-sure, pulmonary wedge pressure, cardiac output, cardiacindex, systemic vascular resistance, and pulmonary vascularresistance. All the thermodilution measurements were ob-
tained in triplicate. In most cases, the ultrasound examina-
www.ajconline.org


etnycvradtscf

c1

R

T

RR

TR

T

M

IL

L

TCr

629Miscellaneous/Right Ventricular Geometry in Pulmonary Hypertension

tion and the right heart catheterization were performed onthe same day.

The data are shown as the mean � SD for the continuousvariables and as the absolute and relative frequencies for thecategorical variables. Pearson’s coefficient was used toevaluate the correlations between the continuous variables.

Figure 1. Parasternal long-axis view of patient with IPAH. Arrows indicatehow RV diameter and RV wall thickness were measured. Note, focus innear gain to better visualize RV wall.

Table 1Clinical, hemodynamic, and echocardiographic parameters according toright ventricular (RV) diameter

Variable RV Diameter (mm) pValue

�36.5(n � 36)

�36.5(n � 36)

Age (years) 50.3 � 16.8 53.1 � 16.6 0.483Gender 0.017

Women 31 21Men 5 15

Functional class III or IV 69% 83% 0.173Systolic pulmonary artery

pressure (mm Hg)76.1 � 22.5 84.0 � 19.3 0.116

Mean pulmonary artery pressure(mm Hg)

48.0 � 13.7 54.0 � 13.8 0.066

Pulmonary capillary wedgepressure (mm Hg)

9.5 � 4.4 10.3 � 4.0 0.453

Pulmonary vascular resistance(Wood Units)

10.7 � 4.6 14.3 � 7.2 0.014

Cardiac index (L/min/m2) 2.4 � 0.6 2.1 � 0.7 0.044ight atrial pressure (mm Hg) 6.2 � 3.4 9.0 � 5.3 0.010ight ventricular wall thickness

(mm)6.5 � 2.0 6.8 � 1.9 0.434

ranstricuspid gradient (mm Hg) 70.0 � 18.8 75.4 � 19.0 0.237ight ventricular fractional area

change28.7 � 10.3% 20.6 � 8.5% 0.001

ricuspid annular plane systolicexcursion (mm)

16.6 � 3.3 13.2 � 4.4 0.000

oderate/severe tricuspidregurgitation

49% 97% 0.000

nferior vena cava collapsibility 85% 33% 0.000eft ventricular end-diastolic

volume (ml)62 � 26 54 � 12 0.131

eft ventricular ejection fraction 61 � 7% 60 � 8% 0.410

The RV wall thickness and RV diameter were dichotomized O

on the basis of their median distribution and the between-group differences in the clinical, hemodynamic, and echo-cardiographic characteristics were compared. The meangroup values were compared using the 2-tailed t test or theMann-Whitney U test and proportions using the Fisherxact test. To identify the independent predictors of RV wallhickness and RV diameter, log-linear models were fitted;oncollinear variables showing p �0.2 on univariate anal-sis were included in the models. The relative risks and 95%onfidence intervals (CIs) were computed. Cumulative sur-ival was calculated using the Kaplan-Meier estimates. Theelative risk of dying and its 95% CI were computed using

Cox model. All survival models included the year ofiagnosis to account for the prolonged enrollment period ofhe study. Cardiac death was the only end point of theurvival analysis; lung transplantation was considered as aensored observation, and those patients were withdrawnrom the analysis at the intervention.

p Values �0.05 were retained for statistical signifi-ance. The computations were made using Stata, version1 (StataCorp, College Station, Texas).

esults

Of the 72 patients, 20 were men and 52 were women.heir mean age was 52 � 16 years. The World Health

able 2linical, hemodynamic, and echocardiographic parameters according to

ight ventricular (RV) hypertrophy

Variable RV Wall Thickness (mm) pValue

�6.6(n � 37)

�6.6(n � 35)

Age (years) 55.0 � 13.4 48.3 � 19.1 0.086Gender 0.116

Women 30 22Men 7 13

Functional class III or IV 76% 76% 1.000Systolic pulmonary artery pressure

(mm Hg)70.8 � 20.8 89.9 � 16.6 0.000

Mean pulmonary artery pressure(mm Hg)

46.0 � 12.7 56.3 � 13.1 0.001

Pulmonary capillary wedgepressure (mm Hg)

9.7 � 4.3 9.6 � 4.1 0.555

Pulmonary vascular resistance(Wood Units)

10.5 � 5.8 14.6 � 6.1 0.004

Cardiac index (L/min/m2) 2.4 � 0.7 2.1 � 0.6 0.031Right atrial pressure (mm Hg) 7.5 � 4.7 7.7 � 4.7 0.837Right ventricular diameter (mm) 35.6 � 6.7 40.2 � 9.9 0.022Transtricuspid gradient (mm Hg) 64.7 � 16.3 81.5 � 17.9 0.000Right ventricular fractional area

change26.3 � 9.3% 22.9 � 11.0% 0.188

Tricuspid annular plane systolicexcursion (mm)

15.1 � 4.3 14.7 � 4.2 0.702

Moderate/severe tricuspidregurgitation

67% 79% 0.293

Inferior vena cava collapsibility 66% 53% 0.328Left ventricular end-diastolic

volume (ml)60 � 26 56 � 15 0.470

Left ventricular ejection fraction 60 � 7% 61 � 9% 0.474

rganization functional class at referral was II for 17 pa-

mmppv2H

Ttdefhepaodrr0R

1aterwggcteapww

rt(�21pC�0woln(tiw

Tptcpw�pe6

D

rtlsp


tients and III-IV for 55 patients. The main echocardio-graphic and hemodynamic characteristics were as follows:RV diameter 37.8 � 8.7 mm; RV wall thickness 6.7 � 1.9

m; tricuspid annular plane systolic excursion 14.9 � 4.2m; moderate to severe tricuspid regurgitation 71.8% of

atients; left ventricular ejection fraction 60 � 8%; meanulmonary artery pressure 51.0 � 13.8 mm Hg; pulmonaryascular resistance 12.5 � 6.3 Wood Units; cardiac index.3 � 0.7 L/min/m2; and right atrial pressure 7.6 � 4.7 mmg.The median RV diameter was 36.5 mm (range 7 to 13).

he patients were accordingly divided into 2 groups: pa-ients with a RV diameter �36.5 mm and those with a RViameter of �36.5 mm. Their clinical, hemodynamic, andchocardiographic characteristics are listed in Table 1. Moreemale patients had a RV diameter of �36.5 mm. The righteart hemodynamic profile of the patients with a RV diam-ter greater than the median was characterized by greaterulmonary vascular resistance, greater right atrial pressure,nd a lower cardiac index. Greater RV dysfunction wasbserved at echocardiography in the patients with a RViameter �36.5 mm. On multivariate analysis, age (relativeisk 1.01, 95% CI 1.00 to 1.02, p � 0.011) and tricuspidegurgitation (relative risk 9.9, 95% CI 1.3 to 77.9, p �.027) were the independent variables associated with theV diameter.

The median RV wall thickness was 6.6 mm (range 3 to3), and the patients were divided into 2 groups: those withRV wall thickness �6.6 mm and those with a RV wall

hickness of �6.6 mm. Their clinical, hemodynamic, andchocardiographic characteristics are listed in Table 2. Theight heart hemodynamic profile of the patients with a RVall thickness greater than the median was characterized byreater systolic and mean pulmonary artery pressures,reater pulmonary vascular resistance, and a slightly lowerardiac index. The clinical and echocardiographic charac-eristics were substantially similar within the 2 groups,xcept for a slightly greater RV diameter in the patients withgreater RV wall thickness. The mean pulmonary artery

ressure (relative risk 1.02, 95% CI 1.00 to 1.04, p � 0.018)as the only variable independently associated with the RVall thickness on multivariate analysis.During the follow-up period, 22 patients died. The death

ate per 100 person-years was 6.6 (95% CI 3.3 to 13.2) forhe patients with a RV diameter of �36.5 mm and 15.995% CI 9.4 to 26.8) for the patients with a RV diameter36.5 mm. The enrollment-year adjusted hazard ratio was

.64 (95% CI 1.06 to 6.57; p � 0.036). The death rate per00 person-years was 11.2 (95% CI 6.3 to 19.7) for theatients with a RV wall thickness of �6.6 mm and 9.8 (95%I 5.3 to 18.3) for the patients with a RV wall thickness6.6 mm. The enrollment-year adjusted hazard ratio was

.88 (95% CI 0.38 to 2.08; p � 0.785). A clear interactionas seen between the RV wall thickness and RV diametern the prognosis. In patients with a RV wall thickness at oress than the median, dilation had a strong negative prog-ostic effect. The death rate per 100 person-years was 5.495% CI 2.0 to 14.5) for the patients with a RV wallhickness of �6.6 mm and RV diameter of �36.5 mm, butt was 23.5 (95% CI 11.8 to 47.1) for the patients with a RV
all thickness of �6.6 mm and RV diameter �36.5 mm. i
he adjusted hazard ratio was 4.23 (95% CI 1.21 to 14.82;� 0.024; Figure 2). In contrast, in patients with a RV wall

hickness greater than the median, dilation had no signifi-ant negative prognostic effect. The death rate per 100erson-years was 8.4 (95% CI 3.2 to 22.5) for the patientsith a RV wall thickness �6.6 mm and RV diameter of36.5 mm and was 11.1 (95% CI 5.0 to 24.6) for the

atients with a RV wall thickness �6.6 mm and RV diam-ter �36.5 mm. Adjusted hazard ratio 1.70 (95% CI 0.46 to.34; p � 0.428; Figure 2).

iscussion

The anatomic characteristics of the pressure overloadedight ventricle have been described using various imagingechniques.4–7 However, the lack of knowledge on the re-ation between RV structure and function and how suchtructural characteristics of the right ventricle affect therognosis in patients with pulmonary arterial hypertension

Figure 2. (A) Kaplan-Meier survival estimates for 36 patients with IPAHand RV wall thickness of �6.6 mm. Continuous line indicates patients withRV diameter �36.5 mm; dashed line, patients with RV diameter �36.5mm. End point of survival analysis was cardiac death. (B) Kaplan-Meiersurvival estimates for 36 patients with IPAH with RV wall thickness �6.6mm. Continuous line indicates patients with RV diameter of �36.5 mm;dashed line, patients with RV diameter �36.5 mm. End point of survivalanalysis was cardiac death.

s still substantial. The novelty of the present study was that

dsfffa

acssap

pRgsmttCr

r

631Miscellaneous/Right Ventricular Geometry in Pulmonary Hypertension

a simple echocardiographic evaluation of the RV geometryprovided relevant clinical and prognostic information forpatients with IPAH. Although both a larger RV diameterand a thicker RV wall at standard echocardiography havebeen associated with a more advanced right heart hemody-namic profile, RV dilation also predicted a poor prognosis.A greater RV wall thickness was not, per se, related to theprognosis, but it might reduce the risk of death associatedwith a dilated right ventricle.

In the present study, tricuspid regurgitation was the onlyhemodynamic determinant of RV dilation in patients withpressure overload of the right ventricle. This result can beexplained by the variable degree of tricuspid regurgitation,which is a common finding among patients with pulmonaryhypertension. Also, enlargement of the RV chamber, par-ticularly in the free wall to septum minor axis, rather than inthe long axis, is the main mechanism of adaptation tovolume overload.8,9 The prognostic relevance of RV dila-tion in patients with pulmonary hypertension is an issue thathas not yet received much attention in published studies,even though in the landmark epoprostenol study, continuousinfusion of prostacyclin for 12 weeks resulted in improvedsurvival and less RV dilation.10 The reason could be that theRV volumes can be accurately assessed only using cardiacmagnetic resonance (CMR).4,5 A study performed at a cen-ter with high CMR experience demonstrated that the RVend-diastolic volume is an independent determinant of prog-nosis in patients with IPAH. In particular, the investigatorsobserved that progressive RV dilation during treatment pre-dicts treatment failure and a poor long-term outcome.11

Such data from patients with IPAH parallel what has beenwell demonstrated in patients with heart failure (i.e.,changes in the left ventricular volumes more accuratelypredict the outcome than do baseline values).12 The evi-ence provided in the present study that the RV diameter attandard echocardiography is related to prognosis can there-ore be considered of clinical relevance and lays the basisor an easy assessment of the changes in RV dilation duringollow-up of patients with IPAH for centers without CMRvailable.

Cardiac hypertrophy is the mechanism of adaptation toltered mechanical or biochemical stimuli in which in-reased pressure overload plays a key role. It has beenuggested that a greater RV hypertrophic adaptation in re-ponse to the chronic increase in RV afterload might bessociated with a better hemodynamic profile and a betterrognosis in patients with IPAH.13,14 This hypothesis has

not received further attention, although a consensus existsthat a better adaptation of the right ventricle to the highafterload is one reason patients with pulmonary arterialhypertension associated with congenital heart disease have asubstantially better prognosis than those with IPAH.15 In theresent series of patients with IPAH, those with a thickerV free wall had an hemodynamic profile characterized byreater pulmonary pressure and vascular resistance and by aubstantially similar cardiac index compared to those withoderate hypertrophy, supporting the concept that the sus-

ained increase in RV wall stress was the main stimulusriggering hypertrophy. These data agree with previousMR studies demonstrating that the RV mass is directly

elated to pulmonary artery pressure and not as strongly

elated to mortality as RV dilation.16–19 Whichever theimaging technique, it must be noted that understanding towhat extent hypertrophy is adaptive or maladaptive is notpossible. The “protective” effects of myocyte hypertrophycould coexist to a variable extent with the “detrimental”effects of perivascular fibrosis of intramyocardial coronaryarteries and varying degrees of interstitial fibrosis, as well ascontractile dysfunction.20–22 Hypertrophy might also be as-sociated with subendocardial ischemia, although a greaterthickness of the RV wall was not necessarily associatedwith the presence or absence of RV perfusion defects atmyocardial scintigraphy.23 Little is known about the cellularand molecular mechanisms that underlie the transition fromcompensated hypertrophy to RV dilation and failure.24

In the present study, the patients with a RV free wallthickness greater than the median had a prognosis similar tothat of the patients with RV wall thickness at or less than themedian, despite having greater pulmonary pressure andgreater vascular resistance. This could imply that the effectsof hypertrophy on survival are not neutral but somehowbeneficial in patients with IPAH. However, the “protective”effects of hypertrophy can be inferred considering the in-teraction observed between the wall thickness and diameter.In patients with a RV wall thickness less than the median, aRV diameter greater than the median value increased by therisk of death �3-fold but in patients with thicker RV walls,a RV diameter greater than the median was not associatedwith a worse prognosis.

We acknowledge that it is necessary to verify to whatextent the evaluation of RV geometry provides significantadditive clinical information compared to the echocardio-graphic assessment of RV function, which has demonstratedprognostic relevance in patients with IPAH. However, thisshould be better explored in a multicenter setting and largerpopulations. In addition, only in a multicenter setting,would it be possible to integrate the echocardiographic,hemodynamic, and functional data of patients with IPAH.Another limitation of the present study was that the enroll-ment of patients occurred for several years, and the treat-ment of the patients changed during that period, possiblyaffecting RV function to a different extent. To avoid thislimitation, we used a statistical analysis that took into ac-count the time of diagnosis. Finally, we also acknowledgethat echocardiography is not the ideal imaging technique toassess the RV structure. The RV mass and volumes arecorrectly measured using CMR, a technique that could alsohelp in identifying fibrosis within the myocardial walls.25

However, the widespread availability of echocardiographyrepresents an unsurpassed advantage that should lead usto consider the 2 techniques complementary, rather thanalternative.

1. Galiè N, Hoeper MM, Humbert M, Torbicki A, Vachiery JL, BarberaJA, Beghetti M, Corris P, Gaine S, Gibbs JS, Gomez-Sanchez MA,Jondeau G, Klepetko W, Opitz C, Peacock A, Rubin L, Zellweger M,Simonneau G; Task Force for the Diagnosis and Treatment of Pulmo-nary Hypertension of the European Society of Cardiology (ESC) andthe European Respiratory Society (ERS), endorsed by the InternationalSociety of Heart and Lung Transplantation (ISHLT). Guidelines forthe diagnosis and treatment of pulmonary hypertension. Eur Heart J
2009;30:2493–2537.


2. Ghio S, Klersy C, Magrini G, D’Armini AM, Scelsi L, Raineri C,Pasotti M, Serio A, Campana C, Viganò M. Prognostic relevance ofthe echocardiographic assessment of right ventricular function in pa-tients with idiopathic pulmonary arterial hypertension. Int J Cardiol2010;140:272–278.

3. Baker BJ, Gammill J, Massengill J, Schubert E, Karin A, Doherty JE.Echocardiographic detection of right ventricular hypertrophy. AmHeart J 1983;105:611–614.

4. Boxt LM, Katz J, Kolb T, Czegledy FP, Barst RJ. Direct quantitationof right and left ventricular volumes with nuclear magnetic resonanceimaging in patients with primary pulmonary hypertension. J Am CollCardiol 1992;19:1508–1515.

5. Katz J, Whang J, Boxt LM, Barst RJ. Estimation of right ventricularmass in normal subjects and in patients with primary pulmonaryhypertension by nuclear magnetic resonance imaging. J Am Coll Car-diol 1993;21:1475–1481.

6. Hoeper MM, Tongers J, Leppert A, Baus S, Maier R, Lotz J. Evalu-ation of right ventricular performance with a right ventricular ejectionfraction thermodilution catheter and MRI in patients with pulmonaryhypertension. Chest 2001;120:502–507.

7. Raman SV, Shah M, McCarthy B, Garcia A, Ferketich AK. Multi-detector row cardiac computed tomography accurately quantifies rightand left ventricular size and function compared with cardiac magneticresonance. Am Heart J 2006;151:736–744.

8. Reynertson SI, Kundur R, Mullen GM, Costanzo MR, McKiernan TL,Louie EK. Asymmetry of right ventricular enlargement in response totricuspid regurgitation. Circulation 1999;100:465–467.

9. Mutlak D, Aronson D, Lessick J, Reisner SA, Dabbah S, Agmon Y.Functional tricuspid regurgitation in patients with pulmonary hyper-tension: is pulmonary artery pressure the only determinant of regur-gitation severity? Chest 2009;135:115–121.

10. Hinderliter AL, Willis PW IV, Barst RJ, Rich S, Rubin LJ, BadeschDB, Groves BM, McGoon MD, Tapson VF, Bourge RC, BrundageBH, Koerner SK, Langleben D, Keller CA, Murali S, Uretsky BF,Koch G, Li S, Clayton LM, Jöbsis MM, Blackburn SD Jr, Crow JW,Long WA; Primary Pulmonary Hypertension Study Group. Effects oflong-term infusion of prostacyclin (epoprostenol) on echocardio-graphic measures of right ventricular structure and function in primarypulmonary hypertension. Circulation 1997;95:1479–1486.

11. van Wolferen SA, Marcus JT, Boonstra A, Marques KM, BronzwaerJG, Spreeuwenberg MD, Postmus PE, Vonk-Noordegraaf A. Prognos-tic value of right ventricular mass, volume, and function in idiopathicpulmonary arterial hypertension. Eur Heart J 2007;28:1250–1257.

12. Cohn JN, Ferrari R, Sharpe N. Cardiac remodeling-concepts andclinical implications: a consensus paper from an international forum on

cardiac remodeling. On behalf of an International Forum on CardiacRemodeling. J Am Coll Cardiol 2000;35:569–582.

13. Bristow MR, Zisman LS, Lowes BD, Abraham WT, Badesch DB,Groves BM, Voelkel NF, Lynch DM, Quaife RA. The pressure-overloaded right ventricle in pulmonary hypertension. Chest 1998;114:101S–106S.

14. Abraham WT, Raynolds MV, Gottschall B, Badesch DB, Wynne KM,Groves BM, Lowes BD, Bristow MR, Perryman MB, Voelkel NF.Importance of angiotensin-converting enzyme in pulmonary hyperten-sion. Cardiology 1995;86(Suppl):9–15.

15. Hopkins WE, Ochoa LL, Richardson GW, Trulock EP. Comparison ofthe hemodynamics and survival of adults with severe primary pulmo-nary hypertension or Eisenmenger syndrome. J Heart Lung Transplant1996;15:100–105.

16. Quaife RA, Lynch D, Badesch DB, Voelkel NF, Lowes BD, RobertsonAD, Bristow MR. Right ventricular phenotypic characteristics in sub-jects with primary pulmonary hypertension or idiopathic dilated car-diomyopathy. J Card Fail 1999;5:46–54.

17. Saba TS, Foster J, Cockburn M, Cowan M, Peacock AJ. Ventricularmass index using magnetic resonance imaging accurately estimatespulmonary artery pressure. Eur Respir J 2002;20:1519–1524.

18. Roeleveld RJ, Marcus JT, Boonstra A, Postmus PE, Marques KM,Bronzwaer JG, Vonk-Noordegraaf A. A comparison of noninvasiveMRI based methods of estimating pulmonary artery pressure in pul-monary hypertension. J Magn Reson Imaging 2005;22:67–72.

19. Vonk-Noordegraaf A, Lankhaar J-W, Gotte MJV, Marcus JT, PostmusPE, Westerhof N. Magnetic resonance and nuclear imaging of the rightventricle in pulmonary arterial hypertension. Eur H J Suppl 2007;9(Suppl H):H29–H34.

20. Katz AM. Cardiomyopathy of overload. N Engl J Med 1990;322:100–110.

21. Weber KT, Janicki JS, Shroff SG, Pick R, Chen RM, Bashey RI.Collagen remodeling of the pressure-overloaded non human primatemyocardium. Circ Res 1988;62:757–765.

22. Weber KT, Pick R, Jalil JE, Janicki JS, Carroll EP. Pattern of myo-cardial fibrosis. J Mol Cell Cardiol 1989;21(Suppl 5):121–131.

23. Gomez A, Bialostozky D, Zajarias A, Santos E, Palomar A, MartínezML, Sandoval J. Right ventricular ischemia in patients with primarypulmonary hypertension. J Am Coll Cardiol 2001;38:1137–1142.

24. Bogaard HJ, Abe K, Vonk Noordegraaf A, Voelkel NF. The rightventricle under pressure: cellular and molecular mechanisms of right-heart failure in pulmonary hypertension. Chest 2009;135:794–804.

25. McCann GP, Gan CT, Beek AM, Niessen HW, Vonk Noordegraaf A,van Rossum AC. Extent of MRI delayed enhancement of myocardial
mass is related to right ventricular dysfunction in pulmonary arteryhypertension. AJR Am J Roentgenol 2007;188:349–355.

Clinically Significant Incidental Findings Among HumanImmunodeficiency Virus-Infected Men During Computed

Tomography for Determination of Coronary Artery Calcium

Nancy Crum-Cianflone, MD, MPHa,b,*, James Stepenosky, MDc, Sheila Medina, MPHa,b,Dylan Wessman, MDd, David Krause, MDd, and Gilbert Boswell, MDc

Those infected with the human immunodeficiency virus (HIV) have a greater risk ofcardiovascular disease and might undergo computed tomographic (CT) scans for earlydetection. Incidental findings on cardiac CT imaging are important components of thebenefits and costs of testing. We determined the prevalence and factors associated withincidental findings on CT scans performed to screen for coronary artery calcium (CAC)among HIV-infected men. A clinically significant finding was defined as requiring addi-tional workup or a medical referral. A total of 215 HIV-infected men were evaluated. Theirmedian age was 43 years; 17% were current tobacco users; the median CD4 count was 580cells/mm3; and 83% were receiving antiretroviral medications. Also, 34% had a positiveCAC score of >0. An incidental finding was noted among 93 participants (43%), with 36(17%) having >1 clinically significant finding. A total of 139 findings were noted, mostcommonly pulmonary nodules, followed by granulomas, scarring, and hilar adenopathy.Most of the incidental findings were stable on follow-up, and no malignancies weredetected. The factors associated with the presence of an incidental finding in the multi-variate model included increasing age (odds ratio 1.6 per 10 years, p <0.01), positive CACscore (odds ratio 2.3, p <0.01), and current tobacco use (odds ratio 2.5, p � 0.02). Inconclusion, incidental findings were common among HIV-infected men undergoing screen-ing CT imaging for CAC determination. The incidental findings were more commonamong older patients and those with detectable CAC. © 2011 Elsevier Inc. All rights
sscw

ifiwm

Although several studies have used coronary artery cal-cium (CAC) scores to measure subclinical heart disease inhuman immunodeficiency virus (HIV)-infected persons,1–3

no study to date has determined the frequency and types ofincidental abnormalities found during these tests. Studiesevaluating the incidental findings detected on coronarycomputed tomographic (CT) imaging (CAC and angiogra-phy) in the general population have been published but did

aInfectious Disease Clinical Research Program, Uniformed ServicesUniversity of the Health Sciences, Bethesda, Maryland; bInfectious Dis-ease Division and Departments of cRadiology and dCardiology, NavalMedical Center San Diego, San Diego, California. Manuscript receivedSeptember 8, 2010; manuscript received and accepted October 11, 2010.

Support for this study (Infectious Disease Clinical Research Program-018) was provided by the Infectious Disease Clinical Research Program, aDepartment of Defense program executed through the Uniformed ServicesUniversity of the Health Sciences. This project was funded in whole, or inpart, with federal funds from the National Institute of Allergy and Infec-tious Diseases, National Institutes of Health, Bethesda, Maryland, underInter-Agency Agreement Y1-AI-5072.

The content of this publication is the sole responsibility of the authorsand does not necessarily reflect the views or policies of the NationalInstitutes of Health or the Department of Health and Human Services,Department of Defense, or Departments of the Army, Navy, or Air Force.The mention of trade names, commercial products, or organizations doesnot imply endorsement by the United States Government.

sE-mail address: [email protected] (N. Crum-Cianflone).

not specifically evaluate HIV-infected subjects.4–18 HIV-infected persons might be at a greater risk of infectious andmalignant conditions,19,20 which could have implications forthe type and number of incidental findings on imaging studies.Incidental findings are important, because they not only affectthe clinical usefulness, but also the cost-effectiveness, ofscreening CT scans. We report the frequency and clinicalsignificance of incidental findings among HIV-infected per-sons undergoing CT screening for CAC determination.

Methods

We evaluated HIV-infected men who were screened forCAC using noncontrast CT imaging of the heart fromDecember 9, 2008 to March 1, 2010. The primary studyobjective was to determine the prevalence of subclinicalcoronary atherosclerosis using the Agatston scoring meth-od.21 The objectives of the present substudy were to de-cribe the prevalence and types of incidental findings on CTcans during standard imaging for CAC scores, and toompare the characteristics of HIV-infected men with andithout incidental findings.The inclusion criteria for study participation were a pos-

tive HIV test (enzyme-linked immunosorbent assay con-rmed by Western blot) and being a military beneficiary,hich included active duty members, retirees, and familyembers. All participants provided written informed con-

ent, and the governing institutional review board approved

www.ajconline.org


p

(Atiopcitepglfit

D

U


the study. The study was registered at ClinicalTrials.gov(study identifier NCT00889577).

The clinical data collected at the time of the CT scanincluded demographics, tobacco use history, body massindex, waist circumference, C-reactive protein (lower limitof detection �0.5 mg/dl; particle enhanced immunoturbidi-metric Assay, Roche, Indianapolis, Indiana), erythrocytesedimentation rate (modified Westergren method), CD4 cellcounts (flow cytometry), plasma HIV RNA levels (RocheAmplicor, undetectable at �50 copies/ml), and antiretrovi-ral therapy use. The diagnosis of hypercholesterolemia (to-tal cholesterol �200 mg/dl), hypertension, and diabetes(determined by the use of medications for these conditions)and a history of an opportunistic infection were recorded.

Participants underwent noncontrast CT imaging using amultidetector CT machine. Prospectively gated, axial,3-mm images were obtained at 120 kV on a Siemens Def-inition Dual Source CT scanner (Siemens Medical Solu-tions, Forsheim, Germany). The scanning protocol capturedimages with a 330-ms gantry rotation time, an individualdetector width of 0.6 mm with a reconstructed section widthof 3 mm and a temporal resolution of 165 ms. Contiguous3-mm-thick sections were reconstructed during peak inspi-ration with a 16- to 25-cm field of view, depending on theheart size. The images were processed on an Impax 6.3workstation (Agfa-Gevaert Group, Mortsel, Belgium). The

Table 1Study population characteristics stratified by incidental findings on cardia

Characteristic* Total Coho(n � 215)

Age (years) 43 (36–50)Ethnicity

White 110 (51%)Black 48 (22%)Other 57 (27%)

Tobacco useCurrent 37 (17%)Ever 109 (51%)Years of use† 12 (5–20)iabetes mellitus 13 (6%)

Hypertension 64 (30%)Hypercholesteremia‡ 66 (31%)Body mass index (kg/m2) 26.7 (24.1–29Waist circumference (cm) 94 (85–100C-reactive protein �0.5 mg/dl 30 (14%)Erythrocyte sedimentation rate (mm/hour) 10 (7–18)Erythrocyte sedimentation rate �20 mm/hour 47 (22%)Presence of coronary atherosclerosis (coronary

artery calcium �0)74 (34%)

Human immunodeficiency virus duration (years) 12 (5 to 19Current CD4 cell count (cells/mm3) 580 (386–72Nadir CD4 cell count (cells/mm3) 260 (138–36

ndetectable human immunodeficiency virusRNA level (�50 copies/ml)

150 (70%)

Current antiretroviral therapy 178 (83%)History of opportunistic infection 16 (7%)

*Categorical variables expressed as n (%) and continuous variables as† Among those with a history of tobacco use.‡ Hypercholesteremia defined as total cholesterol �200 mg/dl.

full field of view reconstructions were performed of the i

entire lungs at the levels from the carina to just below thecardiac apex. The CT images also included the mediasti-num, hilum, and diaphragm. The abdominal organs werenot visualized unless the diaphragm had been shifted cra-nially because of underlying pathology. CAC scoring wasperformed using an Aquarius workstation (TeraRecon, SanMateo, California) and calculated as the sum of all lesionsin each of the coronary arteries using Agatston units, aspreviously described.21 A CAC score of �0 was consideredositive for detectable calcium.

The images were read by a board-certified radiologistG.B.) for incidental noncoronary findings at CAC scoring.ny finding requiring additional clarification was reread by

he same radiologist. Incidental findings were captured onlyf the finding was not solely age-related (e.g., calcificationf the aorta), due to trauma (e.g., old rib fractures), orostoperative. The incidental findings were classified aslinically significant by the need for additional workup,ncluding imaging or medical referral. The need for addi-ional radiologic imaging was determined using availablevidence-based criteria; for example, the management ofulmonary nodules was determined using the Fleischneruidelines.22 A pulmonary nodule was defined as a �3-cmesion in the lung parenchyma and a granuloma as a calci-ed opacity. The patients with emphysema or bronchiec-

asis were recommended for referral to pulmonary med-

uted tomographic (CT) imaging

Incidental Findings p-value

Yes (n � 93) No (n � 122)

48 (39–52) 41 (34–47) �0.01

50 (54%) 60 (49%) 0.6921 (22%) 27 (22%)22 (24%) 35 (29%)

22 (23%) 15 (12%) 0.0454 (58%) 55 (45%) 0.0717 (8–26) 10 (5–18) �0.0110 (11%) 3 (2%) 0.0234 (37%) 30 (25%) 0.0727 (29%) 39 (32%) 0.66

26.8 (23.8–29.6) 26.5 (24.4–29.1) 0.8694 (87–103) 93 (85–98) 0.3114 (15%) 16 (13%) 0.7013 (9–23) 10 (7–15) �0.0126 (28%) 21 (17%) 0.0743 (46%) 31 (25%) �0.01

13 (7 to 21) 9 (5 to 19) 0.08563 (334–682) 600 (457–754) 0.02230 (100–360) 278 (184–367) 0.0659 (63%) 91 (75%) 0.10

74 (80%) 104 (85%) 0.289 (10%) 7 (6%) 0.30

(interquartile range).

c comp

rt

.5))

)9)6)

median

cine. The medical records of the participants with a

http://ClinicalTrials.gov

P

C

635Miscellaneous/Incidental Cardiac Imaging Findings in HIV

clinically significant incidental finding were reviewed inAugust 2010 for clinical outcomes, including data on

Table 2Number and type of incidental findings by location among 93HIV-infected men*

Finding IncidentalFindings (n)

Findings With ClinicalSignificance (n)

Lung 78 37Parenchymal/bronchi

Nodule 23 18Granuloma 19 0Scar 13 0Emphysema 9 9†

Bronchiectasis 6 6†

Parenchymal opacity/consolidation

4 3

Bronchial opacity 2 1Bullae 1 0

PleuraPleural thickening 1 0

Heart 4 2Cardiac valves with

significant calcification2 2

Papillary musclefat/calcification

2 0

Mediastinum 9 1Lymph nodes 7 0Mass 2 1

Hilum/subcarina 18 0Lymph node 17 0Granuloma 1 0

Pericardiac 1 1Lymph node 1 1

Other 29 5Gynecomastia 8 0Hiatal hernia 7 0Axillary adenopathy 7 0Aortic dilation/aneurysm 1 1Thymus mass 1 1Paralyzed hemidiaphragm 1 1Diaphragm eventration 1 0Pectus excavatum 1 0Enlarged pulmonary artery 1 1Large splenic cyst‡ 1 1

Total 139 46

* Some patients had multiple incidental findings.† Medical referral recommended.‡ Cyst was 15 � 13 cm in diameter.

Table 3Final multivariate model for factors associated with incidental findingson computed tomographic (CT) imaging for coronary artery calcium(CAC) determination

Factor OR (95% CI) p-value

Age (per 10 years) 1.6 (1.2–2.2) �0.01ositive coronary artery

calcium score*2.3 (1.3–4.3) �0.01

urrent Tobacco use 2.5 (1.3–4.3) 0.02

CI, confidence interval, OR, odds ratio.* Defined as CAC score of �0.

follow-up imaging.

The statistical analysis included descriptive statistics ofthe prevalence of incidental findings. Categorical variablesare described as numbers with proportions and continuousvariables as medians with interquartile ranges. The inciden-tal findings were divided into “clinically significant” and“not clinically significant,” as defined. The comparisons ofparticipants with and without incidental findings were per-formed using Fisher’s exact and rank-sum tests for categor-ical and continuous variables, respectively. A multivariatelogistic regression model was performed to evaluate thefactors associated with the incidental findings on cardiac CTscanning. The variables with p �0.10 on univariate testingwere placed in the full multivariate model, and a backwardstepwise approach was used to derive the final model.p-values �0.05 were considered statistically significant. Allanalyses were performed using Stata, version 10 (StataCorp,College Station, Texas).

Results

The study population consisted of 215 HIV-infected men(Table 1). An incidental finding was noted for 93 patients(43%), with 36 (17%) having a clinically significant findingthat required follow-up or medical referral. The range ofincidental findings per person was 0 to 5. Of those with anincidental finding, 62 patients had 1 finding, 19 had 2findings, 10 had 3 findings, 1 had 4 findings, and 1 had 5findings.

The number and type of incidental findings are listed inTable 2. A total of 139 incidental findings were noted, withsome participants having multiple findings. The most com-mon findings were pulmonary nodules, followed by granu-lomas and scars. A total of 15 patients (7% of the cohort)had findings consistent with emphysema or bronchiectasis.Several extrapulmonary findings were noted, including gy-necomastia, hiatal hernia, pathologic aortic dilation, and alarge (15 � 13 cm) asymptomatic splenic cyst.

Of the incidental findings, 46 (33%) were deemed clin-ically significant. Most often, these were pulmonary nodules(Table 2). On the follow-up evaluation at a median of 15months (range 6 to 21) after the initial CT scan for CACscoring, 15 of 18 of the pulmonary nodules were reimaged.Of the 15 pulmonary nodules, 13 were unchanged; 1 hadincreased in size (from 7 to 20 mm) but the workup findings,including bronchoscopy with cultures and cytology, havebeen negative; and 1 was determined not to be a nodule onreimaging. The remaining 3 participants with pulmonarynodules failed to undergo the scheduled follow-up imaging.The patients with pneumonia/opacities received antibiotics,and the cases had resolved or were improving for 4 of the 5who underwent repeat imaging. The subjects with emphy-sema or bronchiectasis had no changes in management,except for 1, for whom bronchodilators were prescribed.The aortic dilation detected on imaging remained stable.The single patient with the large splenic cyst underwentsplenectomy because of the high risk of rupture. Only 1participant died during the follow-up period; the cause ofdeath was not associated with the incidental finding (i.e.,nodule) on CT imaging.

The HIV-infected persons with an incidental finding
were older (48 vs 41 years, p �0.01), were current tobacco

olfi

r(rawg

D

iCfdpc

i

f4B

pslH

vut

fir


users (23% vs 12%, p � 0.04), were diabetic (11% vs 2%,p � 0.02), had a greater erythrocyte sedimentation rate (13vs 10 mm/hour, p �0.01), had a positive CAC score (46%vs 25%, p �0.01), and had a lower current CD4 cell count(563 vs 600 cells/mm3, p � 0.02) compared to those with-ut an incidental finding (Table 1). Of the tobacco users, aonger duration of use was also associated with an incidentalnding (p �0.01).

In the final multivariate model, increasing age (oddsatio 1.6 per 10 years, p �0.01), the presence of CACodds ratio 2.3, p �0.01), and current tobacco use (oddsatio 2.5, p � 0.02) were associated with the presence ofn incidental finding (Table 3). No HIV-specific factoras significant in the final model. The final model had aood fit (likelihood-ratio test, 26.7, p �0.01).

iscussion

The present study is the first to evaluate the prevalence ofncidental findings among HIV-infected persons undergoingT imaging for coronary atherosclerosis screening. We

ound that 43% of HIV-infected persons undergoing CACetermination had �1 incidental finding. In the generalopulation, 5 studies have evaluated the prevalence of in-idental findings on CT scans for CAC determination,7–11

with additional studies examining other types of CT cardiacimaging,5,6,12–18 and noted 8% to 53% had incidental find-ngs, with a mean prevalence of 38%.7,8,10,11 Our preva-

lence of incidental findings was similar to that in thesestudies and investigations using similar CT imaging tech-niques (41% to 56%).10,12 However, our study was per-ormed among patients substantially younger (median age3 years) than those in the general population (�60 years).ecause incidental findings increase with age,5,6 our data

suggest that HIV-infected persons might have a greaternumber of incidental findings. Additionally, our HIV-posi-tive cohort had a greater frequency of incidental findings thanHIV-negative military personnel (8%) of similar ages.11

The reason for the greater rate of incidental findingsamong HIV-infected persons might be because of theirgreater risk of pulmonary disease.19 Most incidental find-ings in our study were pulmonary nodules, granulomas,scarring, and bronchiectasis, all suggestive of previous re-spiratory infections. In addition, HIV-infected persons oftenhave greater rates of tobacco use compared to the generalpopulation, which might have contributed to an increasednumber of pulmonary findings.10,23 Finally, HIV-infectedersons might have unique incidental findings not typicallyeen in the general population, including gynecomastia,ikely the result of altered fat distribution patterns due to theIV and/or antiretroviral agents.24,25

In our study, 17% of patients with HIV had a clinicallysignificant finding that required follow-up or medical refer-ral. This compared with a range of 1% to 52% in the generalpopulation.4–6 The wide variation in rates is a result of thearying definitions and fields of view used for imaging. Wesed classifications within recent publications and multide-ector CT imaging to allow for comparability.4,10 Similar to

the general population, the most common significant find-
ings were pulmonary nodules.4,8–10
Several of our participants had incidental findings ofunquestionable clinical significance, including pneumoniatreated with antibiotics and a large splenic cyst. The dis-covery of such findings could result in added value ofscreening CT scans.4 In contrast, the detection of findingssuch as pulmonary nodules, which are often benign orresolve on follow-up imaging,5,17 could result in added

nancial costs, risks (including radiation exposure duringepeat imaging), and increased patient anxiety.5 Most of our

incidental findings were stable over time and did not resultin changes in management. Also, our study did not detectany undiagnosed cancers. As such, an assessment of thefrequency and medical significance of incidental findingsamong specific populations are important.

The factors associated with incidental findings includedincreasing age, current tobacco use, and the presence ofCAC; no HIV-specific factors were identified. Only 1 otherpublished study has evaluated the predictors of incidentalfindings.10 Age and tobacco use are known to be associatedwith pulmonary disease, including scarring and nodules. Ofparticular interest was our finding that the HIV-infectedparticipants with a positive CAC score had a �2-foldgreater frequency of incidental findings after adjusting forage and smoking. The precise relation between CAC andincidental findings is unknown. A possible explanation isthat the increased inflammation associated with previousinfections (which leads to incidental pulmonary findings)might contribute to the development of vascular disease.26

Additional studies are needed to evaluate this potentialassociation. Other explanations include the shared factorsbetween CAC and incidental findings, such as exposure orconcurrent conditions, which were unmeasured in our study.Overall, these data suggest that the benefit of identifyingcalcified coronary plaques should be balanced against thegreater number of incidental findings among these patients.

Our study had potential limitations. As noted in otherstudies, no standard definition for “clinically significant”incidental findings has been determined. Our study did usethe recent classifications4 and standardized criteria.22 Sec-ond, although our report has provided information on theoutcomes of the initial findings, follow-up is ongoing. Twoprevious studies with follow-up data showed that a singledeath in each study was attributed to an incidental findingon screening CT imaging; however, the duration of fol-low-up was limited.5,16 Third, we did not enroll an age-matched HIV-negative control group to determine whetherHIV-infected persons have a greater prevalence of inciden-tal findings but used historical information. Fourth, our HIVpatients had low overall rates of tobacco and illicit drug use;thus, we might have underestimated the prevalence of inci-dental findings compared to other HIV populations. Finally,because only men were evaluated in our study, future stud-ies examining HIV-infected women are advocated.

1. Mangili A, Gerrior J, Tang AM, O’Leary DH, Polak JK, Schaefer EJ,Gorbach SL, Wanke CA. Risk of cardiovascular disease in a cohort ofHIV-infected adults: a study using carotid intima-media thickness andcoronary artery calcium score. Clin Infect Dis 2006;43:1482–1489.

2. Talwani R, Falusi OM, Mendes de Leon CF, Nerad JL, Rich S, ProiaLA, Sha BE, Smith KY, Kessler HA. Electron beam computed tomog-
raphy for assessment of coronary artery disease in HIV-infected men

1

1

1

1

1

637Miscellaneous/Incidental Cardiac Imaging Findings in HIV

receiving antiretroviral therapy. J Acquir Immune Defic Syndr 2002;30:191–195.

3. Mangili A, Jacobson DL, Gerrior J, Polak JF, Gorbach SL, Wanke CA.Metabolic syndrome and subclinical atherosclerosis in patients in-fected with HIV. Clin Infect Dis 2007;44:1368–1374.

4. Jacobs PC, Mali WP, Grobbee DE, van der Graaf Y. Prevalence ofincidental findings in computed tomographic screening of the chest: asystematic review. J Comput Assist Tomogr 2008;32:214–221.

5. Machaalany J, Yam Y, Ruddy TD, Abraham A, Chen L, BeanlandsRS, Chow BJW. Potential clinical and economic consequences ofnoncardiac incidental findings on cardiac computed tomography. J AmColl Cardiol 2009;54:1533–1541.

6. Lee CI, Tsai EB, Sigal BM, Plevritis SK, Garber AM, Rubin GD.Incidental extracardiac findings at coronary CT: clinical and economicimpact. AJR Am J Roentgenol 2010;194:1531–1538.

7. Hunold P, Schmermund A, Seibel RM, Grönemeyer DH, Erbel R.Prevalence and clinical significance of accidental findings in electron-beam tomographic scans for coronary artery calcification. Eur Heart J2001;22:1748–1758.

8. Schragin JG, Weissfeld JL, Edmundowicz D, Strollo DC, FuhrmanCR. Non-cardiac findings on coronary electron beam computed to-mography scanning. J Thorac Imaging 2004;19:82–86.

9. Horton KM, Post WS, Blumenthal RS, Fishman EK. Prevalence ofsignificant noncardiac findings on electron-beam computed tomogra-phy coronary artery calcium screening examinations. Circulation2002;106:532–534.

0. Burt JR, Iribarren C, Fair JM, Norton LC, Mahbouba M, Rubin GD,Hlatky MA, Go AS, Fortmann SP; Atherosclerotic Disease, VascularFunction and Genetic Epidemiology (ADVANCE) study. Incidentalfindings on cardiac multidetector row computed tomography amonghealthy older adults: prevalence and clinical correlates. Arch InternMed 2008;168:756–761.

1. Elgin EE, O’Malley PG, Feuerstein I, Taylor AJ. Frequency andseverity of “incidentalomas” encountered during electron beam com-puted tomography for coronary calcium in middle-aged army person-nel. Am J Cardiol 2002;90:543–545.

2. Gil BN, Ran K, Tamar G, Shmuell F, Eli A. Prevalence of significantnoncardiac findings on coronary multidetector computed tomographyangiography in asymptomatic patients. J Comput Assist Tomogr 2007;31:1–4.

3. Onuma Y, Tanabe K, Nakazawa G, Aoki J, Nakajima H, Ibukuro K,Hara K. Noncardiac findings in cardiac imaging with multidetectorcomputed tomography. J Am Coll Cardiol 2006;48:402–406.

4. Haller S, Kaiser C, Buser P, Bongartz G, Bremerich J. Coronary artery
imaging with contrast-enhanced MDCT: extracardiac findings. AJRAm J Roentgenol 2006;187:105–110.
15. Koonce J, Schoepf JU, Nguyen SA, Northam MC, Ravenel JG. Extra-cardiac findings at cardiac CT: experience with 1,764 patients. EurRadiol 2009;19:570–576.

16. Law YM, Huang J, Chen K, Cheah FK, Chua T. Prevalence ofsignificant extracoronary findings on multislice CT coronary angiog-raphy examinations and coronary artery calcium scoring examinations.J Med Imaging Radiol Oncol 2008;52:49–56.

17. Iribarren C, Hlatky MA, Chandra M, Fair JM, Rubin GD, Go AS, BurtJR, Fortmann SP. Incidental pulmonary nodules on cardiac computedtomography: prognosis and use. Am J Med 2008;121:989–996.

18. Lehman SJ, Abbara S, Cury RC, Nagurney JT, Hsu J, Goela A, SchlettCL, Dodd JD, Brady TJ, Bamberg F, Hoffmann U. Significance ofcardiac computed tomography incidental findings in acute chest pain.Am J Med 2009;122:543–549.

19. Davis JL, Fei M, Huang L. Respiratory infection complicating HIVinfection. Curr Opin Infect Dis 2008;21:184–190.

20. Patel P, Hanson DL, Sullivan PS, Novak RM, Moorman AC, Tong TC,Holmberg SD, Brooks JT. Adult and Adolescent Spectrum of DiseaseProject and HIV Outpatient Study Investigators. Incidence of types ofcancer among HIV-infected persons compared with the general pop-ulation in the United States, 1992–2003. Ann Intern Med 2008;148:728–736.

21. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr,Detrano R. Quantification of coronary artery calcium using ultrafastcomputed tomography. J Am Coll Cardiol 1990;15:827–832.

22. MacMahon H, Austin JH, Gamsu G, Herold CJ, Jett JR, Naidich DP,Patz EF Jr, Swensen SJ; Fleischner Society. Guidelines for manage-ment of small pulmonary nodules detected on CT scans: a statementfrom the Fleischner Society. Radiology 2005;237:395–400.

23. Currier JS. Update on cardiovascular complications in HIV infection.Top HIV Med 2009;17:98–103.

24. Mira JA, Lozano F, Santos J, Ramayo E, Terrón A, Palacios R, LeónEM, Márquez M, Macías J, Fernández-Palacin A, Gómez-Mateos J,Pineda JA, Grupo Andaluz para el Estudio de las EnfermedadesInfecciosas. Gynaecomastia in HIV-infected men on highly activeantiretroviral therapy: association with efavirenz and didanosine treat-ment. Antivir Ther 2004;9:511–517.

25. Strub C, Kaufmann GR, Flepp M, Egger M, Kahlert C, Cavassini M,Battegay M; Swiss HIV Cohort Study. Gynecomastia and potentantiretroviral therapy. AIDS 2004;18:1347–1349.

26. Ross AC, Rizk N, O’Riordan MA, Dogra V, El-Bejjani D, Storer N,Harrill D, Tungsiripat M, Adell J, McComsey GA. Relationship be-tween inflammatory markers, endothelial activation markers, and ca-
rotid intima-media thickness in HIV-infected patients receiving anti-retroviral therapy. Clin Infect Dis 2009;49:1119–1127.

tt

0d

Self-Terminated Ventricular Fibrillation and Recurrent Syncope

Yuval Konstantino, MD*, Angela Morello, MD, Peter J. Zimetbaum, MD, andMark E. Josephson, MD

Ventricular fibrillation (VF) is a lethal arrhythmia that requires immediate cardioversionand is rarely self-terminating. Spontaneous termination is typically associated with moreorganized activation than sustained VF terminated by shock, but the precise mechanism isunclear. In the present case, we describe a patient with recurrent syncope and documentedself-terminating VF, who ultimately underwent implantable cardioverter defibrillator in-sertion. Assessment of the rhythm strip revealed organization of a chaotic rhythm intomonomorphic ventricular tachycardia before termination, in supportive of previous re-ports. In conclusion, self-terminating VF is a very rare condition that can cause
syncope. © 2011 Published by Elsevier Inc. (Am J Cardiol 2011;107:638–640)
badwmA

C

rs

The cardiac causes of syncope, primarily bradyarrhyth-mias and tachyarrhythmias, account for 10% to 20% ofsyncopal episodes.1 Of the tachyarrhythmias, ventricularachycardia (VT) and, less commonly, supraventricularachycardia can cause syncope. In contrast, ventricular fi-

Cardiovascular Institute, Beth Israel Deaconess Medical Center, Bos-ton, Massachusetts. Manuscript received August 22, 2010; manuscriptreceived and accepted October 1, 2010.

*Corresponding author: Tel: (617) 632-27713; fax: (617) 632-7377.E-mail address: [email protected] (Y. Konstantino).

Figure 1. ILR tracing demonstrating VF that organized into monomorphic

002-9149/11/$ – see front matter © 2011 Published by Elsevier Inc.oi:10.1016/j.amjcard.2010.10.025

rillation (VF) is an extremely rare cause of syncope. VF islife-threatening arrhythmia that requires immediate car-

ioversion and is rarely self-terminating. In the present case,e describe a patient with recurrent syncope and self-ter-inating VF detected by implantable loop recorder (ILR).n implantable cardioverter defibrillator was inserted.

ase Report

A 68-year-old man was admitted to our hospital withecurrent syncope. Previously, he had had syncope afteritting up in bed. At that time, he had a prodrome of nausea,

VT, followed by spontaneous termination and sinus bradycardia.

www.ajconline.org


C, as re

639Case Report/Ventricular Fibrillation and Recurrent Syncope

flushing, and dizziness. The cardiac evaluation included anelectrocardiogram with normal findings, an echocardiogramthat revealed normal biventricular function and no wallmotion abnormality, and an imaging stress test that demon-strated a fixed infero-posterolateral defect. Angiographyrevealed 100% occlusion of the right coronary artery and a50% narrowing of the obtuse marginal artery.

One year later, he lost consciousness at rest, without any

Figure 2. VF initiated by early coupling VP

preceding symptoms. Cardiopulmonary resuscitation was

initiated immediately. On medical team arrival, he wasawake and free of symptoms. His vital signs were normal.The electrocardiogram revealed sinus rhythm without evi-dence of previous infarction, acute ischemia, or Brugada,long QT, or short QT syndromes. The electrolytes andcardiac biomarker levels were normal. The echocardiogramdisplayed normal biventricular function with no wall motionabnormality. An electrophysiologic study was performed,

corded on telemetry during hospitalization.

revealing a normal voltage map, without evidence of an

ytsptpeco

kdQ

Vsbmdit

3

4

5

6


endocardial scar. Ventricular arrhythmia was not induced,with extra stimuli delivered from the right ventricular apexand the right ventricular outflow tract, with and withoutisoproterenol. Sinus node function and atrioventricular con-duction were normal. Given recurrent unexplained syncope,an ILR was placed.

Two weeks later, a third syncopal episode occurred. TheILR tracing demonstrated VF, which organized into mono-morphic VT and spontaneously terminated (Figure 1). Dur-ing hospitalization, another syncopal episode occurred, withno preceding symptoms. VF was documented (Figure 2),and brief cardiopulmonary resuscitation was initiated withspontaneous VF termination. Cardioversion or medicationswere not required. Assessment of the initiating ventricularpremature complex (VPC) revealed a Q wave in lead II,suggestive of a VPC originating from an inferior scar. Animplantable cardioverter defibrillator was inserted.

Discussion

In 3 studies that analyzed data from ILRs from a total of365 patients with recurrent syncope, arrhythmia was docu-mented in 20% to 25%. Bradycardia was seen in 15% to19% and tachyarrhythmia in 4% to 6% (VT in 1 to 3% andVF in none).2–4 Yu et al5 described a patient with recurrentsyncope diagnosed as recurrent VF from the ILR findingsThe VF was eliminated by radiofrequency catheter ablationof the triggering VPCs.

Leenhardt et al6 had previously reported a group ofoung patients at high risk of sudden death, with no struc-ural heart disease and normal QT who presented withyncope related to a short-coupled variant of torsade deointes. The initiating VPC had a very short coupling in-erval, and the arrhythmia deteriorated into VF in most ofatients soon after its onset. This was unlikely to be thetiology in our patient, given his older age, history oforonary artery disease, and a much longer coupling interval
f the initiating VPC.
The precise mechanism of self-termination of VF is un-nown. Self-terminating polymorphic VT has been previouslyescribed in patients with coronary artery disease and a normalT interval undergoing an electrophysiologic study,7 but it is

extremely uncommon. Spontaneous reversion to sinus rhythmtypically occurs when the localized chaotic activity is confinedto small areas of the heart and gradually becomes regular.7

Mäkikallio et al8 have found that spontaneously terminatingF displays more organized local activation dynamics than

ustained VF terminated by shock, suggesting that the dynamicehavior of local cardiac activation might be related to theaintenance of ventricular tachyarrhythmias. The present case

emonstrated the organization of the cardiac electrocardiogramnto monomorphic VT, before VF termination, in support ofhe previous reports.

1. Calkins H, Zipes DP. Hypotension and syncope. In: Zipes DP, Braun-wald E, eds. Braunwald’s Heart Disease: A Textbook of CardiovascularMedicine, 7th ed. Philadelphia: Elsevier Saunders; 2005:912.

2. Entem FR, Enriquez SG, Cobo M, Expósito V, Llano M, Ruiz M, JoseOlalla J, Otero-Fernandez M. Utility of implantable loop recorders fordiagnosing unexplained syncope in clinical practice. Clin Cardiol 2009;32:28–31.

. Seidl K, Rameken M, Breunung S, Senges J, Jung W, Andresen D, vanToor A, Krahn AD, Klein GJ. Reveal-investigators. Diagnostic assess-ment of recurrent unexplained syncope with a new subcutaneouslyimplantable loop recorder. Europace 2000;2:256–262.

. Farwell DJ, Freemantle N, Sulke N. The clinical impact of implantableloop recorders in patients with syncope. Eur Heart J 2006;27:351–356.

. Yu CC, Tsai CT, Lai LP, Lin JL. Successful radiofrequency catheterablation of idiopathic ventricular fibrillation presented as recurrent syn-cope and diagnosed by an implanted loop recorder. Int J Cardiol2006;110:112–113.

. Leenhardt A, Glaser E, Burguera M, Nürnberg M, Maison-Blanche P,Coumel P. Short-coupled variant of torsade de pointes: a new electro-cardiographic entity in the spectrum of idiopathic ventricular tachyar-rhythmias. Circulation 1994;89:206–215.

7. Josephson ME. Clinical Cardiac Electrophysiology, 4th ed. Philadel-phia: Lippincott Williams & Wilkins; 2008:516–524.

8. Mäkikallio TH, Huikuri HV, Myerburg RJ, Seppänen T, Kloosterman M,Interian A Jr, Castellanos A, Mitrani RD. Differences in the activation
patterns between sustained and self-terminating episodes of human ven-tricular fibrillation. Ann Med 2002;34:130–135.

stcssAtwroosbwtRcbppmtecHldDhcaoppii

to1mtawtRSO

READERS’ COMMENTS

dwb

tivple

ttintpp2obnbs

Comparison of 600 Versus 300-mgClopidogrel Loading Dose inPatients With ST-Segment ElevationMyocardial Infarction UndergoingPrimary Coronary Angioplasty

The rationale for the use of a doubledose of clopidogrel in the study ofMangiacapra et al1 for those with ST-egment elevation myocardial infarc-ion (STEMI) who underwent primaryoronary angioplasty was to rapidlyuppress platelet activity to avoid sub-equent cardiovascular ischemic events.s is known, other adjuvant antiplatelet

herapy trials for patients with STEMIho undergo primary angioplasty have

evealed similar results, with reductionsf cardiovascular end points and eventsf acute and subacute instant thrombo-is. However, additional benefits haveeen emphasized, especially for thoseith high risk profiles. For example, in

he Korean Acute Myocardial Infarctionegistry, Chen et al2 pointed out thatilostazol-based triple therapy shoulde used for elderly, female, and diabeticatients with STEMI for primary angio-lasty. De Luca et al3 performed aeta-regression analysis of randomized

rials and demonstrated a mortality ben-fit proportional to baseline risk in ab-iximab-adjuvant triple therapy. In thearmonizing Outcomes With Revascu-

arization and Stents in Acute Myocar-ial Infarction (HORIZONS-AMI) trial,angas et al4 reported that Killip class Iad the lowest hazard ratio (0.36), and alopidogrel loading dose of 600 mg hadhazard ratio of 0.67 for the occurrencef cardiovascular events at 30 days afterercutaneous coronary intervention, im-lying that the benefits of a higher load-ng dose of clopidogrel will be bluntedn the condition of Killip class I.5

Recently, Pocock et al6 reported thathe highest death rate from ischemiaccurred on days 0 and 1 (hazard ratio5.57), but the highest death rate fromajor bleeding occurred during days 8

o 30 (hazard ratio 4.80) after primaryngioplasty. This finding is consistentith the results of the Clopidogrel Op-

imal Loading Dose Usage to Reduceecurrent Events/Optimal Antiplatelettrategy for Interventions (CURRENT-
ASIS 7) trial that clopidogrel 600 mg/
Am J Cardiol 2011;107:6410002-9149/11/$ – see front matter © 2011 Elsevier I

ay can be safely used for 1 weekithout increasing the rate of majorleeding.7 Accordingly, the use of a

600-mg loading dose of clopidogrelshould be recommended in patientswith STEMI with high risk profiles forprimary angioplasty. How to identifynonresponders to clopidogrel will be thenext major issue for those who require aloading dose of clopidogrel �600 mgfor primary angioplasty.

Gen-Min Lin, MD

Yi-Hwei Li, PhD

Hualien, Taiwan8 November 2010

1. Mangiacapra F, Muller O, Ntalianis A, TranaC, Heyndrickx GR, Bartunek J, VanderheydenM, Wijns W, De Bruyne B, Barbato E. Com-parison of 600 versus 300-mg clopidogrelloading dose in patients with ST-segment ele-vation myocardial infarction undergoing pri-mary coronary angioplasty. Am J Cardiol2010;106:1208–1211.

2. Chen KY, Rha SW, Li YJ, Poddar KL, Jin Z,Minami Y, Wang L, Kim EJ, Park CG, Seo HS,Oh DJ, Jeong MH, Ahn YK, Hong TJ, Kim YJ,Hur SH, Seong IW, Chae JK, Cho MC, Bae JH,Choi DH, Jang YS, Chae IH, Kim CJ, Yoon JH,Chung WS, Seung KB, Park SJ; Korea AcuteMyocardial Infarction Registry Investigators.Triple versus dual antiplatelet therapy in patientswith acute ST-segment elevation myocardial in-farction undergoing primary percutaneous coro-nary intervention. Circulation 2009;119:3207–3214.

3. De Luca G, Suryapranata H, Stone GW, An-toniucci D, Tcheng JE, Neumann FJ, Boniz-zoni E, Topol EJ, Chiariello M. Relationshipbetween patient’s risk profile and benefits inmortality from adjunctive abciximab to me-chanical revascularization for ST-segment el-evation myocardial infarction: a meta-regres-sion analysis of randomized trials. J Am CollCardiol 2006;47:685–686.

4. Dangas G, Mehran R, Guagliumi G, CaixetaA, Witzenbichler B, Aoki J, Peruga JZ, BrodieBR, Dudek D, Kornowski R, Rabbani LE,Parise H, Stone GW; HORIZONS-AMI TrialInvestigators. Role of clopidogrel loading dosein patients with ST-segment elevation myocar-dial infarction undergoing primary angio-plasty: results from the HORIZONS-AMI(Harmonizing Outcomes With Revasculariza-tion and Stents in Acute Myocardial Infarc-tion) trial. J Am Coll Cardiol 2009;54:1438–1446.

5. Lin GM, Han CL. The loading dose of clopi-dogrel in patients with ST-segment elevationmyocardial infarction undergoing primary an-gioplasty. Am J Emerg Med 2010;28:382–383.

6. Pocock SJ, Mehran R, Clayton TC, NikolskyE, Parise H, Fahy M, Lansky AJ, Bertrand ME,Lincoff AM, Moses JW, Ohman EM, White
HD, Stone GW. Prognostic modeling of indi- c
nc. All rights reserved.

vidual patient risk and mortality impact ofischemic and hemorrhagic complications:assessment from the Acute Catheterization andUrgent Intervention Triage Strategy trial.Circulation 2010;121:43–51.

7. Mehta SR, Tanguay JF, Eikelboom JW, Jolly SS,Joyner CD, Granger CB, Faxon DP, RupprechtHJ, Budaj A, Avezum A, Widimsky P, Steg PG,Bassand JP, Montalescot G, Macaya C, Di Pas-quale G, Niemela K, Ajani AE, White HD, Chro-lavicius S, Gao P, Fox KA, Yusuf S; on behalf ofthe CURRENT-OASIS 7 Trial Investigators.Double-dose versus standard-dose clopidogreland high-dose versus low-dose aspirin in in-dividuals undergoing percutaneous coronary1intervention for acute coronary syndromes(CURRENT-OASIS 7): a randomised facto-rial trial. Lancet 2010;376:1233–1243.

doi:10.1016/j.amjcard.2010.11.001

Long-Term Follow Up ofAtrioventricular Block inTranscatheter Aortic ValveImplantation

We have read with great interest thereport by Roten et al1 regarding predic-ors of atrioventricular (AV) conductionmpairment after transcatheter aorticalve implantation with the CoreValverosthesis (Medtronic, Inc., Minneapo-is, Minnesota). We have published ourarly experience with this prosthesis2

and recently performed electrophysi-ologic studies in some pacemaker-freepatients immediately before and aftervalve implantation. An electrode wasplaced on the His bundle during valveimplantation, and data were continu-ously recorded during the procedure.We agree that the type of AV block isintra- or infrahisian; in fact, our grouphas published for the first time a report ofintrahisian AV block in a patient whounderwent percutaneous CoreValve pros-thesis implantation.3 However, we are ex-remely surprised by the high AV conduc-ion recovery rate, taking in account thatnfrahisian blocks are usually perma-ent. In our series with the first 50 pa-ients who underwent percutaneous im-lantation of the CoreValve prosthesis,acemaker implantation was needed in2 patients (44%), 20 patients becausef complete AV block and 2 patientsecause of first-degree AV block withewly developed left bundle branchlock. Of the 20 patients, considering aubgroup of 15 patients who were dis-
harged alive from the hospital and af-
www.ajconline.org

etpwCedA

ssiAA


ter a median follow-up period of 435days, only 2 patients recovered cardiacrhythm, and the remaining patients hadescape rhythm, with a mean heart rateof 28 beats/min. Compared with Rotenet al’s1 study, these differences may bexplained in part by technical aspects ofhe procedure and differences in therostheses implanted, as in our hospital,e implant only CoreValve prostheses.onsidering our data and those of Rotent al1 (median follow-up period 79ays) together, we can speculate that
V conduction impairment seems to be
lowly progressive over time, whichupports early pacemaker implantationn patients who develop second or thirdV block despite detecting recovery ofV block.

José Rubín, MD, PhD

Pablo Avanzas, MD, PhD

Raquel del Valle, MD

Cesar Morís, MD, PhD

Oviedo, Spain10 November 2010

1. Roten L, Wenaweser P, Delacrétaz E, HelligeG, Stortecky S, Tanner H, Pilgrim T, Kadner
A, Eberle B, Zwahlen M, Carrel T, Meier B,
Windecker S. Incidence and predictors ofatrioventricular conduction impairment aftertranscatheter aortic valve implantation. Am JCardiol 2010;106:1473–1480.

2. Avanzas P, Muñoz-García AJ, Segura J, PanM, Alonso-Briales JH, Lozano I, Morís C,Suárez de Lezo J, Hernández-García JM. Per-cutaneous implantation of the CoreValve self-expanding aortic valve prosthesis in patientswith severe aortic stenosis: early experience inSpain. Rev Esp Cardiol 2010;63:141–148.

3. Rubín J, Avanzas P, Calvo D, Moris C. Intra-hisian block during transcatheter aortic valveimplantation with the CoreValve prosthesis.Rev Esp Cardiol. In press.

doi:10.1016/j.amjcard.2010.11.013

the american journal of cardiology vol 107 issue 4 february 2011

Documents

john mancinifrancis

david hillisdavid

david gertzmihai gheorghiaded

andersonevan appelbaumrichard

dimarcoallen dollarmichael

shahjamshid shiranirobert

vogelron waksmandavid

kostischarles landaurichard