aortic root replacement with cryopreserved allograft for prosthetic valve endocarditis

2002;74:650-659 Ann Thorac SurgB. Pettersson and Delos M. Cosgrove

Joseph F. Sabik, Bruce W. Lytle, Eugene H. Blackstone, Antonino G.M. Marullo, Gosta endocarditis

Aortic root replacement with cryopreserved allograft for prosthetic valve

http://ats.ctsnetjournals.org/cgi/content/full/74/3/650on the World Wide Web at:

The online version of this article, along with updated information and services, is located

Print ISSN: 0003-4975; eISSN: 1552-6259. Southern Thoracic Surgical Association. Copyright © 2002 by The Society of Thoracic Surgeons.

is the official journal of The Society of Thoracic Surgeons and theThe Annals of Thoracic Surgery

by Michael Mack on September 9, 2009 ats.ctsnetjournals.orgDownloaded from

http://ats.ctsnetjournals.org/cgi/content/full/74/3/650

http://ats.ctsnetjournals.org

ORIGINAL ARTICLES: CARDIOVASCULAR

Aortic Root Replacement With CryopreservedAllograft for Prosthetic Valve EndocarditisJoseph F. Sabik, MD, Bruce W. Lytle, MD, Eugene H. Blackstone, MD,Antonino G. M. Marullo, MD, Gosta B. Pettersson, MD, and Delos M. Cosgrove, MDDepartments of Thoracic and Cardiovascular Surgery and Biostatistics and Epidemiology, The Cleveland Clinic Foundation,Cleveland, Ohio

Background. Our strategy has been to treat aortic pros-thetic valve endocarditis (PVE) with radical debridementof infected tissue and aortic root replacement with acryopreserved aortic allograft. This study examines theeffectiveness of this strategy on hospital mortality andmorbidity, recurrent endocarditis, and survival.

Methods. From 1988 through 2000, 103 patients withaortic PVE underwent root replacement with a cryopre-served aortic allograft. Abscesses were present in 78%,and aortoventricular discontinuity was present in 40%.Thirty-two patients had at least one previous operationfor endocarditis. In 23 patients with a history of nativevalve endocarditis, the allograft was implanted after oneepisode (17 patients), two episodes (5 patients), or threeepisodes of PVE (1 patient). In the 80 patients without ahistory of native valve endocarditis, the allograft wasplaced after one previous aortic valve replacement (57patients), two (19), or three (4) previous aortic valve

replacements. Among the 92 patients with positive cul-tures, 52 had staphylococcal organisms, 20 had strepto-coccal, 6 had fungal, 4 had gram-negative, and 6 hadenterococcal organisms. Mean follow-up was 4.3 � 2.9years.

Results. Hospital mortality was 3.9%. Permanent pace-makers were required in 31 patients. Survival at 1 year, 2years, 5 years, and 10 years was 90%, 86%, 73%, and 56%,respectively, with a risk of 5.3% per year after 6 months.Four patients underwent reoperation for recurrent PVEof the allograft (95% freedom from recurrent PVE at >

� 2years). Risk of recurrent PVE peaked at 9 months andthen declined to a low level by 18 months.

Conclusions. A strategy of radical debridement andaortic root replacement with a cryopreserved aortic allo-graft for aortic PVE is safe, effective, and recommended.

(Ann Thorac Surg 2002;74:650–9)© 2002 by The Society of Thoracic Surgeons

The goals of operation for aortic prosthetic valveendocarditis (PVE) are to excise all infected material,

repair any cardiac defects caused by infection, recon-struct the aortic root, place a competent valve, andprevent recurrent infection. It has been our strategy totreat aortic PVE with radical debridement of infectedtissue and replacement of the aortic root and valve witha cryopreserved aortic allograft. This study was under-taken to evaluate that strategy. Goals of the study were to(1) characterize the patients and the nature of the endo-carditis, (2) investigate hospital outcomes and their pre-dictors, (3) determine the rate of endocarditis recurrence,and (4) identify risk factors for death after radicaloperation.

Patients and Methods

Study GroupFrom 1988 through 2000, 103 patients underwent treat-ment of aortic PVE with surgical debridement of infected

tissue and aortic root replacement with a cryopreservedaortic allograft (CryoLife, Inc, Kennesaw, GA) at TheCleveland Clinic Foundation. Patients who had replace-ment of an aortic valve prosthesis with an allograft butdid not have prosthetic valves in other positions wereidentified from our surgical registry. Their medicalrecords were reviewed to confirm and augment datacollected routinely and concurrently with patient care.Use of these data for research has been approved by theclinic’s institutional review board.

Mean age of the patients was 57 � 15 years (range, 24to 84 years). Eighty-two percent were men. Seventy-fourpatients had one previous aortic valve replacement, 24had two, and 5 had three replacements. The sequence ofaortic valve replacements and previous infections is pre-sented in Table 1. Among the 23 patients with a history ofnative valve endocarditis, the allograft was implantedafter one episode (17 patients), two episodes (5 patients),or three episodes of PVE (1 patient). In the remaining 80patients, a cryopreserved aortic allograft was placed afterone previous aortic valve operation (57 patients), two (19),or three previous aortic valve operations (4 patients).

At the time of allograft implantation for PVE, 50.5% ofpatients had a mechanical aortic valve and 49.5% had abioprosthesis (Table 2). Fifty-seven (55%) had been pre-viously implanted in centers other than our own.

Preoperative signs and symptoms included severeprosthetic aortic valve regurgitation in 70% of patients,

Presented at the Forty-eighth Annual Meeting of the Southern ThoracicSurgical Association, San Antonio, TX, Nov 8–10, 2001.

Address reprint requests to Dr Sabik, Department of Thoracic andCardiovascular Surgery, The Cleveland Clinic Foundation, 9500 EuclidAve/F25, Cleveland, OH 44195; e-mail: [email protected].

© 2002 by The Society of Thoracic Surgeons 0003-4975/02/$22.00Published by Elsevier Science Inc PII S0003-4975(02)03779-7



heart failure in 68%, septic embolism in 28%, and hypo-tension in 24% (Table 3).

Characterization of InfectionNinety-two patients were culture positive either positivepreoperative blood cultures or positive cultures of surgi-cal specimens (Table 4). Eighty had positive preoperativeblood cultures. The most common infecting organismswere gram-positive cocci (81 patients): Staphylococcus (52patients), Streptococcus (20 patients), and Enterococcus (6patients). Fungal infections occurred in 6 patients andgram-negative infections in 4. Four patients had twoinfecting organisms. No organisms were cultured preop-eratively or at the time of operation in 11 patients.

In 80 patients (78%), abscess formation was present(this term includes pseudoaneurysms and abscess cavi-ties as well). The abscesses were single in 45 patients(44%), multiple in 30 (29%), and healed in 5 (4.9%). We

were unable to identify any factors associated with singleabscesses. However, multivariable analysis demon-strated that multiple abscesses were more common inpatients with persistent pyrexia (93% versus 55%; p �0.0005), those presenting with dyspnea (97% versus 66%;p � 0.0003), those with short intervals from operation toendocarditis (p � 0.01), and those with a more recentoperation (p � 0.006).

Fistulas were present in 15 patients (15%), and com-plete aortoventricular discontinuity was found in 41(40%). Older patients were at higher risk for the latter(59 � 16 years versus 55 � 14 years, p � 0.003), as werethose presenting with hypotension (41% versus 13%; p �0.003) and those with staphylococcal infections (68%versus 39%; p � 0.009).

The interval from aortic valve replacement to operationfor PVE was 3 months or less in 18 patients (17.5%), 6months or less in 36 (35%), 1 year or less in 55 (53%), 2years or less in 64 (62%), 5 years or less in 76 (74%), and10 years or less in 95 patients (92%). See Appendix 1 fordetails.

Table 1. Sequence of Aortic Valve Replacements andInfections

Sequence No. of Patients

Native valve endocarditis 23After first prosthesis 17After second prosthesis 5a

After third prosthesis 1No native valve endocarditis 80

After first prosthesis 57After second prosthesis 19b

After third prosthesis 4c

No. of episodes of endocarditisOne 68Two 30Three 5

No. of episodes of PVEOne 86Two 17

a In 1, PVE occurred only after second prosthesis. b In 10, PVE oc-curred only after second prosthesis. c In 1, PVE occurred only afterthird prosthesis; in 3, PVE occurred after both second and third prosthe-ses.

PVE � prosthetic valve endocarditis.

Table 2. Prosthesis and Aortic Graft in Place at Time ofCurrent Operation for Prosthetic Valve Endocarditis

Prosthesis or Aortic Graft No. of Patients

ProsthesisMechanical 52

St. Jude Medical 38Other 14

Bioprosthesis 51Stented pericardial 30Stented xenograft 19Allograft 2

Aortic graft 27Interposition graft 14Composite graft (Bentall) 13

Table 3. Progressive Signs and Symptoms

Variable No. of Patients

SymptomsDyspnea 77Persistent pyrexia 68Neurologic event 30Angina 3Syncope 2

SignsPreoperative heart failure 70Severe prosthetic aortic valve

regurgitation72

Hypotension 25Septic embolus 29

Table 4. Bacteriology

Organism No. of Patientsa

Culture positive 92Gram-positive organisms 81

Staphylococcus 52aureus 28epidermidis 22Other 2

Streptococcus 20viridans 10Other 10

Enterococcus 6Other 4

Gram-negative organisms 4Fungus 6Other 4

No organisms cultured 11

a The numbers are not mutually exclusive because multiple organismswere cultured in 4 patients.

651Ann Thorac Surg SABIK ET AL2002;74:650–9 ALLOGRAFT ROOT REPLACEMENT FOR AORTIC VALVE PVE



Operative TechniqueAll operations were performed through a median ster-notomy. Cardiopulmonary bypass was performed withaortic cannulation in most patients, but not uncommonlyusing other routes (Table 5). Right atrial or bicavalcannulation was used for venous drainage.

Infected prosthetic aortic valves were excised, andsurrounding infected tissue was debrided aggressively(Fig 1A). Any intracardiac defects caused by infection ordebridement, such as fistulas or ventricular septal de-fects, were repaired with either bovine or autologouspericardium using monofilament suture. No artificialpatch material was used.

Cryopreserved aortic allografts were implanted as aor-tic roots. Interrupted or continuous 4-0 polypropylenemonofilament suture was used for the proximal anasto-mosis of the allograft to the heart. Often, either autolo-gous or bovine pericardium or saphenous vein wasplaced as a washer in this proximal suture line.

In many cases, infection had destroyed the usuallandmarks of the aortic root. To orient the allograft, thetwo fibrous trigones and anterior mitral valve leaflet onthe allograft were aligned with the patient’s fibroustrigones and anterior mitral valve leaflet.

The coronary ostia were mobilized as buttons andsewn to the allograft with running 5-0 polypropylenemonofilament suture. The distal anastomosis of the allo-graft to the ascending aorta was constructed in end-to-end fashion with running 4-0 polypropylene monofila-ment suture.

If the ascending aorta needed to be replaced, either anaortic valve allograft long enough to do this or anallograft aorta was used. Nonbiological material wasavoided in reconstructing the aortic root or replacing theascending aorta.

In reconstructing aortic roots with allografts, infectedareas of myocardium were excluded from the systemiccirculation. Allografts were sutured to the heart proximalto the infection. The most common site of abscess forma-tion was the mitral-aortic intervalvular fibrosa (see Fig1A). To exclude this abscess cavity from the circulation,the attached anterior leaflet of the mitral valve on theaortic allograft was sutured to the native anterior leafletof the mitral valve (Fig 1B). The abscess cavity was not

closed but allowed to “drain” into the mediastinum. Themediastinum was routinely drained with chest tubes for48 to 72 hours.

Antibiotics were begun as soon as a diagnosis of PVEwas made and continued for at least 8 weeks afteroperation. Patients with fungal infections were placed ona lifelong regimen of oral antifungal agents.

Follow-UpFollow-up was complete for all patients. Mean follow-upwas 4.3 � 2.9 years for survivors (maximum follow-up,12.6 years), and 397 patient-years of total follow-up wereavailable for analysis. Of the survivors, 77% were fol-lowed for 2 years, 30% for 5 years, and 7% for 10 or moreyears. Therefore, time-related depictions were truncatedat 10 years.

Data AnalysisVARIABLES AND END POINTS ANALYZED. To characterize theendocarditis, descriptive statistics and multivariable logis-tic regression were used to identify factors associatedwith a single abscess, multiple abscesses, and aortoven-tricular discontinuity. To investigate hospital outcomes,multivariable logistic regression was used to identifypredictors of respiratory failure, renal failure, and needfor permanent pacemakers. To determine the rate ofrecurrent endocarditis and to identify risk factors for deathafter operation, multivariable analysis was performed inthe hazard function domain [1].

DESCRIPTIVE STATISTICS. Descriptive statistics included themean and standard deviation for continuous variables,and frequencies and percentages for categorical vari-ables. Nonparametric estimates of time-related eventswere obtained using the Kaplan-Meier method. A para-metric method was used to resolve the number of phasesof instantaneous risk for each time-related end point(hazard function) and to estimate shaping parameters [1].

MULTIVARIABLE ANALYSES. Screening of variables for possi-ble relation to findings of interest or end points employedcontingency table methods, t testing, Cox proportionalunivariable analysis, and stratified life-table analysis us-ing the log-rank test. Potential correlates and risk factorswere organized for analysis as in Appendix 2. Continu-ous and ordinal variables were assessed univariably bydecile risk analysis to suggest possible transformations ofscale that best calibrated the variable with respect tomodel assumptions. For the small number of missingvalues, noninformative imputation of the mean valuewas used.

The primary variable selection method for all multiva-riable analyses was bootstrap bagging using 1,000 resam-pled data sets, a forward stepwise procedure, and a pvalue criterion for retention of variables of 0.05 [2].Variables appearing in at least 50% of the bootstrapanalyses were considered reliably identified.

PRESENTATION. Confidence limits of proportions and time-related depictions are accompanied by confidence limits

Table 5. Operative Details

Detail No. of Patients

Cannulation siteAorta 69Femoral artery 24Axillary artery 9Subclavian artery 1

Technique of allograft insertionRoot 92Root with patch 11

Replacement of aortaAscending 30Arch 6

652 SABIK ET AL Ann Thorac SurgALLOGRAFT ROOT REPLACEMENT FOR AORTIC VALVE PVE 2002;74:650–9



equivalent to one standard error (68%). Tables of riskfactors identified in the hazard function domain arepresented with regression coefficients rather than hazardratios because the model is not one of proportionalhazards and because many of the continuous factorswere transformed to meet model assumptions. The latterreason also accounts for presentation of logistic regres-sion coefficients rather than estimated odds ratios.

Results

Hospital Mortality and MorbidityFour patients (3.9%; confidence limits, 2.0% to 7.0%) diedin the hospital. All had early-onset staphylococcal infec-tions. All had abscesses, multiple in 2. Two had aortoven-tricular discontinuity, and 2 had fistulas. Preoperatively,all had neurologic events, and 3 were hypotensive. Twodied of persistent sepsis with multiple organ systemfailure.

Complete heart block requiring a permanent pace-maker occurred in 31 patients (30%). It was more com-mon in patients who had more than one previous oper-ation (51% versus 18%), in those with multiple abscesses(45% versus 22%), and in those with an aortic biopros-thesis in place at the time of operation for PVE (65%versus 43%) (Table 6).

Respiratory failure occurred in 22 patients (21%). Riskfactors were older age (66 � 11 years versus 54 � 15years), multiple abscesses (55% versus 22%), and shorterinterval from previous valve replacement to operation forPVE (median interval of 6 months versus 14 months) (seeTable 6).

Renal failure occurred in 24 patients (23%). Risk factorswere older age (65 � 12 years versus 54 � 15 years) andhigher preoperative creatinine level (1.9 � 1.2 mg/dLversus 1.3 � 0.84 mg/dL) (see Table 6).

Recurrent Aortic Prosthetic Valve EndocarditisFour patients developed recurrent endocarditis. At thetime of allograft insertion, 3 of the 4 had abscesses(multiple in 2), and 1 had aortoventricular discontinuity.All had active gram-positive infections, 1 had Staphylo-coccus aureus and a gram-negative organism, 1 had Staph-ylococcus epidermidis, and 1 had Enterococcus. All requiredreoperation. (Only 1 other patient underwent allograftexplantation, and it was for subvalvular stenosis 2 yearspostoperatively.)

Freedom from recurrent endocarditis at 1 year, 2 years,5 years, and 10 years was 96%, 95%, 95%, and 95%,respectively (Fig 2A). Instantaneous risk (hazard func-tion) of recurrent aortic PVE peaked at 9 months (Fig 2B).

Fig 1. Technique of aortic root replacement and allograft implanta-tion. (A) The infected prosthesis is excised, and all infected and ne-crotic tissue is aggressively debrided. The coronary ostia are pre-pared as buttons. (B) The most common site of abscess formation isthe mitral-aortic intervalvular fibrosa. After the abscess cavity isdebrided, the defect is repaired by suturing the allograft mitral valveto the native mitral valve below the level of infection.4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™




Risk of recurrent endocarditis at 6 months, 1 year, and 2years was 0.25% per year, 0.44% per year, and 0.08% peryear, respectively. Of the 4 patients with recurrent endo-carditis, 1 died 4 months after reoperation and 1, 10 yearsafter reoperation.

SurvivalSurvival at 30 days, 1 year, 2 years, 5 years, and 10 yearswas 96%, 90%, 86%, 73%, and 56%, respectively (Fig 3A).The instantaneous risk of death peaked right after oper-ation and then fell rapidly to a constant rate of 5.3% peryear (Fig 3B). Survival was similar for 10 years to thatexpected for patients undergoing reoperative aortic valvereplacement but without PVE (Appendix 3). Incrementalrisk factors for early death were a shorter interval fromprevious operation to operation for PVE and the presenceof fistulas (Table 7). The constant hazard phase increasedwith increasing age and with higher preoperative creat-inine levels (see Table 7).

Comment

BackgroundProsthetic valve endocarditis is a devastating complica-tion of valve replacement. Its prevalance is 4.1% at 4years after primary valve replacement, with the greatestrisk occurring 1 month to 2 months after operation [3].Medical therapy alone has been disappointing, withmortality as high as 70% [3]. Results obtained withcombined medical and surgical treatment have improvedover time. Early surgical series reported mortality of 20%to 60% [4–6], whereas more recent studies have reported4% to 20% [7–13]. Many factors have contributed to thisimprovement: better and more specific antibiotics, morevalve replacement options, improved methods of myo-cardial protection that permit the longer aortic clampingnecessary for more anatomic aortic root reconstruction,and improved anesthesia management and postopera-tive care.

The goals of surgical therapy for aortic PVE includeremoval of all infected and necrotic material, repair of

any resultant cardiac defects, and placement of a compe-tent valve. However, severe destruction of the aortic rootoccurring with aortic PVE, including abscesses, aortoven-

Fig 2. Recurrent aortic prosthetic valve endocarditis after allograftplacement. (A) Freedom from recurrence. Each circle represents arecurrence. Vertical bars are 68% confidence limits of the Kaplan-Meier estimates. The solid line and its 68% confidence limits (bro-ken lines) are parametric estimates as described in Data Analysis.The number of patients remaining at risk is depicted in parenthe-ses. (B) Hazard function. The solid line represents point estimates,and the broken lines enclose 68% confidence limits.

Table 6. Factors Associated With Postoperative Complications

Complication and Risk Factors Logistic coefficient � Standard Error p Value

Pacemaker insertion (n � 31)� One redo operation 2.2 � 0.59 0.0002Multiple abscesses 1.58 � 0.56 0.005Bioprosthesis in place at time of operation 1.39 � 0.54 0.01

Respiratory failure (n � 22)Older agea 5.2 � 1.70 0.002Multiple abscesses 1.33 � 0.57 0.02Shorter interval from previous valve

operation to operation for PVEb0.141 � 0.068 0.04

Renal failure (n � 24)Older agec 1.38 � 0.48 0.004Higher preoperative creatinine levela 1.60 � 0.58 0.006

a Natural logarithmic transformation. b Inverse transformation. c Squared transformation (age/50)2.





tricular discontinuity, fistulas, and ventricular septal de-fects, makes its reconstruction difficult.

A number of different operations to treat aortic PVEhave been described. These include patch closure ofannular and aortic defects and suturing of the prostheticvalve directly to the patch [14], supracoronary placementof the prosthetic valve with closure of coronary ostia andcoronary artery bypass grafting [15], composite pros-thetic valve conduit replacement and reimplantation ofcoronary ostia [16], the Ross procedure [17], and aorticroot replacement with an allograft [11].

Principal FindingsThe strategy of treating aortic PVE with radical debride-ment and replacement of the aortic root and valve with acryopreserved aortic allograft was effective. Hospitalmortality was low, morbidity was related to aggressiveinfection, freedom from recurrent endocarditis was high,and long-term survival was good.

HOSPITAL MORTALITY. Hospital mortality was lower in ourseries (3.9%) than in other series (10% to 20%), [9, 10, 13,

14, 18] where mechanical or xenograft valves were usedto treat PVE. We believe the improvement was due bothto our strategy of radical debridement of infected tissueand to some features of allografts that make them espe-cially beneficial in treating PVE. These features include(1) their flexibility, which allows allografts to be tailoredto conform to aortic roots distorted by infection anddebridement, (2) their attached anterior mitral valveleaflet, which can be used to repair defects such asfistulas, ventricular septal defects, and abscesses, and (3)their biological nature, which appears to make themrelatively resistant to early recurrent infection [19–21].

Early survival after operation was indirectly related tothe aggressiveness of the infection. All four hospitaldeaths occurred in patients with early, active, destructivestaphylococcal infections. Fistulas and early-onset PVEwere risk factors for early death; aortoventricular discon-tinuity was a risk factor for early operation for PVE.

POSTOPERATIVE MORBIDITY. Postoperative morbidity was re-lated to the aggressiveness and destructiveness of theinfection. Patients with multiple abscesses were morelikely to experience respiratory failure postoperativelyand to require permanent pacemakers. Almost one thirdof our patients, developed heart block requiring a per-manent pacemaker after operation. This complicationwas due both to the aggressiveness of the aortic rootinfection and to the radical debridement necessary toeradicate it.

RECURRENT ENDOCARDITIS. Risk of recurrent endocarditiswas also lower in our series compared with others. At 10years, freedom from recurrent endocarditis was 95%,whereas others have reported 82% [10]. We believe ourlower risk was due to both aggressive debridement ofinfected tissue and resistance of allografts to infection[19–21].

In our series, the pattern of risk of recurrent endocar-ditis also differed from that of other studies. We foundthat risk peaked about 9 months after operation and thendeclined slowly. In previous reports in which aorticallografts were used to treat both native valve endocar-ditis and PVE [19–21], risk did not peak but rather

Fig 3. Survival after surgical treatment of aortic prosthetic valveendocarditis (PVE). The format is similar to that of Figure 2. (A)Survival. The dot-dash-dot line represents survival predicted afteraortic valve re-replacement for indications other than PVE for pa-tients with characteristics identical to those of patients in this study(see Appendix 3). (B) Hazard function.

Table 7. Incremental Risk Factors for Death

Factor

Coefficient� StandardDeviation p Value

Early hazard phaseShorter interval between

previous valve operationand operation for PVEa

0.46 � 0.141 0.001

Fistulas 5.7 � 1.94 0.003Constant hazard phase

Older age at operationb 0.81 � 0.24 0.0008Higher preoperative creatinine

level0.34 � 0.151 0.02

a Inverse transformation. b Exponential transformation, exp(age/50).





remained at a constant low level. Why our pattern wasdifferent is unclear, but it may have been due to thesevere invasiveness of infections observed in our seriesand the fact that all the patients in our series had PVE. All4 patients in whom recurrent endocarditis developed haddeeply invasive and destructive infections. Despite radi-cal debridement, not all infected material may have beenexcised. Postoperative antibiotic therapy may have qui-eted the infection but not completely eradicated it, thusaccounting for the relatively late recurrence of endocar-ditis in a small number of patients.

LONG-TERM SURVIVAL. We believe this strategy for treatingaortic PVE improves survival. Comparing the outcomesin this series with previous reports on endocarditis isdifficult because most series contain heterogeneous pa-tient populations; ie, patients with both native valveendocarditis and PVE, both mitral and aortic valve endo-carditis, and both healed and active endocarditis. In ourstudy, the patient population was fairly homogeneous; allpatients had aortic PVE, and 92 had active culture-positive endocarditis.

Despite these difficulties in comparison, late survival inthis report appears to be better than that in previousreports where mechanical or xenograft valves were usedto treat PVE. If hospital mortality is included in deter-mining late survival (not simply survival after discharge),long-term survival for patients with standard prostheseshas been only about 50% to 60% at 5 years and 30% to40% at 10 years [9, 10, 14, 18, 22, 23]. In a previous studyof PVE from our institution [8], survival including hospitalmortality was similar to that of this report—71% and 52%at 5 and 10 years, respectively. However, a substantialnumber of those patients underwent operation for healedPVE. We believe the improvement in survival in thepresent study is attributable in part to both the lowhospital mortality and the low incidence of recurrentendocarditis (4 patients), which is itself important forlong-term survival.

For additional perspective, we compared time-relatedsurvival of patients in this series with survival of patientswho had the same characteristics undergoing reoperativeaortic valve replacement for noninfectious indications(see Appendix 2 & 3). They were nearly identical (see Fig3). This suggests that when patients with aortic PVE aretreated with aggressive debridement and allograft aorticroot replacement, the survival penalty paid for the pres-ence of active endocarditis is neutralized. These patientsappear to have a “normal” survival comparable to that ofpatients undergoing reoperative aortic valve replacement.

LimitationsAlthough this is the largest series reported to date ofpatients treated surgically for aortic PVE, in absolutenumbers there were only 103 patients, and few hadpostoperative events. Therefore, it is not possible todetermine predictors of hospital death or recurrence ofendocarditis. Also, because half the patients had hadprevious aortic valve replacement at other institutions,referral patterns may have skewed the apparently longintervals between that operation and treatment of PVE or

may have increased the prevalence of invasive PVE. It isalso important to note that follow-up is too short to allowlate allograft structural failure to affect outcomes and toassess whether endocarditis will have an adverse influ-ence on allograft durability.

ConclusionsA strategy of radical debridement of infected and ne-crotic material and aortic root replacement with a cryo-preserved aortic allograft was effective for aortic PVE.Hospital mortality was low, freedom from recurrentendocarditis was high, and long-term survival was good.Postoperative morbidity and early mortality were directlyrelated to the aggressiveness of the infection. Heart blockrequiring a permanent pacemaker was a common com-plication of this aggressive therapy. On the basis of thisstudy, we recommend this treatment strategy for aorticPVE.

We thank Colleen Laffey, RN, for assistance in maintaining theallograft database; Angela Eshelman, AB, for database manage-ment; Linda DiPaola, BA, for constructing the data set; Jeevan-antham Rajeswaran, MS, for performing some of the statisticalanalyses; and Tess Knerik, BS, for expert editorial assistance.

References1. Blackstone EH, Naftel DC, Turner ME Jr. The decomposition

of time-varying hazard into phases, each incorporating aseparate stream of concomitant information. J Am Stat Assoc1986;81:615–24.

2. Breiman L. Bagging predictors. Machine Learning 1996;26:123–40.

3. Ivert TS, Dismukes WE, Corbis CG, Blackstone EH, KirklinJW, Bergdahl LA. Prosthetic valve endocarditis. Circulation1984;69:223–32.

4. Wilson WR, Danielson GK, Giuliani ER, Geraci JE. Pros-thetic valve endocarditis. Mayo Clin Proc 1982;57:155–61.

5. Glazier JJ, Verwilghen J, Donaldson RM, Ross DN. Treat-ment of complicated prosthetic aortic valve endocarditiswith annular abscess formation by homograft aortic rootreplacement. J Am Coll Cardiol 1991;17:1177–82.

6. Donaldson RM, Ross DM. Homograft aortic root replace-ment for complicated prosthetic valve endocarditis. Circula-tion 1984;70(Suppl 1):178–81.

7. Camacho MT, Cosgrove DM. Homografts in the treatment ofprosthetic valve endocarditis. Semin Thorac CardiovascSurg 1995;7:32–7.

8. Lytle BW, Priest BP, Taylor PC, et al. Surgical treatment ofprosthetic valve endocarditis. J Thorac Cardiovasc Surg1996;111:198–210.

9. Jault F, Gandjbakhch I, Chastre JC, et al. Prosthetic valveendocarditis with ring abscesses. Surgical management andlong-term results. J Thorac Cardiovasc Surg 1993;105:1106–13.

10. Larbalestier RI, Kinchla NM, Aranki SF, Couper GS, CollinsJJ Jr, Cohn LH. Acute bacterial endocarditis. Optimizingsurgical results. Circulation 1992;86(Suppl 2):68–74.

11. Dossche KM, Defauw JJ, Ernst SM, Craenen TW, De JonghBM, Brutel de la Riviere A. Allograft aortic root replacementin prosthetic aortic valve endocarditis: a review of 32 pa-tients. Ann Thorac Surg 1997;63:1644–9.

12. Petrou M, Wong K, Albertucci M, Brecker SJ, Yacoub MH.Evaluation of unstented aortic homografts for the treatmentof prosthetic aortic valve endocarditis. Circulation 1994;90(Suppl 2):198–204.

13. Aranki SF, Santini F, Adams DH, et al. Aortic valve endo-carditis. Determinants of early survival and late morbidity.Circulation 1994;90(Suppl 2):175–82.




14. Symbas PN, Vlasis SE, Zacharopoulos L, Lutz JF. Acuteendocarditis: surgical treatment of aortic regurgitation andaortico–left ventricular discontinuity. J Thorac CardiovascSurg 1982;84:291–6.

15. Reitz BA, Stinson EB, Watson DC, Baumgartner WA, Jamie-son SW. Translocation of the aortic valve for prosthetic valveendocarditis. J Thorac Cardiovasc Surg 1981;81:212–8.

16. Frantz PT, Murray GF, Wilcox BR. Surgical management ofleft ventricular–aortic discontinuity complicating bacterialendocarditis. Ann Thorac Surg 1980;29:1–7.

17. Joyce F, Tingleff J, Pettersson G. The Ross operation in thetreatment of prosthetic aortic valve endocarditis. SeminThorac Cardiovasc Surg 1995;7:38–46.

18. David TE, Bos J, Christakis GT, Brofman PR, Wong D,Feindel CM. Heart valve operations in patients with activeinfective endocarditis. Ann Thorac Surg 1990;49:701–5.

19. Haydock D, Barratt-Boyes B, Macedo T, Kirklin JW, Black-stone EH. Aortic valve replacement for active infectiousendocarditis in 108 patients. J Thorac Cardiovasc Surg 1992;103:130–9.

20. McGiffin DC, Galbraith AJ, McLachlan GJ, Stower RE, WongML, Stafford EG. Aortic valve infection: risk factors for deathand recurrent endocarditis after aortic valve replacement.J Thorac Cardiovasc Surg 1992;104:511–20.

21. McGiffin DC, Kirklin JK. The impact of aortic valve ho-mografts on the treatment of aortic prosthetic valve endo-carditis. Semin Thorac Cardiovasc Surg 1995;7:25–31.

22. Mullany CJ, Chua YL, Schaff HV, et al. Early and latesurvival after surgical treatment of culture-positive endocar-ditis. Mayo Clin Proc 1995;70:517–25.

23. Baumgartner WA, Miller DC, Reitz BA, et al. Surgicaltreatment of prosthetic valve endocarditis. Ann Thorac Surg1983;35:87–104.

24. Ferrazzi P, McGiffin DC, Kirklin JW, Blackstone EH, BourgeRC. Have the results of mitral valve replacement improved?J Thorac Cardiovasc Surg 1986;92:186–97.

Appendices

Appendix 1. Timing of Infection

MethodDistribution of intervals between the immediately pre-ceding aortic valve replacement and operation for pros-thetic valve endocarditis (PVE) was characterized by theKaplan-Meier and parametric methods [1]. However,unlike survival-type analysis, these intervals have nodenominator. Thus, the distribution is portrayed by theprobability density function rather than the hazard func-tion, calculated as the product of hazard and survivorshipfunctions.

To obtain probability density functions for subgroupsof patients, the multivariable equation was solved foreach patient across time, with each patient’s individualvalues for variables in the model. Estimates of the sub-group cumulative frequency distribution and probabilitydensity functions were then found by averaging theseindividual curves [24]. Factors associated with earlier andlater time to operation for PVE were identified by multi-variable analysis in the hazard function domain [1].

ResultsThe overall distribution of intervals between immedi-ately preceding aortic valve replacement and operationfor PVE exhibited an early peak 2 months after aorticvalve replacement that fell to a persistent, but low level(Fig A1). Factors associated with a short interval to oper-

ation for PVE included aortoventricular discontinuity, amore symptomatic patient, type of infective organism,and type of prosthesis before onset of PVE (AppendixTable 1). The shortest intervals to operation were ob-served in Staphylococcus epidermidis PVE (Fig A2). Staph-ylococcus aureus infections exhibited somewhat longerintervals and streptococcal infections, the longest inter-vals, rather evenly distributed across time.

Both mechanical prostheses and bovine stented peri-cardial valves exhibited an early peak of short intervals tooperation, but stented porcine xenograft PVE occurredmore evenly across time (Fig A3). However, 67% of thebovine pericardial prostheses had been implanted at ourinstitution compared with only 32% of the porcine xeno-grafts.

CommentAs in other studies, the interval from aortic valve replace-ment to operation for endocarditis was dependent on the

Fig A1. Time interval between immediately preceding aortic valvereplacement and operation for prosthetic valve endocarditis (PVE).(A) Cumulative distribution function, presented as the complementof the usual format. Each circle represents an interval to occurrenceof operation for PVE. Vertical bars are 68% confidence limits of theKaplan-Meier estimates. The solid line and its 68% confidence lim-its (broken lines) are parametric estimates as described in DataAnalysis. The number of patients coming to operation at longer in-tervals is depicted in parentheses. (B) Probability density functionshowing the frequency of occurrence of operation for PVE permonth.




infecting organism [3]. Staphylococcal infections oc-curred early and streptococcal infections, late. Early oc-currence of staphylococcal infections is probably due tohospital-acquired infection at the time of valve replace-ment, and these were most likely to be destructive.

Timing for this study was based on hard data on datesof operation. We did not have secure information aboutthe date of onset of the infections.

Appendix 2. Variables Available for Analysis

DemographicAge (y), height, weight, body surface area, body massindex

Symptoms and SignsNew York Heart Association functional class, emergentoperation, dyspnea, progressive heart failure, hypoten-sion, syncope, preoperative neurologic event, persis-

tent leukocytosis, persistent pyrexia, systemic emboli-zation, progressive aortic regurgitation, intervalbetween previous operation and operation for pros-thetic valve endocarditis

Characteristics of EndocarditisCulture positive or negative, gram-positive or nega-tive, specific infecting organism, presence of singleabscess or multiple abscesses, fistulas, aortoventriculardiscontinuity

Preoperative Medical Management of EndocarditisComplete course of antibiotics, incomplete course, noantibiotics

Previous EndocarditisNative valve endocarditis, sequence of infection withrespect to each operation, total number of occurrences ofendocarditis

Appendix Table 1. Factors Associated With Interval From Prior Valve Replacement to Operation for ProstheticValve Endocarditis

Factor Coefficient � Standard Error p Value

Early onsetMore symptomatic at reoperationa 0.080 � 0.028 0.004Organism other than Streptococcus �1.01 � 0.48 0.04Aortoventricular discontinuity 0.69 � 0.27 0.01Prosthesis other than stented porcine xenograft �2.2 � 0.75 0.003

Late onsetWomen 1.82 � 0.66 0.006Symptoms other than dyspnea �1.86 � 0.53 0.0005Staphylococcus epidermidis 2.8 � 0.69 � 0.0001Stented bovine pericardial prosthesis 1.77 � 0.48 0.0003

a Squared transformation of New York Heart Association functional class.

Fig A2. Frequency of occurrence of operation for aortic prostheticvalve endocarditis according to infecting organism. The figure is de-rived from solutions of the multivariable equation shown in Appen-dix Table 1. Confidence limits (68%) for the frequency of occurrenceof Staphylococcus organisms overlap, but they do not overlap thosefor Streptococcus.

Fig A3. Frequency of occurrence of operation for aortic prostheticvalve endocarditis according to type of prosthesis. The figure is de-rived as in Figure A2. Confidence limits (68%) for the frequency ofoccurrence of endocarditis after aortic valve replacement with bovinepericardial and mechanical prostheses overlap, but they do not over-lap those for porcine xenografts early after operation.




Prosthesis in Place at Time of Prosthetic ValveEndocarditisType of prosthesis (mechanical, biological, allograft), de-vice manufacturer, group of porcine xenograft devices,previous aortic replacement (ascending or arch)

Cardiac-Related ComorbidityFamily history of coronary artery disease, rhythm mech-anism, total number of cardiac operations, presence ofmitral regurgitation

Noncardiac ComorbiditySmoking history, renal disease, preoperative creatininelevel, preoperative blood urea nitrogen level, preopera-tive bilirubin level

ProcedureUse of patch extension to allograft, replacement of ascend-ing aorta or arch, concomitant operations, date of procedure

Appendix 3. Expected Survival

We sought to determine expected survival for a group ofpatients undergoing reoperative aortic valve replacementwho had precisely the characteristics of our 103 patientsbut did not present with endocarditis. For this, a data setof 13,258 aortic valve replacements from nine represen-tative sources containing both primary valve replace-ments and reoperations was used (unpublished data).Multivariable analysis of all-cause mortality includedpatient age, sex, body surface area, New York HeartAssociation functional class, original native aortic valvepathophysiology, concomitant coronary artery bypassgrafting, and date of operation.

The multivariable equation was solved for time-relatedsurvival for each of the 103 patients using patient-specificvalues for model variables. The 103 curves were thenaveraged across time to yield expected survival [24].

DISCUSSION

DR JOHN D. OSWALT (Austin, TX): My colleagues and I inAustin have a policy of using the Ross procedure for infectiveendocarditis. The first reason for this is that we think the viabletissue may provide some added benefit against recurrence ofendocarditis. The second reason concerns longevity. We believethat the autograft valve will survive longer than the allograft. Ofthe patients who died at a later date, ie, those not included in the73% survival rate at 5 years, how many died of a valve-relatedcause? Also, how many patients during follow-up have hadremoval of the allograft because of failure of the graft?

DR SABIK: Five patients in this series have had the allograftremoved. Recurrent endocarditis developed in 4 of them, and all4 required reoperation and allograft replacement. Subvalvularstenosis developed in the other patient, and reoperation wasnecessary to relieve the obstruction. At the time of reoperationin that patient, the allograft valve leaflets appeared normal andpliable.

We have been unable to accurately determine all the causes oflate death, and consequently I do not know how many of the latedeaths were valve related. However, it does not appear to us thatduring the 10-year follow-up, allograft failure has been animportant cause of late morbidity and mortality.

DR RANDALL B. GRIEPP (New York, NY): Dr Sabik, this is, Ithink, the lowest hospital mortality ever reported for acuteprosthetic valve endocarditis, and you and your colleagues atThe Cleveland Clinic are to be congratulated. I will discuss aslightly different approach.

For a number of reasons my associates and I have usedprosthetic material in this situation. Over the same interval, wehave operated on 28 patients with prosthetic valve endocarditis,either after previous aortic valve replacements or previouscomposite replacements. The adverse outcome was 11%; inother words, 89% of the patients were discharged from thehospital without neurological deficit. In that group, we had 1patient who had recurrent endocarditis while still in the hospital

and 1, within the first year, but there were no other instances ofrecurrent endocarditis in these patients, in all of whom aprosthetic conduit was used.

We agree with the first two of your three principles, beginningwith the need for radical debridement of infected tissue. We alsothink it is important to remove a good part of the ascendingaorta so that the conduit lies in vascularized tissue in theposterior pericardium. However, we have been amazed to seethat prosthetic material seems to be very resistant to infectionand to recurrent endocarditis.

One point I consider important is the durability of the repair.I wonder if the 5% ongoing mortality risk for your patients hadanything to do with the aortic homograft per se, for example, interms of the integrity of the valve.

DR SABIK: Prosthetic valve endocarditis clearly compromisesthe long-term outcome of patients after aortic valve replace-ment. Previously published series where mechanical or xeno-graft valves have been used to treat prosthetic valve endocarditisand where hospital mortality has been included in the long-termsurvival calculation have demonstrated 5 and 10-year survival ofonly 50% to 60% and 30% to 40%, respectively. It also appearsthat even when patients survive reoperation for prosthetic valveendocarditis, the long-term outcome is still compromised. Theseless than desirable results are possibly due to the complicationsof recurrent endocarditis, such as risk of reoperation and para-prosthetic leak. In our series, survival, including hospital mor-tality, was 73% at 5 years and 56% at 10 years. This improvedlong-term survival is due to our low hospital mortality andpossibly to the low occurrence of recurrent infection. Allograftshave been shown to have a low relative risk of infection, and inour series, freedom from recurrent endocarditis was 95% at 10years.

As I stated earlier, we do not believe homograft failure hasbeen a significant cause of late mortality and morbidity. How-ever, the time period of this study is only 10 years, and I wouldnot expect much allograft failure during that interval.




2002;74:650-659 Ann Thorac SurgB. Pettersson and Delos M. Cosgrove

Joseph F. Sabik, Bruce W. Lytle, Eugene H. Blackstone, Antonino G.M. Marullo, Gosta endocarditis

Aortic root replacement with cryopreserved allograft for prosthetic valve

& ServicesUpdated Information

http://ats.ctsnetjournals.org/cgi/content/full/74/3/650including high-resolution figures, can be found at:

References http://ats.ctsnetjournals.org/cgi/content/full/74/3/650#BIBL

This article cites 22 articles, 13 of which you can access for free at:

Citations http://ats.ctsnetjournals.org/cgi/content/full/74/3/650#otherarticles

This article has been cited by 24 HighWire-hosted articles:

Subspecialty Collections

http://ats.ctsnetjournals.org/cgi/collection/valve_disease Valve disease

following collection(s): This article, along with others on similar topics, appears in the

Permissions & Licensing

[email protected]: orhttp://www.us.elsevierhealth.com/Licensing/permissions.jsp

in its entirety should be submitted to: Requests about reproducing this article in parts (figures, tables) or

Reprints [email protected]

For information about ordering reprints, please email:


http://ats.ctsnetjournals.org/cgi/content/full/74/3/650

http://ats.ctsnetjournals.org/cgi/content/full/74/3/650#BIBL

http://ats.ctsnetjournals.org/cgi/content/full/74/3/650#otherarticles

http://ats.ctsnetjournals.org/cgi/collection/valve_disease

http://www.us.elsevierhealth.com/Licensing/permissions.jsp


aortic root replacement with cryopreserved allograft for prosthetic valve endocarditis

Documents