Nai(ifaa

Fm

2

Journal of the American College of Cardiology Vol. 49, No. 19, 2007© 2007 by the American College of Cardiology Foundation ISSN 0735-1097/07/$32.00P

Platelet Surface CD62P and CD63, MeanPlatelet Volume, and Soluble/Platelet P-Selectinas Indexes of Platelet Function in Atrial FibrillationA Comparison of “Healthy Control Subjects”and “Disease Control Subjects” in Sinus Rhythm

Anirban Choudhury, MRCP, Irene Chung, MRCP, Andrew D. Blann, PHD,Gregory Y. H. Lip, MD, FACC

Birmingham, United Kingdom

Objectives The aim of this work was to comprehensively study the role of platelets in atrial fibrillation (AF), in relation to theunderlying cardiovascular diseases and type of AF, and to analyze the effect of antithrombotic treatment on dif-ferent aspects of platelet activation.

Background Platelet activation is present in nonvalvular AF, but there is debate whether this is due to AF itself and/or to un-derlying cardiovascular diseases.

Methods A total of 121 AF patients were compared with 65 “healthy control subjects” and 78 “disease control subjects”in sinus rhythm. Platelet activation was assessed using 4 different aspects of platelet pathophysiology: 1) plate-let surface expression of CD62P (P-selectin) and CD63 (a lysosomal glycoprotein) (by flow cytometry); 2) meanplatelet volume (MPV) (by flow cytometry); 3) plasma levels of soluble P-selectin (sP-selectin, enzyme-linkedimmunoadsorbent assay); and 4) total amount of P-selectin per platelet (pP-selectin) (“platelet lysis” assay).

Results Both AF patients and “disease control subjects” had higher levels of CD62P (p � 0.001), CD63 (p � 0.001), andsP-selectin (p � 0.001) compared with “healthy control subjects,” with no difference between AF patients and“disease control subjects.” Patients with permanent AF had higher levels of sP-selectin (p � 0.014) and MPV(p � 0.025) compared with those with paroxysmal AF. The presence of AF independently affected the levels ofCD62P expression, while “high-risk” AF patients (CHADS score �2) had higher levels of CD62P compared withthose with “low risk.” Introducing warfarin resulted in a reduction of pP-selectin (p � 0.013).

Conclusions There is a degree of excess of platelet activation in AF compared with “healthy control subjects,” but no signifi-cant difference between AF patients and “disease control subjects” in sinus rhythm. Platelet activation may dif-fer according to the subtype of AF, but this is not in excess of the underlying comorbidities that lead to AF.Platelet activation in AF may be due to underlying cardiovascular diseases, rather than due to AF per se.(J Am Coll Cardiol 2007;49:1957–64) © 2007 by the American College of Cardiology Foundation

ublished by Elsevier Inc. doi:10.1016/j.jacc.2007.02.038

ab

bccsmuadea

onvalvular atrial fibrillation (AF) is associated with anpproximately 5-fold increase in stroke risk, which is highern association with clinical and echocardiographic variables1,2). The pathophysiology of thromboembolism in AFs multifactorial, but increasing evidence points to theulfillment of Virchow’s triad, with flow abnormalities,bnormalities of the vessel wall (endothelial/endocardialbnormalities), and abnormal blood constituents (platelets

rom the Haemostasis Thrombosis and Vascular Biology Unit, University Depart-ent of Medicine, City Hospital, Birmingham, United Kingdom.

aManuscript received November 27, 2006; revised manuscript received January 29,

007, accepted February 5, 2007.

nd the clotting cascade) present, resulting in a prothrom-otic/hypercoagulable state (3).Platelets play a role in rendering AF more “prothrom-

otic” by interacting with the endothelium, proteins of theoagulation cascade, and inflammatory cells (e.g., mono-ytes) (4). On activation, the normal platelet undergoes aeries of characteristic changes that include shape change,embrane budding, adhesion, aggregation, release of gran-

lar contents, and thromboxane synthesis. Methods aimedt measuring platelet activation detect these changes eitherirectly or indirectly. For example, direct assays (e.g.,lectron microscopy, flow cytometry) allow platelets to bessessed in their physiological milieu of whole blood, and

re extremely sensitive, with the ability to detect as few as

ta

pcpnaiAatd

Abpud

suaawaPapgvPP“

M

OietnesdacshnbbdaBmCtvsccowpuFc3wcdlhenfpfl

osCbta1blm

1958 Choudhury et al. JACC Vol. 49, No. 19, 2007Platelets in Atrial Fibrillation May 15, 2007:1957–64

1% of activated platelets (5).Hence, such methods are partic-ularly useful in monitoring rapidchanges of activation status, but,in order to prevent/minimize ar-tifactual platelet activation, bloodhas to be collected with minimalstasis and analyzed promptly (6).In contrast, “indirect” assaysmostly measure the metabolitesof activated platelets in theplasma or urine (e.g., enzyme-linked immunoadsorbent assay[ELISA]). In general, as thesemetabolites are modified in vivobefore assay, and these assays arearguably not as sensitive in de-tecting dynamic changes as the“direct” assays, but at the sametime are less prone to be affectedby artifactual platelet activationduring blood collection. Hence,

hey are more useful in measuring “baseline” plateletctivation.

Thus, different platelet assays assess different aspects oflatelet pathophysiology, and the results depend on thelinical condition being investigated and the aspect oflatelet activation being measured. Unfortunately, there iso consensus on the best method of measuring plateletctivation, and an “ideal” study of platelet activation shouldnclude at least 1 of each of the “direct” and “indirect” assays.lthough there are numerous studies assessing platelet

ctivation in AF, the results are (unsurprisingly) inconsis-ent, thus reflecting difference between the assays and theifferent aspects of platelet function measured (7–12).The possibility also arises that platelet activation differs in

F depending on the type of AF and the use of antithrom-otic therapy (13–15). It also remains unclear to what extentlatelet activation in AF is due to AF itself and/or to thenderlying cardiovascular diseases (e.g., coronary arteryisease [CAD], hypertension, diabetes, and stroke).The aim of the present study was to comprehensively

tudy the role of platelets in AF, especially in relation to thenderlying cardiovascular diseases, type of AF, and tonalyze the effect of antithrombotic treatment on differentspects of platelet activation. To achieve this, AF patientsere compared not only with “healthy control subjects,” but

lso with “disease control subjects” (in sinus rhythm).latelet activation was assessed by looking into 4 differentspects of platelet pathophysiology: 1) platelet surface ex-ression of CD62P (P-selectin) and CD63 (lysosomallycoprotein) (using flow cytometry); 2) mean plateletolume (MPV) (flow cytometry); 3) plasma levels of soluble-selectin (sP-selectin) (ELISA); and 4) total amount of-selectin per platelet (pP-selectin) (as quantified with a

Abbreviationsand Acronyms

AF � atrial fibrillation

CAD � coronary arterydisease

ELISA � enzyme-linkedimmunoadsorbent assay

MPV � mean plateletvolume

PAF � paroxysmal atrialfibrillation

PBS � phosphate-bufferedsolution

PFP � platelet-free plasma

pP-selectin � total amountof P-selectin per platelet

PRP � platelet-rich plasma

sP-selectin � solubleP-selectin

�TG � beta-thromboglobulin

platelet lysis” assay). i

ethods

utpatients attending our AF clinic in our hospital werenvited to participate in our study. All had AF diagnosed onlectrocardiogram before recruitment. Exclusion criteria forhe study included refusal of consent, any history of malig-ancy, connective tissue disease, thyroid disease, renal dis-ase, use of steroids, hormone replacement therapy, activemoking, and any concomitant “active/acute” cardiovascularisease. “Disease control subjects” were recruited from patientsttending our CAD, hypertension, and general cardiologylinics. “Healthy control subjects” were normotensive and ininus rhythm, who were not taking regular medications andad no evidence of disease on careful history, clinical exami-ation, and basic blood tests. To assess the effect of antithrom-otic therapy, we also recruited AF patients whose antithrom-otic treatment was changed from aspirin to warfarin. Pairedata from these patients were used to determine the effect ofntithrombotic treatment on indexes of platelet activation.lood collection. Blood was collected according to theethod described in the European Working Group onlinical Analysis Consensus Protocol on platelet flow cy-

ometry (6): 25 ml of blood was taken from an antecubitalein with minimal stasis into vacutainers tubes (BD Bio-ciences, Oxford, United Kingdom). Aliquots of blood wereollected into citrated (containing 0.5 ml of 3.8% sodiumitrate) and CTAD (containing citrate, theophylline, aden-sine, and dipyridamole) vacutainers. The citrated samplesere to be used for platelet lysate assay and measuringlasma levels of sP-selectin, while the CTAD samples weresed for flow cytometric work.low cytometry. Blood for platelet flow cytometry wasollected as described in the preceding text (6). CaliBRITE-color beads and FACSComp software (BD Biosciences)ere used to set photomultiplier tube voltages, fluorescence

ompensation, spectral overlap, and sensitivity. Size-ependent forward scatter and 90° side scatter were set at

ogarithmic gain. Mouse IgG2a and IgG1 antikeyhole limpetemocyanin monoclonal antibodies conjugated to phyco-rythrin were used as isotype-negative controls to defineonspecific binding. Monoclonal antibodies were obtainedrom BD Biosciences. Data acquisition and analysis wereerformed with CELLQuest software version 3.1 in theow cytometer (FACScan, BD Biosciences).Flow cytometric analysis was commenced within 15 min

f blood collection, to prevent artificial elevated levelsecondary to in-vitro platelet activation; 100 �l of theTAD sample was diluted with 800 �l of phosphate-uffered solution (PBS). Two separate aliquots of 15 �l ofhis diluted sample were then incubated with 10 �l ofnti–CD42b-FITC (binds to platelet glycoprotein 1b) and0 �l of anti–CD62P-PE (binds to platelet membraneound P-selectin) or anti–CD63-PE (binds to plateletysosomal membrane glycoprotein). After incubation for 30

in, the samples were further diluted with 800 �l of PBS

mmediately before flow cytometric analysis. Results were

eeewPp“wmwa[l

dNlteDtarEofc(atnPawhead“msarcghirsDDmewgn

nTwwit“wfwmc

R

Wiac(c“ia“CcCwbrAcj2(

a0(wacnsA(rlcetEti

1959JACC Vol. 49, No. 19, 2007 Choudhury et al.May 15, 2007:1957–64 Platelets in Atrial Fibrillation

xpressed as the percentage of CD62P- or CD63-positivevents compared with the total number of CD42b-positivevents. The inter- and intra-assay coefficients of variationere �5% and �8%, respectively.latelet lysate. The mass of pP-selectin per platelet (i.e.,latelet P-selectin/pP-selectin) was measured using aplatelet lysate assay,” as follows: platelet-rich plasma (PRP)as obtained by centrifugation of citrated blood within 15in of collection, at 1,000 rpm for 10 min; 1 ml of the PRPas centrifuged for a further 20 min at 3,000 rpm to obtainplatelet pellet. The supernatant (platelet-free plasma

PFP]) was stored at �70°C for measurement of plasmaevels of sP-selectin by ELISA.

The concentration of platelets (from the PRP) wasetermined using ADVIA 120 hematology system (Bayer,ewbury, United Kingdom). Accordingly, the platelet pel-

et was resuspended in saline to achieve a fixed concentra-ion of 200 � 106 platelets per ml, and then lysed with anqual volume of 0.1% Triton X-100 (Sigma/Aldrich, Poole,orset, United Kingdom). Aliquots were stored at �70°C

o allow batch analysis of pP-selectin by ELISA. The inter-nd intra-assay coefficients of variation were �5% and 10%,espectively.LISA for P-selectin. Levels of sP-selectin (1/5 dilutionf plasma) from the PFP and pP-selectin (1/10 dilution)rom the lysed platelet sample were derived from standardurves using the commercial ELISA kit from R&D SystemsAbingdon, Oxon, United Kingdom). The inter- and intra-ssay coefficient of variation were �5% and 10%, respec-ively. The lower limit of sensitivity of the assay was 0.8g/ml.ower calculation. Previous data on markers of plateletctivation reveal that some are normally distributed (14–17)hile others have a non-normal distribution (13,15,18). Weypothesized that indexes of platelet activation will belevated by: 1) 0.25 of a SD for indexes normally distributednd by 50% of the median values in indexes non-normallyistributed, in “disease control subjects” compared withhealthy control subjects”; and by 2) 0.5 SD and 100% of theedian in AF patients compared with “healthy control

ubjects.” Using sP-selectin as the reference molecule, tochieve a difference of p � 0.05 and a power of 0.80, weequired 60 recruits in each group. We therefore recruitedonsecutive patients and controls until we had 60 in eachroup and then exceeded it for extra confidence. We alsoypothesized that levels of indexes of platelet activation will

ntercorrelate with sP-selectin, to a degree of a Spearman’sank correlation of at least 0.4, which we consider asignificant.

ata analysis and statistics. After application of Anderson-arling test, normally distributed data were expressed asean (SD), while non-normally distributed data were

xpressed as median (interquartile range). Categorical dataere analyzed by chi-square test. Differences betweenroups were analysed by Kruskal-Wallis test (for non-

ormally distributed data) or 1-way analysis of variance (for w

ormally distributed data). For intergroup differences,ukey’s post hoc test was used, using log-transformed datahere appropriate. As all research indexes other than MPVere non-normally distributed, all correlations were accord-

ng to Spearman’s rank correlation method. Stepwise mul-iple regression analysis on the combined cohort of AF anddisease control subjects” was performed to determinehether the presence/absence of AF and other clinical

eatures (again, on log-transformed data where appropriate)as independently predictive of the levels of the plateletarkers under investigation. A probability of �0.05 was

onsidered statistically significant.

esults

e invited 150 consecutive patients with AF to participaten our study. Of these, 9 declined our invitation whilenother 20 could not be recruited because of our exclusionriteria. Thus, we recruited a total of 121 patients with AF58 paroxysmal AF [PAF] and 63 permanent AF) andompared them with 78 “disease control subjects” and 65healthy control subjects” (Table 1). There were no signif-cant differences in underlying cardiovascular comorbiditiesnd antiplatelet drug usage between the AF patients anddisease control subjects.”ross-sectional analyses. Both AF patients and “disease

ontrol subjects” had higher levels of CD62P (p � 0.001),D63 (p � 0.001), and sP-selectin (p � 0.001) comparedith “healthy control subjects.” There were no differencesetween AF patients and “disease control subjects” withegards to the markers as discussed earlier (Table 2).lthough AF patients had higher levels of pP-selectin

ompared with both “healthy” and “disease control sub-ects,” this did not achieve significance (p � 0.118) (Table). There was no difference in MPV between the 3 groupsp � 0.213) (Table 2).

Among AF patients, those with permanent AF had moressociated cardiovascular disease (87.3% vs. 61.4%; p �.001), as well as significantly higher levels of sP-selectinp � 0.014) and MPV (p � 0.025) compared with thoseith PAF (Table 3, top). Though permanent AF patients

lso had higher levels of CD62P, CD63, and pP-selectinompared with PAF patients, these small differences wereot statistically significant (Table 3, top). There were noignificant differences in any of the platelet indexes amongF patients on aspirin compared with those on warfarin

Table 3, middle). Atrial fibrillation patients with a higherisk of stroke (CHADS2 score of �2) (19,20) had higherevels of CD62P (p � 0.003) and lower MPV (p � 0.020)ompared with “low-risk” AF patients. There were differ-nces in CD63, sP-selectin, and pP-selectin levels betweenhe 2 groups (Table 3, bottom).ffect of introducing antithrombotic treatment. To de-

ermine the effects of antithrombotic treatment on thendexes of platelet activation, we recruited 11 AF patients

hose antithrombotic therapy was changed from aspirin to

wwacl(Ch3

ppngnC

cp

B

VIb

D � hypep

I

Vl†

1960 Choudhury et al. JACC Vol. 49, No. 19, 2007Platelets in Atrial Fibrillation May 15, 2007:1957–64

arfarin (as per clinical risk stratification). Switching toarfarin therapy was associated with a reduction in the total

mount of pP-selectin (p � 0.013) along with nonsignifi-ant reductions of CD62P and CD63 expression on plate-ets and in the total amount of sP-selectin in the plasmaTable 4).

orrelations. Platelet expression of CD62P and CD63ad a strong positive correlation (r � 0.6; p � 0.001 in allgroups). Among “disease control subjects,” CD62P ex-

aseline Characteristic in Nonvalvular AF Patients and Control Sub

Table 1 Baseline Characteristic in Nonvalvular AF Patients and

“Healthy Control Subjects” “Disease Cont

Numbers 65 78

Age in years 61.02 (10.97) 64.06 (1

Male gender, n (%) 43 (66.2) 57 (7

BMI 26.5 (3.7) 27.8 (3

SBP 133.3 (13.1) 138.9 (1

DBP 78.3 (8.5) 80.2 (9

Cholesterol (mmol/l) 5.5 (0.8) 4.6 (1

WBC (�103/mm3) 6.4 (1.7) 7.1 (2

Hb (g%) 14.0 (1.2) 13.9 (1

Hct (%) 40.6 (33.7) 41.6 (3

Platelets (�108/mm3) 266.9 (56.1) 261.1 (6

Comorbidities

IHD, n (%) — 22 (2

HT, n (%) — 54 (6

CVA/TIA, n (%) — 6 (7

DM, n (%) — 10 (1

PVD, n (%) — 1 (1

Treatment

Warfarin, n (%) — 5 (6

Aspirin, n (%) — 37 (4

Clopidogrel, n (%) — 9 (1

ACEI or ARB, n (%) — 45 (5

Diuretics, n (%) — 23 (3

CCB, n (%) — 20 (2

Beta-blocker, n (%) — 34 (4

Statin, n (%) — 50 (6

alues are expressed as mean (SD) or number (%). Analyses done by chi-square test (for categoricntergroup differences with Tukey’s test (p � 0.05): *difference lies between “healthy” and “diseasetween “disease control subjects” and AF patients.ACEI � angiotensin-converting enzyme inhibitor; AF � atrial fibrillation; ARB � angiotensin rece

BP � diastolic blood pressure; DM � diabetes mellitus; Hb � hemoglobin; Hct � hematocrit; HTressure; TIA � transient ischemic attack; WBC � white blood cell.

ndexes of Platelet Activation in Patients With AF Compared With

Table 2 Indexes of Platelet Activation in Patients With AF Com

Marker of Platelet Activation “Healthy Control Subjects” “Disease Cont

n 65 7

CD62P (% gated) 1.21 (0.69–2.17) 1.71 (1.0

CD63 (% gated) 4.68 (2.10–6.62) 5.26 (3.1

sP-selectin (ng/ml) 40 (30–50) 57 (36

Lysate P-selectin (ng/ml) 300 (180–420) 320 (19

MPV (fl) 7.39 (�0.96) 7.76 (�0

alues are expressed as mean (�SD) or median (1st–3rd interquartile range) and analyzed by 1-og-transformed data) where appropriate. Significant p values are indicated in bold. Intergroup dif

difference lies between “healthy control subjects” and atrial fibrillation (AF) patients.MPV � mean platelet volume; sP-selectin � soluble P-selectin.

ression negatively correlated with pP-selectin (r � �0.34;� 0.005) and MPV (r � �0.26; p � 0.046). These

egative correlations were also noted among the other 2roups, but they were not statistically significant. No sig-ificant correlations were found between sP-selectin andD62P expression or pP-selectin, in any of the 3 groups.Stepwise multiple regression analysis on the combined

ohort of AF and “disease control subjects” revealed thatresence of AF was independently associated with CD62P

trol Subjects

jects” AF Patientsp Value (“Healthy Control Subjects” vs.

“Disease Control Subjects” vs. AF Patients)

121 —

62.58 (8.6) 0.176

91 (76.0) 0.355

27.1 (3.9) 0.144

133.3 (14.4) 0.013*

78.8 (9.2) 0.439

4.8 (0.9) �0.001*†

7.1 (1.7) 0.07

14.6 (1.6) 0.007‡

42.3 (4.3) 0.316

259.9 (66.3) 0.811

29 (23.9) �0.001*†

77 (63.6) �0.001*†

12 (10.0) 0.034†

10 (8.2) 0.014*

2 (1.7) 0.587*

54 (44.6) �0.001†‡

46 (37.2) �0.001*†

7 (5.8) 0.015*

65 (53.7) �0.001*†

43 (35.5) �0.001*†

46 (38.0) �0.001*†

53 (43.8) �0.001*†

35 (28.9) �0.001*†‡

) or 1-way analysis of variance (for continuous data) and Tukey’s post hoc test where appropriate.ol subjects”; †difference lies between “healthy control subjects” and AF patients; ‡difference lies

cker; BMI � body mass index; CCB � calcium-channel blocker; CVA � cerebrovascular accident;rtension; IHD � ischemic heart disease; PVD � peripheral vascular disease; SBP � systolic blood

lthy” and “Disease Control Subjects”

d With “Healthy” and “Disease Control Subjects”

bjects” AF Patientsp Value (“Healthy Control Subjects” vs.

“Disease Control Subjects” vs. AF Patients)

121 —

0) 2.43 (1.30–5.32) <0.001*†

4) 6.08 (3.94–10.72) <0.001*†

59 (42–84) <0.001*†

) 380 (227–600) 0.118

7.56 (�1.38) 0.213

lysis of variance or Kruskal-Wallis test, with intergroup differences by Tukey’s post hoc test (ons with Tukey’s test (p � 0.05): *difference lies between “healthy” and “disease control subjects”;

jects

Con

rol Sub

0.29)

3.1)

.7)

3.9)

.8)

.4)

.1)

.6)

.9)

3.5)

8.2)

9.2)

.7)

2.8)

.2)

.4)

7.4)

1.5)

7.7)

0.8)

5.6)

3.6)

4.1)

al datae contr

ptor blo

“Hea

pare

rol Su

8

0–3.8

7–9.3

–74)

5–540

.98)

way anaference

e0M

D

Oimaa

tcwtmps

eleeeraptm�pslat

d cardio

Eo

Vni

1961JACC Vol. 49, No. 19, 2007 Choudhury et al.May 15, 2007:1957–64 Platelets in Atrial Fibrillation

xpression (p � 0.029), but not CD63 expression (p �.166), sP-selectin (p � 0.097), pP-selectin (p � 0.397), orPV (p � 0.288).

iscussion

ur results suggest that there is excess of platelet activationn AF patients compared with “healthy control subjects” as

easured by sP-selectin levels and expression of CD62Pnd CD63, but not in terms of total amount of pP-selectinnd MPV. However, there was no excess of platelet activa-

Assessment of Platelet Activation Among AF Pa

Table 3 Assessment of Platelet Activation A

Between Paroxysmal a

Marker of Platelet Activation Paroxysmal

Numbers 58

CD62P (% gated) 2.43 (1.04–5.

CD63 (% gated) 5.58 (3.47–10

sP-selectin (ng/ml) 51 (38–70)

Lysate P-selectin (ng/ml) 340 (220–48

MPV (pg) 7.27 (�0.79)

Aspirin and War

Marker of Platelet Activation AF Patients on

Numbers 49

CD62P (% gated) 2.42 (1.13–5

CD63 (% gated) 6.24 (4.00–9

sP-selectin (ng/ml) 54 (42–72)

Lysate P-selectin (ng/ml) 410 (258–54

MPV (pg) 7.38 (�0.90)

Between High-Risk (CHADS Score >2) a

Marker of Platelet Activation CHADS Score

Numbers 80

CD62P (% gated) 1.92 (1.05–4.

CD63 (% gated) 6.37 (4.92–11

sP-selectin (ng/ml) 55 (39–78)

Lysate P-selectin (ng/ml) 390 (248–62

MPV (pg) 7.76 (�1.51)

Values are expressed as mean (�SD) or median (1st–3rd interquartileare indicated in bold. *Paroxysmal AF was defined as patients with dwhile permanent AF was defined where AF was present for �1 year an

Abbreviations as in Table 2.

ffect of Aspirin or Warfarin Therapyn Platelet Activation in Atrial Fibrillation

Table 4 Effect of Aspirin or Warfarin Therapyon Platelet Activation in Atrial Fibrillation

Platelet Marker On Aspirin4 Weeks After

Swapping to Warfarin p Value

Numbers 11 11 —

CD62P (% gated) 3.62 (1.62–6.38) 3.31 (1.08–3.73) 0.554

CD63 (% gated) 5.61 (3.93–7.04) 4.75 (2.35–11.01) 0.676

sP-selectin (ng/ml) 54 (43–74) 48 (42–64) 0.906

Lysate P-selectin(ng/ml)

503 (340–1,038) 360 (190–400) 0.013

MPV (pg) 7.36 (�1.01) 7.36 (�1.41) 0.829

alues expressed as median (interquartile range); paired data analyzed by Wilcoxon (for non-ormally distributed data) or paired t test (for normally distributed data). Significant p values are

hndicated in bold.

MPV � mean platelet volume.

ion (as measured by 4 different methods) in AF patientsompared with “disease control subjects.” This is consistentith the view that platelet activation in AF is more related

o the associated vascular diseases, rather than the arrhyth-ia itself. On a stepwise multiple regression analysis,

resence of AF did not significantly affect CD63, sP-electin, pP-selectin, or MPV levels.

The role of platelets in atherothrombotic diseases is wellstablished (21). However, the relative significance of plate-et activation in the overall atherothrombotic risk in differ-nt cardiovascular diseases seems to vary. Although there isvidence suggesting that platelets are activated in AF, thevidence that it relates directly to the increased thromboticisk in AF is uncertain. Initial studies had reported associ-tions between beta-thromboglobulin (�TG) (a marker oflatelet activation) (22) and CD62P with intra-atrialhrombus and echo-contrast (23). However, a subsequentuch larger study (7) found no association between plasmaTG levels and subsequent thromboembolic events. Also,lasma levels of sP-selectin were unrelated to estimatedtroke risk (14), despite associations between sP-selectinevels and atherothrombotic risk factors, such as smokingnd peripheral vascular disease. Moreover, plasma markershat seem to have prognostic importance in identifying

s

g AF Patients

rmanent AF Patients*

Permanent AF p Value

63 —

2.50 (1.34–5.71) 0.926

6.69 (4.32–11.01) 0.208

64 (44–104) 0.014

400 (242–624) 0.216

7.84 (�1.72) 0.025

se at Baseline

AF Patients on Warfarin p Value

54 —

2.50 (1.32–5.25) 0.798

6.63 (4.60–12.98) 0.219

61 (45–112) 0.146

315 (185–520) 0.086

7.46 (�1.26) 0.697

w-Risk (CHADS Score <2) AF Patients

CHADS Score >2 p Value

41 —

3.78 (2.10–6.99) 0.003

5.79 (4.45–9.92) 0.792

64 (46–100) 0.115

380 (190–520) 0.346

7.20 (�1.01) 0.020

and analyzed by Mann-Whitney or 2-sample t test. Significant p valuested paroxysms of AF lasting �100 beats on 24-h Holter monitoring,version was considered inappropriate or was previously unsuccessful.

tient

mon

nd Pe

AF

32)

.20)

0)

farin U

Aspirin

.35)

.79)

4)

nd Lo

<2

72)

.24)

1)

range)ocumen

igh-risk AF patients (e.g., fibrin D-dimer and von Wille-

bcUltatcbr

siwAhrabpdcpt

aefdacvPigeuetrEmm

wkvnmlgiplah

mp

wcfbrCCdbesep(p(cdtawitaimemf

trtplacpbwr

i“Cehcha

1962 Choudhury et al. JACC Vol. 49, No. 19, 2007Platelets in Atrial Fibrillation May 15, 2007:1957–64

rand factor) (14,24) are directly or indirectly linked to theoagulation cascade rather than platelet activation per se.nsurprisingly, the efficacy of warfarin in preventing the

ikelihood of strokes in AF is significantly superior not onlyo aspirin (25) but also compared with the combination ofspirin and clopidogrel (26). Of note, the 22% risk reduc-ion (of strokes) with aspirin just reaches statistical signifi-ance, and it has been suggested to be secondary to theeneficial role of aspirin on the underlying vascular diseaseather than on AF itself (25).

Over the last decade, the MPV has been established as aimple and practical method of assessing platelet activationn hypertension, CAD, diabetes, and stroke (16,17,27,28),hich are all diseases that frequently coexist with AF.mong patients with established cardiovascular diseases,igher levels of MPV have been shown to identify “high-isk” patients and those more likely to have poorer outcomefter an acute vascular event (29–35). Until now, there haveeen no previous reports on the behavior of MPV inatients with nonvalvular AF, but the lack of significantifference in MPV between AF patients and “healthyontrol subjects” is in keeping with platelet activationlaying a secondary role in rendering AF more “pro-hrombotic.”

The finding of no significant differences in pP-selectinmong patients and control subjects could potentially bexplained by 2 reasons. First, activated platelets result inusion of �-granules with the platelet surface, followed byegranulation. This would explain higher levels of CD62Pnd sP-selectin in AF patients and “disease control subjects”ompared with “healthy control subjects.” However, thisery process might result in depletion of the total amount of-selectin in the platelets, sufficient to render the difference

n pP-selectin statistically insignificant between the 3roups. Second, it has been suggested that P-selectin,xpressed on the surface of the platelet, changes its config-ration upon platelet activation (36,37). Indeed, Semenovt al. (38) demonstrated that the polyclonal antibody used inheir study to detect sP-selectin in platelets had a lowereactivity toward membrane P-selectin. As we have used theLISA that detects sP-selectin, it is possible that the assayight have detected granular P-selectin, but not all of theembrane expressed P-selectin.Our finding of higher levels of sP-selectin in patients

ith permanent AF compared with those with PAF is ineeping with previous reports (13). We also found higheralues of MPV in permanent AF compared with PAF, buto difference between the 2 groups with regards to plateletembrane expression of CD62P and CD63, or pP-selectin

evels. It has been suggested that sP-selectin reflects back-round chronic platelet activation, while CD62P expressions associated with acute changes (39,40). Patients withermanent AF in our study had increased associated under-

ying cardiovascular disease compared with those with PAF,nd it is possible that the associated comorbidities might

ave resulted in excess of background platelet activation, as f

anifested by higher levels of sP-selectin and MPV inermanent AF.Among the different platelet markers, CD62P and CD63

ere strongly correlated in all the 3 groups supporting theoncept that platelet activation results in migration andusion of � and lysosomal granules with the platelet mem-rane with subsequent expression of CD62P and CD63,espectively. The negative correlation noted betweenD62P expression and pP-selectin suggests expression ofD62P (and its subsequent release in the plasma) results inepletion of pP-selectin. However, no correlation was notedetween CD62P and sP-selectin. There can be 3 potentialxplanations for this. First, other than active cleavage orhedding of platelet membrane-bound P-selectin, there isvidence that sP-selectin may be released from thelatelets by “direct” secretion (36,37) or passive diffusion41). Moreover, some contribution from the endothelium isossible, although remote (42–44). Second, Ferroni et al.40), in the setting of cardiopulmonary bypass, reported noorrelation of CD62P and sP-selectin in view of theifferent time course of the rise and fall (normalization) ofhese molecules. Similar lack of correlation between CD62Pnd sP-selectin has been reported in patients presentingith acute chest pain (45) and in patients with myocardial

nfarction after thrombolysis (46) leading to suggestionshat CD62P might reflect platelet activation in response tocute stimuli, while sP-selectin is a better marker in detect-ng circulating activated platelets in clinical settings where

ore “chronic” stimuli are present (39). Third, there is alsoxperimental evidence to suggest that activated plateletsight completely loose surface P-selectin, but continue to

unction and circulate (47).We also found that AF patients whose antithrombotic

herapy was changed from aspirin to warfarin resulted ineduction in pP-selectin along with nonsignificant reduc-ions in 3 out of 4 other platelet markers. Aspirin has beenreviously shown to have no significant effect on sP-selectinevels (48–50), as P-selectin release from platelets isdenosine-diphosphate-dependent and independent of theyclooxygenase pathway. There are also reports of excesslatelet expression of P-selectin after stimulation by throm-in receptor-activating peptide (51). Thus, it is possible thatarfarin by reducing circulating thrombin might result in

eduction of overall platelet activation on AF.Among the different platelet markers, presence of AF

ndependently affected the levels of CD62P expression.High-risk” AF patients had significantly higher levels ofD62P compared with the “low-risk” patients. Pongratz

t al. (23) has previously reported an association betweenigher levels of CD62P with intra-atrial echocardiographyontrast and thrombus. Whether this is a chance finding origher levels of CD62P can truly identify AF patients withhigher stroke risk can only be ascertained in longitudinal

ollow-up studies.

RUBa

R

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

1963JACC Vol. 49, No. 19, 2007 Choudhury et al.May 15, 2007:1957–64 Platelets in Atrial Fibrillation

eprint requests and correspondence: Prof. Gregory Y. H. Lip,niversity Department of Medicine, Dudley Road, City Hospital,irmingham B18 7QH, United Kingdom. E-mail: g.y.h.lip@bham.c.uk.

EFERENCES

1. Kannel WB, Wolf PA, Benjamin EJ, Levy D. Prevalence, incidence,prognosis, and predisposing conditions for atrial fibrillation:population-based estimates. Am J Cardiol 1998;82:2N–9N.

2. Atrial Fibrillation Investigators. Risk factors for stroke and efficacy ofantithrombotic therapy in atrial fibrillation: analysis of pooled datafrom five randomised controlled trials. Arch Intern Med 1994;154:1449–57.

3. Lip GYH. The prothrombotic state in atrial fibrillation: the atrium,the endothelium . . . and tissue factor? Thromb Res 2003;111:133–5.

4. Blann AD. Angiogenesis and platelets: the clot thickens further.Cardiovasc Res 2004;63:192–3.

5. Kestin AS, Ellis PA, Barnard MR, Errichetti A, Rosner BA, Mich-elson AD. Effect of strenuous exercise on platelet activation state andreactivity. Circulation 1993;88:1502–11.

6. Schmitz G, Rothe G, Ruf A, et al. European Working Group onClinical Cell Analysis: consensus protocol for the flow cytometriccharacterisation of platelet function. Thromb Haemost 1998;79:885–96.

7. Feinberg WM, Pearce LA, Hart RG, et al. Markers of thrombin andplatelet activity in patients with atrial fibrillation: correlation withstroke among 1531 participants in the stroke prevention in atrialfibrillation III study. Stroke 1999;30:2547–53.

8. Gustafsson C, Blomback M, Britton M, Hamsten A, Svensson J.Coagulation factors and the increased risk of stroke in nonvalvularatrial fibrillation. Stroke 1990;21:47–51.

9. Lip GY, Lip PL, Zarifis J, et al. Fibrin D-dimer and beta-thromboglobulin as markers of thrombogenesis and platelet activationin atrial fibrillation. Effects of introducing ultra-low-dose warfarin andaspirin. Circulation 1996;94:425–31.

0. Minamino T, Kitakaze M, Sanada S, et al. Increased expression ofP-selectin on platelets is a risk factor for silent cerebral infarction inpatients with atrial fibrillation: role of nitric oxide. Circulation 1998;98:1721–7.

1. Yamauchi K, Furui H, Taniguchi N, Sotobata I. Plasma beta-thromboglobulin and platelet factor 4 concentrations in patients withatrial fibrillation. Jpn Heart J 1986;27:481–7.

2. Furui H, Taniguchi N, Yamauchi K, Sotobata I, Saito H, Inagaki H.Effects of treadmill exercise on platelet function, blood coagulabilityand fibrinolytic activity in patients with atrial fibrillation. Jpn Heart J1987;28:177–84.

3. Li-Saw-Hee FL, Blann AD, Gurney D, Lip GY. Plasma vonWillebrand factor, fibrinogen and soluble P-selectin levels in paroxys-mal, persistent and permanent atrial fibrillation. Effects of cardiover-sion and return of left atrial function. Eur Heart J 2001;22:1741–7.

4. Conway DS, Pearce LA, Chin BS, Hart RG, Lip GY. Plasma vonWillebrand factor and soluble p-selectin as indices of endothelialdamage and platelet activation in 1321 patients with nonvalvular atrialfibrillation: relationship to stroke risk factors. Circulation 2002;106:1962–7.

5. Kamath S, Blann AD, Caine GJ, Gurney D, Chin BS, Lip GY.Platelet P-selectin levels in relation to plasma soluble P-selectin andbeta-thromboglobulin levels in atrial fibrillation. Stroke 2002;33:1237–42.

6. Coban E, Yazicioglu G, Berkant Avci A, Akcit F. The mean plateletvolume in patients with essential and white coat hypertension. Platelets2005;16:435–8.

7. Henning BF, Zidek W, Linder B, Tepel M. Mean platelet volume andcoronary heart disease in hemodialysis patients. Kidney Blood PressRes 2002;25:103–8.

8. Tan KT, Tayebjee MH, Macfadyen RJ, Lip GY, Blann AD. Elevatedplatelet microparticles in stable coronary disease are unrelated todisease severity or to indices of inflammation. Platelets 2005;16:368 –71.

9. Gage BF, Waterman AD, Shanon W, Boechler M, Rich MW,Radford MJ. Validation of clinical classification schemes for predicting

stroke: results from the National Registry of Atrial Fibrillation. JAMA2001:285:2864–70.

0. Fuster V, Rydén LE, Cannon DS, et al. ACC/AHA/ESC 2006guidelines for the management of patients with atrial fibrillation—executive summary: a report of the American College of Cardiology/American Heart Association Task Force and the European Society ofCardiology Committee for Practice Guidelines (Writing Committeeto Revise the 2001 Guidelines for the Management of Patients WithAtrial Fibrillation). J Am Coll Cardiol 2006;48:854–906.

1. Ruggeri ZM. Platelets in atherothrombosis. Nat Med 2002;8:1227–34.

2. Heppell RM, Berkin KE, McLenachan JM, Davies JA. Haemostaticand haemodynamic abnormalities associated with left atrial thrombosisin non-rheumatic atrial fibrillation. Heart 1997;77:407–11.

3. Pongratz G, Brandt-Pohlmann M, Henneke KH, et al. Plateletactivation in embolic and preembolic status of patients with nonrheu-matic atrial fibrillation. Chest 1997;111:929–33.

4. Vene N, Mavri A, Kosmelj K, Stegnar M. High d-dimer levels predictcardiovascular events in patients with chronic atrial fibrillation duringoral anticoagulation therapy. Thromb Haemost 2003;90:1163–72.

5. Lip GYH, Edwards SJ. Stroke prevention with aspirin, warfarin andximelagatran in patients with non-valvular atrial fibrillation: a system-atic review and meta-analysis. Thromb Res 2006;118:321–33.

6. ACTIVE Writing Group on behalf of the ACTIVE Investigators,Connolly S, Pogue J, Hart R, et al. Clopidogrel plus aspirin versusoral anticoagulation for atrial fibrillation in the Atrial fibrillationClopidogrel Trial with Irbesartan for prevention of Vascular Events(ACTIVE W): a randomised controlled trial. Lancet 2006;367:1903–12.

7. Inui Y, Suehiro T, Kumon Y, Hashimoto K. Platelet volume andurinary prostanoid metabolites in non-insulin-dependent diabetesmellitus. J Atheroscler Thromb 1994;1:108–12.

8. Nadar SK, Lip GY, Blann AD. Platelet morphology, soluble P selectinand platelet P-selectin in acute ischaemic stroke. The West Birming-ham Stroke Project. Thromb Haemost 2004;92:1342–8.

9. Endler G, Klimesch A, Sunder-Plassmann H, et al. Mean plateletvolume is an independent risk factor for myocardial infarction but notfor coronary artery disease. Br J Haematol 2002;117:399–404.

0. Glud T, Schmidt EB, Kristensen SD, Arnfred T. Platelet number andvolume during myocardial infarction in relation to infarct size. ActaMed Scand 1986;220:401–5.

1. Huczek Z, Kochman J, Filipiak KJ, et al. Mean platelet volume onadmission predicts impaired reperfusion and long-term mortality inacute myocardial infarction treated with primary percutaneous coro-nary intervention. J Am Coll Cardiol 2005;46:284–90.

2. Yang A, Pizzulli L, Luderitz B. Mean platelet volume as marker ofrestenosis after percutaneous transluminal coronary angioplasty inpatients with stable and unstable angina pectoris. Thromb Res2006;117:371–7.

3. Papanas N, Symeonidis G, Maltezos E, et al. Mean platelet volume inpatients with type 2 diabetes mellitus. Platelets 2004;15:475–8.

4. Bath P, Algert C, Chapman N, Neal B, PROGRESS CollaborativeGroup. Association of mean platelet volume with risk of stroke among3134 individuals with history of cerebrovascular disease. Stroke 2004;35:622–6.

5. Greisenegger S, Endler G, Hsieh K, Tentschert S, Mannhalter C,Lalouschek W. Is elevated mean platelet volume associated with aworse outcome in patients with acute ischemic cerebrovascular events?Stroke 2004;35:1688–91.

6. Schick PK, Konkle BA, He X, Thornton RD. P-selectin mRNA isexpressed at a later phase of megakaryocyte maturation than mRNAsfor von Willebrand factor and glycoprotein Ib-alpha. J Lab Clin Med1993;121:714–21.

7. Johnston GI, Cook RG, McEver RP. Cloning of GMP-140, a granulemembrane protein of platelets and endothelium: sequence similarity toproteins involved in cell adhesion and inflammation. Cell 1989;56:1033–44.

8. Semenov AV, Romanov YA, Loktionova SA, et al. Production ofsoluble P-selectin by platelets and endothelial cells. Biochemistry(Mosc) 1999;64:1326–35.

9. Ferroni P, Pulcinelli FM, Lenti L, Gazzaniga PP. Is soluble P selectindetermination a more reliable marker of in vivo platelet activation than

CD62P flow cytometric analysis? Thromb Haemost 1999;81:472–3.

4

4

4

4

4

4

4

4

4

4

5

5

1964 Choudhury et al. JACC Vol. 49, No. 19, 2007Platelets in Atrial Fibrillation May 15, 2007:1957–64

0. Ferroni P, Speziale G, Ruvolo G, et al. Correlation between plateletactivation and inflammatory response during hypothermic cardiopul-monary bypass. Thromb Haemost 1998;80:58–64.

1. Kostelijk EH, Fijnheer R, Nieuwenhuis HK, Gouwerok CW, deKorte D. Soluble P-selectin as parameter for platelet activation duringstorage. Thromb Haemost 1996;76:1086–9.

2. Kameda H, Morita I, Handa M, et al. Re-expression of functionalP-selectin molecules on the endothelial cell surface by repeatedstimulation with thrombin. Br J Haematol 1997;97:348–55.

3. Khew-Goodall Y, Butcher CM, Litwin MS, et al. Chronic expressionof P-selectin on endothelial cells stimulated by the T-cell cytokine,interleukin-3. Blood 1996;87:1432–8.

4. Gotsch U, Jager U, Dominis M, Vestweber D. Expression ofP-selectin on endothelial cells is upregulated by LPS and TNF-alphain vivo. Cell Adhes Commun 1994;2:7–14.

5. Gurbel PA, Kereiakes DJ, Serebruany VL. Soluble selectin is not asurrogate marker for platelet P selectin: evidence from a multicentrechest pain study group. J Thromb Thrombolysis 2000;10:15–22.

6. Serebruany VL, Gurbel PA. Assessment of platelet activity by mea-

suring platelet derived substances in plasma from patients with acute

myocardial infarction: surprising lessons from the GUSTO-IIIPlatelet Study. Thromb Res 1999;93:149–50.

7. Michelson AD, Barnard MR, Hechtman HB, et al. In vivo trackingof platelets: circulating degranulated platelets rapidly loose surfaceP selectin, but continue to circulate and function. Proc Natl Acad SciU S A 1996;93:11877–82.

8. Li N, Wallen NH, Hjemdahl P. Evidence for prothrombotic effects ofexercise and limited protection by aspirin. Circulation 1999;100:1374–9.

9. O’Connor CM, Gurbel PA, Serebruany VL. Usefulness of solubleand surface-bound P-selectin in detecting heightened plateletactivity in patients with congestive heart failure. Am J Cardiol1999;83:1345–9.

0. Jilma B, Blann A, Pernerstorfer T, et al. Regulation of adhesionmolecules during human endotoxemia. No acute effects of aspirin.Am J Respir Crit Care Med 1999;159:857–63.

1. Goette A, Ittenson A, Hoffmanns P, et al. Increased expression ofP-selectin in patients with chronic atrial fibrillation. Pacing Clin

Electrophysiol 2000;23:1872–5.