("

D lo1lu /*i.lu,I

Pharmacodynamic effectandclinicalefficacyof clopidogrel

inhibitor: with or withouta proton-pump andprasugrel

>@

trials of two randomised an analysis Rozenman, AlonDMichelson, ElliottM Antman,Sabina A Murphy,EricRBates,Yoseph LO'Donoghue, Eugene Braunwold, Michelle Sobotine, Stephen D Wiviott L Mega, Marc 5 Peter Ver Lee, Sondro L Close, Lei Shen, Houtvast, N RaymondW Jessica

5ummary Background Proton-pump inhibitors (PPIs) are often prescribed in combination with thienopyridines. Conflicting data exist as to whether PPIs diminish the efficacy of clopidogrel. We assessed the association between PPI use, measures of platelet function, and clinical outcomes for patients treated with clopidogrel or prasugrel. Methods In the PRINCIPLE-TIMI 44 trial, the primary outcome was inhibition of platelet aggregation at 6 h assessed by light-transmission aggregometry. In the TRITON-TIMI 38 trial, the primary endpoint was the composite of cardiovascular death, myocardial infarction, or stroke. In both studies, PPI use was at physician's discretion. We used a multivariable Cox model with propensity score to assessthe associanon of PPI use with clinical outcomes. Findings In the PRINCIPLE-TIMI 44 trial, 201 patients undergoing elective percutaneous coronary intervention were randomly assigned to prasugrel (n=102) or high-dose clopidogrel (n=99). Mean inhibition of platelet aggregation was significantly lower for patients on a PPI than for those not on a PPI at 6 h after a 600 mg clopidogrel loading dose (23.2t19.5% vs35.2x.20.9%".p=0.02), whereas a more modest difference was seen with and without a PPI after a 60 mg loading dose of prasugrel (69.6113.5%ovs76.7x.12.a%,p=0.05\. In the TRITON-TIMI 38 trial, 13608 patients with an acute coronary syndrome were randomly assigned to prasugrel (n=6813) or clopidogrel (n=6795). In this swdy,3!% @=a5291of patients were on a PPI at randomisation. No associationexisted between PPI use and risk of the primary endpoint for patients treated with clopidogrel (adiusted hazard ratio [HR] 0.94,95% CI 0'8&-1.11) or prasugrel(1.00,0. 84-1. Z0). lnterpretation The current findings do not support the need to avoid concomitant use of PPIs, when clinically indicated, in patients receiving clopidogrel or prasugrel. Funding Daiichi Sankyo Company Umited and Eli tilly and Company sponsored the trials. This analysis had no funding.

Introduction Both clopidogrel and prasugrel are prodrugs that are metabolised by the cytochrome P450 enzyme system to form their active metabolites. Prasugrel achievesgreater and more consistent platelet inhibition than,standard doses or higher doses of clopidogrel.'' This difference might partiy be causedby a more efficient conversion of prasugrel into its activemetabolite.' Proton-pump inhibitors (PPIs) are often administered in combination with thienopyridines to help reduce the risk ofgastrointestinal bleeding, a strategythat is endorsed by existing consensus guidelines.o However, several studies have raised concerns that many PPIs, especially omeprazole, might diminish the antiplatelet effects!7 and the clinicai effectivenesss'oofclopidogrel, possibly through inhibition ofthe hepaticcltochrome P4502C19(CYP2C19) isoenzyme and, therefore, the conversion of clopidogrel into its activemetabolite.By contrast,a preliminary report of data from another trial did not find an association between use of a PPI in combination with clopidogrel and an increased risk of adverse outcomes.tr With these conflicting results, the clinical implications of co-administration of a PPI and ciopidogrel remain

w.thelancet.com

Vol 374 September19, 2009

unclear. Whether concomitant use of a PPI attenuates the pharmacodynamic effect or clinical efficacy of prasugrel is also unknown. Therefore, we assessedthe associationbetween the use of a PPI, measuresof platelet function, and clinical outcomes for patients treated with either clopidogrel or prasugrel in the PRINCIPLE (Prasugrel In Comparison to Clopidogrel for Inhibition of Platelet Activation and Aggregation)-TlMl 44 and TRITON (Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition with Prasugrel)-TIMI 38 trials.ll'z

Lancet2OO9,374t 989-97 Published Online September 1. 2009 D0l:10.1016/S01406736(o9)61525-7 SeeCommentpage952 Brighamand Women's Hospital,Boston,MA, U5A (M L 0'DonoghueM0, ProfE Eraunwald MD, Pr ofEMA n t ma n MD , 5 A MurphyMPH,JL MegaMD, M 5 sabatineMD, 5 DWiviottMD); Universityof Michigan, Ann Arbor, Ml, USA ( Pr ofERB a t e s MD ); EdithWolfson MedicalCentdr, Tel-AvivUniversity,Holon, lsrael(ProfYRozenman MD); Children's Hospital,Bo5ton, MA.U5A(A D Michelson MD); Medirch CentrumAlkmaar, Alkmaar,Netherlands (RWHautvast MD);Northeast CardiologyAssociates,Bangor ME,USA(PN VerLeeMD);and Eli Lilly ResearchLaboratorie5, Indianapolis, lN, USA (5 L ClosePhD,L ShenPhD) Correspondence to: DrMichelle O'Donoghue, TlMlStudyGroup,Brigham andwomen'sHospital, 350 LongwoodAvenue, MA 02115,U5A Bo5ton, modonoghue@partne6,org

Methods Patients The PRINCIPLE-TIMI 44 trial was a double-blind, two-phase crossover study that randomly assigned 201 individuals undergoing cardiac catheterisation with planned percutaneous coronary intervention to prasugrel (n:102; 60 mg loading dose, 10 mg a day maintenance dose) or to high-dose clopidogrel 1n:99; 600 mg loading dose,150 mg a day maintenancedose).1 lnhibition of platelet aggregation with 20 pM ADP by light-transmission aggregometry was to be recorded in

989

I

Artictes

all patients at 30 min (r5), 2 h (t10 min), and 6 h (t30 min) after the loading dose. Patientswho underwent percutaneous coronary intervention entered the maintenance-dose phase, a 28-day comparison of prasugrei 10 mg a day versus clopidogrel 150 mg a day. The trial had a crossoverdesign such that patients were transferred to the alternate therapy on day 15 after randomisation. Measures of platelet function were to be done in these individuals at 18-24 h, day 15, and day 29 after randomisation. For the purpose of the current analysis, results of platelet function studies were restricted to the time preceding the crossoverdate. In this trial, thienopyridine hyporesponsivenesswas defined a priori as less than 20% inhibition of platelet aggregation with 20 pM ADP.' Inhibition of platelet aggregationwas defined as:

1-

maximal platelet aggregation at time t after loading dose maximal platelet aggregation at baseline

x100

platelet functlon studies were done. In the TRITON-TIMI 38 trial, individuals were inciuded in the primary analysis if they were taking a PPI at the time of randomisation. Becauseindividuals might have initiated or discontinued a PPI during the course of follow-up, several additional sensitivity analyses were done to assessthe risk ofadverse outcomes for those on a PPI as captured at different timepoints and with varying durations of follow-up. These additional analyses included a comparison between patients who were recorded to be taking a PPI both at randomisation and at end of follow-up and those who were never exposed to a PPI. We also did landmark analysesat 3 days, 3 months, and 6 months after randomisation to analyse use of a PPI at these different timepoints during follow-up and to show consistency of results. We did futher sensitivity analyses to assess the potential effect of specific PPI subtlpes and other gastric acid suppressive drugs, including H2 receptor antagonists.

Statistical analysis Baselinecharacteristicsare presentedas means (25-75%l Analyses were restricted to those individuals who or medians (interquartile ranges, IQR) for continuous received study drug and did not receive a variables and frequencies for categorical variabies. !7e glycoprotein IIb/IIIa inhibitor (n=197). A1l platelet did comparisons for baseline characteristicswith t tests aggregometry tracings were confirmed by a single reader, or Wilcoxon rank sum tests for continuous variables, and who was unaware of randomised treatment assignment X2tests or Fisher's exact test for categorical variables. at a central core laboratory (Center for Platelet Function In the PRINCIPLE-TIMI 44trial, measures of platelet Studies, University of Massachusetts,Worcester, MA, function at each timepoint were reported as means (SD). usA)., All analyses were stratified by randomised treatment The TRITON-TIMI 38 study was a double-blind, arm. Measures of platelet function for patients who phase 3 trial in which 13608 patients with an acute were or were not on a PPI were compared with a I test coronary syndrome undergoing planned percutaneous for inhibition for platelet aggregation or Fisher's exact coronary intewention were randomly assigned to test for iategorical measures of platelet responsiveness. prasugrel (n=6813; 60 mg loading dose, 10 mg a day We used a multivariable linear regression model to maintenance dose) or clopidogrel (n:6795; 300 mg adjust for any baseline characteristicsthat significantly loading dose, 75 mg a day maintenance dose). The differed between patients who were or were not treated primary endpoint of the TRITON-TIMI 38 trial was the with a PPI. composite of cardiovasculardeath, non-fatal myocardial In the TRITON-TIMI 38 trial, we used a multivariable infarction, or non-fatal stroke.tr Relevant exclusions to Cox proportional hazards model to evaluate the participation in the trial included individuals with an independent association between use of a PPI and the increased risk of bleeding, a history of anaemia, risk of clinical adverse outcomes. Using logistic thromboqtopenia, pathological intracranial findings, or regression modelling with a forward selection algorithm the use of a thienopyridine within 5 days before (p for inclusion <0.20), we developeda propensity score randomisation. Ali primary efficacy and key safety for the use of a PPI. The model included baseline and endpoints were adfudicated by an independent clinical clinical characteristics of sex, ethnic origin, region, events committee who were unaware of the assigned history of peptic uicer disease, history of carotid or treatment; however,the use of a PPI was not concealed. vertebral artery disease, previous myocardial infarction, creatinine clearance, use of angiotensin converting Proton-pump inhibitor use enzyme inhibitor or angiotensin receptor blocker at In both trials, the decision to treat with a PPI was at the randomisation, use of statin at randomisation, index discretion ofthe treating physiciah. At each patient visit event of unstable angina/non-ST-elevation myocardial during the course of follow-up, the names of all infarction or ST-elevation myocardial infarction, baseline concomitant medications were recorded on case-report haemoglobin, systolicblood pressure,and heart rate. We forms, including the start and stop dates. In the used an iterative process to define and verify that strata PRINCIPLE-TIMI 44 trial, patients were classifiedas were balanced on the score (ie, no association existed taking a PPI if they were on a PPI on the same day that between the predicted probability from the model and

\M.thelancet.(om

Vol 374 September19, 2OO9

tr a c I]

\ I I

p tl n +l e 14 ?r

o cl u n ir c( tI al ol

u ir al 1t

oi 1". AT

ir

c1 ri: re fo dr di re w, ca IC

h1 ac m P]

w ca n'i

ArticlesI

treatment with PPI versus no PPI within any stratum) and on covariates(ie, no associationexistedbetween any covariate included in the propensity score model and treatment with PPI versus no PPI within any stratum). and With these iterative methods of definition verification, we divided the predicted exposures into 11 propensity strata, which achieved balance on all 13 covariates. For survival analyses, multivariable models are presented using Cox proportional hazard modeis with the PPI effect within each propensity score stratum measured and averaged across strata to estimate the effect of PPI versus no PPI on the outcome. We assessed the assumption of proportional hazards for the primary endpoint models with Schoenfeld residuals; violations were not seen. These models also included risk adjustment for the following covariates:age, sex, ethnic origin, history of hypertension, history of hypercholesterolaemia, history of diabetes, current tobacco use, index event of unstable angina/non-ST-elevation or ST-elevation myocardial myocardial infarction infarction, previous myocardial infarction, previous coronary artery bypass graft surgery history ofstroke or transient ishaemic attack, family history of coronary artery disease,history ofperipheral artery disease,history ofheart failure, history ofpeptic ulcer disease,history of carotid or vertebral artery disease,creatinine clearance, use of drug-eluting stents, multivessel percutaneous used, use of intervention, tfpe of antithrombin angiotensin converfing enzyme inhibitor or angiotensin receptor blocker at randomisation, use of statin, aspirin, or B blockersat randomisation, body-massindex, baseline haemoglobin, systolicblood pressure, and heart rate. We assessedeffect modification by use of a PPI on the efficacy of prasugrel versus clopidogrel by including an interaction term in the model. As an additional exploratory analysis, we examined whether use of a PPI in combination with either clopidogrelor prasugrel was associatedwith an increased risk of adverse outcomes in patients with a single reduced function CYP2C19ailele. Genotyping methods for the TRITON-TIMI 38 trial have been previously described.toFor the current analysis, patients were divided into two groups on the basis of whether they were wild-type carriers or whether they possessedone reduced-function CYP2C19 allele. We then assessed whether PPI use was associatedwith a higher risk of cardiovascular events in patients who had a single reduced-functron CYP2C19 allele, thereby testing the hypothesis that patients with reduced CYP2C19 enzyrne activity caused by a genetic polymorphism would be more susceptibleto additional CYP2C19inhibition by a PPI and thus have a higher risk ofcardiovascular events. We did a second anaiysis to evaluate the risk of cardiovascularevents for patients on a PPI who were wild-type carriers. Individuals with two reduced-function CYP2C19alleleswere excludedfrom the analysisbecause

m.thelancet.com

Vol 374 September19,2009

they would be expected to have minimai CYP2C19 enzyme activity regardlessof the use of a PPI. We did efficacyanalysesin TRITON-TIMI 38 according to the intention-to-treat principle. For all landmark analyses, patients who had an eficacy or bleeding event after randomisation and before the landmark date were excluded io keep the probability of confounding by indication to a minimum. We did safety analyses for patients who received at least one dose of study drug and were followed for 7 days after discontinuation of drug therapy.Ratesofendpoints are expressedas Kaplan-Meier estimates. All tests were two-sided with a significance value of p<0.05. We did the analyses with Stata/SE version.9.2 (Stata Corp, College Station, TX, USA) and SAS version 9.2 (Cary,NC, USA). Roleofthe funding source The PRINCIPLE-TIMI 44 and TRITON-TIMI 38 trials received grant funding from Eli Lilly and Company and Daiichi Sankyo Company Limited. The current analyses receivedno sourcesof external funding. The TIMI Study Group has an independent copy of the trial databases. The authors wrote all drafts of the manuscript and take full responsibility for its content. The sponsors had the opportunity to review and comment on the manuscript, but had no editorial authority. The corresponding author had full accessto the data and had final responsibility to submit for publication.

Results Of the 201 patients enrolled in the PRINCIPLE-TIMI 44 trial, 53 (26' 4%\ were recorded to be taking a PPI at the time of randomisation. Baselineplatelet aggregationwas similar for oatients who wdre or were not on a PPI before I I

Pruugrel

Clopidogrel

Treatedwith a PPI(n=28)

Nottreatedwith P value a PPI(n=71)

Treatedwith a PPI(n=25)

Nottreatedwith P aPPt(n=77) value

Age (years)

0.18 13(18.3%) 9(32'w") (56.5-69.0) 0.64 63.1 64.1(58.o-71.0)

1.O0 22(28.6%\ 7 (28.ov.) 61.8(54.0-69.0)64.7(570-72.0\ o.24

BM I( k 9/m ')

3v o\t/.2-32.5)

29.0(264-31.0)0.17

0.54 29.3(24.0-33.1)28.5(25.3-30.7)

Diabetes

11(39.3%)

18(25.4v.\

0.22

9 (360%)

24\31-.2%)

0.81

PreviousMl

11(39.3%)

17(23.9v"\

o74

13(52o"/.)

78(23.4k)

0.01

Hypertension

22(786v")

Female

s5 (n.svl

1 oo

23(92.ov")

64 (83.1"/")

0.35

PreviousCABG

6 (21.4v.)

76(22.5v4

1.00

13(16.9v")

1.00

Dyslipidaemia

25(89.3%)

61(85.9"/t

0.75

4 (16.0o/) 24(96oY"\

68 (88.3%)

o.44

Currentsmoker

9 (32'1v") 10(35.7%)

7 (99v") 28 (39.4v")

o.o1

6 (24.ov")

72(75.6v,)

o.37

0 82

9 (36.0%)

29e 77" )

1.O0

25(89.3"/")

61(85.9"/")

0.75

22(88.0ol")

68 (88.1v")

1.00

55(n.5v4

0.42

20(80.07")

62(805"/")

1.00

65(91.6010)

0.46

21(84.o%)

70(90.9%)

0.46

AnginaCCSlll or IV Previousaspirin

Curent p blocker 24 (85.7v1. Currentstatin

24(85.7v")

inhibitor.BMI=body-mass index.Ml=myocardial or mean(25-75"/o). PP|=proton-pump Dataarenumber(o/o) (ardiovascular arterybypassgraft surgery.CC5=Canadian fuciety. infarction.CABG=coronary Toble7: Baselinecharacteristicsof patients in the PRINCIPLE-TIMl44 trial

I

Artictes

administration of study drug in either randomised treatment arm. Thble 1 shows baseline characteristicsof patients who were or were not treated with a PPI and stratified by treatment arm.

_--.__--'_. Prasugrel -+- -PPl g +ppl

o

=50 1

o4o

:E ro

10 0

0l t\\l o .5 h Prasugrel MeanIPA (SD)

PPIPPI+

Absolute A(%) A(%) Relative p value Adjustedp value Clopidogrel PPIMeanIPA

(sD)

PPI+

Absolute A(%) RelativeA(%) p value Adjusted p value

4

| 2h

66.0 3 59 (28.8) 1r8 9) 16.0 60.7 (25.0) (241) 199 53 80 554 0.009 0 38 0.013 o.5o 4 .9 (12.0) 4.9 (16.1) 0 0 0 98 o-79

8

I

t2 16 Hoursafter loadingdose

20

6h

24

1&r4 h

767 (72.4\ 69.6 (135) 7.7 93 0.054 0.051

77.1 ( 12.8) 62.9 ( i 6.5) 8'2 115 0.20 0.11

242 35.2 (20 s) (20.9) 70.4 23.2 (16.)) (19.5) 13.8 12.0 57.0 347 0.003 0.02 0 005 0.08

361 (20.8) 238 (74.4) 72.3 341 003 0.16

Figure1: Mean inhibition of platelet aggregation Inhibitionof plateletaggregation(lPA)to 20 pM ADPat prespecifred timepointsaftera 600 nrg loadingdoseof clopidogrelor 60 mg loadingdoseof prasugrelstratifiedby the useof a PPIin the PRINCIPLE-TlMl 44 trial. PPI=proton-pump inhibitor..p<0.05for comparisonbetweenusersand non-usersof a PPl.

For patients randomly assigned to a 600 mg loading dose of clopidogrel (n:99), the mean inhibition of platelet aggregation was significantly lower in those treated with a PPI at 2 h, 6 h, and 18-24 h after study drug administration than in those not on a PPI (figure 1). Inhibition of platelet aggregation did not substantially differ between the two groups at 30 min (p:0.98). After 15 days of maintenance therapy with clopidogrel 150 mg a day,patients on a PPI continued to have a trend towards lower mean inhibition of platelet aggregation than that of those not treated with a PPI (mean [SD] 29.4% 123.7%l vs48 '8%o,17-6%1,p=0 . 06). For patients randomly assigned to prasugrel (n:102), the mean inhibition of platelet aggregation for those on a PPI was significantly lower 30 min after a 60 mg loading dose than for those not a PPI. However, this difference was not significant at other timepoints during the first day of treatment (figure 1). After 15 days of treatrnent with prasugrel, the mean inhibition of platelet aggregation was significantly lower for patients on a PPI than for those not on a PPI (48.2% [18.4%] vs 66 ' 5/" 115. 5%| p:0 01). 24 h after a 600 mg loading dose of clopidogrel, the proportion of patients with thienopy'ridine hyporesponsiveness was more than two-fold higher for patients on a PPI than for those not on a PPI (50.0%"vs I8.2%, p:0.009, figure 2A). Moreover, after 15 day'sof follow-up, the proportion of patients with thienopylidine hyporesponsivenessto clopidogrel 150 mg a day was more than six-fold higher for patients on a PPI than for thosenot a PPI (50.0% vs7.9%",p=0'0f2, figure 28). By contrast, only a few patients had an inadequate response to prasugrel a{ler 24 h or L5 days of follow-up, irrespective ofthe use ofa PPI. Of the 13 608 patients en-rolledin the TRITON-TIMI 38 trial, 4529 individuals (33 .3%) were recorded to be taking a PPI at the time of randomisation. Of these patients,

I

I

I

lc t: € E c

to

t;

E

z

(n=22)

(n=55)

(lopidogrel600 mg loadingdo:e

(n=19)

(n=53)

Prasugrel 6Omg loadingdose

PPI (n=8)

No PPI (n=38)

Clopidogrelt50 mg a day

Figure2; Proportion of non-responders(definedas IPA<2o7o to 20 FM ADP)6 h (A) and 15 days (B) after clopidogrelor prasugreland stratified by the useofa PPIin PRINCIPLE-TIMl 44 PPI=proton'pumpinhibitor.IPA=inhibition of plateletaggregation.

ww.thelancet.com Vol 374 September19, 2009

Articles

Prasugrel

Clopidogrel Not treated with a PPI P value (n=4538)

Treatedwith a PPI (n=2257) Age (years) Male White

62 (53-70)

60 (s2-69)

<0-001

1587(703"/") 2773(94.1vo)

3390(747'/.) 4161(92.ov")

<0 001

27.5(24-9-30.9)

BNal(kg/m')

ia Hypercholesterolaem Diabetes PreviousMl Previous CABG Carotidor vertebralartery disease heartfailure Previouscongestive

0.004

0.oo7 <0.001

27.8(25.2a7.t

0.83

2902(63.9%)

0.57

1274(56.5v")

25r.6(55.4v")

0.43

7299(67.2v"1

249L(54-9v.)

oo7

547(24.2v")

(22.5v") 1.023

0.72

7o42(23o"k)

0.61

388\17.2"/") 1.75(7 8v1

820(18.r"%)

o 37

534(23.5v") 400(77.6vo\

826(78.2v"\

0,55

322(7.7q.1

0 33

177(7 8v1

364(8.0%)

o75

72O(2.6v")

o.L8

87(3.9v")


154(3.4v")

0.13

73(32v") 97 (4.3v1 21O(9.2v")

190(4.2v")

<0001

272(12.2v"\

445(r0ov") 140(129-150)

<0.001

r n Q5 v")

93 (4.1%)

Previouspeptic ulcerdisease

D9 Oo 2 v")

1 8 6 ( 4 .7%)

<60 mL/min* clearance Creatinine

292(13.1v")

Baselinehaemoglobin (g/L)

117(726-747)

481(108%) 140(129-150)


3317(73.1v")

1713(75.9"/")

o.oo5 0.01

138(127-148) 1705V5 Ov")

169(37%)

0 25

3339(73.5%)

Region

33Lh 2.2%

South America

0.006 0 18
<0.001

North America

0.88

1737(38.3"k)

0.052

0.11

<0.001

IndexdiagnosisUA or NSTElvll

27.8(25.L-3L.0)

o-01

6o (52-59) 3450(760%) 4124(9r.3"/")

2 8 8 3( 6 3 .5%)

7755(J'87v"):

1488(65.9"/")

Hypertension

2 78 ( 2 52-31-.2\

61(s3-70) 1658(73.0%) 2139e4)v.)

Nottreatedwith a PPI p value (n=4541)

875(385"/6) 1468(64.6"/")

828 (36.7v.)

use Currenttobacco

0.001

Treatedwitha PPI (n=2272)

3lOok

30 8%

333%

47v"

2 2'k

4.9v" 22.6v"

WesternEurope

325"/.

22.9v"

33.2"/"

EasternEurope

20.2v"

266%

20.00/o

26.5%

Restoftheworld

12.O./"

14'9/"

lt.40k

15.\qo

56.9v"

54.Ova

59.2vo

56.4"/"

o.15

at randomisation Medications ACE-lor ARB

75.8v"

p Blocker

0.03

0.03

73Ovo

001

751k

13 5v.

82.9v.

76.8v"

<0.001

Statin

8 0 .5 %

777v"

0.008

A sp i ri n

96.7v"

96 5v"

070

97.7Vo

96.8"/"

0.40

89.3%

87.9v"

0.10

882v.

87.gva

o.73 0.004 o.37

duringindexhospitalisation Medications p Blocker Statin

92.9"/"

923%

o.43

93.1vo

97.\vo

AsPirin

99.3v"

99 4v"

0-9o

99.4/.

99 2v"

graftsuIgery' coronary arterybyPass infarctiorlCABG= ppl=proton-pump index.Ml=myocardial inhibitor.BMI=body-mass Dataarenumber(%)or median(tQR). infa myocardial NSTEMI=non-sT-elevation angina. uA=unstable formula' wasestimatedwith the Cock(roft-Gault range.'Creatinineclearance f

Toble2: Baselinecharacteristicsof patients in theTRlTON-TlMl 38 trial

1844 were recorded to be on pantoPrazole, 1675 on omeprazole, 613 on esomeprazole, 441 on lansoprazole, and 66 on rabeprazole. Table 2 shows baseline characteristicsfor patients who were or were not on a PPI and stratified by treatment arm. Patients who were keated *"ith a PPI were significantly more likely to be older, female, white, have an index diagnosis of unstable angina/non-ST-elevationmyocardial infarction, and live in North America or western Europe than those not treated with a PPI. Patients treated with a PPI were Also significantly more likeiy to have a history of peptic ulcer disease and have a lower baseline haemoglobin concentration at the time of randomisation (table 2). Baseline characteristicsbetween randomised treatment arms did not differ

ffi.thelancct.com

Vol 374 SePtember19, 2009

For patients randomly assigned to clopidogrel rateof the primaryendpoint (n:6795),the Kaplan-Meier throughout long-term follow-up was 11'8% for individualson a PPI and72'2%"for thosenot on a PPI HR 0'98, 957"CI 0'84-1'14,p:0'80). For (unadjusted patientsrandomly assignedto prasugrel(n:6813),the rate of the primary endpointwas10'2yafor individuals on a pPI and 9.7%"for those not on a PPI (1'05' O.8g-7,23,p=0'58; figure 3). After adiustmentfor potentialconfoundersand the propensityto treatwith a remainedbetweenuseof PPI,no significant.association both for patients primary endpoint, risk of the a PPI and treatedwith clopidogrel(0'94,0'80-1'11,p:0'46) and for those treated with prasugrel(1 00, 0'84-1'20, with useof a PPI wasnot associated p:0'97). Similariy, 993

Artictes

I

increasedrisk ofmyocardial infarction, stent thrombosis, or a decreasedrisk ofbieeding for patients treated with either clopidogrel or prasugrel (table 3). BecausePPIs could be started or discontinued at any time during foilow-up, we did sensitivity analyses to look for consistencyof our data based on PPI use at different timepoints and a{ler varying durations of follow-up. PPI use at the time of randomisation was not associatedwith an increased risk of cardiovascular death, myocardial infarction, or stroke during 3 days of follow-up for patients treated with clopidogrel (1.00, 0.80-1..271 or pr as ugr el ( 1. 14, 0. 88- 1. 46) . S i m i l a r l y , P P I u s e a t

10

; o

fs

o Numberat risk PPI Clopidogrel-no Clopidogrel-PPl Prasugrel PPI Prasugrel-no PPI

100

200

100

400

3158 1533 1559 1236

2690 1269 1296 2715

Days / ( 1q

2257 2272

4014 1994 2050 4101

3619 1757 1827 3666

FigureJ:Unadjusted cumulative incidenceof CVdeath, Ml, or strokethrough long-term follow-up Patientswereor were not on a PPIin combinationwith clopidogrelor prasugrelin theTRlTON-TlMl38 trial CV=cardiovascular. Ml=myocardialinfarction PPI=proton-pumF inhibiior.

randomisation was not associatedwith an increasedrisk of cardiovasculardeath, myocardial infarction, or stroke during 30 days of foliow-up for patients treated with clopidogrel (0.98, 0.80-1.21) or prasugrel (i.09, 0. 87-L. 37).PPI use was not associatedwith risk of eariy stent thrombosis (Academic Research Consortium definite or probabie <30 days a{ter randomisation) for clopidogrel-treated(1. 17,0. 76-1. 81) or prasugrei-treated patients (0.76, 0.36-1.61). Landmark analyses showed consistency of the results for PPI use at different timepoints during foliow-up, including day 3, 3 months, and 6 months after randomisation and after short-term or long-term follow-up. As an additional sensitivity analysis, we compared individuals who were recorded to be taking a PPI for the entire duration of follow-up (n=2814) with those who were never exposed to a PPI (n:6912). Individuals on a PPI for a period shorter than the entire duration of follow-up were excluded from the analysis. Consistent PPI use was not associatedwith an increased risk of cardiovascular death, myocardial infarction, or stroke in patients treated with either clopidogrel (1. 05, 0 .85-1 . 30) or prasugrel (1. 10,0 . 88-1. 39) compared with those who were never exposed to a PPI. In another sensitivity analysis in which we restricted the study population to those who had not discontinued stuJy drug/ no significant associationwas present between PPI use and risk of the primary endpoint for clopidogrel-treated (0.94, 0.80-1.11) or prasugrel-treatedpatients (0.99, 0.83-1.19). We did additional analyses to assess the effect of individual types ofPPIs and other gastric antacid drugs, including H2 receptor antagonists. Regardlessofthe type of PPI (including omeprazole alone or the exclusion of pantoprazole), no independent association existed between use of these drugs and the risk of myocardial infarction or the composite of cardiovascular death, myocardial infarction, or stroke for patients randomly assigned to clopidogrel or prasugrel in the

Clopidogrel

Prasugrel

Treatedwith a PPI

Not treated with a PPI

Adjusted HR (95x Cl)

Treatedwith a PPI

1r.8y"(25512257)

12.2%(52614538)

0.94(0.80-1.11)

r0.2vo(22o12272)

All-causedeath

2.9v"(5812257)

(139/4s38) 3.3"/"

0 68 (0.47-0.95)

Cvdeath

2.2vo(4412257)

Cvdeath, Ml,orstroke

Nottreatedwith a PPI

1.00(0.84-1.20)

J lYo\o>ltt/2)

0.71(o.47-7.07)

2.2v.(4612272)

2.ov"(8714541)

1.06(0.70-1.62)

0.98(0.82-1.17)

7 7v.(166/2272)

7.3,/,{.37914541) (4614263) 1..7%

1.02(0.84-1.25)

1.7E" (2212159) 4 8"k(9812253)

5.0v"(20514488\

0.92(o 71-1.18)

2.4v"(9514488\

0.97(0.67-1.39)

Ml

9.5v.(20912257)

2.5y"11061453s) 9 gTo(424/4538)

Stent thrombosis (ARCdefinite or probable)

2.4y" (5O1215O)

2'3/"(9214272)

TIMI majoror minor bleeding(non-CABG)

4.6v. (9212234)

3.4v,(r3914482)

1.08(0.75-1.55) 1.13(0.85-1.49)

7.6v"(6514482t

1.20(0.80-1.79)

2.5v"(57t2253)

118%(594/4538)

0.96(0.83-1.12)

r2.6y"(26812272)

TfMl major bleeding (non-CABG) Net clinicafoutcome(death,Ml, stroke, orTlMl majornon-CABGbleeding)

2.4y" @612234) 13 9y" (29912257)

AdjustedHR(95%Cl)

9.7v"(4231454r) 3.ov"(723t4547)

t2.1v"(5].614547)

1.00(0.7r.-1.41)

1.03(0.60-1.76)

0.99(0.85-1.17)

arterybypassgraft surgery. cABG=coronary Tablel; Kaplan-Meierevent ratesfor efficacyand safety endpointsin theTRlTON-TlMl38 trial

ww.th€lancet.com Vol 374 September19, 2009

Artictes I

TRITON-TIMI 38 trial (tabie a). Use of an H2 receptor antagonist or PPI at baselinewas not associatedwith risk of cardiovascular death, myocardial infarction, or stroke for patients randomly assigned to clopidogrel (0.80, 0. 51-1.26 )or p rasug rel( 0. 91,0. 55- 1.5i) . Overall, no significant interaction existed between PPi use, use of clopidogrel or prasugrel, and risk of the primary endpoint of cardiovascular death, myocardial infarction, or stroke. Prasugrel was consistently more effective than clopidogrel in reducing the primary endpoint in patients who were taking a PPI (0.85, 0.71-1.01) or were not on a PPI at baseline (0.80, 0.70-0'90, p interaction:O.57,figure 3). DNA samples were available for 1477patients randomly assigned to clopidogrel and for 1466randomly assigned to prasugrel in the TRITON-TIMI 38 study. A single reduced-function CYP2C19allelewas seenin 357patients randomly assigned to clopidogrel and 372 randomly assigned to prasugrel. Of these, 34%"in the clopidogrel arm were treated with a PPI at randomisation versus 33%"in the prasugrel arm. For patients randomly assignedto clopidogrel and who had a single reduced-function CYP2C1.9 allele, the primary endpoint of cardiovascular death, myocardial infarction, or stroke occurred in I0.2%o (1211201of patients on a PPI and13.0%"(301237)noton a PPI, which was not significant after adjusting for the propensity to be treated with a PPI (0.76, 0'39-1.48). Similarly,for patients randomly assignedto prasugrel, the use of a PPI was not independently associatedwith an increasedrisk of the primary endpoint in patients who were (7.4%", 91122)or were not (9.9%, 241250)treated with a PPI

Clopidogrel

Prasugrel

CVdeath, Ml, or stroke

CVdeath,Ml, or. stroke

Ml

1.02(0.76-1.36)

(n=16/5) Omeprazole

0.91 (0.22-1.15)

0.95 (0,7j-1,2j)

1.04 (0.81-1.34)

(n=1844) Pantoprazole

0.94 (o 74-1 18)

o.97 10.75-124)

1 09 (0 86-1.39)

1 09 (0.83-1.43)

(n=613) Esomeprazole

1..o7(0.75-752\

1.18 (0.81-1.73)

0.86 (0.55 1.33)

o.92 (0.57-1.48)

(n=441) Lansoprazole

1.00 (0 63-1 59)

0.86 (0.51-1.46)

0 98 (0.61-1.5/)

1.08(0.66 1.79)

Cv=cardiovascular. Ml=myocardialinfarction.Rabeprazole is not includedinthetable becauseonly 66 patientswere takingit at randomisation. Icble{: Adjusted hazard ratios (gg% Cl) forthe association between differenttyp6 ofproton-pump inhibitors and the risk of Ml orthe compositeofCV death, Ml, or strokethrough long-term follow-up

observed a trend towards a modest attenuation of the in-vitro anfiplatelet effects of both clopidogrel and prasugrel in the presenceofa PPI during the loading and maintenance phasesof the drugs. PPIs are often administered to patients in combination with thienopyridines to help reduce the risk of bleeding after acute coronary syndrome or after percutaneous coronary intervention. However, several studies have shown that PPIs, especiallyomeprazoie, could diminish the antiplatelet effects of clopidogrel.'? Analyses from three large population-basedstudies have found that use of a PPI is associatedwith an increased risk of adverse eventsin patients treated with ciopidogrel.*'oIn response to these findings, both the European Medicines Agency (EMEA)" and the US Food and Drug Administration (FDA)tureleasedstatements or communications warning ofa potential interaction between PPIs and clopidogrel, and discouraging their combined use in the absenceof a (0.81,0.3s-1.8s). strong indication. Furthermore, for patients randomly assigned to Yet,at this time, scarcedata show a definitive interaction clopidogrel and who did not have a reduced-function between PPI use and the clinical benefit of clopidogrel. CYP2C19 allele (wild-type carriers), the primary endpoint Without a randomised trial, concern existsthat the higher occurred in7.2% (231333\of patients on a PPI and8.4%" cardiovascularevent rates in patients prescribed a PPI in 160l731lnot on a PPI; a differencethat was not significant population-based studies might partly be explained by between groups (0'90, 0 . 55-1 .48). For wild-type carriers differences in baseline comorbidities and indicate a randomly assigned to prasugrel, the primary endpoint higher-risk patient population. By contrast, less occurred in9.|Yo (27 1323)of patients on a PPI and,70'27" heterogeneity might exist in a trial population, which not on a PPI (0.89, 0.57-1.39). These results could help to reduce the probability of confounding by (7217251 were consistentwhen the analysiswas repeatedwith oniy indication. Such a possibility is supported by the fact that those PPIs known to be stronger inhibitors of CYP2C19, we saw more comparable event rates for patients who including esomeprazole,lansoprazole,omeprazole, and were or were not on a PPI in the TRITON-TIMI 38 trial, rabeprazole. even before multivariable adjustment. Why might the use of a PPI interfere with the Discussion antiplateleteffectsor clinical benefits ofthienopyridinesl In a iarge population ofpatients with an acute coronary One hypothesis is that drug absorption is increased in syndrome undergoing percutaneous coronary interan acidic environment, and therefore the use of a PPI or vention, the use ofa PPI was notindependently associated other antacids could diminish or slow drug absorption. with increased risk of adverse clinical outcomes 'for Another hypothesis is that many PPIs inhibit CYP2C19,' patients treated with either clopidogrel or the novei and therefore the use of a concomitant PPI could impede thienopyridine prasugrel.Our findings contrastwith data or prevent the metabolism ofclopidogrel or prasugrel to from some observational studies that reported an their active metaboiites through competition for the increasedrisk ofadverse eventsin patients treated with a same substrate.Genetic polymorphisms of the CYP2C19 PPI in combination with clopidogrei.s 10However, we allele are associatedwith an increased risk of adverse

w.thelancet.com

Vol 374 September19, 2009

I

Artictes

outcomes for patients treated with clopidogrel, thereby supporting the concept that diminished CYP2C19activity might increase the risk of cardiovascular events in Importantly, however,a clopidogrel-treatedpatients.'n'rr2' reduced-functton CYP2C19 allele was not associated with an increased risk of adverse cardiovascular outcomes in patients treated with prasugrel in the TRI TO N- TI M l 38 t r ial. " Our study provides important new information. First, we showed some attenuation of the in-vitro antiplatelet effects of both high-dose clopidogrel and prasugrel in patients treated with a PPI compared with those in patients not treated with a PPL These findings were seen during both the loading and maintenance dosing of the drugs. Becauseprasugrel showed more potent and more consistent inhibition of platelet aggregation than high-dose clopidogrel, aimost no patients met the definition for hyporesponsivenesson the combination of prasugrel and a PPI. By contrast,despitethe use ofhigher loading doses (600 mg) and maintenance doses (150mg) of clopidogrel, the proportion of patients who were hyporesponsive was substantially higher for those randomly assignedto clopidogrelthan for those assigned to prasugrel, and was higher for patients on the combination of clopidogrel and a PPI than for those not on a PPI . Despite the observed attenuation of the in-viko antiplatelet effects of clopidogrel and prasugrel in patients treated with a PPI, we did not show use of a PPI to be associatedwith an increasedrisk ofadverse clinical outcomes after adjusting for potential confounders and the propensify to be treated with a PPI. These findings were consistent regardlessofindividual types of PPIs or the use of an H2 receptor antagonist. A modest attenuation of the antiplatelet effects of either clopidogrel or prasugrel might have been insufficient to translate into an increased risk of adverse outcomes. Although ex"vivo platelet function studies are associated with risk of subsequent ciinical outcomes, these findings emphasise that surrogate endpoints should not be used as a substititute for clinical events. The absence of a clinical drug-drug interaction is similar to a previousiy reported interaction between atorr'astatin and clopidogrel. Atorvastatin attenuatesplatelet inhibition by clopidogrel in a dose-dependent manner;" however, these in-vitro findings are not associated with worse clinical outcomes.t-26 Our anaiysis also assessedthe eficacy of prasugrel versus clopidogrel in the setting of a PPI. We found that prasugrel had a consistent benefit for reducing the risk of cardiovascular death, myocardial infarction, or stroke compared with clopidogrel, irrespective of the use of d PPI. These findings are important because it has been previously questioned whether use of a PPI contributed to the superiority of prasugrel over clopidogrel by diminishing conversion of clopidogrel to its active metabolite in the TRITON-TIMI 38 trial."

996

We assessedwhether PPi use was associatedwith an increased risk of adverse outcomes in patients with a reduced-function CYP2C19allele. If such an interaction existed, it would support a two-hit hypothesis that patients with diminished baseline CYP2C19 activity would be more mlnerable to further suppression of CYP2C19activiry by PPIs. Although the genetic analysis was underpowered to detect a difference, we did not show a higher risk ofadverse outcomes for patients on a PPI with diminished CYP2C19 activity caused by a reduced-function allele. Important limitations of our analysis warant consideration. In the PRINCIPLE-TIMI 44 and TRITON-TIMI 38 trials, the use of a PPI was not randomised and was left to the physician s discretion. Although we did extensive multivariable adjustment for potential confounders and included a propensity score for the decision to treat with a PPI, the possibility of residual confounding remains, which is a potential hazard of all obsewational comparisons. Our analysis was designed post-hoc and the TRITON-TIMI 38 trial was not specifically designed to assessPPI use. Misclassificationof patients according to the use of a PPI could bias the results towards the nuil. Also, PPIs could be initiated or discontinued during the course of follow-up, and compliance cannot'be ensured for individual patients. We addressedthese isfues by doing sensitivity analyses that included landmark analyses to detect PPI use at different timepoints, and modelled varying durations of foilow-up. One advantageof doing this type of analysis within a randomised trial compared with many population-based studies is that a1l endpoints were strictly defined and adjudicated by a blinded clinical events committee. In our analysis,individuai subgroups might have been underpowered to show an associationbetween PPI use and risk of pharmacodynamic or clinical outcomes, if such a relation existed. Other pharmacodynamic studies have previously investigated the effects of different types of PPIs on platelet function.'7" Furthermore, DNA samples were oniy available in a subset of the study population, and this population was further subdivided on the basis ofthe use ofa PPI. Consequently,the genetic analysis was underpowered and shouid be viewed solely as exploratory. Additionally, DNA sampies were unavailable in the PRINCIPLE-TIMI 44 study; therefore, we could not evaluate the effect ofgenetic polyrnorphisms on platelet function tests for either clopidogrel or prasugrel with or without a PPI use. Overali, PPI use was not associatedwith an increased risk of cardiovascular events in patients treated with either clopidogrel or prasugrel. Although only a randomised trial of PPI use can definitively establishthe clinical implications of combining a PPI vrith a thienopyridine, our findings do not support the need to avoid concomitant use of PPIs for gastric protection in patients receiving thienoplridine therapy who are at increasedrisk for gastrointestinalbieeding.

ww.thelancet.com Vol 374 September19, 2009

Artictes I

Contributors MLO'D particlpatedin study design, data analysisand interpretation, witing ofthe report, literature search,and preparation offigures. EB participatedin study design, and data analysisand interiretatron. EMA participatedin study design, data collecion. data analysis and interpretation, and revision ofthe report. SAM participated in study design, data analysisand interpretation, and critical review ERB. yR. and RWH participated in data collection, data interpretation, and revieu. ofthe report. ADM participated in data collection, data inrerpretation, and writing ofthe report. PNVL partjcipated in data collecti;n, data analysis,and critical review. SLC and LS participated in data analysis and interpretation. fLM and MSS participated in data interpretation, and review and revision ofthe report. SDW par-ticipated in design implementation and interpretation, study design, data interpretaton, review and comrnents on figures and rcporl. Confli(ts of interest EB received research grants and honoraria from Eli Lilly and Daiichi Sankyo.EMA receivedresearchgranrs from EIi Lilly, D;iichi Sankyo, and Sanofi Aventis; and consulting fees or paid advisory board feesfrom Sanofi Aventis and lecture fees from Eli Lilly and Sanofi Aventis. SAM received research grants frorn Eli Lilly and Daiichi Sankyo, and consulting fees fiom Eli Lilly. ERB received honoraria frorn Eli Lilly, Daiichi Sankyo, Sanofi Aventis, and Bristol-Myers Squibb. yR received consulting or lecture fees from Eli LiJly, Sanofi Aventis, Medtronrc, Boston Scientific, Pfizer, and Schering-plough. ADM is a consultant to EIi Lilly, Daiichi Sankyo,Sanofi Aventis, and Bristol.Myers Squibb. SLC and LS are employeesand stockholdersof Eli Lillv. I LM received researchgrants from Eli Lilly, Daiichi Sanlgo. rnj john.o,. & johnsorr, and honoraria from Bayer Healthcare.MSS receivedresearch srants from Sanofi Aventis and Astra Zeneca.and honoraria andTor consulting fees from Sanofi Aventis, Bristol-Myers Squibb, Astra Zeneca, and Eli Lilly. SDW received research grants frcm Eli Li)ly, Daiichi Sankyo, Sanofi Aventis, and Schering Plough; consulting fees or paid advisory fees from Astra Zeneca, and Sanofi Aventis; and honoraria from Eli Lilly, Daiiclri Sankyo,Astra Zeneca,and Schering Plough. MO D, RWH, pVL declare that they have no conflicts ofinterest. References 1 fernberg I PayneCD, Winters KJ, et al. prasugrel achievesgreater inhibition ofplatelet aggregation and a lower nte ofnon,re$onders compared with clopidogrel in aspirin-treated patients wit} stable coronary artery disease. Eur Heert J 2006:27: 1,f66_73. 2 Brandt JI Payne CD, Wjviott SD, et al. A comparison of prasugrel and clopidogrel loadrng doses on plitelet function: magnitude of platelet inhibition is related to active metabolite tormation. Am Heart J 2O07;753:66.e9-i6. 3 Wiviott SD, Trenk D, Frelinger AL, et al. prasugrel compared with high loading- and maintenance-dose clopidogrel in patients with planned percutaneous coronary intenention: the prasugrel in Comparison to Clopidogrel for Inhibition of plateldt Activition and Aggregation-Thrombolysis in Myocardial Infarction 44 trial. C ir culation 2007: 176: 2923-32. 4 Bhatt DL, Scheiman f , Abraham NS, er al. ACCF/ACG/AHA 2008 expert consensus document on reducing the gastrointestinal risks ofantiplatelet therapy and NSAID use: i report ofthe American College ofCardiology Foundation Task Force on Clinical Expert Consensus Documents. Citcvlqtiofl 200g., 118: 1894-909. 5

Gilard M, Arnaud B, l.e Gal G, Abgrall JF, Boschat [. Influence of omeprazol on the antiplatelet action of clopidogrel associatedto aspirin. / Thromb Haemost 2006:4:2508-09.

6

Gilard M, Arnaud B, Cornily JC, et al. Influence of omeprazole o n th e a n tj p l a te l e ta cti o n ofclopidogr el associatedu.ith aspir in: the randomized, double:blind OCLA (Omeprazole CLopidogrel Aspirin) study./ A m Coll Cardiot 2008; 5l: i56-60. Sibbing D, Morath T, Stegherr I, et al. lmpact ofproton pump i n h i b i to rs o n th e a n ti p l a teletcffects of clopidogr el. Thr omb H a emost 2009; 70l: 7 74-79. |uurlink DN, Gomes T, Ko DT, et al. A population-based study of the drug interaction between proton pump inhibitors and clopidogrel. CMAJ 2009; 180: 713-i8.

Z

8

www.thelancet.comVo[374 Septemberlg, 2OO9

9

Ho PM, Maddox TM, lfang L, et aJ.Risk of adverse outcomes associatedwith concomitant use ofclopidogrel and proton pump inhibitors following acute coronary syndrome_ /AMA 2009; 301:937-44.

l0

Stanek EJ,Aubert RE, Flockhart DA, et al. A national study ofthe effect ofindividual proton pump inhibitors on cardiovascular outcomes in patients heated with dopidogrel following cororlary stentlng:the ClopidogrelMedco OutcontesStudy.papir presentedat SCAI 32nd Annual 5cjenri6c Sessiotrs.Las Vegas.NV, ZObC. Dunn SP, MacauJayTE, Brennan DM, et al. Baseline protorr purnp inhibitor use is assocatedwith increased cardiovasculu events with and without the use ofdopidogrel in the CREDO trial. Circulat[ott 2008; 118: 815A (abstract).

U

12

Wiviott SD, Bramwald E, McCabe CH, et al. prasusrel versus clopidogrel in patients witJr acute coronary slmdrories. N Engl Med J 2007: 357: 200r,75.

13

Wiviott SD, Antrnan EM, Glbson CM, et al. Evaluation of prasugrel compared with clopidogrel in patients with arute coronary spdromes: design and rationale for the TRial to assessImprovement in Therapeutic Outcomes by optimizing platelet InhibitioN with prasugrel Thrombolysis In Myocardial Infarction lg (TRITON_TIMI 38).Am Heart J 2006;152: 627_35. lr4ega|L, Close SL, Wiviott SD, et al. Cyrodrorne p450 pollmorphisms and response to doprLdogrel.N Enfi Med 2}0gi 360. 35:+42. J European Medicires Agency. public statement on Dossible interachon between clopidogrel and proton-pump jnhibitors. http,77*."*ea. europa.u/rumandocs/PD Fs/ EpAR/ plavix/32895fi09en.pdf (accessed fuly 23, 2009).

74 15

16

Food and Drug Adminishation. Early communication about an ongoing safety review ofdopidogrel bisulfate (marketed as piavix). http://ww fda-govI cderI drugl eaily_comm/cLpidogrelbisulfate. htm (accessedJuly 23,2009). 17 Li XQ, Andersson TB, Ahlstrorn M, Weidolf L. Comparison of inhibitory eflects of the proton pump-itrhibiting drugs omeprazole, esomeprazole, lansoprazole, pantoprazole, and rabeprazole on human cytochrome P450 a cnines. Drug Metab DisposZ}}li3/:82t_Zz 18 Sibbing D, Stegherr J, Latz W, et al. Ctrochrome P45OZC79 loss-of function polymorphism and stent tluombosis followine percutaneous coronary intenention. Eur Heart 2009;30:916-22. J -19 Collet fP, Hulot fS, Pena A, et al. Clrochrome p450 2C19 polymorphism in yormg patients treated u.ith dopidogre) after myocardial infarction: a cohort study. Lancet 20O9;37i: 309-17. 20 Giusti B, Gori AM, Marcucci R, et al. Relation of cy,tochrome p450 2C19 loss-offunction pol;morphism to occunence of drug_eluung coronaq, stent thrombosis. Atn J Cardiol 2009; 101: g06_1i 21 Simon T Verstuyf C, Mary-Krause M, et al. Genetic determinants of respmse to clopidogrel and cardiovascular events. N EngL Med2O09; I 360:363-75. 22

23

24

25

26

27 28

MegaJL, Close SL, Wiviott SD, et al. Clrochrome p450 genetic pol;morphisms and the response to prasugrel: relationihip to phamacokinetic, phamacodynamit, and ilinical outcomes. Circulation 2009; 119; 2553-60. hu WC, Waskell tA, Viratkins pB, et al. Atoruastatin reduces the ability ofclopidogrel to inhibit platelet aggregation: a new drug_drug interaction. Circulation 2003: 107: i2-37. Saw J, Steinhubl SR, Berger pB, et al. tack ofadverse clopidogrelatoryastatin clinical interaction from secondarv analvsis ofa ' randomized, placebo-controlled clopidogr.eltrial. Circulation 2OO3; 108:921..24. Men^bergen H, Gin AK, Schiele R, et al. Comparison of clinical benefits ofdopidogrel therapy in patients with acute coronary syndromes taking atonastatin versus other statin therapies. Am J Cardiol,2003; 92: 285-88. Gorchakova O, von Beckerath N, Gawaz M, et al. AntiDlatelet effects ofa 600 mg loading dose ofclopidogrel are not attenuated in patients receiving atonastatin or simvastatin for at least 4 rveeks priorio coronary arteD srennng. Eur Ileafl J 2004; 25: l89g_902. Eli Ljlly and Co. Efllcient labeling infomarion. hnp://pi.liJly. c om /us /effi enr .pdf{ac c es s ed) uJ y 23.20091. . SillerMania /M, Spiel AO, l:ng IM, Kreiner G, Christ G, lilma B. Effects ofpantoprazole and esomeprazole on platelet inhibition by dopidogrel. Am Heort J 2009:t5z:1€.e141-4i.

997