The Risk of CV Events for Patients Treated with Clopidogrel or Prasugrel in Combination with a Proton Pump Inhibitor Results from the TRITON-TIMI 38 Trial M. O’Donoghue, S.D. Wiviott, E.M. Antman, S.A. Murphy, C.F. Contant, E.R. Bates, Y. Rozenman, J.L. Mega, M.S. Sabatine and E. Braunwald Cardiovascular Division, Brigham and Women’s Hospital

Disclosures l M. O’Donoghue: none l S.D. Wiviott: Grants from Eli Lilly, Daiichi Sankyo, sanofi aventis, Schering Plough. Consulting or advisory board fees: Astra Zeneca, sanofi aventis. Honoraria from Eli Lilly, Daiichi Sankyo, Astra Zeneca, Schering Plough l E.M. Antman: Grants from Eli Lilly, Daiichi Sankyo, sanofi aventis. Consulting or advisory board fees from sanofi aventis. Lecture fees from Eli Lilly, sanofi aventis l S.A. Murphy: Research grants from Eli Lilly, Daiichi Sankyo. Consulting fees from Eli Lilly. l E.R. Bates: Honoraria from Eli Lilly, Daiichi Sankyo, sanofi aventis, Bristol-Myers Squibb l Y. Rozenman: Consulting or lecture fees from Eli Lilly, sanofi aventis, Medtronic, Boston Scientific, Pfizer, Schering-Plough l J.L. Mega: Grants from Eli Lilly, Daiichi Sankyo, Johnson & Johnson. Honoraria from Bayer Healthcare. l M.S. Sabatine: Research grants from sanofi aventis and Astra Zeneca. Honoraria and/or consulting for sanofi aventis, Bristol-Myers Squibb, Astra Zeneca, Eli Lilly l E. Braunwald: Research grants and honoraria from Eli Lilly, Daiichi Sankyo.

Background PPIs are often administered with dual antiplatelet therapy for gastric protectionPPIs are often administered with dual antiplatelet therapy for gastric protection 1 Gilard et al., J Am Coll Cardiol Sibbing et al., Thromb Haemost 2009 Ex vivo studies have shown diminished platelet inhibition in response to clopidogrel in the presence of a PPI 1,Ex vivo studies have shown diminished platelet inhibition in response to clopidogrel in the presence of a PPI 1,2 3 Juurlink et al., CMAJ Ho et al., JAMA 2009 Recent studies have raised concerns that PPIs may attenuate the clinical benefit of clopidogrel 3,4Recent studies have raised concerns that PPIs may attenuate the clinical benefit of clopidogrel 3,4

FDA and EMEA Statements EMEA statement May 29, 2009: “The product information for all clopidogrel-containing medicines should be amended to discourage concomitant use of PPIs unless absolutely necessary”

N S O Cl O CH 3 C Clopidogrel Pro-drugs 3 N S O CH C O F O Thienopyridines: Formation of Active Metabolite Prasugrel Gut Oxidation (Cytochrome P450) Oxidation (Cytochrome P450) Active Metabolite HOOC HS HS N O Cl OCH 3 CYPs: 1A2 2C19 2B6 CYPs: 3A 2C19 2C9 2B6 CH 3 O N S O Cl O C 85% Inactive Metabolites N S O F O HOOC HS HS N O F CYPs: 3A 2B6 2C9 2C19 Active Metabolite Hydrolysis (Esterases) - Proton pump inhibitors - -?-? -

Primary Aim   To examine the association between PPI use and the risk of CV events for patients on either clopidogrel or prasugrel

TITAN ASA n= 13,608 Wiviott et al., NEJM 2007; 357: TRITON-TIMI 38

HR 0.81 ( ) P= Prasugrel Clopidogrel Days Endpoint (%) HR 1.32 ( ) P=0.03 Prasugrel Clopidogrel o EP: CV Death / MI / Stroke TIMI Major NonCABG Bleeds TITAN Wiviott et al., NEJM 2007; 357: TRITON-TIMI 38

Methods l l PPI use was at the discretion of the treating physician and captured on the case-report forms at each patient visit l l Propensity scores were constructed to predict PPI use l l Cox proportional hazards models included potential confounders and were stratified by propensity score strata

Primary and Secondary Analyses PPI use at randomization End of follow-up (median 14.5m) TRITON-TIMI 38 Trial 3 day and 30 day follow-up Landmark day 3 Landmark month 3 Landmark month 6 * All analyses were stratified by randomized treatment arm Additional sensitivity analyses  “Consistent” use of a PPI (defined as patients who were taking a PPI at randomization to end of follow-up)  Use of different types of PPIs  Use of a PPI or H2 receptor antagonist

PPI Use at Randomization n=4529, 33% of study population Type of PPIFrequency Pantoprazole1844 (40%) Omeprazole1675 (37%) Esomeprazole613 (14%) Lansoprazole441 (9.7%) Rabeprazole 66 (1.5%)

Baseline Characteristics PPI used (n=4529) No PPI (n=9079) P value Age (median) 61 y 60 y <0.001 Male72%75%<0.001 Caucasian94%92% <0.001 <0.001 Western Europe or North America66%54%<0.001 UA or NSTEMI75%73%0.007 Diabetes mellitus 24%23%0.14 Hypertension65%64% Hyperlipidemia57%55%0.07 History peptic ulcer disease 9.7%4.1%<0.001 Hemoglobin (g/dl) <0.001

CV death, MI or stroke Days CLOPIDOGREL PPI vs no PPI: Adj HR 0.94, 95% CI PPI use at randomization (n= 4529) Clopidogrel Primary endpoint stratified by use of a PPI

CV death, MI or stroke Days CLOPIDOGREL PPI vs no PPI: Adj HR 0.94, 95% CI PPI use at randomization (n= 4529) Clopidogrel Prasugrel PRASUGRELPPI vs no PPI: Adj HR 1.00, 95% CI Primary endpoint stratified by use of a PPI

ClopidogrelAdj HR (95% CI) Death, MI or stroke 0.94 ( ) CV death 0.71 ( ) MI 0.98 ( ) Stent thrombosis (ARC def/prob) 1.08 ( ) TIMI major or minor bleeding 1.13 ( ) Prasugrel Death, MI or stroke 1.00 ( ) CV death1.06 ( ) MI 1.02 ( ) Stent thrombosis (ARC def/prob) 1.03 ( ) TIMI major or minor bleeding 0.92 ( ) PPI use at randomization Favors PPI PPI worse

Risk of CV events for patients “consistently” on a PPI CV death, MI or stroke Days Defined as: PPI used at randomization and end of f/u (n= 2814) Patients exposed to PPI for shorter durations were censored (n=3882) CLOPIDOGREL PPI vs no PPI: Adj HR 1.05, 95% CI PRASUGREL PPI vs no PPI: Adj HR 1.10, 95% CI Clopidogrel Prasugrel

Short-term outcomes stratified by use of a PPI CV death, MI or stroke Days Clopi - Adj HR 1.00 ( ) Pras - Adj HR 1.14 ( ) PPI use at randomization (n= 4529) THROUGH 3 DAY FOLLOW-UP THROUGH 30 DAY FOLLOW-UP Clopi - Adj HR 0.98 ( ) Pras - Adj HR 1.09 ( ) Clopi Pras

Type of PPI Clopidogrel HR (95% CI) CV death, MI or stroke Prasugrel HR (95% CI) CV death, MI or stroke Omeprazole (n=1675) 0.91 ( )1.04 ( ) Pantoprazole (n=1844) 0.94 ( )1.09 ( ) Esomeprazole (n=613) 1.07 ( )0.86 ( ) Lansoprazole (n=441) 1.00 ( )0.98 ( ) Risk of CV events with different types of PPIs Rabeprazole not included due to small sample size (n=66)

Additional sensitivity analyses No association seen between PPI use and the risk of CV events for:  Use of a PPI or H2 receptor antagonist at randomization  Landmark analyses that captured PPI use at day 3, month 3 and month 6 after randomization  Landmark analyses to capture PPI use at different timepoints 

Limitations l Use of a PPI was not randomized, thus there is the potential for residual confounding l PPIs could be started or stopped during the course of follow-up

Conclusion l l In TRITON-TIMI 38, PPI use was not associated with an increased risk of CV events for patients on clopidogrel or prasugrel l l The current findings do not support the need to avoid concomitant use of PPIs in patients treated with thienopyridines

Available online at on September 1, 2009 THE LANCET