1. Proton Pump Inhibitor Use is Likely a Marker for, Rather than a Cause of, a Higher Risk of Cardiovascular Events: Insights from PLATO Shaun G. Goodman, Robert Clare, Karen S. Pieper, Stefan K. James, José C. Nicolau, Robert F. Storey, Warren J. Cantor, Dominick J. Angiolillo, Steen Husted, Christopher P. Cannon, Ph. Gabriel Steg, Kenneth W. Mahaffey, Jan Kilhamn, Robert A. Harrington, Lars Wallentin, on behalf of the PLATO Trial Investigators

2. The PLATO trial was funded by AstraZeneca Shaun Goodman: Significant research grant support from Astra Zeneca, Bristol Myers Squibb, Daiichi Sankyo, Lilly, Novartis, Sanofi Aventis Modest consultant/advisory board honoraria from Astra Zeneca, Bristol Myers Squibb, Lilly, Merck, Teva Other Author Disclosure Information available in the abstract: Circulation 2010;122:A12092 Ticagrelor is not yet approved for use Disclosures/Conflicts of Interest

3.  Conflicting data exist regarding the potential adverse interaction between clopidogrel and proton pump inhibitors (PPIs)  PPIs inhibit the cytochrome P450 2C19 isoenzyme and conversion of clopidogrel into its active metabolite  In contrast, ticagrelor is an ADP P2Y 12 inhibitor that does not require biotransformation and has no known interaction with PPIs ADP Receptor Antagonists and Proton Pump Inhibitors

4.  To examine the association between proton pump inhibitor (PPI) use and clinical outcomes for acute coronary syndrome (ACS) patients randomized to clopidogrel or ticagrelor Objective

5. PLATO Study Design Primary endpoint: CV death + MI + Stroke Primary safety endpoint: Total major bleeding 6–12-month exposure Clopidogrel (n=9291) If pre-treated, no additional loading dose; if naive, standard 300 mg loading dose, then 75 mg qd maintenance; (additional 300 mg allowed pre PCI) Ticagrelor (n=9333) 180 mg loading dose, then 90 mg bid maintenance; (additional 90 mg pre-PCI) NSTE-ACS (moderate-to-high risk) STEMI (if primary PCI) Clopidogrel-treated or -naive; randomised within 24 hours of index event (N=18,624) James et al Am Heart J 2009;157:599-605

6. PLATO Main Endpoints No. at risk Clopidogrel Ticagrelor 9,291 9,333 8,521 8,628 8,362 8,460 8,124 Months 6,650 6,743 5,096 5,161 4,047 4,1478,219 024681012 12 11 10 9 8 7 6 5 4 3 2 1 0 13 K-M estimated rate (% per year) 9.8 11.7 HR 0.84 (95% CI 0.77–0.92), p=0.0003 Clopidogrel Ticagrelor 024681012 10 5 0 15 Clopidogrel Ticagrelor 11.2 11.6 HR 1.04 (95% CI 0.95–1.13), p=0.43 K-M estimated rate (% per year) Months Major Bleeding CV Death, MI, Stroke 9,186 9,235 7,305 7,246 6,930 6,826 6,6705,209 5,129 3,841 3,783 3,479 3,4336,545 Wallentin et al N Engl J Med 2009;361:1045-57

7.  Pre-specified subgroup analysis  Proton pump inhibitor use was at the physician’s discretion  Multivariable Cox model with propensity adjustment and landmark analysis  The primary endpoint was the 1-year composite of CV death, MI, or stroke Methods

8. Primary and Secondary Analyses PPI use at randomization End of follow-up Median Time to Death/Censoring (IQR) = 358 Days (266, 369) PLATO Trial All analyses were stratified by randomized treatment arm Landmark Day 60 Landmark Day 180 Landmark Days 2, 4, 9, 30 Landmark Day 90  Any PPI vs. Non-PPI Gastric Supressive (e.g., H 2 antagonist) therapy  Any PPI vs. no GI therapy

9. PPI Use at Randomization Type of PPI* Frequency (%)* Omeprazole 3200 (17) Pantoprazole 1967 (11) Esomeprazole 764 (4) Lansoprazole 510 (3) Rabeprazole 97 (<1) 18601 of 18624 (99.9%) patients had documentation regarding PPI use prior to randomization → 6539 (35.2%) were taking a PPI * Type of PPI available in n=6538

10. Selected Baseline and Index Event Characteristics Characteristic No PPI (n=12,060) PPI (n=6539) Median age, years 62 63 Age ≥75 years, %1517 History, % Dyslipidemia Chronic renal disease Peripheral arterial disease 45 3.8 5.8 50 5.1 6.8 Peptic ulcer Percutaneous coronary intervention Coronary-artery bypass grafting Congestive heart failure Chronic obstructive pulmonary disease 0.9 12 5.4 5.7 5.3 2.4 15 6.9 4.6 6.8 ECG at entry, % ST-segment depression5249 Troponin-I positive, % Baseline hemoglobin, median TIMI Risk Score, % ST Elevation MI, ≥3 Non-ST Elevation ACS, ≥5 GRACE Risk of In-Hospital Death, median % 82 141 43 42 1.66 86 138 47 1.71 All p<0.05

11. Primary Outcome by Randomized Treatment and PPI Use 0 2 4 6 8 10 1214050100150200250300350400 Days 9.19 10.92 10.96 13.03 Clopidogrel + PPI (n=3255) Ticagrelor + PPI (n=3284) Clopidogrel + No PPI (n=6020) Ticagrelor + No PPI (n=6040) % of Patients with CV Death/MI/Stroke

12. Unadjusted and Adjusted* Cardiovascular Outcomes by Randomized Treatment and PPI Use Hazard Ratio & 95% CI 0.81.21.6 No PPI Better PPI Worse 1.41.0 Cardiovascular Death, MI or Stroke Unadjusted Ticagrelor 1.23 (1.07,1.41) Clopidogrel 1.22 (1.08,1.39) + P P2Y12 * PPI 0.96 *Propensity Adjusted Clopidogrel 1.20 (1.04,1.38) Ticagrelor 1.24 (1.06,1.45) 0.72 Cardiovascular Death or MI *Propensity Adjusted Clopidogrel 1.20 (1.03,1.40) Ticagrelor 1.26 (1.07,1.48) Unadjusted Ticagrelor 1.24 (1.08,1.44) Clopidogrel 1.27 (1.11,1.45) 0.84 0.94 + P2Y12 inhibitor Treatment * PPI interaction P-value

13. Clopidogrel 0 1 2 3 4 5 6 % of Patients 4.39 3.43 4.90 4.24 Ticagrelor 1-Year Non-CABG PLATO Major Bleeding* by Randomized Treatment and PPI Use * Kaplan-Meier estimates + Propensity-adjusted PPI No PPI PPI + Hazard Ratio 1.30 (95% CI 1.00-1.70) + Hazard Ratio 1.02 (95% CI 0.80-1.29) P interaction for P2Y12 * PPI =0.17

14.  Patients (n=1826) on non-PPI gastrointestinal drugs (e.g. H 2 receptor antagonists) prior to randomization were at similar risk to those on a PPI  Clopidogrel:HR 0.98 (0.79-1.23)  Ticagrelor:HR 0.89 (0.73-1.10)  Patients (n=10236) on no gastric therapy were at significantly lower risk of the primary endpoint  Clopidogrel:HR 1.29 (1.12-1.49)  Ticagrelor:HR 1.30 (1.14-1.49)  Landmark analyses accounting for PPI use (at days 2, 4, 9, 30, 60, 90, and 180) post-randomization showed no increased risk of the primary endpoint in those receiving a PPI  Except in patients who prematurely discontinued study treatment (clopidogrel or ticagrelor) from day 180 post-randomization (PPI vs. no PPI: HR 4.31 [1.70-10.95]) Additional Analyses PPI vs. non PPI GI treatment PPI vs. no GI treatment

15.  Pre-defined subgroup analysis with multiple comparisons → individual subgroups may have been underpowered to show an association between PPI use and clinical outcomes  Use of a PPI was not randomized → potential for residual confounding despite multivariable adjustment and propensity score for the decision to treat with a PPI  PPIs could be initiated or discontinued during the course of follow-up → landmark analyses employed  Different types of PPIs with potentially different effects on CYP2C19 and clopidogrel metabolism Limitations

16.  The apparent association between PPI and clopidogrel use and adverse events is highly confounded  PPI use may simply be a marker for, rather than a cause of, a higher risk of CV events  Regardless of PPI use, ticagrelor was superior to clopidogrel in preventing ischemic events Conclusions