COX-2 FDA Advisory Committee Meeting 2005 Rofecoxib Speaker: Ned S. Braunstein, MD Senior Director, Merck Research Labs

Consultants Marc Hochberg, MD Head, Division of Rheumatology and Clinical Immunology Professor of Medicine, Epidemiology & Preventive Medicine, University of Maryland School of Medicine Marvin A. Konstam, MD Chief of Cardiology, Tufts-New England Medical Center Professor of Medicine, Tufts University School of Medicine Loren Laine, MD Professor of Medicine, University of Southern California School of Medicine Chief, Gastroenterology Section, L.A. County and U.S.C. Medical Center

Objectives of Merck Presentation: Assessing Risk/Benefit of Rofecoxib/COXIBs Provide data on risk/benefit assessment of rofecoxib GI and CV data from NDA through APPROVe Methods used to acquire and analyze data Design considerations for rofecoxib study of CV outcomes New exploratory analyses Risk/benefit conclusions Feb-2005 for the class and unanswered questions Next steps

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

Overview: Rationale for Development of Selective COX-2 Inhibitors NSAID gastropathy Most common cause of drug-related morbidity and mortality Gastroduodenal perforation Gastroduodenal ulcers Upper GI bleeds Class labeled with bolded Warning

The COX-2 Hypothesis (1992) NSAIDs Selective COX-2 Inhibitors COX-1 Arachidonic Acid CO2H COX-1 COX-2 NSAIDs Selective COX-2 Inhibitors Prostanoids Prostanoids Protection of Gastric Mucosa Platelet Function Renal Effects Pain, Inflammation, Fever Gastropathy and Antithrombotic Effects Effects on Na+ Balance Anti-inflammatory and Analgesic Effects

Overview: Key GI Observations with Rofecoxib Demonstrated reduction in clinical upper GI events vs. non-selective NSAIDs Reduction vs. naproxen in VIGOR (outcomes study) Only selective COX-2 inhibitor with modification to NSAID-template GI Warning Consistent reduction vs. each of naproxen, ibuprofen, diclofenac Pooled analysis of 20 OA/RA studies More upper GI events with rofecoxib than placebo Reduced incidence of lower GI events vs. naproxen in VIGOR

Overview: Key Thrombotic CV Safety Observations with Rofecoxib Clinical data on thrombotic CV events for rofecoxib show: Increased risk relative to placebo Seen with long-term use in APPROVe Rates similar to non-naproxen NSAIDs Long-term data limited Increased risk compared to naproxen Apparent after short-term use

Overview: Key Public Health Questions What is risk/benefit of selective COX-2 inhibitors? Relative to placebo Relative to ibuprofen/diclofenac Relative to naproxen Can we identify factors associated with observed increased risk for thrombotic CV events with these drugs? Is observed increased CV risk a class effect of COX-2 inhibition? How big is the class? What are long-term CV effects of traditional NSAIDs?

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe 1998: Data in original NDA 2000: Data in VIGOR sNDA 2000-2004: Ongoing assessment post VIGOR until APPROVe findings APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

Phase IIb/III OA Pooled Analysis: Primary Endpoint Confirmed Clinical Upper GI Events (1998) Pts. With Events 23 22 Cumulative Incidence (%) with 95% CI Month Rofecoxib NSAIDs RR (95% CI): 0.45 (0.25, 0.81) p=0.006 Relative Risk (RR)=Rofecoxib vs. NSAIDs (Ibuprofen, diclofenac, nabumetone).

Investigator-Reported Thrombotic CV Events in Phase IIb/III OA Clinical Studies (1998) Rates per 100 Patient-Years (Number of Patients with Events) Clinical Studies of 6 to 86 Weeks Duration Investigator-reported cardiovascular events† Event Rofecoxib N=3357 2.0 (34) NSAIDs‡ N=1564 2.3 (16) Placebo N=711 2.6 (4) N=2253 2.7 (14) † Cardiac, Cerebrovascular, and Peripheral (Arterial and Venous) Events. ‡ Ibuprofen, diclofenac, and nabumetone.

COX-2 Cardiovascular Questions (1998) What is the clinical importance of inhibiting systemic prostacyclin synthesis without inhibiting platelet thromboxane? Can some NSAIDs, through their effects on COX-1, decrease the risk of thrombotic CV events? Is there a clinical benefit to inhibiting COX-2 mediated inflammation in atherosclerotic plaques?

Vascular Events Adjudication SOP Initiated (1998) Purpose Standardize the evaluation of cardiovascular events Predefined criteria All source documentation collected Blinded, external adjudication committees Improve clarity by eliminating questionable events Pooled analysis of events planned across all studies Increase precision

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe 1998: Data in original NDA 2000: Data in VIGOR sNDA 2000-2004: Ongoing assessment post VIGOR until APPROVe findings APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

VIGOR (Jan-1999 to Mar-2000) 8076 rheumatoid arthritis (RA) patients: Rofecoxib 50 mg QD 2 to 4 times recommended chronic dose Provides rigorous test of GI safety Naproxen 500 mg BID Extend GI findings to additional NSAID Most common NSAID regimen for RA Exclusion Criteria Patients using aspirin Confounds test of COX-2 hypothesis

VIGOR Primary Endpoint (2000) Time to Confirmed Clinical Upper GI Event Cumulative Incidence (%) with 95% CI Month RR (95% CI): 0.46 (0.33, 0.64) p<0.001 56 121 Rofecoxib 50 mg Naproxen 1000 mg Pts. With Events Patients at Risk Rofecoxib 50 mg Naproxen 1000 mg

Confirmed Thrombotic CV Events Rofecoxib vs Confirmed Thrombotic CV Events Rofecoxib vs. Naproxen: The VIGOR Study (2000) Cumulative Incidence (%) with 95% CI Month Rofecoxib 50 mg Naproxen 1000 mg RR (95% CI): 2.38 (1.39, 4.00) p=0.002 Pts. with Events 45 19

VIGOR Confirmed Thrombotic CV Events (2000) Patients with Events (Rates per 100 Patient-Years) (45) (28) (3) (20) (5) (11) (9) (2) (6) 1.7 1.0 0.2 0.7 0.2 0.4 0.3 0.1 0.2 Rofecoxib N=4047 2697 Patient-Years rate‡ (n) Naproxen N=4029 2698 Patient-Years rate‡ (n) Outcome† Any cardiovascular event Cardiac Sudden death Myocardial infarction Unstable angina Cerebrovascular Ischemic CVA Transient ischemic attack Peripheral events (19) (10) (4) (4) (3) (8) (8) (0) (1) 0.7 0.4 0.2 0.2 0.1 0.3 0.3 0.0 0.0 † Patients may be counted in more than one row but are only counted once within a row. ‡ Rate per 100 Patient-Years.

VIGOR Additional Exploratory Analyses Thrombotic CV Events (2000) More hypertension AEs with rofecoxib 50 mg than naproxen 1000 mg Similar relative risk in patients with or without increases in BP during study Similar relative risk in patients with or without baseline risk factors for CV events Merck analyses did not identify significant increases in relative risk over time for rofecoxib vs. naproxen

Alzheimer’s Disease and Mild Cognitive Impairment (MCI) Studies (Sep-2000) Combined analysis of two large ongoing placebo-controlled studies in patients with MCI and Alzheimer’s Disease 25 mg rofecoxib vs. placebo Elderly patient population with increased risk for CV events As of September, 2000: 2091 patients Median duration of therapy: 12.5 months

Investigator-Reported Thrombotic CV Events in the Pooled Alzheimer’s Studies (Sep-2000) Cumulative Incidence (%) with 95% CI Month Rofecoxib 25 mg Placebo RR (95% CI): 0.85 (0.53, 1.35) p>0.05 Pts. with Events 32 40 Patients at Risk Rofecoxib 25 mg 1041 947 879 769 640 359 Placebo 1050 991 930 816 693 376

Cardiovascular Pooled Analysis (Sep-2000) Phase IIb to V (post-marketing) rofecoxib studies 4 weeks duration APTC combined endpoint (MI, CVA, vascular death) Included studies not subject to adjudication Reports of APTC events had high confirmation rates Allowed comparison to published reports Pooled analysis of double-blinded patient-level data stratified by disease Included data on >28,000 patients and >14,000 patient-years

Relative Risk of APTC Events Pooled Analysis: Rofecoxib vs Relative Risk of APTC Events Pooled Analysis: Rofecoxib vs. Placebo or NSAIDs (Sep-2000) Relative Risk with 95% CI Favors Rofecoxib Favors Comparator Rofecoxib vs. Non-Naproxen NSAIDs Relative Risk (Rofecoxib/Comparators) Naproxen Placebo 0.84 (0.51, 1.38) 3867 8364 2918 1.69 (1.07, 2.69) 0.79 (0.40, 1.55) (65) (35) (84) (Pts. with events) PYR PYR = Patient-Years. Konstam et al., Circulation. 2001;104:2280-2288.

Merck’s Conclusions: VIGOR (2000) Clear evidence for GI benefit of rofecoxib Data versus placebo and non-naproxen NSAIDs did not suggest increased CV risk with rofecoxib Thus, weight of evidence most consistent with a cardioprotective effect of naproxen 1000 mg

Arthritis Advisory Committee Conclusions: VIGOR (Feb-2001) Clear evidence for GI benefit of rofecoxib vs. naproxen Data on CV risk inconclusive Both GI and CV data should be described in labeling

Updated VIOXX™ Labeling: GI and CV (Apr-2002) Modified GI Warning VIGOR GI results New CV Precaution VIGOR and Alzheimer’s Disease CV results Clinical Significance Unknown Caution should be exercised when VIOXX™ is used in patients with a medical history of ischemic heart disease Chronic use of 50 mg not recommended Increase in NSAID-type AEs at 50 mg No increased efficacy at 50 mg

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe 1998: Data in original NDA 2000: Data in VIGOR sNDA 2000-2004: Ongoing assessment post VIGOR until APPROVe Study of CV Outcomes Clinical Trials Data Assessment of Data APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

Considerations in CV Outcomes Study Design NSAID-controlled studies OA or RA patients Placebo control not appropriate in patients needing chronic NSAIDs Chronic NSAIDs not appropriate for patients without need Placebo-controlled studies Precludes study of chronic arthritis patients Applicability to arthritis population uncertain Size of studies to exclude 30% increased risk with >95% confidence and with 90% power More than 600 CV events Approximately 25,000 patients for 3 years

Considerations for CV Outcomes Study Design Placebo-Controlled Designs Acute Coronary Syndrome Known GI and renovascular risks of particular concern in this unstable patient population All patients on aspirin Could confound interpretation Chemoprevention Known GI and renovascular risks manageable Patients with broad spectrum of CV risk Aspirin users and non-users Not unlike patients with arthritis

Rofecoxib Study of CV Outcomes (Oct-2002) Prospective combined analysis of 3 studies comparing rofecoxib 25 mg vs. placebo APPROVe: Recurrent adenomatous colon polyps VICTOR: Colon cancer mortality (Oxford University Study) ViP: Incidence of prostate cancer in at-risk patients Separate protocol, analysis plan and safety monitoring board Approximately 25,000 patients Target 20 to 30% patients on aspirin Discussed with regulatory agencies

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe 1998: Data in original NDA 2000: Data in VIGOR sNDA 2000-2004: Ongoing assessment post VIGOR until APPROVe Study of CV Outcomes Clinical Trials Data Assessment of Data APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

Confirmed Clinical Upper GI Events: Relative Risk with 95% CI Final Pooled Analysis Confirms and Broadens GI Safety Benefit for Rofecoxib (2003) Confirmed Clinical Upper GI Events: Relative Risk with 95% CI Favors Rofecoxib Favors Comparator Ibuprofen PYR Relative Risk (Rofecoxib/Comparators) Naproxen* NSAIDs* Diclofenac VIGOR: Naproxen (97) (26) (30) (39) (Pts. With Events) (177) Rofecoxib vs. PYR = Patient-Years. *Excludes VIGOR.

Ph IIb/III OA Investigator-Reported Thrombotic CV Events (2003) Rofecoxib vs. Non-Naproxen NSAIDs Kaplan-Meier Estimates (95% CI) RR (95% CI): 0.98 (0.60, 1.62) p>0.05 Rofecoxib Non-Naproxen NSAIDs† Cumulative Incidence (%) with 95% CI Pts. with Events Patients at Risk Rofecoxib Non-Naproxen NSAIDs† 50 22 † Ibuprofen, diclofenac, nabumetone.

Kaplan-Meier Estimates (95% CI) Alzheimer’s Disease (PN 078 + 091) Confirmed Thrombotic CV Events (2003) Rofecoxib vs. Placebo Kaplan-Meier Estimates (95% CI) RR (95% CI): 1.01 (0.67, 1.53) p>0.05 Rofecoxib 25 mg Placebo Pts. with Events Patients at Risk Rofecoxib 25 mg 1069 878 707 415 318 226 185 Placebo 1074 939 797 463 385 283 243 42 48

Pooled Analysis of APTC Combined Endpoint, Rofecoxib vs Pooled Analysis of APTC Combined Endpoint, Rofecoxib vs. Comparator Agents (2003) Relative Risk with 95% CI (Pts. with events) Favors Rofecoxib Favors Comparator PYR Rofecoxib vs. Placebo 1.14 (0.77, 1.68) 5841 (105) Rofecoxib vs. Non-Naproxen NSAIDs 0.93 (0.51, 1.69) 4222 (47) Rofecoxib vs. Naproxen 1.61 (1.04, 2.50) 10318 (95) Relative Risk (Rofecoxib/Comparators) PYR = Patient-Years.

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe 1998: Data in original NDA 2000: Data in VIGOR sNDA 2000-2004: Ongoing assessment post VIGOR until APPROVe Study of CV Outcomes Clinical Trials Data Assessment of Data APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

Updated Safety Assessment (2004 Pre-APPROVe) Ongoing interest in Thrombotic CV safety of selective COX-2 inhibitors Observational epidemiology studies (10 presented or published) Results mixed Pre-clinical models Applicability to humans uncertain TARGET Thrombotic CV Data from rofecoxib randomized control trials: CV event rates similar to placebo and non-naproxen NSAIDs CV event rate higher than naproxen Similar CV data with other COX-2 selective inhibitors Overall risk benefit favorable for rofecoxib CV outcomes study ongoing

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

APPROVe Colon Polyp Prevention Study Design Rofecoxib 25 mg vs. placebo Approximately 2600 patients Stratified by baseline aspirin use 3-Year on-drug treatment period with 1-year off-drug period Colonoscopies Screening, 1 year, 3 year Follow-up at Year 4 after withdrawal of therapy (assess rebound) Primary endpoint: Cumulative incidence of patients with adenomatous polyps at Year 3 First patient screened in Dec-1999

APPROVe Inclusion/Exclusion Criteria Inclusion Criteria ≥40 years, histologically confirmed large bowel adenoma Prior History MI, PTCA, CABG allowed if >1 year prior to randomization TIA, CVA allowed if >2 years prior to randomization Exclusion Criteria Uncontrolled hypertension (>165/95 mm Hg), angina at rest or minimal activity, CHF at rest

APPROVe Confirmed Thrombotic CV Events (Sep-2004) Kaplan-Meier Estimates (95% CI) RR (95% CI): 1.96 (1.20, 3.19) p=0.007 Placebo Rofecoxib 25 mg Cumulative Incidence (%) with 95% CI Month Patients at Risk Rofecoxib 25 mg Placebo 1287 1123 1050 986 935 898 411 1299 1192 1148 1079 1039 1002 470 Pts. with Events 45 25

APPROVe Confirmed Thrombotic CV Events (Sep-2004) Any Confirmed Thrombotic CV event‡ Cardiac Events Acute myocardial infarction Fatal acute myocardial infarction Sudden cardiac death Unstable angina pectoris Cerebrovascular Events Fatal ischemic cerebrovascular stroke Ischemic cerebrovascular stroke Transient ischemic attack Peripheral Vascular Events Placebo (N=1299) 3315 Patient-Years (25) (11) (8) (3) (1) (4) (7) (0) (6) (2) 0.75 0.33 0.24 0.09 0.03 0.12 0.21 0.00 0.18 0.06 (n) Rate† Rofecoxib (N=1287) 3041 Patient-Years (45) (30) (20) (15) (5) 1.48 0.99 0.66 0.10 0.23 0.49 0.36 0.16 † Rate per 100 Patient-Years. ‡ Patients may be counted in more than one row but are only counted once within a row.

Rofecoxib Assessment CV Thrombotic Events (Post-APPROVe Sep-2004) Increased thrombotic CV risk rofecoxib 25 mg relative to placebo Non-constant relative risk in APPROVe Risk similar to placebo over first approximately 18 months Risk relative to placebo began to increase starting after approximately 18 months Potentially molecule specific Mechanism uncertain At the time, no placebo-controlled data available on other agents to support class effect

Reasons for VIOXX™ Voluntary Withdrawal (30-Sep 2004) Administrative committee communicated recommendation for termination of study treatment September 23, 2004 On the basis of the data, Merck voluntarily withdrew VIOXX™ from the market on September 30, 2004 APPROVe was first clinical trial with rofecoxib that showed an increased cardiovascular risk versus placebo Alternative therapies were available without evidence of a similar cardiovascular risk Merck believed voluntary withdrawal best served interest of patients

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal APPROVe Final Data and Exploratory Analyses of Risk Factors Thrombotic CV Events from CV Outcomes Study Implications of the Data

Post-Hoc Exploratory Analyses of APPROVe (Dec-2004 Data) Many factors assessed in multiple analyses Baseline factors (>10 factors) (e.g., age, gender, individual CV risk factors, etc.) Concomitant aspirin use Blood pressure (>40 analyses) Statistical approach: Tests for treatment-by-subgroup factor interaction, one subgroup factor at a time Multiple subgroup testing Results considered hypothesis generating

APPROVe Confirmed Thrombotic Events: Exploratory Post Hoc Analyses of CV Risk Factors (Dec-2004 Data) Relative Risk with 95% CI (Pts. with events) Favors Rofecoxib Favors Placebo PYR Total Cohort (72) Incr. CV Risk† (38) (34) Not Incr. CV Risk ASA User (16) (56) Non ASA User Hx Sympt. ASCVD (11) (61) No Hx Sympt. ASCVD Hx Diabetes (14) (58) No Hx Diabetes Relative Risk (Rofecoxib/Placebo) PYR = Patient-Years. † 2 or More Risk Factors for CV Disease or a History of Symptomatic Atherosclerotic CV Disease (ASCVD).

Blood Pressure Measurement Methodology in APPROVe BP measured once per visit Every 4 months BP measurements not standardized across sites Time of day and measurement technique varied Between-group difference in change from baseline in mean systolic and diastolic BP values (rofecoxib – placebo) 4 mm Hg systolic 2 mm Hg diastolic

APPROVe Exploratory Post-Hoc Analyses of Blood Pressure (Dec-2004 Data) Multiple BP analyses did not identify consistent patient subgroups or covariates associated with increased relative risk Baseline BP Change from baseline BP On treatment BP Hypertension reported as adverse experience One subgroup with increased relative risk SBP ≥160 mm Hg DBP = diastolic blood pressure. SBP = systolic blood pressure.

Rofecoxib 25 mg Reduces Cumulative Incidence of Colorectal Adenomas in APPROVe (Dec 2004) Endpoint Primary endpoint† (ITT Analysis) Relative Risk (95% CI) 0.76* (0.69, 0.83) Cumulative Incidence (%) (Number of Pts. with Adenomas) Rofecoxib 25 mg N=929 42 (376) Placebo N=956 56 (515) p<0.0001 *p<0.0001 † Primary endpoint evaluated over the entire 3-year study in patients with increased risk of colorectal cancer.

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal APPROVe Final Data and Exploratory Analyses of Risk Factors Thrombotic CV Events from CV Outcomes Study Implications of the Data

Pooled Analysis of Confirmed Thrombotic CV Events CV Outcomes Study (Interim Data Feb-2005) Relative Risk with 95% CI Pts with Events Favors Rofecoxib Favors Placebo PYR CV Outcomes Pooled Analysis (116) (72) APPROVe ViP* (28) (16) VICTOR* Relative Risk (Rofecoxib/Placebo) *Interim data as of Feb 2005.

All-Cause Mortality in Rofecoxib Clinical Program: (Updated Feb-2005) Rates per 100 Patient-Years with 95% CI NSAIDs controlled Placebo Controlled OA IIb/III VIGOR RA IIb/III ADVANTAGE ALZHEIMER* APPROVe VIP  VICTOR  100 Patient-Years Rate per  Rofecoxib 12.5-50 mg  Rofecoxib 25 mg  Rofecoxib 50 mg  Diclofenac + Ibuprofen  Naproxen  Placebo * On-drug population: Interim data as of Feb 2005

Outline of Merck Rofecoxib Presentation Overview Chronology of Rofecoxib GI and CV Safety Prior to APPROVe APPROVe and the Voluntary Withdrawal of VIOXX™ Data Since Withdrawal Implications of the Data

Implications: Key Public Health Questions What is risk/benefit of selective COX-2 inhibitors for established indications? Relative to ibuprofen/diclofenac Relative to naproxen Can we identify factors associated with observed increased risk for thrombotic CV events with these drugs? Duration Patient demographics Dose Is observed increased CV risk a class effect of COX-2 inhibition? How big is the class? What are long-term CV effects of traditional NSAIDs?

Next Steps (Feb-2005) Ongoing assessment of rofecoxib thrombotic CV data Examine additional factors for relationship in APPROVe Patients in APPROVe being followed off-drug Scientific hypotheses for thrombotic CV findings being explored Efforts underway to analyze thrombotic CV data across drugs Comparative outcome studies needed to determine relative risk among agents in relevant populations Etoricoxib vs. diclofenac MEDAL study in >23,000 patients targeted to complete 2006