1. FDA Advisory Committee on COX-2 Inhibitors & NSAIDs Gaithersburg, Maryland February 16-18, 2005 Ernest Hawk, MD, MPH Office of Centers, Training, & Resources DHHS/NIH/NCI FDA Advisory Committee on COX-2 Inhibitors & NSAIDs Gaithersburg, Maryland February 16-18, 2005 Ernest Hawk, MD, MPH Office of Centers, Training, & Resources DHHS/NIH/NCI Celecoxib in Adenoma Prevention - The APC Trial

2. Colorectal Cancer - Global Statistics Worldwide (2000) Worldwide (2000) USA (2005) USA (2005) Incidence Mortality Incidence Mortality 895,000 489,000 895,000 489,000 145,290 56,290 145,290 56,290 7.9 % of cancer deaths (3 rd leading cause) 9.9% of cancer deaths (2 nd leading cause) Parkin M: Lancet Oncology 2:533-543, 2002 Jemal A, et al. CA Cancer J Clin 55:10-30, 2005 Parkin M: Lancet Oncology 2:533-543, 2002 Jemal A, et al. CA Cancer J Clin 55:10-30, 2005

3. 5-20 yrs 5-15 yrs ADENOMA NormalNormalMildMildModerateModerateSevereSevereCancerCancer APC, bcl -2, c- myc HypomethylationCOX-2 APC, bcl -2, c- myc HypomethylationCOX-2 K-ras SMAD 2 SMAD 4 DCC SMAD 2 SMAD 4 DCC P53 E-Cadherin P53 E-Cadherin Carcinogenesis is a Chronic Disease - From Molecular Defect to Dysplasia to Cancer Focus of Screening & Surgical Care Focus of Molecularly-targeted Intervention Adapted from Ilyas et al. Eur. J. Cancer 1999; 35:335-351

4. Evidence for NSAID Efficacy in Reducing Colorectal Cancer Risk Mechanistic:Mechanistic: –Induce apoptosis, r educe angiogenesis & p roliferation – Stimulate immune surveillance In vivo: > 100 studies in intestinal carcinogenesisIn vivo: > 100 studies in intestinal carcinogenesis Epidemiologic: > 30 studies suggesting 30-40% reductions inEpidemiologic: > 30 studies suggesting 30-40% reductions in –Adenoma & c olorectal cancer incidence –Colorectal cancer mortality Randomized trials of aspirin:Randomized trials of aspirin: –3 placebo-controlled trials showing 20-30% reductions in adenoma recurrence Evidence extends to other organsEvidence extends to other organs –Breast, prostate, lung, head & neck, esophagus, bladder, skin, plasma cells Mechanistic:Mechanistic: –Induce apoptosis, r educe angiogenesis & p roliferation – Stimulate immune surveillance In vivo: > 100 studies in intestinal carcinogenesisIn vivo: > 100 studies in intestinal carcinogenesis Epidemiologic: > 30 studies suggesting 30-40% reductions inEpidemiologic: > 30 studies suggesting 30-40% reductions in –Adenoma & c olorectal cancer incidence –Colorectal cancer mortality Randomized trials of aspirin:Randomized trials of aspirin: –3 placebo-controlled trials showing 20-30% reductions in adenoma recurrence Evidence extends to other organsEvidence extends to other organs –Breast, prostate, lung, head & neck, esophagus, bladder, skin, plasma cells

5. Established Safety Concerns of Nonselective COX Inhibitors (NSAIDs) 13 million chronic users (U.S.)13 million chronic users (U.S.) Dyspepsia in 10-20%Dyspepsia in 10-20% Serious GI complications in 1.3%Serious GI complications in 1.3% –Fatalities in 5-10% of these Total direct cost/year > $2 billionTotal direct cost/year > $2 billion ~ 15 th most common cause of death (U.S.)~ 15 th most common cause of death (U.S.) 13 million chronic users (U.S.)13 million chronic users (U.S.) Dyspepsia in 10-20%Dyspepsia in 10-20% Serious GI complications in 1.3%Serious GI complications in 1.3% –Fatalities in 5-10% of these Total direct cost/year > $2 billionTotal direct cost/year > $2 billion ~ 15 th most common cause of death (U.S.)~ 15 th most common cause of death (U.S.) Wolfe, et al: NEJM 1999 Mortality for 7 Selected Disorders (1997)

6. FAP PATHOPHYSIOLOGY FAP PATIENTS NSAID CANCER PREVENTION –Genetic basis of disease –Adenoma-carcinoma sequence –Serial endoscopic surveillance –COX-2 overexpression –Established preclinical models –Genetic basis of disease –Adenoma-carcinoma sequence –Serial endoscopic surveillance –COX-2 overexpression –Established preclinical models –Preclinical data –Observational data –Intervention data –Safety concerns –Preclinical data –Observational data –Intervention data –Safety concerns –Unmet clinical need –Limited cohort –MDACC & St. Mark’s registries –Unmet clinical need –Limited cohort –MDACC & St. Mark’s registries NCI - PHARMACIA COLLABORATIVE TRIAL OF CELECOXIB IN FAP NCI - PHARMACIA COLLABORATIVE TRIAL OF CELECOXIB IN FAP SEARLE/PHARMACIASEARLE/PHARMACIA –COX-2 technology –Celecoxib safety database –Celecoxib preclinical data –COX-2 technology –Celecoxib safety database –Celecoxib preclinical data

7. Phenotype & Natural History of Familial Adenomatous Polyposis (FAP) Germline APC gene defect Autosomal dominant transmission Frequency = 1/10,000 live births ~ 20-30% of cases may be due to spontaneous mutations ~ 100% lifetime cancer risk Average age at death = 42 years Germline APC gene defect Autosomal dominant transmission Frequency = 1/10,000 live births ~ 20-30% of cases may be due to spontaneous mutations ~ 100% lifetime cancer risk Average age at death = 42 years Extracolonic manifestations Extracolonic manifestations Duodenal adenomas/cancer Duodenal adenomas/cancer Desmoids Desmoids Despite standard care Despite standard care RR of death = 3.35 (n = 222) RR of death = 3.35 (n = 222) Celecoxib is the only approved pharmacologic adjunctive therapy Celecoxib is the only approved pharmacologic adjunctive therapy Nugent, KP et al. Dis Colon Rectum, 1993

8. Colorectal Efficacy of Celecoxib in Patients with FAP

9. 2,035 patients with prior sporadic adenomas Randomized 1:1:1 Colonoscopy after 12 & 36 months evaluating recurrence; collection of all adenomas NCI/Pfizer’s APC Trial Celecoxib 200 mg BID x 36 mos. Stratified by aspirin use & clinical center Stratified by aspirin use & clinical center Celecoxib 400 mg BID x 36 mos. Placebo x 36 mos. 91 Clinical Sites - 72 US, 10 Canada, 8 Australia, 1 UK Accrual - November 1999 to March 2002 91 Clinical Sites - 72 US, 10 Canada, 8 Australia, 1 UK Accrual - November 1999 to March 2002

10. Adjudication & Analysis of Serious CV Adverse Events in the APC & PreSAP Trials Recommended by DSMBsRecommended by DSMBs Independent Cardiovascular Safety CommitteeIndependent Cardiovascular Safety Committee –Clinical Endpoint Committee (blinded) Scott Solomon, MDScott Solomon, MD Peter Finn, MDPeter Finn, MD –Cardiovascular Review Committee Marc Pfeffer, MD, PhDMarc Pfeffer, MD, PhD John McMurray, MDJohn McMurray, MD Janet Wittes, PhDJanet Wittes, PhD Recommended by DSMBsRecommended by DSMBs Independent Cardiovascular Safety CommitteeIndependent Cardiovascular Safety Committee –Clinical Endpoint Committee (blinded) Scott Solomon, MDScott Solomon, MD Peter Finn, MDPeter Finn, MD –Cardiovascular Review Committee Marc Pfeffer, MD, PhDMarc Pfeffer, MD, PhD John McMurray, MDJohn McMurray, MD Janet Wittes, PhDJanet Wittes, PhD

11. CV Safety Adjudication & Analysis in the APC & PreSAP Trials - A 3-Step Process PlanningPlanning –Developed standard definitions, heirarchical categorization scheme & statistical analysis plan Data Compilation, Verification, & AdjudicationData Compilation, Verification, & Adjudication –Reviewed SAE forms & source documents –Queried sites to supplement study data –Adjudicated events –Created an ACCESS database of adjudicated CV SAEs AnalysisAnalysis –Obtained randomization codes & relevant baseline data –Analyzed data according to intent-to-treat principles –Presented data to DSMB PlanningPlanning –Developed standard definitions, heirarchical categorization scheme & statistical analysis plan Data Compilation, Verification, & AdjudicationData Compilation, Verification, & Adjudication –Reviewed SAE forms & source documents –Queried sites to supplement study data –Adjudicated events –Created an ACCESS database of adjudicated CV SAEs AnalysisAnalysis –Obtained randomization codes & relevant baseline data –Analyzed data according to intent-to-treat principles –Presented data to DSMB

12. APC Trial - Patient Characteristics at Baseline Baseline characteristic Placebo N = 679 Celecoxib 200 mg BID N = 685 Celecoxib 400 mg BID N = 671 Age in years 59.7  9.759.7  9.459.9  9.4 %% Male69.767.267.7 History of cardiovascular events 47.348.945.8 Myocardial infarction (MI)4.33.24.6 Cerebrovascular disease 2.12.91.9 Congestive heart failure (CHF)2.10.91.6 Angina7.57.36.3 Hypertension40.841.938.7 Diabetes † 9.09.69.5 Current smoker18.017.414.3 Aspirin use31.429.329.8 Lipid-lowering drug use27.127.428.5 Note: Plus-minus values are means ±SD. There were no significant differences among the groups. † Data were missing for one patient in the placebo group. Solomon SD, et al: N Engl J Med 352, 2005

13. Incidence of Hierarchical Cardiovascular Composite Endpoints in the APC Trial EndpointNumber of patients (%)Rate/1000 patient-years Placebo200 mg BID 400 mg BID Placebo200 mg BID 400 mg BID Death from CV causes1 (0.1)3 (0.4)6 (0.9)0.51.42.9 Death from CV causes or MI4 (0.6)12 (1.8)15 (2.2)1.95.87.4 Death from CV causes, MI, or stroke 6 (0.9)15 (2.2)20 (3.0)2.97.39.9 Death from CV causes, MI, stroke, or heart failure 7 (1.0)16 (2.3)23 (3.4)3.47.811.4 Death from CV causes, MI, stroke, heart failure, or angina 11 (1.6)18 (2.6)25 (3.7)5.48.712.5 Death from CV causes, MI, stroke, heart failure, angina, or need for a cv procedure 17 (2.5)26 (3.8)31 (4.6)8.412.715.5 Solomon SD, et al: N Engl J Med 352, 2005

14. Hazard Ratios for Hierarchical Cardiovascular Composite Endpoints in the APC Trial Solomon SD, et al: N Engl J Med 352, 2005 EndpointHazard Ratio with 95% Confidence Interval* 200 mg BID400 mg BID Death from CV causes3.0 (0.3-28.6)6.1 (0.7-50.3) Death from CV causes or MI3.0 (1.0-9.3)3.8 (1.3-11.5) Death from CV causes, MI, or stroke2.5 (1.0-6.4)3.4 (1.4-8.5) Death from CV causes, MI, stroke, or heart failure 2.3 (0.9-5.5)3.4 (1.4-7.8) Death from CV causes, MI, stroke, heart failure, or angina 1.6 (0.8-3.4)2.3 (1.1-4.7) Death from CV causes, MI, stroke, heart failure, angina, or need for a cv procedure 1.5 (0.8-2.8)1.9 (1.0-3.3) *Relative to placebo

15. Number & Percentage of Patients with Individual CV Events in the APC Trial Endpoint Placebo N = 679 n (%) Celecoxib 200 mg BID N = 685 n (%) Celecoxib 400 mg BID N = 671 n (%) Death from CV causes1 (0.1)3 (0.4)6 (0.9) Death from non-CV causes5 (0.7)3 (0.4) Death from any cause6 (0.9) 9 (1.3) Solomon SD, et al: N Engl J Med 352, 2005

16. APC Trial - CV Events by Baseline Subgroup Examined cardiovascular risk in various risk subgroupsExamined cardiovascular risk in various risk subgroups –Age, gender, cv risk factors, diabetes, aspirin use, use of lipid-lowering drug No statistical evidence of a differential hazard by baseline risk factorsNo statistical evidence of a differential hazard by baseline risk factors Few eventsFew events Limited powerLimited power Examined cardiovascular risk in various risk subgroupsExamined cardiovascular risk in various risk subgroups –Age, gender, cv risk factors, diabetes, aspirin use, use of lipid-lowering drug No statistical evidence of a differential hazard by baseline risk factorsNo statistical evidence of a differential hazard by baseline risk factors Few eventsFew events Limited powerLimited power Solomon SD, et al: N Engl J Med 352, 2005

17. Kaplan-Meier Estimates of the Risk of Serious CV Events in the APC Trial by Treatment Arm*

18. Solomon SD, et al: N Engl J Med 352, 2005 *In this analysis, “serious CV events” include death from CV causes, MI, stroke, or heart failure 671

19. Cross-trial Cardiovascular Safety Meta-analysis of Celecoxib TrialPlanned CohortDosing RegimenPlanned Duration APC~ 2,035 pts with prior adenomas 200 or 400 mg BID3-5 years PreSAP~1,560 pts with prior adenomas 400 mg QD3-5 years Selcel~1,600 pts with prior adenomas 400 mg QD3 years ADAPT~2,625 pts at risk for Alzheimer’s dementia 200 mg BID3 years MA27~6,830 post- menopausal pts with ER+ breast cancer 400 mg BID3 years NEI~86 pts with diabetic retinopathy 200 mg BID2 years

20. Long-term NSAID Use May Increase Risks of Cardiovascular Death - A Norwegian Study, 2005 Population-based, nested case-control study of 454 Scandinavian patients diagnosed with oral cancer between 1975 and 2003, and 454 gender- and age-matched controls. (Sudbo J, et al. 2005 - Manuscript submitted)Population-based, nested case-control study of 454 Scandinavian patients diagnosed with oral cancer between 1975 and 2003, and 454 gender- and age-matched controls. (Sudbo J, et al. 2005 - Manuscript submitted) 0 0 1 1 2 2 Hazard Ratio for CV Death in Long-term NSAID Users Compared to Non-users (with 95% Confidence Intervals) Hazard Ratio for CV Death in Long-term NSAID Users Compared to Non-users (with 95% Confidence Intervals) 2.06 3 3 5 5 4 4 6 6 Ketoprofen Piroxicam Indomethacin Naproxen Ibuprofen 1.16 2.86 1.70 2.26 1.84 1.90 Aspirin Any NSAID CV deaths 42 2 2 7 7 10 7 7 4 4 12

21. Issues Arising from the CV Safety Data with Celecoxib in the APC & PreSAP Trials ContextContext –Efficacy –Risk:benefit assessments –Relative GI and CV safety compared to other NSAIDs SafetySafety –Overall safety GI ulcerationGI ulceration –Accuracy & precision Cross-trials meta- analysisCross-trials meta- analysis ContextContext –Efficacy –Risk:benefit assessments –Relative GI and CV safety compared to other NSAIDs SafetySafety –Overall safety GI ulcerationGI ulceration –Accuracy & precision Cross-trials meta- analysisCross-trials meta- analysis –Mechanism(s) DoseDose FrequencyFrequency DurationDuration PharmacokineticsPharmacokinetics –Mitigation or risk management –Risk segregation Baseline risksBaseline risks Metabolic polymorphismsMetabolic polymorphisms Future ResearchFuture Research