1. An epidemiologic perspective on etoricoxib David J. Graham, MD, MPH Office of Surveillance and Epidemiology April 12, 2007

2. Introduction  What is known about NSAID-related UGI hospitalization and mortality?  What is known about NSAID-related cardiovascular risk?  Is diclofenac a reasonable comparator for a drug that will be marketed to millions in the US, many with underlying CV disease?  What is known about the performance of COX-2 selective coxibs compared to other therapies with respect to –GI risk –CV risk  Based on current state of knowledge, should etoricoxib be approved?

3. Deaths due to NSAID gastropathy (1) Singh, Rheumatology 1999; Suppl 56 Estimated total GI deaths while taking NSAIDs, not deaths due to NSAIDs. Case fatality rate estimated @ 17%.

4. Deaths due to NSAID gastropathy (2) Gutthann et al., Epidemiology 1997 First time hospitalization for UGI events among patients ever treated with NSAIDs, Saskatchewan, Canada, 1982-1986

5. Deaths due to NSAID gastropathy (3) US hospitalization and mortality data, 1999-2003 Acute + chronic UGI ulcers, perforations, bleeds Mean population Hospital discharges 1 Mortality 2 284,887,691 332,000 4,714 (1.4%) (includes gastric ulcer, duodenal ulcer, peptic ulcer site unspecified, gastrojejunal ulcer, gastritis and duodenitis ± hemorrhage, perforation, or obstruction: 1 ICD 9 codes 531-535; 2 ICD 10 codes K25-K29) 1 National Hospital Discharge Survey, 1999-2003, available at: http://www.cdc.gov/nchs/about/major/hdasd/listpubs.htm 2 Compressed Mortality File (CMF) compiled from CMF 1999-2003, Series 20, No. 2I 2006 on CDC WONDER On-line Database

6. Meta-analysis of clinical trials of AMI risk with coxib NSAIDs compared to placebo Kearney, et al., BMJ 2006 CoxibPlacebo

7. Meta-analysis of clinical trials of AMI risk with coxibs compared to traditional NSAIDs Kearney, et al., BMJ 2006 RR=0.45 (0.30-0.68) naproxen v. coxib Coxib tNSAID

8. Risk estimates for AMI with various NSAIDs compared to non-use from meta-analysis of observational studies Relative risk Diclo (9) Ibuprof (16) Indo (6) Naprox (15) Pirox (4) Other (19) Celecox (11) Rofecox (11).7.9 1 1.2 1.4 1.7 McGettigan & Henry, JAMA 2006

9. Observational studies of AMI risk with NSAIDs published since JAMA meta-analysis Diclofenac Ibuprofen Naproxen Celecoxib Etoricoxib Rofecoxib.6.8 1 1.2 1.5 2 3 4 Relative risk Andersohn, Circulation 2006 Helin-Salmivera, Eur Heart J 2006 Brophy, Heart 2007

10. Risk of AMI with selected NSAIDs from meta-analyses of observational (▬) and randomized (▬) studies Diclofenac Ibuprofen Naproxen.6.8 1 1.4 2 2.5 Relative risk (9) (26) (16) (24) (15) (42)

11. COX-2 selectivity of various NSAIDs Patrono, et al. J Clin Invest 2001

12. COX-1 and COX-2 selectivity of various pain relievers FitzGerald and Patrono, N Engl J Med 2001

13. Prescription NSAID use in the US, 2000-2006 Source: Verispan, Vector One™: National, Extracted March 2007

14. Prescription non-coxib NSAID use in the US, 2000-2006 Source: Verispan, Vector One™: National, Extracted March 2007

15. Diclofenac as a reference group (1) Rofecoxib vs. Diclofenac Hospitalized AMI: RR = 0.99 (0.82-1.19) Pharmacoepidemiology Drug Saf 2006; (epub November) Etoricoxib vs. Diclofenac APTC events: RR = 0.96 (0.79-1.16) Lancet 2006; (epub November)

16. Diclofenac as a reference group (2) What if an appropriate reference had been used? Drug n Risk (per 100 pyrs) Relative risk (95% CI) Etoricoxib 1960 1.09 (0.72-1.58) 2.72 (1.18-6.27) Naproxen 1497 0.41 (0.16-0.83) Dr. Robert Shibuya, FDA presentation

17. Cumulative probability of recurrent ulcer bleeding in patients treated with celecoxib or diclofenac+omeprazole Chan et al, New Engl J Med 2002 p=0.60 (logrank test)

18. Cumulative probability of recurrent ulcer bleeding in patients treated with celecoxib or diclofenac+omeprazole (2) Chan et al., New Engl J Med 2002

19. Cumulative probability of recurrent ulcer bleeding in patients treated with celecoxib or naproxen+lansoprazole (1) Lai, et al., Amer J Med 2005

20. Cumulative probability of recurrent ulcer bleeding in patients treated with celecoxib or naproxen+lansoprazole (2) Cum prob (95% CI) Celecoxib 3.7% (0%-7.3%) Naproxen+PPI 6.3% (1.6%-11.1%) Difference - 2.6% ( - 9.1%-3.7%) Lai, et al., Amer J Med 2005

21. Summary of published observational studies of NSAID+PPI use StudyOutcome# Cases NSAID+PPI vs. Nonuse NSAID+PPI vs. NSAID Epidemiology 1999 Recurrent UGIB760 (0-1.0) Epidemiology 2001PUBs21050.5 (0.2-1.1) Aging Clin Exp Res 2003UGIB2251.05 (0.2-5.7) BMJ 2005PUBs94070.83 (not avail)0.3 (0.2-0.5) Gut, 2006 Am J Gastro 2007UGIB27770.9 (0.7-1.3)0.13 (0.09-0.19) Pharmacoepi Drug 2007PUBs9791.3 (0.7-2.5)0.4 (0.2-0.7)

22. FD&C Act Sec 505(d) Requirement for approval of a new drug: "...adequate tests by all methods reasonably applicable to show whether or not such drug is safe for use under the conditions prescribed, recommended, or suggested in the proposed labeling thereof"

23. Public health considerations (1)  Cardiovascular disease leading cause of mortality in US –Background risk of AMI in males 65-74: 2% (1 in 50 per year)  This demographic is a large segment of target population for etoricoxib  Potential impact based on RR=2.72 (etoricoxib v. naproxen) –Among males 65-74, risk of 5.4% (1 in 18 per year) –NNH Etoricoxib = 147 person-years –6,800 extra APTC events per 1,000,000 person-years etoricoxib use  If etoricoxib use becomes extensive, multiplier effect on number of premature and excess CV events, compounded year after year –Estimated 88,000-140,000 excess AMI events with rofecoxib

24. Public health considerations (2)  Not all NSAIDs are the same with respect to CV risk –Diclofenac Low use in US Increases AMI risk Inappropriate and deceptive comparator for CV safety –Naproxen More widely used Neutral with respect to AMI risk Does not interfere with beneficial CV effects of aspirin  tNSAID + PPI equivalent to coxibs for UGI outcomes –RCT data –Epidemiologic data  No apparent or demonstrable added benefit to etoricoxib use

25. Conclusions (1)  Diclofenac is an inappropriate comparator for assessment of population cardiovascular risk –Applicant’s program is based on this inappropriate comparator  Etoricoxib probably confers substantial increase in CV risk –Enormous public health and population consequence  Etoricoxib is no more effective for pain relief than tNSAIDs  Naproxen + PPI equivalent to coxibs for gastroprotection –Substantial cardiovascular safety advantage –Substantially less expensive

26. Conclusions (2)  Approval should be based on “...adequate tests by all methods reasonably applicable…” –Demonstration of efficacy: Current “tests” probably adequate Etoricoxib comparable to tNSAIDs for pain relief; no advantage to 60 mg dose over 30 mg –Demonstration of safety: No cardiovascular safety data for 30 mg strength Current “tests” not adequate or reasonable Coxib superiority over naproxen + PPI for UGI outcomes Coxib superiority over naproxen + PPI for CV outcomes