1. James Cross, MS Pharmaceutical Outcomes Research and Policy Program University of Washington ----- Biobehavioral Cancer Fellows Day April 20, 2007 A risk-benefit analysis of celecoxib for the prevention of colorectal cancer

2. 2 The framework & decision problem Nonsteroidal antiinflammatories (NSAIDs)  risk of colorectal adenoma. NSAIDs  risk of gastrointestinal & cardiovascular adverse events. What is the risk-benefit profile of these drugs for colorectal cancer chemoprevention?

3. 3 Colorectal cancer: US estimates for 2006 1 Lifetime risk: 1 in 18 diagnosed with CRC. Diagnoses: 148,610 Deaths: 55,170 1 Ries L et al. SEER Cancer Statistics Review, 1975-2003, NCI. Bethesda MD. http://seer.cancer.gov/csr/1975_2003/ Based on November 2005 SEER data submission.

4. 4 Polyps: precursors to adenocarcinoma Standard colonoscopy Zielinski SL. JNCI 2004. Colonoscopy: current surveillance method among “high-risk” patients.

5. 5 NSAIDs Non-selective inhibitors Ibuprofen, naproxen, diclofenac Aspirin COX-2 selective inhibitors Celecoxib

6. 6 Inflammation in colorectal cancer Ulrich C, Nat Rev Cancer 2006

7. 7 Evidence of benefit: NSAID vs. placebo at 3 years 1 Baron JA, NEJM 2003 2 Bertagnolli MM, NEJM 2006 3 Arber N, NEJM 2006.

8. 8 Evidence of risk (CV, GI) 1 Derry L BMJ 2000. (RCT meta-analysis, n=66,000) 2 USPhysician Health Study NEJM 1989. 3 Solomon SD Circulation 2006. GI bleed risk for ASA: OR: 1.59 (1.40-1.81) 1 OR: 1.77 (1.07-2.94) 2 CV risk for Celecoxib:

9. 9 Current opinion of celecoxib …Due to the increased risk of CV events associated with their use, COX-2 inhibitors are not recommended routinely for sporadic adenomas. -Practice Guideline in Oncology v1.2007, Nat’l Comprehensive Cancer Network …It is reasonable to conclude that celecoxib has no role as a chemopreventive agent either in patients with nonfamilial colonic adenomas or the general population. -Psaty and Potter NEJM 2006

10. 10 Reasons for doing a risk benefit analysis Proposed risks: Cardiovascular (celecoxib) Gastrointestinal (aspirin) Proposed benefits: Colonoscopies not perfectly sensitive Slows carcinogenesis process Decreases # of adenomas Decision problems: Efficacy data based on surrogate endpoint No methodical, quantitative assessment

11. 11 Risk benefit analysis: methods Objective: compare the net health outcomes (risk & benefit) between 3 CRC prevention strategies ASA vs celecoxib vs colonoscopy alone Perspective: societal perspective, 20 years Population: 60 years, prior finding of adenoma Approach: health-state transition model quantifying health outcomes over 20 year period

12. 12 Polyp free (Post- polypectomy) Death Advanced adenoma CRC GI Tox CV Tox Discontinue ASA/COX Health-state transition model of CRC prevention

13. 13 Methods: Model assumptions GI/CV serious adverse events require drug discontinuation. Those who discontinue drug assume health state transitions as though they were receiving only colonoscopy.

14. 14 Methods: RBA estimates 1 Bertagnolli M. NEJM 2006 2 Arber N. NEJM 2006 3 Baron J NEJM 2003

15. 15 Results: Risk and benefit outcomes 1 Bertagnolli M. NEJM 2006 2 Arber N. NEJM 2006 3 Baron J NEJM 2003

16. 16 Results: Net health impact # cancer case/death + # cardiovascular event/death = net health impact COMPARATOR [net health impact – REFERENCE [net health impact] For a cohort of 100,000 (undiscounted):

17. 17 Limitations Scarcity of data: correlation between surrogate endpoints and outcomes used in decision-making. How should this be handled here, & in general? ASA use: How best to model cardioprotective effect of ASA use, which celecoxib users may also take? For calculations, 1 cancer case/death = 1 GI/CV event.

18. 18 Acknowledgements Dissertation committee: Lou Garrison (UW, chair) Scott Ramsey (UW & FHCRC) Dave Veenstra (UW) Funding: Biobehavioral Cancer Prevention & Control Training Program (NCI & UW)