1. CCEB October 30 David J. Margolis MD PhD Associate Professor of Dermatology and Epidemiology Center for Clinical Epidemiology and Biostatistics University of Pennsylvania School of Medicine

2. CCEB Potential Conflict of Interest Not currently a consultant with Novartis and they are not directly paying for my time today. Recently completed a research project with Novartis via grant to Trustees of the University of Pennsylvania (UPENN). A study has been developed by me as a research study under the aegis of UPENN to conduct a 10-year malignancy safety study for Novartis.

3. CCEB Overview I will present my personal perspective As a committee you need to realize that clinical trial methodology and epidemiology methodology are not always the same

4. CCEB Atopic Dermatitis (AD) Common illness of childhood Protean cutaneous manifestations Waxing and waning clinical course Associated with atopic illnesses Mainstay of treatment, corticosteroids Newly released topical immunosuppressives used commonly Health care provider for by generalists, pediatricians, and dermatologists

5. CCEB Post marketing study At labeling, a post-marketing concern about long-term safety with respect to malignancy. All malignancies are rare in childhood. Lymphoproliferative malignancies are of greatest concern since they can occur in transplant patients SEER lymphoma rates (0.1 to 4 per 100,000)

6. CCEB Designs Clinical Trial –Randomization of exposure, blinded observer, determination of outcome, measurement of exposure Cohort Study (Registry) –Prospective data collection and determination of outcome, but exposure is selected by another mechanism

7. CCEB Feasibility (Clinical Trial) Very large study sample with years of follow up Physicians enrolling individuals with AD (e.g. dermatologists) not likely to diagnosis lymphoma For the diagnosis of lymphoma, biannual exam-based follow-up not likely to have impact on outcome but very likely to create concern where it does not currently exists. Randomized therapy over 5 to 10 years is unethical and not feasible Drop out

8. CCEB Feasibility (Epidemiology) Create large registry (cohort) of users with AD (real world diagnosis and use). Follow frequently and rely on patient reports of this catastrophic outcome (lymphoma) and drug exposure. Magnitude of exposure may create a natural comparison group(s). Confirm lymphoma with patient granted access to medical records. Implement protocols to maximize chances to maintain contact (minimize drop out), determination of exposure, and determination of outcome.

9. CCEB Synthesis To date after more than 3 years use, no worrisome signal In any study, it would be impossible to follow 100,000 individuals More reasonable would be 20,000 person-years over 10 years (power to rule out a 2 to 3 times increase rate) Rely on finding no more than an upper limit of risk to ensure safety Evaluate and compare natural exposure variations

10. CCEB Additional benefits A cohort contributing 20,000 person-years would be the largest study of atopic dermatitis on record. Many other natural history questions about AD can be addressed. Academic community is engaged and interested.

11. CCEB Alternative designs Peto trial or large simple trial –Too large, too long Database study –Retrospective in nature –Would probably require another five years of use within a large health network like GPRD before a study could even begin to be developed.

12. CCEB Conclusion Epidemiologic studies are the best way to study rare outcomes and really the traditional method for safety concerns (not a clinical trial). This is a wonderful opportunity to learn more about AD and the long term use of calcineurin inhibitors. –It would be a tragedy to create scientifically unsound barriers in designing this study. –It is a tragedy to continue to delay its implementation.