To: Biomaven who wrote (10041 ) 11/5/2008 9:23:00 AM From: Sidney Street 1 Recommendation Read Replies (1) | Respond to of 10345 Peter - drug holiday bandied about on the IV board. That board has lots of clutter, but here's an interesting exchange between srsmgja and nangasimon there. Nangasimon is pro- but not crazy. I'm still not convinced that in the general MS patient population the risk of PML increases over time. What we might be seeing, as these two agree, is increased occurrences due to larger N with compromised immune systems. One of the key assumptions here - 5% of MS patients have taken Azathioprine, Methotrexate or other immunosuppressants - is roughly consistent with a post by neuro111, who thought it was 3-5%. >>>>>>>>>> Medicine is replete with examples of drugs acting synergistically on a particular receptor for both beneficial and harmful effects. We all know PML rarely presents in a individual who is immunocompetent. Tysabri works by preventing lymphocytes from traversing into the central nervous system. If you add strong immunosuppressant to the equation, the risk of depleting all immunosurveilance increases, which then increases the risk of developing PML. There are still few data points, but I believe a much stronger statisitcal case can be made for a 'stratified risk'. The concept of a stratified risk for Tysabri, greatly inceases it chances for blockbuster sales status, IMO The PML data points thus far: - 2 cases in combo with Avonex out of a total of 500 patients over approx. 2 years - 3 cases with PRIOR RECENT history of strong immunosuppressant use. If we assume that only 5% of the current Tysabri users had prior recent strong immunosuppressant use (a very conservative estimate) that would give us a denominator of approx 1500-2000 total patients........... so 3 cases out of 1500-2000 -1 case in monotherapy WITHOUT prior recent history of strong immunosuppressant use out of approx. 38,000 pateints (commercial and trial patients) >>>>>>>> This poster, srsmgja, then asked nangasimon to think about these numbers using actual statistical analysis; here's the reply: >>>>>> i would say when something is obvious you don't need stats, but to go through the motions with your figures you could use either a Fisher's exact test or a Poisson model to question whether those numbers might have occurred by chance both 2/500 (the avonex combo cases) and 3/2000 (the three prior immunosuppression cases) are high sigificantly different from 1/38000 (the true mono case) P=0.0004