>Surely what you want us to pick is the expected result for the BEST VX-950 arm(s), not the overall response obtained from all 420 VX-950 patients , since the best data will guide P3 development.<
Using the SVR rate of the best PROVE-1/PROVE-2 regimen to guide the phase-3 trial design does make the most sense, but…
1. The SVR data will start to trickle in during early 2007, and VRTX doesn’t want to wait for all of the data before starting phase-3. (Indeed, Boger made it clear on the GS webcast that the PROVE-n trials will overlap with phase-3.)
2. There are 80, 80, 100, and 160 patients in the combined PROVE-1/PROVE-2 trials taking each specific VX-950 treatment regimen (see the table in #msg-11248731). If the differences between these groups in terms of the SVR rate are relatively small, it may be reasonable to treat the differences as noise and use an average figure of the four groups (or perhaps two or three groups) for planning purposes.
In any case, the survey is already written and some people have voted, so I’d better leave it alone. If I were doing it from scratch, I’d do it the way you suggested. Regards, Dew
Market Data 
Markets 
AI
