According to GTx, the phase III study will be a success if the number of responders in the enobosarm arm is greater than the responders in the placebo arm. The difference favoring enobosarm has to be statistically significant as well.
The problem with a responder analysis is that it only tells you the benefit derived by individual responders, but says nothing about the clinical meaningfulness or magnitude of treatment effect for the patients as a whole.
A per-patient responder analysis can be useful but only in the context of knowing if the entire group of patients treated with a drug benefits over a control.
I'll use an extreme example to demonstrate my point.In a two-arm study, let's say 20% of patients treated with a drug respond but the other 80% either don't respond or are made worse off. In the placebo arm, 10% of patients respond but 90% don't respond or fare worse. In this hypothetical study, the number of drug responders is twice that of placebo and could very well be statistically significant even though the overall treatment effect for all patients is clinically meaningless or even negative. GTx will have you believe achieving a positive result via the responder analysis in the enobosarm phase III trials is enough, but that's very misleading. In order to be fully transparent, GTx must also disclose the mean treatment effect for all enobosarm patients compared to all placebo patients. Vertex Pharmaceuticals (VRTX - Get Report) ran into this problem last year with a phase II study of one of its cystic fibrosis drugs. At first, the company only provided investors with responder analysis data and was sharply criticized by analysts and investors for doing so. Later, Vertex provided the mean treatment effect for each arm of the study. Watch closely to see if GTx provides the mean treatment effect for enobosarm and placebo arms of the study, particularly the stair climb power endpoint. If the company doesn't, it's hiding something. Jason B. asks: " Vanda Pharmaceuticals (VNDA - Get Report) presented today