Perhaps my response will put some things in perspective relative to recent discussions/disagreements.
First, please consider the role of the FDA reviewers. Their analysis when considering a drug/biologic for approval is (in their own words):
*Rooted in Science *Codified by Regulation *Validated by Review *Described in Label
The ethical dilemma you hypothesize is observed benefit in the mesenchymal subgroup.
Fine. The analysis of the second Phase 1 data suggests this may be true.
You asked what would happen if an analysis of the Phase 3 data also suggests this benefit in the mesenchymal subgroup.
My response would be: that would be great news for such patients. However, it would require confirmation in a prospectively designed trial.
This may be hard to swallow, but it is true.
To explain, let me quote something else you said:
Linda does like "multiple shots on goal".
So, this is the thing, you don't get "multiple shots on goal" in a Phase 3 registrational trial. If you did, one, or several, of those shots would go in by chance alone, not due to skill.
You must pre-specify your analysis, minimize any bias, and control your error.
The biotech scrap heap is littered with failed trials with promising, even compelling, subgroup results.
Fleming likes to say (and I'm paraphrasing from memory): If I (as a statistician) present clinicians with evidence of treatment benefit in a subgroup, they can, invariably, come up with a biologically plausible reason why that subgroup benefits from the treatment."
But that observation is "data driven". To confirm the observation you need to prospectively study it.
The issue involves "multiplicity". If you look at multiple subgroups in a data set, chance alone suggests some will be positive.
Off the top of my head I can offer 4 examples of this: Soligenix (OrBec) in graft vs host disease Oblimerson in patients with low LDH Celsions HEAT trial in patients ablated for longer than 20(?) minutes Stimuvax in patients expressing high serum MUC1
All 4 of the above had promising subgroup data from their Phase 3 trials (after failing the primary analysis) supported by strongly plausible biologic reasons why those subgroups should indeed benefit from the treatment.
The very promising Oblimerson trial results failed to be repolicated, Celsion is running another trial and Stimuvax was bagged.
The OrBec data were very compelling, it really was. The meeting was so heated it makes debates here look like a church service.
The sponsor had to be removed from the meeting, kicking and screaming.
Ultimately, the FDA demanded another trial in their promising subgroup. Investors couldn't believe the evil FDA would be so cold hearted to desperately needy patients.
The trial proceeded.
And it failed.
The biologically plausible, very promising, subgroup data were a fluke. By demanding a confirmatory trial, the FDA spared those needy patients from an ineffective therapy.
So, no. I don't think there will be an ethical dilemma introduced by promising Mesenchymal subgroup data here (absent a Stat Sig benefit in the primary analysis). The FDA would say "great, fantastic, good work. Now please perform a confirmatory trial to prove it".
*Rooted in Science *Codified by Regulation *Validated by Review *Described in Label
There is no getting around that.
(quick note: the above was off the top of my head, I originally thought it was the oblimersen data that resulted in the elevator/ADCOM fiasco, it was Orbec. Whatever. I'm guessing there are people fact checking my posts hoping if they find an error they can discredit everything i say. Just for the record, the above is my recollection. I spent too much time digging up that photo and transcript to double check my recollection. I need to run to work but will try to address some of the other comments from last night later)
News 
Market Data 
Discover 
