Replies to post #129143 on Biotech Values

[TastyTheElf]

#msg-7959424 Study in Journal of Clinical Oncology and program survival bias

So I printed out the full article referenced here and read it. Maybe I'm obtuse, but the real point of that article seemed to me to be an indictment of "response rates" as an endpoint for clinical trials.

I've read elsewhere there are serious concerns about the meaningfulness of response rates, because it is all too common for positive response rates to fail to translate to a meaningful survival benefit.

The rationale for using that kind of endpoint in a Phase II trial is obviously speed -- you get the trial done faster.

This isn't meant to be an assault on the general validity of the program survival bias thesis -- I'm just not sure that that article is important support for it. I think what it tells me is that I should be wary of response rate as an endpoint, and lean toward Phase II => Phase III sequences where the endpoints were consistent -- ideally OS in both, with a reasonably defined progression perhaps coming in second, and response rate a clear third. (I should be particularly wary if the endpoint changed from a lower type to a higher type.)

Regards,
TGW

[size_14_EEE]

Regarding program-survival bias, phase 3 trials often have a similar internal problem between arms. This problem is often solved by adjusting results for stratification factors. I'm not saying that this would solve all potential discrepancies between phase 2 and 3 trials but some DD into this could be a positive move.

[genisi]

Semi OT on phase II failures potential (perhaps I've posed that two years ago. If so, please del. Btw, FOLD is mentioned).

When failure should be the option

By Gregory A Petsko

http://www.biomedcentral.com/1741-7007/8/61