Biotech Values

[urche]

EXEL

Can anyone shed some light on what is meant by a "randomized discontinuation trial"?
I am trying to understand the transcript of the recent Exelixis Q2 earnings conf call---and therein may lie some of my problem, relying on a poor quality transcript.
http://seekingalpha.com/article/152690-exelixis-inc-q2-2009-earnings-call-transcript?source=yahoo&page=-1

This is the befuddling passage:

We and BMS also believe that XL184 has potential beyond GBM and MTC. We recently initiated a previously planned, randomized discontinuation trial of XL184 in multiple tumor types, including pancreatic, prostate, heterocellular, gastric and GE junctional, melanoma, small cell lung and small cell lung with variant and breast cancers. This trial is designed to identify potential clinical activity across multiple indications as part of our long term plan with BMS to develop XL184 for the treatment of a variety of cancers.

Other than taking a hit on this approach of conducting Phase 2 trials in a small number of tumor types, we are undertaking a more comprehensive trial that should yield information about the potential activity of XL184 across a broad array of oncology indications. We and BMS believe that this approach will provide us with the data we need to maximize both the clinical and commercial opportunities for these very promising compounds.

One of the questioners in the Q+A session asked the question I would have asked and , IMO, came away just as befuddled:

Eric Schmidt - Cowen & Company

Okay. And then in terms of the randomized discontinuation study, that's a fairly novel design as at least in my experience. It's got a bunch of tumor types in there. How do you, how do you work to separate any potential signal from the noise and what size study are we talking about?

Mike Morrissey

You know it's a trial design that I think was initially validated with sorafenib. Certainly that's where the RCC indication came from, and I believe other indications as well. So across the nine, nine different histologies that we're looking at, we project a total of 600 patients in total and again there is, as this works in the classic paradigm, there's various enrollments goals initially to see a signal along with rules for stopping either based upon success or utility in the running phase and then for patients with stable disease are then looking at a randomization to be able to generate a rough PFS [meet] out.

So, in our view and certainly in the view of working closely with BMS, this is a very efficient way to really do some very careful analysis of variety of different tumor types in a very efficient manner. So we're excited about having this trial up and running and looking forward to wrapping it most.

I have not heard of such a study type and wonder if it has any validity. My hunch is that, much like what one can learn from retrospective sub group analysis, this type of study may be useful for hypothesis generation and not much else.

Urche