Biotech Values

[AlpineBV_Miller]

The FDA's argument on PFS was that the differential scheduling in follow-up scans accounted for the statistically-significant PFS benefit.

They presented a number of scenarios to demonstrate how certain differentials would result in the PFS meaure losing statistical significance. However, the data presented contained math errors. When those math errors were corrected, all the outcomes* were statistically significant, even if the actual median PFS difference shrank to only a few days (i.e., not clinically significant).

* My memory, which might be a little foggy since it was three years ago, is that the FDA included some nonsense scenarios like keeping the progression date in the control arm the same and moving the Genasense progression date back to the previous scan before failure was detected. In that instance, the difference was not statistically significant. The ont thing I thought was very odd was the FDA spent all the time io argue this point, but never bothered to calculate the "correct" differential based upon the patient records they had in their hands. For all rational scenarios, however, the correc tmath retained statistical significance.