Re: Feuerstein botches Satraplatin analysis
>I would submit that if the p value for death is substantially the wrong way, then the NDA is at substantial risk.<
It wasn't.
Each component of the composite primary endpoint favored the Satraplatin arm to some degree (the time-to-pain component was even statsig). Where Feuerstein went wrong is in thinking that the proportion of patients who progressed according to primary-endpoint component X in the Satra arm vs the control arm was a valid estimator of time to progression according to component X in the Satra arm vs the control arm. This, of course, is mathematically nonsensical.
In fact, among the various components in the primary endpoint, time to death was the component least strongly related to how well the drug was working! That’s because a death due to prostate cancer was usually preceded by an event corresponding to one of the trial’s other primary-endpoint components: radiologic progression by RECIST, pain progression, or medical intervention. When death occurred before any of these other events (the subgroup Feuerstein was writing about), the death in question was generally due to causes other than the cancer per se. Fewer than 10% of the patients in the final PFS analysis triggered the death component.
>As for MOA - chemo is dangerous. It doesn't seem unreasonable to suppose that two chemo drugs together may kill patients.<
The SPARC data to be presented at upcoming medical conferences will show that Satra had an excellent safety profile for a chemo drug—much superior to Taxotere and to other platinum drugs. The only drug given in the Satra arm other than Satra was prednisone. Regards, Dew
News 
Market Data 
Discover 
