<<Opinions as to likely reaction of share price if interim is not "successful"? What is "success" for the interim? Minimum p value for interim that would still suggest final p value may be stat sig?>>
Typically, companies and DSMBs do not announce the p value of the interim data if the trial continues. You're extremely lucky if the company says that there is a "trend" toward significance. Not sure what DNDN will say if the p value is worse than 0.05 (50% odds, IMO). I don't have a great idea what would happen to the stock price in this case. I imagine there would be a significant drop...just don't know how big it would be, and whether or not there would be somewhat of a recovery after the initial drop.
There is a strong possibility that DNDN files for approval if the p value is between the 0.01 interim alpha guesstimate and the 0.05 typical threshold. I think the price will go up in this case, but IMO the same ilk (Scher, Hussain, Pazdur, Fleming, ASCO, Sendek, Aschoff, Shenouda) who opposed approval before will oppose it again.
If the p value is better than the interim alpha, IMO there will be less opposition to approval (perhaps only Sendek, Aschoff, Shenouda, and one of the other examples above). However, a lot is going to depend on the relationship between the log rank and Cox regression p values. If log rank is >0.05 and Cox is >0.01 but <0.05, IMO there will be significant opposition to approval from all of the elements in the previous paragraph. If log rank is better than 0.05 and Cox is still worse than 0.01, there will be less opposition. If log rank is better than 0.05 and Cox is better than 0.01, the opposition will be lesser still. The price should increase proportionately with better p value. However, we still don't know how much info the company will announce, even if they do announce they are re-filing.
I think we will see a lot of FUD being stirred up by the negative analysts no matter what, especially Aschoff and Shenouda. There will probably be insinuations of faked data, or improper Cox regression calculations. DNDN mgmt has not helped this at all with their secretive policy over the Cox regression methodology for 9902B. We don't know for sure if they will use all 20 covariates in the Halabi nomogram and gradually eliminate the non stat sig factors, or if they will just use the three covariates that were stat sig in both 9901 and 9902A and do a straight calculation with those three.
