Replies to post #277243 on Avid Bioservices Inc (CDMO)

[Protector]

jbainseky, on the biomarker that I will not answer because I don't know. On how long someone lives how-ever, which is key for the median tables and stat.sig. OS, our first endpoint, that I do know.

but how could unblinding have any effect on something like expression of a biomarker and how long someone lives?

It is generally though that patients condition may progress based on a 'believe' (the placebo effect) that they get some kind of treatment. This may slow down, or in 100% imaginary situations even cure, the patients disease. In cancer a slow down means living longer, or if we reverse that we could say event less rapidly.

This placebo effect can occur at ANY MOMENT as long as the patient is not evented yet. So if we consider a patient from start of treatment then he can even BEFORE unblinding and theoretically should not have had any placebo influence because he never knew what arm he was in OR the patients may survive the moment of UNBLINDING.

If he survives that moment then he knows what arm he was in (he must answer the maintenance question if he was a bavi arm patients). Hence at that moment one does NOT KNOW anymore if the patient lives EXTRA/LONGER because he knows that he got bavi and if he decided to participate in the trial must have BELIEVED bavi is a GOOD thing for him.

The other way around patients discovering that they didn't get bavi might due to the same effect now progress faster and live less long aka event sooner and contribute also to a tail separation.

The statistical significant data included patients results of patients that evented AFTER unblinding. And that data cannot go to the FDA. They cut-off at unblinding (what you have then can be used officially but was NOT stat. sig. in our case).