The SAP (Stat Analysis Plan) will define the target P, but the FDA will either agree or not at decision time.
The P value target will NOT be <.05 for the final look, it will be a bit (not much) lower. The reason is that the the early looks spend alpha (certainly the second, first is unclear).
The second line P2 was such a screw-up (for whatever reason) it is really not meaningful to discuss is being stat sig. You can not compare post hock 3 mg vs combined arms and claim stat sig (and they were not anyway). You can not compare 3MG + (placebo with some bavi) to (1mg bavi with some placbo) and claim this means anything at all.
Forgive me, but with a journalism background morphed into engineer/lawyer, and as an editor of a pair of professional journals with a proofreader's eye:
"... It has always been my understanding that a P value greaterthan 0.05 was considered to be statistically significant...."
I think you mean, instead: "lower than" or "better than" to avoid contradicting yourself later.
Well it still does not explain why the Sept 7th results were said NOT to be stat. sig. for INDIVIDUAL arms but were for COMBINED arms (0.015) with the ONLY value below 0.02 while BOTH individual arms had values BELOW (better then, smaller then) 0.05 which if 0.05 is the required FDA's p-value should have been stat. significant.
Furthermore I never heard of an SPA for the PIII.
PS: I suppose you mean smaller then (p<0.05) and not greater then (p>0.05).