"...if it was by such a slim margin that perhaps 8 narrowly defined 3 point MAC events could tip the scale."

First of all, the patient populations and/or trial protocols may be different enough between the Phase 2s and Phase 3 BETonMACE to prevent the "pooling" of MACE from Phase 2s into the BETonMACE pile. For example, there was a recent ACS requirement for BETonMACE that did not exist for the Phase 2s. Also, there may have also been a stricter low-HDL requirement for BETonMACE. Additionally, the SUSTAIN and ASSURE trials were done quite some time before BETonMACE; it is not like pooling two temporally parallel trials.

Second, BETonMACE was all diabetics. There were only 3 3-point MACE events by my count in the pooled SUSTAIN and ASSURE that came from diabetics. The 8 3-point MACE were from non-diabetics and diabetics combined. So IF one were to pool past Phase 2 3-point MACE into the BETonMACE pile, it is most appropriate to only do this on the 3 events from diabetics. Adding 3 Phase 2 events out of ~275 or so from BETonMACE is not likely to make any difference, in my opinion. Unless it was a razor thin margin of BETonMACE missing the primary endpoint with a p-value of like p=0.06ish without those additional 3 Phase 2 events.

BDAZ