In elan's primary analysis, they used a linear model to measure benefit of Bapineuzumab over PBO, insteead of calculating benefit on actual observed data. For some reason, they pre-specified the analysis would be done this way. Maybe because there are usually a lot of dropouts in AD trials.
For me, the remaining concern about bapi ph 2 data is this....
the average 18 mnth decline on ADAS-COG (gold standard for cognitive measurement) in hisotrical controls is about 8 points.
In the bapineuzumab carrier group, the bapi tx pts declined by 7.5 pts and the PBO declined by 8.5pts.
In the non-carrier group, the bapi tx pts declined by 7 pts and the PBO declined by 13pts.
Almost all of the benefit was do to a poorly perfroming PBO.
besides all that, i think the science is very good and makes good sense. i think bapi's chances in earlier AD or MCI would be better. There are a lot of dead neurons by the time patients have mild or moderate AD.