A 3-arm trial of a toxin and its antidote could show the same pattern.
Yes, in theory; however, I don’t think anyone of consequence believes that the above is what occurred in the Ipi trial. Given the way the data turned out (i.e. that the OS HR’s for the Ipi arms relative to gp100 were almost identical), I don’t see any reason why the 3-arm design will present an undue problem for the advisory panel or the FDA itself.
Had the data turned out differently, the 3-arm design might have been a serious problem, so one could say that BMY dodged a bullet.