it's possible that they could set the alpha as high as 0.02 for the interim look. That's what I would do.
That is so wildly different than what I do I would ask your assumptions. I know at least one area where you and I differ is the assumed validity of the 9901/2a Cox Regression, but where else?
Are you assuming the HR seen in the CR is the 'correct' HR?
Are you assuming that the Ego effect is not real? That 9901 and 2a were flukes and in reality both have constant HR?