You should have actually read your second link. It describes exactly what happened to here.
If, however, the decision to change the endpoint is not independent of the trial data, then ‘‘ cherry-picking’’ is a serious concern. New endpoints may be selected because they displayed a trend towards signi?cance, while other candidate endpoints may have been examined but not selected or reported because they failed to display a desirable trend; this increases the chance of false positive (type 1) errors. In the Physicians’ Health Study [10,11], the trial’s data monitoring committee (DMC) recommended termination of the study because interim data seemed unlikely to show any bene?t of aspirin with respect to the primary endpoint, total mortality. At the time this decision was made, there was evidence of bene?t with respect to myocardial infarction. However, the United States Food and Drug Administration did not approve an indication for aspirin for the prevention of myocardial infarction, because t his was not the prespeci?ed primary endpoint.