Let's tease apart some of the sticky-wicket issues of FDA approvals and revocations. For one, how does the FDA measure success?
When designing clinical trials, researchers determine which endpoints or outcomes to measure in order for the FDA to justify its thumbs-up or thumbs-down decision on approval.
For cancer trials, the most obvious outcome is overall survival or mortality. The problem with survival as a primary endpoint is that meaningful results may demand unacceptably high cost ($$ and resources) and length of time.
For example, clinical trials of treatments for a type of cancer with an average life expectancy of, say, 7 years might require decades to complete. Given the rapid pace of progress in our understanding of the biology of cancer, such trials become outdated before providing useful information.
Consequently many trials also look at interim or secondary endpoints, such as tumor response (how much patients' tumors shrink) or progression-free survival (how long before patients' tumors grow). These secondary outcomes can result in an increase in the number of "events" and thus decrease the time and number of patients needed to reveal trends in an intervention's effect.
The wicket gets sticky when researchers disagree about whether favorable interim outcomes correlate with lower mortality or justify approval.
It's possible for a treatment to improve progression-free survival while shortening overall survival: Imagine tumors that remain stable longer on treatment, but once they start to grow they take off, progressing more quickly -- and shortening patients' overall survival.
Healthy Survivors consider the endpoints when assessing news of trial results.