How good is good enough?

A recurring question for companies developing cancer compounds is how much efficacy they need to see in Phase II trials before pulling the trigger on Phase III studies. For many companies, the answer has been "very little." But as might be expected, many of these compounds have gone on to fail in larger, controlled studies.

The latest example of the low predictive value of open-label studies against historical controls is the outcome of two Phase III trials of Alfacell Corp.'s Onconase, a cytotoxic RNase enzyme, which degrades RNA in cancer cells...