what could go wrong?, page-106

"I have generalised percentage result but my question is more on what constitutes a good result."

PFS extension of 6 weeks + is needed in order to make the study meaningful( A Good result 8 weeks +), however if the control also shows unexpected improvements over the company's statistical assumptions, then that could affect things like the Hazard ratio, and perhaps what is needed in order to achieve a statistically significant result. Best call the company to clarify this, and ask them some questions based around different scenarios- eg "What happens if patients in the control arm achieve PFS of 6 months, but the PFS extension in the HA arm is still 8 weeks?"

"Statisitical Assumptions: Generated from published literature in identical patient populations where:
Control arm PFS of 4.0 months versus HA-Irinotecan arm PFS of 5.5 months. Targeting PFS with 6 week difference (Primary Endpoint) 85% power to detect a Hazard Ratio of 0.73 (log-rank test; 2-sided alpha = 0.05) when compared with the Phase II the hazard ratio was 0.46 "

So results in the Phase 3 in terms of improvements in PFS do not have to be greater than the Phase 2, the study is powered to show significance from a smaller difference in PFS.

The Phase 2 study showed a 12 week improvement in PFS, while the Phase 3 study is targeting an 8 week + improvement. Since the Phase 3 study is larger and more robust( among other variables),it' becomes easy to determine if there's a statistical difference between the two arms. As also mentioned above, Phase 3 failure rates in oncology are high, and results in Phase 3 studies are generally less favourable than what was shown in the Phase 2- I assume this would also be the case even when a study meets its end point.