Yeah, it is rare to have ORR as primary endpoint in randomized trial, which is why it is very odd they did that especially considering the touted MOA of the drug.
There is other way to make money too :-)
Yes, crossover can give misleading OS if study drug doesn't work by preventing placebo arm getting more effective subsequent treatments while study drug arm got better subsequent treatments. This is why people shouldn't look past weak PFS primary endpoint data to OS secondary endpoint data in crossover trial. This just shows people shouldn't look top line number alone from small randomized oncology trial. It is the detail that predict drug effectiveness, and whether it could be replicated in larger trial.