It seems like it would useful to have some kind of probationary approval for situations like this. If the data isn't convincing and the only alternative patient

It seems like it would useful to have some kind of probationary approval for situations like this. If the data isn't convincing and the only alternative patients have is death, approve the drug under the condition that the manufacturer must greatly subsidize the cost until they have data that is convicing or the manufacturer loses confidence and withdraws from the market. This might be a terrible idea given how complex and delicate the incentive structures are in healthcare.