Nice break down - however I am confused by your partition of probabilities.
Regarding Senario 1 -FDA ask whether should wait for OS. I think this senario is highly unlikely because FDA almost always ask 2 questions ie whether it is effective or whether it is safe.
Any rate, the sum of probabilities (conditional or marginal) should be 1 regardless of which senario is played out.
Under senario 1 - your 1a and 1b together amounts to 104% (16%+88%) >100%? under senario 2- your 2a and 2b added up to only 76%, this is OK, if there is a third outcome which you did not list which should sum up to 100%?
Howeer, your 1a, 1b, 2a and 2b did sum to 1 which is good, using the numbers you provided, one can figure out the conditional probs and marginal probs for 1 and 2 as follows:
You handicaped a 74% probability for 1), and under 1, your conditional prob should be 0.65 for 1a, and 0.35 for 1b.
Then you handicaped a 26% probability for 2), and under 2, your conditional probs should be 0.73 vs 0.27, for 2a and 2b respectively.
If I were correct in saying that senario 1) has 0 probability, then the accelerated approval probability would be 73%.
Market Data 
Markets 
AI
