i'm not convinced you need SVR data to move to 3 agents - if it becomes clear (and it's starting to) that a 3rd agent is necessary all you need is requisite safety data that said your timeline is probably more accurate than mine
Elaborating on your point about protocol tweaking, trials with three unapproved drugs in combination will likely need at least five arms: the triple combination, each of the three double combinations (to ensure that each drug in the triple is actually needed), and a control arm.