Looking at it from an engineering perspective it's a pretty dumb problem to go after. Coming mostly from my experience with minification of electronics. E.g. the problems you get at 150nm are almost the same you'd get at 140nm or close but going down to 22nm is a whole different story. The effects that you could skim over before begin to matter in a critical way forming a sort of butterfly effect of ridiculous comple…
I think it's likely that a lot of these treatment paths end up cumulatively increasing QALY's (quality-adjusted life years) by more than they increase life expectancy. That is, they add more good time than total time, in part because of a lot of the factors you mention. If we pick up 5 years of QALY and 4 years of increased lifespan, that's a pretty huge win, IMO. It means getting to know grandchildren much more meaningfully as the typical age of parents at childbirth has increased.