Not an expert , but don't antibody-drug-conjugates have many problems in their design that make them less the "ideal" targeting drug we're talking about - but if those are solved , their potential is quite big ?

Also , assuming we have the theoretical tech to sample a big set of cancer cells from a patient - collect a big list of all the unique characteristics, create a large collection of drug molecules targeted at all those characteristics with very very high specificity, and give them to a patient, won't this cure him in very high likelihood ?