Rethinking therapeutic durability in cancer

Cancer therapies are commonly evaluated by target inhibition, pathway modulation, or initial tumor response, yet these measures do not reliably predict which treatments produce lasting disease control. Many pharmacodynamically active therapies produce transient benefit followed by relapse, while others achieve therapeutic durability despite comparable initial activity. This Perspective argues that durability is better explained by the behavior of the treated tumor system than by the depth of target suppression and distinguishes three modes of intervention behavior: buffered suppression, in which adaptive escape remains broadly accessible; adaptive constraint, in which future tumor states are narrowed without eradication; and collapse-dominant control, in which treatment drives a state transition that leaves few routes for persistence. Examples from targeted therapy, differentiation therapy, checkpoint blockade, and CAR-T therapy illustrate how the same modality can produce different intervention behaviors depending on tumor biology. A prospective sequence is also proposed for testing which behavior a candidate therapy is likely to induce.