Trial designs and results supporting treatment de-escalation and escalation.

Trials for escalation and de-escalation of treatment aim to improve patient care, but from different sides of the same coin with respect to disease control and burdens of treatment. De-escalation of therapy is inherently a non-inferiority question. A design with random assignment to standard of care versus de-escalated therapy is typically implemented but ordinarily will require a large sample size. Some research questions of treatment de-escalation might be asked in select patient populations using single-arm designs. Trials of therapy escalation may better inform the care of individual patients by posing a clinically-oriented research question in an enriched patient population, or by planning analyses that exploit the heterogeneity of a broadly-defined enrolled population.