De-escalation trials do not always need to be noninferiority: a case for superiority design de-escalation trials in oncology

De-escalation trials in oncology have received increased attention recently because there is a growing concern that patients with cancer are being overtreated-more patients (than would benefit) are being treated, earlier in the disease course, at a higher dose, for a longer duration, at a higher frequency. Thus, it is important to understand if less treatment allows us to achieve similar outcomes, a strategy referred to as de-escalation of treatment. However, one of the major concerns with such de-escalation strategies is the possibility of compromising treatment efficacy. While de-escalated treatment with lesser therapeutic burden is a worthwhile goal in itself because it leads to less physical, financial, and time toxicities, de-escalation cannot come at a substantial compromise of treatment efficacy. The commonest way to test whether such de-escalation strategies are safe and do not lead to unacceptable compromise in efficacy is through a non-inferiority design trial. Such trials require larger sample size, and thus, more funding and longer time to be completed. However, de-escalation trials do not necessarily need to be non-inferiority design. In this article, I make a case for using superiority design to test de-escalation strategies. This will avoid the limitations of non-inferiority design and make de-escalation strategies more efficient to test.