RE: Meta-analysis of the Relationship Between Dose and Benefit in Phase I Targeted Agent Trials

Extract

I read with interest the article by Gupta et al. describing the relationship between dose and outcome in phase I targeted agent trials ( 1 ). I am concerned, however, about the validity of the survival experience portrayed in Figure 1. According to the figure, participants in the cohort that Gupta et al. studied had estimated survival probabilities of 80%–90% at one year after trial entry. This contrasts sharply with estimates from other published meta-analyses of phase I trials, which generally suggest median survivals of six to nine months ( 2–7 ).

Inspection of the figure suggests that extensive and differential loss to follow-up may have led to biased estimates of survival probabilities. By four months, approximately 75% of the cohort was no longer at risk, in most cases due to censoring rather than to the occurrence of events. Presumably, most trial participants whose data were censored had come off study due to disease progression or toxicity. Because survival among this censored group was undoubtedly much worse than that among patients with longer follow-up, the survival experience reflected in Figure 1 is likely unrepresentative of that of the cohort as a whole.