Abstract
Patients with advanced melanoma treated with immune checkpoint inhibitors can experience ongoing disease control after treatment discontinuation without subsequent systemic anticancer therapy. We previously defined a novel outcome, treatment-free survival (TFS), as the time between protocol therapy cessation and subsequent therapy initiation/death. We assessed the effect of established prognostic variables [lactate dehydrogenase (LDH), programmed death ligand 1 status, BRAF mutation status, performance status, and sex] on TFS in different treatment scenarios: treatment until toxicity/progression with frequent early cessation (nivolumab plus ipilimumab), treatment until toxicity/progression with a well-tolerated regimen (nivolumab), and treatment for a short fixed duration (ipilimumab). Data were pooled from 1077 patients with advanced melanoma treated in the CheckMate 069 and 067 trials. TFS was defined as the area between the Kaplan-Meier curves for time to therapy cessation and time to subsequent therapy initiation/death. TFS was estimated by restricted mean (r-mean) survival time at 36 months since randomization. Clinically meaningful TFS (r-mean TFS 3.7-12.7 months) was observed across all patient subgroups. TFS was longest in patients treated with nivolumab plus ipilimumab. The largest differences in r-mean TFS were observed with LDH in the nivolumab plus ipilimumab and ipilimumab treatment groups (TFS difference 4.7 and 4.9 months, respectively). In the nivolumab group, there was little difference in TFS across subgroups (r-mean TFS 3.7-5.5 months). TFS was sensitive to prognostic subgroup differences; however, duration of treatment affected the sensitivity of TFS. These results provide further support for TFS as a clinical outcome measure.