Estimating Long-Term Survivorship Rates Among Patients With Resected Stage III/IV Melanoma: Analyses From CheckMate 238 and European Organization for Research and Treatment of Cancer 18071 Trials

Abstract

Purpose: Standard-of-care treatments for patients with resected stage III/IV melanoma include the immuno-oncology (IO) agents nivolumab (NIVO) and ipilimumab (IPI). This study used mixture cure models (MCMs) to estimate cure rates among patients treated with NIVO or IPI in the phase III CheckMate 238 (ClinicalTrials.gov identifier: NCT02388906) and European Organization for Research and Treatment of Cancer (EORTC) 18071 (ClinicalTrials.gov identifier: NCT00636168) trials, and to assess the impact of use of adjuvant immunotherapy on cure rates versus watchful waiting.

Methods: MCMs were applied to patient-level recurrence-free survival data from CheckMate 238 and EORTC 18071. Cured patients were assumed to experience no disease recurrence and mortality risks similar to the general population. Uncured patients were at risk of disease recurrence and all-cause death. The survival trend of the cured patients was estimated using life expectancy data for a general population with the same baseline demographic characteristics. A regression model assessed the odds ratios (ORs) of cure across key subgroups on the basis of baseline characteristics of the study populations.

Results: In CheckMate 238, estimated cure rates were 48.3% (95% CI, 41.8 to 54.9) with NIVO and 38.2% (95% CI, 32.7 to 44.1) with IPI. In EORTC 18071, estimated cure rates were 38.0% (95% CI, 32.1 to 44.2) with IPI and 29.2% (95% CI, 24.4 to 34.6) with placebo. In the indirect comparison of the two trials, the odds of cure were significantly higher with NIVO than with placebo (OR, 2.33 [95% CI, 1.49 to 3.65]).

Conclusion: Analyses involving two large phase III trials investigating adjuvant IO treatment for resected melanoma demonstrate higher cure rates for both NIVO and IPI than placebo, with NIVO providing the highest cure rate. Similar cure rates were estimated for patients treated with IPI in both trials, despite staging and dosing differences.

FIG 1.

Estimated cure rates (95% CIs) for the CheckMate 238 and EORTC 18071 trials, including successive data cuts of CheckMate 238 with different lengths of follow-up. EORTC, European Organization for Research and Treatment of Cancer; IPI, ipilimumab; NIVO, nivolumab.

FIG 2.

Kaplan-Meier and MCM-estimated (cured, uncured, and mixed populations) RFS curves in the CheckMate 238 trial. The vertical dashed line indicates the length of minimum follow-up. The annual proportions of patients in each latent subgroup (cured and uncured) are labeled. IPI, ipilimumab; MCM, mixture cure model; NIVO, nivolumab; RFS, recurrence-free survival.

FIG 3.

Kaplan-Meier and MCM-estimated (cured, uncured, and mixed populations) RFS curves for the IPI and placebo arms in the EORTC 18071 trial. The dashed vertical line indicates the length of median follow-up. The annual proportions of patients in each latent subgroup (cured and uncured) are labeled. EORTC, European Organization for Research and Treatment of Cancer; IPI, ipilimumab; MCM, mixture cure model; RFS, recurrence-free survival.

FIG 4.

Observed (smoothed) and MCM-estimated RFS hazard rates in the CheckMate 238 and EORTC 18071 trials compared with background mortality rates. Shaded areas around the observed hazard rates indicate 95% CIs. Vertical dashed lines indicate the lengths of minimum follow-up in CheckMate 238 and median follow-up in EORTC 18071. EORTC, European Organization for Research and Treatment of Cancer; IPI, ipilimumab; MCM, mixture cure model; NIVO, nivolumab; RFS, recurrence-free survival.

FIG 5.

Effects of specific baseline factors on RFS cure rates in a reference population, estimated from the indirect treatment comparison pooling data from the CheckMate 238 and EORTC 18071 trials. Three patients from CheckMate 238 trial were excluded from the analyses because of unknown disease stage. Whiskers indicate 95% CIs for the ORs. CIs excluding 1 imply the statistical significance of the associations. EORTC, European Organization for Research and Treatment of Cancer; IPI, ipilimumab; NIVO, nivolumab; OR, odds ratio; RFS, recurrence-free survival.

FIG A1.

Schematic overview of the MCMs. MCMs, mixture cure models; RFS, recurrence-free survival. ^aParametric distributions were used to represent the probability of an event at different time points as a function of a given set of parameters.