OBJECTIVES: Lack of post-recurrence survival data poses challenges for clinical and economic evaluation of emerging therapies in resected melanoma. This study evaluated an NMA-based approach in predicting PDRS for patients with primary resected Stage II/III melanoma.

METHODS: PDRS was estimated with an NMA of first-line metastatic melanoma treatments and compared against individual patient-level data (IPD) for distant recurrent patients with primary Stage II/III melanoma in randomized clinical trial (RCT) and real-world (RW) settings. In the RCT dataset, patients with Stage IIIA/B/C (n=232) or Stage IIIB/C (n=215) disease treated with either Nivolumab or Placebo across CheckMate-238 and EORTC-18071 trials were pooled into two cohorts. In the RW dataset, patients with Stage II (n=311) or Stage III (n=85) disease from the retrospective US-based Flatiron Health advanced melanoma database were pooled into two cohorts. For each cohort, time-varying treatment effects from parametric distributions derived by the NMA were applied to the survival of the reference treatment and used together with the corresponding post-distant recurrence treatment distribution to generate an aggregate survival curve. Long-term PDRS extrapolations from IPD were obtained from standard parametric distributions and spline-based models. Range of discounted lifetime mean PDRS was estimated across top-3 statistical fits.

RESULTS: For the Stage IIIA/B/C RCT cohort, projections from the NMA and IPD were 2.84-3.37 and 3.52-3.79 years, respectively with marginal impact on the results when Stage IIIA patients were excluded. For the RW cohorts, projections from the NMA and IPD were 4.14-4.72 and 5.46-5.94 years, respectively, for Stage II patients, and 3.82-4.40 and 5.24-5.48 years, respectively, for Stage III patients. The underprediction margin across top-3 statistical fits by the NMA ranged between 0.4-0.41 and 0.86-1.08 years in the RCT and RW cohorts, respectively.

CONCLUSIONS: The NMA-based approach provides a conservative, flexible and scalable framework for estimating PDRS of resected Stage II/III melanoma patients in the absence of IPD.