Prognostic and predictive importance of body mass index and type 2 diabetes in the European Organisation for Research and Treatment of Cancer 1325/KEYNOTE-054 phase III trial of pembrolizumab versus placebo in resected high-risk stage III melanoma

Background: Metabolic conditions, such as obesity and type 2 diabetes mellitus (T2DM), cause changes in immune function that may influence immunotherapy effectiveness and immune-related adverse events (irAEs). Objectives: To investigate the prognostic and predictive effects of BMI and T2DM and investigate the effect of BMI on toxicity using data from the EORTC 1325/KEYNOTE-054 randomised controlled trial. Methods: Pembrolizumab (n = 514) was administered every three weeks for 1 year and prolonged recurrence-free survival (RFS) compared to placebo (n = 505) among patients with resected high-risk stage III melanoma. Here, we used multivariate Cox regression to investigate associations of BMI and T2DM with RFS, and Fine and Gray regression to investigate the association of BMI with the cumulative incidence of irAEs. Results: BMI had an approximately U-shaped association with RFS (p = 0.004) in both treatment arms combined. The RFS hazard ratios (HR)s for BMIs of 20, 30 and 35 kg/m2 (compared to 25 kg/m2) were 1.28 (95 % CI 1.05–1.56), 0.96 (95 % CI 0.88–1.06) and 1.14 (95 % CI 0.98–1.34), respectively. There was no evidence that BMI was associated with pembrolizumab effectiveness (p = 0.20) or the cumulative incidence of irAEs (p = 0.74). T2DM was not associated with RFS (HR 1.01, 95 % CI 0.73–1.40) in both arms combined and there was no evidence of an association between T2DM and pembrolizumab effectiveness (p = 0.83). In summary, in patients with resected high-risk stage III melanoma, BMI was associated with RFS in patients overall and within each treatment group. Conclusions: BMI appeared to confer a prognostic effect but was not predictive of pembrolizumab effectiveness regarding RFS. BMI was not related to irAEs among patients receiving pembrolizumab, while T2DM was not associated with RFS irrespective of treatment.