The combination of anti-CTLA-4 and anti-PD-1 antibodies in patients with metastatic melanoma provides additive benefits over either modality alone. Now, an analysis of biopsy samples from patients with advanced-stage melanoma receiving sequential immune-checkpoint inhibition as part of the CheckMate 064 or 069 trials provides insight into the mechanisms of each class of agent.

Patterns of MHC expression in non-malignant and malignant skin samples obtained during CheckMate 064 were investigated using immunohistochemistry. MHC I was expressed on all non-malignant skin cells and on the majority of malignant cells, although 34 of 92 samples had MHC I expression on <50% of tumour cells. MHC II was detected in Langerhans cells in non-malignant skin samples, although cellular expression was detectable at the invasive margins in tumour biopsy samples.

In CheckMate 064, patients with MHC I expression on <30% of cells were more likely to have disease progression after 13 weeks of single-agent ipilimumab (P = 0.02), whereas a nonsignificant trend was observed between MHC II expression on >1% of cells and response to pembrolizumab (P = 0.052). Furthermore, baseline MHC I expression <50% was found to predict inferior overall survival (OS) in those receiving ipilimumab, despite the trial design allowing crossover to pembrolizumab (HR 0.38, 95% CI 0.18–0.82; P = 0.01). Conversely, baseline MHC II expression >1% was associated with superior OS in patients who initially received pembrolizumab (HR 0.11, 95% CI 0.02–0.83; P = 0.01).

Both associations disappeared in patients receiving concurrent ipilimumab plus nivolumab as part of the CheckMate 069 trial, despite MHC I expression <50% having a 100% negative predictive value for a response to ipilimumab monotherapy in this trial. These findings indicate that MHC I expression is required for a response to ipilimumab, while MHC II expression, likely owing to IFNγ-mediated immune activation, is predictive of a favourable response to pembrolizumab.