Feed

Immunotherapy based on anti-PD1 inhibitors has significantly improved survival in advanced melanoma. However, a significant proportion of patients do not benefit, and predicting response to immunotherapy remains an area of unmet need. Our group previously defined an immune signature able to predict response to anti-PD1 inhibitors in this scenario.

In this study, we analyzed two cohorts of patients with advanced melanoma treated with anti-PD1 inhibitors: the GEM cohort, previously used to validate our immune signature, and Campbell's cohort, which contains data about different immunotherapy schemes. Using the 107 genes that compose our immune signature and consensus clustering, samples were classified as immune-low or immune-high. Then, CIBERSORTx and Ecotyper were used to estimate the proportion of each immune cell type and carcinoma ecotypes in both cohorts.

We confirmed that the immune-low group includes mostly patients who do not response to anti-PD1 inhibitors. We also studied the distribution of carcinoma ecotypes in the immune-high and immune-low groups defined by our immune classification. Ecotypes CE9 and CE10 clustered in the immune-high group, with good response to treatment. The use of combination immunotherapy improved response rate both in immune-low and immune-high tumors. The immune-high group contained a higher number of CD8 T cells, B memory cells and T follicular helper cells.

Our immune-based classification defines an immune-low group of tumors with poor response to anti-PD1 inhibitors. This immune classification is related to carcinoma ecotypes. Finally, a use of a combo scheme improves the rates of response both in immune-high and low groups but in the case of immune-low tumors, our results suggests that a combo treatment approach could be an adequate strategy and should be further explored in these patients. Altogether, our results support the utility of our immune signature in the prediction of response to anti-PD1 inhibitors in advanced melanoma.