An estrogen response-related signature predicts response to immunotherapy in melanoma

Abstract

Background: Estrogen/estrogen receptor signaling influences the tumor microenvironment and affects the efficacy of immunotherapy in some tumors, including melanoma. This study aimed to construct an estrogen response-related gene signature for predicting response to immunotherapy in melanoma.

Methods: RNA sequencing data of 4 immunotherapy-treated melanoma datasets and TCGA melanoma was obtained from open access repository. Differential expression analysis and pathway analysis were performed between immunotherapy responders and non-responders. Using dataset GSE91061 as the training group, a multivariate logistic regression model was built from estrogen response-related differential expression genes to predict the response to immunotherapy. The other 3 datasets of immunotherapy-treated melanoma were used as the validation group. The correlation was also examined between the prediction score from the model and immune cell infiltration estimated by xCell in the immunotherapy-treated and TCGA melanoma cases.

Results: "Hallmark Estrogen Response Late" was significantly downregulated in immunotherapy responders. 11 estrogen response-related genes were significantly differentially expressed between immunotherapy responders and non-responders, and were included in the multivariate logistic regression model. The AUC was 0.888 in the training group and 0.654-0.720 in the validation group. A higher 11-gene signature score was significantly correlated to increased infiltration of CD8+ T cells (rho=0.32, p=0.02). TCGA melanoma with a high signature score showed a significantly higher proportion of immune-enriched/fibrotic and immune-enriched/non-fibrotic microenvironment subtypes (p<0.001)-subtypes with better response to immunotherapy-and significantly better progression-free interval (p=0.021).

Conclusion: In this study, we identified and verified an 11-gene signature that could predict response to immunotherapy in melanoma and was correlated with tumor-infiltrating lymphocytes. Our study suggests targeting estrogen-related pathways may serve as a combination strategy for immunotherapy in melanoma.

Keywords: estrogen; gene signature; immune checkpoint blockade; melanoma; tumor-infiltrating lymphocyte.