Development and validation of an immune and stromal prognostic signature in uveal melanoma to guide clinical therapy

Abstract

The tumor microenvironment is known to play an important role in uveal melanoma. Reliable prognostic signatures are needed to aid high risk patients and improve prognosis. Uveal melanoma tissues from three public datasets were analyzed. RNA sequence data of uveal melanoma and corresponding clinical features were obtained from The Cancer Genome Atlas database. Immune and stromal scores were calculated by applying the “ESTIMATE” algorithm. The samples were divided into high and low immune or stromal score groups. We constructed prognostic models by using the ‘lasso’ package and tested them for 500 iterations. The cell signature was validated in another GSE44295 and GSE84976 datasets. We found that the median survival time of the low immune/stromal score group is longer than that of the high-score group. Thirteen immune cells and one stromal cell were concerned significant in predicting poor overall survival rate. Finally, a four-cell model was identified. Further validation revealed that the low-risk group has a significantly better survival than the high-risk group in another two datasets (P < 0.05). Moreover, the high-risk group is more sensitive to immunotherapy and chemotherapy. Summarizing, the proposed immune cells signature is a promising biomarker for estimating overall survival in uveal melanoma.