Landscape of Infiltrated Immune Cell Characterization in Uveal Melanoma to Improve Immune Checkpoint Blockade Therapy

Numerous studies indicated that tumor-infiltrated immune cells (TIC) in the microenvironment are substantially linked to immunotherapy response and cancer prognosis. However, systematic studies of infiltrated immune cell characterization in uveal melanoma (UM) for prognosis and immune checkpoint blockade therapy are lacking.UM datasets were extracted from open access resources (TCGA and GEO databases). The tumor-infiltrated immune cells in the microenvironment were decoded by using the CIBERSORT algorithm, which was further applied to classify UM patients into subgroups using an unsupervised clustering method. The Boruta algorithm and principal component analysis were used to calculate the TIC scores for UM patients. Kaplan-Meier curves were plotted to prove the prognostic value of TIC scores. Besides, the correlations of the TIC score with clinical features, mutated characteristics, and the immune therapeutic response were subsequently investigated.As a result, we defined three subtypes among 171 UM patients according to the TIC profiles and then calculated the TIC score to characterize the immune patterns for all patients. We discovered that high-TIC score patients with low BAP1 and high EIF1AX mutations have a better prognosis than low-TIC score patients. Activation of immune inflammatory response and increase in immune checkpoint-related genes in high-TIC score patients may account for good prognosis and immunotherapy response. Three melanoma cohorts received immunotherapy, proving that high-TIC score patients have substantial clinical and immune therapeutic improvements. Besides, several potential therapeutic agents were identified in the low-TIC score group.Our study afforded a comprehensive view of infiltrated immune cell characterization to elucidate different immune patterns of UM. We also established a robust TIC-score signature, which may work as a prognostic biomarker and immune therapeutic predictor.