Integrated Transcriptomic Analysis Identifies Immune Remodeling and Prognostic Signatures in Uveal Melanoma

Abstract

Background

Uveal melanoma (UVM) is the most common primary intraocular malignancy in adults and exhibits a high propensity for liver metastasis, often leading to poor prognosis. However, effective prognostic biomarkers and therapeutic strategies for metastatic UVM remain limited.

Methods

We comprehensively analyzed transcriptomic data from both single-cell and bulk RNA sequencing cohorts, integrating data from TCGA and GEO (GSE139829, GSE22138, and GSE84976). After batch effect correction and cell type annotation, differentially expressed genes (DEGs) between primary and metastatic malignant cells were identified. These were intersected with 900 prognosis-related genes from TCGA, and 11 key prognostic genes were selected via least absolute shrinkage and selection operator (LASSO) regression to construct a risk prediction model. Model performance was evaluated across multiple cohorts. Furthermore, immune infiltration was assessed using CIBERSORT, and drug sensitivity was predicted based on chemotherapeutic IC50 values.

Results

The 11-gene risk model effectively stratified UVM patients into high-risk and low-risk groups with distinct survival outcomes. High-risk patients exhibited a more immunosuppressive tumor microenvironment and were associated with altered sensitivity to multiple chemotherapeutic agents. Immune checkpoint gene expression also varied significantly between risk groups, indicating potential implications for immunotherapy response.

Conclusion

This study identifies critical molecular features underlying UVM metastasis and immune remodeling, providing novel prognostic markers and potential therapeutic targets for clinical management of UVM.