Abstract
Uveal melanoma (UVM) prognosis and the possibilities for targeted therapy depend on a thorough understanding of immune infiltration features and the analysis of genomic and immune signatures. Leveraging multi-omics data from The Cancer Genome Atlas and GEO datasets, we employed an unsupervised clustering algorithm to categorize UVM into immune-related subgroups. Subsequent multi-omics analysis revealed two distinct UVM subtypes, each characterized by unique genomic mutations and immune microenvironment disparities. The aggressive UMCS2 subtype exhibited higher TNM stage and poorer survival, marked by elevated metabolism and increased immune infiltration. However, UMCS2 displayed heightened tumor mutational burden and immune dysfunction, leading to reduced responsiveness to immunotherapy. Importantly, these subtypes demonstrated differential sensitivity to targeted drugs due to significant variances in metabolic and immune environments, with UMCS2 displaying lower sensitivity. We developed a robust, subtype-specific marker-based risk scoring system. This system’s diagnostic accuracy was validated through ROC curves, decision curve analysis, and calibration curves, all yielding satisfactory results. Additionally, cell experiments identified the pivotal function of HTR2B, the most crucial factor in this risk model. Knocking down HTR2B significantly reduced the activity, proliferation, and invasion ability of the UVM cell line. These findings underscored the impact of gene and immune microenvironment alterations in driving distinct molecular subtypes, emphasizing the need for precise treatment strategies. The molecular subtyping-based risk assessment system not only aids in predicting patient prognosis but also guides the identification of populations suitable for combined treatment. Molecules represented by HTR2B in the model may serve as effective therapeutic targets for UVM.
Communicated by Ramaswamy H. Sarma
Acknowledgments
We express our sincere gratitude to the patients and researchers who participated in this project and generously provided study data. We would also like to thank Dr. Jianming Zeng from the University of Macau and his bioinformatics team, as well as Biotrainee and Sangerbox, for their valuable contributions, support, and sharing of their experience and codes. Their contributions have greatly enhanced the quality and success of this research.
Author contributions
Yuan Zhang, Ni Shen, and Aimin Jiang contributed equally to this study. Yu Gao and Wei Shen conceptualized and guided the study. Yuan Zhang, Ni Shen and Jiawei Zhao wrote the paper and revised it. Aimin Jiang performed the statistical analysis and visualized the data. Yanzhi Sang and Anbang Wang analyzed and interpreted the data. All authors contributed to the article and confirmed the final version of the manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The data and associated code included in this study are available from public databases and the corresponding author on reasonable request.