Machine learning based regulatory T cells related gene signature for predicting the clinical outcome and immunotherapy benefits in skin cutaneous melanoma

Skin cutaneous melanoma (SKCM) is an aggressive malignancy with variable prognosis and response to immunotherapy. Regulatory T cells (Tregs) play a crucial role in immune suppression within the tumor microenvironment. However, the clinical significance of Treg-related genes in SKCM remains unclear. We constructed a Treg-related gene signature for SKCM using bulk RNA-seq data from TCGA and three GEO cohorts. Machine learning algorithms were employed to identify prognostic genes associated with Tregs, and a Lasso-Cox model was used to derive a Treg-score. The prognostic value of the Treg-score was validated across multiple cohorts. Associations between Treg-score, immune infiltration, immune escape, and response to immunotherapy were assessed in several independent datasets. Drug sensitivity was also predicted. In vitro, MCAM, a key Treg-related gene, was experimentally validated for its role in SKCM cell proliferation and metastasis. The Treg-score effectively stratified patients into high- and low-risk groups with distinct survival outcomes. High Treg-scores correlated with an immune-cold tumor microenvironment, characterized by lower immune cell infiltration and reduced response to immunotherapy. Conversely, low Treg-scores indicated an immune-hot phenotype and higher likelihood of responding to immunotherapy. Drug sensitivity analysis suggested high-risk patients may benefit from chemotherapy and targeted therapies. MCAM was confirmed as a functional driver of SKCM progression. We developed a Treg-related gene signature that predicts prognosis and is associated with immunotherapy benefit in SKCM. The Treg-score provides a valuable tool for risk stratification and personalized treatment decision-making in melanoma.