A Telomere-Maintenance Three-Gene Model for the Prognostic Predictionin Intrahepatic Cholangiocarcinoma and Multi-Cohort Validation

INTRODUCTION: Intrahepatic cholangiocarcinoma (iCCA) is characterized by heterogeneity and poor survival. It remains unclear how telomere maintenance programs influence the prognosis and immune microenvironment in iCCA, and clinically applicable, telomere‑anchored transcriptomic tools are currently lacking. METHODS: Transcriptomes and survival data from TCGA-CHOL, GSE107943, and E-MTAB-6389 cohorts were integrated (n=135). Telomere maintenance gene (TMG) scores were calculated using GSVA and subsequently used for WGCNA module identification. Functional enrichment was analyzed with clusterProfiler (FDR-adjusted P < 0.05). A prognostic model was constructed using univariate Cox and LASSO-Cox regression analyses, and validated by Kaplan-Meier (KM) curve and time-dependent ROC. Immune features were inferred by MCPcounter, CIBERSORT, and ssGSEA algorithms. Drug sensitivity was predicted by pRRophetic, and the resulting IC50 estimates were then correlated with the RiskScore. RESULTS: The TMG score was inversely correlated with the ESTIMATE score (Spearman ρ=-0.2439, P=0.0044). WGCNA identified a TMG-associated module enriched for cellcycle/ mitotic and microtubule functions. Three key TMGs (PTTG1, TSPYL5, PLLP) were integrated to develop a RiskScore that can consistently stratify overall survival (OS) in both the integrated set and all external cohorts, demonstrating a robust accuracy for 1-, 3-, and 5-year prognostic prediction. High-RiskScore tumors exhibited reduced immune infiltration across multiple deconvolution frameworks. The RiskScore was negatively correlated with predicted IC50 for several agents (e.g., pyrimethamine, GNF-2, NSC-87877, CGP-082996, KIN001-135), suggesting higher drug sensitivity in high-risk cases. DISCUSSION: We identified a three-gene, telomere-related risk score for iCCA, which can effectively predict patients' survival and distinguish immune-cold, high-proliferation phenotypes. Potential targeted drugs for high-risk patients were predicted and supported by mechanistic validation. CONCLUSION: A compact telomere-anchored three-gene RiskScore was developed to predict the prognosis, immune contexture, and therapy sensitivity for iCCA.