Genomic Alterations and Tumor Immune Microenvironment as Predictors of Immunotherapy Outcomes in Advanced Biliary Tract Cancer

Background: The response to immunotherapy is limited in advanced biliary tract cancer (BTC). Response prediction is a serious challenge clinically.

Methods: This study included 55 patients with advanced BTC, who received anti-PD-1 treatment. 520 gene panel sequencing was performed in 30 patients, while multiplex circulating cytokine testing was done in 50 patients. Entropy and mutation features were analyzed using the previously optimized pipeline. The repeated LASSO algorithm was used to identify the optimal features. The associations between sequence features and cell communication were explored by analyzing publicly available single-cell transcriptome data of BTC (GSE125449). Cox regression was used to develop an integrated model. Prediction performance was assessed by time-dependent C-index, Kaplan-Meier, and receiver operating characteristic curve determination.

Results: TP53, NRAS, FBXW7 and APC were identified as prognosis related genes. The average C-index of sequence entropy and mutation for overall survival were 0.819 and 0.817, respectively. They were significantly higher than 0.392 and 0.638 in tumor mutation burden and mutation score. Single-cell transcriptome data inferred that TP53, KRAS, and NRAS are enriched in plasmacytoid dendritic cells (pDCs), and that the communication between pDCs and macrophages is mediated through the CXCL signaling pathway. An EM-CXCL10 integrated model shows powerful predictive performance, including on survival status (AUC: 0.863) and objective response rate (AUC: 0.990), in patients receiving immunotherapy.

Conclusion: Through a multi-dimensional strategy, this study presents a promising predictive model for selecting BTC patients prone to immunotherapy. A further clinical study with a large patients cohort is required to validate its predictive performance.

Publication History

Article published online:
20 January 2025