Multiomic Profiling and Machine Learning Reveal Distinct Molecular Subtypes of Cholangiocarcinoma and Identify TNK1 as a Novel Therapeutic Target

Highlight

This study leverages multiomic datasets and machine learning to identify three distinct molecular subtypes of cholangiocarcinoma (CCA) independent of the traditional anatomical classification. Among these, the metabolic cluster exhibits elevated TNK1 kinase activity, which was validated as a therapeutic target with the selective inhibitor TP-5801 in patient-derived xenograft (PDX) models. Furthermore, the immunomodulatory cluster emerged as most responsive to standard gemcitabine/cisplatin chemotherapy, offering potential for subtype-guided treatment stratification.

Study Background