Jeong Hyehyun, Oh Ji-Hye, Ahn Hee-Sung, Ryoo Baek-Yeol, Kim Kyu-Pyo, Jeong Jae Ho, Park Inkeun, Hwang Dae Wook, Lee Jae Hoon, Song Ki Byung, Lee Woohyung, Kim Ki-Hun, Moon Deog-Bog, Song Gi Won, Jung Dong-Hwan, Hong Seung-Mo, Park Chae Won, Baek In-Pyo, Cho You Sook, Kim Kyunggon, Sung Chang Ohk, Yoo Changhoon
Department of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, 05505, Republic of Korea.
Bioinformatics Core Laboratory, Convergence Medicine Research Center, Asan Institute for Life Sciences, Asan Medical Center, Seoul, 05505, Republic of Korea.
J Hepatol. 2025 Aug 11. doi: 10.1016/j.jhep.2025.07.031.
BACKGROUND & AIMS: Cholangiocarcinoma is a heterogeneous disease, and its molecular characteristics and biomarkers are not yet fully understood. We performed comprehensive proteogenomic analyses to investigate the molecular landscape of extrahepatic cholangiocarcinoma (EH-CCA).
Prespecified exploratory analyses were conducted within the STAMP trial - a randomized phase II trial of adjuvant capecitabine or gemcitabine plus cisplatin (GemCis) for patients with resected extrahepatic CCA. Among 101 patients in the intention-to-treat population, 89 were included in the biomarker analysis (45 treated with GemCis, 44 with capecitabine). Surgical specimens were assessed by whole-exome sequencing and proteomics analyses. We performed correlative prognostic and predictive biomarker analyses for disease-free survival (DFS).
Somatic mutations were most frequently detected in TP53 (63%), SMAD4 (20%), and KRAS (18%). Homologous recombination deficiency was present in 10 patients (11%), microsatellite instability in one patient (1%), and actionable alterations, according to ESCAT classification, in 13 patients (15%). PIK3CA and FBXW7 mutations were significantly associated with poor DFS, and 11q13.3 amplification with favorable DFS. Some genomic alterations - including 8q24.21, 3q26.1, and 4p16.3 amplifications, 8p23.1 deletion, and homologous recombination deficiency - showed significant interactions with the adjuvant treatment regimen, favoring GemCis. To integrate the impacts of these potentially predictive multiomic biomarkers, we developed a machine-learning prediction model to calculate the conditional average treatment effect and categorize patients into three groups: highest (favoring GemCis), middle (no predicted difference), and lowest (favoring capecitabine). The model showed significant interaction with adjuvant treatment and was significantly associated with DFS.
We identified several prognostic and predictive biomarkers that may guide adjuvant chemotherapy selection in resected EH-CCA, though further validation is required.
We performed comprehensive proteogenomic analyses within a prospective randomized trial of patients with extrahepatic cholangiocarcinoma. The findings highlighted specific genomic and proteomic alterations that were associated with differential benefits from adjuvant chemotherapy regimens. The integration of multiomics data into a machine-learning model provides a practical framework for stratifying patients and guiding treatment decisions, which may inform future clinical trials and biomarker-driven therapeutic strategies.
NCT03079427.
