Biliary tract cancer (BTC) is a highly aggressive malignancy often diagnosed at advanced stages, leaving patients with limited treatment options and poor survival outcomes. Current diagnostic tools, including imaging and invasive procedures, are insufficient for detecting early-stage BTC with high sensitivity and specificity. To address this unmet need, a new study highlights the potential of a circulating cell-free DNA (cfDNA) fragmentomics and machine learning method to enable noninvasive, early detection of BTC, which could significantly improve patient outcomes.
The study included 163 patients with BTC and 165 healthy individuals who were divided equally into training and validation cohorts. Plasma samples were collected from all participants for analysis using low-depth whole-genome sequencing. The researchers extracted 3 key cfDNA fragmentomics features—copy number variation, fragment size distribution, and promoter fragmentation entropy—to develop a machine learning model for BTC detection. This model was further validated using an external cohort of 55 patients with benign biliary diseases and 18 cases of Tis (carcinoma in situ) or high-grade cases.
Results from the validation cohort demonstrated excellent performance of the cfDNA-based model, achieving an area under the curve (AUC) of 0.96, suggesting that this model is highly effective at distinguishing between individuals with BTC from those who are healthy. At an 86% training specificity cutoff, the model showed a sensitivity of 90.9% and a specificity of 87.9%, highlighting its potential for noninvasive early detection of BTC. Promoter fragmentation entropy emerged as a strong single feature, with an AUC exceeding 0.92. The model’s sensitivity was especially notable for early-stage detection, increasing from 80% in stage I BTC to 95.7% in stage II, demonstrating its potential for identifying BTC at curable stages. In the external validation cohort, the model maintained its high performance, with a sensitivity of 89% for early lesions and a specificity of 89% for benign cases, further highlighting the potential of the methods for detection of early BTC.
In conclusion, the novel cfDNA-based approach for BTC detection shows promise in clinical application. By enabling early and accurate diagnosis, this technology has the potential to transform BTC screening and management, ultimately improving survival outcomes for patients. Further studies are needed to validate these findings in larger, more diverse populations and to explore its integration into routine clinical workflows for BTC detection.
Wang J, Ni X, Zheng Y, et al. Integration of cfDNA fragmentomics for early biliary tract cancer detection. Presented at: 2025 ASCO Annual Meeting. May 30-June 3, 2025; Chicago, IL. Poster 421.
To sign up for our newsletter or print publications, please enter your contact information below.
