PD-6 DNA methylome as a potential biomarker in biliary brushes and bile fluid samples to differentiate between benign and malignant biliary stenosis

The diagnosis of perihilar and distal cholangiocarcinoma (CCA) remains an important clinical challenge. Current detection methods are mostly based on biliary brushing samples with suboptimal sensitivity and/or specificity, leading to late higher mortality rates. We aimed identify a diagnostic biomarker using targeted DNA methylation sequencing in classical training validation study-design. Biliary bile fluid from patients known malignant versus benign stenoses were prospectively collected during endoscopic retrograde cholangiopancreatography (ERCP). Clinical data including baseline patient characteristics follow-up was recorded. All subjected targeted, enzymatic (EM-seq) total 608,293 capture probes targeting genomic regions be hypermethylated cancer. Differential analysis used differentially methylated between samples. Only both brush retained. These train ‘random forest classification’ prediction model ‘training cohort’ samples, 10-fold cross-validation. remaining then as ‘validation set’ test the potential score classifying Receiver operating characteristic curve analyses evaluate performance brush-derived fluid-derived scores differentiating A 43 included November 2019 September 2021. Twenty-eight had stenosis, majority which due ischemic cholangiopathy, while 15 stenosis CCA or pancreatic adenocarcinoma. Average coverage significantly (59.6X 24.9X, respectively; p < 0.001). identified 669 sample types. Brush-derived differentiated specificity 0.913 0.933 (AUC 0.93). Similarly, derived demonstrated 0.961 0.8 0.89). present DNA-methylation that accurately differentiates stenosis. Bile aspiration ERCP is alternative when not feasible. Further larger cohorts warranted.