The role of bile acid subtypes in the diagnosis of cholangiocarcinoma

Objectives To find the potential biomarkers of cholangiocarcinoma, form a biomarker package, evaluate its efficiency, and validate it. Methods R software was used to analyze the differential expression of mRNAs between cholangiocarcinoma and adjacent nontumorous tissues, obtained from The Cancer Genome Atlas (TCGA), and enrich the KEGG pathway. Metabo-Profile Inc. performed the comprehensive bile acid profiling and quantitation. The training set concluded 20 cholangiocarcinoma and 20 nontumorous volunteers. Receiver operating characteristic (ROC) curve and accompanying area under the curve (AUC) was calculated. The top four bile acids formed a new biomarker package. The validation set included 15 cholangiocarcinoma and 15 nontumorous, and the sensitivity and specificity of the new biomarker package were tested. Results Gene expression of 36 cholangiocarcinoma and nine adjacent nontumorous tissues was obtained in January 2020. Totally 9887 differential genes were eligible (logFC >= 1 or <= -1, P < 0.05, and adjust P < 0.01). GO analysis showed that 20 KEGG pathways were enriched, including primary bile acid biosynthesis and bile secretion. Comprehensive bile acid profiling and quantitation showed 15 differential bile acid types, and the ROC-AUC was between 0.953 and 0.750. HDCA, isoLCA, bCDCA, and DCA were selected to form a biomarker package. The Logit (p = cholangiocarcinoma) = 7.898 - 3.70*(1isoLCA) - 0.444*(bCDCA) + 0.415*(HDCA) + 0.041*DCA. Its ROC-AUC was 0.944. In the validation set, the sensitivity was 0.933 and the specificity was 0.867. Conclusion Bile acid types package was efficient to distinguish nontumorous population and cholangiocarcinoma. The difference might be associated to the downregulation of primary bile acid biosynthesis and bile secretion pathway of cholangiocarcinoma.