A genome-wide DNA methylation-based classifier differentiates intrahepatic cholangiocarcinomas from hepatic metastases of pancreatic ductal adenocarcinoma

Differentiating intrahepatic cholangiocarcinomas (iCCA) from hepatic metastases of pancreatic ductal adenocarcinoma (PAAD) is difficult. We hypothesized that DNA-methylation based machine-learning algorithms may be able to perform this task. We assembled genome-wide DNA-methylation data for iCCA, PAAD, and normal bile from publicly available sources. We used an internal cohort consisting of 72 samples to test our classifier. Internal testing of a classifer using 72 FFPE samples after Infinium HD FFPE DNA Restore Kit and analyed using Illumina Infinium MethylationEPIC BeadChip.