Bile-Based Cell-Free DNA Analysis Is a Reliable Diagnostic Tool in Pancreatobiliary Cancer

Abstract

Currently available serum biomarkers for pancreatobiliary cancers lack sensitivity and specificity and ultimate diagnosis still requires invasive procedures for histological confirmation. The detection of tumor-specific genetic aberrations with utilization of cell free DNA (cfDNA) is a less invasive approach than traditional tissue biopsies; however, it has not been implemented into clinical routine. In this study, we investigated bile as a liquid biopsy source in pancreatobiliary cancers and compared its potential as cell-free DNA source to plasma. Blood (n = 37) and bile (n = 21) samples were collected from patients affected by pancreatic ductal adenocarcinoma (PDAC) and extrahepatic cholangiocarcinoma (CCA) or with non-malignant biliary obstructions (blood n = 16; bile n = 21). Panel-based next generation sequencing (NGS) and digital droplet PCR (ddPCR) were applied for tumor mutation profiling. NGS results from matched tumor tissues (n = 29) served as comparison. Sequencing of cfDNA from bile resulted in detection of 96.2% of the pathogenic tumor mutations found in matched tissue samples. On the other hand, only 31.6% of pathogenic tumor mutations found in tissue could be detected in plasma. In a direct comparison, only half of the mutations detected in bile cfDNA were concordantly detected in plasma from the same patients. Panel NGS and ddPCR displayed comparable sensitivity. In conclusion, bile is a suitable source of cfDNA for the diagnosis of pancreatobiliary cancer and performs more reliably than plasma. Although primary diagnosis still requires histologic confirmation, bile-derived cfDNA could offer an alternative if tissue sampling is not feasible and might allow less invasive disease monitoring.

Keywords: ERCP; cell-free DNA; cholangiocarcinoma; liquid biopsy; next generation sequencing; pancreatic cancer.

Figure 1

Next generation sequencing (NGS) analysis in bile samples: (a) Oncoprint showing mutation occurrence of six most frequently mutated genes across all patients. Patients are divided into those with localized and those with metastatic tumors. Concordant mutations are displayed in green, while mutations only detected in tissue or bile are displayed in blue or yellow, respectively. (b) Distribution of mutations within the bile samples. (c) KRAS allele frequency in bile samples. Patients with localized cholangiocarcinoma (CCA) were excluded, as no KRAS mutation was detected in the bile samples from these patients.

Figure 2

NGS analysis in plasma samples: (a) Oncoprint showing mutation occurrence of nine most frequently mutated genes across all patients. Patients are divided into those with localized and those with metastatic tumors. Concordant mutations are displayed in green, while mutations only detected in tissue or plasma are displayed in blue or red, respectively. (b) Distribution of mutations across the plasma samples. (c) KRAS allele frequency in plasma samples. Samples from CCA patients were excluded, as no KRAS mutation was detected in the plasma samples from these patients. *: p < 0.05.

Figure 3

Comparison of NGS results between bile and plasma. (a) Oncoprint showing comparison of mutation occurrence in five top genes across all patients in blood and bile cfDNA samples. Patients are divided into controls, patients with localized and patients with metastatic tumors. KRAS allele frequency in bile and plasma cfDNA in patients with localized (b) and metastatic disease (c). (d) KRAS and TP53 allele frequencies of patient 44 in plasma and bile. (e) CA19-9 serum levels of patients within our cohort. *: p < 0.05; **: p < 0.005.

Figure 4

Comparison of KRAS allele frequency measured by NGS vs. ddPCR. KRAS allele frequency displayed in tumor samples measured by ddPCR (green) and NGS (red) in plasma (a) and bile samples (b). Some patients had to be excluded from analysis because of insufficient quantity of DNA or because of KRAS mutations in codon 61.