Discrimination of pancreato-biliary cancer and pancreatitis patients by non-invasive liquid biopsy

Abstract

Background: Current diagnostics for the detection of pancreato-biliary cancers (PBCs) need to be optimized. We therefore propose that methylated cell-free DNA (cfDNA) derived from non-invasive liquid biopsies serves as a novel biomarker with the ability to discriminate pancreato-biliary cancers from non-cancer pancreatitis patients.

Methods: Differentially methylated regions (DMRs) from plasma cfDNA between PBCs, pancreatitis and clinical control samples conditions were identified by next-generation sequencing after enrichment using methyl-binding domains and database searches to generate a discriminatory panel for a hybridization and capture assay with subsequent targeted high throughput sequencing.

Results: The hybridization and capture panel, covering around 74 kb in total, was applied to sequence a cohort of 25 PBCs, 25 pancreatitis patients, 25 clinical controls, and seven cases of Intraductal Papillary Mucinous Neoplasia (IPMN). An unbiased machine learning approach identified the 50 most discriminatory methylation markers for the discrimination of PBC from pancreatitis and controls resulting in an AUROC of 0.85 and 0.88 for a training (n = 45) and a validation (n = 37) data set, respectively. The panel was also able to distinguish high grade from low grade IPMN samples.

Conclusions: We present a proof of concept for a methylation biomarker panel with better performance and improved discriminatory power than the current clinical marker CA19-9 for the discrimination of pancreato-biliary cancers from non-cancerous pancreatitis patients and clinical controls. This workflow might be used in future diagnostics for the detection of precancerous lesions, e.g. the identification of high grade IPMNs vs. low grade IPMNs.

Keywords: DMRs; Hybridization and capture; IPMN; Methylation; Next-generation sequencing; Non-invasive diagnostics; Pancreatitis; Pancreato-biliary cancer; cfDNA; cfMBD-Seq.

Fig. 1

General workflow. A: DNA was isolated of liquid and solid biopsies derived from different patient cohorts. Sequencing libraries were prepared that underwent methyl-binding domain (MBD) enrichment. The enriched fragments were sequenced by means of NGS to generate MBD-Seq data. B: CfMBD-Seq data together with already published regions from literature and public tissue data were used to identify differentially methylated regions (DMRs) that served for the design of a targeted panel. The panel was used for hybridization and capture with subsequent sequencing to enable high-throughput identification of different patient subgroups

Fig. 2

Machine learning approach M1. 50 most promising DMCs (hybridization and capture approach) combined with CA19-9 values for distinguishing PBC from pancreatitis and controls. A: PCA based on the 50 most informative DMCs combined with CA19-9 values for the conditions control (blue), pancreatitis (black), and PBC (red). Variances explained: PC1 = 56.75%, PC2 = 9.05%. B: ROC curve (AUC = 0.85) of PBC predicition scores for the identification cohort C2. The red dot indicates the determined optimal threshold value for the PBC prediction score that maximizes sensitivity and specificity with a defined minimum sensitivity of 90%. C: Boxplot of PBC prediction scores from the identification cohort C2 with the optimized classification threshold of 0.15 (gray line). D: ROC curve (AUC = 0.88) of PBC prediction scores for the validation cohort C3 including IMPNs. The red dot indicates the threshold value for classifying PBCs and high grade IPMNs with a minimum sensitivity of 90%. E: Boxplot of the PBC prediction scores from the validation cohort C3 including low and high grade IPMNs and the pre-determined PBC classification threshold of 0.15 (gray line). F: Kaplan-Meier curve for the survival of PBC patients from the validation cohort C3. Follow-up of 44 months after diagnosis. Separation of PBC group (n = 10) into two subgroups by the pre-determined PBC classification threshold of 0.15