Detection of pancreatic cancer using a novel blood-based DNA methylation signature.

Abstract

e15546 Background: Pancreatic cancer (PC) is a high mortality malignancy typically found when curative surgery is not an option. Liquid biopsies are a minimally invasive option that may allow earlier detection of PC. Using a bioinformatic analysis of TCGA and GEO we identified a novel set of blood-based biomarkers to detect PC with high sensitivity and specificity (AUROC = 0.999) (Vrba et al, Epigenetics 2019). Here we tested the ability of this DNA methylation signature (MS) to distinguish metastatic PC from pancreatic cysts (Cy) and healthy controls (HC) through quantitative DNA methylation analysis of cell free DNA (cfDNA). Methods: A 10 gene DNA methylation marker set that identifies PC was used to evaluate the cfDNA component of the plasma of PC patients, Cy patients and HC. The ROC analysis and AUC calculations were performed using the R library pROC (Robin et al, BMC Bioinformatics 2011). Plasma samples were collected in cfDNA collection tubes from patients with benign Cy, PC, and HC. Clinical information was collected including age, gender, smoking history, CA 19-9 level, cyst size, presence of liver metastases, and survival status. After processing and cfDNA extraction, qRT-PCR methylation analysis was performed on these prospectively collected samples. The difference in median DNA MS per marker of the full 10-gene marker set between the above groups was analyzed by Wilcoxon rank sum test and ROC analysis. A Cox proportional hazard regression model was used to test the association between MS and survival. Results: A total of 66 samples were collected (PC = 10, Cy = 12, HC = 44). Clinical information was collected on all patients included in the analysis. Using liquid biopsies for cfDNA analysis and a novel set of DNA methylation biomarkers, MS was substantially stronger in PC (median, 13.2) patients than Cy (median, 3.5) patients (p = 0.0001). Importantly, while there was no clear association between MS and age, gender, smoking status, CA 19-9 levels, or presence of liver metastases, higher methylation levels were non-significantly associated with worse overall survival in PC patients (HR, 1.29; 95% CI, 0.97–1.72; p = 0.082). Conclusions: While this is a small sample size, early findings suggest that a DNA MS may reliably distinguish PC from benign Cy. The strength of signal may be a prognostic marker that could play a role in determining treatment approaches. Given its utility in predicting PC, future directions include testing this tool as a blood-based early detection mechanism as well as a predictor for recurrence or response to PC treatment.