Abstract 5880: Leveraging cell-free DNA methylation biomarkers for pancreatic cancer detection and minimal residual disease monitoring

Abstract Introduction: Detecting pancreatic cancer from plasma cell-free DNA (cfDNA) is particularly challenging due to the low circulating tumor DNA (ctDNA) shedding associated with pancreatic tumors, especially in localized cases. Pancreatic cancer often exhibits differential methylation across hundreds or thousands of genes. We applied fragment-level DNA methylation analysis to tissue and plasma samples from a pancreatic ductal adenocarcinoma (PDAC) cohort to obtain differentially methylated fragment regions (DMFRs) alongside traditional differentially methylated regions (DMRs). Mutation-based tumor fractions using orthogonal assay were calculated for comparison. Additionally, we developed a PDAC DNA methylation subpanel using patient data based on DMFRs obtained from fragment-level analysis. Methods: Pancreatic cancer samples from localized (n = 36) and metastatic (n = 117) stages were profiled by methylation-based PredicineEPIC™, PredicineALERT™, and mutation-based PredicineBEACON™ assays. Abnormally methylated fragments were identified using a background model based on healthy plasma donors. A PDAC panel was created using both tissue (n = 45) and plasma (n = 117) samples, as well as public datasets such as TCGA. Results: Fragment-level analysis identified a comparable number of DMFRs in PDAC plasma and tissue samples, with strong overlap between the two. While DMFRs and DMRs were comparable in the tissue samples, significant fewer DMRs were identified in the plasma samples. This underscores the superior sensitivity of fragment-level analysis for detecting abnormal methylation in the liquid biopsy samples. Furthermore, key PDAC-related biomarkers were more distinctively identified using DMFR analysis. Genome-wide DNA methylation losses were also observed in many pancreatic samples, and combining these with location-specific methylation signals enhanced both the robustness and sensitivity of the assay. The incorporation of the PredicineALERT PDAC panel, informed by the methylation fragment-level analysis, improved concordance between methylation and mutation-based tumor fractions in both metastatic and localized PDAC plasma samples. Sensitivity for minimal residual disease (MRD) detection was also markedly improved using PredicineALERT panel compared to genome-wide methylation profiling using PredicineEPIC. Conclusion: This study demonstrates the value of fragment-level cfDNA methylation analysis for detecting and monitoring PDAC using plasma samples. DMFR analysis demonstrated greater sensitivity than traditional DMR analysis. In addition, a PDAC panel designed through fragment-level analysis enhanced correlations between methylation abnormalities and mutation-based tumor fractions, offering a robust tool for early detection and disease monitoring in PDAC. Citation Format: Ziqi Zhu, Giancarlo Bonora, Chao Dai, Kemin Zhou, Binggang Xiang, Shidong Jia, Pan Du. Leveraging cell-free DNA methylation biomarkers for pancreatic cancer detection and minimal residual disease monitoring [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5880.