Development of a comprehensive cfDNA methylation signature for prognostic, predictive, and diagnostic applications in pancreatic ductal adenocarcinoma (PDA).

Abstract

4150 Background: Blood-based biomarkers are promising for predicting outcomes in pancreatic ductal adenocarcinoma (PDA) and could pave the way for developing multi-omic prognostic tools. We previously identified a 15-gene cell-free DNA (cfDNA) methylation signature targeting treatment response (TRg). To enhance its utility, we incorporated additional genes with established prognostic (Prg) and diagnostic (Dxg) value from the literature, creating a composite panel. This study aimed to develop a comprehensive cfDNA methylation signature with predictive, prognostic, and diagnostic value. Methods: Enzymatic methylation sequencing was performed on plasma samples collected between January 2010 and May 2022 from PDA patients undergoing chemotherapy at The Ohio State University. A 206-gene panel (TRg + Prg + Dxg) was analyzed, with methylation levels correlated to overall survival (OS) using univariate and multivariate (MV) Cox regression models. Backward selection identified significant genes, and MV models adjusted for clinical covariates. Patients were stratified into high-risk (Hg) and low-risk (Lg) groups based on median risk scores. Results: Our study cohort had 51 PDA patients, with a median age of 65 (range: 34 -80), 53% females, and 86% Caucasian (12% African-American and 2% others). Stage at diagnosis (Dx-S) distribution: 15 locally advanced (LA), 14 metastatic (Mets), and 22 resectable/borderline resectable (R/BR). 12 patients in BR/R received neoadjuvant therapy (NT) while the rest had upfront surgery (UpS) followed by adjuvant therapy (AT). Ultimately, 21 had resection, and 30 got palliative therapy. First-line chemotherapy (FL) patients received is, 36 FOLFIRINOX (12 PT, 9 NT, and 5 AT), 19 gemcitabine (Gem)/nab-paclitaxel (NP) (16 PT, 1 NT, and 2 AT), and 6 Others. A 25-gene cfDNA methylation signature (15 TRg, 10 Prg, and 1 Dxg) stratified patients into Hg and Lg groups. OS was significantly longer in Lg compared to Hg across all models. Conclusions: The study successfully developed a comprehensive 25-gene cfDNA methylation signature combining predictive, prognostic, and diagnostic markers for PDA. This signature effectively stratifies patients into Hg and Lg, demonstrating significant differences in OS across various clinical models. The results highlight the potential utility of cfDNA methylation as a multi-omic tool to enhance personalized treatment strategies and improve patient outcomes in PDA. Further validation in larger cohorts is warranted to confirm its clinical applicability. Models tested OS*Hg vs. Lg Hazard ratio Signature-alone 7.58 vs. 33 13.5 Plus, FL 8.98 vs. 33 13.2 Plus, FL, Surgery (NAT vs. UpS vs. PT) 7.58 vs. 33 17.3 Plus, Dx-S 7.58 vs. 33 14.4 *In months.