Abstract B045: Plasma cfDNA 5-hydroxymethylcytosine signatures associate with treatment response and molecular subtype in BRCA -mutant pancreatic cancer
Ceyda Coruh, Maria Raitses-Gurevich, Yuhong Ning, Alessandra Cunsolo, Anna Leighton, Hannah Ghaffari, Verena Freidl, Wayne Volkmuth, Shimul Chowdhury, Samuel Levy, Gulfem D. Guler, Talia Golan, Chani Stossel, Dikla AtiasAbstract
Background:
Biomarkers that predict and monitor response to platinum-based chemotherapy and PARP-inhibitor (PARPi) therapy remain an unmet need in BRCA-positive pancreatic ductal adenocarcinoma (PDAC). Furthermore, blood-based assays to classify tumor expression subtypes are limited. Cell-free DNA (cfDNA) 5-hydroxymethylcytosine (5hmC) profiling is a non-invasive epigenomic approach that captures tissue- and cancer-associated signals from plasma. We evaluated whether the Avantect Pancreatic Cancer Test scores associate with clinical response and resistance in BRCA-positive PDAC patients undergoing platinum or PARPi treatment. We also investigated the correlation between the tumor-based expression subtype and plasma cfDNA 5hmC signals using Moffitt signature genes to determine if cfDNA 5hmC serves as a liquid biopsy surrogate for tumor tissue transcriptome-based subtype classification.
Methods:
Plasma samples were collected in EDTA tubes from PDAC patients, a subset of whom harbored BRCA mutations and received platinum or PARPi therapy, obtained at baseline and multiple on-treatment time points (n=69). Patients were classified into three clinically-defined response groups: long-term responders (LTR, n=8), patients with acquired resistance (AR, n=24), and refractory (REF, n=9) patients. Genome-wide cfDNA 5hmC patterns were used to derive Avantect pancreatic cancer scores for comparison across response groups and over time. In parallel, cfDNA 5hmC signals over the Moffitt classical and basal-like genes were used to assign a subtype to each plasma sample, which were compared against subtype calls derived from patient-matched patient-derived xenograft (PDX) expression profiling.
Results:
Baseline plasma 5hmC-based Avantect cancer scores differed by response group, with the lowest scores in LTR patients and progressively higher scores in AR and REF patients. Most LTR patients showed declining cancer scores on treatment, whereas AR patients generally exhibited increasing scores over time. REF patients had the highest baseline cancer scores and little longitudinal change, consistent with persistent disease burden and lack of response. Furthermore, PDAC molecular subtypes defined by PDX-derived expression profiling were largely recapitulated using cfDNA 5hmC signals from Moffitt classical and basal-like gene signatures, supporting the feasibility of subtype classification directly from plasma.
Conclusions:
Plasma cfDNA 5hmC profiling yields a cancer score that dynamically associates with treatment response, acquired resistance, and refractory disease in BRCA-positive PDAC patients receiving platinum or PARPi therapy. cfDNA 5hmC signals utilizing Moffitt subtype genes were largely concordant with PDX-derived expression subtypes in the majority of cases, indicating that plasma cfDNA can capture both cancer burden and molecular subtype information. These findings support further evaluation of plasma 5hmC signatures as a non-invasive biomarker for treatment monitoring, molecular subtyping, and early detection of resistance in this patient population.
Citation Format:
Ceyda Coruh, Maria Raitses-Gurevich, Yuhong Ning, Alessandra Cunsolo, Anna Leighton, Hannah Ghaffari, Verena Freidl, Wayne Volkmuth, Shimul Chowdhury, Samuel Levy, Gulfem D. Guler, Talia Golan, Chani Stossel, Dikla Atias. Plasma cfDNA 5-hydroxymethylcytosine signatures associate with treatment response and molecular subtype in BRCA-mutant pancreatic cancer [abstract]. In: Proceedings of the AACR Conference on Pancreatic Cancer: New Frontiers in Biology and Therapeutic Development; 2026 Sep 25-28; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(18_Suppl_2):Abstract nr B045.