743 Background: Reliable predictors of early progression on palliative chemotherapy (PCT) in pancreatic ductal adenocarcinoma (PDA) could enable more personalized treatment strategies. We previously reported a 15-gene cfDNA methylation signature predictive of overall survival, but it was limited in guiding frontline treatment decisions. Here, we describe a predictive epigenetic signature (PSig) designed to identify patients at high risk (Hg) for failing first-line chemotherapy (FLC). Methods: Targeted enzymatic methylation sequencing was performed on plasma samples from PDA patients receiving chemotherapy at Ohio State University's biorepository (January 2010 to May 2022). Gene methylation levels were assessed for associations with time to progression (TTP, the interval from the first chemotherapy dose to first progression) and total duration of response (tDOR, from first dose to death or last follow-up), using multivariate Cox regression models. Covariates included FLC (FOLFIRINOX vs. gemcitabine/nab-paclitaxel G/NP vs. others) and stage at diagnosis (StD: early-stage ES, locally advanced LA, or metastatic Met). Multivariate models generated risk scores, stratifying patients into Hg and low-risk group (Lg). Kaplan-Meier and log-rank tests were used to evaluate survival differences. Results: Among the study cohort of 51 PDA patients (21 ES, 13 LA, 14 Met), 30 had palliative therapy (3 ES cases progressed on neoadjuvant therapy) – median age was 66 years, 53% male, and 87% Caucasian. FLC distribution: 12 FFX, 16 G/NP, 2 others. PSig included 20 genes for TTP and 16 for tDOR, linked to response to G, NP, irinotecan, and platinum. The performance of Psig models is summarized below. FLC and StD did improve Psig for TTP, while StD strengthened prediction for tDOR. Conclusions: PSig robustly stratifies PDA patients receiving PCT into distinct prognostic groups, predicting both rapid progression and short treatment benefit. Integration with StD enhances prediction for tDOR, underscoring the value of combining molecular and clinical variables. Larger, prospective studies are needed to validate PSig for clinical application. Models tested Hg vs. Lg TTP (in months) HR p-value Hg vs. Lg tDOR (in months) HR p-value Psig alone 2.7 vs. 9.13 15 <0.001 5.07 vs. 15.57 11.7 <0.001 Psig + FLC 2.7 vs. 9.13 42.6 5.07 vs. 15.57 14.9 Psig + StD 2.23 vs. 9.13 49.4 4.57 vs 15.57* 45.5 HR – hazard ratio, *StD was significant.
Manne et al. (Sat,) studied this question.