Longitudinal analysis of circulating tumor DNA and CA19-9 dynamics in predicting disease relapse and monitoring treatment response in stage I-III pancreatic ductal adenocarcinoma: An interim analysis of a prospective observational study.

INTRODUCTION: Postoperative recurrence is the leading cause of mortality in resected pancreatic ductal adenocarcinoma (PDAC), yet reliable tools for early relapse detection and treatment response assessment remain lacking. METHODS: In a prospective cohort of 136 patients with resected stage I-III PDAC receiving adjuvant chemotherapy, we evaluated circulating tumor DNA (ctDNA) and CA19-9 as longitudinal biomarkers across multiple postoperative time windows. RESULTS: ctDNA consistently outperformed CA19-9 as an independent prognostic factor; ctDNA positivity at on-treatment and surveillance assessments achieved a positive predictive value of 91.7%, while persistent negativity identified the lowest-risk patients. Integrating CA19-9 with ctDNA resolved the ctDNA-alone gap in distinguishing treatment clearance from conversion, improving discrimination of responders from non-responders (HR 3.72; P = .008). A time-weighted dynamic ctDNA risk score (MinerVa-dynamic) further stratified ctDNA-negative patients into clinically distinct prognostic subgroups, achieving an area under the curve of 0.87 and 0.82 for one- and two-year disease-free survival prediction, respectively. CONCLUSIONS: These findings support a dual-biomarker longitudinal monitoring framework as a practical, individualized approach to postoperative surveillance and early therapeutic decision-making in PDAC.