Patient-specific tumor-informed circulating tumor DNA (ctDNA) assay to predict cancer recurrence in patients with resected pancreatic cancer.

744 Background: Pancreatic cancer (PC) is one of the leading causes of cancer death. Identifying molecular residual disease (MRD) with tailored tumor-informed ctDNA-based next-generation sequencing (NGS) assays after curative surgery could facilitate the individualized management of patients with resected pancreatic cancer (PC). Here, we prospectively evaluated the clinical performance of tumor-informed ctDNA mutation analysis using a novel Burning Rock Patient Specific Prognostic and Potential Therapeutic Marker Tracking (brPROPHET) approach for assessing MRD in patients with resected PC. Methods: The prospective cohort study recruited patients (n = 20) diagnosed with resectable stage I-III PC. Plasma samples (n = 63) were collected before surgery (baseline), 7-days after surgery (landmark), 30-days after surgery (before any adjuvant therapy), during the adjuvant therapy and follow-up. Individual tumors and matched white blood cells (WBCs) were whole-exome sequenced and somatic mutations were identified. Serial plasma samples were analyzed by a personalized, tumor-informed ctDNA-based NGS assay (brPROPHET) designed to target up to 50 variants per patient. Results: A total of 20 patients (stage I/II 10 [50%]/9[ 45%]) were analyzed. Patients were followed for a median of 190 days (range: 158 days to 396 days). Preoperative ctDNA was detected in 100% (20/20) of the patients. MRD status was analyzed postoperatively on day 7 and day 30 with a positivity rate of 11% (2/18) and 17% (2/12), respectively. Of the 4 patients with known recurrences, 2 had detectable ctDNA (50%, 2/4) at 7 days after surgery, whereas all disease-free patients were ctDNA-negative (100%, 13/13) at this time point. Detection at the landmark timepoint 7 days after surgery was associated with shorter recurrence-free survival (hazard ratio [HR]: 16.87, p=0.00363). The landmark negative predictive value (NPV) and positive predictive value (PPV) were 87.5% and 100%, respectively. When integrating all timepoints, 100% (4/4) of relapse patients were ctDNA positive before relapse, with the median leading time by brPROPHET assay to radiological recurrence was 126 days, whereas all of the patients with constant ctDNA negative status were disease-free (longitudinal PPV 80%, NPV 100%). In one relapsed patient who had elevated preoperative carbohydrate antigen (CA) 19-9 and constant normal postoperative CA19-9 (<27 U/ml), ctDNA was detected prior to radiological relapse, with a lead time of 211 days. Conclusions: Patient-specific tumor-informed ctDNA-based postoperative monitoring predicts disease recurrence at earlier postoperative settings better than the clinical parameter CA19-9, which paves an alternative strategy in the individualized management of patients with resected PC.