Early treatment ctDNA dynamics to predict response to chemotherapy in patients with metastatic pancreatic cancer: A prospective, observational pilot study.

4158 Background: Assessment of chemotherapy response using imaging has limitations in metastatic pancreatic cancer whereas circulating tumor DNA (ctDNA) may offer an alternate assessment of tumor burden with improved lead-time. This study evaluated temporal ctDNA testing as a potential biomarker of treatment response data compared to standard-of-care CT scans in metastatic pancreatic cancer. Methods: In this prospective, observational, single-center trial, patients with metastatic pancreatic ductal adenocarcinoma starting a first-line (1L) or second-line (2L) systemic treatment regimen underwent imaging and frequent blood-based treatment assessment. CT imaging was obtained prior to and after 8 weeks of treatment, with response measured by RECIST 1.1 criteria. CA 19-9, CEA, and longitudinal ctDNA profiling using Signatera were performed at baseline, after 2 weeks, 4 weeks, and 8 weeks. A threshold of 20% decrease achieved or not was used to dichotomize ctDNA response and other cut-offs were evaluated in sensitivity analysis. Progression-free survival (PFS) was measured and compared to biomarker response. The primary objective was to evaluate the association of ctDNA changes at 4 and 8 weeks with PFS. Exploratory objectives included evaluation of response at 2-weeks, and comparison of PFS with CA 19-9 and CEA. Results: Between June and December 2023, 19 patients were enrolled. Sufficient tissue was available to perform tumor-informed ctDNA profiling in 12 of 19 patients (63%) undergoing systemic therapy with 7 patients (58%) starting on 1L therapy and 5 (42%) on 2L therapy. Overall median PFS was 4.0 months (4.2 months 1L, 3.3 months 2L). CtDNA response at 4-weeks was prognostic of PFS with a median decrease in ctDNA of 83.4% (p = 0.0002). CtDNA response at 8-weeks was also prognostic of PFS (p = 0.01). Three patients achieved ctDNA clearance of over 95% by week 4 and had a PFS of 5.9 months. Of, note there was one patient who had a 1-log decrease in ctDNA at 4-weeks, but a 20-fold increase at 8-weeks, whereas CEA and CA 19-9 were stable at 8-weeks, and this patient had subsequent disease progression 1 month later. In an exploratory analysis, ctDNA as soon as 2 weeks was also prognostic of recurrence (p = 0.002). In contrast to ctDNA, CA 19-9 and CEA did not show significance at any timepoint (2 weeks: p = 0.2 and p = 0.8, respectively; 4 weeks: p = 0.7 and p = 0.8, respectively; 8 weeks: p = 0.8 and p = 0.5, respectively). Sensitivity analysis showed that 20% decrease was robust; association at 4-weeks with ctDNA remained significant across a wide threshold range (-70% to +60%). Conclusions: Circulating tumor DNA predicts PFS as early as 2-weeks following treatment. These findings suggest clinical utility in measuring ctDNA in mPDAC as early as 2 weeks following treatment initiation within the context of prospective interventional clinical trials.