Postoperative minimal residual disease detection using tumor-informed circulating tumor DNA analysis in stage II–III gastric cancer: Updated analysis.

428 Background: Circulating tumor DNA (ctDNA) analysis enables detection of minimal residual disease (MRD), offering potential for improved risk stratification and guidance for adjuvant chemotherapy (ACT). We evaluated the clinical utility of tumor-informed ctDNA analysis for postoperative MRD assessment in stage II–III gastric cancer (GC). Methods: This sub-study of the ongoing EXODOX trial (NCT04787354), a phase III randomized study comparing reduced-duration versus standard 6-month XELOX as ACT in patients with pathologic stage II–III GC after curative resection, investigated ctDNA MRD using the CancerDetect assay (IMBdx, Inc.). This tumor-informed bespoke panel (BSP) was generated by whole-exome sequencing of tumor tissue and paired white blood cells, identifying up to 100 patient-specific variants. Whole blood (20 mL) was collected at five postoperative time points: 3–10 weeks (P0), 6 months (P1), 12 months (P2), 18 months (P3), and 24 months (P4). MRD positivity was defined as detection of ≥2 BSP variants in plasma. Results: This analysis included 68 patients enrolled between May 2022 and March 2025. ctDNA MRD was assessed at P0 (n=64), P1 (n=50), P2 (n=36), P3 (n=27), and P4 (n=15). Median age was 66 years (range, 39–96); 45 patients (66.2%) were male; all had adenocarcinoma; 54 patients (79.4%) had stage III disease. At P0, MRD positivity was 43.8% (28/64), with a median of 7 somatic mutations per patient (range, 2–94). MRD positivity was significantly higher in stage III versus stage II patients (52.0% vs. 14.3%, P=0.015). MRD clearance after ACT occurred in 40.0% (8/20) of patients who were ctDNA-positive at P0. Conversely, 5 (19.2%) of 26 patients who were initially ctDNA-negative converted to positive, resulting in an overall positivity rate of 34.0% (17/50) at P1. MRD positivity rates were 27.8% at P2, 25.9% at P3, and 13.3% at P4. After a median follow-up of 13.3 months, 11 patients (16.2%) experienced recurrence. ctDNA positivity was significantly associated with worse recurrence-free survival (RFS) at P0 (HR, 15.8; 95% CI, 2.0–123.4; P=0.01), P1 (HR, 5.5; 95% CI, 1.1–27.4; P=0.04), and P2 (HR, 24.5; 95% CI, 2.6–234.0; P=0.01). Patients who converted from positive to negative after ACT had RFS comparable to those with persistent negativity, while patients with persistent positivity had the poorest outcomes (HR, 11.6; 95% CI, 2.3–58.1; P<0.01). Conclusions: In stage II–III GC, postoperative MRD detected by ctDNA was strongly associated with recurrence risk. MRD clearance after ACT was associated with outcomes similar to persistent negativity, supporting the clinical utility of ctDNA MRD in guiding ACT decisions. Continued follow-up is underway to validate these findings.