Informing optimal duration of adjuvant chemotherapy (ACT) in resected stage I-IV colorectal cancer (CRC) based on early circulating tumor DNA (ctDNA) dynamics.

3501 Background: While ACT is recommended by clinical guidelines for patients (pts) with high-risk stage II and stage III CRC, individual benefit varies significantly and appropriate ACT duration is debatable. Here, we evaluated whether on-ACT ctDNA dynamics can inform recurrence risk and benefit from continued ACT beyond 3 months (mo) in resected stage I-IV CRC. Methods: This analysis included 1,028 pts from the CIRCULATE-Japan GALAXY observational study who received ACT but no neoadjuvant chemotherapy, with >2 ctDNA results available within 6 mo post-surgery. Patients were divided into long-ACT (>3 mo, N = 651) and short-ACT cohort (< 3 mo, N = 377) cohorts. ctDNA was analyzed using a personalized, tumor-informed Signatera assay at two timepoints: a) pre-ACT (≥14 days post-surgery, before ACT), and b) post-3 mo from ACT (short cohort) or 3 mo ±45 days from ACT (long cohort). Disease-free survival (DFS) was measured from the date of surgery and compared between pts who received long vs short ACT, stratified by ctDNA dynamics. Subanalyses comparing ACT duration were landmarked at the second ctDNA timepoint. Results: Of the 1,028 pts (< 1%/17%/75%/7% stages I/II/III/IV), 44% were females; 81% had colon cancer and 19% rectal cancer. Median pt age was 66 (24-89) years. Median follow-up was 30.7 (4.9-55.8) mo. Median ACT duration was 5.7 mo in long-ACT and 2.5 mo in short-ACT cohorts. Pre-ACT sampling was a median of 14 and 20 days before ACT start for long- and short-ACT cohorts, respectively; post-3 mo sampling was a median of 112 days after ACT start for both cohorts. ctDNA dynamics from the pre-ACT to post-3 mo ACT timepoints revealed sustained negativity in 78.7% (N = 809) of pts, ctDNA clearance in 14.7% (N = 151), decreasing but detectable levels in 2.7% (N = 28), rising levels in 2.8% (N = 29), and newly detectable ctDNA on ACT in 1.1% (N = 11). No benefit of short vs long ACT duration was observed among pts with sustained ctDNA negativity (HR 0.71, p = 0.11) or ctDNA clearance (HR 1.06, p = 0.84). Similarly, pts with rising ctDNA experienced no benefit from longer ACT (HR 1.82, p = 0.18). Conversely, within the decreasing but detectable ctDNA subgroup, longer ACT was associated with improved DFS compared with short ACT median DFS 5.9 vs 1.7 mo; HR 3.64, p = 0.012), acknowledging that the analysis was limited by small cohort size and requires further investigation. Conclusions: ctDNA dynamics during ACT provide clinically meaningful stratification of recurrence risk and reveal heterogeneity in treatment benefit. Pts with ctDNA clearance or sustained negativity showed durable disease control and may not require longer ACT. Conversely, pts with partial molecular response showed directional benefit from longer ACT. Pts with molecular progression on ACT did not benefit from extended ACT, indicative of chemotherapy refractory disease and the need for alternative treatment regimens. Clinical trial information: R000044197.