Circulating tumor DNA as a marker of recurrence risk in stage III colorectal cancer: The α-CORRECT study.

302 Background: Patients with stage III CRC undergo surgery and adjuvant chemotherapy. Evaluation of MRD using a highly sensitive and specific, tumor-informed ctDNA assay may help predict which patients will have a recurrence. Methods: Patients with stage III CRC were enrolled after surgery in an observational study across 19 Pennsylvanian hospitals between 2016-2020. Tumor tissue and serial blood samples (every 3 months for years 1-3, biannually for years 4-5) were collected. Following whole exome sequencing (WES) of the tumor and selection of 50 – 200 somatic variants, a personalized MRD assay was used to measure ctDNA status of each sample. Three timepoints were used to analyze the association of ctDNA with recurrence: the single sample post-surgical (PS) and post-definitive therapy (PDT) timepoints, and the surveillance period, which included the PDT and subsequent samples. Primary endpoint was the association of ctDNA status during surveillance with recurrence-free survival (RFS). Results: A total of 124 patients with 1029 ctDNA results were analyzed. Almost all patients (96.0%) received adjuvant chemotherapy, primarily with an oxaliplatin-based regimen. On average, patients had 10 (SD = 4.5) blood draws and the median follow-up was 4.8 years. WES identified 172 (range 50 – 200) somatic variants per patient on average. Patients who were ctDNA positive during surveillance had a significantly higher likelihood of recurrence than patients who remained ctDNA negative (HR 49.6, 95% CI: 16.6 – 148.3; ctDNA status as time-dependent variable). At all timepoints, ctDNA detection was strongly associated with recurrence (Table). In multivariable analyses, ctDNA was strongly prognostic for recurrence but CEA was not. The median lead time between positive ctDNA result and a clinical diagnosis of recurrence was 10.4 months. Conclusions: In this cohort of patients with stage III CRC with long-term follow-up, a tumor-informed ctDNA assay using up to 200 variants was strongly prognostic for recurrence at all timepoints. Hazard ratios (HR), sensitivity, and specificity, with 95% confidence intervals (95% CI), for the association of ctDNA status with recurrence-free survival (RFS) at the surveillance, postsurgical (PS), and post-definitive therapy (PDT) timepoints (three-year RFS for the PS and PDT timepoints are also shown). Timepoint Statistic Result Surveillance(N = 111) HR(95% CI) 49.6**(16.6 – 148.3) Sensitivity(95% CI) 20/22 = 90.9%(72.2 – 97.5%) Specificity(95% CI) 82/87 = 94.3%(87.2 – 97.5%) PS(N = 88) HR(95% CI) 9.6**(3.2 – 29.5) Sensitivity(95% CI) 14/18 = 77.8%(54.8 – 91.0%) Specificity(95% CI) 53/66 = 80.3%(69.2 – 88.1%) RFS at 3 yrsctDNA(+), ctDNA(-) 54.5%, 96.1% PDT(N = 97) HR(95% CI) 16.7**(6.9 – 40.3) Sensitivity(95% CI) 10/21 = 47.6%(28.3 – 67.6%) Specificity(95% CI) 75/76 = 98.7%(92.9 – 99.8%) RFS at 3 yrsctDNA(+), ctDNA(-) 18.2%, 90.0% *P< 0.001, **P<0.0001.