Microscopic residual disease (MRD) monitoring with circulating tumor DNA (ctDNA) across gastrointestinal (GI) malignancies.

Abstract

176 Background: ctDNA is a powerful tool that can detect early evidence of tumor relapse, or MRD. There are limited studies characterizing the performance of ctDNA in non-colorectal GI cancer patients (pts). We describe the clinical outcomes of GI cancer pts who underwent real-world MRD testing at our institution. Methods: We retrospectively analyzed GI cancer pts with ≥1 ctDNA test (Natera, Inc) ordered for MRD testing between June 2020-August 2022. Descriptive statistics were used to characterize rates of ctDNA positivity (ctDNA+) and relapse rates (RR). Median relapse-free survival (mRFS) and overall survival (mOS) were calculated using Kaplan-Meier method, group comparisons by log-rank test, and hazard ratios (HR) by Cox proportional hazard models. Associations between initial mean tumor molecules per milliliter (MTM/mL) (continuous variable) and clinical factors were assessed by Kruskal-Wallis or Wilcoxon Rank-sum tests. Results: ctDNA testing was successful in 222/228 pts (table). The median follow-up was 15 months (mo) (Interquartile Range 8.6-24.7). 61/222 (27.5%) of pts had ≥1 ctDNA+ test of which 59% had relapse on their next scan obtained after a median of 40 days from the first + test. Of all relapsed pts (62/222), ctDNA detected relapse before imaging in 69.4% with a median lead time of 57 days. The positive and negative predictive values of ctDNA for relapse were 82% and 92.5%, respectively. The mRFS was 10.8 mo vs. not reached (NR) (HR 18.6; p<0.0001) and mOS 43.1 mo vs. NR (HR 5.0; p=0.002) in anytime vs. never ctDNA+ pts with similar trends in the colorectal and pancreas subgroups. Of ctDNA+ pts with >1 test, 49.2% (29/59) cleared ctDNA with 79.3% of clearances on/after therapy. Between ctDNA clearers and non-clearers, RR were 72.4% vs. 93.3%, mRFS 12.4 vs. 7.6 mo (HR 0.3; p=0.6), and mOS NR vs. 43.1 mo (log-rank test p=0.01). No significant associations existed between initial MTM/mL and tumor types (p=0.6), relapse status (p=0.2), or number of relapse sites (p=0.4). Conclusions: ctDNA is a prognostic tool able to detect early relapse with high accuracy in GI cancers. The initial MTM/mL in ctDNA+ pts did not correlate with tumor histology or relapse patterns. Pts clearing ctDNA tended to have improved outcomes; however, RRs were higher than anticipated. More pts and longer follow up will help characterize the performance of ctDNA across rarer GI subtypes. [Table: see text]