Circulating tumor DNA (ctDNA) dynamics in liver-limited metastatic colorectal cancer (mCRC) patients resected after first-line systemic treatment.

3544 Background: Liver-limited disease (LLD) occurs in 20-30% of metastatic colorectal cancer (mCRC) patients. Although 20-30% of patients who undergo resection can achieve a long-term overall survival benefit from liver surgery, most patients relapse during the first two years after hepatectomy. ctDNA is a promising tool in detecting the presence of minimal residual disease (MRD) after resection of colorectal liver metastases and a reliable prognostic tool for recurrence. ctDNA and its dynamics may also serve as a prognostic tool in patients candidate to liver resection following upfront chemotherapy. Methods: mCRC patients (N = 116) with initially unresectable LLD and R0/R1 resected after upfront chemotherapy were selected from 3 Italian academic centers. Blood samples were collected prospectively at baseline (T0), pre-surgery (TPrS) and post-surgery (TPoS). T0 samples were evaluable for 82 patients, TPrS for 116 and TPoS for 60. Biobanked plasma samples were analyzed with the Tempus xM MRD assay (xM), a tumor-naïve ctDNA MRD assay that integrates methylation and genomic variant classifiers to deliver a binary MRD call blinded to clinical outcomes. The methylation classifier detects fragments with CRC methylation signatures in differentially methylated regions trained by sequencing CRC and presumed-healthy samples on a 6 Mbp panel. The variant classifier detects highly prevalent CRC variants. Results: Methylation results were available for 60 TPoS patients with a clinical sensitivity of 56.4% and specificity of 100%. TPoS ctDNA status was associated with relapse-free survival (RFS) with the ctDNA- group experiencing longer median RFS (mRFS) than ctDNA+ (HR = 6.7, mRFS > 24 mos vs. 5.5 mos, p < 0.001). Patients who were persistently ctDNA- by methylation calls (n = 20) or converted to negative (n = 13) from TPrS to TPoS experienced longer RFS (mRFS 16.3 mos and > 24 mos respectively). Those who remained persistently ctDNA+ (n = 9) or converted to ctDNA+ (n = 12) had a mRFS of 5.3 and 5.9 mos respectively. Patients with variant allele fraction (VAF) reduction of ≥50% from T0 to TPrS (N = 53) experienced longer RFS than those who had < 50% reduction or increase in VAF (N = 18) (HR 2.21, mRFS 18.8 mos vs. 9.8 mos, p = 0.012). Lastly, patients that remained positive from T0 to TPrS (N = 23) experienced a numerically shorter RFS compared to those who converted to negative (N = 47) (median RFS 10.4 and 15.1 mos, HR 1.65, p = 0.10). Conclusions: xM demonstrates remarkable performance in predicting clinical recurrence and correlation to RFS at TPoS in LLD mCRC patients resected after upfront systemic therapy. Interestingly, patients with a VAF reduction ≥ 50% experience longer RFS following surgery, suggesting a potential role for this tool in multidisciplinary decision making in this setting.