Low-pass whole methylome sequencing–based liquid biopsy for metastatic colorectal cancer monitoring in the VALENTINO trial.

Abstract

3552 Background: The amount of ctDNA is a proxy for metastatic colorectal cancer (mCRC) disease burden, with potentials for prognostic stratification and treatment monitoring. We investigated two methods based on low-pass whole genome and methylome sequencing (WGS and WMS) for ctDNA detection and quantification in the VALENTINO trial. Methods: All patients from the VALENTINO trial - a phase II trial comparing the addition of 5FU to Panitumamb-based maintenance after first line FOLFOX+Panitumamb induction in RAS wild-type mCRC - were eligible. Baseline (BL) and 8-week (8w) plasma samples were collected for low-pass WGS and WMS analysis. Two methods for ctDNA quantification based on DNA methylation (METER) and copy number alterations (ichorCNA) were assayed. Chi-squared, Wilcoxon and Cox regression tests were used. Performances of WMS and variant allele fraction (VAF) of a 14-gene panel were compared. Results: A BL liquid biopsy was available for 154 patients, with 142 also having an 8w assessment. METER and ichorCNA detected ctDNA in 112 (72.7%) and 94 (59.7%) BL samples, respectively; all discordant cases were METER+ but ichorCNA-. Detection rate increased in the presence of liver metastases (86.0% vs 42.6% for METER, 75.7% vs 23.4% for ichorCNA; both p < 0.001) and decreased with peritoneal metastases (55.6% vs 78.0% for METER, 38.9% vs 66.1% for ichorCNA; p = 0.011, p = 0.006). Tumor fraction (TF) of both BL METER and ichorCNA correlated with the diameter of measurable lesions (both p < 0.001) and CEA (p < 0.001 and p = 0.010). Both PFS and OS were shorter after baseline ctDNA detection with METER (mPFS: 10.6 vs 18.6 months, HR: 1.65, p = 0.010; mOS: 28.7 vs 62.2 months; HR: 2.24, 95%CI: 1.37-3.66; p = 0.001) or ichorCNA (mPFS: 10.6 vs 15.0 months, HR: 1.42, 95%CI: 1.00-2.00, p = 0.047; mOS: 27.8 vs 48.4 months, HR: 1.35, 95%CI: 1.29-2.95; p = 0.002). In the multivariate analysis, METER ctDNA detection was the strongest predictor of both PFS and OS (p = 0.005 and p = 0.001) while ichorCNA ctDNA detection was significantly associated with OS but not with PFS (p = 0.002 and p = 0.093). METER ctDNA TF decreased significantly at 8w in patients with CR, PR, or SD (paired Wilcoxon p = 0.015, p < 0.001, p < 0.001) but not PD (p = 0.560) as the best radiological response. Patients without METER ctDNA clearance at 8w had a higher risks of progression (HR: 2.70, 95%CI: 1.63-4.49; p < 0.001) and death (HR: 3.37, 95%CI: 2.00-5.69; p < 0.001). Among 123 patients with both METER and VAF available, concordance was 78.0% and in 10 and 17 patients, respectively, ctDNA was detected only with METER or only with VAF. The mPFS and mOS of discordant cases were longer than METER+ / VAF+ cases and shorter than METER− / VAF− cases. Conclusions: CtDNA quantification with low-pass WMS by METER retains a prognostic significance, can be used for disease monitoring during treatment and refines ctDNA detection based on a restricted gene panel assay. Clinical trial information: NCT02476045 .