Response prediction by mutation- or methylation-specific detection of ctDNA dynamics in pretreated metastatic colorectal cancer

BACKGROUND: Serial analysis of circulating tumor DNA (ctDNA) levels is a promising tool for both relapse prediction in the curative setting, as well as predicting clinical benefit from systemic treatment in metastasic colorectal cancer (mCRC). Most data in this context are derived from treatment naive patients. OBJECTIVE: To predict progressive disease (PD) as early as possible through monitoring of changes in ctDNA levels during systemic treatment in pretreated patients with mCRC. DESIGN: A prospective, single-center, observational study. METHODS: Patients treated beyond first-line were prospectively included between February 2020 and September 2021. Blood for ctDNA detection was taken before every treatment cycle from start of treatment until first restaging by CT-scan. ctDNA was detected by mutation- (mut-ctDNA) and methylation-specific ddPCR. Receiver Operating Characteristic (ROC)-analysis was used to describe sensitivity and specificity for prediction of PD at restaging for all time points. RESULTS: A total of 42 patients were included who all carried a mutation in tumor tissue. Detection rate of mut-ctDNA was 88.1% and 74.4% for meth-ctDNA. Absolute ctDNA levels before treatment were prognostic in terms of overall survival. Levels of ctDNA were significantly higher in patients with PD at restaging. Median time from start of treatment to restaging was 93 days (95% CI 88.8–96). After a median of 19 days of treatment (95% CI 16.1–20.2), a decline of either mutation- or methylation-specific ctDNA levels of ⩽58% predicted PD at restaging with a sensitivity/specificity of 92.9/85.7% and 85.7/100%, respectively. Median time to restaging was 66 days (95% CI 56.8–75.2). There was no significant increase of sensitivity/specificity at later time points of ctDNA measurements. CONCLUSION: Monitoring early changes of ctDNA levels either by mut- or meth-ctDNA allows for early prediction of PD in pretreated patients with mCRC. This has the potential to complement RECIST-based treatment assessment with the aim to switch potentially insufficient treatments as early as possible, which is of particular interest in higher treatment lines.