Purpose: Baseline circulating tumour DNA (ctDNA) is a potential prognostic marker in metastatic colorectal cancer (mCRC) patients. However, few studies have compared ctDNA with the usual prognostic factors, and no ctDNA cut-off has been proposed for daily use in clinical practice.
Patients and methods: Chemotherapy-naive patients with mCRC were prospectively included. Plasma samples were collected at diagnosis and analysed centrally by both NGS and methylation digital PCR. Baseline patient and disease characteristics, treatment regimens, and secondary surgeries were collected. The restricted cubic spline method was used to define the optimal cut-off of ctDNA mutated allelic frequency (MAF). Prognostic values were assessed on overall survival (OS) using Cox models.
Results: From July 2015 to December 2016, 412 patients were included. ctDNA was undetectable in 83 patients (20%). ctDNA was an independent prognostic marker for OS considering the whole study population. The optimal cut-off for ctDNA MAF was 20% with median OS of 16.0 and 35.8 months for patients with MAF ≥20% and<20%, respectively (hazard ratio = 0.40; 95% confidence intervals: 0.31-0.51; P < 0.0001). The independent prognostic value of ctDNA MAF at 20% was confirmed in subgroups defined by RAS/BRAF status or resectability of metastases. Combining ctDNA MAF and carcinoembryonic antigen levels allowed us to define three different prognostic groups with median OS of 14.2, 21.1, and 46.4 months (P < 0.0001).
Conclusion: ctDNA with a MAF cut-off of 20% improves prognostication of chemotherapy-naïve mCRC patients and may be useful in the future for individualised therapeutic decisions and as a stratification factor in clinical trials.
Trial registration: Clinicaltrials.gov, NCT02502656.
Keywords: BRAF; CEA; Circulating tumour DNA; Colorectal cancer; Metastases; Prognostic value; RAS.