Clinical Validity of Circulating Tumor DNA as Prognostic and Predictive Marker for Personalized Colorectal Cancer Patient Management

Simple Summary

CtDNA analysis is a promising tool in liquid biopsy for the detection of tumor recurrence and progression, and is increasingly adopted into clinical practice. Still, guidelines for the accurate clinical interpretation of ctDNA analysis results are largely lacking, especially for tumor mutant variants detected at very low frequencies. Here, we show that cutoff determination for the detection and quantification of low-frequency mutant variants enables the accurate prediction of residual disease, tumor recurrence and progression, even before clinical evidence. CtDNA analysis using these cutoffs outperformed cfDNA and CEA level measurements. With these findings, we highlight the need to thoroughly validate each liquid biopsy assay and define the assay-specific limit of blanks (LOB) and limit of quantifications (LOQ) of BRAF p.V600E and KRAS p.G12/p.G13 assays for clinical interpretation. Our approach enables accurate clinical interpretation to support clinical decision making.

Abstract

Circulating tumor DNA (ctDNA) is a promising liquid biopsy (LB) marker to support clinical decisions in precision medicine. For implementation into routine clinical practice, clinicians need precise ctDNA level cutoffs for reporting residual disease and monitoring tumor burden changes during therapy. We clinically validated the limit of blank (LOB) and the limit of quantification (LOQ) of assays for the clinically most relevant somatic variants BRAF p.V600E and KRAS p.G12/p.G13 in colorectal cancer (CRC) in a study cohort encompassing a total of 212 plasma samples. We prove that residual disease detection using the LOB as a clinically verified cutoff for ctDNA positivity is in concordance with clinical evidence of metastasis or recurrence. We further show that tumor burden changes during chemotherapy and the course of disease are correctly predicted using the LOQ as a cutoff for quantitative ctDNA changes. The high potential of LB using ctDNA for accurately predicting the course of disease was proven by direct comparison to the routinely used carcinoembryonic antigen (CEA) as well as the circulating free DNA (cfDNA) concentration. Our results show that LB using validated ctDNA assays outperforms CEA and cfDNA for residual disease detection and the prediction of tumor burden changes.