Comparison of variant allele frequency and number of mutant molecules as units of measurement for circulating tumor DNA

Quantification of tumor-specific variants (TSVs) in cell-free DNA is rapidly evolving as a prognostic and predictive tool in patients with cancer. Currently, both variant allele frequency (VAF) and number of mutant molecules per mL plasma are used as units of measurement to report those TSVs. However, it is unknown to what extent both units of measurement agree and what are the factors underlying an existing disagreement. To study the agreement between VAF and mutant molecules in current clinical studies, we analyzed 1116 TSVs from 338 patients identified with next-generation sequencing (NGS) or digital droplet PCR (ddPCR). On different study cohorts, a Deming regression analysis was performed and its 95% prediction interval was used as surrogate for the limits of agreement between VAF and number of mutant molecules per mL and to identify outliers. VAF and number of mutant molecules per mL plasma yielded greater agreement when using ddPCR than NGS. In case of discordance between VAF and number of mutant molecules per mL, insufficient molecular coverage in NGS and high cell-free DNA concentration were the main responsible factors. We propose several optimization steps needed to bring monitoring of TSVs in cell-free DNA to its full potential.

Automated summary

Recent studies have shown that quantification of tumor-specific variants (TSVs) in cancer patients is more consistent when using digital droplet PCR compared to next-generation sequencing (NGS). In cases of discordance between variant allele frequency (VAF) and number of mutant molecules per mL plasma, insufficient molecular coverage in NGS and high cell-free DNA concentration are the main contributing factors. Further optimization steps are proposed to fully realize the potential of monitoring TSVs in cell-free DNA.