An ultra-sensitive amplicon-based NGS assay for MRD detection.

Abstract

e13634 Background: Detection of the minimal residual disease (MRD) has proven a promising noninvasive approach for monitoring the tumor remissions and relapses after surgical resection or other therapies. However, circulating tumor DNA (ctDNA) after surgery in the early-stage cancers can be at very low levels, which makes the MRD detection quite challengeable if the plasma sample only contains a few ctDNA copies. In the clinical practice, an ultra-sensitive and tumor-informed multiplexing variant detection system is required for postsurgical cancer surveillance. Methods: In this study we developed an ultra-sensitive amplicon-based NGS approach, namely UMIamp, to provide a personalized and tumor-informed strategy for MRD detection using plasma sample. Firstly, up to 16 clonal somatic variants can be obtained by whole exome sequencing (WES) of the tumor tissue samples. Then optimized primers were designed to uniformly cover the target DNA regions and those target regions were efficiently enriched by linear amplification. With randomly unique molecular identifiers (UMI) attached to DNA templates and ultra-deep NGS sequencing, the somatic target mutations with very low frequency can be identified using a customized bioinformatic pipeline. To validate the performance of the limit of detection (LOD), the reference standard samples were diluted to the allele frequencies of 0.02%, 0.01%, 0.005% and 0.002%, respectively. Each sample contained 14 single nucleotide variants (SNVs). The sensitivity and specificity were calculated based the mutations called. Results: Single site level sensitivity and specificity were 75% and 100% with the allele frequency of 0.02%, while they were 70.73% and 95.08% with the allele frequency of 0.01%. The results showed UMIamp was very sensitive and was able to detect 2 copies from 60ng cfDNA sample at 0.01% mutation level. Even at the 0.005% and 0.002% allele frequency levels there were less than two copies in 60ng cfDNA, UMIamp still achieved single site sensitivities of 24.18% and 23.07%, respectively. In the MRD clinical practice, sample level sensitivity was important to evaluate the detection capability based on more gene mutations. When combined the 14 variants as a unified indicator to detect the sample level MRD status, the positive rates at the VAFs of 0.02%, 0.01%, 0.005% and 0.002% were all 100%, which demonstrated the LOD of UMIamp approach can reach to as low as 0.002%. Such sensitivity will be adequate for post-surgical early-stage cancer surveillance. Conclusions: We developed an ultra-sensitive amplicon-based NGS approach which can effectively detect MRD status at the allele frequency of 0.002%. The developed approach can be applied in a personalized, tumor-informed MRD detection settings in post-surgical early-stage cancers.