Single-stranded linear amplification library preparation method for highly sensitive and low input amount ctDNA detection.

Abstract

e15096 Background: Detection of circulating tumor DNA (ctDNA) or minimum residual disease (MRD) through somatic variant identifications in plasma samples has emerged as a promising approach for monitoring the clinical outcomes after surgical resections or other therapies. However, due to very low ctDNA levels in the early-stage cancers and tumor remissions after therapies, or low plasma volumes and DNA degradation caused low qualities in clinical practice, to sensitively identify somatic mutations with low allele frequencies in those samples remains a challenging technology task. Methods: In this study, we developed a single-stranded, linear amplification based library preparation sequencing (SLA-Seq) method. In the experimental workflow, double stranded molecular templates will be converted into single strands to construct libraries, thus minimize the loss of irregular double stranded DNA molecules. Adapter ligation was carried out in two steps. During the first step, P7 adapters with unique molecular identifiers(UMI) were ligated to the 3’ end of the molecular templates, and the DNA molecules were amplified by linear amplification to improve the DNA utilization efficiency. In the second step, P5 adapters were ligated to the linear amplified PCR products, then amplified by the pre-PCR. The pre-library was enriched by hybridization and magnetic-beads capturing and then subjected to high-throughput sequencing. 30ng of the cfDNA reference standard samples containing mutations of 0.3% and 0.1% allele frequencies were used to validate the performance. Results: In the SLA-Seq method, adapter sequences were carefully optimized to eliminate adapter dimers and thus improved the DNA ligation efficiency. Compared with the double-stranded library preparation method, SLA-seq showed overall higher effective depths after removing duplications if the same start amount of the DNA was used, which indicated that the higher DNA utilization rate was achieved. To evaluate the performance, we applied SLA-Seq on 11 reference standard samples with only 30ng start amount DNA. Among them, 6 samples were diluted to 0.3% allele frequency and 5 samples were diluted to 0.1% allele frequency. Totally they had 24 and 35 mutations, respectively. The sensitivity at the 0.3% allele frequency was 100% (24/24) and it was 94.29% (33/35) at the 0.1% mutation level, which also showed the higher sensitivity achieved by SLA-Seq than double-stranded library preparation method. Conclusions: We developed a single-stranded, linear amplification based library preparation method to improve difficult sample situations such as low input amount and low qualities, and achieved high single site sensitivity at the 0.1% allele frequency. Such features imply SLA-Seq has important clinical significance in the ctDNA/MRD detection.