Abstract 2489: Characterization of breast cancer ctDNA with Omi-Seq

Cancer patients' cell free DNA (cfDNA) contains circulating tumor-derived DNA (ctDNA). The fast dynamic of clearance allows ctDNA to be used as biomarkers for tumor real time profiling. The benefit of non-invasive, cost effective, and easily-repetitive sampling makes liquid biopsy a highly-desirable option for molecular assessment. Deep sequencing has been widely used to investigate subpopulations in complex biological samples. However, there are still limitations of the sequencing technology due to the low yield of cfDNA amount, and errors introduced during sample preparation and sequencing. Approximately 1% of bases are incorrectly identified, depending on the specific platform and sequence context. Although Molecular Identifier can theoretically greatly reduce errors, it suffers from several drawbacks. First, it is difficult to synthesize double strand adapters with randomized Molecular Identifier. Second, due to the difficulty in making high quality Single Molecular Identifier (SMI) adaptors, the ligation efficiency might be compromised and therefore might require large amounts of input DNA. To optimize the liquid biopsy protocol, we first examined if there is any bias in the molecular barcodes' selection during library preparation. Indeed, a skewed trend with molecular barcode selection in ligation process has been detected. We also evaluated the error rates in molecular barcodes and found that there is a substantial amount of errors (~5% of total reads carrying the wrong molecular barcode). To overcome these issues, Omi-seq calculates Digital Molecular Identifier (DMI) using a pool of regular adapters with defined barcodes which can be easily synthesized in high quality. We then analyzed Horizon multiplex I cfDNA reference standard with the lowest available allele frequency (AF) 0.1%. With Omi-error correction, Omi-Seq was able to reproducibly detect the variants. We also evaluated Omi-Seq in real patient samples, Omi-Seq is able to detect variants at AF 0.03%. Omi-Seq was utilized to created cancer profiles for 202 subjects enrolled in The First Affiliated Hospital of Wenzhou Medical University breast cancer study. The gene landscape indicated that the most frequently detected genes are PIK3CA and TP53, followed by ERBB2. The accuracy of Her2 CNV in tissue/FFPE sample is 96% with 85 % sensitivity and 100% specificity. Mutation numbers vary between different molecular subtypes of cfDNA, where the basal-like group had more variants detected than Lumin