ERASE: Elimination of Recurring Artifacts and Stochastic Errors: an NGS Analysis Approach and Its Application to Liquid Biopsy Samples.

e13002 Background: Next generation sequencing (NGS) of blood-derived nucleic acids is an emerging paradigm for determining the mutational status of cancer patients over time. Circulating tumor cells (CTC) and cell-free circulating DNA have been proposed as possible sample types for extracting tumor DNA. A number of sequencing and approaches (i.e. barcoding) have been proposed for increasing sensitivity/specificity for somatic mutation detection but require specialized chemistry and informatics; here we present a novel methodology that is compatible with routine sequencing lab workflows and functions well at allelic frequencies below 1%. Methods: Analytical sensitivity / specificity was determined by DNA spikes and cell spikes into whole blood resulting in mutation at allelic frequencies down to 0.4%. For validation using clinical samples, a number of cancer patient blood samples were enriched for circulating tumor cells (CTCs) using the IsoFlux System. Matched samples were enumerated to determine the CTC load. Cells were lysed andDNA amplified by whole genome amplification. Targeted libraries were sequenced using the NextSeq platform (Illumina) and analyzed using the ERASE algorithm (Elimination of Recurring Artifacts and Stochastic Errors). VarSeq software (Golden Helix Inc.) was used for variant functional interpretation. Results: Analytical DNA spiking experiments demonstrated a detection limit of 0.5%, with no false positives detected across a standard amplicon set aimed at somatic mutation detection. Comparisons of the ERASE methodology results with standard sequencing demonstrates a drastic reduction of a number of different classes of errors by accounting for the statistical significance of the sample mutations compared to background at each position. Clinical samples demonstrated variant detection in a majority of samples, and concordance for the biological replicates tested. Conclusions: The ERASE approach is compatible with routine amplicon based sequencing, and increases the detection floor from about 5% to 0.5% for such samples. It enables detection of somatic variants from liquid biopsy samples.