Clinical utility of whole genome sequencing for upfront risk stratification in AML and MDS patients

Acute Myeloid Leukemia (AML) and Myelodysplastic Syndromes (MDS) are hematologic malignancies that require a quick and accurate risk stratification for guiding treatment decisions. In this study, we retrospectively assessed the accuracy and utility of Whole Genome Sequencing (WGS) using clinical samples previously tested at Labcorp with FISH, Chromosomal banding, microarray, single gene molecular technologies, and/or NGS. Samples were selected based on presence of structural variants (SVs), Copy Number Variants (CNVs), or clinically significant mutations e.g. FLT3-ITD, and included controls and healthy donors. 90 total samples were included with 74 suspected of having AML/MDS. To determine WGS performance, all available clinical testing results were leveraged; samples were also assayed on Labcorp's 141 gene pan-heme NGS assay that reports somatic small variants when VAF >3%, as well as 16 CNVs. Sensitivity and specificity were calculated for single nucleotide variants (SNVs), indels, and CNVs overlapping the pan-heme NGS panel, as well as SVs and CNVs from other clinical testing results. Libraries prepared using Illumina's PCR-free Tagmentation kit were sequenced on a NovaSeq X Plus and analyzed using Illumina's DRAGEN Heme WGS pipeline in Illumina Connected Analytics; variants were scored and interpreted using Illumina Connected Insights. Estimated TAT: 67 hours per batch of 15 samples before director review and sign out. Mean genomic coverage for the cohort was 160X. Excluding events with FISH probe positivity <10%, WGS detected 116/121 SVs and 122/127 CNVs identified by cytogenetic/FISH analysis (sensitivity 0.96 and 0.98, respectively). The comparator targeted NGS assay averages 400X read depth. WGS detected 426/427 (.998) variants >10% VAF and 441/447 (.989) variants >5% VAF. Additional clinically relevant findings were identified with WGS in regions not covered by the comparator assays. Calculating performance for reportable SNVs >10% we observed a positive predictive value (PPV) of 100% with no known false positives (FPs). SV and CNV false negatives (FNs) were driven by variants below the Limit of Detection (LOD) expected for WGS. SNV FPs and FNs were from minor fluctuation between the two methods near the performance threshold. Using the Platinum Genome Cell line NA12878 and healthy donors, specificity was 100% with no FPs detected across 11 replicates and 9 individuals evaluating SNVs/indels within the pan-Heme ROI. Utilizing a WGS approach we observed increased sensitivity compared to similar previously reported studies, likely resulting from improved technology and increased sequencing depth. Compared to current laboratory assays WGS has comparable accuracy, sensitivity, and specificity, with additional benefits in clinical utility, cost reduction, improved workflow, TAT, and robustness.