Abstract Next-generation sequencing (NGS) of relevant genes offers critical insights into diagnostic, prognostic, and therapeutic biomarkers for hematologic malignancies. We present a validated, automated variant oncogenicity workflow for hematologic malignancies, developed at Labcorp and implemented within the GenomOncology (GO) Variant Interpretation Engine software. This system enables high-throughput interpretation of variants across 141 genes included in Myeloid, Lymphoma, and Pan-Heme panels. The workflow integrates the latest ClinGen/CGC/VICC oncogenicity guidelines (Horak et al., 2022) and ACMG/AMP somatic classification criteria (Li et al., 2017) for hotspots, supporting oncogenicity of SNVs, indels, tandem duplications, and select CNVs. Within the oncogenicity workflow, variants are assessed using 22 evidence categories, including population frequency, functional impact, computational predictions, and curated cancer hotspot data. These categories are hierarchically weighted to assign oncogenicity at the earliest point of reliable evidence. The classification logic was refined over 12 iterative cycles using a ground truth dataset of in-house Tier assignments (2018–2022), and implemented as a customizable, template-driven ruleset within the GO platform. We also performed an ongoing prospective quality control (QC) study, which uses a point-based workbench developed in accordance with ClinGen/CGC/VICC (Horak et al., 2022) guideline to assess variant oncogenicity manually. Validation of the oncogenicity workflow was performed on 23,001 myeloid variants and 732 pipeline validation variants, achieving 99% concordance with manual Tier classification in our previous system. In the first 13 weeks following assay launch, we applied this workflow to 7969 unique variants in clinical samples, with a significant improvement in turnaround time (TAT). With the current automated workflow, 99.5% of variants are now classified automatically. We performed a prospective quality control study involving 1,650 randomly selected variants from early clinical testing (49.3% in myeloid genes, 50.7% in lymphoid/other genes) and compared variant oncogenicity between the automated workflow and manual review. Through this study, we identified improvement rules that can be implemented in the automated workflow including gene specific rules and use of subpopulation data in gnomAD v4 database. These enhancements improved the automated workflow to 99.6% concordance between automated and parallel manual review demonstrating both the accuracy and scalability of the automated system. This automated framework enables rapid, reproducible, and scalable variant classification, significantly improving workflow efficiency and supporting consistent clinical decision-making in hematologic oncology.
Gardner et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: