An AML Targeted Duplex Sequencing Assay That Can Detect Measurable Residual Disease (MRD) at a Sensitivity Better Than 0.01% Variant Allele Frequency

• Published in: Transplantation and cellular therapy 2024-02, Vol.30 (2), p.S101-S102
• Main Authors: McElwain, Mark, Olson, Janice M, Zannette, Camila, Nguyen, Karen, Fitzgerald, Devon M, Lee, Isabel Y, Schmidt, Elizabeth, Vavra, Kevin C, Gillis, Laura B, Gordon, Kelly J, Wakeman, Wayne B, Smith, Thomas H, Salk, Jesse J, Higgins, Jake, Harrell, Jill C, Chu, Frances, Meredith, Gavin D
• Format: Article
• Language: English
• ISSN: 2666-6367; 2666-6367
• DOI: 10.1016/j.jtct.2023.12.162

The majority of acute myeloid leukemia (AML) patients relapse after initially achieving remission following conventional therapies. Measurable residual disease (MRD) in AML has emerged as a strong prognostic factor important for managing therapy, which may include hematopoietic stem cell transplant, and for monitoring response and predicting clinical outcomes. Next generation sequencing (NGS) assays have the potential for accurate detection of low frequency mutations in blood or bone marrow, however, PCR and sequencing errors limit accuracy at variant allele frequencies (VAF) below approximately 1%. Duplex sequencing (DS) is an error-corrected NGS (ecNGS) method that greatly reduces errors by comparing complementary DNA strands to each other to eliminate PCR and sequencing artifacts. We previously demonstrated using a 29-gene panel and a retrospective AML cohort that DS can identify approximately twice as many patients with MRD compared to flow cytometry (Dillon LW, et. al., 2023, Haematologica), and showed superior prediction of clinical outcomes. Here, we report the analytical validation of DuplexSeq™ AML-XP, a 36-gene AML targeted assay* informed by 2022 European LeukemiaNet (ELN) recommendations. We tested samples consisting of contrived human genomic DNA carrying 25 targeted coding-sequence variants plus DNA extracted from AML-positive peripheral blood and bone marrow. 82% of targeted regions exhibit sequencing depth >80% of the panel-wide mean while >99.5% of targeted regions exhibit depth >20% of the mean. The assay detected variants down to 0.00055% VAF with a calculated limit of detection (LoD) of 0.0098% VAF for single-nucleotide variants (SNV), and insertions and deletions (indel). In the LoD study, for variants above 0.0098% VAF, overall percent agreement (OPA) was 97.8%, positive percent agreement (PPA) 98.3%, and negative percent agreement (NPA) 97.7%. Background detection rate in DNA samples derived from healthy normal specimens established an assay limit of blank (LoB) of 0% VAF. Data were linear from 0.0016% to 96.1% VAF with R2=0.95. All six expected FLT3-ITD variants, and all five expected NPM1 insertions were detected. The assay was highly repeatable and reproducible across two operators, two lots of reagents, and three independent runs of library preparation. With an LoD of 0.0098% for SNVs and indels, and an LoB of 0, this novel AML assay represents a highly sensitive, specific, accurate, and precise ecNGS assay for detecting MRD in A