Abstract
Relapse is one of the major challenges in clinical treatment of acute myeloid leukemia (AML). Though minimal residual disease (MRD) monitoring plays a crucial role in quantitative assessment of the disease, molecular MRD analysis has been mainly limited to patients diagnosed with gene fusions and NPM1 mutations. Here, we report a longitudinal ultra-sensitive mutation burden (UMB) monitoring strategy for accurate MRD analysis in AML patients regardless of genetic abnormality types. Using a Quantitative Blocker Displacement Amplification (QBDA) sequencing panel with limit of detection below 0.01% variant allele frequency (VAF), a hazard ratio of 14.8 (p < 0.001) is observed in cumulative incidence of relapse analysis of 20 patients with ≥ 2 samples during complete remission (CR). The ROC area under curve (AUC) is 0.98 when predicting relapse within 30 weeks of CR timepoint 2 (N = 20). Furthermore, we demonstrate quantitating VAF below 0.01% is essential for accurate relapse prediction.