Personalized fusion measurable residual disease detection in acute myeloid leukemia using droplet digital PCR

Acute myeloid leukemia (AML) is an aggressive cancer, of which 30-40% are driven by gene fusions. Effective therapy selection often requires accurate assessment of disease burden-termed measurable residual disease (MRD) detection. MRD is typically assessed with multiparameter flow cytometry (MPFC) or RT-qPCR, which are limited by accuracy, cost, turnaround time or applicability. MPFC has a reported limit of detection of 1:1,000 to 1:10,000, but 20% of MRD negative patients by MPFC relapse within weeks to months. MPFC requires an invasive bone marrow biopsy, so testing is infrequent and critical treatment decisions are often based on a single measurement (e.g. end of induction). In fusion driven AML, the fusion itself is a robust biomarker of disease. Unfortunately, only the most common gene fusions (e.g. BCR-ABL and PML-RARA) have RT-qPCR-based MRD assays. But there are hundreds of gene fusions that drive AML that do not have a molecular MRD assay. To address this limitation, we have developed personalized gene fusion MRD detection using droplet digital PCR (ddPCR). We utilize diagnostic sequencing to define a patient's unique fusion, design their personalized MRD assay and validate the assay to a limit of detection of 1 in 100,000. We applied this approach in a cohort of 20 AML and acute lymphoblastic leukemia (ALL) patients harboring diverse gene fusions including KMT2A (with partners including AF9 , AF10 , AF4, ELL, ENL , AF6 , AF1q ), RUNX::RUNX1T1 , PAX5::MLLT3, EP300::KMT2A and BCR::ABL . We focused on KMT2A fusions due to their heterogeneity and clinical need for molecular MRD given the development of Menin inhibitors. Using personalized ddPCR assays, we analyzed 155 patient specimens (median 4.5, range 1-24 per participant) collected at diagnosis, throughout treatment, and during surveillance from bone marrow aspirates (n=31) and peripheral blood samples (n=124). We identified disease recurrence in eight patients by ddPCR 38-103 days before standard of care MPFC-based testing or clinical relapse. In our study, every patient with increasing fusion burden relapsed and every patient who remained in long-term remission cleared their fusion. Fusion transcript abundance was similar in peripheral blood and bone marrow specimens obtained concurrently, demonstrating non-invasive peripheral blood sampling may be able to replace bone marrow biopsy for fusion-based MRD detection. In conclusion, personalized gene fusion MRD testing by ddPCR is highly sensitive and specific for disease recurrence. This approach is ideal for real-time clinical decision-making given the low limit of detection, non-invasive testing from the peripheral blood, and rapid turnaround time (1-2 days). Future work is focused on building trials to assess clinical utility. Long-term this has the potential to improve treatment selection and drive better patient outcomes.