Clinical translation of single cell measurable residual disease (scMRD) assay in real world Acute Myeloid Leukemia (AML) patients

Abstract In this study, we aimed to analyze the prognostic potential of a single-cell MRD (scMRD) assay in a prospective cohort of real world acute myeloid leukemia (AML) patients treated with either chemotherapy or hypomethylating agent plus venetoclax (HMA+Ven) therapy in routine clinical care. This assay has been reported to have a sensitivity of approximately 0.01% in retrospective samples (Robinson et al., 2023), but its utility in prospective clinical samples is unknown. We hypothesize that the incorporation of multiomics into MRD detection and further single cell characterization of AML, can address limitations intrinsic to clinical MRD assays. To study scMRD's clinical applicability, we prospectively enrolled 21 sequential patients receiving routine induction therapy for AML at the Sylvester Comprehensive Cancer Center (SCCC) of the University of Miami. Of this cohort, a total of 25 samples were collected and processed for scMRD using Mission Bio's TapestriⓇplatform. Bone marrow aspirates were enriched for CD34+ and CD117+ cells (Miltenyi) and were then multiplexed, encapsulated, and barcoded to generate single-cell genomic and protein libraries using Tapestri's v3 chemistry. The samples were sequenced on an Illumina NovaSeqⓇ X Plus. The median age of the cohort was 60 (range 42-89) and the majority (62%) received induction with cytarabine and anthracycline, while 28% received HMA and venetoclax and 10% received other therapies. Of the processed samples, 17 had AML and 4 had other myeloid malignancies (CMML [1], MDS [1], or MPN [2]). The scMRD assay successfully called a total of 66,515 cells (mean 2892 cells/sample). Clinical outcomes were collected for each patient and then used to assess for similarities and discrepancies between the scMRD and clinical assay MRD calls. Of special interest, is the variation in detection found in patients that had undergone non-intensive chemotherapy treatments, including targeted therapies and HMA+Ven, prior to or at the time of scMRD sample collection (n=4). A case which highlights this variation is an AML patient harboring RUNX1::PRDM16, ASXL1, U2AF1, and FLT3-TKD variants and a complex karyotype who received HMA+Ven. At the time of sample collection, this patient had tested MRD negative by multiparameter flow cytometry (MFC) but positive by next-generation sequencing (NGS) assays. The scMRD assay detected 345 mutant cells (67.1% of 514 cells), with 3 mutant clones harboring U2AF1, chr20:31023453:A/* and FLT3 variants, as well as aberrant surface protein phenotypes. The patient went on to receive an allogenic stem cell transplant, with post-transplant relapse treated with HMA+Ven, followed by donor lymphocyte infusion. He progressed and was given cladribine, idarubicin, cytarabine (CLIA); the patient remains refractory as of June 2025. Despite discordant clinical assay results at the time of collection, the scMRD assay successfully detected multiple mutant variants which were also identified by NGS at the time of relapse. This case indicates the assay's ability to both correctly identify variants as residual leukemia despite discordant clinical MRD calls, and to predict persistent disease status and/or relapse. Similar early detections were found in AML patients treated with targeted therapy approaches. Overall, we have demonstrated the prognostic potential and applicability of the scMRD assay in a clinical setting. The integration of single cell resolution into existing MRD practices offers novel opportunities to capture early AML clonality and disease progression in recently treated patients. This insight into clonal progression also opens the door to further exploring disease patterns unique to targeted therapy and HMA+Ven treated patients. Several patients (n=4) treated with HMA+Ven exhibited scMRD-facilitated early detection of mutant clones that persisted and eventually relapsed. Further investigation into this pattern can potentially identify variables of special interest when interpreting both clinical and scMRD assay MRD results, such as variability in time of sample collection, targeted therapy history/timeline, and mutant clonal variability and complexity. The scMRD assay's high sensitivity and rapid turnaround time (<1 week) in a high-volume setting supports its feasibility in a clinical setting. Further prospective studies are warranted to assess scMRD's predictive and prognostic potential in AML patients, including those treated with HMA+Ven and targeted therapies.