P453: novel insights into minimal residual disease of acute myeloid leukemia using primary samples and a clinically relevant pdx model.

Background: Despite typically high response rates to initial chemotherapy, the rate of mortality in acute myeloid leukemia (AML) remains high due to frequent and hard to control relapses. Persistence of therapy resistant Leukemic Stem Cells (LSCs) harboring clonal outgrowth capacity and present within Minimal Residual Disease (MRD) during complete remission stage are thought to be the origin of relapse. However, identifying the exact cellular composition of MRD in patients during remission has been notoriously difficult due to the exceedingly low number of resistant leukemia cells in MRD stage hiding within the vast majority of healthy blood cells. Thus, the molecular and cellular mechanisms responsible for the functional maintenance of LSCs in patient MRD remains poorly understood. Aims: Clinical MRD sample analysis is limited by the difficulty of accessing such samples in a longitudinal manner. Therefore, we established and characterized an MRD human xenograft mouse model to circumvent the difficulty to analyze the MRD fractions of AML patients. Based on the knowledge attained, precise strategies targeting or preventing the appearance of resistant persister leukemic cells including LSCs will be developed. Methods: A unique set of four longitudinally collected triplet samples (diagnosis, remission, and relapse) were used for this analysis. Established in vivo patient derived xenograft (PDX) models from the longitudinally triplet samples were treated with combination chemotherapy (Cytarabine and Daunorubicin) in order to mimic the MRD state in the clinical setting. Using flow cytometry and RNA-seq, we analysed MRD fractions in both primary samples and the corresponding PDX models. Results: First, we determined the in vivo regimen and the optimum sub-lethal dose of Cytarabine (AraC) + Daunorubicin. Administration of AraC 30mg/kg/day for 4 consecutive days combined with 2 days of Daunorubicine 2.5mg/kg/day was determined as the most efficient treatment to observe a significant reduction in total AML cell tumor burden in the bone marrow and spleen at day 8. Additionally, we determined week 2 as the optimal time point to observe tumor regrowth mimicking the clinical MRD state. Our data show that this combination treatment (4 + 2) is well tolerated in NSG mice at a dose and schedule analogous to the clinically used 7 + 3 AML patient treatment regime, and thus allowing us to study mechanisms involved in drug resistance. Importantly, PDX generated from diagnosis samples, but not from MRD or relapse samples, displayed sensitivity to chemotherapy. Moreover, the residual AML-MRD cells surviving chemotherapy in PDX mice from diagnostic samples displayed a higher OxPhos metabolism compared to untreated controls, as we observed a higher mitochondrial membrane potential and mitochondrial mass as assessed by FACS assays TMRE and MTG respectively. In addition, MRD cells expressed higher level of the myeloid chemokine receptors. Collectively, these data correlate with the clinical data of the analysed longitudinal triplet samples as both primary patient and PDX MRD samples were enriched on High OxPhos signature and presented a higher gene expression of the myeloid chemokine receptors compared to the respective diagnosis samples. Summary/Conclusion: Taken together, we have established PDX mouse models that recapitulates the metabolic properties and the phenotypic features that we identified at the clinical MRD state. Single-cell multi-omics technologies will now be applied to gain novel insights into the cellular identity, heterogeneity, and molecular mechanism of therapy-resistant LSCs in AML. Keywords: Acute myeloid leukemia, Minimal residual disease (MRD), Chemoresistance