P447: BCL2-INHIBITION TARGETS LEUKEMIC STEM CELLS IN ACUTE MYELOID LEUKEMIA INDEPENDENT OF MONOCYTIC BLAST POPULATIONS

Background: Treatment with Hypomethylating agents (HMA) in combination with the BCL-2 inhibitor Venetoclax (VEN) has recently become the standard of care for AML patients unsuitable for intensive induction chemotherapy and shows superior results to treatment with Azacitidine (AZA) alone (DiNardo et al., 2020, NEJM). However, upfront resistance and relapse remain major obstacles. It has recently been proposed that monocytic differentiation associated with increased expression of antiapoptotic-protein MCL-1 predicts resistance to AZA/VEN treatment in AML (Pei et al., 2020 Cancer Discovery). However, other studies did not find impaired outcome in patients with monocytic AMLs treated with HMA/VEN (DiNardo et al., 2021, Lancet Oncology; Stein et al., 2021 Blood Advances). Aims: We aimed to understand apoptotic dependencies linked to differentiation states in AML subpopulations in combined with clinical data to understand and treatment effects and resistance upon HMA/VEN therapy. Methods: We performed retrospective analysis of >50 patients at Heidelberg University hospital to identify risk factors of treatment failure. For functional investigation, we studied 24 AML cell lines and generated Xenografts in NSG mice of 12 AML patients to study mature and leukemic stem cell (LSC)-like primary AML subpopulations. We characterized these samples further by RNA sequencing and BCL-2 family level assessment. To functionally validate our findings, we performed ex vivo drug treatment and subpopulation specific BH3 profiling. Results: In our clinical cohort the only factors associated with HMA/VEN resistance were previous myelodysplastic syndrome (MDS)/myeloproliferative neoplasm (MPN) and complex karyotype, but not monocytic differentiation. The CD64+CD11b+, mature blast population making up >50% of leukemic cells in monocytic and <20% in primitive samples showed high levels of resistance to AZA/VEN therapy in both primitive and monocytic leukemias but did not engraft when transplanted into NSG mice, arguing they do not contain LSCs. In contrast, we found immature GPR56+ LSCs to be sensitive to AZA/VEN treatment irrespective whether they were derived from monocytic or primitive primary AMLs. LSCs from monocytic and primitive AMLs initiated disease in NSG mice, highlighting that targeting LSCs is essential to treat AML. Next, we investigated expression of BCL2, MCL1 and BCL-xL levels in the same primary patient samples and observed high MCL1 expression in monocytic AML samples. However, MCL1 expression predominantly derived from the mature population whereas LSCs expressed robust levels of BCL2 but comparatively low levels of MCL1, independent of whether monocytic or primitive AMLs were analyzed. Additionally, transcriptomes of LSCs from either monocytic or primitive samples did not display significant differences. We next performed BH3 profiling and found LSC to be dependent on BCL2 and mature blasts on MCL1 in all samples studied. In line, ex vivo HMA/VEN treatment specifically eliminated LSCs. We further monitored blast clearance in longitudinal patient samples and found robust loss of LSC within hours in patients, while mature blasts persisted for days. Summary/Conclusion: We find that LSCs, independent of the overall mature blast content, are sensitive to HMA/VEN treatment and highlight that genetic risk factors and disease history are relevant predictors of treatment failure.