Patient-Derived Acute Myeloid Leukemia (AML) Ex Vivo platform for Evaluation of Novel Molecular Targeted Therapeutic Agents

Acute myelogenous leukemia (AML) is the most common acute leukemia in adults. A hematologic cancer, the disease is highly heterogeneous, with multiple subtypes. Despite advances in treatment, the long-term survival for AML remains poor and the development of novel treatments is an unmet need. Due to the highly divergent subtypes and mutation profiles in AML, the use of patient-derived models may improve drug discovery and development. To address this, we have established a short-term culture system that supports the growth of primary AML cells ex vivo to permit the evaluation and/or screening of candidate therapeutic agents. Our AML bank is comprised of patient-derived specimens across a range of subtypes (which includes M1, M2, M4, M5, NOS and others), and includes models with common mutations in FLT3 (ITD), IDH1/IDH2 and NPM. Primary AML specimens were characterized for common mutations by TruSight sequencing and for surface marker expression by flow cytometry. The ex vivo assay system was evaluated for the ability to support the survival and expansion of primary AML specimens. Among these, the majority of the models showed evidence of proliferation, a few models had no net expansion, and some failed to survive. Extension of the culture period for up to 14 days was feasible, with most models having equal or increased cell numbers by the end of the culture period (8 of 10 models evaluated). All models stably expressed CD33 throughout the assay. To verify the applicability of this system for drug testing, a standard of care agent cytarabine (Ara-C) was assessed for each of the AML models. A small cohort of models with distinct mutations were tested for sensitivity towards small molecule inhibitors venetoclax, gilteritnib, glasdegib, panobinostat and Ibrutinib. Cell growth/viability was assessed using Cell titer-Glo assay. Concentration-dependent responses to Ara-C were observed across multiple models (IC₅₀ 10 nM to 150 nM), indicating a range of relatively sensitive to resistant AML models. Heterogeneous concentration-dependent responses were observed across multiple patient-derived ex vivo models when treated with venetoclax, glasdegib, gilteritnib, ibrutinib and panobinostat, with both sensitive and resistant models identified. A cohort of models were evaluated in vivo for sensitivity to Ara-C. Systemic engraftment of the patient-derived xenograft AML models (into NOG or NOG-EXL mice) was evaluated by flow cytometry. When engrafted was confirmed, AML-bearing mice were randomized into Ara-C or vehicle control groups. Two weeks later, mice were evaluated for the presence of human CD45⁺CD33⁺AML cells. Models that were relatively sensitive to Ara C ex vivo (IC₅₀ < 30 nM) generally showed a greater in vivo response as evidenced by a significant reduction in the mean circulating AML cells versus models with IC₅₀ values >100 nM that had no response to Ara C in vivo. Over thirty distinct patient-derived primary, unpassaged models of AML have been studied both in ex vivo culture and as in vivo systemic models. The diversity in these AML models is reflective of patient diversity, enhancing their utility in the evaluation of novel therapeutic candidates. These data indicate the feasibility of utilization of these primary models, ex vivo as well as in vivo, for drug discovery for AML, from screening to preclinical efficacy modeling.