A defined culture model for ex vivo expansion of primary AML cells preserves leukemic stem cells and clonal architecture for functional and therapeutic studies

Abstract First three authors contributed equally. Introduction: Acute myeloid leukemia (AML) is highly heterogeneous, and its ex vivo culture has been challenging, slowing its study. Here, we present a study of primary AML cases using a serum-free, defined culture system with cytokines/small molecules, maintaining functional leukemia stem cells and clonal architecture in extended culture. Methods: Primary human AML cells (n=30) were collected from blood/bone marrow after informed consent at PSU. Bulk mononuclear cells were cultured in serum-free medium under five conditions (C1–C5) with varying combinations of cytokines and small molecules. C2 consists of SCF, FLT3-L, TPO, IL-6, SR1 and UM729. C3 consists of SCF, IL-3, FLT3-L, G-CSF, SR1 and UM729. Cultured cells were transplanted into irradiated NRG-S mice, in some cases transduced with GFP-luciferase lentivirus. Stemness (CD34/CD123) and differentiation (CD33/CD13/CD11b) were assessed by flow cytometry. AML DNA was Results: Among five conditions (C1-C5) tested in the first six pilot AML cases, C5 showed the highest expansion (389-fold) while C2 had 7-fold and C3 77-fold, but best maintained CD34%, leading to their validation. CD34 expression was used as a surrogate endpoint for “stemness”. Validation in a larger cohort (n=23) confirmed these trends, with mean CD34+ fold expansion of 7.1 (C2), 8.9 (C3), and 0.14 (C5). Cells cultured in C2 showed fold change in CD34+ cells from 0.04 to 75-fold (11 cases <1-fold, 6 >1-fold). C3 ranged from 0.02 to 33-fold (10 <1-fold, 8 >1-fold). CD34% increased in 5/14 (C2) and 6/16 (C3) cases by day 14, but not in C5 - likely due to myeloid differentiation from GM-CSF. AML-MDS (per WHO 5th edition) cases in C2 showed a CD34+ expansion 0.01–5000-fold; 5 cases ≥3-fold expansion, while 6 showed <1-fold. In C3, 6 cases showed ≥3-fold expansion, and 5 showed minimal (<1-fold) growth over 7-14 days. NPM1-mut AML (typically low CD34) showed limited expansion. C2 yielded ~2-fold CD34 increase in 2/6 cases and C3 showed 2-21-fold in 3/6. All inv16 cases (n=4) failed expansion. Myeloid markers (CD11b/13/14) increased across conditions, most in C5 (GM-CSF+). No pure lymphoid (CD3/CD19+) expansion was observed. Overall, different responses of AML cases to distinct culture conditions reflect the disease's heterogeneity. Transplantability of cultured cells (Pts. 329/1265) was assessed by transplanting cells into NRG-S mice immediately after thaw (day 0), and after 4/5 or 8 days of culture in C2, C3, and C5 conditions. Engraftment in bone marrow was 60% in the day 0 group, and higher in day 4 (C2: 51%, C3: 70%) and day 8 (C2: 89%, C3: 77.5%). C5 yielded the lowest engraftment (~20%). Secondary transplants using day 8 cohorts exhibited high engraftment rates (~90%) in C2, moderate (~50%) in C3, and none in C5 (p < 0.001). Accordingly, C2 and C3 were selected for further studies. NGS was performed on 11 AML cases to compare mutational profiles of uncultured (day 0) and C2-cultured (day 14) cells. All variants, including pathogenic drivers and VUS, showed stable MAFs during culture. MAFs changes (day 0 to 14, median%):NPM1 (n=4): 43.4 to 46.3 (p=0.485)FLT3-ITD (n=2): 14.9 to 12.4 (p=0.667)FLT3-TKD (n=2): 34.8 to 32.8 (p=0.666)DNMT3A (n=6): 47.8 to 49.7 (p=0.121)IDH2 (n=3): 47.2 to 49.5 (p=0.1) Other mutations (IDH1, TET2, NRAS/KRAS, KIT, ASXL1, CEBPA, PTPN11, TP53) also showed stable MAFs. Four AML cases were lentiviral-GFP-luciferase transduced in C2 or C3 culture (1-25 days of culture) and engrafted into NRG-S mice, showing progressive growth on imaging. GFP-sorted Pt. 1265-Luc cells established secondary PDXs with similar progression. Sequencing showed stable MAFs in labeled PDX cells. This platform was used to evaluate Azacytidine and Venetoclax in AML. Azacytidine induced apoptosis, reduced stemness, and cell cycle arrest, while Venetoclax caused apoptosis alone. Low-CD34% cases (n=3) showed poor initial growth. After 10-day C2/C3 expansion and CD34+ sorting, Pt.1335 improved in C2/C3 while Pt.1256 did not. Conclusion: This system reproducibly supports the maintenance of many AML cells in extended culture while preserving leukemia stem cells and clonal architecture. Culture conditions are defined but may be improved through optimization. This platform may provide more clinically relevant disease modeling, biological investigations, and preclinical studies.