Acute Myeloid Leukemia (AML) in a 3D Bone Marrow Niche Showed High Performance for in Vitro and In Vivo Drug Screenings

Chemotherapy still remains the pillar of treatment of children with AML, a disease in which refinements in diagnostic approaches, minimal residual disease monitoring, and patient stratification have resulted into remarkable progresses during the past decade. However, most of the recently tested, novel anti-leukemia agents failed during pre-clinical and clinical validation phases, and one main limit in AML field is the inappropriateness of current preclinical models used to study drug efficacy, this jeopardizing the advance of phase II and III clinical trials, especially for children. In light of this consideration, we aimed at creating novel robust in vitro and in vivo approaches to discover or to re-assess alternative treatments to improve the portfolio of agents active in childhood AML. For this purpose, we developed new protocols for long-term 3D-AML cultures to perform more predictable high throughput drug screening in vitro, and, once identified the best compounds, to create new pre-clinical in vivo models. We set up the bone marrow (BM) endosteal niche by using a biomimetic 3D structure, made up of engineered hydroxyapatite and collagen I, where we seeded mesenchymal stromal cells derived either from AML patients (AML-MSCs) or from healthy BM donors (h-MSCs), together with osteoblasts, endothelial cells and finally AML blasts. We studied AML cell proliferation and clonogenicity cultured in 3D. We obtained results from twenty 3D long-term cultures of different primary AML, confirming blast proliferation up to 21 days. Clonogenic potential and immunophenotype preservation of the original AML blasts was also documented. At the same time, we compared AML-MSCs with h-MSCs, finding that AML-MSCs exhibited a higher proliferation rate (40% increase proliferation at 72 and 96 hours, p<0.001), and commitment to osteogenic differentiation, this latter occurring after 7 days with respect to 21 days of h-MSCs (p<0.01). To better characterize AML-MSCs features supporting AML cell growth, we cultured AML-MSCs and h-MSCs in an inflammatory environment (IL1β, IL6 and TNFα), observing that AML-MSCs did not exert anti-inflammatory activity by HUVEC tube formation assay (n=6, p<0.001). This latter finding was supported by a peculiar secretome profile defined by mass-spectrometry revealing factors as SERPINE1, CHI3L1, TIMP1, and PTX3 being differentially secreted. Thus, a drug targeting of AML-MSCs would be desirable, and we performed a screening of 480 compounds. This screening identified 17/480 active compounds capable of reducing AML-MSCs proliferation without toxicity over h-MSCs and AML blasts. We identified one main compound able to selectively reduce AML-MSCs proliferation, that, when combined to novel therapeutic agents for AML, such as the Quizartinib, I-BET inhibitor and Dasatinib in the 3D cultures, showed a synergistic effect (CI=0.5, p<0.05) towards FLT3-ITD, MLL-rearranged or c-KIT mutated AML. We then proceeded with two pilot studies in order to define the use of this 3D-AML cultures in in vivo setting. We implanted these organoids in the back of NSG mice and monitored leukemia engraftment in the scaffolds, as well as AML dissemination in peripheral blood. We documented the niche being vascularized and well organized by immunohistochemical staining for mCD31, hCD45 and hOPN, whereas we had a low rate of success in AML dissemination out of the niche (1 AML out of 14 implanted in 15 months). On the contrary, we observed that AML cells proliferated in the 3D up to 9 months after implantation, suggesting that this system is more suitable for an easy and quick in loco drug testing. Thus, we implanted AML, after being transduced with luciferase, in the 3D niche in NSG mice and monitored luciferase activity during intra-peritoneal drug treatments. We evaluated some new and old compounds known to target leukemia cells and documented a significant reduction of luciferase in the 3D when a drug was active, supporting our 3D scaffolds as a novel useful in vivo system to screen selected drugs before the prioritization of the best one to be used in a pre-clinical setting in PDXs. In conclusion, our data support the possibility to work with long-term 3D cultures of AML in vitro to identify new drugs, and we attribute to AML-MSCs a crucial supportive role to be further considered in in vivo settings for novel combo strategies. Finally, these findings could help test new compounds to be validated in future clinical trials.