High Throughput and Long-Term Human Bone Marrow Perivascular Organoids to Study Tumor Microenvironment and Drug Sensitivity in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is a blood cancer originating in the bone marrow (BM), establishing a tumor microenvironment (TME) and spreading systemically. Current therapeutic strategies are hindered by the lack of representative in vitro models that can accurately replicate bone marrow architecture and TME, sustain long-term cultures, and be scalable to meet the demands of high-throughput experimentation. Existing organoid platforms often involve seeding AML blasts into supernatants, requiring subsequent homing and engraftment processes, which undermines the crucial three-dimensional architecture necessary for accurate disease modeling. Similarly, iPSC-derived bone marrow models suffer from limitations in tissue maturation, functionality, and reproducibility. Here, we present the first human bone marrow perivascular organoid capable of co-seeding AML cells, human endothelial cells (EC), and mesenchymal stem cells (MSCs) supporting long-term culture and high throughput capabilities, and offering a promising tool for advancing AML research by enhancing high throughput drug screening and facilitating the testing of novel therapeutic approaches.

We developed a protocol to generate BM organoids to model AML. Human umbilical vein endothelial cells (HUVECs), MSCs, and AML cell lines were combined with Matrigel and Methocellulose and seeded into 384-well plates. Control organoids lacking AML cells were seeded in parallel.

To evaluate stromal compositions, 400 AML cells (Kg1-a) were seeded at EC:MSC ratios of 25/75 (25%), 50/50 (50%), and 75/25 (75%). Organoids successfully formed across all ratios, demonstrating robustness in modeling distinct bone marrow sections. Varying AML cell concentrations (200, 400, and 800 cells per organoid) at these ratios showed proportional disease burden in 25% and 50% stromal compositions by CD45+ cells (n=3), while 75% reached saturation at 400 cells per organoid. This customization capability allows modeling of different disease stages. ATP luminescence confirmed viable cell counts stabilizing by day 14 in 50% stromal organoids and by day 28 in 25% and 75% compositions (N=3, n=4). Long-term viability was indicated by >50% signal retention on day 60. Cellular composition analysis through flow cytometry showed stable proportions of CD45+ AML, CD90+ MSC, and CD31+ EC in 50% and 25% stromal organoids from day 14, with 75% stromal organoids initially at ~40% AML, ~40% MSC, ~20% EC shifting to ~75% AML, ~5% MSC, ~20% EC by day 60. Extension to NOMO-1, Kasumi-1, and OCI-AML5 cell lines highlighted cell line-specific kinetics, affirming our technology’s robustness.

Confocal microscopy analysis of whole organoids showed bone marrow-like architecture across all stromal ratios, characterized by EC forming sinusoidal-like interconnected vasculature networks along with MSCs. Perivascular alpha-smooth muscle cells and fibronectin were observed within the endogenous extracellular matrix. Cancer cells formed proliferative niches within the vasculature and in association with MSCs. Comparison to no-AML controls revealed upregulation of AML characteristic adhesion molecules VCAM1 and E-selectin, indicating the inflammatory microenvironment generated in situ.

Organoids were utilized to assess the dose response to cytarabine treatment, revealing that Kg1-a cells exhibited significantly lower levels of apoptosis within the 1 to 100 μM concentration range (p < 0.0001) compared to cells cultured in 2D, suggesting a potential chemoprotective effect conferred by the 3D culture environment.

In conclusion, we successfully developed a bone marrow organoid model capable to simulate varying stromal compositions for studying AML dynamics. AML cell concentrations can be adjusted to reflect different disease stages and observed long-term viability, stable cellular compositions, and an inflammatory microenvironment characteristic of AML TME. The organoids showed a significant chemoprotective effect with cytarabine treatment compared to 2D culture, highlighting their potential for drug sensitivity testing in AML therapy research.