Acute Myeloid Leukemia Induced Remodelling of the Human Bone Marrow Niche Predicts Clinical Outcome

The hematopoietic stem cell (HSC) niche consists of different cellular and non-cellular constituents which regulate HSC maintenance and retention in the bone marrow. It has been shown in a number of murine models of myeloid neoplasia how leukemia infiltration alters the HSC niche to reinforce malignancy. Acute myeloid leukemia (AML) is characterized in human by a high relapse rate indicating that leukemia initiating cells are protected by its niche. However, despite our knowledge in murine models little is known about the bone marrow architecture in human and the impact of the leukemic niche on clinical outcome.

In this study, we combined immunohistochemical stainings (IHC) with protein and global gene expression analyses together with clinical data to dissect the human bone marrow architecture in AML and assess its clinical impact. Human bone marrow was collected from AML patients at first diagnosis and matching non-leukemic donors. To evaluate the bone marrow architecture CD271⁺ mesenchymal stem and progenitor cells (MSPCs) were automatically quantified on bone marrow sections. In fact, AML patients showed 1.5-fold increase in bone marrow MSPCs compared to non-leukemic donors (Median (IQR), AML: 5.5% (2.8-9.5), n=36; control: 3.7% (2.1-5.7), n=58; p < 0.01). MSPCs proved to produce reticular fibers, an extracellular matrix protein frequently associated with different malignancies. In AML bone marrow these fibers were also found to be more abundantly expressed (Median (IQR), AML: 3.4% (1.8-4.5), n=37; control: 1.6% (1.1-3.3), n=19; p < 0.05).

Next, to globally assess the gene expression profile of MSPCs in AML bone marrow we performed microarray analyses (Clariom^TM S Human Assay) of freshly isolated uncultured lineage^- CD146⁺ CD271⁺ MSPCs. Strikingly, HSC-regulating genes in particular CXCL12, ANGPT1 and VCAM1 showed lower expression in AML MSPCs which correlated with the degree of hematopoietic failure in AML patients. Along with the increased number of MSPCs, geneset enrichment analysis (GSEA) revealed higher proliferation of MSPCs in AML. In murine models loss of quiescence of MSPCs was previously found to be due to bone marrow sympathetic neuropathy. We therefore measured catecholamines and neurotrophic factor in the bone marrow extracellular fluid of AML patients and non-leukemic donors at first diagnosis. In fact, noradrenalin and brain-derived neurotrophic factor (BDNF) showed a 2-fold (p=0.26) resp. 4-fold (p<0.0001) lower expression in AML bone marrow. Importantly, BDNF is proved to be essential for sympathetic neuron proliferation and differentiation.

In order to get an overview of alterations of canonical pathways in bone marrow MSPCs upon AML infiltration, we applied QIAGEN's Ingenuity^® Pathway Analysis software. Several of the major differently regulated pathways proved to involve differentiation and mineralization of MSPCs. We therefore assessed bone metabolism in AML patients at first diagnosis and quantified serum osteocalcin levels. Notably, AML patients showed 30% lower osteocalcin levels than non-leukemic donors (Median (IQR), AML, 12.15ng/ml (7.53-16.28) n=58; control, 17.2ng/ml (12.5-23.45) n=31; p < 0.05). To evaluate if the deficiency in osteoblast mineralization is specifically due to AML infiltration we performed in vitro co-culture assays. Both MSPCs and an osteoblast-like cell line (SaOS2) showed significant impaired mineralization in presence of certain AML cell lines as well as primary human AML cells, while healthy mononuclear cells did not affect mineralization. Strikingly, this AML-induced defect in osteoblast mineralization proved to be of clinical significance. Patients with low osteocalcin levels (<11ng/ml) showed inferior overall survival with 1-year survival rate of 38.7% while patients with high osteocalcin levels reached 66.8% (n=58; median duration of follow-up 9.7 months).

In summary, we globally characterized the bone marrow architecture in AML patients in comparison to non-leukemic donors and assessed its clinical significance. This increasing understanding of the human AML bone marrow microenvironment might open the window for new niche-targeted therapies to eradicate leukemic stem cells and eventually decrease the high relapse rate in AML.