High-Dimensional Mass Cytometric Analysis of the Bone Marrow Compartment Identifies a Disease-Specific Immune Cell Network in Aplastic Anemia

Idiopathic acquired aplastic anemia (AA) is considered an immune-mediated syndrome of bone marrow failure since approximately 70% of patients respond to immunosuppressive therapy (IST). However, the immune response that underlies the pathogenesis of AA remains incompletely understood. Previous studies have extensively studied the immune cell composition in peripheral blood in AA, but a comprehensive overview of the immune landscape at the site of hematopoietic stem and progenitor cell (HSPC) destruction, the bone marrow, is lacking. We investigated the immune composition of the bone marrow in AA to interrogate which immune populations may be implicated in the pathogenesis of AA.

A mass cytometric analysis of bone marrow aspirates was performed using a 39-marker panel for broad immunophenotyping of all major immune lineages and HSPCs. Cryopreserved bone marrow nuclear cells from adult AA patients (n=7) before start of first-line IST with horse-derived anti-thymocyte globulin (hATG) in combination with ciclosporin and age- and sex-matched healthy donors (HDs, n=7) were studied. 6 of 7 AA patients responded to IST. For 3 IST responders paired bone marrow samples taken at six months after start of IST were also included.

A Hierarchical Stochastic Neighborhood Embedding (HSNE) analysis in Cytosplore revealed that the bone marrow in AA pre-IST features a distinct immune cell composition compared to HDs. At the major immune lineage level, HSPC and myeloid cell depletion (0.6% vs 33% of all immune cells, p<0.001), decreased B-cell frequencies (9% vs 16%, p=0.007), increased CD4⁺ T-cell frequencies (37% vs 28%, p=0.04) and increased CD8⁺ T-cell frequencies (37% vs 20%, p=0.007) were observed. Detailed analysis of these 4 lineages identified 121 phenotypically distinct cell clusters, 45 of which were differentially present (p<0.05) in AA patients pre-IST compared to HDs. To capture the relationships between these 45 immune clusters a Spearman's rank correlation analysis was performed using the cluster frequencies. Two cell networks were identified (Figure 1A), one of which (network AA) was significantly increased in frequency in AA pre-IST (22% vs 5% of all immune cells, p=0.001; Figure 1B). This network consisted of CD16⁺ myeloid cells, CCR6⁺⁺ B-cells, Th17-like CCR6⁺ memory CD4⁺ T-cells and KLRG1⁺ terminally differentiated effector memory (EMRA) CD8⁺ T-cells, compatible with a state of chronic inflammation. Successful treatment with IST largely normalized the immune cell composition at the immune lineage level and strongly reduced the levels of CD16⁺ myeloid cells. However, at six months after start of IST the frequencies of CCR6⁺⁺ B-cells, CCR6⁺ memory CD4⁺ T-cells and KLRG1⁺ EMRA CD8⁺ T-cells were not normalized, which could indicate that ciclosporin is still needed.

This study provides a unique overview of the immune landscape in bone marrow in AA pre- and post-IST at a single-cell level. Previously reported cell subsets could be further defined and possible new players that contribute to the pathogenesis of AA were identified. Our findings show that AA is characterized by high frequencies of immune cells compatible with a state of chronic inflammation and provide a basis for future studies to characterize the function of the identified subsets in the pathogenesis of AA.

No relevant conflicts of interest to declare.