Imaging Mass Cytometry Reveals the Spatial Architecture of Myelodysplastic Syndromes and Secondary Acute Myeloid Leukemias

Histologic review of bone marrow trephine biopsies is a central component of the diagnostic and treatment response evaluation of hematologic malignancies. Well-validated antibody reagents are routinely used for immunohistochemistry of these samples to provide additional insight into abnormal antigen expression. However, current immunohistochemistry staining protocols are typically limited to only one or two markers simultaneously. Dysplastic changes in cellular morphology and dyssynchronous expression of lineage markers are common features of myelodysplastic syndromes, myeloproliferative neoplasms and secondary acute leukemias.

We have integrated the use of multiple diagnostic validated antibody clones with additional antibodies for hematologic lineages and structural proteins to create a 30-marker panel for imaging mass cytometry (IMC). Antibodies included in this panel identify myeloid, lymphoid, erythroid, macrophage, vascular, megakaryocyte and stromal markers as well as markers of cellular proliferation and apoptosis. Through conjugation to elemental metal tags, the entire panel is stained simultaneously on the tissue sample, then acquired by time-of-flight mass spectrometry on a Hyperion instrument (Fluidigm). Antibody staining concentrations and antigen retrieval conditions were optimized for formalin-fixed paraffin-embedded (FFPE) bone marrow to obtain consistent staining for all markers on the panel. Redundant markers for cell populations were selected to provide further internal validation of the observed staining patterns. After data acquisition, cell segmentation algorithms using CellProfiler and ilastik were applied to quantify marker expression in single cells and Phenograph in HistoCAT was used for cell population clustering. Cluster identities for all cells are associated with the original image location in order to plot the spatial arrangement of populations.

Using this highly multiplexed panel, we have imaged sets of bone marrow specimens from patients with normal bone marrow morphology and those with myeloid malignancies. We initially confirmed the staining patterns expected for each antibody patterns of co-expression of lineage markers in normal bone marrow samples. We then extended this panel to examine biopsies from patients with myelodysplastic syndrome, myelofibrosis, and secondary acute myeloid leukemia. We found a clear population of CD71+ CD235a+ erythroid cells with strong expression of the proliferative marker Ki67 located within erythroid islands in normal bone marrow samples and MDS. Cell markers of apoptosis and DNA damage are scattered at low frequency throughout the bone marrow in samples with normal bone marrow morphology, but increased clusters of the DNA damage marker phospho-H2AX are observed in selected cases of myelodysplastic syndromes.

Overall, this IMC imaging approach is able to extend the current clinical immunostaining for myeloid malignancies by identifying all major bone marrow cell populations. Through highly multiplexed analysis of bone marrow cell populations, the spatial architecture of cell populations and stromal structures can be elucidated, including erythroid islands, lymphoid aggregates and changes in vascular structures with increasing severity of myelofibrosis. In ongoing studies, the development of these imaging techniques for analysis of archived FFPE bone marrow samples is being applied to translational research on hematologic diseases.