Tracking Bone Marrow Regeneration after CAR-T Cell Therapy in Myeloma at Spatial Single-Cell Resolution Reveals Determinants of Prolonged Cytopenia

Tracking bone marrow regeneration after CAR-T cell therapy in myeloma at spatial single-cell resolution reveals determinants of prolonged cytopenia

Background: Anti-BCMA CAR-T cell therapy is a widely used treatment strategy of relapsed/refractory multiple myeloma. Despite its notable efficacy and substantial overall response rate, this treatment is frequently accompanied by severe undesirable side effects. In particular, the pathophysiology of impaired bone marrow (BM) regeneration post therapy, which is critical for patient recovery, is not understood.

Methods:

In this study, we analyzed BM biopsies of anti-BCMA CAR-T cell treated patients (n=16) at spatial single-cell resolution by combining single-cell RNA-sequencing with spatial transcriptomics using 10x Xenium in situ sequencing. To optimize cell type coverage of the single-cell RNA-sequencing data, we did negative selection of the highly abundant cell types to prevent under-representation of the rare but nonetheless crucial cell types, utilizing magnetic separation method involving magnetic-beads-conjugated antibodies and subsequent FACS sorting to enrich for the non-hematopoietic cells and hematopoietic stem and progenitor cells (HSPCs). Compared to conventional aspirates, the use of BM biopsies has immense advantage of recovering more cell types and providing a more physiologically representative view of the BM microenvironment.

To perform spatial transcriptomics, we selected a panel of 427 marker genes from the single-cell RNA-sequencing dataset to target all the subpopulations of the various cell types. The Xenium platform then provides single-cell resolution illustration of these marker gene expressions. By integrating with the single-cell RNA-sequencing data, we were able to therefore annotate all the cell subpopulations. This provided a unique opportunity to comprehensively characterize the niche diversity and organization within the BM biopsy.

Results: Our data recovered 20 different cell types including all major hematopoietic and stromal cells of the BM microenvironment, and revealed substantial inter-patient heterogeneity of cell states and BM architecture, which was associated with hematopoietic regeneration post therapy. Focusing on prolonged cytopenia, one of the most common and potential fatal side effects of the therapy, our integrated spatial analysis strategy revealed an inflammatory mesenchymal cell population, which was over-represented in cytopenic patients. Ligand-receptor interaction analysis together with spatial niche information derived from our in-situ sequencing data suggests that inflamed mesenchymal cells may serve as a signaling hub to relay inflammatory signaling to the myeloid cells within the BM. Using in vitro culture assays with primary human cells, we demonstrate that in the presence of mesenchymal cells, T cell-secreted cytokines activate monocytes and neutrophils more potently, suggesting the presence of an inflammatory feedforward loop. The resulting sustained inflammatory state leads to diminished functional output of the HSPCs as a putative cause of prolonged cytopenia. Gene set enrichment analysis indicates a prevalent senescence signature in the HSCs from cytopenic patients, suggesting that HSPC malfunction might associate to their senescence state.

Conclusions: Our study unravels a potential mechanism to maintain prolonged inflammation in the BM microenvironment and suggests that this inflammatory state may underlie prolonged cytopenia by inducing senescence of HSPCs.