Spatial biomarkers of CAR-T treatment response in relapsed refractory multiple myeloma

Abstract Autologous chimeric antigen receptor (CAR) T-cell therapy has shown promise for relapsed/refractory multiple myeloma (MM), but disease recurrence remains problematic. Understanding the mechanisms behind CAR-T cell failure is essential for enhancing therapeutic efficacy and achieving durable remissions. To investigate the impact of anti-BCMA CAR-T cell treatment on the bone marrow (BM) environment and identify spatial biomarkers of response, we performed longitudinal spatial transcriptomic profiling using the Xenium platform on 40 MM biopsies, including 17 paired pre- and post-CAR-T samples and 6 singletons. Untreated MM and age-matched donor normal cohorts were also used for comparison. All samples were processed with the Human Multi-Tissue and Cancer panel plus 100 custom genes and analyzed using Xenium software, Seurat, and custom pipelines. Cell types were annotated via label transfer from internal and external single-cell datasets and confirmed by marker gene expression. We profiled over one million cells spanning immune, stromal, erythroid, and megakaryocytic lineages. Plasma cell (PC) abundance in our spatial cohort, as determined by transcript profiles, correlated well with clinical histology-based PC infiltration. PCs localized in either diffuse or dense regions independent of treatment response. In nearly all cases, CAR-T therapy led to both reduced PC proportions and reduced expression of transcripts encoding the targeted BCMA on remaining PCs, demonstrating both effective tumor clearance as well as the opportunity for antigen escape under selective pressure. Within the tumor microenvironment (TME), myeloid cells made up the largest proportion (25-50% of total cells), and upregulated immunosuppressive markers including checkpoint inhibitors and M2-macrophage polarization markers following CAR-T treatment. CD68+ KCNMA1+ M2-like macrophages were observed in direct physical contact with post-treatment PCs. Due to their high relative abundance and ubiquity throughout the BM, myeloid cells have the potential to be a major immunosuppressive influence in the post-treatment milieu. Spatial analysis further revealed two major types of PC neighborhoods: one enriched with B cells and osteoblasts, and another with T cells. In post-treatment samples, CD8+ T cells upregulated transcripts encoding the immune checkpoint molecule TIM-3, further suggesting immune dysfunction that could compromise treatment efficacy. BM architecture largely remained stable before and after treatment. Megakaryocyte-dense regions would be observed in both pre-and post-CAR-T timepoints with similar frequency and cellular organization. The majority of post-CAR-T samples showed a modest increase in stromal cells, although specific stromal subtypes varied across patients. However, the post-CAR-T upregulation of stromal cell-cycle arrest and cellular senescence markers, most strikingly in endothelial cells and even more significant when compared to untreated MM and normal controls, suggest that non-hematopoietic tissues are also negatively affected by treatment exposure. As BM stromal cells provide critical support roles to hematopoietic niches, their cumulative dysfunction in relapsed/refractory myeloma may dampen response to later lines of therapy. Our findings highlight spatially organized microenvironments that may modulate CAR-T efficacy and support the development of improved therapeutic strategies.