P-203: Ability to perform spatial transcriptomics in FFPE decalcified bone marrow samples of patients with precursor myeloma

Introduction: Spatial transcriptomics (ST) provides unprecedented insights into complex multicellular architecture by mapping the expression of hundreds to thousands of genes across spatial coordinates. While ST unravel spatial immune contextures to predict disease progression and (immune)therapy response, its application in bone marrow (BM)-derived hematologic cancers has been limited by RNA degradation resulting from required decalcification processes. To evaluate the feasibility of current technologies, we performed ST on 14 BM formalin-fixed, paraffin-embedded (FFPE) samples from healthy individuals and myeloma precursor patients, comparing four ST platforms (CosMx, Xenium, Visium CytAssist, Visium HD) and two decalcification methods (EDTA, RapidCal Immuno).

Methods: Samples were collected from participants with MGUS, SMM, or MM upon local IRB approval. Eight archived FFPE sections were decalcified with RapidCal Immuno (Statlab). Four of these were analyzed with the CosMx platform (1000-plex panel), three sections from the same FFPE cores were analyzed with the Xenium platform (377-plex panel), and one independent section with Visium HD. Six additional BM samples were EDTA-decalcified, four of which were analyzed with Xenium, one with Visium CytAssist, and one with Visium HD.

Results: Matching FFPE blocks enabled direct comparison of CosMx and Xenium platforms. Samples analyzed with CosMx exhibited no tissue detachment, yielding 8,448-12,100 cells each with a median of 72-94 genes per cell. While one sample detached during the Xenium run, the remaining samples yielded 5,027-28,334 cells with a median of 11-17 genes per cell. Despite these differences, both technologies detected all major hematopoietic and stroma-derived cell types after Leiden clustering. Xenium identified minor cell populations (e.g., T cells, megakaryocytes) more robustly than CosMx. Interestingly, we detected CCND1^high plasma cells in a patient with t(11;14), unlocking the ability to map distinct myeloma cell clones within their immune microenvironment.

Next, we compared the effect of decalcification on ST quality. In Xenium, EDTA increased the signal detection five to ten-fold compared to standard methods. Applying Visium CytAssist to one EDTA-decalcified sample resulted in a median of 1014 genes (maximum 5361 genes) per 50 μm-wide spot. Moreover, in ongoing analyses of two samples analyzed by Visium HD as the most advanced ST capturing ~18,000 genes across single cells, we will evaluate if Visium HD has the potential to be an ultimate solution for BM samples.

Conclusions: Here, we demonstrate that ST is feasible on decalcified BM samples, while minor technical challenges remain due to tissue detachment and acid-based decalcification. However, despite these limitations, ST technologies can be applied to archival decalcified samples. In the future, improved processing and decalcification strategies can enhance the accessibility of BM samples for advancements in spatial biology.