(PA-233) An Integrated Spatial and Single-cell Multiomics Platform for Translational Research in Multiple Myeloma

Multiple Myeloma (MM) is a genomically heterogeneous cancer of malignant plasma cells (PC) residing in the bone marrow. Despite recent therapeutic advances, it is incompletely understood why some patient populations remain largely therapy-refractory or relapse, even with targeted immunotherapies. Whilst MM classification based on genomic features and gene expression profiling (GEP) is well-described, the role of spatial interactions between tumour cells and the bone marrow tumour microenvironment (TME) in dictating treatment responses is insufficiently defined. Although spatial heterogeneity in MM has been demonstrated, it is unknown whether spatially and genomically distant clones interact differently within their immune and stromal compartments.

Isolated in situ ‘omics’ assays provide limited information due to biased selection of marker panels or discordance between transcriptional and proteomic profiles. We hypothesize that a combined, systematic approach to integrate multiple layers of information (multiplexed imaging, unbiased LC/MS and targeted proteomics, transcriptomics, genomics) within their spatial bone marrow and single-cell context represents a technology platform to advance our understanding of the interactions between tumour and microenvironment activation or exhaustion phenotypes and how they shape responses to therapy.

Methods:   

In this proof of concept, technology development platform, matched bone marrow trephines and aspirates from newly diagnosed patients (n=4) were subjected to a workflow that encompassed:

1. On formalin fixed, paraffin-embedded sections: a) Xenium spatial transcriptomics (10X Genomics, custom panel), b) multiplexed immunofluorescence or imaging mass cytometry panels to deep phenotype tumour and immune cells, c) laser capture microdissection of regions of interest (ROIs) followed by both liquid chromatography/mass spectrometry (LC/MS) proteomics and d) targeted-region genome panel and shallow whole-genome sequencing.

2. On aspirate samples: single-cell a) LC/MS proteomics, b) multiplexed mass cytometry, c) long-read transcriptomics for transcript isoform identification, and d) bulk CD138+ targeted genomic sequencing.

A data analysis pipeline utilising machine-learning to integrate data is being implemented.

Results:   In this proof of concept study we successfully developed an experimental pipeline that allows disease phenotyping in unprecedented detail, and correlation of in situ tumour clonal heterogeneity with the immune and stromal tumour microenvironment.

Conclusions:   

Our next translational step is the deployment of this workflow to longitudinal clinical trial in situ samples with correlation to single cell data from bone marrow aspirates. Taken together, application of this translational platform will provide new insights into MM pathobiology and both tumour and microenvironment-mediated therapeutic response and resistance mechanisms.