AUTHORS
Denis Ohlstrom, MSc, Marina E Michaud, MSc2, Mojtaba Bakhtiari3, Junia Vieira dos Santos, MD, PhD4, William Cain Pilcher Jr, BSc5, Alden Staub6, Sarthak Satpathy7, Katherine E Ferguson8, Sowmitri Karthikeya Siddhartha Mantrala9, Seunghee Kim-Schulze, PhD10, Zhihong Chen, PhD11, Sagar Lonial, MD, FACP12, Melissa Kemp, PhD13, Daniel W. Sherbenou, MD, PhD14, Alessandro Lagana, PhD15, David L. Jaye, MD16, Ajay K Nooka, MD, MPH17, Samir Parekh, MD18, Manoj Bhasin, PhD19
AFFILIATIONS
1 Emory University, Atlanta, Georgia, United States
2 Emory University, Atlanta, Georgia, United States
3 Emory University, Atlanta, Georgia, United States
4 Icahn School of Medicine at Mount Sinai, New York, New York, United States
5 Emory University, Atlanta, Georgia, United States
6 Emory University, Atlanta, Georgia, United States
7 Emory University, Atlanta, Georgia, United States
8 Emory University, Atlanta, Georgia, United States
9 Emory University, Atlanta, Georgia, United States
10 Icahn School of Medicine at Mount Sinai, New York, New York, United States
11 Icahn School of Medicine at Mount Sinai, New York, New York, United States
12 Emory University, Atlanta, Georgia, United States
13 Georgia Institute of Technology, Atlanta, Georgia, United States
14 University of Colorado Anschutz Medical Campus, Denver, Colorado, United States
15 Icahn School of Medicine at Mount Sinai, New York, New York, United States
16 Emory University, Atlanta, Georgia, United States
17 Emory University, Atlanta, Georgia, United States
18 Icahn School of Medicine at Mount Sinai, New York, New York, United States
19 Emory University, Atlanta, Georgia, United States
ABSTRACT
Introduction
Multiple myeloma (MM) progression and treatment resistance are strongly influenced by interactions between malignant and non-malignant cells in the bone marrow (BM) microenvironment. However, existing spatial sequencing approaches have been limited in their ability to resolve interactions underlying MM survival signaling, cell adhesion-mediated drug resistance , and immune dysfunction. To characterize these processes, we developed an integrated spatial transcriptomics (ST) framework to improve spatial resolution, transcriptomic depth, and biological interpretation.
Methods
We performed Visium HD ST on 21 MM BM biopsies and developed an empirical Bayesian framework to identify niche-specific transcriptional programs. Spatial findings were orthogonally validated using Xenium ST data. Spatially derived programs were mapped onto 258 scRNA-seq profiles to extend transcriptomic characterization.
Results
Application of our empirical Bayesian framework to 21 clinically collected BM biopsies confidently assigned 198,717 spatial bins to major malignant and immune (e.g. T, NK) cell populations and resolved recurrent neighborhoods, including MM- and immune-rich niches.
To define pathologic signaling programs, we compared ligand expression across neighborhoods. Plasma cells in MM-dense niches expressed increased non-canonical Wnt (ncWnt) ligands, including WNT5B (Log2Fold Change[L2FC]=0.95, adjusted p [adj.p]=1.78x10-27). Spatial proximity analyses found downstream ncWnt signaling was strongest near ligand-expressing MM cells, with ligand enrichment and spatial dependence validated in Xenium data.
Projection of spatial programs onto MM scRNA-seq profiles revealed preferential activation of the Ca2+/NFAT branch of ncWnt signaling (L2FC=4.10, adj.p=1.78x10-163), driving increased expression of genes linked to MM survival (DEPTOR, L2FC=1.66, adj.p=1.49x10-206) and CAM-DR (CDH2 L2FC=1.83, adj.p=1.05x10-28). Elevated WNT5B was additionally associated with earlier disease progression (HR=1.39, p=1.0x10-5), plausibly associating spatial signaling with TR and early disease progression.
Immune cells within plasma-rich niches exhibited reduced cytotoxicity and increased exhaustion-associated phenotypes. T cells had increased LAG3 (L2FC=1.18, adj.p=3.44x10-19) and enrichment of FOXP3 activity. Integration of immune and MM spatial markers yielded a 15-gene model associated with accelerated progression in independent discovery (HR=2.003, p=4.4x10-5) and validation cohorts (HR=1.484, p=0.0262).
Conclusions
Integrated spatial multi-omics combining Visium HD, Xenium, and scRNA-seq enables high-resolution characterization of niche-specific transcriptional programs in MM. Our approach revealed MM-rich niches characterized by ncWnt signaling, TR-associated transcriptional states, and T-cell suppression. These spatial signaling networks represent potential therapeutic vulnerabilities and demonstrate the prognostic relevance of niche-resolved transcriptional profiling in MM.
DISCLOSURES
Denis Ohlstrom, MSc: No relevant conflicts of interest to declare.