Technology Spotlight
Spatial Transcriptomics in Hematology: A New Frontier Free

Over the past 50 years, rapid advances in high-throughput “omics” technologies, including the application of unbiased single-cell transcriptomic, genomic, and epigenomic assays, have transformed our understanding of hematology in health and disease. As an early adopter of molecular technology, driven in part by the accessibility of hematologic tissues in a liquid cell state, hematology has frequently been the first clinical specialty to benefit from technological advances. Spatial transcriptomics (ST) represents the latest evolution in the molecular toolkit. With a plethora of abstracts presented at the 2024 ASH Annual Meeting, this is a timely moment to review the principles, methods, limitations, and clinical potential of ST in hematology.

Spatial analysis has long been routine in hematopathology, conventionally by hematoxylin and eosin staining and multiplexed immunofluorescent or immunohistochemical panels. These techniques remain central to current World Health Organization diagnostic criteria — for instance, in identifying aberrant megakaryocyte clustering in myeloproliferative neoplasms (MPNs) or the abnormal localization of immature progenitors in myelodysplastic syndromes (MDS).¹  The role of the microenvironment in promoting hematologic disease and shaping response to therapy is also well-established, exemplified by the inflammatory microenvironment in Hodgkin lymphoma or the fibrosis-promoting stromal niche in myelofibrosis.^2,3  While richly annotated single-cell atlases now map hematologic health and disease with high resolution,^4,5  clinical diagnostics still rely heavily on limited plex morphology-based methods that are subject to inter-observer variability.⁶ 

ST offers a path to bridge this gap. By profiling thousands of gene expression products in situ in intact tissue architecture, ST overcomes the multiplexing, specificity, and steric hindrance limitations of conventional approaches. Multiple ST platforms are in ongoing development, with ever-increasing multiplexing capacity, transcriptomic coverage, and sensitivity. The concurrent evolution of a suite of novel computational tools enables the integration of these large and complex datasets with other “omics” and histopathological data. This integrative approach compensates for the limited number of features measured by ST, permitting high-resolution, spatially resolved profiling at unprecedented scale and accuracy, with the potential of diverse clinical applications in hematology (Figure).

Key differentiating features between the array of commercially available platforms include the nature of the readout (imaging versus sequencing), resolution (subcellular to “cellular microniche”), coverage (full transcriptome to targeted panels), sensitivity, compatibility with formalin-fixed paraffin-embedded tissue, and ease of integration with standard histological techniques. The unit for ST data analysis is a “multicellular neighborhood,” enabling the microenvironmental context to be interrogated (Figure).^7,8 

A key choice when selecting a platform is prioritizing gene throughput or sensitivity. Sequencing-based ST methods (e.g., 10X Visium, Stereo-seq, Curio Seeker, and DBiT-seq) rely on barcoded slides to spatially index messenger ribonucleic acid (RNA) capture.^9-11  Tissue placed on the slide initiates contact with a barcoded primer, which captures the poly(A) tail in an unbiased manner, enabling unbiased, whole-transcriptome coverage, including isoform and mutation detection, at cost of sensitivity. Most current assays generate around 100 unique transcripts per micrometer — and coverage — generally in the order of 10 mm,²  although the assays are rapidly evolving.

Imaging-based techniques include in situ hybridization (ISH) of fluorescent probes binding to a panel of known targets with serial imaging (e.g., Xenium, CosMx, and MERFISH) or in situ sequencing (ISS) technologies (e.g., HybISS, STARmap, and ExSeq) — where RNA molecules are directly sequenced in fixed tissues by use of hybridizing “padlock probes” that undergo circularization, driving rolling circle amplification for localized amplicon formation.^12-17  Localized amplicons undergo sequencing in situ by ligation or synthesis, with imaging occurring after each cycle. ISH and most ISS approaches (bar ExSeq and FISSEQ) rely on a priori knowledge of targets of interest but can provide subcellular resolution of whole samples, limited only by the optical diffraction limit (around 100 nm). Predesigned and variably customizable panels target 500 to 1,000 genes dependent on technology — although larger, more flexible panels are in development. The advantages of each platform have recently been comprehensively reviewed.^18-20 

Several barriers limit the translation of ST from a research innovation to clinical adoption. These include their high cost, data volume, and storage demands, as well as the need for specialized expertise in both wet lab procedures and bioinformatic analysis. Methodological standardization remains an unmet need: key variables such as sample processing, data normalization, and benchmarking of data quality across platforms lack consensus. For instance, while single-cell RNA sequencing uses standard reference experiments — such as mixed-species samples (e.g., murine and human cells) to benchmark assay performance — ST lacks equivalent standards, limiting cross-platform comparability. Moreover, intratumoral heterogeneity and sampling bias are persistent challenges across all tissue-based assays.

Nonetheless, at ASH 2024, numerous studies showcased the growing potential of ST to address unmet clinical needs in hematology, highlighting that the spatial arrangement of cells affects disease biology and therapeutic response. In MDS, ST revealed that remodeling of the adaptive immune microenvironment is intrinsic to MDS development.²¹  In juvenile myelomonocytic leukemia (JMML), a disease with a dire prognosis at relapse, spatial transcriptomics highlighted previously unrecognized cell-to-cell interactions of conventional dendritic cells and JMML cells at disease recurrence.²²  In acute myeloid leukemia, ST highlighted the role of undifferentiated stromal cells in maintaining leukemic stem cells, implicating the bone marrow microenvironment as a therapeutic target.²³  Meanwhile, spatial mapping in MPNs has generated novel insights into tissue architecture, supporting a growing role for artificial intelligence in spatial data interpretation.²⁴  Beyond disease characterization, ST is also being applied to the evaluation of treatment responses and resistance mechanisms, highlighting that clinically actionable information can be obtained from this approach.^25,26 

The trajectory of whole-genome sequencing offers a useful precedent for how complex technologies can transition into routine clinical workflows.²⁷  As spatial “omics” continue to evolve — now including modalities such as histone modification mapping and profiling of open chromatin^28,29  — ST is poised to become a core component of multidimensional disease profiling. The future of hematology will be mapped not only in genes, but in space.