A potential tissue-based biomarker in gut GVHD

Patel et al used spatial transcriptomic analysis of gastrointestinal (GI) tract biopsies at the onset of graft-versus-host disease (GVHD) to determine that high expression of ubiquitin-specific protease 17 (USP17L) RNA by colonic hematopoietic cells represents a potential prognostic biomarker for GVHD.¹ 

Severe acute GVHD of the lower GI tract causes high mortality and morbidity after allogeneic stem cell transplantation (SCT).² In particular, steroid-refractory gut GVHD portends a dismal outcome. Because of the nature of its interface with environmental microorganisms, the GI tract is immunologically active at steady state. Following allogeneic SCT, inflammatory signals in the GI tract are amplified and alloantigen presentation to donor T cells is enhanced locally, promoting donor T-cell activation and severe GVHD.³ The mechanisms governing steroid responsiveness of gut GVHD during immunosuppressive treatment remain largely unknown. Innovative high-dimensional technology that retains spatial integrity may be useful to interrogate this question. Spatial transcriptomics is a rapidly progressing technology and merges transcriptome profiling and tissue imaging.⁴ In the current study, tissue was interrogated with barcoded RNA probes, and selected areas of formalin-fixed, paraffin-embedded tissue (eg, immune cell aggregations) were analyzed to configure spatial transcriptomes.

Patel et al conducted spatial transcriptomic analysis on colonic biopsy samples from 32 patients at the time of acute GI tract GVHD diagnosis using the Nanostring GeoMx Digital Spatial Profiler,⁵ one of several currently available commercial platforms. They prepared 2 serial sections from each paraffin block: one section for hematoxylin and eosin (H&E) staining and the other for the GeoMx platform. The H&E-stained sample was used for pathologic grading. The latter was stained with the large panel of RNA probes that are tagged by UV light-cleavable indexing oligonucleotides (gene barcodes) to detect >18 000 genes, and fluorophore-conjugated lineage-specific antibodies to visualize cell subsets, such as hematopoietic/immune cells (anti-CD45), epithelial cells (anti-pancytokeratin), and endothelial cells (anti-CD31). On the basis of this, they then selected lineage-enriched areas (termed areas of interest [AOIs]¹). Then, photocleavable gene barcodes bonded to RNA probes were released from an AOI by UV light projection and collected into a single well of a 96-well plate. This process was repeated for every defined AOI. UV-cleaved barcodes were quantified with next-generation sequencing, and the RNA expression data were mapped to a spatial reference (AOI location on fluorescence image).¹ 

In comparison to single-cell RNA sequencing (scRNAseq) of dissociated tissues, spatial transcriptomics enables the analysis of gene expression data of cells and their neighboring microenvironment. If small but specific cell populations play a significant role only in juxtaposition to other cell lineages, this will likely be missed by scRNAseq of the same dissociated tissue. Because available discovery spatial transcriptomics has not yet reached true single-cell resolution, the ability to address cellar interactions is somewhat limited. New targeted RNA probe-based platforms, however, do now offer subcellular resolution and may well provide more informative data. In conjunction with pathologic assessment and clinical outcome information, the authors of the current study could distinguish AOI clusters in which gene signatures were associated with high-grade pathologic changes and steroid resistance. First, they generated gene expression heat maps of AOIs for each lineage (immune cells [ICs], epithelial cells, and vascular endothelial cells), and separated clear 2 clusters therein (eg, IC1 vs IC2). Because paired serial sections were assessed by pathologists, individual AOIs and their gene expression data could be associated with local pathology. Thus, the authors identified which of 2 clusters (AOIs) included higher-grade pathology. Among immune cell clusters, they found that IC1 was associated with high-grade pathology and steroid resistance, which was associated with high expression of USP17L family genes. Higher expression of USP17L family genes correlated with higher nonrelapse mortality and lower overall survival. Little is known about the USP17L gene family members, which are deubiquitinases. In particular, the current study cannot assign any causative relationship to outcome, nor were results validated in additional patient cohorts. Nevertheless, another deubiquitinase, ovarian tumor deubiquitinase 1, activates and differentiates CD4⁺ T cells to T helper cell (Th) 1, Th17, and Th2 via its ubiquitin-cleavage function, resulting in the exacerbation of GVHD.⁶ Understanding any functional role of the USP17L genes in immune cells may thus help shed light on intestinal GVHD progression and steroid resistance.

Risk-predictive biomarkers for acute GVHD that can guide therapeutic interventions are crucial to the field. The Mount Sinai Acute GVHD International Consortium algorithm probability is based on 2 serum biomarkers, suppressor of tumorigenesis 2 (ST2) and regenerating islet-derived 3α (REG3α), and is useful to predict prognosis in systemic acute GVHD.⁷ A tissue-based biomarker for intestinal GVHD has not yet been developed, although RNA sequencing of biopsies has been reported in prior studies.⁸ As opposed to GeoMx analysis, simple USP17L RNA in situ hybridization in isolation did not predict GVHD outcome in the current study. Thus, future clinical trials will likely require additional markers or improved staining protocols to detect USP17L RNA or protein in isolation to verify USP17L as a biomarker on intestinal biopsies.

Conflict-of-interest disclosure: The author declares no competing financial interests.