The Pulse
MD Anderson’s SPACER Maps Immune-Cell Signals Across 37 Datasets
MD Anderson researchers used SPACER to analyze 37 spatial gene-activity datasets and identify signals linked to immune-cell recruitment. The model connected tumor antigen presentation with T-cell presence and found that mucins were linked t
AI.info Team · Written with AI ·
Across 37 spatial gene-activity datasets, MD Anderson researchers report that their SPACER model was four to 21 times more effective than existing methods at identifying biologically meaningful signals linked to cell recruitment in benchmark tasks. The range does not describe a general improvement in prediction accuracy: it applies to the study’s benchmarks for finding recruitment signals. The team announced the findings on October 13, 2026, and the study appeared in Nature Biomedical Engineering.
SPACER is short for Spatial Analysis of Cellular Engagement and Recruitment. It analyzes spatial transcriptomics, which measures gene activity while preserving information about where cells sit within a tissue. Rather than treating each cell as an isolated data point, the model examines local groups of neighboring cells to infer which signals are associated with immune cells entering or avoiding a region.
SPACER looks for local recruitment cues
The researchers applied the model to 37 datasets: 17 high-definition and 20 lower-definition spatial transcriptomics datasets, covering tumors and heart tissue from models of myocarditis. The method uses a multiple-instance learning design to connect gene expression in a neighborhood of resident cells with the presence or absence of another cell type nearby. Its interpretable structure is intended to help researchers examine which genes contribute to the model’s predictions, rather than receiving only a forecast from an opaque system.
That distinction matters because spatial maps can show where immune cells are, but do not by themselves explain why those cells gather in one area and not another. SPACER’s estimates are associations inferred from tissue data; they do not show that a specific gene alone caused a cell to move. The authors tested some of their findings against other experimental data, including immunopeptidomics, spatial T-cell receptor sequencing and single-cell sequencing.
Antigen presentation drew attention; mucins were linked to exclusion
In tumor datasets, SPACER linked genes involved in antigen presentation with greater T-cell presence. Antigen presentation is how cells display protein fragments for immune cells to inspect. The paper also reports that genes encoding immunogenic peptides were among signals associated with T-cell recruitment, while recruitment patterns for B cells and macrophages involved different molecular programs.
“On the other hand, expression of mucins in the tumor cells was found to repel T cell localization.”
Jia Yao et al., authors of the SPACER study
Mucins are large, sugar-coated proteins that many cancers produce. The researchers found that higher mucin levels were associated with poorer responses to immune checkpoint inhibitors and poorer patient outcomes across several melanoma and gastric cancer datasets. The results point to a possible physical barrier, but the associations do not establish that mucins directly caused treatment failure or that reducing them would improve outcomes.
The myocarditis result broadens the test
Applying SPACER to preclinical data on immunotherapy-related myocarditis, or heart inflammation, highlighted areas and molecular signals associated with T-cell recruitment. The model suggested that CD4-positive T cells may have a larger role in severe inflammation than their numbers alone would imply. That result extends the method beyond tumors, but it remains a finding from preclinical data, not evidence that SPACER can diagnose or guide treatment for patients with myocarditis.
The team says the approach could help researchers identify candidate biomarkers and targets for further study, including in autoimmune disease, tissue repair and wound healing. The study does not establish a clinical test or show that using SPACER to choose cancer treatment improves patient outcomes. Its immediate contribution is a way to turn spatial tissue data into testable hypotheses about why immune cells gather in some places and are scarce in others.