Fig 2: CIVO + GeoMx + CosMx Workflow. A, Drugs are delivered intratumorally in a columnar fashion to spatially defined positions in head and neck tumors using the CIVO microinjector. The position of each site of drug injection is identified and tracked from the initial injection procedure, through surgical resection, and ultimately spatial profiling by co-injection of fluorescent-tracking microspheres (FTM). Following incubation for 4 to 96 hours in the living intact tumor, the tumor is surgically resected and cross-sectioned perpendicular to the injection column plane. Tumors exhibiting distinct injection sites, as visualized by the presence of the FTM (fluorescent green in appearance) and verified as located in tumor tissue are processed as formalin-fixed, paraffin-embedded (FFPE) sections for downstream analysis via spatial profiling technologies and conventional IHC/ISH assays. B, Regions from FFPE sections containing trackable sites of drug target engagement are analyzed using GeoMx Digital Spatial Profiling RNA assays. Identification of localized regions of target engagement was enhanced by staining with the MIL113 antibody that specifically recognizes formation of subasumstat-SUMO adducts (yellow) and verified for inhibition of the SUMO pathway using an antibody specific for SUMO 2+3. To establish drug-induced responses across a compositionally diverse TME, multiple regions of interest (ROI) are placed at sites within and outside of graded areas of MIL113 staining (from high to low surrounding the FTM) for transcriptional profiling using probe sets such as the Cancer Transcriptome Atlas (1,800+ gene targets). C, Differential gene expression analysis was performed to determine drug-specific gene expression signatures across the patient tumor sample set. Similarly, regions within CIVO microdosed tumors can be analyzed using spatial molecular imaging such as Nanostring's CosMx technology. Gene expression at the cellular level is determined through iterative ISH and imaging for up to 1,000 gene targets. Segmented cells are assigned to cell types and cell type–specific expression levels can be compared at the cellular level within the spatial context of the intact TME. Drug-induced changes in transcript expression, detected by either DSP or SMI, are then validated at each spatially defined position of drug exposure within the TME by conventional IHC or ISH.

Fig 3: SiGra accurately identifies spatial domains in the single-cell spatial profiles of NanoString CosMx SMI.a Spatial regions of ground truth and those detected by different methods, including SiGra, BayesSpace, and Seurat. b Boxplot of the adjusted Rand index (ARI) scores of six methods in all 20 FOVs of lung cancer tissue. In the boxplot, the centreline, box limits and whiskers denote the median, upper and lower quartiles, and 1.5× interquartile range, respectively. Source data are provided as a Source Data file. c Spatial regions of two FOVs detected by different methods, including SiGra, BayesSpace, and Seurat.

Fig 4: Histological hallmarks of glioblastoma characterized by single-cell spatial transcriptomics.A Illustration of the spatial transcriptomics workflow. Archived formalin-fixed paraffin-embedded tissue sections from resected patient tumors (n = 5 patients) were screened for histological hallmarks of glioblastoma tumors. Spatial Molecular Imaging (CosMx, NanoString Technologies) was performed on all tumors using both gene transcripts and cell masks for cell profiling. B Uniform Manifold Approximation and Projection (UMAP) of all cells with individual plots for each patient sample using the same UMAP coordinates for each plot. C Mean expression of modified Neftel signatures across all malignant cells. D Dotplot of differential expression testing for all cell types. Color represents the average expression and dot size is the percentage of cells expressing the gene. E Correlation matrix of the distributions of cell types and malignant states across all patient regions of interest. F Composition of cells within our study and other single-cell RNA sequencing studies. G Heatmap of cell composition in each study, with values scaled and centered for each cell type. H Proportions of selected cell types within our data (red) (n = 1) and other studies (blue) (n = 16). Box plots show interquartile range (IQR), with the middle line indicating the median, and whiskers representing 1.5-fold IQR, and all individual points are shown. Source data are provided as a Source Data file. Panel A Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig 5: Demonstration of the two-stage approach of SMART. A An example field of view of the single-cell Nanostring CosMx data in human non-small cell lung cancer. B The SMART-predicted cell type composition of the example field of view. C A heatmap showing the Pearson correlation coefficients between the predicted and the GT cell type proportions of each cell type. D A comparison between the one-stage approach and the two-stage approach on predicting T cell subtypes (top) and dendritic cell subtypes (bottom)