Discover More from Precious FFPE Samples 

Using a CosMx™ SMI prototype, we generated this open-source dataset from eight FFPE non-small-cell lung cancer (NSCLC) tissue samples to highlight the power of spatial molecular imaging. 

The CosMx SMI platform, shipping now, combines superior sensitivity, specificity, robust cell segmentation and high plex panels to enable researchers to perform Cell Atlasing / Cell Typing, Biomarker Discovery and Ligand-Receptor Interaction. ​ 

How it Works

This dataset was generated with a 960-plex CosMx RNA panel run on a CosMx SMI instrument prototype. CosMx RNA Assays go beyond simple cell typing to deconvolution of multiple tissue types and key aspects of cell signaling and cell state, with a focus on ligands and receptors that enable communication between cells. ​ 

Data Summary
Interactive Visualization of Data
Download Data
Data Summary

This data set contains 8 different samples from 5 NSCLC tissues​

  • Sex: 3 female, 2 male
  • Race: White
  • Age: 60+​
  • Grade: G1-G3​
  • Stage: 4 IIIA, 1 IIB

* Note: Non-Small Cell Lung Cancer data was collected on a prototype of CosMx SMI. See Liver FFPE Dataset for data collected on the commercially available CosMx SMI.

Interactive Visualization of Data
Download Data

All Data Files

Individual Datasets

Discover and Map Cell Types​

Data from CosMx SMI create a comprehensive spatial cell atlas of tissues by defining the subcellular expression map of 960 genes, identifying 18 different cell types, including four sub-types of T-cells, and mapping cell types across FFPE tissue sections.

Mapping FFPE sections is easy with the spatial molecular imager
Cell type map of NSCLC tissues. The map displays 135,707 cells across ~20 mm2 tissue section. Color denotes cell type. (A) UMAP projection. (B) Spatially resolved cell-type map. 

Phenotyping of the Tissue Microenvironment

Spatial location of cell types enables characterization of the tissue microenvironment. Based on analysis of each cell’s 200 closest neighbors, CosMx SMI data partitions tissues into distinct neighborhood clusters or niches. Analysis of these neighborhoods reveal how some neighborhoods are specific to a single tissue, while others are shared across all tissues analyzed.

SMI data partitioned FFPE tissue sections into multiple distinct clusters
Organizational map of NSCLC tissue. Color denotes neighborhood cluster, “niche”. (A) UMAP projection. (B) Spatially resolved neighborhood cluster map. 

In-Depth Characterization of the Tumor Microenvironment

CosMx SMI data allows tumor characterizations not just by gene expression, cell types, and spatial features, like niches, but also by the tendency of immune cells to invade into the tumor. As an example, the NSCLC dataset indicates macrophages in Lung 6 tissue are primarily surrounded by non-tumor cells, while mast cells are more likely to be surrounded by tumor cells.

High-level tumor characteristics. (A) Abundance of each cell type within each tissue. (B) Abundance of each niche within each tissue. (C) Invasiveness of each cell type within each tumor. Cells are scored for the frequency of tumor cells within their 100 nearest neighbors. Shapes show the density of this score within each cell type.  

Differential Expression Pattern of a Cell Type Based on Spatial Location

The data can be used to analyze changes in gene expression patterns of any given cell type based on spatial context as shown here with macrophages. For example, within the NSCLC dataset, macrophages express more SPP1 in the tumor interior and tumor-stroma boundary than they do in more immune-rich settings.

NSCLC gene expression of macrophages based on spatial context. (A) Heat map for expression of 960 genes across all niches. (B) Spatial map for expression of SPP1. Color shows SPP1 expression level. 

Ligand Receptor Analysis

CosMx SMI enables analysis for the enrichment of pairwise ligand-receptor expression between interacting cell types. When examining the NSCLC dataset, of these, 16 were significantly enriched at the interface between tumor cells and T cells. Interestingly one of these pairs, the immune checkpoint PD-L1/PD-1 (CD74/PDCD1), exhibited a high degree of variation in Tumor to T cell co-expression. 

NSCLC paired ligand-receptor expression between interacting tumor and T-cells. (A) 100 unique ligand-receptor pairs are included in the CosMx Human Universal Cell Characterization RNA panel. (B) 16 ligand-receptor pairs exhibited spatial significance in lung cancer tumors.  

Detect with Confidence

Single cell in situ data was generated from three serial sections of the “Lung 5” sample. While these three FFPE sections include different cells, they capture the same regions of the tumor microenvironment. The analysis of these three replicates shows high correlation between each section, indicating a high level of reproducibility.

SMI generates highly reproducible data from 3 replicates. For each replicate, total transcripts of each gene were recorded and compared.  

Request a Quote

Contact our helpful experts and we’ll be in touch soon.

Technology Publication:
High-Plex Multiomic Analysis in FFPE Tissue at Single-Cellular and Subcellular Resolution by Spatial Molecular Imaging