{
  "abstract": "Background Lung cancer accounts for 12% of all cancers and 20% of cancer-related deaths. Rapid developments in the field of spatial biology are helping uncover new mechanisms of disease, patient-to-patient variability, and reveal immune-based biomarkers of the tumor microenvironment (TME) predictive of response to immunotherapy. Using a combination of protein- and RNA-based -omic technologies paired with supervised and unsupervised learning approaches, we have characterized the spatial distribution, phenotypes, and functional states of tumor, immune, and stromal cells in the TME of human non-small cell lung cancer (NSCLC).Methods 5 µm formalin-fixed paraffin-embedded (FFPE) tissue sections were prepared from a sample of human NSCLC exhibiting a lepidic growth pattern (T2a, N1, M0). Serial sections underwent: (1) multiplex immunofluorescence (mIF) staining using a 30-antibody panel, (2) multi-omic RNAscope labeling with Lunaphore COMET, and (3) spatial transcriptomics via the 10x Genomics Visium CytAssist platform. Immunofluorescence imaging data were preprocessed using QuPath and MCMICRO and analyzed at the pixel-level using an innovative approach to multiplex tissue image analysis (MORPHÆUS) which uses a class of generative AI models called variational autoencoders (VAEs) to infer cell states and generate pathology-level annotations. Visium data were analyzed using the 10x Genomics Loupe Browser.Results NSCLC was found to contain multiple populations of tumor, immune and stromal cells, with immune cells found in both tertiary lymphoid structures (TLSs) and diffusely scattered throughout the TME. Depending on the type of immune cells, the TLSs could be differentiated between mature and immature TLSs. Cancerous epithelial cells exhibited lower and variable expression of the cell adhesion molecule E-Cadherin relative to normal epithelial cells, suggesting an increased potential to undergo epithelial-to-mesenchymal transition (EMT) and metastasis. Consistent with an EMT-permissive TME, αSMA + (and FAP+) cancer-associated fibroblasts were observed surrounding both immune and tumor cells, with their proportions varying by cell state. Transcriptomic data further highlighted an EMT-permissive TME, revealing substantial TGF-β1 expression and M2-like macrophages interacting with tumor cells.Conclusions These findings reveal an EMT-permissive and immunosuppressive TME in NSCLC, underscoring key opportunities for therapeutic targeting through the modulation of fibroblast activity, and reprogramming macrophage polarization. Future analysis aims to determine the composition of immune cells and checkpoint markers (e.g., regulatory T cells expressing CTLA-4 and TIM3) facilitating such immunoregulation along with changes in the transcriptomic signature within and around tumor cells.",
  "authors": [
    {
      "affiliations": [
        "Nikon Bioimaging Lab, Lexington, MA, USA"
      ],
      "name": "Gaurav N Joshi"
    },
    {
      "affiliations": [
        "Oregan Health and Science University, Portland, OR, USA"
      ],
      "name": "Gregory J Baker"
    },
    {
      "affiliations": [
        "Nikon Bioimaging Lab, Lexington, MA, USA"
      ],
      "name": "Nicholas Sciascia"
    },
    {
      "affiliations": [
        "Nikon Bioimaging Lab, Lexington, MA, USA"
      ],
      "name": "Junya Yoshioka"
    },
    {
      "affiliations": [
        "Nikon Bioimaging Lab, Lexington, MA, USA"
      ],
      "name": "Michael T Yang"
    },
    {
      "affiliations": [
        "Nikon Bioimaging Lab, Lexington, MA, USA"
      ],
      "name": "Fumiki Yanagawa"
    }
  ],
  "title": "1254 Mapping the tumor landscape: spatial multi-omics analysis reveals tumor plasticity and immune suppression in non-small cell lung cancer",
  "uid": "0bd4433f-bfeb-5784-b299-5a1995d94715"
}