Background The tumor microenvironment (TME) of non-small cell lung cancer (NSCLC) undergoing immune checkpoint inhibitor (ICI) treatment is poorly understood. Spatially-resolved single-cell analyses are necessary to identify co-enriched cellular interactions and expression programs that signal potential resistance to ICI treatment. Here, we developed a spatial multi-omics analysis platform that integrates spatial proteomics, spatial transcriptomics, and bulk RNA-seq on paired tissue sections. Using this workflow, we were able to analyze NSCLC tumor tissue samples, providing insights into key signatures of response to immunotherapy.

Methods Spatial data was generated from a tissue microarray (TMA) consisting of a cohort of treatment-naive NSCLC patients (n=42) that were eventually treated with second-line ICI treatment and had clinical follow-up information available. Clinical information regarding this cohort is available in the attached clinical data (table 1).

We developed a spatial multiomic solution that integrates spatial proteomic and spatial transcriptomic data. We used Akoya’s PhenoCycler-Fusion multiplex immunofluorescence imaging technology (CODEX) to generate initial immune profiling of NSCLC tumor samples and NanoString’s GeoMX Digital Spatial Profiler to generate spatial transcriptomic data from the same patient sample. Data analysis was performed, including cell segmentation, quality-control filtering, phenotype assignment, and pairwise interaction analysis.

Results In the spatial proteomic analysis performed with CODEX, we showed that in a pairwise interaction analysis focusing on Responders (R) vs. Non-Responders (NR) treated with Nivolumab, NR displayed increased interactions between CD68+ macrophages and CD8+ T-cells compared to R. In our follow-up analysis with GeoMX, we found enrichment of several pathways, including macrophage migration inhibitory factor, interferon signaling, leukocyte cell-cell adhesion, TCR signaling pathways, and interleukin 6, correlated with the increased CD68+ macrophages and CD8+ T-cells interactions. Whereas several anti-inflammatory cytokines and chemokines, including IL-10, TGF-beta, CCL17, CCL18, CCL22, and CCL24, in the GeoMX analysis.

Conclusions In this study, we co-analyzed spatial proteomics and transcriptomics data from TMA sections of NSCLC tissues. This allowed us to correlate analysis results across assays. Both immune signaling pathways and multiple pro-inflammatory pathways were correlated with interactions between CD8+ T-cells and CD68+ macrophages. Surprisingly, T-cell and macrophage signaling motifs are more significantly associated with patients’ non-responsiveness to Nivolumab than anti-inflammatory signatures typically associated with an immunosuppressive TME.

Acknowledgements Pasteur Hospital, 30 Voie Romaine, 06000 Nice, France, for NSCLC tumor samples.