胸膜间皮瘤免疫化疗反应的空间架构

Introduction: Pleural mesothelioma (PM) is a rare and aggressive malignancy originating from the mesothelial cells of the pleural lining, with a well-established etiological association with asbestos exposure. Although immunochemotherapy is being applied as a first-line treatment, most patients fail to respond to the therapy. Previous single-cell RNA sequencing studies on PM have revealed cellular heterogeneity but lack spatial resolution and comparative analyses of immune checkpoint inhibitor (ICI) responses, limiting understanding of tumor microenvironment (TME)-mediated resistance mechanisms. This study aims to comprehensively identify the cellular composition and spatial architecture of the TME, as well as mechanisms responsible for ICI resistance in PM. Methods: Spatial transcriptomics analysis using the Xenium Prime 5K platform was conducted on baseline samples from 21 mesothelioma patients who received neoadjuvant immune checkpoint inhibitors (nivolumab/pembrolizumab), stratified as Responders (CR/PR, n=7) and Non-responders (PD/SD, n=14). Transcript-aware cell segmentation was performed with Proseg, followed by clustering and downstream analysis using Scanpy and Seurat. Copy number variations (CNV) were inferred with insituCNV, and pathway analysis was conducted via PROGENy and KEGG. Conserved spatial niches were identified using the BuildNicheAssay() function in Seurat v5. Results: Spatial transcriptomics identified distinct 36 cell clusters, including mesothelial, immune, stromal, and normal alveolar cells. Mesothelial cells exhibited chromosome 22 deletions, with CNV scores inversely correlating with treatment response. Mesothelial cells of non-responders displayed “Active & Proliferative” phenotype with the high MAPK/PI3K/VEGF activities, whereas responders demonstrated “Stressed & Adaptive” state with elevated p53/TRAIL pathway activities. Responders demonstrated increased B cells, proliferating T cells, CD8 effector memory T cells, and exhausted T cells (CD8 Tex), whereas non-responders showed elevated tumor-associated macrophages. Among the 12 recurrent spatial niches, two tumor-reactive immunity niches of CD8 Tex and B cell-enriched tertiary lymphoid structure-like regions were prominent in responders. Immunosuppressive niches featuring mesothelial cells and SFRP2+ cancer-associated fibroblasts with active TGFbeta pathways predominated in non-responders. Conclusion: This study defines the TME architecture, mesothelial cell functional states, and cell type proportions associated with therapeutic response in PM. Cell type proportions and spatial niche composition featuring tumor-reactive immunity in responders versus TGFbeta-driven immunosuppression in non-responders demonstrate predictive value for treatment outcomes and establish novel therapeutic targets to enhance immunotherapy responses.