转移性三阴性乳腺癌在免疫治疗期间肿瘤与微环境的共同演化

Cancer cells are constantly shaped by the immune system and the dynamic microenvironment surrounding them. Decoding this co-evolution is essential for uncovering the forces that drive tumor progression and determine therapeutic response. Metastatic triple-negative breast cancer (mTNBC) is the most aggressive subtype of breast cancer with a clear need for better treatment options. While recent immunotherapy trials for mTNBC have shown promise in some PD-L1-positive tumors, the reasons for limited and inconsistent responses remain unclear. To investigate these dynamics, we established a multimodal, longitudinal dataset that combines spatial proteomics (MIBI, 40-plex protein) and spatial transcriptomics (CosMx, 6k-plex) from patients with mTNBC enrolled in the phase II TONIC trial in which nivolumab was given to patients regardless of PD-L1 status (NCT02499367). We collected 400 tissue samples from 110 patients, including both pre- and on-treatment biopsies, enabling detailed analysis of tumor-microenvironment interactions and evolution during immunotherapy. Patients with a complete or partial response, or with stable disease lasting longer than 24 weeks, were classified as responders. Using the SpaTopic algorithm, we extracted ten spatial niches (cellular neighborhoods) in our dataset, six of which were dominated by cancer cells. Comparing these with gene expression-based cell clusters revealed a significant link between tumor cell states and their spatial niches. For instance, tumor inflammatory response gene sets are associated with the immune infiltration niche, while hypoxia and epithelial-mesenchymal transition (EMT) gene sets are related to the tumor core. We also defined genomic clusters based on copy-number alterations inferred from spatial transcriptomics data. Within the same genomic cluster, diverse cellular states and spatial niches were observed, indicating that cancer can adapt non-genetically to different niches. Notably, responders exhibited more organized and consistent changes in cellular states and spatial niches before and after immunotherapy, in contrast to non-responders. For example, cancer cells from responders showed increased expression in inflammatory response genes, while non-responders exhibited patient-specific cellular states. This multi-dimensional approach advances our understanding of tumor-immune interactions in mTNBC and offers potential strategies to enhance immunotherapy response.