Colorectal cancer (CRC) globally ranks among malignancies with highest incidence and recurrence rates, afflicting almost one million subjects each year. Despite the wealth of studies carried out in CRC correlating the degree and type of immune infiltration with invasive capability, spatial relationships governing the interplay between immune and tumor cells in the CRC tumor microenvironment (TME) have been thus far explored at coarse-grained scale. In this work, we employ spatial profiling technologies to investigate the cellular ecosystem of metastatic and non-metastatic CRC, providing a comprehensive picture of the spatial architecture of immune cell responses at play and elucidating their role in the progression towards an invasive cancer phenotype.

To identify links between spatial determinants of the CRC ecosystem and the process of tumor metastasization, we assembled an initial retrospective cohort of primary tumor FFPE samples from 16 CRC patients with either overt or undetectable metastases after surgery. Samples were processed for targeted spatial transcriptomics with the latest version of the CosMx™️ SMI platform, profiling the expression of ~6,000 different genes in situ resulting in a comprehensive collection of more than 2.2 million single-cell transcriptomes coupled to their spatial location.

Leveraging a large-scale scRNA-seq cohort, we uncover the transcriptional diversity of malignant cells across more than 100 CRCs at single-cell resolution. Notably, cancer cells exhibited a state marked by features of TNF signaling and hypoxia response, active in primary tumors prone to form metastases. This transcriptional state also correlated with increased IFN response, inflammation and EMT programs, suggesting an intimate link between the metastasization potential of these cells and immune compartments of the CRC TME. Integrating spatial information with the depth provided by scRNA-seq data facilitated cell type annotation and the identification of malignant cell states directly in situ, revealing how metastatic CRC spatially orchestrates expression of these gene programs. To further connect features of the CRC TME with the development of pro-metastatic cancer cell states, we correlated the abundance of local ligand-receptor interactions in the CRC TME with the activity of transcription factors linked to the emergence of invasive cancer cell states.

By combining large-scale scRNA-seq data and targeted spatial approaches we gathered an unbiased look into the molecular and physical architecture of metastatic and non-metastatic CRC. Cancer cells displayed gene programs which are intimately linked to the surrounding composition of the tumor bed they reside in. Ultimately, this work provides initial mechanistic insights into how TME signals drive metastatic potential in CRC.