The tumor microenvironment (TME) evolves during malignant cell formation, influences the immune cells, and remodels the extracellular matrix (ECM) and vasculature to orchestrate a supportive environment for tumor growth. The composition and functional state of the TME can vary considerably among different organs due to the presence of unique tissue-resident cells. Here, we collected 4,254,586 cells from 735 patient samples across 36 cancer types, including normal, precancerous, primary, and metastatic tumors, and constructed a pan-cancer resource named TabulaTIME to depict the heterogeneity of the TME. Our integrated analyses reveal that CTHRC1 is a hallmark of ECM-related cancer-associated fibroblasts (CAFs) that are prevalent in different cancer types. Spatiotemporal analyses further indicated that CTHRC1 positive CAFs are located at the leading edge between the malignant and normal regions and prevent immune infiltration. Moreover, we identified a profibrotic macrophage subtype marked by SLPI expression that is involved in ECM remodeling. Interestingly, SLPI positive macrophages colocalized with CTHRC1 positive CAFs to form unique spatial ecotypes with shared upstream regulators including TGFB1 and IL1B. Finally, we demonstrate that TabulaTIME can be utilized to analyze ecotype composition in large-scale cohorts and evaluate immune cell infiltration and patient prognosis. This work establishes a comprehensive single-cell landscape of the heterogenous TME encompassing both immune and stromal cells and offers a potential therapeutic strategy for targeting the profibrotic ecotype in cancer treatment.
Annotation
TabulaTIME employed SELINA, which is a deep learning-based framework for single-cell assignment, to predict the cell types of queried cells, after pre-trained SELINA models by the integrated pan-cancer datasets.