A new study in Nature presents adetailed map of over 2.6 million cells from 30 different types of tumors across 12 organ systems. Researchers from the Human Tumor Atlas Network (HTAN) used single-cell RNA sequencing and spatial transcriptomics to analyze both cancer cells and surrounding immune and stromal cells. Their goal was to better understand how tumors differ – and overlap – at the cellular level, regardless of where they start in the body.
The team found that four core malignant cell state behaviors, or "programs," are shared across many cancers. These include stress responses, cell proliferation, metabolic changes, and a shift toward more mobile, invasive cell types (known as epithelial-to-mesenchymal transition or EMT). Some of these programs were found across multiple cancers, while others were more specific to certain tumor types. Importantly, many tumors contained more than one of these programs at once.
The researchers also studied the tumor microenvironment, including immune cells like T cells and macrophages, as well as fibroblasts. Using spatial transcriptomics, they showed where different cell types are located within the tumor. For example, they observed that exhausted T cells often sit near tumor cells with strong EMT features – suggesting interactions that could affect disease progression or treatment response.
This work may support future development of diagnostics that classify tumors based not just on their tissue of origin, but on the cellular programs they use. The findings could also help identify new biomarkers for immunotherapy or guide how spatial transcriptomics is used in pathology labs.
The dataset is available online for researchers and clinicians. While more work is needed to connect these findings to patient outcomes, the atlas offers understanding of tumor behavior at the single-cell level and could inform future integration of spatial transcriptomics into diagnostic workflows.