Non-invasive profiling of the tumour microenvironment with spatial ecotypes (www.nature.com)

Researchers have developed an innovative machine-learning framework for non-invasive profiling of the tumor microenvironment (TME) by identifying distinct multicellular ecosystems known as spatial ecotypes (SEs). By analyzing over 10 million single-cell and spatial transcriptomic data from various human cancer types, particularly melanomas and carcinomas, the study identified nine conserved SEs, each associated with unique biological characteristics and clinical outcomes, notably including responses to immunotherapy. This method addresses previous challenges in profiling SEs, which traditionally required invasive tumor biopsies and limited their analysis to a narrow range of markers. The significance of this breakthrough lies in its potential to enhance cancer risk stratification and personalize therapies by utilizing liquid biopsies, specifically analyzing cell-free DNA (cfDNA) from patients. The findings reveal that levels of SEs in cfDNA correlate strongly with tumor biopsy-confirmed SE levels and patient responses to immune checkpoint inhibitors. By integrating high-resolution spatial data with machine learning, the framework not only allows for comprehensive profiling of TME dynamics but also facilitates improved forecasting and monitoring of therapeutic responses, paving the way for more effective cancer treatments.