Spatial EcoTyper: Noninvasive and High-Resolution Profiling of Spatially Defined Tumor Microenvironments Using A Machine Learning Framework

Stanford researchers in Prof. Aaron Newman's lab have created a machine learning framework to profile spatial ecotypes in the tumor microenvironment (TME), providing a new way to monitor cancer pathogenesis and immunotherapy response noninvasively. This technology enhances risk stratification and personalizes therapy across multiple types of carcinomas.

Understanding the spatial microenvironments within tumors is crucial for predicting disease progression and response to therapy. However, current methods to profile the identified multicellular ecosystems, or spatial ecotypes (SEs), are hindered by a lack of single-cell resolution, limitation to predefined cell types/states, and difficulty in performing cross-platform analyses. Also, conventional profiling methods require invasive tumor biopsies, which can introduce significant sampling bias and are not suitable for longitudinal studies. Additionally, while cell-free DNA (cfDNA) shows promise for non-invasive profiling, no liquid biopsy assay currently exists for TME assessment.

The inventors have developed a machine learning framework, termed Spatial EcoTyper, that directly profiles SEs from single-cell spatial transcriptomics data. Current systems face challenges in profiling SEs due to the lack of single-cell spatial resolution, limited community detection, and difficulties in cross-platform analyses. Spatial EcoTyper not only overcomes these challenges but also enables the noninvasive assessment of SEs using plasma cfDNA. This pioneering technology provides a more accurate and broader understanding of the tumor microenvironment, exhibiting already great potential in improving immunotherapy response assessment compared to existing methods.

Figure Description: Multimodal profiling of spatial ecotypes in human cancer. Top: Discovery and clinical characterization of spatially co-localized cell states in human tumors, termed spatial ecotypes (SEs). Bottom: Recovery of SEs in plasma cell-free DNA and the use of noninvasive SE profiling for immunotherapy response assessment.

Stage of Development:
Prototype