Related Experiment Video
Updated: May 8, 2026
09:19
Spatial Profiling of Protein and RNA Expression in Tissue: An Approach to Fine-Tune Virtual Microdissection
Published on: July 6, 2022
Non-invasive profiling of the tumour microenvironment with spatial ecotypes
Wubing Zhang1,2, Erin L Brown1,2, Abul Usmani3
1Institute for Stem Cell Biology and Regenerative Medicine, Stanford University, Stanford, CA, USA.
Nature
|May 6, 2026
Summary
Researchers developed a machine-learning framework to identify spatial ecotypes (SEs) in the tumor microenvironment (TME). These SEs, found in diverse cancers, predict immunotherapy response and can be detected in blood cell-free DNA (cfDNA).
Area of Science:
- Oncology
- Computational Biology
- Genomics
Background:
- Multicellular programs within the tumor microenvironment (TME) are crucial for cancer development and treatment response.
- Identifying and clinically profiling these complex TME ecosystems remains a significant challenge.
Purpose of the Study:
- To develop a machine-learning framework for profiling spatially-dependent cell states and multicellular ecosystems, termed spatial ecotypes (SEs).
- To identify conserved SEs across diverse human carcinomas and melanomas and assess their biological and clinical significance.
Main Methods:
- Integrated over 10 million single-cell and spot-level spatial transcriptomes from human carcinomas and melanomas.
- Utilized machine learning to identify nine conserved spatial ecotypes (SEs).
- Validated SE distinguishability using DNA methylation profiling and deep learning for cfDNA recovery.
Main Results:
- Identified nine conserved SEs with distinct biology, geospatial features, and clinical outcome associations, including links to immunotherapy response.
- Demonstrated that SEs are distinguishable by DNA methylation.
- Showed SEs are recoverable from plasma cell-free DNA (cfDNA) using deep learning.
- Found strong associations between cfDNA-derived SE levels and immunotherapy response in melanoma patients.
Conclusions:
- Revealed fundamental units of TME organization through conserved spatial ecotypes (SEs).
- Established a multimodal platform for profiling both solid and liquid TMEs.
- Highlighted the potential of SEs for improved cancer risk stratification and personalized therapy, particularly in relation to immunotherapy.
