Non-invasive profiling of the tumour microenvironment with spatial ecotypes

2026-05-06 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Artificial Intelligence & Machine Learning, Health & Medical Research, Data Science & Analytics · Depth: Expert, extended

Summary

Researchers developed a machine-learning framework, Spatial EcoTyper, for non-invasive profiling of the tumor microenvironment (TME) using spatial ecotypes (SEs). By integrating over 10 million single-cell and spot-level spatial transcriptomes from diverse human carcinomas and melanomas, the framework identified nine broadly conserved SEs, each with unique biology, geospatial features, and clinical outcome associations, including several linked to immunotherapy response. A deep learning component, Liquid EcoTyper, was developed to recover these SEs from plasma cell-free DNA (cfDNA). In a study of nearly 100 patients with melanoma, cfDNA-derived SE levels showed strong associations with immunotherapy response, outperforming traditional biomarkers like tumor mutational burden (TMB) and PD-L1 expression. This multimodal platform offers implications for improved cancer risk stratification and therapy personalization by enabling both solid and liquid TME profiling.

Key takeaway

For oncology researchers and clinicians evaluating cancer treatment strategies, this work demonstrates that liquid biopsy-derived spatial ecotype profiling offers a powerful, non-invasive method to predict immunotherapy response. You should consider integrating these machine learning-driven SE analyses into your diagnostic and monitoring workflows, particularly for melanoma, as they provide more granular insights into TME organization and may outperform existing biomarkers like TMB and PD-L1 expression for patient stratification.

Key insights

A new machine learning framework enables non-invasive profiling of tumor microenvironment spatial ecotypes from liquid biopsies, predicting immunotherapy response.

Principles

• Multicellular ecosystems (SEs) are fundamental units of tissue organization.
• SEs exhibit conserved cellular, spatial, and clinical features across cancer types.
• cfDNA methylation profiles can reflect TME spatial organization.

Method

Spatial EcoTyper uses data fusion and statistical learning on spatial transcriptomes to identify SEs. Liquid EcoTyper, a deep learning framework, infers SE levels from cfDNA methylation data.

In practice

• Use Liquid EcoTyper for non-invasive TME assessment.
• Profile SEs to predict immunotherapy response in melanoma.
• Integrate spatial and single-cell data for robust ecotype discovery.

Topics

• Spatial Ecotypes
• Tumor Microenvironment
• Spatial Transcriptomics
• Cell-free DNA
• Liquid Biopsy

Code references

• immunogenomics/presto
• digitalcytometry/spatialecotyper

Best for: AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• Defining cancer spatial ecotypes — Spatial EcoTyper uses ML to define nine robust spatial ecotypes in the tumor microenvironment, revealing cell-type-agnostic expression signatures.
• Mitosis Detection in the Wild: Multi-Tumor and Context-Aware Generalization in the MIDOG 2025 Challenge — "In the wild" mitosis detection demands models robust to diverse biological and contextual variance.
• Atlas H&E-TME: Scalable AI-Based Tissue Profiling at Expert Pathologist-Level Accuracy — Atlas H&E-TME provides scalable, expert-level AI-based tissue profiling from H&E slides, validated by a robust dual framework.
• GigaTIME: Scaling tumor microenvironment modeling using virtual population generated by multimodal AI — GigaTIME uses multimodal AI to convert common H&E slides into virtual mIF images, enabling population-scale precision immuno-oncology research.
• Profiling of extracellular vesicles from primary hepatocytes, organoids, and mash patients identifies cell injury-specific signatures — Extracellular vesicle protein signatures, identified across multi-models and validated by machine learning, offer non-invasive MASH diagnostics.
• SegTME-UNI2: A Foundation Model-Based Framework for Generalisable Multiclass Cell Segmentation and LLM-Driven Tumour Microenvironment Characterisation in Histopathology — SegTME-UNI2 integrates foundation models and pseudo-labeling for robust, generalizable multiclass cell segmentation and TME characterization.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.