Cellular architecture and neighborhood-informed virtual spatial tumor profiling from histopathology

Area of Science:

• Oncology
• Computational Biology
• Bioinformatics

Background:

• The tumor microenvironment (TME) significantly influences cancer progression and treatment resistance.
• Understanding the spatial organization of the TME is crucial for clinical applications but remains challenging.
• Current methods for TME analysis lack comprehensive spatial profiling capabilities.

Purpose of the Study:

• To develop an artificial intelligence (AI) platform for inferring tumor ecological habitats from standard histopathology images.
• To establish a scalable method for spatial profiling of the TME that bridges single-cell data with clinical insights.
• To enable prediction of patient outcomes and treatment responses based on TME spatial architecture.

Main Methods:

• Developed CANVAS, an AI platform integrating 41-plex spatial proteomics data from 18 million cells across 457 non-small cell lung cancer patients.
• Defined 10 reproducible cellular neighborhoods (CNs) representing conserved TME spatial organization.
• Employed multimodal alignment and foundation-model-based morphological encoding to predict CN-anchored habitat structures from H&E slides.

Main Results:

• CANVAS successfully identified 10 distinct cellular neighborhoods (CNs) reflecting conserved TME spatial organization.
• The platform accurately predicted CN-anchored habitat structures from H&E slides.
• Clinical evaluation in over 5,000 patients across 9 cancer types demonstrated CANVAS's utility in prognostic modeling, spatial ecotype stratification, and immunotherapy outcome prediction.

Conclusions:

• CANVAS provides a clinically scalable platform for spatial profiling of the tumor microenvironment using H&E histopathology.
• The AI-driven approach bridges single-cell spatial proteomics with population-level insights for precision oncology.
• CANVAS facilitates enhanced understanding of TME architecture for improved cancer diagnosis, prognosis, and therapeutic strategies.