Diffusion model comparison identifies distinct tumor sub-regions and tracks treatment response

Area of Science:

• Biomedical Imaging
• Radiotherapy Research
• Tumor Microenvironment Analysis

Background:

• Tumor response to treatment is often assessed using MRI biomarkers derived from pre- and post-treatment data.
• Tumors exhibit spatial and temporal heterogeneity, necessitating region-specific and time-varying models for accurate analysis.
• Current MRI biomarker models may not fully capture the complex and dynamic nature of tumor microenvironments.

Purpose of the Study:

• To evaluate the spatial variation in the suitability of two diffusion-weighted MRI (DW-MRI) models within tumors.
• To assess how model suitability changes in response to radiotherapy.
• To infer qualitatively different tumor microenvironments based on model performance.

Main Methods:

• Acquisition of DW-MRI data in CT26 subcutaneous allografts before and after radiotherapy.
• Comparison of restricted and time-independent diffusion models at the voxel level.
• Technical and biological validation of the restricted diffusion microstructural model (%MM) using simulations and histology.

Main Results:

• Observed spatial and radiotherapy-related variations in model suitability within tumors.
• A decrease in %MM from 64% at baseline to 44% at 6 days post-radiotherapy in the treated group.
• %MM showed a negative correlation with histological necrosis but overestimated its extent.

Conclusions:

• Tumor diffusion modeling reveals spatial heterogeneity and radiotherapy-induced changes, suggesting distinct and evolving tumor sub-regions.
• The %MM biomarker is validated and shows correlation with tumor necrosis, though it overestimates the necrotic fraction.
• These findings support the use of voxel-level DW-MRI model analysis for characterizing dynamic tumor microenvironments.