3D spatial sampling to quantify morphologic heterogeneity in isocitrate dehydrogenase-wildtype glioblastoma

Area of Science:

• Neuropathology
• Digital Pathology
• Machine Learning

Background:

• Current diagnostic models for brain tumors rely on single H&E slides, overlooking intratumoral heterogeneity.
• Diffuse gliomas exhibit significant epigenetic, genetic, and transcriptional variability within patients.
• The impact of this heterogeneity on diagnostic model development is not well understood.

Purpose of the Study:

• To quantitatively assess morphologic intratumoral heterogeneity in glioblastoma (GBM).
• To investigate the relationship between morphologic variations and molecular alterations in GBM.
• To inform the development of more accurate neuropathology diagnostic models.

Main Methods:

• Acquired 92 distinct tissue samples from 10 patients with isocitrate dehydrogenase-wildtype GBM.
• Quantified cell density, nucleus area, and nucleus circularity from whole-slide H&E images.
• Performed tumor-level and sample-level mutation profiling.

Main Results:

• Significant morphologic variations (cell density, nucleus area, circularity) were observed both between patients and within individual tumors.
• Mutations in tumor protein 53 (TP53) correlated with larger nucleus area and decreased nucleus circularity.
• Morphologic features did not show association with regional tumor location.

Conclusions:

• Intratumoral morphologic heterogeneity is a significant factor in glioblastoma.
• TP53 mutations may influence observable morphologic characteristics.
• Future H&E-based diagnostic and prognostic models require training datasets that include multiple spatially distinct samples per patient to account for heterogeneity.