Integrated Transcriptomic and Histological Analysis of TP53/CTNNB1 Mutations and Microvascular Invasion in...

Area of Science:

• Oncology
• Genomics
• Computational Pathology

Background:

• Hepatocellular carcinoma (HCC) exhibits significant molecular and histopathological heterogeneity.
• Mutations in TP53 and CTNNB1, and microvascular invasion (MVI) are key prognostic indicators in HCC.
• The relationship between these molecular alterations and tumor morphology is not well understood.

Purpose of the Study:

• To investigate transcriptomic changes associated with TP53/CTNNB1 mutations and MVI in HCC.
• To determine if these molecularly defined states are reflected in routine histopathology using deep learning.

Main Methods:

• Differential gene expression analysis of RNA sequencing data from HCC samples.
• Machine learning-based feature selection to identify transcriptional programs.
• Weakly supervised deep learning on whole-slide images to infer histological patterns.

Main Results:

• Distinct gene expression profiles were identified for TP53-mutated, CTNNB1-mutated, and MVI-positive HCC.
• These profiles involve pathways related to proliferation, metabolism, and invasion.
• Deep learning models successfully identified morphological patterns associated with these molecular states from histology.

Conclusions:

• Coherent biological programs underlie key risk determinants in HCC.
• The phenotypic effects of these molecular alterations are detectable in routine histopathology.
• Integrative morpho-molecular approaches can enhance HCC risk assessment.