Data-Driven Radiogenomic Approach for Deciphering Molecular Mechanisms Underlying Imaging Phenotypes in Lung
Sarah Fischer1,2, Nicolas Spath1,3, Mohamed Hamed1
1Institute for Biostatistics and Informatics in Medicine and Ageing Research, Rostock University Medical Center, Ernst-Heydemannstr. 8, 18057 Rostock, Germany.
International Journal of Molecular Sciences
|March 11, 2023
Summary
Radiogenomics links lung tumor imaging features with molecular profiles to reveal heterogeneity. This study maps radiogenomic associations, identifying potential image biomarkers for genetic variations in tumors.
Area of Science:
- Radiogenomics
- Cancer Research
- Molecular Imaging
Background:
- Tumor heterogeneity in lung nodules is evident in radiological images and molecular profiles.
- Radiogenomics aims to connect quantitative image features with gene expression to understand tumor heterogeneity.
- Challenges exist in linking imaging and genomic data due to different acquisition techniques.
Purpose of the Study:
- To analyze image features and transcriptome profiles to uncover molecular mechanisms behind lung tumor phenotypes.
- To construct a radiogenomic association map (RAM) linking imaging characteristics with molecular signatures.
- To identify potential image biomarkers for underlying genetic variations in lung tumors.
Main Methods:
- Analysis of 86 quantitative image features (shape, texture, size) from lung tumors.
- Integration with transcriptome and post-transcriptome profiles from 22 lung cancer patients.
- Construction of a radiogenomic association map (RAM) and gene regulatory networks.
Main Results:
- A RAM was built, linking tumor morphology, shape, texture, and size with gene and miRNA signatures.
- Gene ontology processes like 'regulation of signaling' showed distinct radiomic signatures on CT images.
- Gene regulatory networks involving TAL1, EZH2, and TGFBR2 were implicated in lung tumor texture formation.
Conclusions:
- Radiogenomic approaches can identify image biomarkers reflecting genetic variations, offering insights into tumor heterogeneity.
- Specific gene ontology processes and regulatory networks are associated with observable CT image phenotypes.
- The methodology is adaptable for other cancer types to enhance mechanistic understanding of tumor phenotypes.
