Abstract
Addressing the significant level of variability exhibited by pancreatic cancer necessitates the adoption of a systems biology approach that integrates molecular data, biological properties of the tumors, medical images, and clinical features of the patients. In this study, a comprehensive multi-omics methodology was employed to examine a distinctive collection of patient dataset containing rapid autopsy tumor and normal tissue samples as well as longitudinal imaging with a focus on pancreatic cancer. By performing a whole exome sequencing analysis on tumor and normal tissues to identify somatic gene variants and a radiomic feature analysis to tumor CT images, the genome-wide association approach established a connection between pancreatic cancer driver genes and relevant radiomic features, enabling a thorough and quantitative assessment of the heterogeneity of pancreatic tumors. The significant association between sets of genes and radiomic features revealed the involvement of genes in shaping tumor morphological heterogeneity. Some results of the association established a connection between the molecular level mechanism and their outcomes at the level of tumor structural heterogeneity. Because tumor structure and tumor structural heterogeneity are related to the patients' overall survival, patients who had pancreatic cancer driver gene mutations with an association to a certain radiomic feature have been observed to experience worse survival rates than cases without these somatic mutations. Furthermore, the association analysis has revealed potential gene mutations and radiomic feature candidates that warrant further investigation in future research endeavors.
Introduction
Pancreatic ductal adenocarcinoma (PDAC) is notoriously lethal and highly heterogeneous. While the "Big Four" gene mutations (KRAS, TP53, CDKN2A, and SMAD4) are well-documented, they do not fully explain the vast differences in tumor appearance and behavior between patients. This study introduces a "radioGWAS" framework to bridge the gap between a tumor’s genetic blueprint and its physical manifestation on medical imaging. By identifying how specific somatic mutations drive structural "chaos" in CT scans, researchers aim to develop non-invasive "virtual biopsies" to track disease progression and predict patient survival.
Methods and Data Integration
The researchers utilized a unique dataset from the University of Nebraska Medical Center’s Pancreatic Cancer Rapid Autopsy Program, consisting of 26 patients with high-quality tissue samples and longitudinal CT imaging.
The multi-omics methodology involved:
- Whole Exome Sequencing (WES): Identifying somatic mutations by comparing tumor and normal tissue.
- Radiomic Feature Extraction: Extracting 944 quantitative features from CT scans, including shape, texture, and wavelet-filtered intensity distributions.
- Sequence Kernel Association Test (SKAT): Applying this "burden test" to correlate the total mutational load of specific genes with individual radiomic features.
Validation was performed using the TCGA Pancreatic Cancer cohort (185 patients) to assess the clinical relevance and survival impact of the identified gene-image associations.
Results: Linking the Radiome to the Genome
The study identified 132 genes with mutations in at least three patients and discovered several significant associations ($P < 10^{-4}$) between these genes and radiomic textures. Interestingly, the most common mutations (KRAS and TP53) were not the primary drivers of visible heterogeneity. Instead, "long-tail" driver genes appeared to be the architects of the tumor's complex imaging phenotype.
Specific findings include:
- Biological Mirroring: The "Busyness" radiomic feature, reflecting rapid pixel intensity changes, was linked to genes like NOTCH1, JUN, and KDR, which are involved in cell death and angiogenesis.
- Structural Heterogeneity: Mutations in EGFR were associated with image "Skewness," suggesting a link between cell proliferation pathways and asymmetric tumor density.
- Survival Correlation: Patients with mutations in genes associated with the "Contrast" feature (e.g., CDKN2A, PRKG1) showed significantly shorter survival times (FDR-adjusted $P = 0.0348$).
Discussion and Future Directions
This "proof-of-concept" study demonstrates that radiomics can reflect the underlying genetic landscape of pancreatic cancer. However, limitations include the small primary cohort and the temporal gap between the final CT scan and tissue collection at autopsy. Future research will focus on larger-scale validation and applying this framework to metastatic lesions to understand how the radiome-genome link evolves during cancer progression. Tightening data synchronization with immediate pre-autopsy imaging remains a priority for future iterations of this methodology.
Summary updated