Exploratory analysis of CT-derived radiomic features and adipose tissue distribution in relation to lipid metabolism pathway activity in ovarian serous cystadenocarcinoma
Introduction: Ovarian cancer is characterized by high recurrence rates and complex metabolic reprogramming. Lipid metabolism pathways may influence tumor imaging phenotypes and contribute to biologic heterogeneity. Objective: This study investigated whether computed tomography (CT)-derived radiomic features are associated with metabolic activity profiles in ovarian cancer. Methods: This retrospective radiogenomic study included 68 patients from the Cancer Genome Atlas Ovarian Cancer cohort with available contrast-enhanced CT imaging and matched transcriptomic data. Patients were stratified into high and low metabolic activity groups according to lipid metabolism-related pathway expression profiles. A two-dimensional radiomic approach was adopted using the axial CT slice showing the largest tumor area. Tumor regions were manually segmented, and handcrafted radiomic features were extracted using a custom pipeline developed in Python. Statistical comparisons between groups were performed using the Mann–Whitney U test with Benjamini–Hochberg false discovery rate (FDR) correction. Results: Several radiomic features showed nominal or borderline differences between high- and low-metabolic-activity groups, including glcm_ASM_std (p = 0.027), shape_perimeter_mm (p = 0.045), glrlm_rln (p = 0.049), shape_minor_axis_length_px (p = 0.050), and intensity_p10_HU (p = 0.053). Tumors with higher metabolic activity tended to exhibit larger morphologic dimensions and subtle texture alterations. Subcutaneous adipose tissue (SAT) demonstrated the strongest inverse correlation with LipidScore (Spearman rho = −0.24), and a non-significant trend toward higher SAT values was observed in the low-lipid-metabolism group (p = 0.072). However, none of the radiomic features remained statistically significant after FDR correction. Conclusion: CT-derived radiomic features showed weak exploratory associations with metabolic activity-related molecular profiles in ovarian cancer. Although no robust associations were confirmed after multiple-comparison correction, the observed trends suggest potential radiogenomic interactions related to tumor metabolism that warrant further investigation in larger cohorts.
Authors
- Federico Greco (ORCID: https://orcid.org/0000-0002-9477-0238)
- Chiara Pallara (ORCID: https://orcid.org/0000-0003-3005-343X)
- Roberta Montefrancesco
- Arnaldo Scardapane (ORCID: https://orcid.org/0000-0002-6116-5701)
- Chiara Noemi Greco (ORCID: https://orcid.org/0009-0007-1468-4625)
- Luca Leo
- Luca Giordano
Institutions
- University of Salento (IT)
- Ospedale Vito Fazzi (IT)
- University of Bari Aldo Moro (IT)
Publication Details
- Journal
- Eurasian Journal of Medicine and Oncology
- Published
- 2026-11-16
- DOI
- https://doi.org/10.36922/ejmo026230260
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00