DOI: 10.1097/mnm.0000000000002245 ISSN: 0143-3636

18F-FDG PET/CT radiomics for prediction of BRCA mutations in high-grade serous ovarian cancer

Fangmei Jiang, Shuyong Liu, Chunli Liu, Wenrui Zhang, Xinyan Liu, Xiang Wu, Juan Xu, Ruoyao Zou
Show PDF Cite

Objective

This study aimed to develop and validate a 18 F-fluorodeoxyglucose PET/computed tomography (CT)-based radiomics model for predicting BRCA mutation status in patients with advanced high-grade serous ovarian cancer (HGSOC).

Methods

A retrospective study of 124 patients with HGSOC who underwent preoperative PET/CT scans and BRCA mutation testing was conducted. The cohort was divided into training and internal validation sets using a 7 : 3 ratio. An external validation set of 25 patients was used to confirm model generalizability. Radiomic features were extracted from PET and CT images and used to build predictive models using logistic regression, random forest, support vector machine, and decision tree classifiers. The models were evaluated for their predictive performance using metrics such as the area under the receiver operating characteristic curve (AUC), accuracy, precision, sensitivity, and F1 score. A combined model incorporating radiomics, PET/CT metabolic parameters, and clinical variables was also constructed. In addition, the radiomics models were used to predict progression-free survival (PFS) and overall survival in the training and internal validation cohorts.

Results

The logistic regression model showed the best performance, with AUCs of 94.79, 87.37, and 85.06% in the training, internal, and external validation cohorts, respectively. The combined model achieved the highest AUCs of 0.968, 0.932, and 0.960 in the training, internal validation, and external validation cohorts, respectively. Significant differences in PFS between risk groups were observed, while overall survival differences were not statistically significant.

Conclusion

The PET/CT-based radiomics model accurately predicts BRCA mutation status in HGSOC, potentially aiding in personalized treatment strategies.

More from our Archive