CT radiomics signature: a potential biomarker for fibroblast activation protein expression in patients with pancreatic ductal adenocarcinoma
Yinghao Meng, Jieyu Yu, Mengmeng Zhu, Jian Zhou, Na Li, Fang Liu, Hao Zhang, Xu Fang, Jing Li, Xiaocheng Feng, Li Wang, Hui Jiang, Jianping Lu, Chengwei Shao, Yun Bian
Abdom Radiol (NY) . 2022 Apr 22. doi: 10.1007/s00261-022-03512-6. Online ahead of print.
Purpose: To develop and validate a radiomics model to predict fibroblast activation protein (FAP) expression in patients with pancreatic ductal adenocarcinoma (PDAC).
Methods: This retrospective study included consecutive 152 patients with PDAC who underwent MDCT scan and surgical resection from January 2017 to December 2017 (training set) and from January 2018 to April 2018 (validation set). In the training set, 1409 portal radiomic features were extracted from each patient's preoperative imaging. Optimal features were selected using the least absolute shrinkage and selection operator (LASSO) logistic regression algorithm, whereupon the extreme gradient boosting (XGBoost) was developed using the radiomics features. The performance of the XGBoost classifier performance was assessed by its calibration, discrimination, and clinical usefulness.
Results: The patients were divided into FAP-low (n = 91; 59.87%) and FAP-high (n = 61; 40.13%) groups according to the optimal FAP cutoff (45.71%). Patients in the FAP-low group showed longer survival. The XGBoost classifier comprised 13 selected radiomics features and showed good discrimination in the training set [area under the curve (AUC), 0.97] and the validation set (AUC, 0.75). It also performed well in the calibration test and decision-curve analysis, demonstrating its potential clinical value.
Conclusions: The XGBoost classifier based on CT radiomics in the portal venous phase can non-invasively predict FAP expression and may help to improve clinical decision-making in patients with PDAC.
Read Full Article Here: https://doi.org/10.1007/s00261-022-03512-6