A nomogram to preoperatively predict the aggressiveness of non-functional pancreatic neuroendocrine tumors based on CT features
Xiaoding Shen,Fan Yang,Taiyan Jiang,Zhenjiang Zheng,Yonghua Chen,Chunlu Tan,Nengwen Ke,Jiajun Qiu,Xubao Liu,Hao Zhang,Xing Wang
DOI: https://doi.org/10.1016/j.ejrad.2023.111284
IF: 4.531
2024-01-06
European Journal of Radiology
Abstract:Objectives To develop a nomogram to predict the aggressiveness of non-functional pancreatic neuroendocrine tumors (NF-pNETs) based on preoperative computed tomography (CT) features. Methods This study included 176 patients undergoing radical resection for NF-pNETs. These patients were randomly divided into the training (n=123) and validation sets (n=53). A nomogram was developed based on preoperative predictors of aggressiveness of the NF-pNETs which were identified by univariable and multivariable logistic regression analysis. The aggressiveness of NF-pNETs was defined as a composite measure including G3 grading, N+, distant metastases, and/ or disease recurrence. Results Altogether, the number of patients with highly aggressive NF-pNETs was 37 (30.08%) and 15 (28.30%) in the training and validation sets, respectively. Multivariable logistic regression analysis identified that tumor size, biliopancreatic duct dilatation, lymphadenopathy, and enhancement pattern were preoperative predictors of aggressiveness. Those variables were used to develop a nomogram with good concordance statistics of 0.89 and 0.86 for predicting aggressiveness in the training and validation sets, respectively. With a nomogram score of 59, patients with NF-pNETs were divided into low-aggressive and high-aggressive groups. The high-aggressive group had decreased overall survival (OS) and disease-free survival (DFS). Moreover, the nomogram showed good performance in predicting OS and DFS at 3, 5, and 10 years. Conclusion The nomogram integrating CT features helped preoperatively predict the aggressiveness of NF-pNETs and could potentially facilitate clinical decision-making.
radiology, nuclear medicine & medical imaging