• Fang Huang and Meihua Fang.
• Department of Oncology, The Affiliated Wuxi People's Hospital of Nanjing Medical University, Wuxi, P. R. China.
• Medicine (Baltimore). 2023 Sep 29; 102 (39): e34702e34702.

AbstractLiver was the most common site of distant metastasis in patients with gastric cancer (GC). The prediction model of the risk of liver metastasis was rarely proposed. Therefore, we aimed to establish a prediction model for liver metastasis in patients with GC. In this retrospective cohort study, we extracted demographic and clinical data of all the GC patients from the Surveillance, Epidemiology, and End Results registration database from 2010 to 2015. Patients were divided into training set (n = 1691) for model development and testing set (n = 3943) for validation. Univariable and multivariable logistic regression analyses were carried out on the training set to screen potential predictors of liver metastasis and constructed a prediction model. The receiver operator characteristics curves with the area under curve values were used to assess the predictive performance of the liver metastasis prediction model. And a nomogram of the prediction model was also constructed. Of the total 5634 GC patients, 444 (7.88%) had liver metastasis. Variables including age, gender, N stage, T stage, Lauren classification, tumor size, histological type, and surgery were included in the liver metastasis prediction model. The study results indicated that the model had excellent discriminative ability with an area under curve of 0.851 (95% confidence interval: 0.829-0.873) in the training set, and that of 0.849 (95% confidence interval: 0.813-0.885) in the testing set. We have developed an effective prediction model with 8 easily acquired predictors of liver metastasis. The prediction model could predict the risk of liver metastasis in GC patients and performed well, which would assist clinicians to make individualized prediction of liver metastasis in GC patients and adjust treatment strategies in time to improve the prognosis.Copyright © 2023 the Author(s). Published by Wolters Kluwer Health, Inc.

7 keywords...

hide…