Abstract
BACKGROUND: Liver metastasis is a common complication in gallbladder cancer (GBC). We design this study to develop models for predicting the development of liver metastasis in GBC patients and evaluate the risk of mortality in these patients with liver metastasis. METHODS: GBC patients from Surveillance Epidemiology and End Results (SEER) between 2010 and 2016 were included in this study. Logistic regression was performed to discover risk factors and construct predictive models for liver metastasis in GBC patients. Cox regression was utilized to find risk factors of mortality in GBC patients with liver metastasis. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate the performance of the constructed predictive models. RESULTS: Multivariate logistic regression confirmed that T stage, N stage, and tumor grade were risk factors for liver metastasis in GBC patients. Composed of these factors, the model for predicting the development of liver metastasis had AUCs of 0.707 and 0.657 in the training cohort and testing cohort, respectively. Multivariate Cox regression showed that surgery of the primary site and chemotherapy were independently associated with the mortality of GBC patients with liver metastasis. Composed of these two factors, the predictive model for 1-year mortality of GBC patients with liver metastasis had AUCs of 0.734 and 0.776 in the training cohort and testing cohort, respectively. CONCLUSION: The predictive models that we constructed are helpful for surgeons to evaluate the risk of liver metastasis in GBC patients and the survival condition of those with liver metastasis. Surgery of the primary site and chemotherapy should be provided for GBC with liver metastasis.