A novel nomogram for predicting liver metastasis in patients with gastrointestinal stromal tumor: a SEER-based study

BACKGROUND: Liver metastasis (LIM) of gastrointestinal stromal tumor (GIST) is associated with poor prognosis. The present study aimed at developing and validating nomogram to predict LIM in patients with GIST, thus helping clinical diagnosis and treatment. METHODS: The data of GIST patients derived from Surveillance, Epidemiology, and End Results (SEER) database from 2010 to 2016, which were then screened by univariate and multivariate logistic regression for the construction of LIM nomogram. The model discrimination of LIM nomogram was evaluated by concordance index (C-index) and calibration plots, while the predictive accuracy and clinical values were measured by decision curve analysis (DCA) and clinical impact plot. Furthermore, we validated predictive nomogram in the internal testing set. RESULTS: A total of 3797 patients were enrolled and divided randomly into training and validating groups in a 3-to-1 ratio. After logistic regression, the significant variables were sex, tumor location, tumor size, N stage and mitotic rate. The calibration curves showed the perfect agreement between nomogram predictions and actual observations, while the DCA and clinical impact plot showed the clinical utility of LIM nomogram. C-index of the nomogram was 0.812. What’s more, receiver operating characteristic curves (ROC) also showed good discrimination and calibration in the training set (AUC = 0.794, 95% CI 0.778–0.808) and the testing set (AUC = 0.775, 95% CI 0.748–0.802). CONCLUSION: The nomogram for patients with GIST can effectively predict the individualized risk of liver metastasis and provide insightful information to clinicians to optimize therapeutic regimens.