RESUMEN
Background: Gastroenteropancreatic neuroendocrine neoplasm (GEP-NEN) is a rare heterogeneous tumor. Liver metastasis seriously affects the prognosis of GEP-NEN. However, few tools are existed to predict GEP-NEN complicated with synchronous liver metastasis. Aims: To analyze the risk factors of synchronous liver metastasis in patients with GEP-NEN and establish a nomogram to predict synchronous liver metastasis in patients with GEP-NEN. Methods: A total of 10 973 pathologically confirmed patients with GEP-NEN from Jan. 2010 to Dec. 2017 were collected from SEER database and divided randomly into training set (n=7 511) and test set (n=3 462). Both groups were divided into liver metastasis group and non-liver metastasis group according to the occurrence of liver metastasis. Multifactorical logistic regression analysis was used to identify the risk factors of liver metastasis in patients with GEP-NEN. R software was used to establish and verify the nomogram of liver metastasis in GEP-NEN patients. Results: Liver metastasis was associated with gender, age, race, primary tumor site, degree of differentiation, tumor diameter, T3/4 stage, and lymph node metastasis in patients with GEP-NEN. The results of multivariate logistic regression analysis showed that primary tumor site (small intestine and pancreas), differentiation degree (poorly differentiated and undifferentiated), diameter of tumor ≥ 5 cm, T3/4 stage and lymph node metastasis were independent risk factors affecting liver metastasis in patients with GEP-NEN (P< 0.001). The concordance index of internal validation for nomogram was 0.838 (95% CI: 0.826-0.849), and the concordance index of external validation was 0.847 (95% CI: 0.829-0.864). Conclusions: GEP-NEN patients with primary tumor site in small intestine or pancreas, poor differentiation and undifferentiation, diameter of tumor ≥5 cm, T3/4 stage and lymph node metastasis are more likely to develop liver metastasis which suggested that such patients need to be alert for the occurrence of liver metastasis and need more aggressive treatment. The calibration curves fits are good for both the training and test sets, and can help clinicians to make individualized prediction for whether the GEP-NEN patient has synchronous liver metastasis at the initial diagnosis.