胃神经内分泌瘤患者预后Nomogram预测模型的构建

[摘要] 目的：基于美国国立癌症研究所（NCI）的监测、流行病学和最终结果（SEER）数据库数据分析影响胃神经内分泌瘤（G-NEN）患者预后的相关因素，并构建Nomogram预测模型用于个体化预测G-NEN患者预后。方法：收集SEER数据库2010 年至2015 年有完整随访资料的2 720 例G-NEN患者的临床数据，基于生存分析确定独立危险因素并构建Nomogram 预测模型，采用一致性指数（C-index）和校准曲线评估模型准确性，采用受试者特征曲线下面积（AUC）比较该模型与第7 版AJCC TNM分期评估法的预测价值。结果：2 720 例G-NEN患者的1、3、5 年生存率分别为88.14%、79.09%、71.86%。多因素COX回归分析显示，性别、年龄、婚姻状况、是否伴发其他肿瘤、组织学类型、肿瘤分级、T分期、M分期及是否手术是影响G-NEN患者生存时间的独立危险因素。新构建的Nomogram预测模型C-index 为0.816，显著高于7 版AJCC TNM分期评估法的0.702（P<0.001），且1、3、5 年校准曲线显示预测生存率与实际生存率之间具有良好的一致性。新构建的Nomogram 预测模型1、3、5 年AUC分别为0.800、0.811及0.820，显著高于第7 版AJCC TNM分期评估法的0.650、0.688 及0.698（Z=6.600、8.058、9.632，均P<0.0001）。结论：构建的预测G-NEN患者预后的Nomogram模型具有较高的预测价值，能够个体化预测G-NEN患者的生存率，有助于临床治疗决策和临床研究方案的选择。

Construction of Nomogram prediction model for prognosis of gastric neuroendocrine neoplasms

[Abstract] Objection: To analyze the factors affecting the prognosis of patients with gastric neuroendocrine neoplasms (G-NENs) by using the surveillance of National Cancer Institute (NCI) of America, Epidemiology and End Results (SEER) database, and to construct a prognostic Nomogram model for individualized prediction of prognosis in patients with G-NENs. Methods: The clinical data of 2720 G-NENs patients with complete follow-up data from 2010 to 2015 in the SEER database were collected. The prognostic Nomogram model was constructed based on independent risk factors determined by survival analysis. The consistency index (C-index) and calibration curve were used to evaluate its accuracy. Area under the curve (AUC) was used to compare the evaluation value between the Nomogram and the 7th edition of AJCC TNM staging. Results: The 1-, 3-, and 5-year survival rates of 2,720 patients with G-NENs were 88.14%, 79.09%, and 71.86%, respectively. Multivariate COX regression analysis showed that gender, age, marital status, other associated tumors, histological type, tumor grade, T stage, M stage, and surgery were independent risk factors affecting survival time of GNENs patients. The C-index of newly constructed Nomogram prediction model was 0.816, which was significantly higher than 0.702 of the 7th AJCC TNM staging (P<0.001), and the 1-, 3- and 5-year calibration curves showed a good agreement between predicted survival and actual survival. The AUC for 1-, 3- and 5-year survival by Nomogram prognostic model was 0.800, 0.811, and 0.820,which was higher than 0.650, 0.688 and 0.698 of the 7th AJCC TNM staging, and the differences were statistically significant (Z=6.600, 8.085, 9.632, all P<0.0001). Conclusion: The Nomogram prediction model drawn in this study has a high prognostic value and can individually predict the survival rate of G-NENs patients, which is helpful for clinical treatment decision-making and clinical research options.