IgA肾病进展至ESRD风险模型的验证与新模型建立

发布时间：2018-03-15 08:12

本文选题：IgA肾病　切入点：终末期肾病　出处：《上海交通大学》2015年博士论文　论文类型：学位论文

【摘要】：目的:研究日本Goto、日本Utsunomiya、法国Berthoux和我国瑞金模型对IgA肾病进展至ESRD风险的预测能力,并初步建立符合我国IgAN特点的ESRD风险预测模型。方法:选取上海交通大学医学院附属新华医院1998年6月至2014年6月间IgAN确诊患者535例,完善基线资料及随访信息采集。其中356例结局明确的患者纳入模型验证组,分别评价四种模型对验证组患者ESRD发生风险的预测能力。535例患者全部纳入模型建立组,通过单因素及多因素危险分析筛选独立危险因素,利用Kaplan-Meier法和Cox比例风险模型建立新的风险预测模型。以分辨力及校准度评价各模型的预测能力。以受试者工作特征曲线下面积诊断模型分辨力,AUC0.9表示模型分辨力较高,p0.05表示差异有统计学意义。采用Hosmer-Lemeshow检验对各模型进行拟合优度检验,p0.05表示模型校准度较好。结果:验证组356例患者中有38例进展至ESRD。Goto模型对2年、5年内ESRD发生风险预测的分辨力较高(2年AUC=0.958,5年auc=0.950,p0.05),10年风险的分辨力稍低(auc=0.876,p0.05)。utsunomiya模型对2年、5年、10年风险的分辨力总体较高(2年auc=0.923,5年auc=0.933,10年auc=0.915,p0.05)。berthoux模型的分辨力均较低(2年auc=0.819,5年auc=0.808,10年auc=0.780,p0.05)。瑞金模型对2年、5年、10年风险预测的分辨力均较高(2年auc=0.946,5年auc=0.939,10年auc=0.931,p0.05)。四种模型的校准度均较好(p0.05)。模型建立组535例患者中有40例进展至esrd,151例失访。单因素分析发现,男性、年龄增长、egfr下降、初始血肌酐升高、收缩压升高、舒张压升高、尿蛋白排泄量升高、贫血、低白蛋白血症、高尿酸血症、总胆红素降低、血清c3降低、严重肾小管萎缩或间质纤维化、新月体形成比例增高、肾小球球性硬化比例增高等15项指标是igan进展至esrd的危险因素。多因素cox回归分析提示,肾小球球性硬化比例(β=2.204,hr=9.062,p=0.004)、细胞性新月体及纤维细胞性新月体形成比例(β=2.004,hr=7.422,p=0.039)和egfr(β=-0.069,hr=0.934,p0.001)是我国igan患者进展至esrd的独立危险因素。新模型预测的分辨力高于上述四种模型(2年auc=0.964,5年auc=0.965,10年auc=0.959,p0.001),校准度亦较高(?2=9.905,p=0.272)。结论:goto模型2年、5年风险预测能力优于其他三种模型,瑞金模型10年风险预测能力最佳,Utsunomiya模型预测能力总体较好,较上述两种模型稍差,Berthoux模型预测能力最差。肾小球球性硬化比例、细胞性新月体及纤维细胞性新月体形成比例和eGFR是我国IgAN患者进展至ESRD的独立危险因素。新模型对2年、5年、10年内ESRD发生风险具有较高分辨力和校准度,新模型危险分层可实现对患者进行合理评估及分层管理,总体预测能力较佳。
[Abstract]:Objective: to study the predictive ability of Goto, Utsunomiya, Berthoux and Ruijin models in China for predicting the risk of progression of IgA nephropathy to ESRD. Methods: 535 patients diagnosed with IgAN from June 1998 to June 2014 in Xinhua Hospital affiliated to Shanghai Jiaotong University Medical College were selected. To improve baseline data and follow up information collection, 356 patients with definite outcome were included in the model validation group, and the predictive ability of four models to predict the risk of ESRD in the validation group. 535 patients were all included in the model building group. Independent risk factors were screened by univariate and multivariate risk analysis. A new risk prediction model was established by using Kaplan-Meier method and Cox proportional risk model. The predictive ability of each model was evaluated by resolution and calibration degree. The model resolution was expressed by the area diagnosis model resolution under the operating characteristic curve of subjects and AUC0.9. Hosmer-Lemeshow test was used to test the goodness of fit for each model. Results: 38 out of 356 patients in the validation group had advanced to ESRD.Goto model within 2 and 5 years. The resolution of the genetic risk prediction was higher (2 years AUCX 0.958, 5 years aucmated 0.950 p0.05a, 10 year risk resolution was slightly lower than that of 10 years risk model 0.876 p0.05U. Utsunomiya model for 2 years, 5 years, 10 years risk resolution for 2 years, 5 years auction 0.923, 5 years auc0. 933, 10 years auc0. 915 p0.05n.berthoux model were lower (2 years auc0. 0. 808, 0. 808 in 5 years, 0. 780 p 0. 05. 0. 05%.) the resolution of 10 years risk was higher than that of 10 years of auction (0. 923, 0. 933, 0. 915, 0. 808, 0. 808, 0. 780 p 0. 05) respectively. The model had higher resolution for 2, 5 and 10 years risk prediction (2 years aucnus 0.946, 5 year aucmor 0.939, 10 year auction 0.931 p0.05). The calibration degree of the four models was better. 40 of 535 patients in the model-building group had advanced to ESR 151 cases. Univariate analysis showed that, In males, age increased, initial serum creatinine increased, systolic blood pressure increased, diastolic blood pressure increased, urinary protein excretion increased, anemia, hypoalbuminemia, hyperuricemia, total bilirubin decreased, serum c3 decreased. Fifteen indexes, such as severe tubular atrophy or interstitial fibrosis, increased crescent shape and glomerular sclerosis, were the risk factors for the progression of igan to esrd. The ratio of glomerular glomerulosclerosis (尾 -2.204), the ratio of cellular crescents to fibrocytic crescents (尾 2.004 ~ 7.422p0. 039) and egfr- (尾 -0.069? HR0.934 / p0.001) are independent risk factors for the progression to esrd in igan patients in China. The resolution predicted by the new model is higher than that of the four models (2 years auc0.964 / 5). A year of auction 0.965, a 10-year auction of 0.959 p0.001g, and a higher calibration degree? 2 / 9.905 / 0.272. Conclusion the 5 year risk forecasting ability of the two year model is better than that of the other three models, and the best risk forecasting ability of Ruijin model is 10 years. The Utsunomiya model has a better forecasting ability. The prediction ability of Berthoux model was worse than that of the above two models. The proportion of cellular crescents and fibrocytic crescents and eGFR are independent risk factors for the progression to ESRD in IgAN patients in China. The new model has high resolution and calibration for the risk of ESRD in 2, 5 and 10 years. The new model can realize reasonable assessment and management of patients with risk stratification, and the overall predictive ability is better.
【学位授予单位】：上海交通大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：R692.31

【相似文献】