2型糖尿病颈动脉粥样硬化患者合并糖尿病肾病的Nomogram预测模型

Nomogram prediction model of type 2 diabetes mellitus-carotid atherosclerosispatients complicated with diabetic kidney disease

摘要:

目的：分析影响2型糖尿病颈动脉粥样硬化（type 2 diabetes mellitus-carotid atherosclerosis，T2DM-CAS）患者合并糖尿病肾病（diabetic kidney disease，DKD）的独立危险因素，构建一个个性化的临床预测模型，预测T2DM-CAS患者发生DKD的风险。方法：选取2型糖尿病（type 2 diabetes mellitus，T2DM）患者883例，收集患者的基本特征、实验室检查、辅助检查和伴发疾病情况。应用LASSO回归优化筛选变量，通过多因素logistic回归建立预测模型。依靠受试者工作特征曲线（receiver operating characteristic curve，ROC曲线）、校准曲线和Hosmer-Lemeshow拟合优度检验验证和评价模型的区分度和校准度；决策曲线分析（decision curve analysis，DCA）评估临床有效性。结果：预测模型图的预测变量包括糖尿病病程、收缩压（systolic blood pressure，SBP）、空腹血糖（fasting plasma glucose，FPG）、甘油三酯（triglycerides，TG）、血尿素氮（blood urea nitrogen，BUN）、血肌酐（serum creatinine，Scr）、胱抑素C（cystatin C，Cys C）和糖尿病视网膜病变（diabetic retinopathy，DR）。该模型具有较好的预测能力，ROC曲线下面积（area under the curve，AUC）为0.831（95％CI=0.800~0.863），内部验证AUC为0.825（95％CI=0.766~0.884）。Hosmer-Lemeshow拟合优度检验显示出较好的拟合度（P=0.822）。DCA显示风险阈值为30％，预测模型在临床上是有益的。结论：包含8个预测变量的列线图（Nomogram）模型可用于预测T2DM-CAS患者的DKD发病风险。

Abstract:

Objective：To analyze the independent risk factors affecting type 2 diabetes mellitus-carotid atherosclerosis（T2DM-CAS） patients complicated with diabetic kidney disease（DKD），and construct a personalized clinical prediction model to predict the risk of DKD in T2DM-CAS patients. Methods：A total of 883 patients with T2DM were selected in the study，and their basic characteristics，laboratory tests，auxiliary examinations and concomitant diseases of the patients were collected. LASSO regression was applied to screen the optimized variables by running cyclic coordinate descent. Multivariate logistic regression analyses were applied to build a prediction model，which incorporated the selected features. It relied on the receiver operating characteristic curve（ROC curve），calibration curves，and Hosmer-Lemeshow test to validate and evaluate the discrimination and calibration of the clinical prediction model；while the decision curve analysis（DCA） was used to evaluate its clinical validity. Results：A multivariable prediction model included diabetes duration，systolic blood pressure（SBP），fasting plasma glucose（FPG），triglycerides（TG），blood urea nitrogen（BUN），serum creatinine（Scr），cystatin C（Cys C） and diabetic retinopathy（DR）. This clinical prediction model demonstrated very good discrimination with an AUC of 0.831（95%CI=0.800-0.863），while the internal validation AUC was 0.825（95%CI=0.766-0.884）. The Hosmer-Lemeshow test showed very good fitting degree（P=0.822）. DCA showed the risk threshold of 30% and demonstrated a clini－cally effective prediction model. Conclusion：A Nomogram model with eight clinical predictor variables can be used to predict the risk of DKD in T2DM-CAS patients.