心脏移植术后早期急性肾损伤风险的动态列线图预测模型构建和评价  

Construction and evaluation of dynamic nomogram model prediction model for early acute renal injury risk after heart transplantation

作　　者：陈烨[1,2] 蒋英硕 朱新悦[1,2] 杜文心 陈鑫 娄晟[1,2] 孙建国 朱君荣[1,2] CHEN Ye;JIANG Yingshuo;ZHU Xinyue;DU Wenxin;CHEN Xin;LOU Sheng;SUN Jian-guo;ZHU Junrong(Department of Pharmacy,Nanjing First Hospital,China Pharmaceutical University,Nanjing 210006,Jiangsu,China;Department of Pharmacy,Nanjing First Hospital,Nanjing Medical University,Nanjing 210006,Jiang-su,China;Department of Cardiothoracic Surgery,Nanjing First Hospital,Nanjing Medical University,Nanjing 210006,Jiangsu,China;Key Laboratory of Drug Metabolism and Pharmacokinetics,China Pharmaceutical University,Nanjing 210009,Jiangsu,China)

机构地区：[1]中国药科大学南京市第一医院药学部,江苏南京210006 [2]南京医科大学附属南京医院(南京市第一医院)药学部,江苏南京210006 [3]南京医科大学附属南京医院(南京市第一医院)心胸外科,江苏南京210006 [4]中国药科大学药物代谢动力学重点实验室,江苏南京210009

出　　处：《中国临床药理学与治疗学》2024年第11期1272-1279,共8页Chinese Journal of Clinical Pharmacology and Therapeutics

基　　金：江苏省老年医学学会2023年度临床药学专项基金科研项目(JGS2023ZDYX003)。

摘　　要：目的:分析并筛选心脏移植(heart transplant,HT)术后发生急性肾损伤的危险因素,建立预测HT术后早期发生急性肾损伤(acute kidney injury,AKI)的动态列线图预测模型。方法:回顾性分析2012年10月至2024年6月在南京市第一医院接受HT术受者的临床资料,根据术后7 d内是否发生AKI分为AKI组和非AKI组,按8∶2划分训练集和测试集。采用Lasso回归和多因素Logistic回归筛选影响因素,通过R建立可视化的动态列线图模型。采用Bootstrap法重复抽样1 000次,对模型进行内部验证。使用受试者工作特征曲线下面积(AUC-ROC)、校准曲线、Hosmer-Lemeshow拟合优度检验评价模型的准确性和区分度,并将列线图模型同Cleveland评分进行对比。结果:多因素Logistic回归结果显示,房颤史(OR=9.647,95%CI=1.961~47.470)、术中血管活性药物评分(OR=1.094,95%CI=1.012~1.183)、术中输入红细胞或血浆(OR=10.200,95%CI=1.727~60.238)、术后中心静脉压(OR=1.548,95%CI=1.186~2.021)和术后使用万古霉素(OR=25.082,95%CI=2.122~296.417)是心脏移植后急性肾损伤(HT-AKI)的5个独立危险因素,动态列线图模型在测试集中AUC=0.842(95%CI:0.676~0.971),校准曲线为斜率接近1的斜线,Brier评分为0.173;Hosmer-Lemeshow拟合优度检验(χ2=5.658,P=0.685)提示模型预测性能较好。且本模型显示出比Cleveland评分更好的区分度。结论:本研究发现了影响HTAKI的术前、术中、术后的重要危险因素,建立的动态列线图模型能够较准确识别AKI高风险人群,并方便临床使用。AIM:To analyze and screen the risk factors for acute kidney injury(AKI)following heart transplantation(HT),and to establish a dynamic no-mograms prediction model to forecast early AKI af-ter HT.METHODS:A retrospective analysis was con-ducted on clinical data from HT recipients at Nan-jing First Hospital from October 2012 to June 2024.Patients were divided into AKI and non-AKI groups based on AKI occurrence within 7 days post-sur-gery,with a 8:2 ratio for training and testing sets.Lasso regression and multivariable logistic regres-sion were used to select influencing factors.A dy-namic nomogram model was visualized using R.In-ternal validation was performed using 1000 boot-strap samples.Model accuracy and discrimination were evaluated using the area under the receiver operating characteristic curve(AUC-ROC),calibra-tion curves,and the Hosmer-Lemeshow goodness-of-fit test.The nomogram model was compared with the Cleveland score.RESULTS:The results of a multivariable logistic regression indicate that a his-tory of atrial fibrillation(OR=9.647,95%CI=1.961-47.470),vasoactive inotropic score(OR=1.094,95%CI=1.012-1.183),intraoperative transfusion of red blood cells or plasma(OR=10.200,95%CI=1.727-60.238),postoperative central venous pressure(OR=1.548,95%CI=1.186-2.021),and postoperative use of vancomycin(OR=25.082,95%CI=2.122-296.417)are independent risk factors for HT-AKI.The dynamic nomogram model achieved an AUC of 0.842(95%CI:0.676-0.971)in the test set,with a calibration plot showing a slope close to 1 and a Brier score of 0.173.The Hosmer-Lemeshow good-ness-of-fit test(χ2=5.658,P=0.685)suggests good predictive performance of the model.Moreover,this model demonstrates superior discriminative ability compared to the Cleveland score.CONCLU-SION:This study identified preoperative,intraoper-ative,and postoperative risk factors influencing the occurrence of HT-AKI.The developed dynamic no-mogram model accurately identifies high-risk indi-viduals for early HT-AKI and is convenient for clini-cal use.

关 键 词：心脏移植 急性肾损伤 预测模型 动态列线图 

分 类 号：R969[医药卫生—药理学]