个体化预测经桡动脉介入治疗术后桡动脉狭窄或闭塞风险的列线图模型构建

Establishment of an Individual Nomogram Model for Predicting the Risk of Radial Artery Stenosis or Occlusion after Transradial Artery Interventional Therapy

目的 构建预测经桡动脉介入治疗术后桡动脉狭窄或闭塞(radial artery stenosis or occlusion, RAO)风险的列线图模型。方法 选择2019年6月至2021年6月行经桡动脉介入治疗术的254例患者为研究对象,并按照7∶3分为建模队列(n=178)与验证队列(n=76)。根据是否发生RAO分为RAO组和非RAO组,收集患者临床资料。多因素logistic回归分析经桡动脉介入治疗术后RAO的影响因素。R3.6.3软件及rms程序包构建预测经桡动脉介入治疗术后RAO发生风险的列线图模型。采用校准曲线、受试者操作特征曲线(receiver operator characteristic curve, ROC curve)对Nomogram模型进行内部验证(建模队列)以及外部验证(验证队列)。结果 建模队列178例经桡动脉介入治疗术后患者发生RAO的有42例,未发生RAO的有136例,发生率23.60%。RAO组糖尿病患者比例及鞘管留置时间显著高于非RAO组,桡动脉内径及肝素使用量显著低于非RAO组(P均<0.05)。多因素logistic回归分析结果显示,桡动脉内径、肝素使用量、鞘管留置时间、糖尿病均是发生RAO的影响因素(P均<0.05)。列线图模型显示,桡动脉内径每降低0.5 mm,增加13.8分的权重;肝素使用量每降低10 IU/kg,增加5.1分的权重;鞘管留置时间每增加5 min,增加15.2分的权重;糖尿病增加21.3分的权重。Hosmer-Lemeshow拟合度检验显示,校准曲线斜率接近1,ROC曲线下面积为0.860(95%CI:0.800~0.907,χ^(2)=22.318,P=0.004),灵敏度、特异度分别为76.19%、86.03%,外部验证的曲线下面积为0.747(95%CI:0.670~0.815,χ^(2)=6.143,P=0.631),灵敏度、特异度分别为82.76%、57.02%。结论 糖尿病、桡动脉内径、肝素用量及鞘管留置时间均是经桡动脉介入治疗术后RAO发生的影响因素,基于这四项因素构建的列线图模型在用于个体化预测经桡动脉介入治疗术后RAO发生风险具有较高的临床意义。

Objective To construct a nomogram model to predict the risk of radial artery stenosis or occlusion(RAO) after transradial artery interventional therapy. Methods A total of 254 patients who underwent transradial artery interventional therapy in our hospital from June 2019 to June 2021 were selected as the study objects, and were divided into modeling cohort(n=178) and validation cohort(n=76) according to 7∶3. The patients were divided into RAO group and N-RAO group according to whether RAO occurred, and clinical data were collected;multivariate Logistic regression analysis was used to analyze the influencing factors of RAO after radial artery interventional therapy. R3.6.3 Software and rms package were used to build a nomogram model to predict the risk of RAO after transradial artery interventional therapy. Verification of nomogram internally(modeling queue) and externally(validation queue) using calibration curve and receiver operator characteristic curve(ROC curve). Results Among 178 patients who underwent radial artery interventional therapy, 42 had RAO, and 136 had no RAO, with an incidence of 23.60%;the proportion of diabetes and the sheath indwelling time in RAO group were obviously higher than those in N-RAO group, and the diameter of radial artery and the amount of used heparin were obviously lower than those in N-RAO group(all P<0.05);multivariate Logistic regression analysis showed that radial artery diameter, heparin dosage, sheath indwelling time and diabetes were all the influencing factors for RAO(all P<0.05);the nomogram model showed that the weight of 13.8 points increased for every 0.5 mm decrease in the diameter of the radial artery;the weight of 5. 1 points increased for every 10 IU/kg decrease in heparin dosage;the weight of 15. 2 points increased for every 5 min increase of sheath indwelling time;the weight of 21. 3 points increased for the presence of diabetes;H-L fit test showed that,χ^(2)= 22. 318,P = 0. 004,slope of calibration curve approached 1,area under ROC curve was 0. 860( 95% CI: 0. 800-0. 907),sensitivity and specificity were 76. 19% and 86. 03% respectively,external verification χ^(2)= 6. 143,P = 0. 631,the area under the ROC curve was 0. 747( 95% CI: 0. 670-0. 815),the sensitivity and specificity were 82. 76% and 57. 02% respectively. Conclusion Diabetes,radial artery diameter,heparin dosage and sheath indwelling time are all influencing factors of RAO after radial artery intervention. The nomogram model based on these four factors has high clinical significance in individualized prediction of the risk of RAO after radial artery intervention therapy.