下肢动脉硬化闭塞症介入治疗后下肢深静脉血栓形成列线图模型建立及验证

Establishment and verification of a nomogram model used for predicting lower extremity deep venous thrombosis after interventional treatment of lower extremity arteriosclerosis obliterans

目的 建立下肢动脉硬化闭塞症(LEASO)介入治疗后下肢深静脉血栓(DVT)形成列线图模型,并对模型进行外部验证。方法 以2020年1月至2022年12月日本血管外科学会建立的肢端缺血患者数据库(JCLIMB)中接受腔内治疗的LEASO患者434例为建模组,以本院2020年1月至2023年5月接受介入治疗的LEASO患者60例为验证组。通过电子病历系统获取患者性别、年龄、合并糖尿病、合并高血压、合并冠心病、体质量指数(BMI)、Fontaine分期、D-二聚体、凝血酶原时间(PT)、活化部分凝血活酶(APTT)、血小板淋巴细胞比值(PLR)和中性粒细胞淋巴细胞比值(NLR)等临床资料。使用R软件建立列线图模型,并以校准曲线、决策曲线(DCA)和受试者工作特征曲线评价预测模型。结果 建模组发生DVT与未发生DVT患者BMI、Fontaine分期、D-二聚体、PLR和NLR比较差异有统计学意义(P<0.05);多因素logistics回归分析结果发现,BMI、Fontaine分期、D-二聚体、PLR和NLR是LEASO患者并发DVT的独立影响因素(P<0.05)。基于logistics回归分析结果建立列线图模型,经校准曲线分析显示模型预测值与实际值基本一致,表明该列线图模型有较好的准确度。以净收益为纵坐标,以阈值为横坐标,None线代表所有LEASO患者均不会发生DVT,Nomogram model表示DVT实际发生曲线,阈值范围0.09~1.00,表明以该模型进行决策有较好的临床收益,具备临床实用性。ROC分析结果显示,本研究建立的列线图模型预测建模组、验证组LEASO患者发生DVT曲线下面积分别为0.912和0.834,灵敏度分别为96.50和和99.20,特异度分别为76.60和66.70。结论 以BMI、Fontaine分期、D-二聚体、PLR和NLR建立的LEASO患者并发DVT预测模型对于DVT发生有较好的预测价值,适宜在临床中推广应用。

Objective To establish a nomogram model which is used for predicting lower extremity deep venous thrombosis(DVT)after interventional treatment of lower extremity arteriosclerosis obliterans(LEASO),and to make an external validation of the model.Methods A total of 434 LEASO patients receiving intraluminal therapy between January 2020 and December 2022,who were retrieved from the database of patients with limb ischemia(JCLIMB)established by the Japanese Society of Vascular Surgery,were collected and used as modeling group,and other 60 LEASO patients,who received interventional treatment at the Hefei Binhu Hospital of China between January 2020 and May 2023,were collected and used as validation group.Through the electronic medical record system,the clinical data of the patients were obtained,which included gender,age,coexisting diabetes mellitus,coexisting hypertension,coexisting coronary heart disease,body mass index(BMI),Fontaine stage,D-dimer,prothrombin time(PT),activated partial thromboplastin(APTT),platelet lymphocyte ratio(PLR),and neutrophil-to-lymphocyte ratio(NLR).The nomogram model was established by using R software,and the prediction model was evaluated by the calibration curve,the decision curve analysis(DCA)and the receiver operating characteristic(ROC)curve.Results The differences in BMI,Fontaine stage,D-dimer,PLR and NLR between DVT patients and non-DVT patients were statistically significant(P<0.05).Multivariate logistic regression analysis showed that BMI,Fontaine stage,D-dimer,PLR and NLR were the independent influencing factors for DVT in LEASO patients(P<0.05).The nomogram model was established based on the results of the logistic regression analysis.The calibration curve analysis showed that the predicted values of the model was basically consistent with the actual values,indicating that this nomogram model carried a good accuracy.Taking the net income as the ordinate and the threshold as the abscissa,the none line represented that DVT would not occur in all LEASO patients.The nomogram model represented the actual curve of DVT,and the threshold range was 0.09-1.00,indicating that using this model for decision-making had good clinical benefits,suggesting the clinical utility of this model.ROC analysis revealed that using this nomogram model,which was established in this study,to predict the area under the DVT curve(AUC)of LEASO patients,the AUC in the modeling group and in the validation group was 0.912 and 0.834 respectively,the sensitivity was 96.50 and 99.20 respectively,and the specificity was 76.60 and 66.70 respectively.Conclusion The prediction model of DVT in LEASO patients that is established on the BMI,Fontaine stage,D-dimer,PLR and NLR has a good predictive value for the occurrence of DVT,therefore,this model is worth popularizing in clinical practice.(J Intervent Radiol,2024,33:780-784)