含CT影像特征的主动脉夹层动脉瘤患者术后附壁血栓形成的列线图模型构建

Construction of containing CT imaging features Nomograms model of postoperative intraluminal thromus formation in patients with aortic dissective aneurysm

目的探讨含CT影像特征的主动脉夹层动脉瘤(ADA)患者术后附壁血栓(ILT)形成的列线图模型构建。方法回顾性选取2020年5月至2022年2月济宁医学院附属医院诊治的120例Stanford B型ADA患者,均行覆膜支架腔内修复术及多层螺旋CT(MSCT)检查。按7∶3比例将患者随机分为建模人群(84例)、验证人群(36例)。在建模人群中统计术后1个月内ILT形成情况,通过单因素和多因素Logistic回归模型分析ADA患者术后ILT形成的影响因素,并构建术后ILT形成的列线图预测模型,分别在建模人群及验证人群中对预测模型进行内部及外部验证。结果在建模人群中,术后有ILT形成23例(27.38%),其中5例ILT位于主动脉弓部,18例ILT位于腹主动脉。单因素分析结果显示,术后ILT形成组与无ILT形成组性别构成、年龄、体质量指数(BMI)、吸烟、饮酒、高血压、高脂血症、破口直径、破口距左锁骨下动脉距离、重要分支受累、动脉瘤壁不均匀增厚、平扫呈低密度、手术时机比较差异无统计学意义(P>0.05);两组糖尿病、内壁不规则、伴有钙化斑块、术后抗凝治疗及B型脑钠肽(BNP)、纤维蛋白原(Fib)、D-二聚体(D-D)、C反应蛋白(CRP)水平[43.48%(10/23)比11.48%(7/61)、86.96%(20/23)比57.38%(35/61)、91.30%(21/23)比62.30%(38/61)、21.74%(5/23)比57.38%(35/61)、(523.60±128.74)ng/L比(271.83±109.65)ng/L、(3.82±0.96)g/L比(2.85±0.83)g/L、(601.37±75.62)μg/L比(389.20±68.79)μg/L、(0.63±0.19)mg/L比(0.48±0.15)mg/L]比较差异有统计学意义(P<0.05)。Logistic多因素分析结果显示,糖尿病、内壁不规则、伴有钙化斑块、术后抗凝治疗和BNP、Fib、D-D、CRP水平均为Stanford B型ADA患者术后ILT形成的影响因素(P<0.05)。根据影响因素构建Stanford B型ADA患者覆膜支架腔内修复术后ILT形成的列线图预测模型,该模型在建模人群、验证人群的一致性指数分别为0.903、0.894,具有良好的区分度,且在预测Stanford B型ADA患者覆膜支架腔内修复术后ILT形成方面具有良好的临床效用。结论通过建立含CT影像特征的Stanford B型ADA患者覆膜支架腔内修复术后ILT形成的列线图预测模型,有助于临床早期筛选并识别ILT形成的高危患者。

Objective To investigate the construction of containing CT imaging features Nomograms model of postoperative intraluminal thromus(ILT)formation in patients with aortic dissective aneurysm(ADA).Methods One hundred and twenty patients with Stanford type B ADA treated with overlapping stent endoluminal repair and multilayer spiral CT(MSCT)examination in Affiliated Hospital of Jining Medical College from May 2020 to February 2022 were selected.The patients were divided into the modeling population(84 patients)and the validation population(36 patients)according to a 7∶3 ratio.The factors influencing postoperative ILT formation in ADA patients were analyzed by univariate and Logistic multifactor regression models,and the prediction model of postoperative ILT formation was constructed based on the influencing factors.Results In the modeled population,the rate of ILT formation within 1 month after luminal repair with overlapping stents was 27.38%(23/84),including 5 cases in the aortic arch and 18 cases in the abdominal aorta.In the modeled population,the results of univariate analysis showed that the sex,age,body mass index(BMI),smoking,drinking,hypertension,hyperlipidemia,rupture diameter,rupture distance from left subclavicular artery,involvement of important branches,uneven thickening of aneurysm wall,low density on plain scan and operation timing between the ILT formation group and non-ILT formation group had no statistically significant(P>0.05).The diabetes mellitus,irregular inner wall,calcified plaque,postoperative anticoagulant therapy,B-type brain natriuretic peptide(BNP),fibrinogen(Fib),D-dimer(D-D)and C-reactive protein(CRP)between the two groups had statistical differences:43.48%(10/23)vs.11.48%(7/61),86.96%(20/23)vs.57.38%(35/61),91.30%(21/23)vs.62.30%(38/61),21.74%(5/23)vs.57.38%(35/61),(523.60±128.74)ng/L vs.(271.83±109.65)ng/L,(3.82±0.96)g/L vs.(2.85±0.83)g/L,(601.37±75.62)μg/L vs.(389.20±68.79)μg/L,(0.63±0.19)mg/L vs.(0.48±0.15)mg/L,P<0.05.The results of Logistic multifactor regression analysis showed that diabetes mellitus,irregular inner wall,calcified plaque,postoperative anticoagulant therapy and BNP,Fib,D-D CRP levels were influential factors for postoperative ILT formation in Stanford type B ADA patients(P<0.05).The C-index of the model was 0.903 and 0.894 for the modeled and validated populations,respectively,which had good discrimination and was good at predicting ILT formation after operation in Stanford type B ADA patients.The model had good clinical utility in predicting postoperative ILT formation in Stanford B ADA patients.Conclusions The Nomograms model can help to screen and identify patients with high risk of ILT formation at an early clinical stage.