基于监测和最终结果数据库的冠心病PCI术后冠脉微循环损伤预测模型的建立与验证

Establishment and verification of prediction model of coronary microcirculation injury after percutaneous coronary intervention based on SEER database

目的基于监测和最终结果(SEER)数据库构建冠心病经皮冠状动脉介入治疗(PCI)术后冠脉微循环损伤(CMI)预测模型,并进行临床验证。方法选取SEER数据库中2021年1月至2022年12月在南阳市中心医院接受PCI术治疗的106例冠心病患者作为建模样本,收集其临床资料,统计术后CMI发生情况,根据术后是否发生CMI分为损伤组47例和未损伤组59例。另选取2023年1~6月我院收治的62例冠心病患者作为外部验证样本,根据PCI术后是否发生CMI分为损伤患者25例与未损伤患者37例。比较建模样本中两组患者的临床资料,采用Logistic多因素回归分析冠心病PCI术后CMI的影响因素,构建预测模型,采用受试者工作特征曲线(ROC)及曲线下面积(AUC)评价模型预测价值。结果损伤组患者的周围血管病史、高血压史、糖尿病史、吸烟史和饮酒史的比例明显高于未损伤组,差异均有统计学意义(P<0.05);损伤组患者的冠脉病变长度、冠脉狭窄率、预扩张次数、预扩张时间、同型半胱氨酸(Hcy)、D-二聚体(D-D)水平明显长/高于未损伤组,差异均有统计学意义(P<0.05);经Logistic多因素回归分析结果显示,高血压史、糖尿病史、冠脉病变长度、冠脉狭窄率、预扩张次数、预扩张时间、Hcy、D-D均为CMI发生的独立预测因子(P<0.05);根据上述独立预测因子构建预测模型,该预测模型拟合度较好,校准度为0.892,一致性指数为0.917,预测冠心病PCI术后CMI的AUC为0.947(95%CI:0.910~0.985),敏感度为95.74%,特异度为81.36%。外部验证显示,该预测模型预测冠心病PCI术后CMI的AUC为0.969(95%CI:0.939~1.000),敏感度为96.00%,特异度为86.49%。结论根据高血压史、糖尿病史、冠脉病变长度、冠脉狭窄率、预扩张次数、预扩张时间、Hcy、D-D等资料构建冠心病PCI术后CMI的预测模型预测价值可靠,能为临床防治提供指导信息。

Objective To establish a prediction model of coronary microcirculation injury(CMI)after percutaneous coronary intervention(PCI)based on the Surveillance,Epidemiology,and End Results(SEER)database,and to conduct clinical verification.Methods A total of 106 patients with coronary heart disease who underwent PCI treatment at Nanyang Central Hospital from January 2021 to December 2022 in the SEER database were selected as the modeling sample.Their clinical data were collected,and the incidence of CMI after surgery was statistically analyzed.According to whether CMI occurred after surgery,they were divided into an injury group of 47 cases and a non-injury group of 59 cases.A total of 62 patients with coronary heart disease from January 2023 to June 2023 were selected as external validation samples,and were divided into injury patients(25 cases)and non-injury patients(37 cases)according to whether CMI occurred after PCI.The clinical data of the two groups of patients in the modeling sample were compared,and the factors influencing CMI after PCI in coronary heart disease were analyzed using logistic multivariate regression analysis.A prediction model was constructed,and the ROC curve and area under the curve(AUC)were used to evaluate the predictive value of the model.Results The proportions of peripheral vascular disease history,hypertension history,diabetes history,smoking history,and drinking history in the injury group were significantly higher than those in the non-injury group(P<0.05).The length of coronary lesions,coronary stenosis rate,pre-dilation frequency,pre-dilation time,homocysteine(Hcy),D-dimer(D-D)levels in the injury group were significantly longer/higher than those in the non-injury group(P<0.05).Logistic regression analysis showed that hypertension history,diabetes history,coronary lesion length,coronary stenosis rate,pre-dilation times,pre-dilation time,Hcy,D-D were independent predictors of CMI(P<0.05).Based on the independent predictive factors mentioned above,a prediction model with a good fit was constructed,with a calibration of 0.892,a consistency index of 0.917,and an AUC of 0.947(95%CI:0.910-0.985)for predoi dicting postoperative CMI in coronary heart disease after PCI,a sensitivity of 95.74%,and a specificity of 81.36%.External validation showed that the AUC of this predictive model for predicting CMI after coronary heart disease PCI was 0.969(95%CI:0.939-1.000),with a sensitivity of 96.00%and a specificity of 86.49%.Conclusion The prediction model of CMI after PCI based on hypertension history,diabetes history,length of coronary lesions,rate of coronary stenosis,pre-dilation times,pre-dilation time,Hcy,D-D is reliable and can provide guidance information for clinical prevention and treatment.