冠状动脉CT显示粥样硬化斑块列线图预测模型的构建及斑块稳定性的影响因素

Construction of nomogram prediction model of atherosclerotic plaque shown on coronary CT and the influencing factors of plaque stability

[目的]探讨胸部计算机断层成像(CT)显示冠状动脉粥样硬化斑块(CAP)列线图模型的构建及斑块稳定性的相关影响因素。[方法]将2020年1月—2021年10月接受CT扫描的心血管疾病高危人群的资料进行回顾性横断面研究,收集患者的基本情况(包括性别、年龄等)、既往史、合并疾病、血压、血生物化学指标等方面的信息。根据CT显示冠状动脉情况分为有CAP和无CAP两组。LASSO回归交叉验证法对数据降维,构建列线图模型并验证,决策曲线评估模型的应用价值,最优尺度回归分析评估CAP-CT值的影响因素。[结果]纳入分析的CAP患者240例,非CAP患者52例;调节参数(λ+1se)后,年龄、尿酸、估算的肾小球滤过率(eGFR)为非零特征的变量,有统计学意义的年龄及eGFR构建列线图模型,ROC曲线下面积为0.759(95%CI:0.691~0.828)、灵敏度为73.8%、特异度为71.2%;Hosmer-Lemeshow检验显示模型拟合度较好(χ2=11.846,P=0.158),预测准确性为82.2%;Bootstrap重抽样内部验证平均绝对误差为0.029,校正曲线与理想曲线基本拟合,决策曲线显示列线图模型在风险阈值为0.10~0.40时有良好的净获益;合并糖尿病和β2微球蛋白(β2-MG)是影响CAP-CT值的独立风险因素(均P<0.05),重要性分别为0.121、0.564。[结论]重视肾脏功能的保护及血糖达标将有利于降低CAP的形成及增加CAP稳定性,尤其在选择用药时,应该更多关注肾功能。

Aim To discuss the construction of the nomogram model of coronary atherosclerotic plaque(CAP)shown on chest computed tomography(CT)and the influencing factors of plaque stability.Methods The data of patients at high risk of cardiovascular disease that received CT examination from January 2020 to October 2021 were collected for a retrospective cross-sectional study.Basic information(including gender and age),medical history,complications,blood pressure,blood biochemical indicators,etc.of the patients were collected.According to their coronary artery conditions shown by CT imaging,the patients were divided into CAP group and CAP-free group.The data were then reduced by using LASSO regression cross-validation.The nomogram model was built and validated,and its application value was evaluated by the decision curve.In the end,the influencing factors of CAP-CT value were evaluated by optimal scaling regression.Results There were 240 patients with CAP and 52 patients without CAP included in the research.After adjusting the parameter(λ+1se),age,uric acid,estimated glomerular filtration rate(eGFR)were non-zero variables.The statistically significant age and eGFR were used to build the nomogram model.Area under ROC curve was 0.759(95%CI:0.691~0.828),sensitivity was 73.8%,specificity was 71.2%.Hosmer-Lemeshow test indicated good model fit(χ2=11.846,P=0.158).The prediction accuracy was 82.2%.Mean absolute error of Bootstrap resampling internal verification was 0.029.The calibration curve basically fits the ideal curve.According to the decision curve,the nomogram model shows good net benefits within the risk threshold of 0.10~0.40.Combined diabetes mellitus andβ2-microglobulin(β2-MG)were independent risk factors for CAP-CT values(both P<0.05),with the importance of 0.121 and 0.564,respectively.Conclusion Emphasis on the protection of renal function and glycemic compliance will help to reduce the formation of CAP and increase CAP stability.In particular,more attention should be paid to renal function when selecting medications.