结缔组织病相关间质性肺疾病合并肺动脉高压的预后模型开发与验证
Development and validation of a prognostic model for connective tissue disease-associated interstitial lung disease complicated with pulmonary hypertension
樊铭薇 1蒋天赐 1李鹏飞 1王羽 1赵文静 1吴如昊 1田晓英 1张梦婷 1程哲1
作者信息
- 1. 郑州大学第一附属医院呼吸与危重症医学科(河南郑州 450052)
- 折叠
摘要
目的 调查影响结缔组织病相关间质性肺疾病合并肺动脉高压(connective tissue disease-associated interstitial lung disease complicated with pulmonary hypertension,CTD-ILD-PH)的预后独立危险因素,使用机器学习构建预测1、3、5年死亡风险的列线图模型,为临床诊疗该病提供一定的证据.方法 对2011年2月—2021年6月在郑州大学第一附属医院接受治疗的CTD-ILD-PH患者进行筛查.将最小绝对收缩选择算子(least absolute shrinkage and selection operator,Lasso)、单因素 Cox 回归和多因素 Cox 回归分析相结合,筛选CTD-ILD-PH患者预后独立危险因素,然后构建列线图预后模型.采用1 000次重抽样的Bootstrap法进行内部验证,采用受试者操作特征曲线(receiver operating characteristic curve,ROC曲线)和一致性指数(Harrell's C指数)评估模型的预测效能,校正曲线对模型拟合进行评价,决策曲线分析(decision curve analysis,DCA)评估模型的临床实用性,测试集进行外部验证.结果 研究纳入313例患者,随访期间108例死亡.采用Lasso-Cox方法,筛选出白蛋白、丙氨酸转氨酶(alanine transaminase,ALT)、红细胞体积分布宽度(red cell volume distribution width,RDW)、年龄、吸烟史、农村及肺动脉收缩压独立危险因素,训练集中Harrell's C指数为0.802,ROC曲线下面积为0.880(95％CI0.833～0.928).在内部验证中,平均Harrell's C指数为0.791.校准曲线结果表明预测值与实际观测结果之间存在高度一致性.DCA分析证实模型具有良好的实用性.外部验证结果表面模型具有良好的预测效能和临床实用性.结论 白蛋白降低、ALT升高、RDW升高、高龄、吸烟史、农村和肺动脉收缩压较高是CTD-ILD-PH患者的预后独立危险因素.本CTD-ILD-PH患者1、3、5年病死率的预后模型可以为今后CTD-ILD-PH的死亡风险评估提供一定参考价值.
Abstract
Objective To investigate independent prognostic factors influencing the prognosis of connective tissue disease-associated interstitial lung disease with pulmonary hypertension(CTD-ILD-PH),and construct a nomogram model using machine learning to predict 1-,3-and 5-year mortality risks,providing evidence for clinical diagnosis and treatment.Methods Patients diagnosed with CTD-ILD-PH and treated at the First Affiliated Hospital of Zhengzhou University from February 2011 to June 2021 were screened.The least absolute shrinkage and selection operator(Lasso),univariate Cox regression,and multivariate Cox regression analyses were combined to identify independent prognostic factors for CTD-ILD-PH patients.A novel nomogram prognostic model was constructed and internally validated using 1 000 bootstrap resamples.The receiver operating characteristic(ROC)curve and Harrell's C-index assessed the predictive performance of the model.Calibration curves evaluated the model fit.Decision curve analysis(DCA)assessed the clinical utility of the model,and external validation was conducted using a separate test set.Results The study included 313 patients,with 108 deaths observed during the follow-up.Using the Lasso-Cox method,albumin,alanine aminotransferase(ALT),red cell volume distribution width(RDW),age,smoking history,rural residence,and pulmonary artery systolic pressure were identified as independent prognostic factors.The Harrell's C-index in the training set was 0.802,and the area under ROC curve was 0.880(95％CI 0.833-0.928).Internal validation showed an average Harrell's C-index of 0.791.Calibration curves indicated high consistency between predicted and observed results.DCA confirmed the model's good clinical utility.External validation results demonstrated the model's favorable predictive performance and clinical utility.Conclusions Our research suggest that lower albumin,elevated ALT,elevated RDW,advanced age,smoking history,rural areas and higher pulmonary artery systolic blood pressure are independent prognostic risk factors for patients with CTD-ILD-PH.In this study,a prognostic model was developed for the first time to predict 1-,3-and 5-year mortality in CTD-ILD-PH patients,which provides some reference value for future mortality risk assessment in CTD-ILD-PH.
关键词
肺动脉高压/结缔组织病/间质性肺疾病/预后模型Key words
Pulmonary hypertension/connective tissue diseases/interstitial lung disease/prognostic model引用本文复制引用
基金项目
国家自然科学基金面上项目(82170037)
出版年
2024
NETL
NSTL
万方数据