冠脉CTA特征联合临床危险因素预测TAVR术后不良结局
该多中心队列纳入406例TAVR术前CCTA患者，随访中位768天，MACE发生12.6%、全因死亡13%。联合模型（年龄、肌酐、SSS、LAD-FAI）预测MACE的C-index为0.756、1年AUC 0.863、3年AUC 0.814，优于临床模型，SHAP显示LAD-FAI最重要。
要点
- LAD-FAI是TAVR术后MACE最重要预测因子，阅片时应常规测量。
- 联合临床因素与CCTA特征（SSS、LAD-FAI）可提升预测效能。
- 联合模型C-index 0.756、1年AUC 0.863、3年AUC 0.814。
核心结论
- 联合CCTA特征与临床因素可提高TAVR术后不良结局的预测效能。
- LAD-FAI是TAVR术后MACE最重要的预后因素。
研究设计与样本
- 多中心队列研究，纳入406例TAVR术前CCTA候选者。
- 主要终点为MACE，次要终点为全因死亡，中位随访768天。
- 采用单/多因素Cox分析构建临床模型、影像模型和联合模型。
- 模型效能用时间依赖AUC和Harrell C-index评价，嵌套模型用似然比χ2检验比较。
- 联合模型可解释性用SHAP方法评价。
诊断标准与关键数值
- MACE独立临床预测因素：年龄（HR 1.082，P<0.001）和血清肌酐（HR 1.010，P<0.001）。
- MACE独立影像预测因素：节段狭窄评分SSS（HR 1.077，P<0.001）和LAD-FAI（HR 1.070，P<0.001）。
- 联合模型在测试队列中C-index为0.756（临床模型0.713）。
- 联合模型1年AUC为0.863（临床模型0.779），3年AUC为0.814（临床模型0.760）。
- 似然比检验证实联合模型拟合更优（P=0.020）。
临床应用与阅片要点
- TAVR术前CCTA除评估瓣环和冠脉狭窄外，应测量LAD-FAI并计算SSS。
- 将年龄、血清肌酐与CCTA特征联合可用于术后MACE风险分层。
- LAD-FAI作为最重要预后因子，应在报告中提示其预测价值。
局限性
- 原文未明确报告研究局限性。
- 研究为多中心队列但未报告外部验证队列，联合模型的可推广性需进一步验证。
以上解读依据论文摘要（全文未开放获取或许可不允许翻译转载）。
英文原文（摘要）· 点击展开
RATIONALE AND OBJECTIVES: To explore the value of coronary computed tomography angiography (CCTA)-derived features combined with clinical factors for predicting adverse outcomes after transcatheter aortic valve replacement (TAVR).
MATERIALS AND METHODS: A multicenter cohort of 406 pre-CCTA TAVR candidates were enrolled. The primary endpoint was major adverse cardiac events (MACE), and the secondary endpoint was all-cause mortality. Univariable and multivariable Cox analyses were used to identify the independent clinical and CCTA-derived predictors for post-TAVR MACE, and the clinical model, the imaging model and the combined model were constructed. Model performance was evaluated with time-dependent area under the receiver operating characteristic (ROC) curve (time-dependent area under the curve (AUC)) and Harrell's concordance index (C-index). Likelihood ratio χ2 test was used to compare the nested models. The interpretability of the combined model was evaluated by using SHapley Additive exPlanations (SHAP) method.
RESULTS: The median follow-up was 768 days. MACE occurred in 12.6%, all-cause mortality in 13%. Univariable and multivariate analysis identified age (hazard ratio (HR) 1.082, P < 0.001) and serum creatinine (HR 1.010, P < 0.001) as independent clinical predictors, segment stenosis score (SSS, HR 1.077, P < 0.001) and left anterior descending (LAD)-fat attenuation index (FAI), HR 1.070, P < 0.001) as independent imaging predictors of MACE. The combined model outperformed the clinical model with higher C-index (0.756 vs 0.713), 1-year AUC (0.863 vs 0.779) and 3-year AUC (0.814 vs 0.760) in test cohort. Likelihood ratio test confirmed better fit of the combined model (P = 0.020). SHAP analysis showed LAD-FAI was the most important prognostic factor for post-TAVR MACE.
CONCLUSION: The combined model incorporating CCTA features and clinical factors improves predictive efficacy for adverse outcomes in patients after TAVR.
引用
- Zhu H, Deng Y, Zhang S, et al. Combining Coronary CTA Features with Clinical Risk Factors for Predicting Adverse Clinical Outcomes Following Transcatheter Aortic Valve Replacement. Acad Radiol. 2026. PMID: 42031596. doi:10.1016/j.acra.2026.04.001
供执业医师工作参考，不替代临床判断；数值与分级以指南原文为准。