冠状动脉CT血管成像放射组学模型识别易损斑块并预测心血管事件
本研究基于冠状动脉CT血管成像(CCTA)开发放射组学特征(RS),以血管内超声(IVUS)为参考标准定义易损斑块。RS在训练、验证、内部和外部测试集中AUC分别为0.81、0.75、0.80和0.77。
要点
- CCTA放射组学特征(RS)可用于无创识别IVUS定义的易损斑块。
- RS在训练集AUC 0.81、验证集0.75、内部测试集0.80、外部测试集0.77。
核心结论
- 基于CCTA的斑块放射组学特征能够检测与未来不良心脏结局风险增加相关的易损斑块。
研究设计与样本
- 开发集按7:3随机分为训练集和验证集,并在三家三级医院进行内部和外部测试以评估诊断性能。
- 前瞻性队列纳入2018年4月至2019年3月疑似冠心病患者,共708例患者1020个病变(平均年龄62岁±11,498例男性),用于评估RS对MACE的预测价值。
- 采用多变量Cox回归分析评估RS和传统解剖斑块特征(如节段累及评分)对MACE的预测价值。
诊断标准与关键数值
- RS在训练集、验证集、内部测试集和外部测试集的AUC分别为0.81、0.75、0.80和0.77。
临床应用与阅片要点
- CCTA放射组学特征可作为无创工具辅助识别易损斑块和预测MACE风险。
- RS≥1.07提示未来MACE风险增加,可考虑在报告中提示临床关注。
- RS结合传统解剖斑块特征(如节段累及评分)可能提供额外风险分层信息。
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
Background A noninvasive coronary CT angiography (CCTA)-based radiomics technique may facilitate the identification of vulnerable plaques and patients at risk for future adverse events. Purpose To assess whether a CCTA-based radiomic signature (RS) of vulnerable plaques defined with intravascular US was associated with increased risk for future major adverse cardiac events (MACE). Materials and Methods In a retrospective study, an RS of vulnerable plaques was developed and validated using intravascular US as the reference standard. The RS development data set included patients first undergoing CCTA and then intravascular US within 3 months between June 2013 and December 2020 at one tertiary hospital. The development set was randomly assigned to training and validation sets at a 7:3 ratio. Diagnostic performance was assessed internally and externally from three tertiary hospitals using the area under the curve (AUC). The prognostic value of the RS for predicting MACE was evaluated in a prospective cohort with suspected coronary artery disease between April 2018 and March 2019. Multivariable Cox regression analysis was used to evaluate the RS and conventional anatomic plaque features (eg, segment involvement score) for predicting MACE. Results The RS development data set included 419 lesions from 225 patients (mean age, 64 years ± 10 [SD]; 68 men), while the prognostic cohort included 1020 lesions from 708 patients (mean age, 62 years ± 11; 498 men). Sixteen radiomic features, including two shape features and 14 textural features, were selected to build the RS. The RS yielded a moderate to good AUC in the training, validation, internal, and external test sets (AUC = 0.81, 0.75, 0.80, and 0.77, respectively). A high RS (≥1.07) was independently associated with MACE over a median 3-year follow-up (hazard ratio, 2.01; P = .005). Conclusion A coronary CT angiography-derived radiomic signature of coronary plaque enabled the detection of vulnerable plaques that were associated with increased risk for future adverse cardiac outcomes. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by De Cecco and van Assen in this issue.
引用
- Chen, Pan, Wang, et al. A Coronary CT Angiography Radiomics Model to Identify Vulnerable Plaque and Predict Cardiovascular Events. Radiology. 2023. PMID: 36786701. doi:10.1148/radiol.221693
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。