Wallace Nicolai, Wong Karren, Desmarais Taylor, Park Jin Joo, Krummen David, Lim Woo-Hyun, Campagnari Claudio, Yagil Avi, Greenberg Barry
Cardiology Department, University of California-San Diego, La Jolla, California, USA.
Cardiology Department, University of California-San Diego, La Jolla, California, USA; Cardiovascular Center, Division of Cardiology, Department of Internal Medicine, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Seongnam, Republic of Korea.
J Am Coll Cardiol. 2025 Aug 5;86(5):374-395. doi: 10.1016/j.jacc.2025.05.061.
Implantable cardioverter-defibrillators (ICDs) protect patients from sudden cardiac death (SCD). Landmark trials demonstrating their efficacy for primary prevention in patients with heart failure (HF) used reduced left ventricular ejection fraction (LVEF) as a major inclusion criterion and current recommendations for ICD implantation rely on this variable in patient selection. However, contemporary medical management has reduced the risk of SCD in patients with reduced LVEF so that an increasingly large proportion of this population never requires the protection offered by the device. Although SCD is the major cause of cardiovascular mortality in HF patients with preserved LVEF, ICDs are not recommended for primary prevention in this subset of the population. Advances in patient management, diagnostic testing, and data processing over the past 30 years have made it apparent that recommendations for ICD use for primary prevention of SCD are no longer optimal. This review summarizes the declining incidence of SCD and reasons for the widening gap between risk of SCD and current guideline recommendations for use of ICDs in the HF population. It discusses limitations in our ability to predict risk of SCD that need to be addressed and the potential impact of ongoing clinical trials on recommendations for ICD use for primary prevention of SCD. Patient-related variables including those available from diagnostic tests that could be used to generate prediction models that more accurately identify magnitude of risk of SCD in individual patients are identified. The use of artificial intelligence processing of data from diagnostic tests to facilitate and standardize extraction of predictive variables and the use of machine learning algorithms for developing risk prediction models are discussed. The review concludes by describing a dynamic approach for generating novel risk prediction models that could better align risk of SCD with the benefits of ICD implantation in patients with HF and that could evolve over time as additional treatment strategies that alter risk of SCD are introduced in the future.
植入式心脏复律除颤器（ICD）可保护患者免受心源性猝死（SCD）。具有里程碑意义的试验证明了其在心力衰竭（HF）患者一级预防中的疗效，这些试验将降低的左心室射血分数（LVEF）作为主要纳入标准，目前ICD植入的建议在患者选择中依赖于这一变量。然而，当代医学管理已降低了LVEF降低患者发生SCD的风险，以至于该人群中越来越大比例的人从未需要该设备提供的保护。尽管SCD是LVEF保留的HF患者心血管死亡的主要原因，但不建议在这部分人群中进行ICD一级预防。过去30年中患者管理、诊断测试和数据处理方面的进展表明，ICD用于SCD一级预防的建议已不再是最佳选择。本综述总结了SCD发病率的下降以及SCD风险与当前HF人群ICD使用指南建议之间差距扩大的原因。它讨论了我们预测SCD风险能力的局限性，这些局限性需要得到解决，以及正在进行的临床试验对ICD用于SCD一级预防建议的潜在影响。确定了与患者相关的变量，包括可从诊断测试中获得的变量，这些变量可用于生成预测模型，以更准确地识别个体患者SCD风险的大小。讨论了使用人工智能处理诊断测试数据以促进和标准化预测变量的提取，以及使用机器学习算法开发风险预测模型。综述最后描述了一种动态方法，用于生成新的风险预测模型，该模型可以更好地使SCD风险与HF患者ICD植入的益处相匹配，并且随着未来引入改变SCD风险的其他治疗策略，该模型可能会随着时间的推移而演变。
