中文摘要（关键词）

第一部分 代谢指标在心脏性猝死一级预防患者危险分层中的预测价值

背景：尽管心血管疾病的诊疗已取得显著进展，心脏性猝死（Sudden cardiac death, SCD）因其突发性和高风险人群识别不足，仍面临重大挑战。现有危险分层工具，如左室射血分数、心电图指标及血清生物标志物，在预测准确性、适用范围及中低风险人群识别方面存在局限性。代谢指标（血糖、血脂、胰岛素抵抗等）不仅是代谢综合征的核心特征，也是心血管疾病发生与进展的关键危险因素。研究表明，代谢异常与动脉粥样硬化、心力衰竭、心律失常等心血管事件密切相关，但其在SCD高风险人群中的预测价值尚不明确。本研究拟在植入型心律转复除颤器（Implantable cardioverter defibrillator, ICD）一级预防患者中，探讨非酒精性脂肪性肝病（Non-alcoholic fatty liver disease, NAFLD）和甘油三酯葡萄糖指数（Triglyceride-glucose index, TyG指数）对室性心律失常及心血管事件的预测价值，以优化SCD风险分层。

方法：本研究回顾性纳入符合一级预防指征的ICD植入患者，术前均完成非增强CT及空腹血糖、甘油三酯检测。NAFLD通过非增强CT的肝脏和脾脏衰减值比值（Liver-to-spleen attenuation ratio, L/S比值）诊断。主要终点为ICD记录的持续性室性心动过速和/或心室颤动，次要终点为心源性死亡、心脏移植及心力衰竭再住院的复合事件。通过竞争风险回归模型和累积发生率函数估算事件发生率，并利用Fine-Gray检验分析TyG指数和NAFLD与室性心律失常及心血管事件的风险关联。

结果：本研究最终纳入649例患者，平均年龄58.5±12.5岁，其中男性490例（75.5%），缺血性心肌病患者271例（41.8%），存在NAFLD患者159例（24.5%）。平均随访34.3±11.9月随访后，181例（27.9%）患者出现室性心律失常事件，104例（16.0%）患者出现心血管事件。限制性立方样条提示，随着TyG指数升高，发生主要终点和次要终点的风险概率随之上升；随着L/S比值升高，终点事件的发生风险概率随之下降。校正混杂因素后，Fine-Gray竞争风险模型提示TyG指数在室性心律失常事件发生的风险比（Hazard ratio, HR）为2.46（95%置信区间，95% CI：1.73-3.49），且TyG指数最高组相较于最低组的事件发生的HR为2.86（95% CI：1.77-4.63）；心血管事件发生的HR为2.35（95% CI：1.48-3.75），TyG指数最高组相较于最低组的事件发生的HR为3.01（95% CI：1.53-5.93）。NAFLD在室性心律失常事件发生的HR为4.29（95% CI：2.99-6.16）；心血管事件发生的HR为2.62（95% CI：1.64-4.20）。

结论：在一级预防植入ICD患者中，TyG指数和NAFLD与发生室性心律失常和心血管事件存在显著的相关性，其或可作为新型标志物为SCD高危群体的风险分层提供临床价值。

第二部分 代谢指标在心脏性猝死二级预防患者中预测室性心律失常和心血管事件的价值

背景：心脏性猝死（Sudden cardiac death, SCD）二级预防患者发生恶性室性心律失常风险较高，尽管植入型心律转复除颤器（Implantable cardioverter defibrillator, ICD）已被广泛证实能够显著改善临床预后，但针对该类患者发生室性心律失常和心血管事件的风险预测和干预研究相对有限，且尚未得出明确结论。近年来，代谢指标在心血管疾病的发生、发展及预后评估中的作用逐渐引起了广泛关注。代谢异常不仅与心血管疾病的病理生理过程密切相关，还可能为心血管疾病的预测和干预提供新的思路。本研究旨在探讨非酒精性脂肪性肝病（Non-alcoholic fatty liver disease, NAFLD）和甘油三酯葡萄糖指数（Triglyceride-glucose index, TyG指数）在SCD二级预防人群中的预测价值。

方法：本研究为回顾性研究，纳入符合二级预防指征的ICD植入患者。主要终点事件为ICD记录的持续性室性心动过速和/或心室颤动；次要终点事件为心源性死亡、心脏移植及心力衰竭再住院的复合事件。通过连续变量与分类变量两种维度，使用Fine-Gray检验分析TyG指数和NAFLD与室性心律失常及心血管事件的风险关联。

结果：本研究最终纳入265例患者，平均年龄为58.5±13.0岁，其中男性188例（70.9%），缺血性心肌病患者129例（48.7%），NAFLD患者63例（23.8%）。平均随访36.1±12.8个月后，共有84例（31.7%）患者发生室性心律失常事件，45例（17.0%）患者发生心血管事件。Fine-Gray竞争风险模型分析结果显示，TyG指数在室性心律失常事件发生中的风险比（Hazard ratio，HR）为4.68（95%置信区间，95% CI：2.78-7.89），且TyG指数最高组相较于最低组的HR为5.44（95% CI：1.77-4.63）；心血管事件发生的HR为3.38（95% CI：1.39-8.23），TyG指数最高组相较于最低组的HR为3.67（95% CI：1.35-9.95）。NAFLD与室性心律失常事件发生的HR为2.97（95% CI：1.49-5.92）；与心血管事件发生的HR为1.10（95% CI：0.47-2.58）。

结论：在二级预防植入ICD的患者中，TyG指数与室性心律失常及心血管事件的发生存在显著的正相关性，NAFLD与室性心律失常事件的发生呈现显著相关性，且二者均呈现线性关系。基于这些发现，降低TyG指数和改善NAFLD可能有助于减少ICD二级预防患者发生室性心律失常和心血管事件的风险。

第三部分 代谢指标对植入ICD患者发生室性心律失常和心血管事件的风险模型的构建

背景：植入式心律转复除颤器（Implantable cardioverter defibrillator, ICD）是预防心脏性猝死（Sudden cardiac death, SCD）的重要治疗手段。然而，ICD植入后患者仍面临室性心律失常和心血管事件的高风险。尽管传统临床指标（如左室射血分数、既往心肌梗死史等）已被广泛应用于风险分层，但其预测能力有限，难以精准识别高危患者。代谢紊乱不仅与心血管疾病的发生和进展密切相关，但其在植入ICD患者中对室性心律失常和心血管事件的预测效能尚未明确。本研究旨在通过代谢指标构建植入ICD患者发生室性心律失常及远期预后的风险预测模型。

方法：本研究为回顾性纳入符合一级预防和二级预防指征的ICD植入患者。采用血清学检测与影像学评估相结合的方法，测定受试者的甘油三酯葡萄糖指数（Triglyceride-glucose index, TyG指数）及非酒精性脂肪性肝病（Non-alcoholic fatty liver disease, NAFLD）状态。研究的主要终点为持续性室性心动过速和/或心室颤动。次要终点为心源性死亡、心脏移植及心力衰竭再住院的复合终点。采用多因素Cox比例风险回归分析筛选与终点事件显著相关的临床变量。在一级预防和二级预防人群中分别构建术后生存预测模型，并基于该模型绘制列线图。通过计算TyG指数和NAFLD的C统计量，与传统临床指标进行比较，并评估将代谢指标纳入模型后对预测能力的提升效果。进一步采用综合判别改进指数和净重分类改进指数评估模型的重分类和判别能力。

结果：本研究最终纳入914例患者，平均年龄为58.5±12.7岁，其中男性678例（74.2%）。缺血性心肌病患者400例（43.8%），NAFLD患者222例（24.3%）。平均随访34.8±12.2个月后，共265例（29.0%）患者发生室性心律失常事件，149例（16.3%）患者发生心血管事件。回归分析结果显示，年龄、NYHA III/IV级、QRS波时限、对数转换的N末端脑钠肽前体、超敏C反应蛋白、天门冬氨酸氨基转移酶、γ-谷氨酰转移酶、白细胞计数、糖化血红蛋白、白蛋白、高密度脂蛋白胆固醇、TyG指数、NAFLD、降脂药及β受体阻滞剂的使用均与终点事件显著相关。矫正曲线显示，列线图预测的生存概率与实际观察结果具有良好的一致性。基于TyG指数和NAFLD构建的风险预测模型在预测能力方面表现良好。在一级预防人群中，该模型预测室性心律失常的C统计量为0.75（95%置信区间，95% CI：0.71-0.79），预测心血管事件的C统计量为0.67（95% CI：0.60-0.73）。在二级预防人群中，预测室性心律失常的C统计量为0.72（95% CI：0.65-0.79），预测心血管事件的C统计量为0.65（95% CI：0.55-0.75）。与传统预测指标（左室射血分数≤35%、QRS波时限）相比，该模型在净重分类改进指数及综合判别改进指数方面表现出显著提升。

结论：在植入ICD的患者中，代谢指标（TyG指数和NAFLD）与室性心律失常及心血管事件的发生风险显著相关。基于该指标构建的ICD术后生存预测模型表现出良好的预测效能，或可作为新型代谢标志物，为该类患者的室性心律失常及心血管事件风险评估提供临床价值。

ABSTRACT (Keywords)

Part I Predictive value of metabolic biomarkers in risk stratification of patients for primary prevention of sudden cardiac death

Background: Despite significant advances in the diagnosis and treatment of cardiovascular diseases, the prevention of sudden cardiac death (SCD) remains a major challenge, primarily due to its sudden onset and insufficient identification of high-risk population. Existing risk stratification methods, such as left ventricular ejection fraction, electrocardiographic parameters, and serum biomarkers, have limitations in their predictive accuracy, applicability, and in identifying individuals at low to moderate risk. Metabolic biomarkers (such as glucose, lipids, and insulin resistance) are not only the core features of metabolic syndrome but also key risk factors for the occurrence and progression of cardiovascular diseases. Studies have shown that metabolic abnormalities are closely associated with cardiovascular events such as atherosclerosis, heart failure, and arrhythmias, yet their predictive value in high-risk SCD population remains unclear. This study aims to investigate the predictive value of non-alcoholic fatty liver disease (NAFLD) and the triglyceride-glucose index (TyG index) for ventricular arrhythmias and cardiovascular events in patients with implantable cardioverter-defibrillator (ICD) for primary prevention.

Methods: This retrospective study included patients who had the ICD implantation for primary prevention and underwent preoperative unenhanced computed tomography (CT), fasting glucose, and triglyceride testing. NAFLD was diagnosed using the liver-to-spleen attenuation ratio (L/S ratio). The primary endpoint was the occurrence of sustained ventricular tachycardia and/or ventricular fibrillation recorded by the ICD, while the secondary endpoint was a composite of cardiovascular death, heart transplantation, and heart failure rehospitalization. Event rates were estimated using a competing risks regression model and cumulative incidence functions, with the Fine-Gray test applied to analyze the risk associations between TyG index, NAFLD, and ventricular arrhythmias and cardiovascular events.

Results: A total of 649 patients were included in the final analysis, with a mean age of 58.5±12.5 years, including 490 males (75.5%) and 271 patients with ischemic cardiomyopathy (41.8%). Among these, 159 patients (24.5%) had NAFLD. After a mean follow-up of 34.3±11.9 months, 181 patients (27.9%) experienced ventricular arrhythmia events, and 104 patients (16.0%) experienced cardiovascular events. The restricted cubic spline analysis indicated that the probability of primary and secondary endpoint events increased as the TyG index increased. Conversely, the risk of endpoint events decreased as the L/S ratio increased. After adjusting several variates, the Fine-Gray competing risks model indicated that the TyG index was significantly associated with the risk of ventricular arrhythmias, with a hazard ratio (HR) of 2.46 (95% confidence interval [CI]: 1.73-3.49), and the highest TyG index group had a HR of 2.86 (95% CI: 1.77-4.63) compared to the lowest group. The HR for cardiovascular events was 2.35 (95% CI: 1.48-3.75), with the highest TyG index group having a HR of 3.01 (95% CI: 1.53-5.93). NAFLD was significantly associated with ventricular arrhythmias (HR=4.29, 95% CI: 2.99-6.16) and cardiovascular events (HR=2.62, 95% CI: 1.64-4.20).

Conclusions: In patients with ICD implantation for primary prevention, the TyG index and NAFLD are significantly associated with the occurrence of ventricular arrhythmias and cardiovascular events. These may serve as novel biomarkers for risk stratification in high-risk SCD populations.

Part II Predictive value of metabolic biomarkers in risk stratification of patients for secondary prevention of sudden cardiac death

Background: Patients had implantable cardioverter-defibrillators (ICD) implantation for secondary prevention are at high risk for malignant ventricular arrhythmias. Although the application of ICD has been widely proven to significantly improve clinical outcomes, research on risk prediction and intervention for ventricular arrhythmia and cardiovascular events in this patient population remains limited. In recent years, metabolic biomarkers have garnered increasing attention for their role in the occurrence, progression, and prognostic assessment of cardiovascular diseases. Metabolic abnormalities are closely linked to the pathophysiology of cardiovascular diseases and may provide new insights for predicting and intervening in these conditions. This study aims to explore the predictive value of non-alcoholic fatty liver disease (NAFLD) and triglyceride-glucose index (TyG index) in patients undergoing secondary prevention for sudden cardiac death (SCD).

Methods: This study retrospectively included patients with ICD implantation for secondary prevention. The primary endpoint events were ICD-recorded sustained ventricular tachycardia and/or ventricular fibrillation. Secondary endpoint included the composite event of cardiovascular death, heart transplantation, and rehospitalization due to heart failure. Competing risk regression models and cumulative incidence functions were applied to assess event rates, and Fine-Gray tests were utilized to analyze the associations between TyG index, NAFLD, and the risks of ventricular arrhythmias and cardiovascular events, considering as continuous and categorical variables.

Results: A total of 265 patients were included in this study, with a mean age of 58.5±13.0 years. Among them, 188 were male (70.9%), 129 had ischemic cardiomyopathy (48.7%), and 63 had NAFLD (23.8%). After a mean follow-up of 36.1±12.8 months, 84 patients (31.7%) experienced ventricular arrhythmia, and 45 patients (17.0%) experienced cardiovascular events. Fine-Gray competing risk model analysis revealed that the hazard ratio (HR) for TyG index in the occurrence of ventricular arrhythmia was 4.68 (95% confidence interval [CI]: 2.78–7.89), with the highest TyG index group having an HR of 5.44 (95% CI: 1.77–4.63) compared to the lowest group. The HR for cardiovascular events was 3.38 (95% CI: 1.39–8.23), with the highest TyG index group showing an HR of 3.67 (95% CI: 1.35–9.95) compared to the lowest group. NAFLD was associated with a HR of 2.97 (95% CI: 1.49–5.92) for the occurrence of ventricular arrhythmia, and a HR of 1.10 (95% CI: 0.47–2.58) for cardiovascular events.

Conclusions: In patients with ICD implantation for secondary prevention, the TyG index is significantly positively correlated with the occurrence of ventricular arrhythmias and cardiovascular events, while NAFLD is significantly correlated with the occurrence of ventricular arrhythmias. Both biomarkers show linear relationships with these outcomes. Based on these findings, reducing the TyG index and improving NAFLD may help reduce the risk of ventricular arrhythmias in patients undergoing ICD implantation for secondary prevention.

Part III Risk prediction model of ventricular arrhythmias and cardiovascular events in patients with ICD implantation

Background: The implantable cardioverter defibrillator (ICD) is an important therapeutic intervention for the prevention of sudden cardiac death (SCD). However, patients remain at high risk of ventricular arrhythmias and cardiovascular events following ICD implantation. Although traditional clinical indicators, such as left ventricular ejection fraction and prior myocardial infarction history, have been widely used for risk stratification, their predictive value is limited, making it challenging to accurately identify high-risk patients. Metabolic disorders are now closely linked to the onset and progression of cardiovascular diseases but the predictive efficacy of metabolic indicators for ventricular arrhythmias and cardiovascular events in patients with ICD remains unclear. This study aims to construct a risk prediction model for ventricular arrhythmias and long-term prognosis in patients with ICD based on metabolic indicators.

Methods: This retrospective study enrolled patients who underwent ICD implantation for primary and secondary prevention indication. Serum testing and imaging assessments were combined to measure the triglyceride-glucose index (TyG index) and non-alcoholic fatty liver disease (NAFLD) status in participants. The primary endpoint was sustained ventricular tachycardia and/or ventricular fibrillation. The secondary endpoint was a composite of cardiac death, heart transplantation, and heart failure rehospitalization. Multivariate Cox proportional hazards regression analysis was used to screen clinical variables significantly associated with endpoint events. Postoperative survival prediction models were constructed separately in primary and secondary prevention cohorts, and nomograms were developed based on these models. The C-statistics of the TyG index and NAFLD were calculated and compared with traditional clinical indicators to evaluate the improvement in predictive performance after incorporating metabolic indicators. Further, the integrated discrimination improvement and net reclassification improvement indices were used to assess the reclassification and discrimination capabilities of the models.

Results: A total of 914 patients were included, with a mean age of 58.5±12.7 years, including 678 males (74.2%). Among them, 400 patients (43.8%) had ischemic cardiomyopathy, and 222 patients (24.3%) had NAFLD. After a mean follow-up of 34.8±12.2 months, 265 patients (29.0%) experienced ventricular arrhythmia events, and 149 patients (16.3%) experienced cardiovascular events. Cox regression analysis revealed that age, NYHA class III/IV, QRS duration, log-transformed N-terminal pro-brain natriuretic peptide, high-sensitivity C-reactive protein, aspartate aminotransferase, gamma-glutamyl transferase, white blood cell count, glycated hemoglobin, albumin, high-density lipoprotein cholesterol, TyG index, NAFLD, and the use of lipid-lowering drugs and beta-blockers were significantly associated with endpoint events. Calibration curves demonstrated good agreement between the predicted survival probabilities by the nomogram and the observed outcomes. The risk prediction model based on the TyG index and NAFLD exhibited strong predictive performance. In the primary prevention cohort, the C-statistics for predicting ventricular arrhythmias and cardiovascular events were 0.75 (95% confidence interval [CI]: 0.71–0.79) and 0.67 (95% CI: 0.60–0.73), respectively. In the secondary prevention cohort, the C-statistics for predicting ventricular arrhythmias and cardiovascular events were 0.72 (95% CI: 0.65–0.79) and 0.65 (95% CI: 0.55–0.75), respectively. Compared to traditional predictive indicators (LVEF≤35%, QRS duration), the model showed significant improvements in integrated discrimination improvement and net reclassification improvement.

Conclusion: In patients with ICD, metabolic biomarkers (TyG index and NAFLD) are significantly associated with the risk of ventricular arrhythmias and cardiovascular events. The postoperative survival prediction model based on these indicators demonstrates strong predictive performance and may serve as novel metabolic biomarkers, offering clinical value for risk assessment of ventricular arrhythmias and cardiovascular events in this population.