【作者】 张帅；

【导师】 张澄； 吕慧霞； 王茜；

【作者基本信息】 山东大学 ， 内科学（心血管病）（专业学位）， 2025， 硕士

【摘要】 研究背景心力衰竭(heart failure,HF)是全球患者及医疗卫生系统面临的重大挑战,尤其是射血分数降低的心力衰竭(heart failure with reduced ejection fraction,HFrEF),其病情进展迅速,远期预后不良。神经激素激活被认为是HFrEF的主要病理生理机制之一,表现为左心室(left ventricle,LV)收缩功能障碍和心腔扩张。心脏磁共振(cardiac magnetic resonance,CMR)目前是评估心脏功能、量化心肌容积和评估心肌纤维化的无创“金标准”,相比超声心动图,其测量结果更为精准且具有良好的重复性。CMR可以识别HF中的高危特征,例如缺血性瘢痕、局灶性和弥漫性纤维化。近年来发展起来的CMR特征跟踪技术(CMR feature tracking technique,CMR-FT)可基于标准cine-CMR图像可靠评估心肌应变。与左心室射血分数(left ventricle ejection fraction,LVEF)相比,CMR-FT获得的LV应变被认为是评估亚临床阶段的LV收缩功能障碍的敏感技术,并有助于预测HF患者的生存结局。CMR-FT可获取包括左心房(left atrium,LA)、LV和右心室(right ventricle,RV)在内多个心腔的收缩和舒张功能应变指标,如LA储存应变、LA导管应变、LA收缩应变、左心室整体纵向应变(LV global longitudinal strain,LV-GLS)、左心室整体周向应变(LV global circumferential strain,LV-GCS)、左心室整体径向应变(LV global radial strain,LV-GRS)和右心室整体纵向应变(right ventricle global longitudinal strain,RV-GLS)。此外,左心室全局功能指数(left ventricle global structural and functional index,LVGFI)最近被认为作为 LV 功能的新型综合性指标,亦获得关注。心外膜脂肪组织(epicardial adipose tissue,EAT)是一种特殊的内脏脂肪,其遗传特征与棕色脂肪组织高度相似。EAT具备动态的棕色脂肪样产热功能,既可能通过能量代谢和机械缓冲对邻近心肌产生保护作用,也可能通过旁分泌或血管分泌方式释放促炎和促纤维化因子对心肌产生不利影响。呈现双相病理生理作用。相关机制可能参与HF的发生与进展。研究证明,CMR容积法已被证实是一种可重复、可靠的EAT定量评估手段。既往研究发现,在HFrEF和射血分数轻度降低的心力衰竭(heart failure with mildly reduced ejection fraction,HFmrEF)患者中,EAT厚度的增加与心脏超声参数如LV-GLS呈正相关。此外,CMR通过钆延迟增强(late gadolinium enhancement,LGE)技术可实现心肌瘢痕的精准可视化与定量分析,尤其在评估心肌纤维化方面具有独特优势。无论是在缺血性或非缺血性心肌病中,细胞外基质的重构是病理性心肌重塑的核心特征,常表现为胶原异常沉积和间质纤维化灶形成。LGE模式分析可辅助鉴别LV功能障碍的潜在病因(如缺血性瘢痕多呈心内膜下节段性分布,而心肌炎后瘢痕常表现为心外膜下或中层斑片状强化),且LGE阳性已被证实为HFrEF患者不良预后的独立预测因子。目前,对于HFrEF患者中基于CMR定量分析的EAT体积与多维度心功能参数的相关性研究的研究仍较为缺乏。这些功能参数包括LV功能指标(LVEF、LVGFI、LV-GLS、LV-GRS和LV-GCS)、RV功能指标(RV-GLS)、LA功能指标(LA储存应变、LA导管应变和 LA 收缩应变)、N 末端 B 型利钠肽原(N-terminal pro-B-type natriuretic peptide,NT-proBNP)和心室重塑指标等。尽管HFrEF的诊疗在近年来取得了显著进展,例如“新四联”药物治疗策略的广泛应用、心脏移植术后五年生存率的提升,以及左心室辅助装置(left ventricular assist device,LVAD)等机械循环支持技术的创新发展,该人群的长期预后仍不容乐观。这一矛盾凸显出现有风险分层体系的局限性,尤其是基于影像学特征(如EAT定量参数联合LGE)的高危人群早期识别能力亟待加强。然而,目前尚无研究探讨CMR衍生的EAT体积与LGE的协同效应对HFrEF患者预后(包括全因死亡和因心力衰竭再住院复合终点)的预测价值。因此,本研究旨在探讨HFrEF患者中EAT体积与LA、LV和RV功能参数的相关性,并进一步探讨EAT体积联合LGE在预测预后和进行风险分层中的临床应用价值。研究目的1.探讨在HFrEF患者中EAT体积与LA、LV与RV功能,LV重构和NT-proBNP的关联性分析。2.探讨在HFrEF患者中EAT体积联合LGE对预后和危险分层的价值。研究方法本研究属于回顾性研究,第一部分从2018年9月-2023年2月,连续筛查了 375例在山东大学齐鲁医院完成CMR检查且LVEF≤40%的患者。根据纳入和排除标准,最终276名HFrEF患者纳入分析。使用Circle CVI 42软件定量评估左右心功能、心肌应变、EAT体积和心肌纤维化程度(大于正常心肌的5个标准差的区域)。使用Spearman或Pearson相关性分析方法,探究EAT体积和标准化的EAT体积(indexed EAT体积)与心血管风险因素之间的相关性。采用多因素线性回归方法评估了 EAT体积和indexed EAT体积与HFrEF患者的LA功能指数(LA储存应变、LA导管应变和LA收缩应变)、LV 功能指数(LVEF、LVGFI、LV-GLS、LV-GRS 和 LV-GCS)、RV 功能指数(RV-GLS)和NT-proBNP之间的关系。第二部分:进一步对276名患者进行电话随访,生成Kaplan-Meier无事件生存曲线,并根据EAT体积和LGE中位数进行分层,通过Log-rank检验评估组间不良事件的差异。建立了多因素Cox 比例生存回归模型,分别评估EAT体积和LGE与主要终点的关联性。为了研究EAT体积和LGE对主要终点的联合预测价值,本研究将患者分为4组:第1组:LGE<5%且 EAT体积>44.5ml;第 2 组:LGE<5%且 EAT体积<44.5ml;第 3 组:LGE≥5%且 EAT 体积>44.5ml;第 4 组:LGE≥5%且 EAT 体积<44.5ml。生成相应的 Kaplan-Meier曲线,然后进行Log-rank检验。通过多因素Cox 比例生存回归模型进一步分析联合预测值。为了检验EAT体积和LGE之间的相互作用,计算了 C指数评估模型中加入EAT体积和LGE对模型的改善。同时为了数据呈现更加直观,本研究将LV-GLS、LV-GCS和RV-GLS得到的负数取绝对值进行表达。采用 SPSS 25.0、GraphPad Prism 9.0、R 4.2.0 和 Origin 2024 软件进行分析及绘图。研究结果1.HFrEF患者中EAT与LA、LV和RV功能,心室重塑指标和NT-proBNP关联分析1.1基线特征本研究共纳入276例HFrEF患者,平均年龄45.41±14.56岁,其中213名(77.2%)为男性,平均体质指数(body mass index,BMI)为25.51±4.15kg/m2。根据EAT体积的三分位数将患者平均分为3组(第1组:n=92,EAT体积<38.82 ml;第2组:n=91,38.82<EAT 体积<49.15ml;第 3 组:n=93,EAT 体积≥49.15ml)。EAT 体积越大,男性比例(P=0.023),BMI(P<0.001)和低密度脂蛋白胆固醇(low-density lipoprotein cholesterol,LDL-C)(P=0.008)水平越高,而NT-proBNP水平越低(P=0.001)。在CMR参数中,EAT 体积越高的患者 LVEF、LVGFI、LV-GLS、LV-GRS、LV-GCS、LA 储存应变、LA导管应变、LA收缩应变、RV-GLS越高,LVEDVi和LVESVi越低(所有P值均<0.05)。同时空腹血糖(fasting plasma glucose,FPG)和左心室舒张末期质量(LV end-diastolic mass,LVEM)也存在统计学意义(所有P值均<0.05)。其他指标无明显差异。1.2相关性分析采用Spearman或Pearson相关性分析了 HFrEF患者的EAT体积与传统心血管风险因素之间的相关性。EAT 体积与 BMI(r=0.328,P<0.001)、LDL-C(r=0.154,P=0.010)、LVEF(r=0.206,P=0.001)、LVGFI(r=0.283,P<0.001)、LV-GLS(r=0.258,P<0.001)、LV-GRS(r=0.192,P=0.001)、LV-GCS(r=0.203,P=0.001)、LVEM(r=0.144,P=0.016)、右心室射血分数(right ventricular ejection fraction,RVEF)(r=0.136,P=0.024)、RV-GLS(r=0.171,P=0.004)、LA 储存应变(r=0.301,P<0.001)、LA 导管应变(r=0.248,P<0.001)和 LA 收缩应变(r=0.266,P<0.001)呈正相关,同时与In(NT-proBNP)(r=-0.242,P<0.001)、LVEDVi(r=-0.124,P=0.040)和 LVESVi(r=-0.172,P=0.004)呈负相关,其他指标不存在统计学差异。1.3多因素直线回归多因素线性回归模型用于研究HFrEF患者EAT体积与LA、LV和RV功能是否存在独立关联性。EAT 体积与 LVEF、LV-GLS、LV-GRS、LV-GCS、LVGFI、RV-GLS 和LA应变参数(储存、导管和收缩应变)呈正相关,而与NT-proBNP呈负相关。模型1中的β值(95%CI)如下:LVEF,0.20(0.10-0.30);LVGFI,0.22(0.14-0.31);LV-GLS,0.08(0.05-0.12);LV-GRS,0.11(0.05-0.16);LV-GCS,0.06(0.03-0.09);RV-GLS,0.12(0.05-0.19);LA 储存应变,0.27(0.18-0.36);LA 导管应变,0.14(0.09-0.20);LA 收缩应变,0.13(0.08-0.18);ln(NT-proBNP),-0.03(-0.04to-0.02)。在进一步调整混杂因素后,EAT体积仍与更佳的LA功能指数显著相关(LA储存应变,β=0.29,P<0.001;LA 导管应变,β=0.15,P<0.001;LA 收缩应变指数,β=0.13,P<0.001),同时与更佳的RV功能指数(RV-GLS,β=0.13,P<0.001)及LV功能指数(LVEF,β=0.24,P<0.001;LVGFI,β=0.25,P<0.001;LV-GLS,β=0.09,P<0.001;LV-GRS,β=0.13,P<0.001;LV-GCS,β=0.06,P<0.001)呈正相关,且与较低的 ln(NT-proBNP)水平(β=-0.02,P=0.001)显著负相关。1.4二次分析在二次分析中,本研究在多因素线性回归模型中检验了 EAT体积与心房功能、心室功能和NT-proBNP之间的关系,并对体重指数(体重过轻和正常体重-BMI<23.9kg/m2;超重-BMI:24-27.9kg/m2;肥胖-BMI≥28kg/m2)和 LVEF(低 LVEF 组<25%和高 LVEF组≥25%)进行了分层。在不同组别中,EAT体积与更好的LA、LV和RV功能相关。1.5敏感性分析进一步进行了额外的敏感性分析,使用多元线性回归方程以进一步验证EAT体积与心房和心室功能之间的关系。因此,将两组人排除在分析之外:第一组:糖尿病患者;第二组:钠-葡萄糖协同转运蛋白2抑制剂(sodium-glucose cotransporter-2 inhibitors,SGLT2i)使用的患者。在排除两组患者后,EAT体积与LA、LV和RV功能之间仍然存在相关性,这可能表明这种相关性与药物和糖尿病无关。2.在HFrEF中EAT体积和LGE对预后预测和分层价值2.1 基线特征本研究进一步对第一部分的患者进行电话随访,口头得到知情同意,其中24名患者失访或者电话拒绝随访,记录有无主要终点的发生及发生时间(包括全因死亡和心力衰竭再住院的复合终点),最终研究共纳入252例患者,平均年龄为45.46±14.46岁,其中191名(75.8%)为男性,平均BMI为25.61 ±4.17kg/m2。根据有或无发生主要终点事件进行划分的患者基线特征。发生主要终点事件的患者更有可能患有糖尿病(P=0.026)和高脂血症(P=0.003),纽约心脏学会(New York Heart Association,NYHA)分级较高(P=0.003),且年龄(P=0.020)、NT-proBNP(P=0.004)、FPG(P=0.029)、LDL-C(P=0.029)水平也更高。发生主要终点人群其高密度脂蛋白胆固醇(high-density lipoprotein cholesterol,HDL-C)(P=0.002)和估算的肾小球滤过率(estimated glomerular filtration rate,eGFR)(P=0.041)更低。药物的使用(包括β受体阻滞剂、肾素血管紧张素转化酶抑制剂(angiotensin-converting enzyme inhibitor,ACEI)/血管紧张素Ⅱ受体拮抗剂(angiotensin receptor blocker,ARB)/血管紧张素受体脑啡肽酶抑制剂(angiotensin receptor-neprilysin inhibitor,ANRI)和 SGLT2i 两组也存在显著差异(所有P值<0.05)。CMR 参数中,两组在 LVSDVi、LVEF、LV-GLS、LV-GRS、LV-GCS、LA 储存应变、LA导管应变、LA收缩应变、RV-GLS、EAT体积和LGE存在显著差异(所有P值<0.05)。其他变量不存在统计学差异。2.2主要终点的单因素Cox回归分析观察到年龄、糖尿病、NYHA 分级、ln(NT-proBNP)、LDL-C、HDL-C、eGFR、LVEF、β受体阻滞剂和ACEI/ARB/ANRI药物使用与主要终点发生有显著的统计学相关性(所有P值<0.05)。EAT体积和LGE值每增加1个单位,主要终点发生的未调整HR(95%CI)分别为 0.97(0.94-0.99)和 1.09(1.06-1.11)。相比于 EAT 体积>44.5ml 病人,EAT 体积<44.5ml发生主要终点的风险增加120%。相比于LGE<5%的病人,LGE≥5%发生主要终点的风险增加109%。2.3 EAT体积和LGE对主要终点的预测价值在研究队列中(中位随访时间:22个月,四分位距:12-35个月),57例(22.4%)患者发生了主要终点。根据EAT体积和LGE范围的中位数绘制了 Kaplan-Meier生存曲线。EAT低于中位数的患者主要终点的发生率更高(Log-rank检验,P=0.005)。此外,与LGE值较低的患者相比,LGE值≥5%的患者发生主要终点的风险更高(Log-rank检验,P=0.008)。在调整其他心血管危险因素后,EAT体积和LGE预测价值仍然显著。EAT体积和LGE作为连续性变量纳入模型中,EAT体积每增加1个单位,主要终点发生风险降低4%。同样LGE每增加1个单位,主要终点发生风险增加8%。EAT体积和LGE作为分类变量纳入模型中,相比于EAT体积>44.5ml病人,EAT体积<44.5ml的患者主要终点发生风险升高174%。相比LGE<5%的患者,LGE≥5%的患者发生主要终点风险显著升高(HR=2.34,95%CI 1.32-4.13,P=0.009)。2.4 EAT体积和LGE对主要终点风险的联合预测为了研究EAT体积和LGE对主要终点的联合预测价值,本研究将患者分为4组:第 1 组:LGE<5%且 EAT 体积>44.5ml,第 2 组:LGE<5%且 EAT 体积<44.5ml,第 3 组:LGE≥5%且EAT体积>44.5ml,第4组:LG≥5%且EAT体积<44.5ml。在模型3中,相比第1组患者,第2组患者和第4组患者发生主要不良事件风险升高,调整后的HR(95%CI)分别为2.84(1.02-7.91)和4.99(1.91-13.00),在调整年龄和性别后,与LGE<5%且EAT体积>44.5ml的患者相比,LGE≥5%且EAT体积<44.5ml的患者发生主要终点的风险最高(HR=4.52,95%CI 1.73-11.83,P=0.008)。进一步调整混杂因素(包括年龄、性别、BMI、高血压、糖尿病、NYHA分级、TC和LDL-C)后,与第1组患者相比,第2组、第3组和第4组患者的主要终点调整后HR(95%CI)分别为3.28(1.13-9.52)、2.83(1.06-8.28)和 6.59(2.46-17.67)。2.5模型预测效能绘制受试者工作特征曲线(receiver operating characteristic,ROC),并使用曲线下面积(area under curve,AUC)估计EAT体积和LGE对HFrEF主要不良事件的预测价值。EAT 体积的 AUC(95%CI)为 0.636(0.557-0.714);LGE 的 AUC(95%CI)为 0.663(0.572-0.754)。本研究进一步为了评估EAT体积和LGE是否对主要终点具有增量预测值,比较有和没有EAT体积、LGE和同时加入EAT体积和LGE的完全校正模型,并计算C统计量。EAT体积加入模型中,C统计量从0.654增加到0.682;LGE加入模型中,C统计量从0.654增加到0.725。当同时在模型中加入EAT体积和LGE,C统计量(95%CI)由 0.654(0.574-0.734)增加到 0.752(0.672-0.832)。研究结论1.在HFrEF患者中,较高的EAT体积与更好的左右心室和LA功能显著相关,同时与较低程度的LV重塑及NT-proBNP水平相关。2.在HFrEF患者中,较低的EAT体积和较高的LGE是不良预后的独立相关因素。3.将EAT体积和LGE相结合,可实现对HFrEF患者的危险分层,并提高预测模型的预测效能。更多还原

【Abstract】 BackgroundHeart failure(HF)remains a major global health challenge for both patients and healthcare systems,particularly in cases of heart failure with reduced ejection fraction(HFrEF),which is characterized by rapid disease progression and poor long-term outcomes.Neurohormonal activation is considered one of the key pathophysiological mechanisms underlying HFrEF,manifesting as impaired left ventricular(LV)systolic function and chamber dilation.Cardiac magnetic resonance(CMR)is the noninvasive gold standard for assessing myocardial function,quantifying cardiac volumes,and evaluating myocardial fibrosis.Compared with echocardiography,CMR provides more accurate and reproducible measurements.It can identify high-risk features in HF patients,such as ischemic scarring and both focal and diffuse fibrosis.In recent years,CMR feature tracking(CMR-FT)technology has emerged as a reliable method for quantifying myocardial strain based on standard cine-CMR images.Compared to left ventricular ejection fraction(LVEF),LV strain derived from cardiac magnetic resonance feature tracking(CMR-FT)is considered a more sensitive technique for detecting subclinical LV systolic dysfunction and is helpful in predicting outcomes in HF patients.CMR-FT can assess strain indices reflecting systolic and diastolic function across multiple cardiac chambers,including the left atrium(LA),LV,and right ventricle(RV),such as LA reservoir strain,LA conduit strain,LA contraction strain,LV global longitudinal strain(LV-GLS),LV global circumferential strain(LV-GCS),LV global radial strain(LV-GRS),and RV global longitudinal strain(RV-GLS).Additionally,the left ventricular global structural and functional index(LVGFI)has recently gained attention as a novel integrated measure of LV performance.Epicardial adipose tissue(EAT)is a unique form of visceral fat that shares genetic characteristics with brown adipose tissue.EAT exhibits dynamic brown fat-like thermogenic activity.It may exert protective effects on the adjacent myocardium through energy metabolism and mechanical buffering,while it may also exert detrimental effects by releasing pro-inflammatory and pro-fibrotic factors via paracrine or vasocrine pathways.Thus,it demonstrates a biphasic pathophysiological role.These biphasic pathophysiological effects suggest that EAT may be involved in the development and progression of HF.CMR-based volumetric techniques have been validated as reproducible and reliable methods for quantifying EAT.Previous studies have reported that increased EAT thickness is positively associated with echocardiographic parameters such as LV-GLS in patients with HFrEF and heart failure with mildly reduced ejection fraction(HFmrEF).Moreover,CMR late gadolinium enhancement(LGE)enables precise visualization and quantification of myocardial scarring,providing a unique advantage in assessing myocardial fibrosis.In both ischemic and nonischemic cardiomyopathies,extracellular matrix remodeling—characterized by abnormal collagen deposition and interstitial fibrotic lesions—is a hallmark of pathological myocardial remodeling.LGE pattern analysis assists in differentiating the etiology of LV dysfunction(e.g.,subendocardial segmental distribution in ischemic scars vs.subepicardial or mid-wall patchy enhancement in post-myocarditis fibrosis).Notably,LGE positivity has been identified as an independent predictor of adverse outcomes in patients with HFrEF.To date,limited studies have investigated the correlation between CMR-quantified EAT volume and multidimensional cardiac function parameters in HFrEF patients.These parameters include LV function(LVEF,LVGFI,LV-GLS,LV-GRS,and LV-GCS),RV function(RV-GLS),LA function(LA reservoir,conduit,and booster strain),N-terminal pro-B-type natriuretic peptide(NT-proBNP),and indices of ventricular remodeling.Despite recent advances in HFrEF treatment—such as the widespread adoption of the "four-pillar"pharmacological regimen,improved 5-year survival following heart transplantation,and innovations in mechanical circulatory support(e.g.,left ventricular assist devices,LVADs)—long-term prognosis in this population remains suboptimal.This highlights the limitations of existing risk stratification systems and underscores the urgent need to improve early identification of high-risk individuals,particularly through imaging-based markers such as combined EAT quantification and LGE.However,no studies have yet examined the prognostic value of the combined effect of CMR-derived EAT volume and LGE on outcomes in HFrEF patients,including all-cause mortality and HF-related rehospitalization.Therefore,this study aims to investigate the associations between EAT volume and LA,LV,and RV functional parameters in HFrEF patients,and to further evaluate the clinical utility of EAT volume combined with LGE in predicting prognosis and enhancing risk stratification.Objectives1.To investigate the association between EAT volume and LA,LV,and RV function,LV remodeling,and NT-proBNP in HFrEF patients.2.To explore the prognostic value and risk stratification potential of EAT volume combined with LGE in HFrEF patients.MethodsThis study is a retrospective analysis.In the first phase,a total of 375 patients who underwent CMR imaging and had a LVEF ≤40%were consecutively screened at Qilu Hospital of Shandong University from September 2018 to February 2023.After applying inclusion and exclusion criteria,276 HFrEF patients were enrolled for final analysis.Cardiac function,myocardial strain,EAT volume,and myocardial fibrosis(defined as areas with signal intensity>5 standard deviations above that of normal myocardium)were quantitatively assessed using the Circle CVI 42 software.Spearman or Pearson correlation analysis was used to explore the relationships between EAT volume,indexed EAT volume(normalized to body surface area),and cardiovascular risk factors.Multivariable linear regression was performed to assess the associations between EAT volume indexed EAT volume and LA functional parameters(reservoir,conduit,and contraction strain),LV functional indices(LVEF,LVGFI,LV-GLS,LV-GRS,and LV-GCS),RV function(RV-GLS),and NT-proBNP levels in patients with HFrEF.In the second phase,telephone follow-up was conducted for the 276 patients,and Kaplan-Meier survival curves were generated based on the median values of EAT volume and LGE.The Log-rank test was used to compare differences in primary events between groups.A multivariable Cox proportional hazards regression model was constructed to evaluate the associations of EAT volume and LGE with the primary endpoint.To investigate the combined predictive value of EAT volume and LGE for the primary endpoint,patients were stratified into four groups:Group 1:LGE<5%and EAT volume>44.5 ml;Group 2:LGE<5%and EAT volume<44.5 ml;Group 3:LGE≥ 5%and EAT volume>44.5 ml;Group 4:LGE≥ 5%and EAT volume<44.5 ml.Kaplan-Meier curves were generated for each group,and differences were assessed using the Log-rank test.Further multivariable Cox regression analysis was conducted to examine the combined predictive value of EAT volume and LGE.To evaluate the interaction between EAT volume and LGE,we calculated the C-index to determine the improvement in model performance after incorporating EAT volume and myocardial fibrosis.For clarity in data presentation,the absolute values of LV-GLS,LV-GCS,and RV-GLS were used.All statistical analyses and visualizations were performed using SPSS 25.0,GraphPad Prism 9.0,R software 4.2.0,and Origin 2024.Results1.Association between EAT volume and LA,LV,and RV function,LV remodeling indices,and NT-proBNP in HFrEF patients1.1 Baseline characteristicsA total of 276 patients with HFrEF were included in this study,with a mean age of 45.41± 14.56 years.Among them,213(77.2%)were male,and the mean body mass index(BMI)was 25.51 ± 4.15 kg/m2.Patients were stratified into three groups based on EAT volume tertiles:the lowest tertile(n=92,EAT volume<38.82 ml),the middle tertile(n=91,38.82 ≤EAT volume<49.15 ml),and the highest tertile(n=93,EAT volume≥ 49.15 ml).A larger EAT volume was associated with a higher proportion of male patients(P=0.023),higher BMI(P<0.001),and elevated levels of low-density lipoprotein cholesterol(LDL-C)(P=0.008),while NT-proBNP levels were significantly lower(P=0.001).Among the CMR parameters,patients with higher EAT volumes exhibited significantly better values of LVEF,LVGFI,LV-GLS,LV-GRS,LV-GCS,LA reservoir strain,LA conduit strain,LA booster strain,and RV-GLS,as well as lower LVEDVi and LVESVi(all P-values<0.05).In addition,fasting plasma glucose(FPG)and LV end-diastolic mass(LVEM)were also significantly associated with EAT volume(both P-values<0.05).No significant differences were observed in other parameters.1.2 Correlation analysisSpearman or Pearson correlation analysis was performed to assess the relationship between EAT volume and cardiovascular risk factors.EAT volume was positively correlated with BMI(r=0.328,P<0.001),LDL-C(r=0.154,P=0.010),LVEF(r=0.206,P=0.001),LVGFI(r=0.283,P<0.001),LV-GLS(r=0.258,P<0.001),LV-GRS(r=0.192,P=0.001),LV-GCS(r=0.203,P=0.001),LVEM(r=0.144,P=0.016),right ventricular ejection fraction(RVEF)(r=0.136,P=0.024),RV-GLS(r=0.171,P=0.004),LA reservoir strain(r=0.301,P<0.001),LA conduit strain(r=0.248,P<0.001),and LA booster strain(r=0.266,P<0.001).Conversely,EAT volume was negatively correlated with ln(NT-proBNP)(r=-0.242,P<0.001),LV end-diastolic volume index(LVEDVi)(r=-0.124,P=0.040),and LV end-systolic volume index(LVESVi)(r=-0.172,P=0.004).No statistically significant correlations were found for other parameters.1.3 Multivariable linear regression analysisA multivariable linear regression model was employed to examine whether EAT volume is independently associated with LA,LV,and RV function in HFrEF patients.EAT volume was positively correlated with LVEF,LV-GLS,LV-GRS,LV-GCS,LVGFI,RV-GLS,and LA strain parameters(reservoir,conduit,and booster strain),while negatively correlated with NT-proBNP.The β coefficients(95%CI)in model 1 were as follows:LVEF:0.20(0.10-0.30);LVGFI:0.22(0.14-0.31);LV-GLS:0.08(0.05-0.12);LV-GRS:0.11(0.05-0.16);LV-GCS:0.06(0.03-0.09);RV-GLS:0.12(0.05-0.19);LA reservoir strain:0.27(0.18-0.36);LA conduit strain:0.14(0.09-0.20);LA booster strain:0.13(0.08-0.18);ln(NT-proBNP):-0.03(-0.04 to-0.02).After further adjustment for confounding factors,EAT volume remained significantly associated with better LA functional indices,including LA reservoir strain(β=0.29,P<0.001),LA conduit strain(β=0.15,P<0.001),and LA booster strain(β=0.13,P<0.001).Additionally,it showed a positive correlation with improved RV function(RV-GLS,β=0.13,P<0.001)and LV functional parameters,including LVEF(β=0.24,P<0.001),LVGFI(β=0.25,P<0.001),LV-GLS(β=0.09,P<0.001),LV-GRS(β=0.13,P<0.001),and LV-GCS(β=0.06,P<0.001).Moreover,EAT volume was significantly and inversely associated with ln(NT-proBNP)levels(β=-0.02,P=0.001).1.4 Secondary analysisIn a secondary analysis,the relationship between EAT volume and atrial function,ventricular function,and NT-proBNP was further examined using a multivariable linear regression model.The analysis was stratified by BMI(underweight and normal weight-BMI<23,9 kg/m2;overweight-BMI:24-27.9 kg/m2;obese-BMI≥ 28 kg/m2)and LVEF(severely reduced LVEF group:<25%;mildly reduced LVEF group:≥25%).Across all subgroups,EAT volume remained significantly associated with better LA,LV,and RV function.1.5 Sensitivity analysisAdditional sensitivity analyses were conducted using multivariable linear regression to further validate the relationship between EAT volume and atrial/ventricular function.Two patient groups were excluded:(1)Patients with diabetes;(2)Patients using sodium-glucose cotransporter-2 inhibitors(SGLT2i).Even after excluding these two groups,the associations between EAT volume and LA,LV,and RV function persisted,suggesting that these correlations are independent of diabetes and SGLT2i treatment.2.Prognostic value and risk stratification of combined EAT volume and LGE in HFrEF2.1 Baseline characteristicsA follow-up study was conducted on the patients from the first phase,with informed consent obtained via telephone.A total of 24 patients were lost to follow-up or declined participation.The final study cohort included 252 patients,with an average age of 45.46±14.46 years,and 191(75.8%)were male.The mean BMI was 25.61±4.17 kg/m2.Patients were categorized into two groups based on the occurrence of the primary endpoint(a composite of all-cause mortality and heart failure rehospitalization).Patients who experienced the primary endpoint were more likely to have diabetes(P=0.026)and hyperlipidemia(P=0.003).a higher New York Heart Association(NYHA)functional class(P=0.003),and elevated levels of age(P=0.020),NT-proBNP(P=0.004),fasting plasma glucose(FPG)(P=0.029),and low-density lipoprotein cholesterol(LDL-C)(P=0.029).In contrast,they had lower levels of high-density lipoprotein cholesterol(HDL-C)(P=0.002)and estimated glomerular filtration rate(eGFR)(P=0.041).Significant differences were observed in the use of medications,including β blockers,angiotensin-converting enzyme inhibitor(ACEI)/angiotensin receptor blocker(ARB)/angiotensin receptor-neprilysin inhibitor(ANRI),and SGLT2i(all P values<0.05).Regarding CMR parameters,significant differences between the two groups were found in LVSDVi,LVEF,LV-GLS,LV-GRS,LV-GCS,LA reservoir strain,LA conduit strain,LA booster strain,RV-GLS,EAT volume,and LGE(all P values<0.05).No significant differences were observed in other variables.2.2 Univariate Cox regression analysis for the primary endpointUnivariate Cox regression analysis identified age,diabetes,NYHA class.ln(NT-proBNP),LDL-C,HDL-C,eGFR,LVEF.β blocker.and ACEI/ARB/ANRI use as significant predictors of the primary endpoint(all P values<0.05).For EAT volume and LGE:Each unit increase in EAT volume was associated with a 3%reduction in primary endpoint risk(HR=0.97,95%CI:0.94-0.99);Each unit increase in LGE was associated with a 9%increase in primary endpoint risk(HR=1.09,95%CI:1.06-1.11).Compared to patients with an EAT volume>44.5 ml.those with an EAT volume<44.5 ml had a 120%increased risk of experiencing the primary endpoint.Similarly,compared to patients with LGE<5%.those with LGE≥5%had a 109%increased risk of experiencing the primary endpoint.2.3 Predictive value of EAT volume and LGE for the primary endpointDuring a median follow-up of 22 months(IQR:12-35 months).57 patients(22.4%)experienced the primary endpoint.Kaplan-Meier survival curves were generated based on the median values of EAT volume and LGE.Patients with lower EAT volume had a significantly higher incidence of the primary endpoint(Log-rank test.P=0.005).Patients with LGE≥ 5%had a significantly higher risk of the primary endpoint compared to those with LGE<5%(Log-rank test,P=0.008).After adjusting for other cardiovascular risk factors:Each 1-unit increase in EAT volume was associated with a 4%decrease in primary endpoint risk;Each 1-unit increase in LGE was associated with an 8%increase in primary endpoint risk.When EAT volume and LGE were analyzed as categorical variables:Patients with EAT volume<44.5 ml had a 174%increased risk of experiencing the primary endpoint;Patients with LGE≥ 5%had a significantly higher risk of the primary endpoint(HR=2.34,95%CI:1.32-4.13,P=0.009),compared to those with LGE<5%.2.4 Combined effect of EAT Volume and LGE on primary endpoint riskTo investigate the combined predictive value of EAT volume and LGE for the primary endpoint,patients were categorized into four groups:group 1:LGE<5%and EAT volume>44.5 ml;group 2:LGE<5%and EAT volume<44.5 ml;group 3:LGE≥ 5%and EAT volume>44.5 ml;group 4:LGE≥ 5%and EAT volume<44.5 ml.In Model 3,compared to group 1,the risk of primary events was significantly higher in group 2(HR=2.84,95%CI:1.02-7.91)and group 4(HR=4.99,95%CI:1.91-13.00).After adjusting for age and sex,patients with LGE≥ 5%and EAT volume<44.5 ml had the highest risk of the primary endpoint compared to those with LGE<5%and EAT volume>44.5 ml(HR=4.52,95%CI:1.73-11.83,P=0.008).Further adjustments for confounding factors(age,sex,BMI,hypertension,diabetes,NYHA class,TC,and LDL-C)yielded the following adjusted HRs(95%CI)for the primary endpoint:group 3:HR=3.28(1.13-9.52);group 2:HR=2.83(1.06-8.28);group 4:HR=6.59(2.46-17.67).2.5 Predictive Performance of the ModelTo assess the predictive value of EAT volume and LGE for primary event in HFrEF,we constructed receiver operating characteristic(ROC)curves and estimated the area under the curve(AUC).The AUC(95%CI)was 0.636(0.557-0.714)for EAT volume and 0.663(0.572-0.754)for LGE,respectively.To evaluate the incremental predictive value of EAT volume and LGE,we compared fully adjusted models with and without these parameters and calculated the C-statistic:Adding EAT volume to the model increased the C-statistic from 0.654 to 0.682;Adding LGE increased the C-statistic from 0.654 to 0.725;When both EAT volume and LGE were included in model,the C-statistic increased from 0.654(95%CI:0.574-0.734)to 0.752(95%CI:0.672-0.832),demonstrating improved predictive performance.Conclusions1.In patients with HFrEF,a higher EAT volume is significantly associated with better biventricular and LA function,as well as with lower degrees of LV remodeling and reduced NT-proBNP levels.2.In patients with HFrEF,lower EAT volume and higher levels of LGE are independently associated with worse outcomes.3.The combination of EAT volume and LGE enables effective risk stratification in HFrEF patients and enhances the predictive performance of prognostic models.更多还原