Prognostic value of epicardial and pericardial adipose tissue in patients with chronic heart failure Authors: Yoran Crum, Dirk J. van Veldhuisen, Michelle Lobeek, Adriaan A. Voors, Michiel Rienstra, Thomas M. Gorter Published: 2026 ABSTRACT BackgroundCentral obesity, rather than overall obesity, may influence heart failure (HF) outcomes. Alternative anthropometric measures remain limited in assessing the actual amount and distribution of visceral adiposity. We hypothesize that epicardial and pericardial adipose tissue (EAT and PAT) are easily obtainable on standard echocardiography and provide prognostic value in patients with HF.MethodsIn total, 572 patients with HF with reduced, mildly reduced and preserved ejection fraction (HFrEF, HFmrEF, and HFpEF, respectively) were included. EAT (the adipose tissue between the myocardium and the visceral layer of the pericardium) and PAT (the adipose tissue outside the parietal layer of the pericardium) thickness were measured on echocardiography and expressed in mm. Patients were followed for the combined outcome of all-cause mortality and HF hospitalization.ResultsMean age was 67 ± 14 years, 38% were women, 346 had HFrEF, 100 had HFmrEF and 126 had HFpEF. Mean body mass index (BMI) was 27 ± 6 kg/m2, mean EAT thickness 4.8 ± 2.1 mm, and mean PAT thickness 5.7 ± 2.8 mm. During a median follow-up of 3.2 ± 1.9 years, both EAT and PAT thickness were associated with an increased risk for the combined outcome [HR 1.15 (95% CI 1.03–1.30), p = 0.006 and HR 1.25 (95% CI 1.10–1.41), p < 0.001, respectively], whereas BMI was not. In a multivariable model, only PAT thickness remained associated with poor outcome [HR 1.49 per SD (95% CI 1.10–2.03), p = 0.011].ConclusionIncreased PAT thickness, more than EAT, assessed on routine echocardiography, may serve as a valuable prognostic marker in chronic HF.Supplementary InformationThe online version of this article (https://doi.org/10.1007/s12471-026-02077-z) contains supplementary material, which is available to authorized users. What’s new? Previous studies have primarily focused on the role of general obesity in patients with heart failure (HF), whereas obesity is defined according to body mass index (BMI). However, BMI is limited in assessing the actual burden of adipose tissue (AT). In addition, alternative anthropometric measures, such as waist circumference and waist-to-height ratio, also remain limited in assessing the amount and distribution of AT. Fat depots of increasing interest are epicardial and pericardial adipose tissue (EAT and PAT, respectively), which are situated close to the heart. In the present, large study population, we demonstrated that EAT and PAT measured on routine echocardiography are associated with poor outcome, irrespective of overall obesity. These specific AT depots may thus serve as easily obtainable prognostic markers in patients with HF, irrespective of overall obesity. Introduction Obesity has been repeatedly shown to be an important risk factor for the development of heart failure (HF), but the prognostic value of overall obesity in established HF remains disputed [1, 2]. The association between obesity and HF has been generally based on increased body mass index, which is limited in assessing the burden of adipose tissue [3]. Furthermore, alternative anthropometric indices of central adiposity, such as waist circumference and waist-to-height ratio, have been associated with poor outcomes [4, 5]. Nevertheless, these alternative indices are also limited in assessing the actual amount and distribution of adipose tissue (AT) [6]. Specifically, AT depots in proximity to the heart (e.g., epicardial and pericardial adipose tissue [EAT and PAT]) may have clinical relevance in patients with HF [7]. EAT is located within the visceral layer of the pericardium and directly adjacent to the myocardium and the coronary arteries. Recent studies have shown that EAT is associated with disease severity, HF hospitalization, and mortality in patients with HF [8, 9]. In contrast, less is known about PAT in patients with HF, which differs in anatomical location, embryological origin, and blood supply. PAT may represent a more overall unhealthy cardiometabolic profile (Fig. 1; [6, 10]).Fig. 1Infographic Infographic While EAT and PAT have both been suggested to be involved in the development of HF, the association of these very distinct fat depots with clinical outcomes remains unclear. Therefore, we aimed to assess the association between EAT and PAT measured on standard echocardiography with clinical outcomes in patients with established HF. We also investigated whether these AT depots show a differential association with outcomes in patients with HF with reduced versus mildly reduced and preserved ejection fraction. Methods All patients with new-onset HF included in the present study originated from a prospective observational cohort that has been described previously in detail [11, 12]. In brief, new-onset HF patients were referred to the specialized HF outpatient clinic of our tertiary care hospital between 2012 and 2018. Patients were subsequently classified as having HF with reduced ejection fraction (HFrEF), with mildly reduced ejection fraction (HFmrEF), and with preserved ejection fraction (HFpEF), according to the left ventricular ejection fraction (LVEF). HF had been diagnosed by a cardiologist no more than 3 months before inclusion, according to the European Society of Cardiology Guidelines for HF [13]. Patient characteristics were recorded at baseline visits [12]. Participants were categorized into four BMI categories: 1) BMI < 25 kg/m2; 2) BMI 25–29.9 kg/m2; 3) BMI 30–34.9 kg/m2; and 4) BMI ≥ 35 kg/m2. The study complied with local and national regulations regarding retrospective clinical research, as confirmed by the medical ethical evaluation committee, and was conducted in accordance with the principles of the Declaration of Helsinki. Outcome assessment The primary outcome measure was time to the first event of either HF hospitalization or all-cause mortality. Outcome data were extracted from the hospital’s electronic health record until the censor date of 1 September 2018. Echocardiography Echocardiography was performed as part of routine clinical care. All echocardiograms were digitally stored and transferred to the Echo Core Lab for additional measurements using EchoPAC software (Version B12). In total, 60 echocardiograms were excluded due to insufficient image quality. EAT and PAT thickness were measured alongside the right ventricle (RV) of the parasternal short-axis and parasternal long-axis views, as illustrated in Fig. 2. The measurements were obtained over two cardiac cycles at the end of systole and were subsequently averaged. The papillary muscles/interventricular septum and aortic annulus were used as anatomic reference points for these assessments [14]. PAT was defined as the echolucent area outside the pericardium, in which the visceral and parietal layers can be seen sliding over each other on the echocardiographic image loops, facilitating a clear distinction between EAT and PAT. Patients were divided into quartiles according to EAT and PAT thickness. All EAT and PAT measurements were performed by one experienced investigator (Y.C.). For interobserver variability, > 10% of the measurements were independently re-evaluated by another experienced investigator (T.M.G.), demonstrating differences of ≤ 5% between measurements. Intra-observer variability was likewise ≤ 5%.Fig. 2Location of epicardial- and pericardial tissue depots and echocardiographic assessment. a, b Show the parasternal long-axis and short-axis views on the echocardiogram. c, d Show the different layers of adipose tissue (i.e. EAT and PAT) surrounding the myocardium. White: EAT, Red: PAT. Using the aortic annulus and interventricular septum as reference points. The orange line indicates the pericardium, and blue line indicates the myocardial surface. Ao Aorta; EAT epicardial adipose tissue; PAT pericardial adipose tissue; LA left atrium; LV left ventricle; RV right ventricle Location of epicardial- and pericardial tissue depots and echocardiographic assessment. a, b Show the parasternal long-axis and short-axis views on the echocardiogram. c, d Show the different layers of adipose tissue (i.e. EAT and PAT) surrounding the myocardium. White: EAT, Red: PAT. Using the aortic annulus and interventricular septum as reference points. The orange line indicates the pericardium, and blue line indicates the myocardial surface. Ao Aorta; EAT epicardial adipose tissue; PAT pericardial adipose tissue; LA left atrium; LV left ventricle; RV right ventricle Statistical analysis Data were presented as mean ± standard deviation, median [interquartile range], or numbers (%). Differences for continuous variables between two groups were tested using the Independent samples t‑test or Mann-Whitney U test, according to distribution. The chi-square test was used for categorical variables. Intergroup differences were tested using one-way ANOVA or the Kruskal-Wallis test. Post hoc Bonferroni correction was applied to adjust for multiple comparisons. Correlations between continuous variables were tested using Pearson correlation analysis. Hazard ratios (HR) for continuous variables are depicted as per standard deviation with 95% confidence interval (CI). Cox proportional hazards regression analyses were performed to assess the prognostic value of EAT and PAT, adjusting for confounders identified from previous literature and variables associated with outcome in univariable analysis (p < 0.05), using a backward selection approach [15]. For EAT and PAT, spline curves were plotted up to the 95th percentile of the distribution to minimize the impact of outliers. Associations between outcomes and EAT, PAT, and BMI groups were assessed using log-rank tests. For all statistical analyses, a two-tailed p < 0.05 was considered statistically significant. SPSS Statistics 27 version (IBM SPSS Statistics, NY, USA) was used to conduct statistical analyses, and spline curves were constructed using R. Results In total, 572 patients with HF and available LVEF assessment had echocardiograms of sufficient quality for EAT and PAT measurements. The study population consisted of 346 (61%) patients with HFrEF, 100 (17%) with HFmrEF, and 126 (22%) with HFpEF. Most patients were men (n = 355, 62%). As shown in Tab. 1, patients with HFpEF and HFmrEF were older, more frequently had atrial fibrillation and hypertension, had a higher BMI, and had lower NT-proBNP levels compared with patients with HFrEF.Table 1Baseline CharacteristicsTotal (n = 572)HFrEF (n = 346)HFmrEF (n = 100)HFpEF (n = 126)P for trendAge, years67 ± 1466 ± 1469 ± 15§70 ± 14§ 0.003Sex, women %217 (38%)131 (35%)35 (33%)51 (52%) † 0.002Body mass index (kg/m2)27 ± 627 ± 527 ± 529 ± 7§†< 0.001HF characteristicsNYHA Class 0.4I182 (30%) 53 (14%)20 (19%)17 (13%)II325 (52%)203 (54%)56 (52%)63 (46%)III100 (16%)107 (29%)29 (27%)51 (38%)IV 7 (1%) 6 (2%) 1 (1%) 3 (2%)ComorbiditiesHypertension216 (34%)114 (31%)44 (41%)56 (41%) 0.027Ischaemic heart disease237 (38%)143 (38%)49 (45%)43 (32%)† 0.089Atrial fibrillation, history241 (38%)130 (35%)39 (36%)66 (49%)§† 0.016Diabetes mellitus138 (22%) 72 (19%)27 (25%)38 (28%) 0.083COPD 85 (14%) 35 (9%)17 (16%)31 (23%) §< 0.001Cardiothoracic surgery 94 (15%) 46 (12%)22 (20%)25 (18%) 0.055Laboratory testseGFR, ml/min/1.73 m267 ± 2567 ± 2569 ± 2664 ± 26 0.4NT-proBNP, ng/l1538 [558–3679]1637 [693–3770]1013 [359–2121]§963 [359–2037]§< 0.001MedicationBeta-blocker use534 (85%)333 (89%)90 (83%)105 (77%)§ 0.003ACEi/ARB use521 (83%)325 (87%)91 (84%)101 (74%)§ 0.003Loop diuretic use437 (70%)261 (70%)72 (67%) 99 (73%) 0.6Echocardiographic parametersLVEF (%)38 ± 1328 ± 845 ± 3§55 ± 4§†< 0.001LAVI43 ± 1544 ± 1642 ± 1243 ± 15 0.5LVMI107 ± 32111 ± 32102 ± 28§100 ± 34§ 0.002e′ lateral8.3 ± 3.27.7 ± 3.28.7 ± 3.19.1 ± 3.3§< 0.001e′ septal5.9 ± 2.35.5 ± 2.16.2 ± 2.3§6.7 ± 2.5§< 0.001e′ mean7.1 ± 2.56.7 ± 2.47.4 ± 2.28.0 ± 2.6§< 0.001E/e′12 ± 512 ± 611 ± 412 ± 5 0.4EAT4.8 ± 2.14.5 ± 2.14.9 ± 1.85.4 ± 2.1§< 0.001PAT5.7 ± 2.85.4 ± 2.45.5 ± 2.96.4 ± 3.5§† 0.004Data are mean ± SD, median [interquartile range] or n (%). Final column reflects overall differences in EAT tertiles. ANOVA was used for normally distributed parameters and Chi-squared was used for categorical parametersACEi angiotensin-converting enzyme inhibitor; ARB angiotensin receptor blocker; BMI body mass index; COPD chronic obstructive pulmonary disease; DBP diastolic blood pressure; eGFR estimated glomerular filtration rate; HF heart failure; HFrEF heart failure with reduced ejection fraction; HFmrEF heart failure with mildly reduced ejection fraction; HFpEF heart failure with preserved ejection fraction; LAVI left atrial volumetric index; LVEF left ventricular ejection fraction; LVMI left ventricular mass index; MRA mineralocorticoid receptor antagonist; NT-proBNP N-terminal pro B‑type natriuretic peptide; NYHA New York Heart Association; RVD right ventricular dysfunction; SBP systolic blood pressure§ p0.05 versus HFrEF; † p0.05 versus HFmrEF Table 1Baseline CharacteristicsTotal (n = 572)HFrEF (n = 346)HFmrEF (n = 100)HFpEF (n = 126)P for trendAge, years67 ± 1466 ± 1469 ± 15§70 ± 14§ 0.003Sex, women %217 (38%)131 (35%)35 (33%)51 (52%) † 0.002Body mass index (kg/m2)27 ± 627 ± 527 ± 529 ± 7§†< 0.001HF characteristicsNYHA Class 0.4I182 (30%) 53 (14%)20 (19%)17 (13%)II325 (52%)203 (54%)56 (52%)63 (46%)III100 (16%)107 (29%)29 (27%)51 (38%)IV 7 (1%) 6 (2%) 1 (1%) 3 (2%)ComorbiditiesHypertension216 (34%)114 (31%)44 (41%)56 (41%) 0.027Ischaemic heart disease237 (38%)143 (38%)49 (45%)43 (32%)† 0.089Atrial fibrillation, history241 (38%)130 (35%)39 (36%)66 (49%)§† 0.016Diabetes mellitus138 (22%) 72 (19%)27 (25%)38 (28%) 0.083COPD 85 (14%) 35 (9%)17 (16%)31 (23%) §< 0.001Cardiothoracic surgery 94 (15%) 46 (12%)22 (20%)25 (18%) 0.055Laboratory testseGFR, ml/min/1.73 m267 ± 2567 ± 2569 ± 2664 ± 26 0.4NT-proBNP, ng/l1538 [558–3679]1637 [693–3770]1013 [359–2121]§963 [359–2037]§< 0.001MedicationBeta-blocker use534 (85%)333 (89%)90 (83%)105 (77%)§ 0.003ACEi/ARB use521 (83%)325 (87%)91 (84%)101 (74%)§ 0.003Loop diuretic use437 (70%)261 (70%)72 (67%) 99 (73%) 0.6Echocardiographic parametersLVEF (%)38 ± 1328 ± 845 ± 3§55 ± 4§†< 0.001LAVI43 ± 1544 ± 1642 ± 1243 ± 15 0.5LVMI107 ± 32111 ± 32102 ± 28§100 ± 34§ 0.002e′ lateral8.3 ± 3.27.7 ± 3.28.7 ± 3.19.1 ± 3.3§< 0.001e′ septal5.9 ± 2.35.5 ± 2.16.2 ± 2.3§6.7 ± 2.5§< 0.001e′ mean7.1 ± 2.56.7 ± 2.47.4 ± 2.28.0 ± 2.6§< 0.001E/e′12 ± 512 ± 611 ± 412 ± 5 0.4EAT4.8 ± 2.14.5 ± 2.14.9 ± 1.85.4 ± 2.1§< 0.001PAT5.7 ± 2.85.4 ± 2.45.5 ± 2.96.4 ± 3.5§† 0.004Data are mean ± SD, median [interquartile range] or n (%). Final column reflects overall differences in EAT tertiles. ANOVA was used for normally distributed parameters and Chi-squared was used for categorical parametersACEi angiotensin-converting enzyme inhibitor; ARB angiotensin receptor blocker; BMI body mass index; COPD chronic obstructive pulmonary disease; DBP diastolic blood pressure; eGFR estimated glomerular filtration rate; HF heart failure; HFrEF heart failure with reduced ejection fraction; HFmrEF heart failure with mildly reduced ejection fraction; HFpEF heart failure with preserved ejection fraction; LAVI left atrial volumetric index; LVEF left ventricular ejection fraction; LVMI left ventricular mass index; MRA mineralocorticoid receptor antagonist; NT-proBNP N-terminal pro B‑type natriuretic peptide; NYHA New York Heart Association; RVD right ventricular dysfunction; SBP systolic blood pressure§ p0.05 versus HFrEF; † p0.05 versus HFmrEF Baseline Characteristics Data are mean ± SD, median [interquartile range] or n (%). Final column reflects overall differences in EAT tertiles. ANOVA was used for normally distributed parameters and Chi-squared was used for categorical parameters ACEi angiotensin-converting enzyme inhibitor; ARB angiotensin receptor blocker; BMI body mass index; COPD chronic obstructive pulmonary disease; DBP diastolic blood pressure; eGFR estimated glomerular filtration rate; HF heart failure; HFrEF heart failure with reduced ejection fraction; HFmrEF heart failure with mildly reduced ejection fraction; HFpEF heart failure with preserved ejection fraction; LAVI left atrial volumetric index; LVEF left ventricular ejection fraction; LVMI left ventricular mass index; MRA mineralocorticoid receptor antagonist; NT-proBNP N-terminal pro B‑type natriuretic peptide; NYHA New York Heart Association; RVD right ventricular dysfunction; SBP systolic blood pressure § p0.05 versus HFrEF; † p0.05 versus HFmrEF Baseline characteristics by BMI, EAT, and PAT The mean BMI in the entire cohort was 27 ± 6 kg/m2, mean EAT thickness was 4.8 ± 2.1 mm and the mean PAT thickness was 5.7 ± 2.8 mm. As demonstrated in the Electronic Supplemental Material (ESM) Figure S1, EAT and PAT thickness ranged from 1.3 to 22.8 mm and 1.3 to 20.8 mm, respectively. Tab. 1 and Figure S2 showed that HFpEF patients had significantly higher EAT and PAT thickness than patients with HFrEF and HFmrEF. Tables S1 and S2 showed that patients with higher EAT and PAT thickness had significantly higher BMI and were more likely to be older. In addition, patients with both higher EAT and PAT thickness had ischaemic heart disease significantly more often, while LVEF was higher. These patients also had a lower glomerular filtration rate (eGFR) (Tables S1 and S2). Finally, Figure S3 shows that women had significantly higher EAT thickness, whereas men had higher PAT thickness. In contrast, patients with high BMI did not show a significantly higher prevalence of ischaemic heart disease or lower eGFR (Table S3). Conversely, patients with a high BMI were more often women and more frequently had hypertension, diabetes mellitus, and atrial fibrillation, but also had higher LVEF (Table S3). Echocardiographic measures Table S1 and S2 showed that most echocardiographic parameters, such as E/e′ and ventricular mass index, did not differ across EAT and PAT quartiles. However, higher PAT thickness was significantly associated with lower left atrial volume and lower lateral e′ (p=0.019 and p=0.022, respectively). Patients with high BMI had lower left atrial volume index and higher septal and lateral e′ (Table S3). Clinical outcomes by BMI, EAT, and PAT In total, 217 patients (35%) experienced the combined outcome during a mean follow-up of 3.2 ± 1.9 years. Univariate Cox regression analysis demonstrated that higher EAT and PAT thickness were associated with a higher risk of the combined outcome [HR 1.15 (95% CI 1.03–1.30), p= 0.013; HR 1.25 (95% CI 1.10–1.41), p 0.001, respectively]; Tab. 2. Furthermore, for both EAT and PAT, this association remained significant after adjustment for BMI. However, the only association with PAT remained significant in the multivariate regression model 2, when adjusted for age, sex, BMI, HF type, and 14 additional univariably predictors of outcome [HR 1.49 (95% CI 1.10–2.03), p= 0.008], whereas this association was not seen for EAT thickness (Tab. 2). Moreover, restricted spline analyses showed approximately linear associations between EAT and PAT thickness with the risk of the combined outcome, with increasing EAT and PAT thickness associated with a higher HR (Fig. 4a and 3b). The prognostic cutoff for PAT was approximately 6 mm and for EAT approximately 5 mm. In contrast, a U-shaped association was observed between BMI and the combined outcome, as shown in Fig. 4c.Table 2Univariate and multivariate Cox regression modelling.Combined outcomeUnivariate HR (95% CI)p-valueMultivariate HR (95% CI) Model 1p-valueMultivariate HR (95% CI) Model 2p-valueAge, years1.06 (1.04–1.07)< 0.0011.06 (1.05–1.07)< 0.0011.02 (0.98–1.05) 0.1Sex, Women1.16 (0.89–1.52) 0.3Body mass index (kg/m2)1.00 (0.97–1.02) 0.8HF characteristicsHF type1.19 (1.02–1.39) 0.0281.20 (1.03–1.41) 0.0191.26 (0.87–1.84) 0.2Ischaemic heart disease1.49 (1.14–1.94) 0.0031.50 (1.14–1.94) 0.0031.67 (0.92–3.03) 0.094NYHA Class3.10 (2.37–4.06)< 0.0011.48 (0.92–2.34) 0.1ComorbiditiesHypertension1.32 (1.01–1.73) 0.0441.38 (1.08–1.82) 0.0251.21 (0.68–2.16) 0.5Atrial fibrillation, history1.37 (1.05–1.78) 0.0221.38 (1.05–1.82) 0.0201.0 (0.5–2.1) 1.0Diabetes mellitus1.60 (1.19–2.14) 0.0021.71 (1.26–2.30)< 0.0011.89 (1.01–3.50) 0.048COPD2.03 (1.47–2.78)< 0.0011.99 (1.44–2.78)< 0.0012.89 (1.43–5.71) 0.003Cardiothoracic surgery1.54 (1.10–2.15) 0.0111.51 (1.08–2.10) 0.0161.03 (0.50–2.12) 0.9Laboratory testseGFR, ml/min/1.73 m20.41 (0.34–0.51)< 0.0010.45 (0.37–0.56)< 0.0010.50 (0.22–1.16) 0.2NT-proBNP, ng/l1.67 (1.48–1.89)< 0.0011.72 (1.52–1.94)< 0.0011.72 (1.50–2.00)< 0.001MedicationBeta-blocker use1.13 (0.89–1.96) 0.2ACEi/ARB use0.78 (0.56–1.09) 0.2Loop diuretic use2.16 (2.16–3.00)< 0.0012.80 (1.92–4.19)< 0.0011.06 (0.51–2.20) 0.9Echocardiographic parametersLVEF (%)1.01 (1.00–1.02) 0.068LAVI1.02 (1.01–1.02) 0.0021.02 (1.01–1.02) 0.0031.02 (1.00–1.04) 0.034LVMI1.01 (1.00–1.01) 0.6e′ lateral0.96 (0.91–1.01) 0.1e′ septal0.87 (0.79–0.94)< 0.0010.82 (0.67–0.95)< 0.0010.78 (0.61–1.10) 0.064e′ mean0.92 (0.86–0.99) 0.0270.92 (0.86–0.99) 0.0251.22 (0.95–1.57) 0.1E/e′1.09 (1.05–1.12)< 0.0011.09 (1.05–1.12)< 0.0011.04 (0.98–1.12) 0.2EAT thickness, mm1.15 (1.03–1.30) 0.0131.18 (1.05–1.32) 0.0061.07 (0.87–1.32) 0.5PAT thickness, mm1.25 (1.10–1.41)< 0.0011.27 (1.13–1.44)< 0.0011.49 (1.10–2.03) 0.008CI Confidence interval (Logarithmic scale for NT-pro-BNP and eGFR; Per SD for EAT and PAT) EAT epicardial adipose tissue; HR hazard ratio; PAT pericardial adipose tissueMultivariate model 1 is used to adjust for BMI. Multivariate model 2 is used to adjust for age, sex, BMI, HF type and all univariate associated variables (i.e. ischaemic heart disease, NYHA class, eGFR, NT-proBNP, loop diuretic use, LAVI, e′ septal, e′ mean, E/e′ and presence/history of hypertension, atrial fibrillation, diabetes mellitus, COPD and cardiothoracic surgery)Fig. 3Associations between body mass index, epicardial and pericardial adipose tissue with outcome. Kaplan Meier curves depicted for the survival of combined outcome. BMI groups (a) p = 0.9; EAT quartiles (b) p = 0.04; PAT quartiles (c) p 0.001. BMI body mass index; COPD chronic obstructive pulmonary disease; EAT epicardial adipose tissue; HF heart failure; HFrEF heart failure with reduced ejection fraction; HFmrEF heart failure with mildly reduced ejection fraction; HFpEF heart failure with preserved ejection fraction; PAT pericardial adipose tissueFig. 4Spline curves for the hazard ratio of the combined outcome. This figure shows the relationship between the hazard ratio of the combined outcome and a EAT, b PAT and c BMI. Panels A and B were displayed up to the 95th percentile of the distribution. Odds ratio (solid line) and 95% confidence intervals (colored dashed lines) were based on the Cox proportinal hazard model. BMI body mass index; EAT epicardial adipose tissue; PAT pericardial adipose tissue Table 2Univariate and multivariate Cox regression modelling.Combined outcomeUnivariate HR (95% CI)p-valueMultivariate HR (95% CI) Model 1p-valueMultivariate HR (95% CI) Model 2p-valueAge, years1.06 (1.04–1.07)< 0.0011.06 (1.05–1.07)< 0.0011.02 (0.98–1.05) 0.1Sex, Women1.16 (0.89–1.52) 0.3Body mass index (kg/m2)1.00 (0.97–1.02) 0.8HF characteristicsHF type1.19 (1.02–1.39) 0.0281.20 (1.03–1.41) 0.0191.26 (0.87–1.84) 0.2Ischaemic heart disease1.49 (1.14–1.94) 0.0031.50 (1.14–1.94) 0.0031.67 (0.92–3.03) 0.094NYHA Class3.10 (2.37–4.06)< 0.0011.48 (0.92–2.34) 0.1ComorbiditiesHypertension1.32 (1.01–1.73) 0.0441.38 (1.08–1.82) 0.0251.21 (0.68–2.16) 0.5Atrial fibrillation, history1.37 (1.05–1.78) 0.0221.38 (1.05–1.82) 0.0201.0 (0.5–2.1) 1.0Diabetes mellitus1.60 (1.19–2.14) 0.0021.71 (1.26–2.30)< 0.0011.89 (1.01–3.50) 0.048COPD2.03 (1.47–2.78)< 0.0011.99 (1.44–2.78)< 0.0012.89 (1.43–5.71) 0.003Cardiothoracic surgery1.54 (1.10–2.15) 0.0111.51 (1.08–2.10) 0.0161.03 (0.50–2.12) 0.9Laboratory testseGFR, ml/min/1.73 m20.41 (0.34–0.51)< 0.0010.45 (0.37–0.56)< 0.0010.50 (0.22–1.16) 0.2NT-proBNP, ng/l1.67 (1.48–1.89)< 0.0011.72 (1.52–1.94)< 0.0011.72 (1.50–2.00)< 0.001MedicationBeta-blocker use1.13 (0.89–1.96) 0.2ACEi/ARB use0.78 (0.56–1.09) 0.2Loop diuretic use2.16 (2.16–3.00)< 0.0012.80 (1.92–4.19)< 0.0011.06 (0.51–2.20) 0.9Echocardiographic parametersLVEF (%)1.01 (1.00–1.02) 0.068LAVI1.02 (1.01–1.02) 0.0021.02 (1.01–1.02) 0.0031.02 (1.00–1.04) 0.034LVMI1.01 (1.00–1.01) 0.6e′ lateral0.96 (0.91–1.01) 0.1e′ septal0.87 (0.79–0.94)< 0.0010.82 (0.67–0.95)< 0.0010.78 (0.61–1.10) 0.064e′ mean0.92 (0.86–0.99) 0.0270.92 (0.86–0.99) 0.0251.22 (0.95–1.57) 0.1E/e′1.09 (1.05–1.12)< 0.0011.09 (1.05–1.12)< 0.0011.04 (0.98–1.12) 0.2EAT thickness, mm1.15 (1.03–1.30) 0.0131.18 (1.05–1.32) 0.0061.07 (0.87–1.32) 0.5PAT thickness, mm1.25 (1.10–1.41)< 0.0011.27 (1.13–1.44)< 0.0011.49 (1.10–2.03) 0.008CI Confidence interval (Logarithmic scale for NT-pro-BNP and eGFR; Per SD for EAT and PAT) EAT epicardial adipose tissue; HR hazard ratio; PAT pericardial adipose tissueMultivariate model 1 is used to adjust for BMI. Multivariate model 2 is used to adjust for age, sex, BMI, HF type and all univariate associated variables (i.e. ischaemic heart disease, NYHA class, eGFR, NT-proBNP, loop diuretic use, LAVI, e′ septal, e′ mean, E/e′ and presence/history of hypertension, atrial fibrillation, diabetes mellitus, COPD and cardiothoracic surgery) Univariate and multivariate Cox regression modelling. CI Confidence interval (Logarithmic scale for NT-pro-BNP and eGFR; Per SD for EAT and PAT) EAT epicardial adipose tissue; HR hazard ratio; PAT pericardial adipose tissue Multivariate model 1 is used to adjust for BMI. Multivariate model 2 is used to adjust for age, sex, BMI, HF type and all univariate associated variables (i.e. ischaemic heart disease, NYHA class, eGFR, NT-proBNP, loop diuretic use, LAVI, e′ septal, e′ mean, E/e′ and presence/history of hypertension, atrial fibrillation, diabetes mellitus, COPD and cardiothoracic surgery) Associations between body mass index, epicardial and pericardial adipose tissue with outcome. Kaplan Meier curves depicted for the survival of combined outcome. BMI groups (a) p = 0.9; EAT quartiles (b) p = 0.04; PAT quartiles (c) p 0.001. BMI body mass index; COPD chronic obstructive pulmonary disease; EAT epicardial adipose tissue; HF heart failure; HFrEF heart failure with reduced ejection fraction; HFmrEF heart failure with mildly reduced ejection fraction; HFpEF heart failure with preserved ejection fraction; PAT pericardial adipose tissue Spline curves for the hazard ratio of the combined outcome. This figure shows the relationship between the hazard ratio of the combined outcome and a EAT, b PAT and c BMI. Panels A and B were displayed up to the 95th percentile of the distribution. Odds ratio (solid line) and 95% confidence intervals (colored dashed lines) were based on the Cox proportinal hazard model. BMI body mass index; EAT epicardial adipose tissue; PAT pericardial adipose tissue In Fig. 3a, Kaplan-Meier curves survival-free period of the combined outcome are depicted. No association with the combined outcome was observed across the BMI groups or EAT quartiles (Log-rank p = 0.90; Fig. 3a and Log-rank p = 0.10; Fig. 3b, respectively). Conversely, patients with higher PAT thickness were significantly at risk for the combined outcome (Log-rank p 0.001, respectively; Fig. 3c). Furthermore, Figure S4 demonstrated Kaplan-Meier curves for the combined outcome, stratified by the HF groups. Patients with high PAT thickness within the HFrEF group and especially the HFpEF subgroup were significantly more at risk for the combined outcome than patients with lower PAT thickness (Log-rank p = 0.019 and p = 0.008, respectively), whereas this association was not present in patients with HFmrEF. In addition, no differences were observed according to EAT quartiles within the HF groups (Figure S5). Discussion The main finding of the present study was that both EAT and PAT thickness, measured by standard echocardiography, were associated with poor outcomes in patients with established HF, whereas this association was not observed for higher BMI. We also demonstrated that these AT depots were differentially associated with outcomes across HF phenotypes, including HFrEF, HFmrEF, and HFpEF. Additionally, we showed that patients with high EAT and PAT thickness had different clinical characteristics compared to patients with high BMI. These findings suggest that increased EAT and PAT thickness are independently associated with poor clinical outcomes in patients with HF and predominantly in HFpEF. The association observed between high EAT thickness and an increased risk of all-cause mortality and HF hospitalization aligns with previous studies [16, 17]. For instance, Pugliese et al. demonstrated in a cohort of 188 HFpEF patients that high EAT thickness was significantly associated with an increased risk for HF hospitalization and cardiovascular death [16]. However, we have now demonstrated similar findings in a larger, all-comer HF population. It is suggested that EAT, in physiological conditions, can serve as a metabolic reservoir due to its protective brown-fat features. However, excessive EAT accumulation may induce a shift towards a pro-inflammatory phenotype characterized by the release of a broad range of adipocytokines that may damage the myocardium and subsequently cause, e.g., HF and poor outcomes [18, 19]. Additionally, it was seen that EAT thickness was significantly higher in HFpEF patients than their HFrEF counterparts, which was in line with a previous study [20]. To date, less attention has been given in clinical research to the potential associations between PAT and clinical outcome [21]. However, interest is increasing, as colleagues from the SUMMIT trial have recently demonstrated that Tirzepatide significantly reduces PAT, more than EAT [22]. In the present study, we have demonstrated that higher PAT thickness was significantly associated with an increased risk of the combined outcome independent of BMI, age, sex, and other variables associated with outcome in univariable analysis. It has been hypothesized that an increased burden of PAT promotes multiple deleterious metabolic processes, including insulin resistance, aldosterone secretion, and systemic inflammation [22, 23]. Ultimately, this may result in a higher risk of poor outcome, whereas PAT thickness can thus be used as an important and easily obtainable prognostic marker in chronic HF [24]. Clinical implications Recent studies have advocated the use of alternative anthropometric measures such as waist-to-hip ratio, waist-to-height ratio, and body roundness index [4, 25]. Sub-studies from the PARAGON-HF and PARADIGM-HF trials demonstrated that these anthropometric measures more closely reflect the amount and distribution of body fat and provide prognostic value in HF, beyond overall obesity [4, 23, 25, 26]. However, these anthropometric measures do not account for the characteristics of different fat depots and only provide an estimate of ‘central obesity’ [6]. Unlike these anthropometric measures, EAT and PAT reflect local fat accumulation and may directly influence the myocardium due to their metabolically active and pro-inflammatory profile. Therefore, they may provide a more accurate reflection of cardiovascular risk and yield additional clinical relevance. Although CMR is more precise for the quantification of EAT and PAT volumes, standard echocardiography may be a useful bedside alternative to easily assess the thickness of these fat depots [27]. Although echocardiography cannot measure the entire volume of EAT and PAT, the assessment of both EAT and PAT based on a thickness on 2D echocardiography should therefore be regarded as an easily obtainable, anthropometric measurement that may serve as an important screening tool in patients with HF [16, 28]. For PAT thickness, the prognostic cutoff point was approximately 6 mm, and for EAT this was 5 mm. Limitations This study has several limitations. First, it was conducted in a tertiary HF referral centre, which may have introduced referral bias, and most of the study population included HFrEF patients. Second, we did not have data to adjust for visceral adipose tissue, and anthropometric measures such as waist-to-hip ratio and waist-to-height ratio were also not available. Third, CMR and computed tomography (CT) generally provide a more precise quantification of the total amount of EAT and PAT. However, CMR and CT are often not readily available in clinical practice. In contrast, routine echocardiography provides a quick and easily accessible alternative for assessing AT thickness, which may serve as an important prognostic marker, although it is limited by lower resolution and inability to capture total fat volume [27, 29]. Conclusion This study demonstrates that both EAT and PAT thickness were associated with poor outcomes, irrespective of BMI, in an unselected population of HF. In a multivariable model, only PAT thickness remained significantly associated with poor outcome. The association between PAT thickness and poor outcome was mainly observed in HFpEF. These data suggest that PAT thickness, more than EAT, may serve as an easily obtainable prognostic marker in patients with HF, especially HFpEF and irrespective of overall obesity. Supplementary Information ESM1: Supplementary material 1