Pericardial Fat, Intrathoracic Fat, and Measures of Left Ventricular Structure and Function
The Framingham Heart Study
Background— Pericardial fat has been implicated in the pathogenesis of obesity-related cardiovascular disease. Whether the associations of pericardial fat and measures of cardiac structure and function are independent of the systemic effects of obesity and visceral adiposity has not been fully explored.
Methods and Results— Participants from the Framingham Heart Study (n=997; 54.4% women) underwent chest and abdominal computed tomography and cardiovascular magnetic resonance imaging between 2002 and 2005. Pericardial fat, intrathoracic fat, and visceral adipose tissue quantified from multidetector computed tomography, along with body mass index and waist circumference, were examined in relation to cardiovascular magnetic resonance measures of left ventricular (LV) mass, LV end-diastolic volume, and left atrial dimension. In women, pericardial fat (r=0.20 to 0.35, P<0.001), intrathoracic fat (r=0.25 to 0.37, P<0.001), visceral adipose tissue (r=0.24 to 0.45, P<0.001), body mass index (r=0.36 to 0.53, P<0.001), and waist circumference (r=0.30 to 0.48, P<0.001) were directly correlated with LV mass, LV end-diastolic volume, and left atrial dimension. In men, pericardial fat (r=0.19 to 0.37, P<0.001), intrathoracic fat (r=0.17 to 0.31, P<0.001), visceral adipose tissue (r=0.19 to 0.36, P<0.001), body mass index (r=0.32 to 0.44, P<0.001), and waist circumference (r=0.34 to 0.44, P<0.001) were directly correlated with LV mass and left atrial dimension, but LV end-diastolic volume was not consistently associated with adiposity measures. Associations persisted after multivariable adjustment but not after additional adjustment for body weight and visceral adipose tissue, except for pericardial fat and left atrial dimension in men.
Conclusions— Pericardial fat is correlated with cardiovascular magnetic resonance measures, but the association is not independent of or stronger than other ectopic fat stores or proxy measures of visceral adiposity. An important exception is left atrial dimension in men. These results suggest that the systemic effects of obesity on cardiac structure and function may outweigh the local pathogenic effects of pericardial fat.
Received October 15, 2008; accepted January 26, 2009.
Pericardial fat is an ectopic fat depot associated with measures of adiposity1–7 that may exert a paracrine effect on nearby anatomic structures. We have previously shown that pericardial fat, but not intrathoracic fat, is associated with coronary artery calcification.7 Local toxic effects of pericardial fat may also manifest as abnormalities of left ventricular (LV) structure and function. Previous small studies have demonstrated that pericardial fat is associated with measures of LV mass,8–10 left atrial (LA) size, and impaired diastolic filling10,11 and is negatively correlated with cardiac index.12 These prior studies, however, are limited by their small sample size, use of echocardiography to estimate the thickness of epicardial fat, and lack of adjustment for important covariates. Therefore, whether the association of pericardial fat and measures of LV structure and function is independent of the systemic effects of obesity has not been fully explored.
Clinical Perspective p 1591
Thus, we sought to examine the correlation of pericardial fat, intrathoracic fat, visceral abdominal fat, and measures of LV structure and function by cardiovascular magnetic resonance (CMR). Given the lack of anatomic contact between intrathoracic fat and visceral abdominal fat and these measures, we hypothesized that only pericardial fat would be associated with measures of LV structure and function.
In 1948, the Framingham Heart Study original cohort was enrolled, totaling 5209 men women and men 28 to 62 years of age. The offspring and spouses of the original cohort were enrolled in the Offspring Study in 1971.13,14 The present analysis is made up of offspring cohort participants who participated in both the multidetector computed tomography (MDCT) and CMR substudies.
Between 1998 and 2001, 3539 offspring cohort participants attended the seventh examination cycle of the Framingham Heart Study. As part of the MDCT substudy, 1418 (40.1%) underwent MDCT scanning from 2002 to 2005 for coronary and abdominal aortic calcium assessment, of whom 1372 had interpretable pericardial fat measures.15 In addition, 1794 participants underwent CMR during a similar time period.16 Of the 1418 participants who underwent MDCT imaging, 1372 had interpretable pericardial fat measures, 1036 also had CMR measures, 1006 of whom attended the seventh examination cycle, and 7 had an incomplete covariate profile, resulting in a total sample size of 997.
The study protocol was approved by the institutional review boards of the Boston University Medical Center, Massachusetts General Hospital, and Beth Israel Deaconess Medical Center. All subjects provided written informed consent.
MDCT Scan Protocol and Analysis
In a supine position, participants underwent radiographic assessment with 8-slice MDCT (LightSpeed Ultra, General Electric, Milwaukee, Wis). The heart was imaged on average with 48 contiguous 2.5-mm slices with a prospectively ECG triggered scanning protocol (120 kVp; 400 mA; temporal resolution, 330 ms). In addition, 25 slices 5 mm thick (120 kVp; 400 mA; gantry rotation time, 500 ms; table feed, 3:1) were acquired covering 125 mm beginning at the level of S1.
Total thoracic and pericardial fat tissue volumes were measured with an offline workstation (Aquarius 3D Workstation, TeraRecon Inc, San Mateo, Calif). Because absolute Hounsfield units (HU) pixel values correspond to properties of the imaged tissues, we used a predefined image display setting (window width, −195 to −45 HU; window center, −120 HU) that identifies pixels that correspond to adipose tissue. Total thoracic and pericardial fat volumes were measured with a semiautomatic segmentation technique. Total thoracic fat volume included adipose tissue located in the pericardium and in the thorax from the level of the right pulmonary artery to the diaphragm and the chest wall to the descending aorta. Pericardial fat volume was defined as adipose tissue located within the pericardial sac. Interreader reproducibility was excellent (interclass correlation coefficient for total thoracic fat, 0.98; for pericardial fat, 0.95).7 Intrathoracic fat was created as a derived variable from the difference between total thoracic and pericardial fat, allowing the creation of 2 distinct fat depots; this differs from our prior definition of intrathoracic fat.7 Visceral adipose tissue (VAT) volumes were quantified with the above-mentioned image display windows17; interclass correlations were 0.99.17
CMR Protocol and Analysis
CMR imaging was performed with a 1.5-T whole-body scanner (Philips Medical Systems, Best, the Netherlands). Short-axis cine images encompassing the left ventricle were obtained using a steady-state free precession sequence18 with contiguous 10-mm slices and a temporal resolution of 30 to 35 ms.
CMR data were analyzed by a single observer blinded to clinical data using a commercial workstation (EasyVision 5, Philips Medical Systems). LV endocardial borders were traced manually at end systole and end diastole. LV epicardial borders were traced at end diastole; LV end-diastolic volume (LVEDV) and LV myocardial volume were calculated using the summation of disks method. LV mass was calculated by multiplying an accepted myocardial density (1.05 g/cm3) by the calculated volume of myocardium. LA dimension was measured in the AP direction from an axially oriented image.
Risk Factor and Covariate Assessment
Risk factors were obtained from the seventh Framingham offspring examination (1998 to 2001). Waist circumference was measured at the level of the umbilicus; body mass index (BMI) was defined as weight (kilograms) divided by height squared (meters). Fasting plasma glucose, total and high-density lipoprotein cholesterol, and serum triglycerides were measured on fasting morning samples. Diabetes was defined as fasting plasma glucose ≥126 mg/dL or treatment for diabetes (hypoglycemic agent or insulin). Impaired fasting glucose was defined as fasting plasma glucose of 100 to 125 mg/dL in the absence of treatment for diabetes mellitus. Current smokers were defined as those who smoked on average at least 1 cigarette per day in the past year. Alcohol consumption was quantified via physician-administered questionnaires. Women were considered postmenopausal if menses had stopped for ≥1 year. Hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or antihypertensive treatment.
Because of differences in the distribution of pericardial fat and CMR variables, all analyses were stratified by sex. All adiposity measures were standardized to a mean of 0 and an SD of 1, within each sex, to facilitate comparison of regression coefficients between different fat depots. We considered a full panel of adiposity traits in relation to CMR variables: BMI, waist circumference, VAT, pericardial fat, and intrathoracic fat. In particular, because intrathoracic fat is not in direct anatomic contact with the myocardial structures examined here, it was used as a natural control. Sex-specific age-adjusted Pearson correlations between all adiposity measures and LA dimension, LVEDV, and LV mass were calculated. Next, multivariable regression models were constructed with each adiposity variable as the exposure and the CMR variables modeled as the outcomes. Two models were considered: the multivariable-adjusted model, which included adjustment for age, smoking, alcohol, menopause, hormone replacement therapy, systolic blood pressure, hypertension treatment, and height, and a second model that was additionally adjusted for body weight. Because of the inclusion of body weight in body surface area calculations, we did not index our measures to body surface area. Models using BMI as an exposure did not additionally adjust for height. Formal sex interactions were tested in the first model.
Several secondary analyses were performed. Models examining intrathoracic and pericardial fat in relation to CMR variables were additionally adjusted for VAT, for diabetes and cardiovascular disease (CVD), and in models excluding prevalent CVD.
SAS version 8.0 (SAS Institute Inc, Cary, NC) was used to perform all computations. A 2-tailed value of P<0.05 was considered significant.19
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Study Sample Characteristics
Overall, the study sample comprised 542 women and 455 men with a mean age of 60 years. Overall, 26.3% of women and 27.0% of men were obese. Additional study sample characteristics can be found in Table 1.
All adiposity measures were correlated to CMR measures in women, with correlations ranging from 0.28 to 0.53 (LA dimension), 0.20 to 0.36 (LVEDV), and 0.35 to 0.48 (LV mass) (Table 2). Similar correlations were observed in men, except for LVEDV, which was not correlated with VAT, pericardial fat, or intrathoracic fat.
In women, all adiposity measures were associated with LA dimension (Table 3). Per 1-SD increase in pericardial fat, LA dimension was 1.18 cm larger (P<0.0001). In contrast, per 1-SD increase in intrathoracic fat, LA dimension was 1.37 cm larger (P<0.0001; P for difference between pericardial and intrathoracic fat=0.22). After adjustment for body weight, all associations between fat depots and LA dimension were attenuated (all P>0.18). Among men, per 1-SD increase in pericardial fat, LA dimension was 1.74 cm larger (P<0.0001) compared with 1.47 cm larger per 1-SD increase in intrathoracic fat (P<0.0001; P for difference between pericardial and intrathoracic fat=0.12). After adjustment for body weight, only the association between pericardial fat and LA dimension remained significant (P=0.002).
All measures of adiposity were associated with LVEDV in women. In particular, per 1-SD increase in BMI, LVEDV was 6.88 cm3 larger (P<0.0001). Both pericardial fat and intrathoracic fat were associated with LVEDV; the regression coefficient was larger for intrathoracic (3.73) compared with pericardial (3.01) fat, although the probability value for the difference between pericardial and intrathoracic fat was not significant (P=0.29). In men, VAT, pericardial fat, and intrathoracic fat were not associated with LVEDV.
All measures of adiposity were associated with LV mass in women. Pericardial fat and intrathoracic fat had similar associations with LV mass (4.33 versus 4.48 g per 1-SD increase in fat; P for difference between pericardial and intrathoracic fat=0.85). After adjustment for body weight, pericardial fat remained associated with LV mass (P=0.01), although the magnitude of the association was decreased. In men, all measures of adiposity were associated with LV mass. Similar to women, there was no difference in the magnitude of association between pericardial fat, intrathoracic fat, and LV mass (3.83 versus 3.75 g per 1-SD increase in fat; P for difference between pericardial and intrathoracic fat=0.98).
Models examining the relation between pericardial fat and intrathoracic fat in relation to CMR measurements were additionally adjusted for VAT (Table 4). After adjustment for VAT, nearly all associations were attenuated, except the association between pericardial fat and LA dimension in men (P=0.0004).
When models were additionally adjusted for diabetes and CVD, results were not materially different (Table 4). Similarly, when analyses first excluded individuals with CVD, results were similar (Table 4).
In the present community-based study of nearly 1000 participants undergoing contemporaneous CMR and MDCT examinations, we found that pericardial fat volume was associated with LV mass, LVEDV, and LA dimension in women and with LV mass and LA dimension in men. These associations persist after multivariable adjustment but not after accounting for body weight or VAT, with the exception of LA dimension in men. There is a similar pattern of association of intrathoracic fat with LV structure and function after multivariable adjustment but not after adjustment for body weight or VAT. Finally, BMI and waist circumference also are associated with CMR measures after multivariable adjustment. These results suggest that any potential local pathological effects of pericardial fat on LV structure and function are overwhelmed by the systemic effects of obesity. Our findings do not suggest that pericardial fat is a better correlate of cardiac structure and function than VAT or other more easily conducted anthropometric measures of adiposity.
In the Context of the Current Literature
Pericardial fat has been found to correlate with LV mass across a range of BMI values8–10 and with impaired diastolic filling and atrial enlargement in morbid obesity.11 It has been proposed that the direct anatomic proximity of pericardial fat to the myocardium allows a paracrine interaction that may affect cardiac morphology and function. Such local effects are hypothesized to render pericardial fat a stronger correlate of cardiac structure and function than more general measures of adiposity.20 In the present study, we confirmed that pericardial fat is associated with CMR measures but observed that it is no more correlated than other ectopic fat depots and proxy measures of adiposity. We additionally found that the association between pericardial fat and CMR measures is attenuated once VAT is taken into account. These data suggest this may not be the case for pericardial fat and that the systemic effects of generalized adiposity may overwhelm the local effects of pericardial fat.
Adiposity may affect LV structure and function via mechanical, paracrine, and systemic processes. Compression of the heart by pericardial or intrathoracic fat deposits may decrease LV diastolic filling, leading to atrial dilation. The presence of impaired diastolic function and increased LA dimension without LV hypertrophy in uncomplicated obesity suggests a possible mechanical role for regional adiposity in cardiac structure and function that is independent of systemic obesity-related disorders such as diabetes and hypertension.21 Our finding that pericardial fat remained a significant correlate of LA dimension in men after adjustment for body weight and VAT supports this notion. We did not, however, observe an association between pericardial fat and LVEDV in men.
Direct contact between adipose tissue and the myocardium may affect LV structure and function via paracrine secretion. Pericardial fat lies directly on the myocardium and shares the same coronary blood supply, with no fascia separating the 2 layers.22 Samples of pericardial fat from 42 patients undergoing coronary artery bypass graft surgery showed increased mRNA and protein levels of chemokine (monocyte chemoattractant protein-1) and inflammatory cytokines (interleukin [IL]-1β, IL-6, IL-6sR, and tumor necrosis factor-α) relative to subcutaneous fat and independent of obesity, diabetes, and statin, or angiotensin-converting enzyme inhibitor use.23 Another coronary artery bypass graft study found pericardial fat to be a source of adiponectin and resistin, in addition to monocyte chemoattractant protein-1, IL-6, IL-6sR, and tumor necrosis factor-α.24 However, pericardial fat concentrations of inflammatory biomarkers in coronary artery bypass graft patients did not correlate with plasma concentrations, suggesting that the release of such markers by pericardial fat, a relatively small fat depot, is not great enough to be detected systemically.23
Pericardial fat also has been found to correlate with coronary artery calcification after multivariable and VAT adjustment, suggesting that fat depots in anatomic contact with the vasculature may exert local pathological effects.7 In light of our findings in the present study that associations between pericardial fat and CMR measures do not generally persist after adjustment for body weight and VAT and that pericardial fat, which is in direct anatomic contact with myocardium, is no more correlated with measures of LV structure and function than intrathoracic fat, it appears that the paracrine effects of pericardial fat may be more pronounced for coronary artery calcification than for general measures of LV structure and function.
Compared with the possible mechanical and paracrine effects of pericardial fat, the systemic effects of obesity on cardiac structure and function have been well described. Obesity is strongly associated with diabetes, hypertension, dyslipidemia, and CVD.25,26 Hypertension is an independent risk factor for LV hypertrophy and increased LV mass, but cardiac hypertrophy is observed even in normotensive obese patients.27 This may be due to hemodynamic changes resulting from the increased blood volume and flow required to adequately perfuse increased body mass.
Strengths and Limitations
Compared with earlier work on pericardial fat and measures of LV structure and function, strengths of this study include the large sample size that includes a wide BMI range, reducing the risk of ascertainment bias. Additional strengths include detailed longitudinal assessment of covariates, minimizing the risk of misclassification, and the quantification of pericardial fat volumes rather than fat thickness. Some limitations warrant discussion. In particular, the cross-sectional study design limits inferences of causality, and the predominantly white study sample may limit generalizability to other ethnic groups. Unmeasured factors such as sleep apnea may partially account for our findings. Finally, because computed tomography scanning was performed without heart rate control, it is possible that measurements of pericardial fat may have been affected by motion artifacts.
Pericardial fat is correlated with measures of LV structure and function but not independently of or more strongly than other ectopic fat stores and proxy measures of visceral adiposity. An important exception is LA dimension in men. These results suggest that the systemic effects of obesity on cardiac structure and function may outweigh the local pathogenic effects of pericardial fat.
Sources of Funding
This work was supported by the National Heart, Lung and Blood Institute’s Framingham Heart Study (N01-HC-25195 and R01-HL70279).
Iacobellis G, Ribaudo MC, Assael F, Vecci E, Tiberti C, Zappaterreno A, Di Mario U, Leonetti F. Echocardiographic epicardial adipose tissue is related to anthropometric and clinical parameters of metabolic syndrome: a new indicator of cardiovascular risk. J Clin Endocrinol Metab. 2003; 88: 5163–5168.
Wheeler GL, Shi R, Beck SR, Langefeld CD, Lenchik L, Wagenknecht LE, Freedman BI, Rich SS, Bowden DW, Chen MY, Carr JJ. Pericardial and visceral adipose tissues measured volumetrically with computed tomography are highly associated in type 2 diabetic families. Invest Radiol. 2005; 40: 97–101.
Rosito GA, Massaro JM, Hoffmann U, Ruberg FL, Mahabadi AA, Vasan RS, O'Donnell CJ, Fox CS. Pericardial fat, visceral abdominal fat, cardiovascular disease risk factors, and vascular calcification in a community-based sample: the Framingham Heart Study. Circulation. 2008; 117: 605–613.
Kankaanpaa M, Lehto HR, Parkka JP, Komu M, Viljanen A, Ferrannini E, Knuuti J, Nuutila P, Parkkola R, Iozzo P. Myocardial triglyceride content and epicardial fat mass in human obesity: relationship to left ventricular function and serum free fatty acid levels. J Clin Endocrinol Metab. 2006; 91: 4689–4695.
Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: Framingham study, 18-year follow-up. In: Kannel WB, Fordon T, eds. The Framingham Study: An Epidemiological Investigation of Cardiovascular Disease. Washington, DC: Department of Health, Education, and Welfare; 1973. DHEW publication No. NIH 74–599; section 30.
Fox CS, Massaro JM, Hoffmann U, Pou KM, Maurovich-Horvat P, Liu CY, Vasan RS, Murabito JM, Meigs JB, Cupples LA, D'Agostino RB Sr, O'Donnell CJ. Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation. 2007; 116: 39–48.
Oyama N, Gona P, Salton CJ, Chuang ML, Jhaveri RR, Blease SJ, Manning AR, Lahiri M, Botnar RM, Levy D, Larson MG, O'Donnell CJ, Manning WJ. Differential impact of age, sex, and hypertension on aortic atherosclerosis: the Framingham Heart Study. Arterioscler Thromb Vasc Biol. 2008; 28: 155–159.
Maurovich-Horvat P, Massaro J, Fox CS, Moselewski F, O'Donnell CJ, Hoffmann U. Comparison of anthropometric, area- and volume-based assessment of abdominal subcutaneous and visceral adipose tissue volumes using multi-detector computed tomography. Int J Obes (Lond). 2007; 31: 500–506.
Finn JP, Simonetti OP. Pulse sequence design in MRI. In: Edelman RR, ed. Clinical Magnetic Resonance Imaging. 2nd ed. Philadelphia, Pa: WB Saunders Co; 1996: 168–169.
SAS/STAT User’s Guide, Version 8. Cary, NC: SAS Institute Inc. 2000.
Mazurek T, Zhang L, Zalewski A, Mannion JD, Diehl JT, Arafat H, Sarov-Blat L, O'Brien S, Keiper EA, Johnson AG, Martin J, Goldstein BJ, Shi Y. Human epicardial adipose tissue is a source of inflammatory mediators. Circulation. 2003; 108: 2460–2466.
Hubert HB, Feinleib M, McNamara PM, Castelli WP. Obesity as an independent risk factor for cardiovascular disease: a 26-year follow-up of participants in the Framingham Heart Study. Circulation. 1983; 67: 968–977.
Pericardial fat, or fat that surrounds the heart, may be associated with obesity-related cardiovascular disease. We explored whether the associations of pericardial fat and measures of cardiac structure and function are linked. We measured pericardial fat in participants from the Framingham Heart Study and assessed measures of cardiac structure and function. We found that multiple different measures of fat were associated with cardiac measures of structure and function, but none persisted after accounting for overall body weight and visceral abdominal fat, the most metabolically active fat depot. An important exception was the relation of pericardial fat and left atrial dimension in men. These results suggest that the systemic effects of obesity on cardiac structure and function may outweigh the local pathogenic effects of pericardial fat.
Guest Editor for this article was Robert H. Eckel, MD.