Association between Hepatic Triglyceride Content and Left Ventricular Diastolic Function in a Population-based Cohort: The Netherlands Epidemiology of Obesity Study

Introduction

Obesity has reached epidemic proportions during the past decades and is a well-established risk factor for various diseases, including nonalcoholic fatty liver disease (NAFLD) and cardiovascular disease (1). NAFLD covers a spectrum of simple steatosis to nonalcoholic steatohepatitis and cirrhosis (2). It is the most common liver disease, with a prevalence of 20%–30% in the general population, increasing to 70%–90% among persons who are obese or have type 2 diabetes (3). NAFLD is therefore considered as a hepatic manifestation of the metabolic syndrome. NAFLD, and in particular nonalcoholic steatohepatitis, has been associated with increased prevalence (4,5) and incidence (3) of cardiovascular disease and with decreased myocardial phosphate metabolism (6). Cardiac disease is increasingly observed in persons with obesity, including subclinical impairment of left ventricular (LV) diastolic function, a precursor to overt heart failure (7). Abnormalities of diastolic function have a major role in exercise intolerance in patients with heart failure (8). Nevertheless, limited data are available concerning the relationship between NAFLD and myocardial function. Individuals with metabolic syndrome have a twofold risk of developing cardiovascular disease (9). Potential underlying mechanisms, however, remain unclear. Cardiac myocyte contractile function is decreased by macrophages (10). Although speculative, it can be hypothesized that macrophage recruitment in individuals with NAFLD could be present not only in the liver, but also in the heart, causing diastolic dysfunction. Alternatively, it has been suggested that increased intrahepatic cytokine expression plays a key role in the progression of cardiovascular disease (3).

Furthermore, the individual components that define the metabolic syndrome (three of the following five: high waist circumference, high serum triglyceride level, decreased serum high-density lipoprotein cholesterol level, high blood pressure, high fasting plasma glucose level) are also considered risk factors of both NAFLD and cardiovascular disease. This means that these components, and possibly the metabolic syndrome itself, may be responsible for an observed association between hepatic triglyceride content and diastolic function (3). In other words, it is yet unclear whether the association between NAFLD and cardiovascular disease is a consequence of shared risk factors within the metabolic syndrome or exists independently of these risk factors.

Localized hydrogen 1 (¹H) magnetic resonance (MR) spectroscopy is a sensitive, quantitative, noninvasive method for measuring hepatic triglyceride content (11). Cardiac MR imaging is a highly accurate and reproducible technique for assessing LV diastolic function (12,13). We hypothesized that hepatic triglyceride content is associated with LV diastolic function in a population-based cohort independent of the possible confounding effect of the metabolic syndrome. Therefore, the purpose of this study was to investigate the association between hepatic triglyceride content and LV diastolic function while taking potential confounding factors into account, including the components of the metabolic syndrome.

Materials and Methods

Study Population

Men and women aged 45–65 years with a self-reported body mass index (BMI) of 27 kg/m² or higher from Leiden and the surrounding area (Midwest of the Netherlands) were eligible to participate in +the Netherlands Epidemiology of Obesity (NEO) study. In addition, all inhabitants aged 45–65 years of one municipality (Leiderdorp) were invited irrespective of their BMI, which allowed for a reference distribution of BMI. The study population of the present analysis consists of participants who had undergone MR imaging and ¹H MR spectroscopy. Exclusion criteria for this analysis were a history of cardiovascular disease, history of liver disease, alcohol consumption of more than 10 units per day, and use of statins and/or other lipid-lowering drugs. The study was approved by the medical ethics committee of the Leiden University Medical Center, and all participants provided written informed consent.

Study Design

The present study is a cross-sectional analysis of baseline data from the NEO study. The NEO study is a population-based, prospective cohort study. Detailed information about the study design and data collection is available in Appendix E1 (online).

Metabolic Syndrome

Characteristics of the metabolic syndrome were based on the updated National Cholesterol Education Program Adult Treatment Panel III definition (14) (see Appendix E1 [online]).

MR Studies

MR imaging and spectroscopy were performed with a 1.5-T whole-body MR unit (Philips Medical Systems, Best, the Netherlands). More detailed information, including imaging parameters, can be found in Appendix E1 (online).

MR Spectroscopy

Hepatic ¹H MR spectra were obtained as described previously (15). In short, an 8-mL voxel was positioned in the right lobe of the liver. A point-resolved spectroscopy sequence was used to acquire spectroscopic data during continuous breathing with automated shimming. Spectra were obtained with and without water suppression. Spectral data were fitted by using Java-based MR user interface software (jMRUI, version 3.0; developed by A. van den Boogaart, Katholieke Universiteit Leuven, Leuven, Belgium) (16). Mean line widths of the spectra were calculated. The resonances that were fitted and used for calculation of the triglycerides were methylene (peak at 1.3 ppm, [CH₂]_n) and methyl (peak at 0.9 ppm, CH₃). The hepatic triglyceride content relative to water was calculated with the following formula: (signal amplitude of methylene + methyl)/(signal amplitude of water) × 100.

MR Imaging

Abdominal visceral adipose tissue (VAT) and subcutaneous adipose tissue areas were quantified with a turbo spin-echo MR imaging protocol. At the level of the fifth lumbar vertebra, three transverse images were acquired during one breath hold. The entire heart was imaged in the short-axis orientation by using electrocardiographically gated breath-hold balanced steady-state free precession imaging to assess LV dimensions and mass. To determine diastolic function, an electrocardiographically gated gradient-echo sequence was performed with velocity encoding to measure blood flow across the mitral valve. Diastolic parameters included peak filling rates of the early filling phase (E) and atrial contraction (A) and their ratio (E/A ratio). Image postprocessing was performed with in-house–developed software packages (MASS and FLOW; Leiden University Medical Center, Leiden, the Netherlands), and decisions were made by consensus between two experienced observers (R.L.W. and H.J.L., with 5 and >15 years of experience in cardiovascular MR imaging, respectively).

Statistical Analyses

Additional information about the statistical analyses is provided in Appendix E1 (online). Baseline characteristics of participants are summarized as means ± standard deviations, medians and 25th and 75th percentiles, or as percentages according to categories of BMI. To this extent, participants were stratified into subgroups according to World Health Organization criteria (BMI of <25 kg/m², 25–30 kg/m², and ≥30 kg/m²). Comparisons among groups were tested with the two-tailed independent samples t test or the χ² test where appropriate. Multivariate linear regression analyses were used to study the association between hepatic triglyceride content and E/A ratio while adjusting for potential confounders in different models (Appendix E1 [online]). Hepatic triglyceride content showed a right-skewed distribution; to use this variable in the regression analysis, a log-transformation was applied (log hepatic triglyceride content). We examined the presence of interaction between hepatic triglyceride content and BMI in their association with E/A ratio by including product terms to the final model. Regression (β) coefficients, 95% confidence intervals (CIs), and P and R² values were reported. P < .05 was considered indicative of a statistically significant difference. Statistical analysis was performed with software (SPSS for Windows, version 17.0 [SPSS, Chicago, Ill] and STATA, version 12 [STATA, College Station, Tex]).

Results

Between September 3, 2008, and September 28, 2012, 6673 participants were included in the NEO study, of whom 2580 underwent MR imaging and MR spectroscopy. Of those 2580 subjects, 1207 underwent cardiovascular MR imaging. Cardiovascular MR imaging failed because of technical errors in 35 participants. In another 246 participants, ¹H MR spectroscopy of the liver could not be completed owing to technical errors (n = 241) or because the participant felt unwell (n = 5), for example because of claustrophobia. In our high-throughput study protocol, only a limited time slot was available per subject, and this did not allow for repeat imaging when technical failures were recognized. The failure rate of MR spectroscopy was not related to age, sex, BMI, waist circumference, VAT, total body fat, or LV E/A ratio. Participants in whom MR spectroscopy was unsuccessful had higher subcutaneous adipose tissue compared with those who successfully underwent MR spectroscopy (mean, 738 cm³ ± 298 vs 701 cm³ ± 290, respectively; P = .04). Ultimately, 926 participants successfully underwent cardiovascular and abdominal MR imaging and ¹H MR spectroscopy of the liver. Participants with a history of cardiovascular disease (n = 52), liver disease (n = 9), and alcohol consumption of more than 10 units per day (n = 7) and those taking statins and other lipid-lowering drugs (n = 100) were consecutively excluded. Furthermore, participants with missing data (n = 44) were excluded. Finally, 714 participants (45% men, 98% white) were included in the present analysis (Fig 1). The subjects had a mean age of 55.3 years ± 6.2 (men, 54.8 years ± 6.6; women, 55.6 years ± 5.8; P > .05), a median BMI of 25.6 kg/m² (25th and 75th percentiles, 22.9 kg/m² and 27.9 kg/m²), and a median hepatic triglyceride content of 2.60% (25th and 75th percentiles, 1.30% and 6.05%). The hepatic triglyceride content ranged from 0.3% to 62.9%. Mean line widths of the spectra were 44.7 Hz ± 14.9 (water, 4.7 ppm), 46.7 Hz ± 18.9 (methylene, 1.3 ppm), and 43.2 Hz ± 18.4 (methyl, 0.9 ppm).

Figure 1:

Download as PowerPointOpen in Image Viewer

Participant characteristics are shown in Table 1 and Figure 2. The median hepatic triglyceride content was highest in the obese subgroup. Furthermore, the prevalence of the metabolic syndrome was markedly higher in the obese subgroup (P < .05).

Figure 2:

Download as PowerPointOpen in Image Viewer

LV end-diastolic volume and LV mass indexed to body surface area were higher in the obese subgroup (P < .05) (Table 2). Although cardiac output was higher in the overweight and obese subgroups compared with the normal-weight subgroup, the cardiac index was similar among groups. Ejection fraction was similar between the normal-weight participants and overweight participants but was slightly higher in normal-weight compared with obese participants. Diastolic function was lower in the obese subgroup, as demonstrated by a lower E/A ratio compared with the normal-weight and overweight subgroups (P < .05).

There was no statistically significant interaction between log hepatic triglyceride content and the three World Health Organization categories or with BMI as a continuous variable in their association with LV E/A ratio. However, there was a significant interaction between log hepatic triglyceride content and the binary variable BMI less than and greater than 27 kg/m² (P = .012), meaning that the association between log hepatic triglyceride content and LV E/A ratio is different below and above a BMI of 27 kg/m². Because of this arbitrary cutoff, we show the results stratified according to the World Health Organization categories of normal weight, overweight, and obesity.

Table 3 shows the association between hepatic triglyceride content and LV diastolic function in the entire study population and according to BMI. A significant crude inverse association was observed in the total study population (β: −0.170; 95% CI: −0.273, −0.068), normal-weight participants (β: −0.336; 95% CI: −0.509, −0.163), and obese participants (β:−0.209; 95% CI: −0.298, −0.121) (model 1). These associations diminished after adjusting for confounding factors in model 2. The addition of all components of the metabolic syndrome (model 3) or the metabolic syndrome as a single variable (model 3a) to the multivariate linear regression model did not alter the results. After additional adjustment for VAT and total body fat in the final model (model 4), the inverse association between hepatic triglyceride content and E/A ratio was significant only in the obese subgroup (β:−0.109; 95% CI: −0.186, −0.032), which represents a decrease in mean E/A ratio of 0.109 for a 10-fold increase in hepatic triglyceride content. In other words, in two otherwise-identical obese individuals with hepatic triglyceride contents of 1.5% and 15%, the difference in E/A ratio is estimated to be 0.109. This applies on any 10-fold difference in hepatic triglyceride content. Additional adjustment for the presence of the metabolic syndrome above the five individual components did not alter the results.

Discussion

After stratification in BMI categories according to the World Health Organization, hepatic triglyceride content was significantly associated with diastolic function independent of confounding factors including the metabolic syndrome, VAT, and total body fat in obese adults aged 45–65 years. This association was not statistically significant in normal-weight and overweight subgroups. Future studies with larger sample sizes should reveal to what extent associations between hepatic triglyceride content and diastolic function exist and differ in normal-weight, overweight, and obese persons.

A large Korean epidemiologic study recently reported that NAFLD was associated with subclinical diastolic dysfunction independent of the metabolic syndrome (17). Furthermore, LV diastolic function was impaired in normotensive nondiabetic patients with NAFLD compared with control subjects (18), and NAFLD was associated with the early diastolic velocity of the mitral annulus at tissue Doppler imaging in a multivariate stepwise regression analysis including metabolic and multiple echocardiographic variables (19). In addition, patients with hypertension and NAFLD had a higher prevalence of diastolic dysfunction (E/A ratio, <1) compared with hypertensive patients without NAFLD, and NAFLD was associated with diastolic function (20). NAFLD was also associated with early diastolic dysfunction in patients with well-controlled type 2 diabetes mellitus after adjustment for age, sex, triglyceride level, hemoglobin A1c level, and hypertension (21). Most studies assessed NAFLD by using criteria for abdominal ultrasonography. Quantification of hepatic steatosis could therefore often not be assessed. Furthermore, most studies were limited by small sample sizes, whereas the current study aimed to investigate the relationship between hepatic triglyceride content and diastolic function in a large sample of the general population. The association between hepatic triglyceride content and LV E/A ratio in the normal-weight subgroup could not be detected with statistical significance, probably because of the small number of individuals in the normal-weight subgroup (n = 91). Therefore, future studies with larger sample sizes should reveal to what extent associations between hepatic triglyceride content and diastolic function exist and differ in normal-weight, overweight, and obese persons.

The complex interrelationships among NAFLD, the metabolic syndrome, visceral obesity, and cardiovascular complications make it difficult to distinguish the causal links underlying the increased risk of cardiovascular disease among patients with NAFLD and/or the metabolic syndrome (3). For example, subjects who meet the diagnostic criteria for the metabolic syndrome have multiple risk factors for cardiovascular disease (22). The findings from our study suggest that hepatic triglyceride content contributes to subclinical impairment of LV diastolic function independently of the components of the metabolic syndrome, at least in obese persons. Further study is required to unravel the clinical relevance of the observed small association.

Causal pathways between fatty liver and diastolic function are speculative, but inflammatory cytokines, lipids, and advanced glycation end-products may play an important role (23). An important link between hepatic steatosis and cardiac disease has been described by Rijzewijk et al (6), who reported an association between hepatic steatosis and decreased myocardial perfusion, glucose uptake, and high-energy phosphate metabolism, but not with LV function, in type 2 diabetes. In addition, interstitial myocardial fibrosis has been associated with myocardial systolic and diastolic function in patients with diabetes (24). These findings may indicate an early alteration in myocardial tissue and/or vascular properties. Perseghin et al (25) reported that fatty liver was associated with changes in LV energy metabolism in healthy obese individuals. An animal study provided evidence that decreased adenosine triphosphate synthesis may be responsible for LV dysfunction in obesity (26). It may be hypothesized that NAFLD causes myocardial tissue alterations that lead to diastolic dysfunction. Diastolic dysfunction reflects increased LV diastolic stiffness, which is recognized as the earliest manifestation of LV dysfunction in diabetes mellitus, and is caused by fibrosis, deposition of advanced glycation end-products, and increased cardiomyocyte resting tension (27). Structural myocardial changes have been found in this study an are reflected by decreased LV end-diastolic index and increased LV mass in obesity.

Our study adds to the present knowledge that fatty liver was inversely associated with LV diastolic function independent of the metabolic syndrome and abdominal visceral adiposity, which may suggest subclinical impaired LV relaxation. The importance of subclinical effects is the potential reversibility of the pathophysiologic process, and the possibility to detect and follow up before overt cardiovascular failure is apparent. Increased intrahepatic cytokine expression is likely to play a key role in the progression of NAFLD (3,23,28) and cardiovascular disease (29,30).

Strengths of this study are the large study population and the availability of ¹H MR spectroscopy to quantify hepatic steatosis in combination with cardiac MR imaging. To the best of our knowledge, the current study is the first to report associations between hepatic triglyceride content and diastolic dysfunction in a general Western population. Further strengths are the availability of information about the components of the metabolic syndrome and other potential confounding variables, including total body fat and VAT.

A few limitations of this study should be addressed. No imaging modality is currently able to depict subtle histologic changes of inflammation and thus help differentiate simple steatosis from nonalcoholic steatohepatitis. Therefore, liver biopsy is the standard of reference for differentiating these two stages of NAFLD (31). For ethical reasons, we could not perform liver biopsies in this study and therefore could not classify nonalcoholic steatohepatitis. To assess NAFLD, we measured hepatic triglyceride content by using localized ¹H MR spectroscopy. Because of limited MR protocol time per participant in this large population-based cohort study, we did not correct for individual T2 relaxation times and we could not perform repeat imaging when technical failures were recognized. Proton-density fat fraction measurement with MR imaging recently showed good diagnostic accuracy for quantifying steatosis (32,33). Unfortunately, this technique was not yet validated at the time our study started in 2008. Optimally, such an advanced MR imaging technique would be better than MR spectroscopy, not least because MR spectroscopy is technically demanding. Participants in whom MR spectroscopy was unsuccessful had slightly higher amounts of subcutaneous adipose tissue, which potentially could have introduced selection bias. Finally, the observational, cross-sectional nature of our study precludes a causal interpretation of our results.

In conclusion, we showed that hepatic triglyceride content was associated with decreased diastolic function; however, adjustments for confounding factors attenuated this association. Only in persons with obesity could an association independent of the metabolic syndrome and abdominal visceral adiposity be demonstrated significantly. Therefore, confounding factors seem to largely explain the relationship between hepatic triglyceride content and diastolic function, but fatty liver itself could, at least in obesity, pose a risk of myocardial dysfunction above and beyond known cardiovascular risk factors that are clustered within the metabolic syndrome. Prospective follow-up research is required to study the effect of hepatic steatosis on incident cardiovascular events.

Disclosures of Conflicts of Interest: R.L.W. disclosed no relevant relationships. R.d.M. disclosed no relevant relationships. M.d.H. disclosed no relevant relationships. S.l.C. disclosed no relevant relationships. F.R.R. disclosed no relevant relationships. J.W.J. disclosed no relevant relationships. J.W.A.S. disclosed no relevant relationships. A.d.R. disclosed no relevant relationships. H.J.L. disclosed no relevant relationships.

Author Contributions

Author contributions: Guarantor of integrity of entire study, H.J.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, R.L.W., A.d.R., H.J.L.; clinical studies, R.L.W., R.d.M., F.R.R., J.W.J., J.W.A.S., A.d.R., H.J.L.; statistical analysis, R.L.W., R.d.M., M.d.H., S.l.C., F.R.R., H.J.L.; and manuscript editing, R.L.W., R.d.M., F.R.R., J.W.J., J.W.A.S., A.d.R., H.J.L.