The Effects of Age and Aerobic Fitness on Myocardial Lipid Content

Abstract

Background

Aging and sedentary lifestyles lead to cardiac atrophy, stiffening and impaired diastolic function. Both conditions are marked by increased adiposity which can lead to ectopic fat deposition in non-adipocyte tissues including the myocardium. The effect of excess intra-myocardial fat on cardiac function in non-obese individuals is unknown.

Methods and Results

Cardiac lipid content was measured by magnetic resonance spectroscopy in 153 healthy non-obese subjects with varying fitness levels quantified by peak oxygen uptake (VO₂peak) during treadmill exercise. Cardiac function (echo) and left ventricular (LV) filling pressures (right heart catheterization) were measured under varying preloads. LV stiffness was calculated from acurve fit of the diastolic portion of the pressure-volume (PV) curve. The strongest clinical predictors of lipid content were body mass index (BMI) (β, 95% CI: +0.03, 0.001 to 0.06) and VO₂peak (−0.02, −0.03 to −0.009) (R²=0.14; p<0.001). Subjects in the highest quintile had smaller LV end-diastolic volumes (68±13 vs. 58±12 ml/m², p<0.01), decreased peak early mitral annular and increased peak late mitral inflow velocities. There were no differences in LV stiffness but a leftward shift in the PV curve suggested a less distensible ventricle with increasing myocardial lipid levels. After adjusting for age, fitness and BMI, echocardiographic and morphometric differences amongst groups were attenuated and no longer significant.

Conclusions

BMI and fitness levels are the strongest predictors of myocardial lipid content in non-obese humans. Cardiac lipid content is associated with decreased ventricular distensibility and may provide a causal mechanism linking changes in LV function related to age and fitness.

Aging and inactivity are associated with significant changes in cardiovascular physiology. Previous work by our group has shown that the aged heart becomes smaller, stiffer and less distensible. (1, 2) Extreme inactivity also recapitulates a similar cardiac phenotype as observed during spaceflight or prolonged bed rest. (3, 4) In contrast, habitual exercise training results in above average measures of peak VO₂ adjusted for age and sex, larger stroke volume, increased cardiac mass and improved measures of ventricular compliance. (5, 6)

A common relationship shared between aging and sedentariness is changes in body composition. As individuals age or become sedentary, percentage body fat increases with a concomitant decrease in lean body mass. (7) Excess adiposity and metabolic derangements may directly affect the cardiovascular system via accumulation of intra-myocardial triglycerides leading to lipotoxicity. Typically associated with obesity related diseases, the accumulation of ectopic fats in non-adipocyte tissues can lead to organ dysfunction via buildup of toxic fatty acid metabolic intermediaries. (8)

While the abnormalities associated with lipid accumulation in hepatic, pancreatic and connective tissues have been well documented, the effects of cardiac steatosis on cardiac structure and function are still in the early stages of clinical investigation, supported by a growing body of animal models of myocardial fatty acid accumulation showing significant declines in cardiac function. (9–13) Small clinical studies in humans have suggested an association between lipid content and abnormal diastolic function, myocardial strain and increased LV mass in diabetic and obese patients. (14–18)

In older and unfit individuals, cardiac steatosis may be one causal link contributing to the development of diastolic abnormalities and forms the basis for our current study investigating the effects of a sedentary lifestyle and aging on cardiac structure and function. We hypothesized the accumulation of myocardial triglycerides increases with age and sedentariness and would be associated with ventricular stiffening and impairments in diastolic function in non-obese and non-diabetic individuals.

Methods

Study recruitment

Study subjects were recruited from the Dallas Heart Study, a population based cohort of over 6,000 individuals, enriched by a random sampling of employees of Texas Health Resources, a large healthcare provider in the Dallas-Fort Worth metroplex, as previously described. (19) Subjects were also recruited prospectively from the Aerobic Center Longitudinal Database at the Cooper Clinic in Dallas; a cohort of individuals with extensively recorded exercise histories over 25 years. All study procedures were approved by the Institutional Review Boards of the University of Texas Southwestern Medical Center at Dallas and Texas Health Resources. After providing informed consent, all subjects underwent testing as outlined below. Volunteers were excluded if they had elevated 24 hour ambulatory blood pressure > 140/90 mmHg. In addition, volunteers were excluded if they had a history of cardiovascular disease (e.g. stroke, myocardial infarction, atrial fibrillation, etc.), diabetes, COPD, former or current smokers and body mass index (BMI) > 30 kg/m2. All subjects underwent a screening maximal exercise stress echo prior to enrollment. Subjects were excluded if they had evidence of coronary ischemia by ECG and echocardiography. Of 353 subjects screened, 191 were excluded for indications listed above. The remaining 162 subjects were predominantly Caucasian (>90%).

Body Composition and Physical Activity Level

Body composition was determined by body density using underwater weights as described previously. (5) Subjects were also given activity monitors (Acitical, Phillips, USA; RT3; Stayhealthy, USA) which reported total daily activity as METS over 7 days. Total time spent sedentary and active (by METS) were reported in minutes per day. A threshold of 4 METS was used to quantify minutes per day spent in fitness related activities.

Exercise Testing

A modified Astrand-Saltin incremental treadmill protocol was used to determine peak exercise capacity. Measures of ventilatory gas exchange were made by use of the Douglas bag technique. (5) Gas fractions were analyzed by mass spectrometry (Marquette MGA 1100), and ventilatory volume was measured by a Tissot spirometer. Maximum oxygen uptake was defined as the highest oxygen uptake measured from at least a 40s Douglas bag. To normalize for differences in sex and age, predicted peakVO₂ values were calculated adjusting for age and sex as previously reported. (20)

Cardiac MRI

All subjects underwent cardiac MRI to assess cardiac morphometric parameters in addition to myocardial triglyceride content. Subjects did not undergo MRI if they had cochlear implants or were claustrophobic. Images were obtained using a 1.5-tesla Gyroscan INTERA whole-body system (Philips Medical Systems, Best, The Netherlands) equipped with spectroscopy and cardiac packages. LVEDV and mass were measured as previously reported using a steady-state free-precision imaging sequence. (21) Tissue triglyceride content was measured using 1H nuclear magnetic resonance (NMR) spectroscopy and quantified as a percentage of water content in a region of interest located in the mid-ventricular septum. During acquisition of spectroscopic data, patients breathed freely. The spectroscopic signal was acquired with cardiac triggering at end systole and respiratory gating at end expiration. A PRESS sequence (Point-RESolved Spectroscopy) was used for spatial localization, and the interpulse delay was defined by the length of a respiratory cycle. NUTS software (Acorn NMR, Fremont, Calif) was used to process data. The areas under the signals from water and methylenes of fatty acids in triglycerides were quantified by a line-fitting procedure, and the values were corrected for spin-spin relaxation. (17, 22)

Intra-cardiac pressure/volume measurements

Right heart catheterizations were performed via an antecubital vein approach as described previously. (5) A 6Fr Swan-Ganz catheter (Edwards Lifesciences) was placed under sterile conditions into the pulmonary artery using fluoroscopic guidance. Subjects were then placed in a lower body negative pressure (LBNP) chamber to decrease LV preload. Recordings of right atrial pressure (RAP), PCWP, and left ventricular end-diastolic volumes (LVEDV) using 3-dimensional echocardiography were made at baseline after 20 minutes of supine rest and again after 5 minutes each of LBNP of −15 mmHg followed by −30 mmHg. In addition, cardiac echocardiography as described below was also performed during each condition. Study participants were then subjected to two “loading” conditions to increase PCWP with rapid normal saline infusion at 15 ml/kg and 30 ml/kg at flow rates of approximately 200 ml/min. Pressure and echocardiographic measurements were made after each infusion. Cardiac output and SV were calculated using a modified acetylene gas re-breathing technique. (23) The diastolic portion of the pressure-volume (PV) curve was constructed using PCWP and corresponding LVEDVs under the experimental unloading and loading conditions. The ventricular stiffness constant was calculated by an exponential curve fit of the end-diastolic pressure-volume relationship as previously described. (24)

Echocardiography

Echocardiographic images were acquired on an iE33 machine (Philips) and measured offline using Xcelera cardiovascular image management system (Philips). Images were obtained under baseline, LBNP and saline loading conditions. Tissue Doppler images were obtained in the apical four-chamber view by placing a 2-mm sample volume on the septal and lateral mitral annulus. Mitral inflow velocities were also measured in the four-chamber view by placing a 2-mm pulsed wave sample volume at the tips of the mitral leaflet during valve opening. Propagation velocity (Vp) was obtained by color M-mode imaging with a sampling area extending from the mid portion of the left atrium to the apex and measuring the slope of the early diastolic aliasing velocity. IVRT was measured as the time between aortic valve closure and mitral valve opening by placing a 4-mm sample within the LV outflow tract from a five-chamber view. LV volumes under the different loading conditions were determined by offline 3-dimensional reconstruction. (QLab, version 9.0; Phillips)

Statistical Analysis

Statistical analysis was performed using commercially available software (SPSS, Sigmaplot v12.0). All reported variables are presented as means with standard deviations. After dividing subjects into quintiles by cardiac lipid content, the distribution of a majority of variables was skewed. A nonparametric Jonckheere-Terpstra test was therefore used to determine the effect of increasing ranked quintiles of lipid content on measured variables. Pearson correlation of clinical variables with cardiac lipid was performed and multiple linear regression modeling was then used to determine the clinical variables that best predicted myocardial lipid content. Regression analysis was also performed on echocardiographic and MRI variables. A two way repeated measures analysis of variance (ANOVA) was used to identify differences in pressure and volume parameters under varying loading conditions between the lowest and highest quintiles of cardiac lipid content. To adjust the cardiac lipid quintile comparisons for age and fitness related differences in cardiac function; hemodynamic, echocardiographic and morphometric parameters were modeled using age, sex and peakVO2 as covariates in analysis of covariance models (ANCOVA). Data transformations were employed to meet parametric analysis assumptions as necessary. A p value less than 0.05 was considered statistically significant.

Results

Baseline characteristics

Cardiac lipid content was available in 153 out of 162 enrolled subjects who had no contraindications to MRI. In the nine volunteers who did not have lipid measures, five had technically limited scans due to motion artifact or excessive heart rate variability, two had contaminated signal from epicardial fat and two had levels below detection limits. The mean age of all subjects was 62 years and ranged from 26 to 86 years with an inter-quartile range of 61 and 69 years. Subjects were stratified by lipid content into quintiles. Lipid content ranged from 0.31% ± 0.10 in the lowest quintile to 1.61% ± 0.62 in the highest quintile. (Table 1, Figure 1) Subjects in the highest quintiles had larger BMI compared to the lowest quintile in addition to higher body fat composition. There were no differences in HgbA1c% or age between groups. As comparison, myocardial 1H-NMR spectroscopy from a previously published study from our group showed a range for triglycerides of 0.46% ± 0.30 in lean (BMI<25 kg/m²) individuals to 1.06% ± 0.62 in type 2 diabetic subjects. (22)

Table 1.

Baseline clinical characteristics by quintile of cardiac lipid content

Quintile (n)	Q1 (30)	Q2 (30)	Q3 (30)	Q4 (32)	Q5 (31)
Lipid (%) ‡	0.31 ± 0.10	0.56 ± 0.07	0.76 ± 0.06	1.04 ± 0.10	1.61 ± 0.62
Age (yrs)	60.8 ± 14.3	60.9 ± 12.4	59.4 ± 12.0	62.6 ± 10.5	66.6 ± 10.7
% Women	43	37	60	53	39
Peak VO2 (ml/kg/min) ‡	31.7 ± 7.8	29.8 ± 7.1	27.8 ± 7.2	27.5 ± 7.7	25.0 ± 7.4
Predicted VO2 (%)	130 ± 47	117 ± 31	110 ± 39	111 ± 30	111 ± 35
BMI (kg/m2) ‡	23.1 ± 2.8	24.7 ± 2.8	24.6 ± 3.1	24.7 ± 3.3	26.0 ± 2.1
Body Fat (%) †	28.6 ± 8.1	28.1 ± 7.0	29.7 ± 9.0	31.6 ± 8.1	33.5 ± 8.1
HgbA1c (%)	5.3 ± 0.3	5.3 ± 0.3	5.2 ± 0.4	5.5 ± 0.3	5.4 ± 0.3
Minutes per day > 4 METS †	29 ± 27	24 ± 19	16 ± 15	19 ± 26	14 ± 20
Minutes per day sedentary	1142 ± 115	1196 ± 148	1179 ± 97	1183 ± 73	1165 ± 84

^*p < 0.05;

^†p < 0.01;

^‡p< 0.001 (Jonckheere-Terpstra trend across increasing quintiles)

Figure 1. Localized Cardiac ¹H Magnetic Resonance Spectroscopy.

Representative ¹H spectra from a subject with high (5^th Q) and low (1^st Q) cardiac lipid % with peak signal intensity at 1.4 ppm. Myocardial triglyceride content was calculated as a percentage of area relative to myocardial water intensity at 4.8 ppm.

Clinical Predictors of Cardiac Steatosis

In general, fitness and daily activity levels decreased with increasing quintiles of cardiac lipid content. Subjects in the highest quintile had the lowest measured peakVO₂. (Figure 2) The distribution of peakVO₂ was large and likely driven by differences in age, body weight and fitness levels across all subjects. To adjust for the wide distribution and to distinguish if the relationship between cardiac lipid content and fitness was due to age and sex related differences, peakVO₂ was calculated as a percentage of age and sex predicted VO₂ scaled to ideal body weight. There was a trend towards higher percent predicted peakVO₂ for subjects with the lowest cardiac lipid content. (Jonckheere-Terpstra test p=0.06) There was also an association between increasing lipid content and daily activity level. Subjects in the lower quintiles were more likely to spend additional minutes per day engaged in activities greater than 4 METS. (Table 1) There was no difference in amount of time spent sedentary between groups.

Figure 2. Peak VO₂ by Cardiac Lipid Quintile.

Distribution of peak VO₂ by lipid quintiles with 1^st quintile containing subjects with lowest levels of cardiac lipid. Solid boxes represent interquartile (25–75%) range and solid line represents median value within quintile. Error bars encompass 5–95% range and outliers beyond this range are signified by solid black dots.

Simple linear regression of clinical characteristics and cardiac lipid content across all 153 subjects showed an association between BMI, body fat, peakVO₂, % of predicted peakVO₂ and minutes per day spent in activities greater than 4 METS. (Supplemental Table 1) In our multiple regression models, peakVO₂ and BMI were significant variables in predicting lipid content (beta coefficient, 95% CI: −0.02, −0.03 to −0.009 and +0.03, 0.001 to 0.06, respectively; R²=0.14, p<0.001). To better understand the association between peakVO₂ and cardiac lipid content and to account for confounding factors (age, sex, BMI) that influence peakVO₂which may not have been fully adjusted for, the multiple regression model was re-run using percent predicted peakVO₂ instead of measured peakVO₂. In this model, percent predicted VO₂ (β= −0.47, 95% CI −0.82 to −0.12) and age (β=0.013, 0.005 to 0.021) were the most significant variables (R²=0.15; p<0.001) that predicted cardiac lipid content suggesting that even “relative” fitness, when adjusted for sex and age expected values, remained determinative of intra-myocardial lipids.

Effects of Cardiac Lipid Content on Cardiac Structure and Function

MRI was used to assess baseline cardiac structure and function. With increasing quintiles of cardiac lipid, LV end diastolic and stroke volumes indexed to BSA decreased while LV ejection fraction slightly increased. (Table 2) There was no correlation to BSA indexed LV mass (Jonckheere-Terpstra p=0.07) or end systolic volumes.

Table 2.

MRI derived LV morphometric and echocardiographic parameters by lipid quintile

Quintile	Q1	Q2	Q3	Q4	Q5
LVEF (%) *	67 ± 6	69 ± 7	69 ± 7	70 ± 6	71 ± 6
LVEDV (ml)	123 ± 29	116 ± 32	112 ± 28	108 ± 26	109 ± 28
LVEDVi (ml/m2) †	68 ± 13	62 ± 15	62 ± 12	60 ± 11	58 ± 12
LVSV (ml)	82 ± 20	79 ± 21	76 ± 18	75 ± 19	76 ± 19
LVSVi (ml/m2) *	45 ± 8	42 ± 9	43 ± 8	41 ± 7	40 ± 9
LVMi (g/m2)	57 ± 12	56 ± 13	52 ± 9	53 ± 11	52 ± 11
TDI E’lat (cm/s) **	11.4 ± 2.7	10.9 ± 2.4	11.2 ± 2.3	10.5 ± 2.8	9.2 ± 2.4
TDI A’lat (cm/s)	8.7 ± 2.0	9.5 ± 2.5	8.7 ± 2.4	9.7 ± 2.7	9.3 ± 2.4
TDI S’lat (cm/s)	9.2 ± 1.7	9.3 ± 1.8	9.0 ± 1.4	9.6 ± 2.4	8.5 ± 1.5
E wave (cm/s)	67 ± 15	65 ± 14	74 ± 17	65 ± 13	67 ± 15
A wave (cm/s) ‡	59 ± 20	58 ± 19	63 ± 22	67 ± 19	69 ± 15
E/A ratio †	1.25 ± 0.46	1.23 ± 0.51	1.32 ± 0.57	1.01 ± 0.30	1.01 ± 0.30
IVRT (ms)	115 ± 27	124 ± 27	110 ± 20	110 ± 22	116 ± 21
Vp (cm/s)	46 ± 12	45 ± 11	45 ± 10	43 ± 8	44 ± 8

^*p < 0.05;

^†p < 0.01;

^‡p < 0.001 (Jonckheere-Terpstra trend across increasing quintiles)

LVEDVi - LVEDV indexed to BSA; LVSV - left ventricular stroke volume; LVSVi - LVSV BSA indexed; LVMi - left ventricular mass BSA indexed; TDI E’lat - tissue doppler E’ lateral wall velocity; TDI A’lat - A’ lateral wall velocity; TDI S’lat - S’ lateral wall velocity; E – peak early mitral inflow velocity; A – peak late mitral inflow velocity; IVRT - isovolumic relaxation time; Vp - mitral inflow propagation velocity.

Hemodynamic and Echocardiographic Parameters

There was no correlation between cardiac lipid content and ventricular stiffness derived from the end-diastolic portion of the pressure-volume curve. Stiffness constants derived from PV curves drawn using group averaged values for indexed EDV and PCWP were similar between subjects with low and high cardiac lipids. (Figure 3) Interestingly, the PV curve for the highest quintile (Q5) was shifted leftwards suggesting decreased LV distensibility similar to changes seen with healthy aging or decreased fitness. (1, 5) While indexed EDVs were significantly lower under each of the loading conditions with increasing cardiac lipids (p<0.01 for lipid quintile effect; 2 way repeated measures ANOVA), there were no differences in PCWP across quintiles. (Supplemental Table 2) There was a slight increase in right atrial pressure across all preload conditions with increasing lipid content (Jonckheere-Terpstra test p<0.05) suggestive of a differential effect of myocardial steatosis on right and left ventricular hemodynamic function.

Figure 3. End-diastolic pressure volume relationship in lowest (Q1) and highest (Q5) quintiles of cardiac lipid content.

End diastolic volumes (EDV) were obtained from 3-dimensional echocardiography and indexed to BSA. Stiffness constants (s) were derived from an exponential curve fit. While LV stiffness for subjects in the highest and lowest quintiles were similar, indexed EDV for the highest quintile was smaller under varying preloads (main effect by lipid quintile p<0.01; 2 way repeated measures ANOVA) suggesting a less distensible LV with increasing myocardial lipid content. There were no differences in PCWP between quintiles. PCWP – pulmonary capillary wedge pressure. Values are mean ± SE.

Under baseline conditions, subjects with the highest level of cardiac lipid content had lower peak early mitral annular velocity (TDI E’ lateral) and higher late diastolic mitral inflow (A wave) in addition to lower E/A ratios compared to those with lower levels. (Table 2) There were no differences in measures of ventricular relaxation as assessed by IVRT or mitral inflow propagation velocity. The association between decreasing peak early mitral annular velocity and cardiac lipid content remained consistent with increasing ventricular loading conditions. Similarly, the association between increasing late diastolic mitral inflow and cardiac lipid content also persisted across loading conditions. (Table 3)

Table 3.

Correlation and beta coefficients for cardiac lipid content on select echocardiographic parameters across varying left ventricular pre-load conditions

Echo Variable	Loading Condition	r	Beta (95% CI)	P value
TDI E’ lateral	−30 mmHg LBNP	−0.08	−0.019 (−0.056, 0.018)	0.32
	Baseline	−0.19	−0.039 (−0.072, −0.006)	0.02
	+30 ml/kg NS bolus	−0.24	−0.052 (−0.087, −0.017)	0.004
TDI A’ lateral	−30 mmHg LBNP	−0.003	−0.0006 (−0.037, 0.036)	0.976
	Baseline	−0.007	−0.002 (−0.038, 0.035)	0.931
	+30 ml/kg NS bolus	0.10	0.02 (−0.012, 0.052)	0.212
E wave	−30 mmHg LBNP	0.01	0.001 (−0.007, 0.009)	0.863
	Baseline	0.02	0.001 (−0.005, 0.006)	0.808
	+30 ml/kg NS bolus	0.18	0.005 (0.001, 0.009)	0.026
A wave	−30 mmHg LBNP	0.17	0.006 (0.0001, 0.012)	0.045
	Baseline	0.16	0.004 (0.000, 0.009)	0.05
	+30 ml/kg NS bolus	0.32	0.008 (0.004, 0.012)	<0.001

CI – confidence interval, LBNP - lower body negative pressure; +30 NS - 30 ml/kg normal saline rapid infusion.

To determine the independent effects of cardiac lipid content on echocardiographic and morphometric parameters, ANCOVA analysis using cardiac lipid quintiles and defined covariates of age, BMI and peakVO₂ was performed. Once adjusted for these covariates, many of the observed relationships were attenuated and no longer statistically significant. Nearly all the interactions were highly non-significant (much greater then p=0.25) and did not contribute meaningful information to the model. Therefore our final model does not include interaction terms. There were no differences by ANCOVA amongst quintiles with regards to mitral annular tissue Doppler, mitral inflow velocities or filling pressures. Indexed EDV was also no longer significantly different across lipid quintiles suggesting the differences in distensibility seen in Figure 3 were likely reflective of differences in age and peak fitness levels amongst the groups.

Discussion

The primary findings of this current study are 1) myocardial steatosis can vary greatly across “metabolically healthy” individuals and can approach or exceed levels previously reported in diabetic and metabolic syndrome patients; 2) clinical predictors of cardiac steatosis included BMI and peakVO₂; however these associations ceded to age after adjusting peakVO₂ for age and sex; 3) myocardial steatosis was associated with decreased ventricular distensibility and markers of early diastolic dysfunction.

Adiposity and the heart

Myocardial steatosis is a compelling means of bridging the connection between metabolic disorders and associated abnormalities in diastolic function; which in turn could form the basis for the increased incidence of heart failure observed in patients with obesity related disorders. While frequently associated with co-morbid conditions of coronary artery disease, hypertension and renal dysfunction, patients with diabetes and obesity are twice as likely to develop heart failure compared to lean controls even after adjusting for baseline differences in co-variates. (25, 26) Advances in nuclear magnetic resonance spectroscopy have opened insight into changes in myocardial composition in patients with metabolic disorders independent of ischemia and ventricular hypertrophy. This concept of metabolic remodeling has focused primarily on the presence of intra-cardiac triglyceride accumulation as a surrogate marker of a lipotoxic heart. Whether myocardial triglycerides are direct mediators of abnormal cardiac function or represent a biomarker of altered cardiac metabolism remains a key question. The presence of intra-myocardial fat has previously been correlated with decreased ventricular strain rates, increased LV mass, and decreased early peak mitral inflow velocity in small studies of diabetic and obese individuals suggesting that cardiac lipid content has adverse effects on myocardial structure leading to diastolic dysfunction and raising the possibility of a therapeutic target to reverse diastolic abnormalities.

However the confounding impact of diabetes and obesity on measures of cardiac function can be difficult to distinguish from the independent effects of intra-cardiac triglycerides. Diabetes is known to cause alterations in myocardial calcium signaling and endothelial dysfunction, factors that are difficult to assess clinically. (27) Measuring the effects of cardiac steatosis on myocardial structure and function in “metabolically healthy” individuals reduces potential confounders and may help to elucidate a clearer causal correlation. The findings of our study provide insight into the changes in ventricular function with increasing levels of cardiac lipid content and suggest a plausible causal mechanism linking the effects of aging, aerobic fitness and total body adiposity on cardiac function.

Fitness, aging and adiposity

To date, no studies have analyzed the impact of physical activity on cardiac steatosis. Previous studies have shown metabolic syndrome, visceral adiposity and impaired glucose tolerance to be important clinical predictors of myocardial steatosis. (22, 28) The accumulation of triglycerides in metabolically active tissues, particularly muscle, further decreases the response to insulin mediated glucose uptake. In contrast, metabolic changes associated with increasing fitness are essentially the antithesis of changes that occur in metabolic syndrome and include improved insulin sensitivity and decreased total body adiposity. (29) Individuals with better fitness would be expected to have improved markers of metabolic health but because of the co-linearity of many variables associated with peakVO₂, it is difficult to distinguish the independent effects of one clinical factor over others. Individuals with high aerobic fitness are more likely to be younger, have low BMI and body fat percentage. In a regression model, the correlation between aerobic fitness (peakVO₂) and myocardial triglycerides was more strongly associated than BMI or body fat % suggesting that peakVO₂ was likely the primary driving factor.

Age also was an important factor in the development of cardiac steatosis. Even after scaling for age and gender predicted VO₂, age remained significantly associated with cardiac lipid content. Both low aerobic fitness and advanced age can be associated with impaired insulin sensitivity and although not directly measured in our study, these associations may mirror similar metabolic changes and development of cardiac steatosis seen in obese or diabetic subjects. As individuals age or become more sedentary, the heart is likely not spared from the metabolic milieu of increased body adiposity and BMI. The accumulation of myocardial fat with aging, decreasing aerobic fitness and increasing BMI thus may represent a metabolic change in organ function, eventually leading to development of increased ventricular diastolic stiffening or prolonged myocardial relaxation times.

Adiposity and cardiac function

Myocardial lipid accumulation was associated with premature markers of abnormal diastolic function. Subjects with the highest quintiles of cardiac lipid content were more likely to have abnormalities in early diastolic myocardial relaxation, lower E/A mitral inflow ratios, smaller indexed EDV and decreased LV distensibility. These observations were consistent with expected changes in cardiac function based on clinical predictors of intra-myocardial triglyceride accumulation (poor aerobic fitness, advanced age). After adjusting for co-variates of age, peakVO₂ and BMI, the association between myocardial fat and diastolic function was attenuated. While aging and low cardiovascular fitness share similar diastolic phenotypes, the causal connection leading to diastolic abnormalities shared between these two discrete conditions is unknown. The accumulation of cardiac lipid could provide a plausible mechanism for the increased observation of diastolic dysfunction in older or unfit individuals and may have deleterious consequences beyond acting just as a biomarker for impaired total body metabolic homeostasis. Pre-clinical work has shown lipotoxicity leads to increased production of reactive oxygen species and apoptosis. (30) Prolonged or accelerated accumulation of toxic metabolites could be a key mediator in the decline in diastolic function with age and poor fitness. Although our study did not show an increase in LV stiffness using invasive measures with lipid accumulation, subtle changes in diastolic function may be the precursors leading to the development of more significant diastolic abnormalities and symptoms of heart failure.

Several questions remain regarding the consequences of excess cardiac fat content in individuals with concomitant metabolic and cardiac co-morbidities. As the prevalence of diabetes and obesity continues to rise, there is a greater need to better characterize the consequences of excess adiposity on cardiovascular physiology. Patients with diabetes or ventricular hypertrophy may accumulate myocardial lipids at an earlier age and develop the metabolic phenotype of an older and sedentary heart. Whether these metabolic changes are pathologic remains unanswered and other studies have even raised the possibility of a beneficial effect of obesity in heart failure, a concept termed the “obesity paradox.” (31) Future studies may help to answer these questions by use of agents to lower myocardial triglyceride content in conjunction with a comprehensive assessment of myocardial function. In addition, long term prospective studies of cardiac fat content and the development of diastolic abnormalities and heart failure may also be helpful in establishing a causal connection.

Limitations

Our study is limited by incomplete characterization of the metabolic profile of participants. Both visceral fat and insulin sensitivity have previously been correlated to myocardial triglyceride content but were not measured in our current study. (22, 28) Fasting glucose was also unavailable. The majority of our subjects were elderly with only 25% of participants below 60 years of age. This may have underestimated the association between age and cardiac lipid content given that younger subjects tended to have lower cardiac lipid content.

Overall, we believe our study provides further insight into the cardiac metabolic adaptations that develop with aging and sedentariness. Our findings suggest lipid content in a healthy population is more widely distributed than previously reported and the accumulation of myocardial lipids is associated with low aerobic fitness, increased body mass index and advanced age. Decreased ventricular distensibility and early markers of diastolic abnormalities were seen with increasing levels of myocardial triglyceride suggesting myocardial fatty acid metabolism may be a viable therapeutic target in improving cardiac function in diabetic and obese patients.