Left ventricular (LV) systolic function does not deteriorate with normal ageing; in contrast, markers of diastolic function were found to be progressively impaired with ageing.1 The early mitral peak flow (E) is reduced in older individuals; meanwhile the late, atrial trans-mitral peak flow (A) increases with age, reflecting compensation of atrial function to maintain LV filling.1 It was proposed that LV diastolic dysfunction in normal ageing was the effect of a decrease in cardiac mass caused by a continuous loss of functioning muscle.2

Recently, it was hypothesised that a primary alteration of cardiac energy metabolism might cause the functional alteration, suggesting that metabolic therapy may be a promising new tool for the treatment of heart failure.3 4 Altered cardiac high energy phosphates (HEP) metabolism was reported in humans with congenital cardiac disease and in more common pathological conditions,5 and we have also recently reported that it may be detected in insulin-resistant overweight/obese individuals,6 7 in spite of normal LV geometry and function.

It may be hypothesised that physical exercise, by means of direct or indirect metabolic or haemodynamic effects, may improve the LV functional alterations associated with ageing. In this respect data are contradictory: some studies suggested that in humans endurance exercise training augments early diastolic filling at rest both in healthy young and older sedentary individuals,8 9 but other data supported the hypothesis that the impairment of diastolic function is intrinsic to normative ageing and may not be reversed.10 The aim of our study was to explore this issue in middle-aged men. In particular, using cardiac magnetic resonance imaging (MRI) and ³¹P magnetic resonance spectroscopy (MRS) in a cross-sectional fashion, we wanted to establish (1) whether the LV diastolic dysfunction occurring in sedentary middle-aged subjects was also paralleled by alterations of cardiac HEP metabolism, and (2) whether the long-lasting habit of perform endurance exercise is associated with the attenuation of age-related functional and metabolic features.

METHODS

Subjects

Twenty healthy, young and 25 healthy, middle-aged, non-obese men participated to the study. They were segregated in those with a sedentary lifestyle (11 young and 14 middle-aged) and those undergoing regular exercise activity (9 young and 11 middle-aged). The physically active men were recruited in different environments: the nine young subjects were track runners participating in national competitions and were recruited within the Federazione Italiana Atletica Leggera (FIDAL)-Lombardia Section; the 11 middle-aged subjects were amateur athletes regularly engaged in non-professional regional competitions throughout the year. All subjects were engaged in a sustained aerobic endurance training. The sedentary subjects were recruited in order to be comparable in terms of age and body mass index (BMI) within the outpatient services of the Center of Nutrition/Metabolism of the San Raffaele Scientific Institute. All study subjects were healthy, with no previous history of diabetes, hypertension, dyslipidaemias, vascular events or of dilated cardiomyopathy, of a pathological ejection fraction, of resting ECG markers of cardiac ischaemia and no features compatible with the NYHA classes for heart failure. They were not taking any medications. Their anthropometric and laboratory characteristics are summarised in table 1. Recruited subjects gave their informed written consent after explanation of purpose, nature and potential risks of the study. The protocol was approved by the ethics committee of the Istituto Scientifico H San Raffaele.

Experimental procedures

Subjects were instructed to consume an iso-caloric diet and to abstain from heavy exercise activity for 3 days before the MRI-MRS studies. All volunteers underwent the protocol between 07:30 and 9:30 on Wednesdays in the resting state after a 10-hour overnight fasting period.

Cardiac ³¹P-MRS

Cardiac ³¹P-MRS was performed at rest using a 1.5T whole-body scanner (Gyroscan Intera Master 1.5 MR System; Philips Medical Systems, Best, The Netherlands). ³¹P spectra were obtained by means of a 10-cm-diameter surface coil used for transmission and detection of radio frequency signals at the resonance frequency of ³¹P (at 1.5-T, 25.85 MHz) as previously described.6 7 11 Briefly, after appropriate positioning of the surface coil on the chest, localised homogeneity adjustment was performed using the body coil and ECG-triggering by optimising the 1H-MRS water signal. Afterwards the transmitter-receiver was switched without time delay to the ³¹P frequency and manual tuning and matching of the ³¹P surface coil was performed to adjust for different coil loading. The radio frequency level was adjusted to obtain a 180° pulse of 40 ms for the reference sample at the centre of the ³¹P-surface coil. The acquisition of ³¹P-MR spectra was triggered to the R-wave of the ECG, with a trigger delay time of 200 ms and a recycle time of 3.6 seconds. Image selected in-vivo spectroscopy (ISIS) volume selection in three dimensions (3D-ISIS) based on 192 averaged free induction decays was employed. The volume of interest was oriented avoiding inclusion of chest wall muscle and diaphragm muscle (typical volume size: 5 (caudo-cranial) × 6 × 6 cm³). Acquisition time was 10 minutes. Adiabatic frequency-modulated hyperbolic secant pulses and adiabatic half-passage detection pulses were used to achieve inversion and excitation over the entire volume of interest. Examination time was 40–45 minutes.

Cardiac MRI

MRI studies were performed with the above described scanner using an enhanced gradient system with a maximum gradient strength of 30 mT/m and a maximum gradient slew rate of 150 mT.m⁻¹.s⁻¹. The Cardiac Research software patch (operating system 9) was used. The examination was performed using a five-element cardiac phased-array coil (SENSE-cardiac) and retrospective ECG-triggering obtained with Vectorcardiogram system (VCG) and using standard MRI methodology as previously described.6 7 11

Calculations

³¹P-MRS analysis

³¹P-MR spectra were transferred to a remote SUN-SPARC workstation for analysis. The spectra were quantified automatically in the time domain, using Fitmasters. The ATP level was corrected for the ATP contribution from blood in the cardiac chambers based on a previous study.10 Depending on the repetition time (TR), phosphocreatine/ATP (PCr/ATP) ratios had to be corrected for partial saturation effects and T1-values obtained from inversion recovery experiments on the human LV were used. Based on these data and a repetition time of 3.6 seconds a saturation correction factor of 1.35 was obtained and applied to all “blood corrected” myocardial PCr/ATP ratios.12 An estimate of the signal-to-noise ratio of each spectrum was obtained from the relative Cramer-Rao standard deviation (rCRSD) calculated for the PCr/ATP, which is a commonly reported index of accuracy of the spectral quantification.13

MRI analysis

Image analysis was performed using an image-processing workstation (EasyVision; Philips Medical Systems) by using the cardiac analysis software package as previously described.6 7 11

Analytical determinations

We measured glucose concentration with the glucose oxidase method (Beckman Coulter, Fullerton, CA, USA). Triglycerides, total cholesterol and HDL cholesterol were measured as previously described.6 7 11 Plasma insulin (intra-assay and inter-assay coefficient of variation <3% and 6%, respectively; cross-reactivity with C-peptide and proinsulin <1%) was measured with radioimmunoassay (Linco Research, St Charles, MO, USA).6 7 11 We estimated insulin sensitivity (homeostatic model assessment, HOMA2-%S) and secretion (HOMA2%-B) by the updated computer model homeostasis model assessment available from www.ocdem.ox.ac.uk.14

Statistical analysis

Data in text, tables and figures are mean (SD). Analysis was performed using the SPSS software (version 10.0). Parameters did not show a skewed distribution (Kolmogorov-Smirnov test of normality). One-way ANOVA with Tukey post-hoc analysis was used for comparison among groups. Independent samples t test was performed when comparing young and middle-aged individuals regardless of exercise habits. Two-tailed Pearson’s correlation was performed to establish partial correlation coefficients between variables. Statistical significance was defined as a p value <0.05.

RESULTS

Anthropometric and biochemical characteristics of study groups

The anthropometric features of the study groups are summarised in table 1. Age was different depending on the recruitment criteria. BMI and body surface area were not different between groups. Arterial blood pressure was within the normal range and was not different among groups. Total cholesterol was higher in the middle-aged subjects and high-density lipoprotein (HDL)-cholesterol was lower in the subgroups of sedentary individuals with respect to the physically active. Fasting plasma glucose was not different among groups even if it was higher in the middle-aged than in the young individuals when data were plotted regardless of exercise habits (5.05 (0.55) vs 4.66 (0.33) mmol/l; p = 0.006). Fasting plasma insulin was higher in the sedentary subgroups when compared with the physically active. Insulin sensitivity (HOMA2-%S) was lower in the subgroups of sedentary individuals in comparison with the physically active but no difference was detected within the subgroups paired for the habitual physical activity.

LV anatomical features

Morphological parameters of the LV are shown in table 2. Not surprisingly, the end-diastolic wall mass (EDWM) was higher in the physically active subgroups of individuals when compared with the sedentary subjects and this feature was maintained also when corrected for the body surface area. Also the end-diastolic volume (EDV) was higher in the physically active individuals, but in this case a statistical difference was detected only with respect to the middle-aged sedentary individuals.

LV function

Systolic function

Parameters of systolic function were not different among the four groups (one-way ANOVA; table 2). When a comparison was performed between the 25 sedentary individuals and 20 physically active regardless age, the stroke volume was higher (92 (15) vs 80 (13) ml; p<0.01) and the cardiac output lower (4.4 (0.) vs 5.4 (1.3) l/min; p<0.01) in the physically active than in the sedentary, meanwhile the ejection fraction was not different (60% (5%) vs 62% (5%); p = 0.31).

Diastolic function

Parameters of diastolic function, also in table 2, were different among groups. When a comparison was performed between the 25 sedentary individuals and 20 physically active regardless of age, the E/A peak flow rate was higher in the physically active men in comparison with the sedentary (2.50 (0.70) vs 2.15 (0.58); p = 0.05) because of a higher E-peak filling rate (498 (66) vs 443 (64) ml/s; p<0.02) with a similar A-peak filling rate (219 (53) vs 223 (49) ml/s; p = 0.81). In detail, the E-peak filling rate was reduced in the middle-aged sedentary individuals not only with respect to the sedentary (p<0.05) and physically active young subjects (p<0.02), but also with respect to the middle-aged physically active individuals (p = 0.05). The A-peak filling rate was increased in the subgroups of middle-aged individuals in comparison with the subgroups of young individuals, regardless of the physical exercise habit. As a consequence of this pattern, the E/A peak flow ratio was reduced in the sedentary middle-aged subjects in comparison with all the other study groups and was significantly reduced also in the physically active middle-aged subjects in comparison with the subgroups of young subjects.

LV PCr/ATP ratio

The PCr/ATP ratio, assessed by means of 31P-MRS was reduced in the sedentary middle-aged individuals when compared to the physically active subgroups of young (p = 0.02, table 2) and middle-aged (p<0.05; table 2) individuals, but not with respect to the sedentary young subjects (p = 0.11; table 2). The accuracy was excellent: the mean rCRSD was not different among groups (16% (6%), 12% (3%), 11% (3%) and 17% (7%) in the sedentary young and middle-aged subjects and in the physically active young and middle-aged subjects respectively; p = 0.98). In our setting, the intra-examination coefficient of variance (CV) (assessed by studying subjects twice and consecutively on the same session without changing the position of the surface coil) is 6% (3%). Inter-examination CV (studied by performing the acquisition in two separate occasions, with a time interval of 7–16 days with no effort to minimise variability) is 11% (5%). The data of inter-assay and intra-assay variability were the means obtained in subgroups of 19 individuals recruited for different protocols such as diabetic patients, obese subjects and individuals with heart failure.6 11 15

Correlation analysis

Age and LV morphological features

Pearson’s correlation analysis showed that age was inversely associated with the LV EDV (r = −0.49; p<0.01) and showed a trend with the end-diastolic wall mass (EDWM) (r = −0.31; p = 0.09) when plotting the data of the sedentary individuals, but these findings were lacking when plotting the data of the physically active individuals or when plotting the data of the entire (sedentary and physically active individuals) population.

Age and LV systolic function

The analysis showed that when plotting the data of the sedentary individuals, age was inversely associated with the LV stroke volume (r = −0.43; p<0.03) but not with ejection fraction and cardiac output, but these findings were lacking when plotting the data of the physically active individuals or when plotting the data of the entire population.

Age and LV diastolic function

In the sedentary population age was strongly associated with E peak flow (r = −0.72; p<0.001; fig 1A), A peak flow (r = 0.70; p<0.001; fig 1b) and E/A ratio (r = −0.87; p<0.001) but not with the DT. In the physically active individuals the association of age with the A peak flow (r = 0.62; p<0.004; fig 1B) remained robust; meanwhile with the E peak flow (r = −0.46; p<0.04; fig 1A) and with the E/A ratio (r = −0.69; p<0.004) it was lessened.

Age and metabolic parameters

When age was plotted against whole body parameters (plasma glucose and insulin concentration, HOMA2%-S) and local cardiac metabolic parameters (PCr/ATP ratio) only plasma glucose showed a significant correlation (r = 0.48; p<0.002).

Insulin sensitivity and LV function and energy metabolism

When an association between insulin sensitivity with LV morphological, functional and metabolic characteristics was explored, HOMA2%-S was found to be associated with the PCr/ATP ratio (r = 0.56; p<0.001 in the entire population and separately in the sedentary r = 0.39; p<0.04 and physically active individuals r = 0.50; p<0.02; fig 2). HOMA2%-S was not associated with morphological (EDV and EDWM) and parameters of systolic function (stroke volume, ejection fraction and cardiac output), but showed a weak association with the E/A ratio in the entire population (r = 0.37; p<0.04).

DISCUSSION

This study showed that the LV diastolic performance of sedentary middle-aged men was reduced compared with that of the young counterpart. Middle-aged men undergoing habitual aerobic oxidative training showed a better LV diastolic function when compared with their sedentary counterparts even if their parameters of diastolic function were still lower than that of the young sedentary individuals. The impairment of diastolic function in middle-aged men was paralleled by a lower ³¹P-MRS-derived marker of LV energy metabolism, which was preserved in the middle-aged men habitually performing aerobic physical activity.

In detail, in middle-aged sedentary men the E-peak filling rate was lower than in all the other study groups and the physically active middle-aged subjects were characterised by an improvement of this parameter; in parallel the A-peak filling rate was higher in the middle-aged sedentary subjects as a reflection of atrial compensation of the impaired early peak. Interestingly, the A-peak filling rate was also higher in the subgroup of middle-aged physically active individuals, resulting in the fact that LV diastolic performance of these subjects was still not comparable to that of men 20 years younger, either sedentary or physically active. It may be speculated looking at figure 1 (upper panel) that the E-peak filling rate inevitably drops during time, but the rate of deterioration of this parameter may be attenuated but not fully reversed by the regular performance of aerobic physical activity. In parallel, the A-peak filling rate also increases with ageing (fig 2, lower panel) but in this case the beneficial effect of exercise is less evident with this parameter, suggesting that the atrial phase of the diastolic function still needs to compensate for the age-dependent reduction of the E-peak filling rate.

Mitochondria have been a central focus of several theories of ageing as a result of their critical role in bioenergetics, oxidant production and regulation of cell death.16 The assessment of cardiac energy metabolism could be measured in myocardial specimens obtained during a biopsy, however this is problematic because the instability of these molecules, and the principal method for measuring ATP and PCr metabolism is 31P-MRS.17 Using this technique, we are reporting that along with progressive alteration of diastolic function, the LV of sedentary, middle-aged (45 (6) years) men is characterised by lower resting PCr/ATP ratio, reflecting and supporting the hypothesis of an age-related decline of cardiac high energy phosphate metabolism.18 This is in agreement with the finding that a decline in mitochondrial oxidative and phosphorylation energy production was detected using a similar 31P-MRS technique, also in the skeletal muscle of older (70 (2) year-old) individuals.19

Long-term physical exercise is a countermeasure to age-related cardiac mitochondrial dysfunction20; therefore in the present work we also assessed LV energy metabolism in the subgroup of middle-aged individuals undergoing aerobic physical exercise regularly for many years. They had a better PCr/ATP ratio than that of the sedentary subjects. Physical exercise is a powerful tool to improve whole body insulin sensitivity, modulating substrate metabolism in the muscle,21 and ageing is characterised by a progressive impairment of insulin sensitivity at the level of the skeletal muscle, which is associated with the impairment of mitochondrial function.19 It is possible, therefore, that a metabolic impairment may also be involved in the age-related functional and metabolic alterations of the LV and that physical exercise may be beneficial through a metabolic effect modulated at the level of other peripheral tissues (adipose tissue, liver, muscle and β-cell) or directly at the level of the cardiac muscle. With respect to this issue, it is known that there is a strong epidemiological association between insulin resistance and heart failure22 especially in the elderly with diabetes.23

Whether insulin resistance has a role into the pathogenesis of heart failure or whether it is a consequence of heart failure remains uncertain.24 In this study we reported a significant association of the surrogate marker HOMA2-%S with the PCr/ATP ratio in the sedentary individuals,6 but also in the physically active individuals (fig 2). These data may support the hypothesis that the abnormality of energy metabolism in the ageing heart may contribute to the progressive impairment of diastolic function, even if we noticed that the long-lasting, regular habit of performing physical exercise in middle-aged men resulted in a divergent outcome in these individuals. While the PCr/ATP ratio was super-compensated, the diastolic function was improved but still remained altered compared with the young sedentary subjects. Our feeling is that even if energy metabolism and diastolic function are tightly linked each other, in this context while cardiac energy metabolism is quickly susceptible to the beneficial effects of exercise, the age-dependent reduced diastolic function reflects long-lasting processes involving the decrease in cardiac mass because of a continuous loss of functioning muscle.2 These anatomical modifications are probably less prone to be fully reversed by the effects of exercise.

We want to stress that the assessment of the effects of exercise performed for such a long period of time (20–25 years) may be determined only with a cross-sectional approach. In support of this observation, it was recently shown, in a different setting of patients with dilated cardiomyopathy,25 that a beneficial effect of physical exercise on the ejection fraction was evident after long-term (8 months) but not after short-term26 exercise training. A more important limitation is that we used a surrogate marker of insulin sensitivity and we cannot exclude that the described relation between insulin sensitivity and LV PCr/ATP ratio may be improved using the clamp-derived index. Finally, it is interesting that the above-described difference in LV function and energy metabolism was detectable when comparing young (25 years old) with middle-aged (45-year-old) individuals; using a comparison with skeletal muscle,19 these differences might be amplified in much older healthy individuals.

In conclusion, this study demonstrated that in men the age-related alteration in LV diastolic function was associated with a reduction of a semi-quantitative marker of cardiac energy metabolism. Moreover, long-lasting, habitual physical activity was associated with the attenuation of this detrimental age-related effect on the LV even if full restoration of the diastolic function could not be achieved.