Adherence to a Mediterranean dietary pattern in early life is associated with lower arterial stiffness in adulthood: the Amsterdam Growth and Health Longitudinal Study

Authors

  • R. J. J. van de Laar,

    1. Department of Internal Medicine, Maastricht University Medical Centre+, Maastricht, the Netherlands
    2. CARIM School for Cardiovascular Diseases, Maastricht University Medical Centre+, Maastricht, the Netherlands
    3. CAPHRI School for Public Health and Primary Care, Maastricht University Medical Centre+, Maastricht, the Netherlands
    Search for more papers by this author
  • C. D. A. Stehouwer,

    1. Department of Internal Medicine, Maastricht University Medical Centre+, Maastricht, the Netherlands
    2. CARIM School for Cardiovascular Diseases, Maastricht University Medical Centre+, Maastricht, the Netherlands
    Search for more papers by this author
  • B. C. T. van Bussel,

    1. Department of Internal Medicine, Maastricht University Medical Centre+, Maastricht, the Netherlands
    Search for more papers by this author
  • M. H. Prins,

    1. Department of Internal Medicine, Maastricht University Medical Centre+, Maastricht, the Netherlands
    2. CAPHRI School for Public Health and Primary Care, Maastricht University Medical Centre+, Maastricht, the Netherlands
    3. Department of Clinical Epidemiology and Medical Technology Assessment, Maastricht University Medical Centre+, Maastricht, the Netherlands
    Search for more papers by this author
  • J. W. R. Twisk,

    1. Department of Epidemiology and Biostatistics, VU University Medical Center, EMGO Institute for Health and Care Research, Amsterdam, the Netherlands
    Search for more papers by this author
  • I. Ferreira

    Corresponding author
    1. Department of Internal Medicine, Maastricht University Medical Centre+, Maastricht, the Netherlands
    2. CARIM School for Cardiovascular Diseases, Maastricht University Medical Centre+, Maastricht, the Netherlands
    3. CAPHRI School for Public Health and Primary Care, Maastricht University Medical Centre+, Maastricht, the Netherlands
    4. Department of Clinical Epidemiology and Medical Technology Assessment, Maastricht University Medical Centre+, Maastricht, the Netherlands
    • Correspondence: Isabel Ferreira, Departments of Internal Medicine, Clinical Epidemiology and Medical Technology Assessment, Maastricht University Medical Centre+; P. Debyelaan 25, PO Box 5800, 6202 AZ Maastricht, the Netherlands.

      (fax: +31-43-3874419; e-mail: i.ferreira@maastrichtuniversity.nl).

    Search for more papers by this author

Abstract

Objectives

To investigate whether adherence to a Mediterranean dietary pattern during adolescence and early adulthood affects arterial stiffness in adulthood, and the extent to which any such association may be attributed to a beneficial impact of this diet on cardiovascular disease risk factors such as blood pressure, central fatness and dyslipidaemia.

Setting

The Amsterdam Growth and Health Longitudinal Study.

Design and subjects

We compared longitudinal levels of adherence to a Mediterranean dietary pattern (aMED score with range 0–9) during adolescence and adulthood (two to eight repeated measures obtained between the ages of 13 and 36) between individuals with different levels of arterial stiffness in adulthood. The study population included 373 (196 women) apparently healthy adults in whom properties of the carotid, brachial and femoral arteries were assessed using ultrasonography at 36 years of age.

Results

After adjustments for potential confounders, individuals with stiffer carotid arteries (defined on the basis of the most adverse tertile of, for instance, the distensibility coefficient) had lower aMED scores (−0.32, 95% CI −0.60; −0.06) and were less likely to have adhered to this dietary pattern (aMED score ≥5, odds ratio 0.69, 95% CI 0.50; −0.94) during the preceding 24 years compared with those with less stiff arteries. Differences in aMED scores were already present in adolescence and were only in part explained by the favourable associations between the Mediterranean dietary pattern and other cardiovascular disease risk factors (up to 26%), particularly mean blood pressure (up to 19%).

Conclusions

Promoting the Mediterranean diet in adolescence and early adulthood may constitute an important means of preventing arterial stiffness in adulthood.

Introduction

There is much evidence to support a protective effect of a Mediterranean dietary pattern, which is characterized by high intake of plant foods and olive oil, moderately high intake of fish, low-to-moderate intake of meat and poultry, and regular but moderate consumption of alcohol [1], on cardiovascular disease (CVD) [2-10]. Indeed, greater adherence to a Mediterranean dietary pattern correlates favourably with several CVD risk factors (RFs), such as elevated blood pressure (BP) [2-4], total body fatness and central fat distribution [4, 5], lipid profile [3, 4] and insulin resistance [3, 4], and with lower incidence of CVD and cardiovascular and all-cause mortality [6-10].

The evidence supporting the protective effects of a Mediterranean dietary pattern on incident CVD has been derived mainly from studies conducted amongst middle-aged and older individuals. However, the pathophysiological processes underlying the development of CVD start much earlier [11]. In this line, we [12-15] and others [16, 17] have shown that the presence of RFs such as elevated BP and central fatness at a young age is associated with adult levels of arterial stiffness (i.e. the loss of the elastic properties and cushioning function of large arteries), which may lead to poorer cardiovascular outcome [18, 19]. With ageing, arterial stiffening leads to an increase in systolic BP (SBP) [20] because higher end-systolic pressures must be generated for the same net stroke volume. This leads to increased decay of arterial pressure and volume during systole, causing a reduced arterial volume at the onset of diastole, which in turn causes a greater decrease in diastolic BP (DBP). As such, arterial stiffness may lead to an increased risk of stroke, the development of left ventricular hypertrophy and a decrease in coronary perfusion and heart failure [19]. Indeed, it has been shown that stiffness levels of mainly central (i.e. elastic) arteries, such as the aorta [21] and the carotid arteries [22-25], predict incident CVD and mortality across different populations.

A recent systematic review of dietary and nutrient interventions demonstrated that omega-3 and soy isoflavone supplementation reduced arterial stiffness, thus supporting a role of diet in the prevention of arterial stiffening and its related sequelae. However, this evidence was derived from small studies that were mainly focused on individual foods or nutrients and confined to clinical populations [26]. The extent to which (better adherence to) a Mediterranean dietary pattern throughout the course of adolescence/young adulthood impacts favourably on levels of arterial stiffness later in life is unknown. From a primary prevention point of view, this may be of particular relevance because lifelong adherence to a (un)healthy diet may start at a young age [27]. In addition, the complex interplay between different foods and nutrients may result in synergistic effects on health [28], and therefore, evaluation of dietary patterns, such as the Mediterranean diet, rather than the individual dietary components, may better reflect individuals' dietary habits and capture its impact on cardiovascular health [29].

From an aetiological point of view, the question arises as to whether any beneficial impact of a Mediterranean dietary pattern on arterial stiffness may be due to (i.e. mediated by) or independent of the protective effects of this diet on traditional CVD RFs, some of which (e.g. BP and central fatness) are known determinants of arterial stiffness [15]. In an effort to answer this question, we used mediation analysis [30, 31] to quantify the extent and significance of mediation by these RFs in the association between adherence to a Mediterranean diet and arterial stiffness.

We have therefore investigated these issues in a longitudinal cohort of individuals whose usual dietary intake and CVD RFs were repeatedly assessed from adolescence to adulthood, and levels of arterial stiffness were determined at 36 years of age.

Subjects and methods

All subjects participated in the Amsterdam Growth and Health Longitudinal Study (AGAHLS), an observational study that started in 1976–1977 with a group of approximately 600 boys and girls from two secondary schools in the area of Amsterdam, the Netherlands. This cohort was established to enable the investigation of growth, lifestyle and health of adolescents, and the longitudinal relationships between lifestyle and biological RFs extending into adulthood, as described in detail elsewhere [32]. Briefly, the mean age of the subjects at the beginning of the study was 13.1 ± 0.8 years. Since then, repeated measurements of lifestyle (including dietary, physical activity and smoking habits), anthropometric (height, weight and skinfold measurements) and biological (BP and blood lipid levels) RFs were obtained according to standard procedures two to eight times up to the age of 36 years, over a total follow-up period of 24 years [13, 32-34]. In the year 2000, when the mean age of participants was 36.5 ± 0.5 years, biophysical properties of three large arteries were evaluated for the first time in 373 participants (196 women) [12, 13, 15, 35]; these subjects comprise the sample of the present study.

The medical ethical committee of the VU University Medical Center (Amsterdam, the Netherlands) approved the study, and all subjects gave their written informed consent (provided by parents when participants were 13–16 years old).

Arterial stiffness

Biophysical properties of the large arteries were measured according to guidelines and using reproducible methods [18, 36, 37], as described in detail elsewhere [12, 13, 15, 35]. Briefly, all subjects had abstained from smoking and consumption of caffeine-containing beverages on the day of the measurements, which took place after subjects had been resting in a supine position for 15 min in a quiet temperature-controlled room. Properties of the right common carotid (10 mm proximal to the beginning of the carotid bulb) and the brachial (20 mm above the antecubital fossa) and common femoral (20 mm proximal to the flow divider) arteries were obtained with the use of an ultrasound scanner equipped with a 7.5-MHz linear array probe (Pie Medical, Maastricht, the Netherlands). The ultrasound scanner was connected to a personal computer equipped with an acquisition system and a vessel wall movement detector software system (Wall Track System 2, Pie Medical). This integrated device enabled measurement of arterial diameter (D), distension (ΔD) and intima–media thickness (IMT) as described in detail previously [36, 37]. SBP, DBP and mean BP (MBP) were assessed in the left arm at 5-min intervals throughout the period of ultrasound imaging using an oscillometric device (Colin Press-Mate, model BP-8800, Komaki-City, Japan). Brachial pulse pressure (PP) was defined as SBP–DBP, and PP at the level of the carotid and femoral arteries was calculated by calibration of the distension waveforms [38]. The mean values of D, ΔD and local PP from three consecutive measurements were used to estimate the distensibility (DC) and compliance (CC) coefficients of each artery as follows:

display math(1)
display math(2)

The DC reflects the elastic properties, whereas the CC reflects the buffering capacity of the artery at given operating local pressures. From D, DC and IMT, the carotid Young's elastic modulus (Einc), an estimate of the intrinsic elastic properties of the vessel wall, was calculated as follows:

display math(3)

Degree of stiffness

In contrast to the Einc, higher values of the DC and CC indicate lower arterial stiffness. Therefore, DC and CC values were multiplied by −1 (i.e. reversed) prior to categorizing subjects into three groups on the basis of the sex-specific tertiles (T) of the stiffness estimate: less stiff (T1), intermediate (T2) and stiffer arteries (T3) [13, 15, 35].

Dietary assessment

From 13 to 36 years of age, usual dietary intake was measured using a cross-check dietary history (face-to-face) interview, as described previously [39, 40]. In brief, participants were asked to recall their usual food intake during the previous month by reporting frequency, amount and method of preparation of foods and drinks (including alcoholic beverages) consumed. Amounts were reported in household measures or grams, and models were used to illustrate portion sizes. During the measurements in 2000, a newly developed interviewer-administered, computer-assisted, cross-check dietary history method was introduced [40]. The questions and interview structure were based on the original face-to-face interview. The two methods of dietary assessment showed a high level of comparability; though, it is likely that introduction of the computer-assisted method caused a reduction in interviewer bias [40]. Total energy intake was calculated using the 1996 Dutch Food Composition Table [41].

Mediterranean diet score

For each subject and for each measurement round, we calculated the alternate Mediterranean diet (aMED) score [6, 10, 42], which is an adapted version of the original score developed by Trichopoulou et al. [8] and contains the following nine dietary components: vegetables (excluding potatoes), fruits, whole grains, nuts, legumes, fish, red and processed meats, ratio of monounsaturated-to-saturated fat and alcohol. Subjects scored one point whenever the intake of each component was above the sex-specific median, except in the case of red and processed meat, for which one point was attributed if the intake was below the sex-specific median, and alcohol intake, for which one point was scored if participants abstained from drinking during adolescence (ages 13–16) or if the level of consumption was 5–25 g day−1 during adulthood (ages 21–36) [10]. This distinction seemed to be appropriate because underage drinking may outweigh any beneficial effects of (moderate) alcohol consumption. The aMED score thus ranged from 0 to 9 with higher values, indicating a higher degree of adherence to a Mediterranean dietary pattern.

Statistical analyses

We used generalized estimating equations (GEEs) to investigate the mean differences in the aMED score throughout the 24-year longitudinal period between subjects with different levels of arterial stiffness at the age of 36 [13, 15, 35]. GEEs properly take into account the correlation between repeated observations within the same individual and have the advantage of being able to analyse varying numbers of and unequally time-spaced observations [43]. All analyses were adjusted for sex, height and time (model 1). Interaction terms between stiffness groups and time were added to this model to ascertain whether the magnitude of the differences in the aMED score emerged at any specific age during the course of young life. Results were displayed graphically and reflect the life-course trajectories of the aMED score by levels of arterial stiffness at age 36 [13, 15, 35, 43]. The mean differences in the aMED score between groups with different levels of arterial stiffness were further adjusted for potential confounders such as total energy intake (kcal), habitual physical activity (metabolic equivalents per week) and smoking (yes/no) (model 2).

Generalized estimating equation analyses were also used to ascertain longitudinal associations between the aMED score and CVD RFs (i.e. BP, body fatness and blood lipid levels) throughout the 24-year period; results were expressed by standardized longitudinal regression coefficients to allow direct comparison of the strength of the associations between the score and each RF.

An exchangeable correlation structure was used in all analyses, and statistical significance was set at < 0.05 (two sided). GEE analyses were carried out using stata software version 9.2 (STATA Corp, College Station, TX, USA).

Mediation analyses

The mediating role of a variable (mediator) hypothesized to be involved in the pathway between a given determinant (Mediterranean diet) and a given outcome (arterial stiffness) is ascertained by quantitative evaluation (i.e. extent and significance) of the attenuation in the strength of the association between determinant and outcome after adjustment for the potential mediator [30, 31]. We hypothesized that RFs such as MBP, central fatness and/or dyslipidaemia, all of which were previously shown to be associated with arterial stiffness in this study population [12, 15], could play a mediating role in the association between Mediterranean diet and arterial stiffness. Therefore, we additionally adjusted the differences in the levels of lifetime aMED score between subjects with stiffer versus less stiff arteries (as obtained in model 2 described earlier) for these potential mediators, both one at a time (models 3a–c) and simultaneously (model 4). We used the ab product coefficient method to quantify the magnitude of the attenuations because of these adjustments and the respective 95% confidence intervals (CIs); the latter enable appreciation of whether mediation by any of the RFs investigated was significant (if so, the 95% CI around the magnitude of attenuation would not include 0) [30, 31]. These analyses were conducted with the RMediation package written for use in the r statistical program (version 2.13.0, The R Foundation for Statistical Computing 2011) [44].

Results

Characteristics of the study population and dietary consumption throughout the 24-year period are shown in Tables 1 and 2, respectively. Levels of arterial stiffness across tertiles of the stiffness are presented in Table 3; these data are shown for men and women separately because sex-specific cut-off values were used in the analyses. All subsequent data are shown for men and women together as there was no evidence for sex interactions.

Table 1. Characteristics of the study population throughout the 24-year study period
VariablesSexAge, yearsOverall
1314151621273236
  1. Data presented as mean ± SD, median (interquartile range) or percentage. METs, metabolic equivalents.

  2. aMeasured in sitting position after rest for ≥5 min; bCalculated as (2*DBP + SBP)/3; cSum of the skinfold thickness of the triceps, biceps, subscapular and suprailiac; dRatio calculated as (subscapular + suprailiac)/sum of four skinfold measurements; eSI conversion factor from mmol L−1 to mg dL−1: divided by 0.0259.

Lifestyle
Total energy intake, 1000 kcal day−1M2.72 ± 0.532.85 ± 0.492.99 ± 0.633.00 ± 0.633.11 ± 0.652.83 ± 0.623.00 ± 0.742.97 ± 0.742.93 ± 0.66
F2.22 ± 0.452.22 ± 0.502.22 ± 0.492.18 ± 0.472.21 ± 0.522.17 ± 0.482.26 ± 0.462.32 ± 0.492.24 ± 0.48
Habitual physical activity, 1000 METs per weekM4.87 ± 0.184.59 ± 0.194.09 ± 0.183.66 ± 0.203.35 ± 0.212.82 ± 0.213.30 ± 0.244.34 ± 0.273.95 ± 0.23
F4.00 ± 0.173.49 ± 0.113.36 ± 0.143.39 ± 0.113.23 ± 0.213.16 ± 0.193.51 ± 0.225.37 ± 0.363.86 ± 0.24
Smoking, %M0.08.814.19.227.733.324.428.018.4
F3.112.813.925.231.620.016.819.417.1
Anthropometric and biological
Systolic blood pressure, mmHgaM124.7 ± 9.4124.7 ± 8.4127.9 ± 9.2131.7 ± 9.7133.4 ± 10.7136.5 ± 10.9135.2 ± 12.5138.6 ± 13.9132.1 ± 12.3
F124.6 ± 9.4122.2 ± 9.7122.6 ± 9.9121.3 ± 8.9124.9 ± 10.2123.4 ± 9.1124.9 ± 10.2124.6 ± 11.2123.7 ± 10.0
Diastolic blood pressure, mmHgaM74.1 ± 7.974.3 ± 7.671.1 ± 8.073.2 ± 8.579.0 ± 8.282.4 ± 9.585.9 ± 8.289.2 ± 10.880.0 ± 11.1
F76.7 ± 7.577.3 ± 7.373.8 ± 7.875.9 ± 7.678.4 ± 8.580.0 ± 7.683.5 ± 9.082.0 ± 9.279.1 ± 8.9
Mean blood pressure, mmHgbM91.0 ± 6.891.1 ± 6.390.0 ± 86.792.7 ± 7.397.1 ± 7.3100.4 ± 8.5102.3 ± 8.7105.7 ± 10.997.3 ± 10.3
F92.7 ± 7.092.2 ± 6.890.1 ± 7.091.0 ± 7.193.9 ± ±8.394.5 ± 7.397.3 ± 8.596.2 ± 9.194.0 ± 8.2
Body mass index, kg m−2M17.4 ± 1.418.1 ± 1.518.9 ± 1.719.6 ± 1.521.4 ± 1.722.6 ± 2.124.0 ± 2.624.8 ± 2.721.3 ± 3.5
F18.0 ± 2.118.8 ± 2.319.5 ± 2.420.0 ± 2.421.3 ± 2.621.9 ± 2.522.8 ± 3.123.4 ± 3.421.0 ± 3.4
Sum of four skinfolds, mmcM26.9 ± 8.826.0 ± 8.426.4 ± 8.528.6 ± 7.935.3 ± 11.836.8 ± 13.641.8 ± 17.147.2 ± 15.435.2 ± 15.1
F36.8 ± 12.639.8 ± 14.743.6 ± 15.047.4 ± 17.152.7 ± 17.146.2 ± 16.952.0 ± 19.755.5 ± 19.647.6 ± 18.3
Skinfold ratiodM0.49 ± 0.060.52 ± 0.060.55 ± 0.050.58 ± 0.050.64 ± 0.050.62 ± 0.050.63 ± 0.060.65 ± 0.060.59 ± 0.08
F0.49 ± 0.050.51 ± 0.060.51 ± 0.060.52 ± 0.060.53 ± 0.060.51 ± 0.060.51 ± 0.070.50 ± 0.070.51 ± 0.06
Total cholesterol, mmol L−1eM4.43 ± 0.684.30 ± 0.714.10 ± 0.664.00 ± 0.614.50 ± 0.854.90 ± 0.894.97 ± 0.945.17 ± 0.994.63 ± 0.92
F4.50 ± 0.774.39 ± 0.704.46 ± 0.734.43 ± 0.844.85 ± 0.725.25 ± 0.994.88 ± 0.784.84 ± 0.854.71 ± 0.84
HDL cholesterol, mmol L−1eM1.46 ± 0.281.43 ± 0.271.25 ± 0.211.29 ± 0.191.15 ± 0.201.19 ± 0.221.21 ± 0.271.21 ± 0.281.28 ± 0.27
F1.44 ± 0.291.40 ± 0.251.35 ± 0.271.41 ± 0.311.40 ± 0.301.63 ± 0.371.60 ± 0.351.58 ± 0.351.49 ± 0.33
Total-to-HDL cholesterol ratioM3.12 ± 0.693.08 ± 0.653.36 ± 0.743.16 ± 0.594.01 ± 0.964.25 ± 1.064.30 ± 1.244.52 ± 1.363.80 ± 1.17
F3.21 ± 0.703.22 ± 0.693.42 ± 0.803.26 ± 0.823.58 ± 0.813.33 ± 0.823.19 ± 0.813.20 ± 0.903.27 ± 0.81
Table 2. Daily intake of the individual components of the Mediterranean diet throughout the 24-year study period
ComponentSexAge, yearsOverall
1314151621273236
  1. Data are presented as mean ± SD, median[interquartile range] or percentage. aMED, alternate Mediterranean diet score.

  2. aAmongst drinkers; bCriterion met when alcohol consumption is 0 g day−1 in adolescence and 5–25 g day−1 in adulthood.

n, men/women 113/12491/11099/11086/10666/7969/80160/186173/195 
aMED 4.29 ± 1.534.23 ± 1.484.10 ± 1.633.86 ± 1.553.71 ± 1.553.88 ± 1.624.77 ± 1.745.21 ± 1.684.41 ± 1.69
Vegetables, g day−1M91[72–127]127[102–147]124[101–155]126[100–149]146[116–191]186[143–221]166[130–231]164[114–216]139[105–184]
F85[68–109]109[90–140]105[90–130]110[83–134]120[95–162]164[127–220]160[124–218]166[125–229]128[95–172]
Fruits, g day−1M194[130–290]185[130–250]187[112–267]177[118–279]215[113–301]177[71–313]234[134–358]232[124–369]205[121–310]
F242[175–341]216[143–397]253[156–347]198[129–343]251[146–366]159[92–233]260[169–390]254[150–366]236[144–356]
Nuts, g day−1M0[0–7]2[0–7]0[0–7]0[0–7]0[0–11]0[0–14]4[0–15]3[0–13]1[0–9]
F0[0–7]0[0–4]0[0–7]0[0–7]0[0–5]0[0–8]1[0–9]2[0–9]0[0–7]
Legumes, g day−1M0[0–5]0[0–6]0[0–6]0[0–6]0[0–5]0[0–5]0[0–8]0[0–10]0[0–6]
F2[0–5]0[0–5]2[0–6]0[0–6]0[0–6]0[0–5]0[0–9]0[0–8]0[0–6]
Whole grains, g day−1M81[4–163]80[0–163]94[0–176]132[43–198]137[63–197]103[59–168]133[84–188]155[90–193]121[43–182]
F66[3–134]90[12–132]89[0–137]99[21–137]91[51–143]90[63–124]103[74–141]132[101–171]101[56–143]
Meat, g day−1M116[96–138]109[86–136]111[86–149]104[84–130]135[109–161]127[91–162]116[95–141]91[62–122]111[85–139]
F106[85–131]105[87–130]100[77–123]93[71–118]108[77–137]100[72–133]84[53–108]66[44–87]92[66–119]
Fish, g day−1M14[0–29]0[0–29]0[0–14]0[0–14]0[0–15]0[0–18]8[0–26]11[0–35]0[0–29]
F0[0–29]0[0–14]0[0–21]0[0–21]0[0–20]0[0–29]11[0–25]14[0–29]0[0–26]
Monounsaturated-to-saturated fat ratioM0.90[0.84–0.99]0.95[0.83–1.03]0.92[0.81–1.02]0.89[0.80–1.02]0.94[0.81–1.05]0.94[0.82–1.08]0.97[0.86–1.08]1.00[0.86–1.14]0.94[0.83–1.05]
F0.93[0.83–1.03]0.94[0.85–1.02]0.92[0.82–1.00]0.87[0.77–0.97]0.92[0.76–0.98]0.85[0.76–1.01]0.91[0.81–1.03]0.94[0.85–1.07]0.91[0.80–1.01]
Alcohol drinkers, %M13.317.633.345.969.782.688.890.859.0
F13.714.533.350.068.463.873.174.451.4
Consumption, g day−1aM1.6[0.8–2.9]2.5[1.7–3.4]3.1[1.6–9.1]5.7[2.7–11.4]13.6[6.4–25.2]12.5[3.6–19.6]11.5[6.0–21.0]16.9[5.8–26.7]10.4[3.7–20.9]
F1.1[0.7–1.9]1.6[0.8–3.7]2.8[1.3–4.9]3.7[1.9–6.5]5.3[2.7–10.7]7.9[4.7–13.1]6.9[2.7–15.5]10.5[4.3–20.8]5.7[2.5–12.0]
Criterion met, %bM87.482.466.754.737.944.956.346.859.9
F85.885.565.550.034.238.839.841.554.3
Table 3. Arterial stiffness levels at the age of 36 according to sex-specific tertiles of each stiffness estimate
Stiffness estimatesSexTertile 1 (Less stiff)Tertile 2 (Intermediate)Tertile 3 (Stiffer)
  1. Data are means ± SD.

  2. aDistensibility and compliance coefficients were multiplied by −1 (i.e. reversed) prior to categorization into tertiles, so that, in agreement with the carotid Young's elastic modulus, higher tertiles indicate higher arterial stiffness.

Men/women, n 59/6559/6659/65
Common carotid artery
Distensibility coefficient, 10−3 kPa−1aM32.0 ± 3.425.8 ± 1.420.7 ± 2.1
F34.7 ± 4.326.5 ± 1.719.9 ± 2.3
Compliance coefficient, mm2 kPa−1aM1.35 ± 0.201.04 ± 0.050.80 ± 0.01
F1.22 ± 0.180.91 ± 0.060.65 ± 0.09
Young's elastic modulus, 103 kPa−1M0.35 ± 0.040.45 ± 0.020.60 ± 0.09
F0.30 ± 0.040.41 ± 0.030.56 ± 0.08
Brachial artery
Distensibility coefficient, 10−3 kPa−1aM22.0 ± 10.510.5 ± 1.56.2 ± 1.6
F26.2 ± 7.914.1 ± 1.97.9 ± 2.3
Compliance coefficient, mm2 kPa−1aM0.30 ± 0.100.17 ± 0.020.10 ± 0.03
F0.24 ± 0.050.14 ± 0.020.07 ± 0.2
Common femoral artery
Distensibility coefficient, 10−3 kPa−1aM9.0 ± 3.15.5 ± 0.53.3 ± 0.8
F12.9 ± 3.87.4 ± 0.94.7 ± 0.9
Compliance coefficient, mm2 kPa−1aM0.78 ± 0.220.48 ± 0.050.29 ± 0.06
F0.78 ± 0.190.45 ± 0.050.29 ± 0.06

Lifetime adherence to a Mediterranean dietary pattern and arterial stiffness at the age of 36

After adjustment for sex, height and time, subjects with stiffer carotid arteries at the age of 36 had significantly lower mean aMED scores during the 24-year study period, compared with subjects with less stiff arteries: −0.36 (95% CI −0.62; −0.10), −0.56 (95% CI −0.83; −0.29) and −0.31 (95% CI −0.58; −0.04) when stiffness levels were determined on the basis of the DC, CC and Einc, respectively (Table 4, model 1). Of note, such differences were already present during adolescence and remained fairly stable throughout the study period, particularly for comparisons based on the carotid DC and CC (Fig. 1). After adjustment for potential confounders (i.e. total energy intake and other lifestyle variables), the mean differences in the aMED scores were attenuated to −0.32 (95% CI −0.58; −0.06), −0.52 (95% CI −0.79; −0.25) and −0.27 (95% CI −0.54; −0.01), respectively, but remained statistically significant (model 2).

Figure 1.

The alternate Mediterranean diet score throughout the study period by tertile (T) of the carotid distensibility (DC) and compliance (CC) coefficients and Young's elastic modulus (Einc) at the age of 36. Data are adjusted for sex, height and time. Error bars indicate standard errors of the means. *P < 0.05; †P < 0.01; ‡P < 0.001 for T3 (stiffer) versus T1 (less stiff arteries).

Table 4. Mean difference in the alternate Mediterranean diet score throughout the study period as a function of carotid stiffness at the age of 36
Model adjustmentsDCMediation (%)dCCMediation (%)dEincMediation (%)d
β95% CIβ95% CIβ95% CI
  1. β, longitudinal regression coefficient: indicates mean differences in Mediterranean diet score throughout the 24-year period between subjects with stiffer (i.e. in the highest tertile; T3) versus less stiff (i.e. in the lowest tertile; T1) carotid arteries as defined on the basis of the distensibility (DC) or compliance (CC) coefficients (both reversed by multiplying by −1) or Young's elastic modulus (Einc).

  2. a< 0.05, b< 0.01 and c< 0.001; dExpresses attenuation in magnitude of β shown in model 2 after further adjustment for the potential mediators separately (models 3a–c) or together (model 4);% mediation = (βmodels 3 or 4 − βmodel 2)/βmodel 2*100; eIndicates statistically significant mediation – for further details see Table S2.

1Sex, height and time−0.36−0.62; −0.10b−0.56−0.83; −0.29c−0.31−0.58; −0.04a
2Model 1 +  total energy intake, physical activity and smoking−0.32−0.58; −0.06a−0.52−0.79; −0.25c−0.27−0.54; −0.01a
Potential mediators
3aModel 2 +  mean blood pressure−0.26−0.53; 0.0119e−0.48−0.76; −0.20b8e−0.23−0.50; 0.0518e
3bModel 2 +  skinfold ratio−0.31−0.57; −0.04a5−0.51−0.78; −0.25c2−0.26−0.53; 0.015
3cModel 2 +  total-to-HDL cholesterol ratio−0.31−0.57; −0.04a7−0.53−0.80; −0.26c3−0.26−0.53; 0.026
4Model 2 +  all variables in models 3a–c−0.25−0.52; 0.0326e−0.49−0.76; −0.21c10e−0.21−0.49; 0.0624e

When adherence to a Mediterranean dietary pattern was defined by aMED score ≥5 (i.e. the median value in this population), subjects with stiffer carotid arteries were less likely to have adhered to this diet throughout the 24-year study period than their counterparts with less stiff arteries: odds ratio (OR) 0.68 (95% CI 0.50; 0.94), 0.64 (95% CI 0.46; 0.87) and 0.68 (95% CI 0.50; 0.93) for comparisons based on the DC, CC and Einc, respectively (Fig. 2). Similar trends, albeit mostly nonsignificant, were observed for each of the components comprising the aMED score, except nuts, legumes and the monounsaturated-to-saturated fatty acids ratio. Clustering of the components in the aMED score thus seemed to better capture the favourable associations with carotid arterial stiffness than each component separately.

Figure 2.

Longitudinal odd ratio of adherence to the alternate Mediterranean diet (aMED; score ≥5) and favourable intake of each of its components throughout the 24-year study period for subjects with stiffer (in the highest tertile; T3) versus less stiff (in the lowest tertile; T1, i.e. the reference category) carotid arteries; stiffness levels were defined on the basis of the distensibility (DC) or compliance (CC) coefficients (both reversed by multiplication by −1), or the Young's elastic modulus (Einc) at the age of 36. Whiskers indicate the 95% confidence intervals. Data are adjusted for sex, height, time, total daily energy intake, habitual physical activity and smoking. aFavourable intake of each of the aMED components defined as consumption above the sex-specific median, except for meat intake (defined as consumption below the sex-specific median) and alcohol consumption (defined as abstinence during adolescence and 5–25 g day−1 during adulthood). bAnalyses with different components of the aMED were mutually adjusted for each other.

Qualitatively, similar differences in lifetime aMED score and likelihood of adherence, albeit of lower magnitude, were observed for comparisons based on the estimated level of brachial and femoral stiffness at 36 years (Table S1).

Longitudinal associations between the aMED score and other CVD RFs

Higher levels of the aMED score were significantly associated with lower levels of BP (SBP, DBP and MBP) and total cholesterol throughout the longitudinal period. Inverse but weaker associations were also found with anthropometric measures of total and central fatness, and with HDL and total-to-HDL cholesterol ratio (Fig. 3).

Figure 3.

Longitudinal associations between the alternate Mediterranean diet (aMED) score and cardiovascular disease risk factors. Point estimates indicate change in risk factor (in SD) per two-point increase in the aMED score; whiskers indicate the 95% confidence intervals. Data were first adjusted for sex, height and time (model 1, black lines) and then additionally for potential confounders (i.e. total daily energy intake, habitual physical activity and smoking; model 2, grey lines).

Lifetime adherence to a Mediterranean dietary pattern and arterial stiffness at the age of 36: mediation role of CVD RFs

The mean differences in aMED between subjects with stiffer compared with those with less stiff carotid arteries (Table 4, model 2) were significantly attenuated after further adjustment for MBP (models 2 vs. 3a), but less so after adjustment for the skinfold ratio (reflecting a central body fat distribution; models 2 vs. 3b) or the total-to-HDL cholesterol ratio (models 2 vs. 3c), using simple mediation analyses. When analyses included adjustment for MBP, the skinfold ratio and total-to-HDL cholesterol ratio simultaneously (i.e. multiple mediation model), the differences between subjects with stiffer versus less stiff carotid arteries as estimated on the basis of the DC, CC or Einc were attenuated by 26% (i.e. from −0.32 to −0.25), 10% (i.e. from −0.52 to −0.49) and 24% (i.e. from −0.27 to −0.21), respectively (models 2 vs. 4). Furthermore, in these multiple mediation analyses, MBP accounted for the greatest portion of the attenuations (i.e. 19%, 8% and 17%, respectively) and was the only mediator that attenuated the lifelong differences in aMED score between subjects with stiffer versus less stiff carotid arteries to a significant extent (for further details regarding mediation analyses, see Table S2).

Replacing adjustment for the skinfold ratio by body mass index (BMI) or sum of the skinfold measurements and for total-to-HDL cholesterol ratio by total cholesterol did not essentially affect the results of these analyses (data not shown).

Discussion

The first main finding of this study was that subjects with stiffer carotid arteries, and to a lesser extent brachial and femoral arteries, at the age of 36 had adhered to a lesser extent to a Mediterranean dietary pattern during the preceding 24 years, suggesting a favourable association between a Mediterranean diet throughout adolescence and early adulthood and arterial stiffness in adulthood. Secondly, greater adherence to a Mediterranean dietary pattern was associated with lower levels of BP, total cholesterol and BMI throughout the study period. Thirdly, the beneficial association with MBP explained in part (up to 19%) the favourable association between the Mediterranean dietary pattern and arterial stiffness, which remained largely independent of the other CVD RFs.

To the best of our knowledge, this is the first study to investigate the associations between intake of a Mediterranean diet throughout adolescence and early adulthood and arterial stiffness in adulthood. Our findings are in line with those of previous studies, showing that high intake of fruits and vegetables in adolescence and young adulthood was associated with lower aortic stiffness in adulthood [45], and that a dietary pattern low in fruits, sweetening and dairy products and high in meat and alcohol (identified by principal component analyses) was associated with higher aortic stiffness 7.5 years later in middle-aged individuals [46]. In contrast to the latter study, we investigated adherence to a dietary pattern defined a priori (Mediterranean diet) instead of individual foods or dietary patterns defined a posteriori by means of exploratory data analyses techniques [29]. Investigating a dietary pattern may capture better synergistic health effects and reduce confounding bias by separate foods or nutrients, which are more prone to show associations because of chance [28, 29]. In addition, we obtained repeated dietary data throughout a period of 24 years from adolescence to young adulthood. This enabled a more robust assessment of risk exposure [47], appreciation of its changes over time and identification of potential critical periods early in life when exposure could be related to arterial stiffness in adulthood.

Indeed, trajectory analyses revealed that adherence to a Mediterranean dietary pattern, as depicted by the calculated aMED score, declined from adolescence into young adulthood in all individuals and differed already during adolescence and throughout follow-up between individuals with different levels of arterial stiffness at 36 years. In addition, greater adherence to a Mediterranean dietary pattern was associated with lower levels of other CVD RFs, such as BP, total cholesterol and BMI, all of which have been shown to track from young to older age [34, 48]. Despite the fact that the aMED score was inversely (but not significantly) associated with HDL cholesterol, these data emphasize a potential preventive role of adherence to a Mediterranean diet starting at a young age. Preventive strategies targeting children and their parents may, given the shared environment and parents' primary responsibility for their children's dietary intake [49], thus be more beneficial as a means to engage in long-term healthy dietary habits [27]. In addition, targeting the time of transition between adolescence and young adulthood, when individuals become increasingly more independent and responsible for their own dietary choices, may be particularly important, given the decline in dietary quality observed during this period.

The protective association between lifelong adherence to a Mediterranean dietary pattern and arterial stiffness found in the present study applied to both the elastic carotid and the muscular brachial and femoral arteries, suggesting a favourable impact throughout the arterial tree. The effect estimates were relatively smaller for the muscular arteries, however, which may reflect underestimation given the larger measurement error with which properties of these arteries (mainly distension) were measured [12]. Alternatively, structural differences between elastic and muscular arteries may account for these findings [35]. Nevertheless, the observation of significant associations between adherence to a Mediterranean dietary pattern and stiffness estimates of mainly the carotid artery may be clinically most relevant as stiffness levels of elastic arteries in particular (e.g. the carotid and the aorta) have been shown to predict incident CVD [21-25]. In addition, differences in stiffness levels between the groups being compared (i.e. T3 vs. T1 of the carotid stiffness estimates) correspond to those previously observed over one decade of ageing [50], suggesting that physiologically meaningful differences in arterial stiffness levels at age 36 were studied. Our findings thus suggest that lower arterial stiffness may in part explain the beneficial impact of a Mediterranean diet on CVD and related mortality [6-10].

We have previously shown that the life-course levels of MBP and central (i.e. skinfold ratio) rather than total (i.e. BMI or the sum of skinfolds) fatness were the main independent biological RFs associated with arterial stiffness in adulthood [15]. In line with these earlier observations, we have now shown that the beneficial association between the Mediterranean diet and MBP constituted a significant biological pathway through which this dietary pattern could affect arterial stiffness. However, the mediating role of MBP accounted for only a small part (up to 19%) of the associations between the Mediterranean diet and carotid stiffness, and all RFs considered up to 26%, thus suggesting that other factors may also play a role. Such factors may include less low-grade inflammation, reduced endothelial dysfunction [3, 42], reduced oxidative stress [51] and/or improved insulin sensitivity [3] and thus need to be further investigated as potential mechanisms linking a Mediterranean dietary pattern to arterial stiffness. However, longitudinal measures of these factors were not available in this cohort.

The aMED score calculated in the present and other studies [6, 10, 42] differs from the traditional [2, 8] and modified [9] Mediterranean diet scores originally proposed by Trichopoulou et al. by separating fruits from nuts, excluding dairy products, using whole grains instead of total cereals and applying the same criterion for moderate alcohol consumption in both sexes. The modified Mediterranean diet also takes into account polyunsaturated fat intake. We chose to use the aMED score because recent evidence indicates that quality of carbohydrates is an important determinant of cardiovascular health [52]. Nevertheless, additional analyses showed high correlations between the three scores (standardized longitudinal regression coefficients all >0.7, < 0.001), and repeating analyses using the two other scores instead yielded essentially the same results as those reported (data not shown).

Some limitations of this study should be considered. First, our study was confined to participants in the AGAHLS attending the follow-up measurement round in 2000 in whom complete data on arterial properties could be assessed. These subjects did not significantly differ those who dropped outs at any of the earlier time-points with regard to the RFs considered and the consumption of foods comprising the aMED score, except for lower alcohol consumption at the mean ages of 14, 15 and 21 years. It is thus unlikely that selection bias has affected the validity of our findings. Secondly, longitudinal data on dietary intake were collected prospectively but were self-reported, which might be subject to misclassification and recall bias [53]. Given that throughout the study period individuals were unaware of their arterial stiffness levels at the age of 36, it is likely that any such bias was nondifferential. Moreover, if differential bias has occurred, it most probably resulted from over-reporting of healthy dietary habits by those with unhealthier lifestyles [53]. In both cases, the effect sizes as reported in the present study may have been underestimated. Thirdly, the aMED score is a relative measure that depends on the distribution of food intake in the study population as it is calculated on the basis of the median values of each of its components (except for alcohol consumption for which predefined cut-off levels were used). Although valid from an aetiological point of view, this does not enable direct extrapolation of our findings to other cohorts with other distributions. Similarly, we investigated a young, apparently healthy Caucasian population, and thus, extrapolation of the present findings to older, high-risk populations and/or those of other ethnicities should be performed with caution. Finally, given the observational design of the study and that arterial stiffness was only ascertained at the age of 36, our findings do not enable causality to be inferred.

In conclusion, we showed that adherence to a Mediterranean dietary pattern throughout the course of adolescence and early adulthood was associated with lower arterial stiffness in young adulthood, as well as with lower BP, BMI and total cholesterol during the 24-year study period. Promoting a Mediterranean dietary pattern amongst children, and throughout the course of adolescence and young adulthood, may constitute an important means of preventing adverse levels of CVD RFs and accelerated arterial stiffening in adulthood.

Acknowledgements

We thank all the participants of the AGAHLS. We also acknowledge all the members of the AGAHLS not listed as co-authors of this paper, for their time and expertise throughout the years. Part of the work was funded by a postdoctoral research grant (#2006T050) from the Netherlands Heart Foundation (to IF). The AGAHLS was supported by research grants from the Foundation for Educational Research, the Dutch Prevention Fund, the Netherlands Heart Foundation, the Dutch Ministry of Public Health, Well Being and Sport, the Dairy Foundation on Nutrition and Health, the Netherlands Olympic Committee/Netherlands Sports Federation, Heineken BV and the Scientific Board on Smoking and Health.

The study sponsors had no role in the study design, the collection, analysis or interpretation of the data, the writing of the report or the decision to submit the paper for publication.

Conflict of interest

None of the authors has any conflicts of interest to declare.

Ancillary