In our population of healthy French children, the cardiovascular risk factors most strongly associated with overweight or adiposity parameters were fasting plasma triglycerides, insulin concentrations and blood pressure (positively) and plasma HDL cholesterol concentration (negatively). After adjusting for subcutaneous fat, an association between these risk factors and abdominal fat distribution persisted for triglycerides and blood pressure in girls only, but mostly disappeared for insulin concentration.
A limitation of our study is that adiposity was assessed with anthropometric measurements, which are less accurate than Dual X-ray Absorptiometry (DEXA). However, DEXA is not applicable to large epidemiological studies. Moreover, waist circumference is considered as an appropriate marker of abdominal obesity in children (22
), and the sum of skinfolds, a good indicator of overall adiposity (35
). Furthermore, our sample is not representative of the French children population. Indeed, the included children lived in two towns in northern France. They underwent a special nutritional education program at school during five years and accepted to participate in a longitudinal epidemiological study (25
). The prevalence of overweight (including obesity) was around 10% whereas it was closer to 15% in contemporary samples of French children (16
). This relatively low prevalence is probably due to the low participation rate of the overweight children in the longitudinal FLVS II study compared to the FLVS I (8.0 vs
13.5 %), rather than to the effect of the nutritional education program. Indeed, we documented a rise in mean BMI and in the prevalence of obesity in girls in cross-sectional studies of 5–12 year-old children performed before and 8 years after the beginning of the program in the schools (15
). To our knowledge, no other descriptive report on cardiovascular risk factors in overweight French children has been published so far. Because of these limitations, we believe that our reported prevalences and correlations represent a minimal estimate of the true situation.
We can therefore conclude from our study that overweight, as defined by IOTF, is associated with biological signs of insulin resistance and its associated dyslipidemia. The prevalence of clustered risk factors is low in the overweight non-obese children. Only four children in the overall overweight non-obese population presented two or more of the metabolic syndrome risk factors (0.90%), three among the overweight children (7.7%). A German study of overweight children also found noticeable prevalences of dyslipidemia with figures even higher than in our population; 16% for high triglycerides levels (defined as > 1.7 mmol/L) and 17% for low HDL cholesterol (39
). However, these children were selected in obesity centers and had probably a mean BMI (not reported) higher than our overweight children.
WC was the adiposity parameter most strongly associated with cardiovascular risk factors in our study as in most other studies (19
). It has been widely demonstrated that WC is a good predictor of cardiovascular risk factors in adults (40
), and it is now used in the adult definitions of metabolic syndrome (43
). Several studies showed that, in children, WC is also a good anthropometric parameter to evaluate cardiovascular and metabolic risks (20
Another indicator of abdominal adiposity is waist-to-height ratio. Several studies reported that this was a better indicator of cardiovascular risk factors than BMI or WC itself (20
), but others have not found any difference (21
). In our study, the correlations between the cardiovascular risk factors and WC or waist-to-height ratio were roughly similar. However, total cholesterol in girls and LDL cholesterol in both genders were significantly associated with waist-to-height ratio and not with WC, similar to a previous report (48
When cardiovascular risk factors were associated with height, correlation pattern between WC or waist-to-height ratio and cardiovascular risk factors were different. This could be a strength of using waist-to-height ratio as a parameter less influenced by height. After taking age and Tanner stage into account, WC was still closely related to height in our population (r=0.30; p<0.0001). Moreover, WC was dependent on age and gender, whereas waist-to-height ratio was not, as shown in another study (21
). Hence, as suggested before (40
), and confirmed more recently (48
), one particular advantage of the waist-to-height ratio might be that effects, independent of age, sex and height could be identified. It may be possible to define a unique cutoff for all children to define high WC whatever the age and gender. For example, Ho et al. suggested that in adults, a simple message that one’s WC should not exceed half the stature could be recommended to the public (49
). This threshold has already been used in a population of young adult students (50
). This parameter should be validated in large cohorts of children as a simple measurement allowing the screening of children at risk for cardiovascular diseases.
It is still not clear whether the effects of abdominal fat on cardiovascular risk factors are independent of the effects of total body fat (51
). We evaluate in our study whether WC was associated to cardiovascular risk factors independently of subcutaneous fat mass level in children. Other studies have looked at the differential relationship of fat localization measures and cardiovascular risk factors. One such study found that trunk skinfolds predicted cardiovascular disease risk factors to the same extent as total fat mass by DXA, and in some cases independently of total fatness (23
). In another study, authors included both the percent body fat and fat distribution in a stepwise multiple linear regression analysis and found that fat distribution was a more important independent correlate of cardiovascular risk factors (high triglycerides, low HDL cholesterol, high systolic blood pressure, high left ventricular mass) than percent fat mass (52
These two studies did not perform analyses separately according to gender, whereas we consider that it was more appropriate. Indeed, boys and girls present quite different growth pattern in fat mass, lean mass and fat distribution especially during puberty. Even in our relatively small population, some interactions were significant between anthropometric parameters and gender in relation with some biological parameters (e.g. in the relation between waist-to-height ratio and triglycerides: p=0.02).
The associations found only in girls between waist circumference parameters and cardiovascular risk factors after taking the sum of skinfolds into account were primarily surprising. A central fat distribution is considered as a male specific pattern and an explanation for the high prevalence of cardiovascular disease in men compared to women (53
). We hypothesized that the fat distribution is more homogeneously centrally distributed in boys and its effect on cardiovascular risk factors would not be distinguished from that of subcutaneous fat mass. Conversely in girls, there is more variability in the fat distribution, from a gynoid to an android pattern, for a given level of total fat mass. However, we only found a slightly higher correlation between waist circumference and sum of four skinfolds in boys (r=0.83) than in girls (r=0.68).
One particular result from our study was that the very high positive correlations between anthropometric parameters and insulin concentration disappeared after the adjustment for subcutaneous fat mass, suggesting that subcutaneous fat mass has a role in the relation between abdominal fat distribution and hyperinsulinaemia in these children. Furthermore, plasma glucose was more correlated with anthropometric parameters in girls than in boys. This may be in agreement with the higher prevalence of type 2 diabetes in adolescent girls than in adolescent boys (54