A Prospective Study on the Prevalence of Metabolic Syndrome Among Healthy French Families: Two cardiovascular risk factors (HDL cholesterol and tumor necrosis factor-α) are revealed in the offspring of parents with metabolic syndrome Free

The STANISLAS cohort is a 10-year family study. Its main objective is to investigate the impact of genetic and environmental factors on the variability of cardiovascular risk factors (4,14). Recruitment of families from Northeastern France started in 1993 (t₀) by an invitation for a health examination in the Center for Preventive Medicine at Vandoeuvre-lès-Nancy. At this time, 1,006 presumably healthy families (4,295 individuals) were recruited and responded to the following selection criteria: families of French origin, consisting of two parents and at least two biological children aged 4 or more, with members of the family free from serious and/or chronic illnesses. All participants gave their written informed consent. This cohort was approved by the Local Ethics Committee of Nancy, France. In 1998 (t₊₅), the participants came back for the second health screening, with a participation rate of 75%.

In this study, we investigated 371 nuclear families present at both visits, with at least one child aged ≤20 years. They consisted of 371 fathers, 371 mothers, 282 sons, and 313 daughters. Children >20 years old were excluded to avoid the influence of age on the studied variables. The subjects were present at t₀ and t₊₅ and were selected according to the following criteria: glucose ≤8 mmol/l, apolipoprotein (apo) E ≤200 mg/l, cholesterol or triglycerides ≤10 mmol/l, orosomucoid or haptoglobulin ≤3 g/l, C-reactive proteins ≤30 mg/l, and aspartate aminotransferase or alanine aminotransferase activities ≤200 units/l. At a second time point, we stratified our study and considered only individuals who did not take the following drug treatments: antiangina drugs, antihypertensive drugs, antiarythmics, cardiotonics and cardiac analeptics, vasodilatators, antidiabetic drugs, lipid-lowering drugs, diuretics, and anti-inflammatory drugs. The population was then composed of 256 fathers, 296 mothers, 265 sons, and 286 daughters.

According to NCEP ATP criteria (15), a participant has metabolic syndrome if he or she has three or more of the following criteria: waist circumference >102 cm in men and >88 cm in women; triglyceride levels ≥1.70 mmol/l; HDL cholesterol concentration <1.04 mmol/l in men and <1.30 mmol/l in women; blood pressure ≥130/85 mmHg; and fasting glucose value ≥6.1 mmol/l. Metabolic syndrome was calculated according to the following three different interpretations of the NCEP ATP definition (D): D1 does not take drugs into account; D2 includes antihypertensive drugs; and D3 includes all drugs that treat metabolic syndrome-related abnormalities (i.e., antihypertensive, lipid-lowering, and antidiabetic drugs).

Data dealing with tobacco, alcohol, weight at birth, size at birth, current use of medications, oral contraceptive intake, and hormonal replacement therapy were assessed by a questionnaire on the living habits of each participant. Smoking status was described as nonsmokers, smokers, and ex-smokers. Alcohol consumption was expressed in grams per day. BMI was calculated according to the Quetelet’s formula of weight divided by height (kilograms per meters squared) (2). Blood pressure was calculated as the mean of three measurements taken under standardized conditions with a sphygmomanometer, with the subject in a supine position.

Blood samples were taken from subjects after an overnight fast. Venous blood was collected without anticoagulant in Vacutainer tubes (Becton Dickinson, Rutherford, NJ) containing a gel for serum separation. These samples were centrifuged at 1,000g for 15 min at room temperature and were analyzed for all constituents except apoCIII and insulin concentrations. Serum aliquots were stored in 250-μl straws in liquid nitrogen at −196°C before being analyzed for these two last constituents.

HDL cholesterol was assayed after precipitation of other lipoproteins with phosphotungstate-magnesium on the COBAS-Mira analyzer (Roche Diagnostics, Basel, Switzerland), and total cholesterol, as well as glucose and triglycerides, was assessed enzymatically with an AU 640 automated device (Olympus, Rungis, France). Current values of fasting serum insulin concentration were evaluated by a Microparticular Enzymatic ImmunoAssay on an IMx Abbott automated device (Abbott Laboratories, Abbott Park, IL). The method and reference values for the STANISLAS cohort have already been detailed elsewhere (16). LDL cholesterol was calculated from Friedewald’s formula. Serum apoA1 and apoB as well as haptoglobulin, orosomucoid, and C-reactive protein concentrations were measured by immunonephelometry with a Behring Nephelemeter Analyzer II (Dade Behring, Marburg, Germany). Serum apoE and apoCIII concentrations were assessed by immunoturbidimetry with a COBAS-Mira analyzer with Daïchi reagents. Lastly, interleukin-6 and tumor necrosis factor-α (TNF-α) were measured in plasma with commercially available enzyme-linked immunosorbent assay (R&D System, Abington, U.K.) in samples stored in liquid nitrogen until use.

Statistical analyses were performed with the SAS package program version 8.2 (SAS Institute, Cary, NC). Data for BMI, waist circumference, triglycerides, LDL lipoprotein cholesterol, HDL cholesterol, apoE, insulin, haptoglobin, orosomucoid, interleukin-6, and TNF-α concentrations were log-transformed to reduce skewing.

Descriptive statistics were computed with means and SD for continuous measurements. Crude prevalence was determined at t₀ and t₊₅ as the frequency of men and women presenting metabolic syndrome as defined by the NCEP ATP.

Biological parameter values were compared among offspring of parents who had or currently have metabolic syndrome versus age- and sex-matched control subjects (children with parents without metabolic syndrome) by ANOVA tests. These control subjects were randomly chosen. Because individuals within a family are not independent, ANOVA tests were performed by an SAS GENMOD procedure with a family factor as a repeated statement. GENMOD is based on the generalized estimating equation (17). P < 0.05 was regarded as statistically significant.

Clinical and metabolic characteristics of our studied population at t₀ are summarized in Table 1. As shown in Table 2, the prevalence of metabolic syndrome increased over the 5-year interval (t₀ to t₊₅) and was higher in men than in women. It rose from 5.9 to 7.2% in men and from 2.1 to 5.4% in women (with D1). Moreover, 40.9% of men and 50% of women with the diagnosis of metabolic syndrome at t₀ also had the diagnosis of metabolic syndrome at t₊₅. Global incidence at 5 years for metabolic syndrome was 5.2 and 4.4% in men and women, respectively. Compared with the D2 or D3 definition, prevalence was higher when the D1 definition was used. In addition, we observed that if we excluded individuals under drug treatment from our study, the prevalence was reduced to 3.5% at t₀ and to 4.7% at t₊₅ in men and to 1% at t₀ and to 4.4% at t₊₅ in women.

Figure 1 presents the distribution of metabolic syndrome criteria among fathers and mothers according to D1, D2, and D3 definitions and in fathers and mothers without any drug treatment.

Our data showed that whichever NCEP ATP definition was used, the characteristics mainly found at both t₀ and t₊₅ in the individuals affected by metabolic syndrome were elevated blood pressure and hypertriglyceridemia in men and abdominal obesity (increased waist circumference), low HDL cholesterol, and hypertension in women. For instance, when D1 was applied to men at t₀, 90.9% (92.6% at t₊₅) had elevated blood pressure and 95.5% had hypertriglyceridemia. Of women at t₀, 87.5% (75% at t₊₅) had low HDL cholesterol levels, 75% (85% at t₊₅) had abdominal obesity, and 100% had elevated blood pressure (70% at t₊₅). Hyperglycemia was the less frequent risk factor for metabolic syndrome in both men and women (data not shown).

The investigation of potential clinical (Table 3) and biological (Table 4) risk factors for CVD in the children of parents who had or currently have metabolic syndrome defined by D1 gave the following results: TNF-α concentrations were significantly higher (P = 0.009), whereas HDL cholesterol at t₀ (P = 0.03) and apoE concentrations at t₀ (P = 0.03) and t₊₅ (P = 0.05) were significantly lower in children of parents with metabolic syndrome than in age- and sex-matched control subjects. Additionally, insulin concentration was higher in children of parents with metabolic syndrome than in control subjects (P = 0.05). The same comparison study was performed by applying the D2 and D3 definitions. Only significant results with D3 definitions are shown (Tables 3 and 4). Moreover, we have tested the influence of alcohol, exercise, and smoking on the concentration of HDL cholesterol. No influence of exercise and smoking was found. On the other hand, we observed a significant interaction between alcohol consumption and parental metabolic syndrome on HDL cholesterol concentration with D1 (P = 0.009), D2 (P = 0.004), and D3 (P = 0.0015). This could be tested only at t₊₅ because the number of children who drank at t₀ was not sufficient (Table 4). Our result showed that, given a daily quantity of alcohol consumed, a child with a parent who has metabolic syndrome has an HDL cholesterol concentration lower than that of a matched control subject. In any case, children who drink alcohol have higher HDL cholesterol concentrations than nondrinkers.

The CVD markers that showed statistical significance when either D1, D2, or D3 definitions were applied are summarized in Fig. 2. In this graphical representation, each circle corresponds to a version of metabolic syndrome definition (D1, D2, or D3 definitions). The intersection of these three circles permitted us to see that among all factors, HDL cholesterol, TNF-α, and the interaction between alcohol consumption and parental metabolic syndrome on HDL cholesterol were significant among the three comparison studies.

In this article, we give a first estimate of the prevalence of metabolic syndrome in the STANISLAS cohort according to NCEP ATP criteria. We took care to compare our results only with results from the literature that were produced according to NCEP ATP criteria because some discrepancies have been observed depending on the definition chosen. The NCEP ATP definition of metabolic syndrome was chosen rather than the World Health Organization or European Group for the Study of Insulin Resistance criteria because it relies on variables that are easily ascertainable by physicians and thus is convenient operationally (18).

The NHANES III study of metabolic syndrome involving 4,265 men and 4,549 women observed metabolic syndrome frequencies of 24.0 and 23.4% in American men and women, respectively (2). Among Framingham Offspring Study white subjects, the NCEP ATP prevalence was 24% and among San Antonio Heart Study white subjects, it was 23% (6). Therefore, the American prevalence is globally sixfold higher than in the STANISLAS cohort at t₀. However, the gap is progressively reduced because at t₊₅ the American prevalence is about fourfold higher. Importantly, even when assessing healthy adults without medication, we observed that metabolic syndrome was incontestably present.

The French DESIR (Data from an Epidemiological Study on the Insulin Resistance Syndrome) study, which included volunteers from central-western France, observed that at baseline, 10% of men and 7% of women had metabolic syndrome compared with 5.9% of men and 2.1% of women in our population (19). Age ranges were comparable between DESIR and our study.

We found that 40.9% of men and 50% of women (D1 definition) had this syndrome at baseline and at 5 years. In the French DESIR study, only 12 and 8%, respectively, had metabolic syndrome at baseline and at 3 years (19).

Age-adjusted characteristics of the different populations described above are compared with those of the STANISLAS cohort, shown in Table 5. The differences in these characteristics can be explained by differences in age, geographical regions, and selection methods. Here, our results are in agreement with those obtained in DESIR.

Our results confirm the well-established increase in metabolic syndrome with age and that metabolic syndrome is generally more prevalent in men than in women (21). In our study, metabolic syndrome prevalence has increased by ∼3 points in 5 years (Table 2).

In our population, we found that elevated blood pressure and hypertriglyceridemia in men and abdominal obesity and hypertension in women were the more frequent criteria. Hypertension was also the criteria more present in the French DESIR (19) and MONICA (Monitoring Cardiovascular Disease) (21) studies. Central obesity is the first criteria in the U.S. (2).

We observed that HDL cholesterol and TNF-α concentrations were the two parameters that were significantly different among children of parents with metabolic syndrome and control children when we compared results obtained with three different interpretations of the NCEP ATP definition involving or not involving some specific medication related to the criteria considered for diagnosing metabolic syndrome. To our knowledge, this is the first time that such an investigation was performed in a French population. The only investigation we found presenting a similar scheme was performed by Pankow et al. in an American population (22). However, this study did not analyze as many lipid parameters as we did and did not investigate inflammatory molecules at all. These authors found a significantly higher insulin concentration and higher parameters for obesity (including waist circumference and waist-to-hip ratio) in children of parents with metabolic syndrome. Similar associations, although not consistently significant, were observed in our study using the three definitions D1, D2, and D3 (see Fig. 2). Pankow et al. found no significant differences in the lipid levels that they investigated, including HDL cholesterol. On the contrary, in our study, HDL cholesterol was lower and TNF-α was higher in children of parents with metabolic syndrome than in control subjects, and this was consistently found by applying the three definitions. Low HDL cholesterol is known as a risk factor for coronary artery disease in adults (23,24). It is considered to be an important parameter of metabolic syndrome in adults because it is included in the three major definitions proposed at this time (15,25,26). In our study, it appears to be an important factor to follow during childhood. Additionally, we reported an interaction by using D1, D2, and D3 definitions, consisting of a significantly less important alcohol-related HDL cholesterol increase in children of parents with metabolic syndrome compared with control subjects. It is well known that alcohol intake leads to an increase in HDL cholesterol (27,28). Moreover, it has been suggested that metabolic syndrome is negatively associated with light alcohol consumption in middle-aged Korean adults (29). However, the mechanism by which parental metabolic syndrome acts on HDL cholesterol homeostasis in children who drink alcohol remains to be elucidated.

The proinflammatory cytokine TNF-α has been implicated in insulin resistance (30). It has also been suggested to be a cardiovascular risk factor (31). Levels fourfold higher were observed in insulin-resistant obese humans compared with healthy humans (32). Obese adolescents have been shown to exhibit higher TNF-α and its soluble receptor concentrations than nonobese ones, and TNF-α concentration has been positively correlated with several components of metabolic syndrome in obese adolescents (33). These results suggest that this cytokine could be an early marker of the inflammatory state predisposing to metabolic syndrome development during childhood. Therefore, low HDL cholesterol and TNF-α may play an important role in the development of metabolic syndrome in childhood, at least in our population. These two parameters deserve specific attention as targets for prevention and treatment.

We are aware of some limitations to our study. Although the STANISLAS cohort consists of individuals who were apparently in good health at their inclusion in the study, we cannot exclude the possibility that silent CHD could have been present in some subjects; however, none of the tests and questionnaires collected supported such occurrences. Moreover, as indicated by its inclusion criteria (see research design and methods), the STANISLAS cohort is not a group representative of the French population. All the individuals enrolled are insured by French Social Security (covering >80% of the French population). They were invited to the Centre for Preventive Medicine for a health check-up at Vandoeuvre-lès-Nancy (France) and agreed to participate in a 10-year study and to come back for a medical examination every 5 years.

However, describing metabolic syndrome in such a group is interesting because it shows that this disorder can be isolated even in a healthy cohort and that more significant differences appear in the offspring of presumably healthy individuals affected by metabolic syndrome.

Our results suggest that metabolic syndrome occurs insidiously, even in healthy individuals. Moreover, our study results underline the importance of early management of children of parents with metabolic syndrome to prevent obesity, which is more and more frequent in childhood and is clearly a serious aggravating circumstance for development of metabolic syndrome, type 2 diabetes, and CVD in adulthood.

The NCEP ATP is perhaps not the best definition for the STANISLAS cohort. Nevertheless, it has the advantage of enabling the detection of metabolic syndrome, even in a healthy cohort. At present, there are no well-accepted criteria for diagnosing metabolic syndrome. A more age-sex-ethnic geographic definition is required, including guidelines that would be recognized worldwide. In this context, the understanding of metabolic syndrome mechanisms in a familial healthy population would be of interest. Therefore, one purpose of our cohort should be to further the establishment of metabolic syndrome reference values for the French and European populations.

This study was supported by grants from Abbott, Dade Behring, Roche Molecular Systems, and Daïchi Pure Chemicals.

The authors are grateful to Dr. Daniel Lambert, Dr. Monique Vincent-Viry, Suzanne Droesch, and all the workers from the Center for Preventive Medicine for their help in collecting data. We also thank the families of the STANISLAS cohort for their participation in the study.