INTRODUCTION

At present, we know that the pathological processes that lead to the development of atherosclerosis and coronary disease begin in childhood.1 These progress asymptomatically and generally without clinical manifestations until adulthood. Intravascular ultrasound imaging studies have detected atherosclerotic lesions in the coronary arteries of 17% of individuals aged less than 20 years,2 demonstrating the very early onset of this disease. These processes seem to be associated with cardiovascular risk factors in the first decades of life.3

The Four Provinces study4 (4P) was an epidemiological project investigating such risk factors in prepubertal children that may explain the incidence of coronary disease in adults. It was a comparative cross-sectional study which investigated cardiovascular risk factors in 1275 children aged 6-8 years living in 4 Spanish provinces. In 1993,4 there was a significant difference in adult coronary mortality in these provinces which comprised Cádiz and Murcia, with high mortality, and Madrid and Orense, with low mortality. The children were selected from clusters of schools through random sampling and were stratified by sex and socioeconomic level.4

We investigated diet (food and nutrient intake), anthropometric variables (weight, prevalence of obesity, etc), plasma lipid levels, vitamins, and insulin, and genetic factors clearly related to lipid levels, such as the apolipoprotein-E genotype (APOE). Fieldwork and data collection were conducted during 1998-2000 in the same season in each province. We summarize the study's main findings in this review article.

ANTHROPOMETRIC VARIABLES AND OBESITY

Analysis of the anthropometric variables showed that obesity was very prevalent among the children included in the study. Child obesity is reaching alarming proportions and is becoming a significant public health problem.5 Using the cut-off points proposed by Cole et al,6 the estimated prevalence of obesity for each age and sex was 9.4% in boys and 10.5% in girls.7 One of the most interesting results of the 4P study was the significant difference in the prevalence of obesity between the provinces studied. The percentage of children who had a Body Mass Index (BMI) above 20.1 was almost double in the high-mortality provinces (15.9% in Cádiz and 14.1% in Murcia) than in the low-mortality provinces (8% in Madrid and 9.4% in Orense).8 Associated with this, the average levels of glucose and triglycerides in the children in the high-mortality provinces were statistically higher than those in the children in the low-mortality provinces, which indicates that these children have different metabolic profiles.

When the metabolic changes associated with obesity in prepubertal children of both sexes were analyzed, we found that the obese children had higher plasma triglyceride and lower high-density lipoprotein cholesterol (HDL-C) levels than the nonobese children. No differences were found in plasma glucose and low-density lipoprotein cholesterol (LDL-C) levels.7 As expected, insulin levels and the homeostasis model assessment index were significantly higher in the obese children of both sexes than in the nonobese children; however, free fatty acid (FFA) levels were lower in the obese children than in the nonobese children, with a statistically significant difference in girls.7 Our data demonstrate that some of the metabolic features typical of obesity in adults (high triglyceride, insulin and homeostasis model assessment levels, and low HDL-C levels) are present in obese prepubertal children. However, the metabolic features highly characteristic of adult obesity, such as high glucose and FFA levels, are not present in obese prepubertal children. These data indicate that obesity-related metabolic disorders become established in different ways as a function of age and sex, apparently depending on the chronology of sexual maturation. This is supported by reports in the literature that show that the metabolic changes associated with obesity differ according to the sex and age of the children studied.9,10

BIOCHEMICAL VARIABLES

Table shows the mean values of glucose and lipid levels observed in the children studied. Plasma glucose and ApoA-1 levels were higher in the boys, and triglyceride, LDL-C, and ApoB levels were higher in the girls.11

Glucose and Triglycerides

Analysis of the biochemical variables showed that there were significant differences between provinces. In line with the lipid theory of atherosclerosis, differences in plasma lipid levels associated with ApoB (total cholesterol and LDL-C) should be expected; however, our data suggest otherwise. Interestingly, we found that plasma glucose, triglyceride and ApoA-1 levels in the high-mortality provinces were clearly higher than those in the low-mortality provinces.11 Some 21% and 11.7% of the child population in Cádiz and Murcia, respectively, had glucose values >100 mg/dL versus 8.9% and 3.7% in Orense and Madrid, respectively. Similarly, the percentage of children with triglyceride levels >100 mg/dL in the high-mortality provinces (13.5% in Cádiz and 13.6% Murcia) was considerably higher than the percentage in the low-mortality provinces (7.7% and 8.5% in Madrid and Orense, respectively). In the high-mortality provinces, we found that, in addition to the aforementioned, weight and BMI values were higher and there was a positive correlation between BMI and glucose and triglyceride levels. High BMI and high glucose and triglyceride levels are the most characteristic clinical features of insulin resistance. It is well-known that insulin resistance is a cardiovascular risk factor12 and its presence in children could be a clear marker of future coronary mortality. However, we did not find any differences in mean insulin values between provinces, due to the fact that insulin resistance is probably not yet established at this age. We found that the prepubertal girls had higher plasma insulin levels than the boys, which indicates that the girls had begun to be more resistant to this, although other manifestations of insulin resistance were not yet present at this age.13 We found higher levels of dehydroepiandrosterone sulfate (DHEA-S),14 the only hormone present in children of this age, which has a clear association with insulin values.

Total Cholesterol and Low-Density Lipoprotein Cholesterol

Mean total cholesterol levels were 181.1 mg/dL and 183.7 mg/dL in boys and girls, respectively (Table). The percentage of children with cholesterol levels above the 200 mg/dL limit was high in all the provinces11 and ranged between 19.2% in Madrid and 26.6% in Murcia (Figure 1A). The percentage of children with LDL-C levels >130 mg/dL was 13% in Madrid and 22% in the remaining provinces (Figure 1B). Previous studies15 performed in Madrid showed that hyperlipidemia was prevalent and this was confirmed in the 4P study.11 These findings can be generalized to Spain as a whole, as confirmed by abundant data by region16-19 and in a metaanalysis of childhood cholesterolemia.20

Figure 1. Percentage of children in each range in relation to total cholesterol, low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) in each province studied.

It may seem illogical to recommend a level of 200 mg/dL total cholesterol as optimum in adults given that practically a quarter of school-age children are above this limit. Lack of concern regarding this problem may be based on the results of some studies demonstrating that in 40% of children these high levels do not persist after sexual maturation.21,22 However, while accepting that some of these children reach standard lipid levels at puberty, we now know that atherosclerotic lesions begin to form early in life2 and may determine risk in adults.

High-Density Lipoprotein Cholesterol

We also observed high levels of HDL-C, however, together with these high values of total cholesterol and LDL-C. The level of HDL-C was 59 mg/dL in the prepubertal population.11 Some 38% of the children in Orense and 47%-49% of the children in the remaining provinces had levels higher than 60 mg/dL (Figure 1C). Very high levels of HDL-C have been observed in all the studies involving school-age children conducted in Spain.15,18,19 This situation may be based on dietary and/or genetic factors. Cholesterol and total fat, both saturated and monounsaturated, increase HDL-C levels. The results of our dietary pattern survey8 shows that we cannot attribute the cause of these high values only to olive oil consumption, since total and saturated fat consumption is also very high.

An important question is whether these HDL-C levels remain high in the adult population thereby contributing to low coronary mortality in Spain.23,24 As confirmed in the post-pubertal population, HDL-C levels decrease in boys and are high in girls,15 and will persist in the latter until menopause when the levels return to equilibrium.

However, while accepting that lipid levels can change in children during puberty and later in maturity, some studies have established an inverse correlation between HDL-C levels in children and the incidence of coronary disease in the countries studied.25 In this regard, HDL-C levels in Spanish children are higher than in children from other countries with greater coronary mortality.

Three different studies conducted by our group comparing plasma lipid levels in the school-age population in the Community of Madrid (Barrio de Pilar study26 in 1987, NICAM study15 in 1993, and children from Madrid in the 4P study11 in 1999) (Figure 2) showed that in little more than 1 decade cholesterol levels had risen by 15 mg/dL in children of 6-8 years.27 A decrease of 7 mg/dL28 has been observed in North American children of the same age. Studies on trends in coronary mortality in the USA showed that the decrease in cholesterol levels in the child and adult population was associated with a decrease in coronary heart disease mortality in adults.29 In Spain, total plasma cholesterol and LDL-C levels have gradually increased. However, recent analyses of HDL-C levels have demonstrated some very high stable values, which indicates that the low coronary mortality in Spain could be related to lipid values in Spanish children.28

Figure 2. Changes in plasma levels of total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) between 1987 and 1999 in prepubertal children (6 to 8 years) in the Community of Madrid. Taken from Garcés et al.27

Antioxidant Vitamins

We measured the plasma levels of fat-soluble antioxidant vitamins (vitamin E, carotenes, and retinol) in our prepubertal population. Without going into detail, we found that these correlate with plasma lipid levels and have very acceptable values, higher than those in other countries, and are similar to those found in populations of a similar age in Italy.30

DIET

We analyzed diet by administering a Food Consumption Frequency Questionnaire.8 We found that participants had a high-calorie diet, with a high intake of fats (47% of the energy intake) (Figure 3), especially saturated ones (17%).31 It was also a high-protein (17% of energy intake), low-carbohydrate (38% of energy intake)31 diet (Figure 3). As can seen in Figure 3, this diet is far from meeting the current recommendations on macronutrient intake.31 In contrast, micronutrient intake, in particular antioxidant vitamins and minerals, was not deficient and was more than that recommended in all cases except in the case of vitamin B6.31

Figure 3. Macronutrient intake as a percentage of dietary energy in children in the 4P study compared to the recommendations of the Spanish Society of Community Nutrition (SENC).

We observed that the mean energy intake was higher in the high-mortality provinces.8 This diet seems to contribute to the anthropometric and biochemical variables described in the children studied and is consistent with the high prevalence of overweight and obesity.

We assessed the association between saturated fat intake and the children's anthropometric and biochemical variables. Low saturated fat intake is associated with a better lipid profile (more HDL-C and low LDL-C).32 Furthermore, children with low saturated fat intake, despite consuming more calories, do not weigh more and have a more varied diet. These data suggest the following: a varied diet is associated with greater calorie intake, does not lead to obesity, and has a better biochemical profile.33 The hypothesis that the child who has a more varied diet, despite consuming more calories, does not gain weight and has a better risk profile, is very attractive.

Regarding nutrients in the diet of Spanish children, the most important source of carbohydrates was bread; that of proteins, whole milk; that of fats, olive oil; and that of saturated fat, fried potatoes.34 This food pattern lies between the typical Mediterranean diet and that in Anglo-Saxon countries since, although consumption of fruit, vegetables, fish, and olive oil is quite high, so is consumption of meat and milk derivatives, pastry products, snacks, precooked meals, and carbonated drinks.31 This pattern probably stems from economic development in the Mediterranean region which has led to an influx of high-density energy foods and fast foods that include animal products while replacing cereals. If this food pattern is maintained until adulthood, it could lead to greater coronary risk than is current and contribute in the future to Spain losing its position in Europe regarding its relatively low levels of coronary mortality.

GENETIC DETERMINANTS

Apolipoprotein E Genotype

The 4P study also analyzed how APOE gene polymorphism (APOE genotypes) affects lipid variables. Among the genetic alterations associated with lipid metabolism, the APOE genotype has gained particular prominence within the problem of atherosclerosis, given its key role in lipid metabolism, and because it is one of the genetic determinants that has been consistently associated with variations in lipid levels in the population and risk of coronary disease.

This is a polymorphic gene with 3 codominant alleles (ε2, ε3, ε4) that give rise to 6 genotypes: ε2ε2, ε2ε3, ε3ε3, ε3ε4, ε4ε4 and ε2ε4. Several population studies (epidemiological studies35,36 and case-control studies37) have shown that the ε4 allele, and in particular the ε3ε4 genotype, are strongly associated with coronary disease. The relationship between the APOE genotype and atherosclerosis is basically due to its effect on lipid levels,38 although it can have other effects.

Results from the study population showed that the ε3ε4 genotype is associated with significantly higher values of total cholesterol (Figure 4), LDL-C and ApoB.39 However, we found that these polymorphisms had different effects on these lipid variables depending on sex39 and birth weight,40 indicating that these effects were influenced by hormones. In fact, we have shown that DHEA-S levels modify the influence of the APOE genotypes on plasma lipid levels.41 Based on this data, we can conclude that the lipid levels in the study population are the result of complex interactions between diet, genes, and hormone levels.

Figure 4. Plasma levels of total cholesterol in prepubertal children in relation to the APOE genotype.

When we investigated APOE genotype frequency we found a low prevalence of the ε4 allele (10.1%) and the ε3ε4 genotype (16.8%).42 This low prevalence makes biological sense in a Mediterranean country with low coronary mortality and is similar to that of other low-mortality Mediterranean countries, as well as being lower than in North European countries, where coronary mortality is greater.35,36 Interestingly, however, the distribution of genotypes by province varies considerably, with a very similar pattern in Cádiz and Madrid (16% of ε3ε4) but with very different prevalences in Orense and Murcia (12.9% and 22.5%, respectively) (Figure 5).42 The lowest frequency of the ε4 allele and the ε3ε4 genotype was found in Galicia (where Orense is located), which is similar to regions with a Mediterranean environment (Figure 5), which fits in with the low coronary mortality in this region. Galicia43 is geographically isolated, with high emigration and low immigration, and has maintained its linguistic, economic, and cultural identity. Galicia has low genetic diversity. Murcia lies at the opposite extreme,44 where the ε4 allele and ε3ε4 genotype are highly prevalent, similar to that in countries with high mortality (Figure 5), which again fits with the very high coronary mortality found there. The identical distribution of APOE genotypes in Madrid and Cádiz poses a very interesting problem. Both areas have experienced multiple civilizations throughout history and this population mixture has gradually produced identical genetic frequencies. This explanation is strengthened by observing that 16% of the prevalence of the ε3ε4 genotype is the mean of the total group under study. Spain is not homogeneous regarding the distribution of the APOE genotype. The ε4 allele has low expression in this Mediterranean country, but it has areas where there is high expression of this allele and a high incidence of coronary mortality, such as Murcia, and areas of low prevalence of the allele and low coronary mortality, such as Orense. We can conclude that this genetic determinant is associated with coronary mortality in relatively isolated populations and loses its predictive value in genetically heterogeneous populations.

Fig. 5. Frequency of the ε3ε4 genotype by province in the Four Provinces study

Our results have practical importance. Despite protective factors in children in the Spanish population, such as high HDL-C levels or the generally low prevalence of the ε3ε4 genotype, which is clearly related to cardiovascular risk, we should control the risk factors obviously related to excess weight (high levels of glucose, triglyceride, and cholesterol, etc), through educational interventions aimed at improving the current diet of children. The prevention of obesity in children plus physical exercise will help to reduce the alterations associated with excess weight, and prevent both the onset of adult obesity and the increase in mortality coronary in Spain. Thus, our findings provide new arguments for initiating coronary prevention programs from childhood onwards.

THE FOUR PROVINCES STUDY RESEARCHERS

M.A. Lasunción, H. Ortega, D. Gómez-Coronado (Servicio de Bioquímica, Hospital Ramón y Cajal, Madrid); J.M. Martín Moreno, L. Gorgojo (Universidad de Valencia); M.A. Royo (Instituto Carlos III); F. Rodríguez Artalejo (Departamento de Medicina Preventiva, UAM); A. Gil (Universidad Rey Juan Carlos, Madrid); O. Fernández (Complejo Hospitalario de Orense); A. Mangas and A. Macías (Universidad de Cádiz) and J. Fernández Pardo (Hospital General Universitario de Murcia); M. Benavente, R. Rubio, J.L. del Barrio, E. Viturro, B. Cano, I. de Oya, L. López Simón (Laboratorio de Lípidos, Fundación Jiménez Díaz).

ABBREVIATIONS
FFA: free fatty acids
DHEA-S: dehydroepiandrosterone sulfate
HDL-C: high-density lipoprotein cholesterol
LDL-C: low-density lipoprotein cholesterol
BMI: body mass index

The researchers taking part in the Four Provinces study are listed at the end of the article.

This study received financial support from the Consejo Oleícola Internacional, Fundación Pedro Barrié de la Maza, Fundación Eugenio Rodríguez Pascual, Fundación Ramón Areces, Comunidad de Madrid (08.4/0006/1997, 08.4/0012.1/2003), Fondo de Investigación Sanitaria (FIS 02/3104, PI020994).

Correspondence: Prof. M. de Oya.
Unidad de Lípidos. Fundación Jiménez Díaz.
Avda. Reyes Católicos, 2. 28040 Madrid. España.
E-mail: moya@fjd.es