点击显示 收起
1 From the Department of Pediatrics, San Paolo Hospital, Milan, Italy, and the Departments of Pharmacologic Sciences and Informative Systems, University of Milan, Italy.
2 Presented at the symposium Fat Intake During Childhood, held in Houston, June 8–9, 1998. 3 Address reprint requests to C Agostoni, Department of Pediatrics, San Paolo Hospital, 8 Via A di Rudinì I-20142, Milano, Italy. E-mail: agostoc{at}tin.it.
ABSTRACT
The fat intake of Italian infants has peculiar characteristics that begin quite early because their mothers' milk has a monounsaturated fat content (45%) at the upper limit of the values found in Europe. Comparison studies in breast-fed and formula-fed infants were conducted to evaluate growth and developmental correlates and differences in fat intakes in the early months of life. Breast-fed infants have higher blood lipid concentrations at 4 mo of age than do formula-fed infants. The addition of long-chain polyunsaturated fatty acids (LCPUFAs) and cholesterol to formulas for term infants may affect concentrations of circulating blood lipids as well as the LCPUFA composition of the lipids during the breast-feeding period. The addition of LCPUFAs does not seem to affect the growth rate of formula-fed infants. Although an initial benefit of LCPUFA feeding on eye-hand coordination was observed, this effect was not sustained; by 24 mo, different feeding groups had similar developmental scores. Other peculiarities of the Italian experience are presented, including body weights from infancy to early childhood in 147 children, the nutrient densities of different diets in Italian schoolchildren, and the effects of nutritional education on dietary intakes. The diets of these children were high in animal protein and supplied 30–35% of energy from fats throughout childhood. Both the dietary protein intakes at 1 y of age and parental body mass indexes were associated with 5-y body mass index values. Classroom education may be useful to lower the plasma lipid concentrations in healthy, primary school–age children. It is not known whether this early modification can be maintained and whether it influences the later development of cardiovascular disorders.
Key Words: Children • breast-feeding • formula-feeding • growth • development • blood lipids • dietary fats • cholesterol • dietary proteins • Italy
INTRODUCTION
The biochemical and functional effects of different intakes of nutrients, particularly fats, in infants and children has been the object of several studies conducted by our group in recent years. Most studies concerned infants from the referring area of the San Paolo Hospital. The main purpose was to investigate the effects of dietary patterns early in life on lipid biochemical status and the correlates with growth and developmental indexes at later assessments. All the studies, in agreement with the 1983 revision of the Helsinki Declaration, were approved by the Department of Pediatrics Ethical Committee. The data presented focus on the quality of fats supplied to breast-fed and formula-fed infants, the amounts and quality of nutrients supplied at the end of the weaning process, the dietary habits of Italian schoolchildren, and the effects of intervention studies.
BREAST-FED AND FORMULA-FED ITALIAN INFANTS: FAT SUPPLY, BLOOD LIPID CONCENTRATIONS, GROWTH, AND NEURODEVELOPMENTAL CORRELATES
Differences in blood lipid concentraions (1, 2), growth variables (3–5), and neurodevelopmental abilities (6–10) between breast-fed and formula-fed infants were reported in several studies. It has been hypothesized that the different composition of fat supplied to breast-fed and formula-fed subjects could partly account for these differences. Human milk fats include both cholesterol (2.6–3.9 mmol/L, or 100–150 mg/L) and long-chain polyunsaturated fatty acids (LCPUFAs), mainly consisting of 0.5–0.6% arachidonic acid (20:4n-6) and 0.2–0.3% docosahexaenoic acid (22:6n-3) (11). The dietary supply of LCPUFAs is associated with differences in the fatty acid composition of the brain (12) and modifications in insulin sensitivity (13). We conducted 2 studies to determine the fat composition of human milk in lactating women from our area throughout prolonged lactation and to assess the blood lipid and fatty acid concentrations, the standardized growth indexes, and the developmental attainment in breast-fed infants, standard formula-fed infants, and infants fed a formula supplemented with cholesterol and LCPUFAs.
FAT COMPOSITION OF HUMAN MILK
Subjects and methods
Ninety-five mothers who gave birth to healthy, full-term infants (37–42 wk gestational age) in our hospital in 1995 and who fully breast-fed for 3 mo were eligible for the study. The sample size was calculated to include 10 mothers still breast-feeding when the offspring reached 1 y of age on the basis of prevalence data in our area (14). The design provided consecutive 24-h collections of breast milk during the first day of breast-feeding (colostrum), at the end of the first month, and at 3, 6, 9, and 12 mo. All mothers were encouraged to fully breast-feed for as long as possible and were instructed to express breast milk into sterile vials at the end of each feeding. [The fatty acid composition of human milk does not change during a single feeding over a 24-h period (15, 16).] Breast-milk samples were immediately frozen after collection. Aliquots from each meal on the days of collection were then pooled and analyzed for total fatty acids. Milk total lipids were extracted with chloroform-methanol (2:1, by vol) containing butyl hydroxytoluene as an antioxidant according to the method of Folch et al (17). Fatty acids were analyzed as methyl esters by gas chromatography with a capillary column (model SP-2560,100 m long, 0.25 mm internal diameter, 0.2 mm film thickness; Supelco, Bellefonte, PA) with a programmed temperature from 150 to 220°C). Statistical analysis included repeated-measures analysis of variance (ANOVA) and nonparametric tests for between-group comparisons. SPSS for WINDOWS (SPSS Inc, Chicago) was used for the analyses.
Results and discussion
Of 95 women recruited at delivery, we used data from 44 mothers who fully breast-fed for 3 mo for a preliminary analysis. The remaining 51 (54%) mothers stopped breast-feeding when their infants were 0–3 mo of age. Ten mothers were still breast-feeding at 12 mo (Table 1). Milk fatty acid composition at 3 mo was not significantly different (Kruskal-Wallis test) between the women who stopped breast-feeding before 12 mo and those who breast-fed for 12 mo. Compared with European data (18), we found a higher percentage of monounsaturated fatty acids and a lower concentration of -linolenic acid (18:3n-3). Total saturated fatty acids showed an increasing trend, whereas monounsaturated fatty acids decreased throughout the 12 mo of lactation. Although the concentrations of the 18-carbon precursors (ie, linoleic acid and -linolenic acid) remained constant, the highest concentrations of the major LCPUFAs were in the colostrum, which abruptly decreased at 1 mo to concentrations that were maintained until 12 mo. Similar results were found by others in follow-up studies on milk composition (19, 20). The high concentrations of monounsaturated fatty acids (75% of fats) in the milk samples were explained by the widespread use of olive oil, which is rich in oleic acid (18:1n-9), in the Italian diet (21).
View this table:
TABLE 1. Fat composition of human milk throughout 12 mo of lactation in a sample of women from Milan who breast-fed for 3 mo
LIPID AND FATTY ACID STATUS AT 4 MO OF AGE IN BREAST-FED AND FORMULA-FED INFANTS: GROWTH AND NEURODEVELOPMENTAL CORRELATES
Subjects and methods
Sixty infants born in our clinic between September 1992 and August 1993 whose mothers chose not to breast-feed were randomly assigned at birth to receive either a standard infant formula (SF group; n = 31) or a formula enriched in LCPUFAs and cholesterol (EF group; n = 29). The formula composition was described previously (22). As a reference group, we studied 30 infants born in the same period who were fully breast-fed for 4 mo [human milk (HM) group; n = 30]; this group had a median breast-feeding duration of 8 mo (range: 4–14 mo). At 4 mo, 59 infants (HM group, n = 15; SF group; n = 23; EF group; n = 21) whose parents gave informed consent were assessed for their blood lipid status with enzymatic methods and for their fatty acid status with high-resolution capillary gas chromatography of the blood lipid extracts. LDL-cholesterol concentrations were determined with Friedewald's formula (23).
Weight and length were measured by trained personnel with standardized techniques at birth and at 1, 2, 3, 4, 6, 9, and 12 mo of age. The z scores of anthropometric indexes (WAZ, weight-for-age; LAZ, length-for-age; and WLZ, weight-for-length) were calculated by using the ANTHRO pediatric anthropometry software program (version 1.01, 1990; Centers for Disease Control and Prevention, Atlanta). The z scores represent the distance in SD units from the normative reference medians for age and sex of the National Center for Health Statistics and the World Health Organization .
Neurodevelopmental performance was assessed with Brunet-Lezine's scale (24) at 4 (n = 86), 12 (n = 81), and 24 (n = 81) mo of age, allowing for the classification of the 4 developmental areas covered by the test (posture and gross motor function, eye-hand and fine-motor coordination, language, and social reactions). Scores on Brunet-Lezine's neurodevelopmental test rate the performance on items of graded difficulty. Item batches of 10 are assigned to 1 mo of real age. Real age is defined as the postnatal age in weeks corrected by subtracting 40 wk from the postconceptional age. Developmental age is calculated as the sum of credits earned with the performance of each item in the battery. A credit counts as 1 and represents 3 d of life. The developmental quotient (DQ) is then calculated according to the following formula:
DIETARY FAT INTAKES AT WEANING AND CORRELATES IN LATER CHILDHOOD
In a nutritional survey conducted in 1989 of healthy infants born in Milan (36), it was found that the average diet at 1 y of age was adequate in energy but high in protein (ratio of animal to vegetable protein: 2:1). The average daily intake of protein was 5 g/kg body wt. Fat accounted for 30% of total energy (14% saturated fat, 10% monounsaturated fat, and 3% polyunsaturated fat). A study of toddlers from central Italy found a dietary lipid content that was slightly higher (33% of energy) (37). A long-term follow-up study began in 1991 to address in a healthy infant population from our area the effects of the early dietary pattern on later fatness development and lipemic indexes.
Subjects and methods
A sample of 164 healthy singleton infants (37–42 wk gestational age, weight at birth 2500 g) was recruited and assessed at 1, 5, and 8 y and during adolescence. The infants had no disease or malformation and were randomly selected among those born in our maternity ward between July and December 1991. Data from the 5-y assessment are now available. Dietary habits at 1 and 5 y of age were assessed with a food-frequency questionnaire and 24-h dietary recalls (38, 39). The database of the Italian National Institute of Nutrition was used to quantify nutrient intakes (21). Subjects did not follow any dietary advice and were free to regulate their dietary choices with no restriction. Serum lipid concentrations were determined as described above. Anthropometric measurements were made with standard techniques. The body mass index (BMI; in kg/m2) was calculated for children at each visit and at the 1-y control for parents. The midparental BMI (average of the mother's and father's BMI) was also calculated. Informed consent was obtained from parents when their infants were 8 mo old. The sample size was calculated on the basis of a 12.5% expected prevalence of overweight (40) to include 20 subjects overweight at 5 y of age (BMI > 90th centile of reference age- and sex-adjusted curves) (41). Children were categorized into 3 categories on the basis of their 5-y BMIs (BMI 75th centile, 75th centile < BMI 90th centile, BMI > 90th centile). Statistical analysis was performed by ANOVA and a post hoc Bonferroni's test.
Results and discussion
Follow-up data were obtained for 147 subjects (89%). The prevalence of a BMI greater than the 90th centile at 5 y of age was 23%, which was higher than expected. The subjects consumed a mildly hyperenergetic diet at 5 y and had a very high intake of protein at 1 y (21% of energy in the whole population). The age-corrected Italian recommended dietary allowance of protein is 8–10% of energy (42). The daily fat intake was slightly <35% of energy at ages 1 and 5 y.
An association was found between the 5-y BMI and parental BMIs (Table 5). Subjects in the upper centiles of the 5-y BMI had a significantly higher protein intake at the 1-y assessment, even after adjustment for the midparental BMI. No associations were found between any blood lipid concentrations at 5 y of age and the nutritional indexes at 1 y. The lowest quartile of blood non-HDL cholesterol was weakly associated with lower intakes of proteins (P = 0.09), saturated fat (P = 0.06), polyunsaturated fat (P = 0.04), and cholesterol (P = 0.02), and with higher intakes of carbohydrate (P = 0.02) at the 5-y assessment (data not shown).
View this table:
TABLE 5. Daily energy, nutrient, and cholesterol intakes at 1 and 5 y of age and parental BMI in different BMI categories of children at 5 y of age1
Although a genetic component of obesity has been shown (43), findings with respect to dietary intake are often controversial. One report showed an association between high dietary protein intakes in the first years of life and later development of obesity (44). Our findings seem to confirm that, besides parental overweight, an early high protein intake is a predictor of the development of obesity in children. A high protein intake could stimulate the production of insulin-like growth factors (45), particularly of insulin-like growth factor I, which is normally low up to the age of 8 y (46). The association of low fat and protein intakes at 5 y with low blood non-HDL-cholesterol concentrations requires further investigation.
FAT INTAKES IN ITALIAN SCHOOLCHILDREN
Local dietary habits in Italian schoolchildren were investigated by others (47–49), but no data concerning the entire Italian school population were available until a few years ago. Below is a report on a recently published (50) nationwide survey of nutritional patterns in this population.
Subjects and methods
A sample of 50000 schoolchildren aged 7–10 y was selected from the major regional areas of Italy to represent socioeconomic differences. The sample was assessed with use of a food-frequency questionnaire (39). The response rate was 70% (n = 35072). No significant differences in response rates were observed among the different regional areas. There were no demographic differences (age, sex, and socioeconomic status) between responders and nonresponders. The statistical analysis was descriptive, ie, no statistical tests were provided.
Results and discussion
The average diet of Italian schoolchildren is rich in protein (particularly animal proteins), and lipids constitute 32–35% of the total energy intake (predominantly saturated fatty acids followed by monounsaturated fatty acids) (Table 6). The highest protein, lipid (mainly saturated fatty acids), and cholesterol intakes were reported in central Italy.
View this table:
TABLE 6 . Daily energy, nutrient, and cholesterol intakes in the entire sample of Italian schoolchildren (n = 35072) and by regional area1
Although comparisons with earlier observations were difficult to make, the results were similar to those of local studies (47–49). Moreover, no relevant differences existed between our data and the data supplied by the Italian National Institute of Nutrition for the adult population in 1984 and 1991 (51). The main food categories (meat, cereals, fish, and green vegetables) were similarly distributed in both children and adults. These findings may indicate a relative homogeneity of dietary choices within Italian families. The diet thus categorized is one intermediate to that typical of northern European industrialized nations and that typical of developed countries in the Mediterranean area.
NUTRITIONAL EDUCATION AND BLOOD LIPID CONCENTRATIONS IN PRIMARY SCHOOL CHILDREN
In the 1980s, the potential future consequences of high blood cholesterol concentrations in childhood alerted pediatricians to the need for early, preventive, dietary measures. A pilot intervention trial was performed in healthy children of primary school age and the results were published in 1991 (52). The intervention was justified by the observation (53) that between 30% and 40% of children of this age in the Milan area had plasma cholesterol concentrations >4.65 mmol/L (180 mg/dL), the upper limit of the accepted range (54); a total of 15% of the children had concentrations >5.17 mmol/L (200 mg/dL).
Subjects and methods
All 820 schoolchildren in the second and fourth grades (ages 7 and 9 y, respectively) of 5 primary schools randomly selected in Milan were considered eligible for the trial. Randomization allocated 3 schools for intervention and 2 schools as the control group. Parents were asked for their permission to measure the plasma lipid concentrations of their children at the beginning and end of the school year (after 9 mo). In the intervention schools, the teachers attended a course presented by a trained physician on the relations between high blood cholesterol concentrations, dietary habits, and health risks. Teachers were also given instructional materials, including lesson plans prepared by a pedagogist, and posters, booklets, and leaflets to be distributed to the children. The materials indicated the need to reduce the consumption of animal foods and to increase the consumption of foods of vegetable and marine origin. It was suggested that meat should be consumed 3–4 times/wk and fish 3 times/wk, that cheese consumption be limited, and that fruit be substituted for commonly consumed snacks.
No effort was made to change the meals provided by the schools. Teachers also encouraged the children to speak with their parents about what they were learning at school. Parents were not directly contacted in any phase of the study. These instructional and informative materials were not provided to the teachers or children in the 2 control schools and the teachers in these schools did not attend the course provided to the teachers in the intervention schools. Lipid concentrations were measured as described previously. Repeated-measures ANOVA and Student's t test for paired data were used for the statistical analysis.
Results and discussion
A total of 424 (52%) children [n = 264 (53%) in the intervention group and 160 (50%) in the control group] had blood drawn at baseline and at 9 mo. The ANOVA showed a significant positive effect of intervention on the reduction in plasma total cholesterol (P < 0.0001). Plasma total cholesterol (
CONCLUSIONS
Some of our findings can be extended from our studies to the general issue of pediatric nutrition:
REFERENCES