点击显示 收起
1 From the Child Health Research Institute, Women's and Children's Hospital, North Adelaide, Australia (MM, RAG, and KR); the Department of Paediatrics, University of Adelaide, Adelaide, Australia (MM and RAG); and the Department Paediatrics and Child Health, Flinders University and Medical Centre, Adelaide, Australia (RAG and TU)
2 Supported by grants from Wyeth Nutrition; The Financial Markets for Children, Australia; New Zealand Milk; and Senior Research Fellowships from the National Health and Medical Research Council of Australia (to MM and RAG). 3 Reprints not available. Address correspondence to M Makrides, Applied Nutrition Group, Child Health Research Institute, Women's and Children's Hospital, 72 King William Road, North Adelaide, SA 5006, Australia. E-mail: makridesm{at}mail.wch.sa.gov.au.
ABSTRACT
Background: Adequate growth is an important indicator of health and well-being in infants.
Objective: Our objective was to determine the effect of supplementing infant formula with long-chain polyunsaturated fatty acids (LCPUFAs) on the growth of term infants.
Design: Using the methodology outlined by the Cochrane Collaboration, we reviewed all known randomized controlled trials that involved LCPUFA supplementation of infant formula fed to term infants. Outcome measures were weight, length, and head circumference. Original data obtained from the investigators of published trials were used. Outcomes were analyzed with fixed-effects or random-effects model meta-analyses and were reported as weighted mean differences with 95% CIs.
Results: We identified 14 eligible trials that had data available for meta-analysis (1846 infants). Trial quality was generally high. Meta-analysis showed no significant effect of LCPUFA supplementation on infant weight, length, or head circumference at any assessment age. Similarly, subgroup analyses showed that supplementation with only n3 LCPUFAs (no arachidonic acid) had no significant effect on infant weight, length, or head circumference. The source of LCPUFA supplementation (phospholipid or triacylglycerol) also did not significantly affect infant growth.
Conclusion: We found no evidence that LCPUFA supplementation of infant formula influences the growth of term infants in either a positive or a negative way.
Key Words: Systematic review meta-analysis docosahexaenoic acid DHA arachidonic acid AA infant growth
INTRODUCTION
Growth is the cornerstone assessment of nutritional health and well-being in preverbal children. Growth failure can be one of the first clinical signs that indicate underlying pathology, whereas growth excess may indicate predisposition to obesity and associated chronic health problems (1, 2). Understanding the nutritional factors that alter growth are of interest to health workers and public health authorities alike.
An important change to the composition of infant formula in the past 5 y has been the addition of long-chain polyunsaturated fatty acids (LCPUFAs). The key LCPUFAs added have been docosahexaenoic acid (DHA, 22:6n3) and arachidonic acid (AA, 20:4n6). Both DHA and AA are found in human milk, and infants fed unsupplemented formulas have lower concentrations of DHA and AA in their plasma and erythrocytes than do infants who are breastfed or fed supplemented formulas (3, 4). Some randomized controlled trials have shown that infants fed formulas with DHA alone or DHA in combination with AA have better performance in visual and developmental tests than do unsupplemented infants (5-9). However, it is also recognized that DHA and AA are integral structural components of all cells and may, therefore, have other health effects.
The effect of DHA and AA on infant growth has been somewhat controversial. Some of the early trials (which used fish oil containing only n3 LCPUFA) suggested that preterm infants fed formulas supplemented with n3 LCPUFA alone weighed less and were shorter than preterm infants fed standard, unsupplemented formulas (10-12). Later trials that intervened with formulas containing both DHA and AA have shown positive effects of supplementation on weight and length gains (13-16) as well as negative effects on linear growth (17). Although most trials that involved term infants have shown no effect of LCPUFA supplementation on growth, the observations in preterm infants have resulted in significant debate and raised many questions about the type and source of LCPUFA supplementation for both term and preterm infants. It has also resulted in many formulas for term infants being supplemented with a balance of DHA and AA that favors n6 LCPUFA.
This paper reports the systematic review and meta-analysis of growth in randomized controlled trials that involved LCPUFA interventions in term infants. Subgroup analyses assessed whether there were differential effects of supplementation on growth depending on the type and source of LCPUFAs and whether there were differential effects of LCPUFA supplementation between boys and girls because of their differing growth patterns. A separate paper will report the meta-analysis that relates to growth outcomes of preterm infants.
SUBJECTS AND METHODS
Search strategy
The Medline database and the proceedings of relevant conferences known to us were searched for randomized trials of dietary LCPUFA intervention in term infants. The following terms were used to search the Medline database from 1960 to July 2004 limited to randomized clinical trials: fatty acid, unsaturated; omega fatty acid; or n3 fatty acid. The reference lists of retrieved publications were also searched for relevant trials. The last literature search was completed in July 2004. A total of 21 trials were evaluated for inclusion (5, 7, 8, 18-35).
Selection of trials
Trials that involved term infants were eligible for inclusion if they had a randomized design, growth was reported as an outcome measure, the LCPUFA intervention commenced within 14 d of birth, and the test diets were fed for at least 12 wk. Fifteen of the original 21 trials met these criteria (Figure 1). Two studies were excluded because they only assessed the effect of LCPUFA supplementation on infant blood concentrations of LCPUFAs (35, 36), 1 because the study started when the infants were >14 d old (33), 2 because the intervention diets were fed for <12 wk (32, 34), and 1 did not involve an LCPUFA intervention (37). One trial was subsequently excluded because mean weight, length, and head circumference data were not available from either the paper or the investigators (31). Therefore, a total of 14 trials were included in the meta-analysis (Figure 1).
FIGURE 1.. Randomized controlled trials (RCTs) identified and used in the systematic review and meta-analysis. Ref, reference; LCPUFA, long-chain polyunsaturated fatty acid.
Data abstraction and collection
The characteristics of included studies and the data were abstracted independently by 2 investigators. As mean weight, length, and head circumference data were not often reported in publications, investigators were contacted to provide data that were incomplete. Growth data from the included trials were abstracted on an intention-to-treat basis. Investigators were also asked to provide data separately for boys and girls, and we retrieved this information for 12 of 14 trials.
Financial support to conduct the systematic review was obtained from industry, charitable organizations, and government sources. The funding agencies had no influence on the abstraction and analysis of the data. Data from multicenter, industry-sponsored trials were supplied by Nutricia, Mead Johnson, and Abbott Laboratories.
Analysis
We compared the weight, length, and head circumference of LCPUFA-treated with untreated infants at 4 and 12 mo of age. These primary time points were chosen because 4 mo often represents the end of feeding formula as a sole source of nutrition and 12 mo often coincides with the end of formula feeding. When data were available, weight, length, and head circumference of infants were also compared at 3, 6, and 9 mo of age. Meta-analyses were computed by using the Metaview Program available in REVIEW-MANAGER 4.1 (Cochrane Collaboration). A fixed-effects model was used when there was no significant heterogeneity between studies, and a random-effects model was used when significant heterogeneity existed. Each study was weighted in the meta-analysis according to the SD and number of infants. Subgroup analyses were conducted to determine whether there were differential effects according to sex, type, and source of LCPUFA supplementation. For example, we compared weight, length, and head circumference of infants treated with only n3 LCPUFA with the control group, infants fed LCPUFAs from phospholipid sources with the control group, and infants fed LCPUFAs from triacylglycerol sources with the control group.
RESULTS
Summary of included studies
A total of 14 trials conducted across Europe (n = 6), North America (n = 6), and Australia (n = 2) were included in our review, and these trials are summarized in Table 1 (5, 7, 8, 18-30). All trials reported a randomized design, but it was not always clear from the published papers whether there was adequate concealment of allocation. All trials were placebo controlled and compared standard cow milkbased formula with no LCPUFAs with equivalent formulas containing LCPUFAs. The energy and protein concentrations of test formulas across all trials were comparable and approximated 280 kJ/100 mL and 1.5 g/100 mL, respectively. Doses of n3 LCPUFA in the supplemented formulas ranged from 0.1% to 1.0% of total fatty acids, and the dose of n6 LCPUFA ranged from 0 to 0.72% of total fatty acids (Table 1). LCPUFA supplements were from a variety of sources, including fish oils, egg phospholipid fractions, egg triacylglycerol fractions, and algal and fungal oils. Eleven trials clearly reported that both parents and trial assessment staff members were blinded to each infant's group allocation (5, 7, 8, 18-22, 25-29). Most trials intervened with their test diets through to 4 mo of age (5, 7, 8, 18-27, 29, 30) and also assessed the growth of infants at this age (5, 7, 8, 18-24, 27, 30). Six of the included trials reported <20% attrition of randomly assigned infants through to 4 mo of age (5, 7, 18, 19, 22, 27, 30) (Table 1).
View this table:
TABLE 1. Trials included in the systematic review of long-chain polyunsaturated fatty acid (LCPUFA) supplementation1
Primary analysis
Most trials and the majority (1050 of 1662, 63%) of infants were assessed at 4 mo of age. At this time point the summary statistic indicated no differences in weight, length, or head circumference between infants allocated to either control or LCPUFA-supplemented formulas (see diamond symbols, Figures 24), despite individual studies that reported statistically significant effects on head circumference. Note that at 4 mo all included trials were actively feeding the test formulas, and infants were likely to have received few other sources of dietary LCPUFAs (38). Similarly, at 12 mo of age no differences were observed in the summary statistic for weight, length, or head circumference between groups, although 1 study reported an effect on weight at this time (Figures 24). However, by this stage only 4 trials (653 infants) continued intervention with the test formulas. Two included trials had no growth measurements either at 4 or 12 mo but had measured growth at either 3 mo (29) or 6 and 9 mo (25, 26). Comparisons of weight, length, and head circumference at these ages also revealed no differences between LCPUFA-treated infants and control infants. Overall, mean weight, length, and head circumference from the meta-analysis at 3, 4, 6, 9, and 12 mo of age approximated the 50th percentiles on standardized growth charts (39, 40).
FIGURE 2.. Meta-analysis forest plots for infant weight for all trials and for the subgroups intervening with only n3 long-chain polyunsaturated fatty acid (LCPUFA). Trials are displayed according to increasing dose of n3 LCPUFA, regardless of the presence of n6 LCPUFA. Trials marked with an asterisk (*) had ceased dietary interventions at the indicated assessment. WMD, weighted mean difference.
FIGURE 3.. Meta-analysis forest plots for infant length for all trials and for the subgroups intervening with only n3 long-chain polyunsaturated fatty acid (LCPUFA). Trials are displayed according to increasing dose of n3 LCPUFA, regardless of the presence of n6 LCPUFA. Trials marked with an asterisk (*) had ceased dietary interventions at the indicated assessment. WMD, weighted mean difference.
FIGURE 4.. Meta-analysis forest plots for infant head circumference for all trials and for the subgroups intervening with only n3 long-chain polyunsaturated fatty acid (LCPUFA). Trials are displayed according to increasing dose of n3 LCPUFA, regardless of the presence of n6 LCPUFA. Trials marked with an asterisk (*) had ceased dietary interventions at the indicated assessment. WMD, weighted mean difference.
Subgroup analyses
Six trials intervened with a formula that contained only n3 LCPUFA, and all of these trials also assessed growth at 4 mo of age (Table 1). Weight, length, and head circumference at 4 and 12 mo of age did not differ between groups (Figures 24). No consistent pattern was observed to indicate that dose of n3 LCPUFA influenced weight, length, or head circumference. No differences were observed in weight, length, or head circumference between the groups at either 4 or 12 mo of age in the subgroup analyses related to sex, phospholipid sources of LCPUFAs, or triacylglycerol sources of LCPUFAs (data not shown).
DISCUSSION
Our paper is the most comprehensive meta-analysis of the effect of LCPUFA supplementation of infant formula on the growth of term infants to date. It includes unreported details from published studies (most trials did not publish mean growth data for boys and girls) as well as data from the largest trial that are only available in abstract form. Most trials were conducted to a high standard, all enrolled infants in the first 2 wk of life, and all but one fed the test formulas until infants were at least 4 mo. This high degree of compatibility between trial protocols gave us confidence in the legitimacy of combining the data. Because studies were conducted in different countries with different population groups, the results of the meta-analysis can be generally applied to formula-fed infants. The main difference between the trial designs was in the type and source of LCPUFA supplementation tested, and this allowed us to statistically explore the effectiveness of a range of fatty acid types and sources through subgroup analyses.
The combined data show that no effect of LCPUFA supplementation of infant formula was observed on the growth of term infants at any age. This observation was not influenced by the type of supplementation (n3 LCPUFA alone or n3 + n6 LCPUFA), the source of supplementation (triacylglycerol or phospholipid), or sex. Our results are also consistent with the Cochrane systematic review that assessed the effect of LCPUFA interventions on the outcomes of term infants, although the Cochrane review contained less growth data and undertook a less extensive analysis (41). Because the relative merits of the outcomes of systematic reviews compared with individual adequately powered trials are often discussed, our observations are particularly noteworthy because there is congruency between the results of individual well-powered trials [7 of 14 included trials according to independently published criteria (42)] and the results of the systematic review.
One of the most hotly debated issues that relates to LCPUFA supplementation of infant formula is whether n3 LCPUFA could be added without a source of AA. Much of this debate stems from the early observations of growth deficits in preterm infants who received formulas that contained only n3 LCPUFA compared with control formulas (10-12). It was hypothesized that the depression of plasma AA caused by dietary n3 LCPUFA supplementation may be a factor that contributes to the growth deficit because both observational data (43) and 1 randomized trial (44) indicated an association between plasma AA and weight and length. However, in term infants there is no evidence of any reduction in weight, length, or head circumference associated with dietary n3 LCPUFA supplementation in the absence of AA according to 6 trials. The subgroup analysis of those trials had dietary interventions that ranged in n3 LCPUFA concentration from 0.1% to 1.0% total fatty acids, although most infants consumed formulas with 0.30.45% n3 LCPUFA as total fatty acids (5, 7, 20, 27, 30). Despite the limited data there was no suggestion of an n3 LCPUFA dose effect on any indicator of growth. All infants consumed the test formulas for at least 4 mo, and trials showed a mean reduction in plasma AA of 25% compared with control infants. Because growth is the main criteria used to assess nutritional health and well-being of infants, the fact that formulas supplemented with only n3 LCPUFA are capable of supporting adequate infant growth, despite reductions in AA status, indicates that such formulas are nutritionally adequate. This finding is of considerable relevance because sources of AA for use in infant formulas are expensive and may add to the cost of infant formulas.
The often antagonistic actions of n3 and n6 LCPUFAs in a variety of metabolic systems has caused a reluctance on the part of some experts to endorse the supplementation of infant formula with n3 LCPUFA without the addition of AA. DHA and AA are never absent from human milk. Although the concentration of AA in breast milk depends to some extent on the maternal diet, the amount of DHA in breast milk is almost entirely driven by dietary DHA. Claims of an optimum AA-to-DHA ratio of 2:1 have been based on amounts seen in the breast milk of some mothers in industrialized Western countries who consume diets low in n3 fats but rich in n6 fats (United States, Australia). It must be emphasized that the AA:DHA of human milk from women in other countries (and even American and Australian women who include fish in their diets) varies widely, and many Asian mothers have AA:DHA of 0.4:1, whereas the milk of many European mothers is closer to 1:1 (45, 46). Thus, the AA:DHA of 2:1 seen in some human milk, and suggested to be optimal, seems to only be part of a continuum that is diet driven and may have little relevance to the infant in relation to growth.
Subgroup analyses that investigate the effect of formula LCPUFA supplementation with either phospholipid or triacylglycerol sources also indicated little effect on infant growth. Although subgroup analyses are important, these data need to be interpreted with caution because of the comparatively smaller sample sizes and the increased number of comparisons. With this caveat in mind, our meta-analysis results are consistent with metabolic and biochemical studies that show little difference in the absorption of DHA and AA from both phospholipid and triacylglycerol sources (47) and that infant plasma and erythrocyte concentrations of DHA and AA depend on the dose of dietary supplementation rather than the source of supplementation (27, 32).
A decade after the publication of the first recommendations for LCPUFA supplementation of infant formulas (48-50) and the first published randomized trials of LCPUFA interventions, we have a wealth of quality information about growth and development. However, the impetus to add LCPUFAs to infant formula has been to improve the developmental outcome of infants, and this has resulted in the growth data from many trials often being inadequately reported, even though there has been some controversy about the effect of dietary LCPUFAs on growth. Our systematic review has clearly shown that LCPUFA supplementation of infant formulas for term infants does not affect growth and in this regard is safe. This review has attempted to exclude the possibility of publication bias by seeking details of growth data that were unpublished or not published in full. The positive response from both the investigators and the industry has allowed the completion of a comprehensive review in which there can be a high level of confidence.
ACKNOWLEDGMENTS
MM and RAG designed the study and were responsible for coordinating funding. MM and RAG supervised KR and TU, who assisted with data abstraction and data analysis. MM and RAG wrote the paper with contributions from coauthors. MM and RAG have collaborated with the formula industry on clinical trials related to LCPUFA interventions, but the formula industry and the funding agencies for this project had no role in its design, analysis, and interpretation. All data were managed and analyzed in Adelaide, Australia. The authors had no associated financial interests. The International LCPUFA Investigators donated unpublished data and commented on the interim analyses and the final draft of the manuscript.
The International LCPUFA Investigators include Carlo Agostoni, Department of Pediatrics, The San Paolo Biomedical Institute, University of Milan, Milan, Italy; Nancy Auestad, Ross Products Division, Abbott Laboratories, Columbus, OH; Eileen Birch, Retina Foundation of the Southwest, Dallas, TX; Susan E Carlson, University of Kansas Medical Center, Kansas City, KS; Tamas Decsi, Department of Pediatrics, University Medical School of Pécs, Pécs, Hungary; Deborah Diersen-Schade, Mead Johnson Nutritionals, Evansville, IN; William Goldman, Wyeth Nutritionals International, Philadelphia, PA; JS Forsyth, Department of Child Health, University of Dundee, Dundee, United Kingdom; Robert Hall, The Children's Mercy Hospital, University of Missouri, Kansas City, MO; Cheryl L Harris, Mead Johnson Nutritionals, Evansville, IN; Dennis Hoffman, Retina Foundation of the Southwest, Dallas, TX; Sheila M Innis, British Columbia Research Institute for Children's and Women's Health, University of British Columbia, Vancouver, Canada; Berthold Koletzko, Dr von Hauner Children's Hospital, University of Munich, Muenchen, Germany; Alexandre Lapillonne, Rene Descartes University, Paris, France, and Children's Nutrition Research Center, Baylor College of Medicine, Houston, TX; Kimberly L Merkel, Mead Johnson Nutritionals, Evansville, IN; Michael Montalto, Ross Products Division, Abbott Laboratories, Columbus, OH; James Moorcraft, Royal Glamorgan Hospital, Llantrisant, United Kingdom; Geraint Morris, Singleton Hospital, Swansea, United Kingdom; Ricardo Uauy, Institute of Nutrition and Food Technology (INTA), University of Chile, Santiago, Chile; John CK Wells, Nutricia Baby Food Division, Trowbridge, United Kingdom; and Peter Willatts, Department of Psychology, University of Dundee, Dundee, United Kingdom.
REFERENCES