Omega-3 long-chain polyunsaturated fatty acid (n-3 LCPUFA) supplementation during infancy may reduce adult cardiovascular risk as observed in animals. We assessed the effect of n-3 LCPUFA supplementation in infancy on growth, body composition, and cardiometabolic risk factors at 5 years of age.
Infants were randomly assigned to a daily supplement of n-3 LCPUFA or olive oil (control) from birth to 6 months (n = 420). Measurements included weight, length, cord blood adipokines at birth and anthropometry, skinfolds, blood pressure, heart rate, fasting blood adipokines, and biochemistry at 5 years.
The infants who received n-3 LCPUFA had a smaller waist circumference at 5 years (coefficient: 1.1 cm; 95% confidence interval [CI]: 0.01 to 2.14), which remained significant after adjustments for confounders (coefficient: 0.8 cm; 95% CI: 0.19 to 1.30). Five-year-old boys who received n-3 LCPUFA supplementation as infants had a 21% reduction in insulin concentrations (ratio: 0.79; 95% CI: 0.66 to 0.94) and a 22% reduction in insulin resistance (ratio: 0.78; 95% CI: 0.64 to 0.95) compared with the control group. There were no other differences in growth and cardiometabolic risk factors between the groups for the whole cohort at birth, 2.5, or 5 years.
Supplementation with n-3 LCPUFA in infancy revealed a reduction in waist circumference at 5 years. Boys in the n-3 LCPUFA group showed reduced insulin concentrations and insulin resistance at 5 years, which may have beneficial outcomes for later health. No effects were seen in girls. Longer term follow-up of the cohort is warranted to determine whether these differences are maintained into adolescence.
Increasing dietary omega-3 polyunsaturated fatty acids has been associated with attenuating cardiometabolic risk factors in adults. Supplementation during pregnancy or early life in animal studies has provided beneficial effects on cardiometabolic risk factors.
Supplementing omega-3 polyunsaturated fatty acids in infancy resulted in attenuated 5-year waist circumference and gender-specific effects (plasma insulin concentrations and insulin resistance were reduced at 5 years in boys who received supplementation). These differences are relevant if maintained to adulthood.
Interactions between the developing infant and the environment during the critical development period in early life determine the propensity to develop disease later in life.1 Perinatal nutritional exposures are critical for the ongoing developmental maturation of multiple organ systems and optimal physiologic functions.2 Animal data has been used to support the ability of early environment, in particular nutritional status, in influencing cardiometabolic risk.3 For example, the early administration of omega-3 long-chain polyunsaturated fatty acids (n-3 LCPUFAs), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) can be used to reverse some of the fetal programming of hypertension in rats.4,5
In children with impaired fetal growth, n-3 LCPUFA supplementation from 6 months to 5 years of age was associated with less intimal medial thickening,6 an early marker of atherosclerosis in adults. In children born with reduced birth weight, serum n-3 LCPUFA concentrations were inversely associated with systolic blood pressure (SBP).7 Several long-term randomized trials of n-3 LCPUFA supplementation have been reported. A dietary intervention used to increase n-3 LCPUFA and decrease omega-6 LCPUFA intakes in term infants from infancy to 5 years did not result in significant improvements in arterial structure and function, blood pressure (BP), high-sensitivity C-reactive protein (hs-CRP), and lipoproteins at 8 years of age.8 On the contrary, preterm girl infants who were randomly assigned to n-3 LCPUFA-supplemented formula increased in weight, adiposity, and BP at 10 years of age.9 The only human study in which authors reported beneficial long-term effects of n-3 LCPUFA formula supplementation in infancy revealed significantly lower BP at 6 years of age.10 The varied outcomes (some significant and others nonsignificant) after n-3 LCPUFA intake in infancy was used to reduce cardiometabolic risk could be due to the variations in the composition of n-3 LCPUFA administered, dosage and duration of administration, length of follow-up, and other methodological aspects.8,10,–16
With the current study, we aimed to ascertain if the effects of feeding n-3 LCPUFA in infancy were sustained beyond the intervention period. Specifically, with our study, we aimed to determine whether n-3 LCPUFA supplementation from birth to 6 months of age would favorably influence growth, body composition, and cardiometabolic risk factors in 5-year-old children.
Trial Design and Study Population
Authors of the Infant Fish Oil Supplementation Study investigated the effects of n-3 LCPUFA supplementation on neonatal immune development, allergy, and neurodevelopment outcomes in the children.17,–21 Subjects were stratified for maternal allergy (asthma versus other allergy), parity (first child versus ≥2 children), and paternal allergy (allergic versus nonallergic) with block randomization. A follow-up was undertaken at 5 years to assess the effects of n-3 LCPUFA supplementation in infancy on markers of vascular health and cardiometabolic risk in childhood. The study design, methodology, and inclusion and exclusion criteria have been published.17,21
Briefly, 420 allergic (otherwise healthy), nonsmoking women with low fish intake were recruited at 36 weeks of pregnancy between 2005 and 2008 from private and public metropolitan antenatal clinics in Perth, Western Australia (Fig 1). Maternal allergy was defined by a positive skin prick test result to at least 1 of a defined panel of allergens (milk, peanut, house dust mite, cat, grass, and mold; Hollister-Stier Laboratories, Spokane, WA) and a history of allergic disease (asthma, allergic rhinitis, or eczema). This study was conducted according to the guidelines written down in the Declaration of Helsinki. All procedures involving human subjects were approved by the ethics committees of participating hospitals within the metropolitan area of Perth, Western Australia. Written informed consent was obtained from all the mothers.
Randomization and Intervention
The 420 healthy term infants were randomly assigned to receive either a daily 650 mg supplement of encapsulated n-3 LCPUFA in the form of ethyl esters (280 mg DHA and 110 mg EPA) or olive oil (OO) (66.6% omega-9 oleic acid) as the control from birth to 6 months. Both capsules were from Ocean Nutrition Ltd, Mulgrave, Canada. Capsules were purchased in 1 batch in 2005, and the composition was regularly tested by an independent laboratory during the trial. The fatty acid composition of the capsules remained unchanged over the course of the study, and peroxide and acid levels remained compliant with Australian standards.
Staff independent from those performing the allocation dispensed image and scent-matched n-3 LCPUFA or OO capsules to participants. Participants and study personnel were blinded to the groups for the duration of the study. Capsules were pierced and squirted into the infant’s mouth in the morning immediately before breastfeeding or in the formula during their first daily feed. During the study, the Ocean Nutrition product was discontinued, and the final 27 children received similar n-3 LCPUFA capsules containing 250 mg DHA and 60 mg EPA or OO (provided by nuMega Ingredients Pty Ltd, Brisbane, Australia). This brand substitution was endorsed and supervised by the ethics committee at Princess Margaret Hospital.
The dose implemented is significantly higher than the average intake of DHA and EPA from breast milk within Australia, equating to ∼72.45 mg of DHA and 31.5 mg EPA per day (assuming DHA is 0.23% and EPA is 0.010% total fatty acids,22 750 mL per day intake, with 4.2 g of fat per 100 mL). The relatively high dose of fish oil was chosen to make the likelihood of observing an effect as great as possible. Compliance with the treatment regimen was determined by capsule diaries, a count of the returned capsules, and infant erythrocyte DHA and EPA composition at 6 months. The children were assessed at birth, 6 months, 2.5 years, and 5 years. Questionnaires at 5 years were used to ascertain child smoking exposure in the home and physical activity status. Detailed clinical examination was performed at birth, 2.5, and 5 years.
The primary end points with repeated measures at birth, 2.5, and 5 years were weight, height, BMI, and head circumference. Other primary outcomes measured only at 5 years were arm and waist circumferences, skinfolds (abdominal, suprailiac, subscapular, and triceps), BP and heart rate (HR), fasting bloods analyzed for glucose, insulin, and lipids (cholesterol, high-density lipoprotein cholesterol [HDLC], low-density lipoprotein cholesterol [LDLC], and triglycerides). Insulin resistance was represented by the homeostasis model assessment of insulin resistance (HOMA-IR).23 Secondary end points at 5 years included hs-CRP, leptin, and adiponectin. Leptin and adiponectin were also measured at birth.
Blood Collection and Processing
Cord blood was collected at birth from the placental vessels. Peripheral blood at 5 years was obtained early in the morning by venipuncture after an overnight fast where consented. Blood was collected into heparinized tubes, processed immediately, and the plasma was stored at −80°C. Phospholipid fatty acid composition was measured in erythrocyte cell membranes on available samples as previously published.24
Clinical and Biochemical Measurements
Clinic BP and HR were measured at a seated position by using a Dinamap (CARESCAPE V100 Monitor; General Electric Healthcare, Little Chalfont, United Kingdom) with an appropriate sized cuff. Height was measured by using a calibrated model 240 stadiometer (Seca; Birmingham, United Kingdom) to the nearest 0.1 cm. Weight was measured by using a professional medical scale (Tanita BWB-600; Tanita Australia, Australia) to the nearest 0.1 kg in light clothing, and BMI was calculated. Skinfolds were measured by using a manual skinfold caliper (John Bull; British Indicators Pty Ltd, Burgess Hill, United Kingdom) at 4 sites on the subjects’ left side (abdominal, suprailiac, subscapular, and triceps). Two measurements were recorded, and a third measurement was obtained if there was a difference of 10 mm between measurements. A mean reading was obtained from the 2 or 3 readings. Head, waist, and arm circumferences were measured by using a Seca girth measuring tape (model 201). Two measurements were obtained, and a third measurement was obtained if the measurements had a difference of 1 mm. A mean reading was obtained from the 2 or 3 readings.
Serum glucose, total cholesterol, HDLC, and triglycerides were enzymatically determined on an Architect c16000 Analyzer (Abbott Laboratories, Lake Forest, IL). LDLC was derived from the Friedewald et al25 calculation. hs-CRP was measured by an immunoturbidimetric method on the Architect c16000 Analyzer. Fasting insulin was determined on an Architect i2000SR Analyzer. HOMA-IR was calculated by (fasting glucose × fasting insulin)/22.5.23 Serum adiponectin levels were quantitated by using a Human Total Adiponectin sandwich enzyme-linked immunosorbent assay kit (Research & Diagnostics Systems, Inc, Minneapolis, MN). Serum leptin levels were quantitated by using a Human Leptin Quantikine enzyme-linked immunosorbent assay kit.
Data were summarized by using counts, percentages, means, SDs, medians, and interquartile ranges. Log transformations were applied to outcomes when the data were not approximately normal, and geometric means and geometric SDs were calculated. Between-group differences in characteristics at 5 years, differences between participants at 5 years, and those lost to follow-up for dichotomous data were assessed by using a χ2 test. For continuous data, group differences were assessed by using t tests for normal data, and Mann–Whitney U tests were used for data that could not achieve normality. Outcomes measured at multiple time points were analyzed by using linear mixed models with maximum likelihood estimation (MLE) to produce unbiased estimates (mixed-effects maximum likelihood regression). MLE is used to handle missing data by using observed data as well as intraclass correlation to inform the likelihood function that is then maximized to obtain an estimate of the mean at a given time point. MLE, therefore, retains participants if they provided data for at least 1 time point, thus avoiding attrition bias. Differences in the pattern over time between intervention groups were tested by the interaction of group and time in the model. Outcomes available only at 5 years of age were analyzed by using linear regression. hsCRP that were >10 mg/L were omitted because this may represent an acute inflammatory response and may not reflect basal healthy concentrations.26 Adjustments for sex, birth weight, maternal age at birth, cigarette smoke exposure during pregnancy, childhood BMI, physical activity performed at 5 years, and actual age at the 5-year visit were performed by including these variables in a multivariable model. Outcomes at 5 years of age were stratified by sex to test for the differential effects for boys and girls. Fatty acids with limited follow-up measurements were analyzed by using a regression with a clustered arrangement. For these outcomes, a per person cluster adjustment was made to the variance to account for repeated measures on a subject. Models were bootstrapped to provide more robust P values in which there were slight departures from normality. A P value of <.05 (2-tailed) was considered statistically significant for all analyses. All statistical analyses were performed by using the Stata 13.1 statistical package (Stata Corp, College Station, TX).
We used power calculations to estimate that the sample size of 300 subjects would detect a change in BMI of 0.6 and 0.08 mmol/L triglycerides between groups over time with >90% power at a significance level of P = .05.
A total of 420 infants were randomly assigned, with 218 assigned to the n-3 LCPUFA group and 202 assigned to the OO group (Fig 1). At the completion of the intervention period, there were no significant differences between erythrocyte DHA or EPA levels (P = .73 and P = .07, respectively) or plasma phospholipid DHA or EPA levels (P = .16 and P = .12, respectively) between the 2 groups of participants receiving n-3 LCPUFA capsules from different suppliers, so all children were included in the final analysis. At 6 months of age, oleic acid, arachidonic acid, and DHA did not differ between supplementation groups (data not shown). Of the infants in the n-3 LCPUFA group, 70.2% were exposed to formula, compared with 61.9% in the OO group (no significant difference; P = .127). Of those who received formula, 54.3% in the n-3 LCPUFA group compared with 43.9% in the OO group received a formula containing n-3 LCPUFA (no significant difference; P = .164). At 5 years, 165 participants in the n-3 LCPUFA group and 157 in the OO group remained in the study. Children were more likely to participate in the study at the 5-year follow-up if their mothers were older (P < .001), educated for ˃12 years (P < .001) at the time of recruitment, and if they had a higher capsule adherence during the intervention (P = .002) (Supplemental Table 7). These biases for follow-up occurred equally between control and placebo groups. There were no other significant differences in the demographic and clinical characteristics of the participants lost to follow-up compared with those retained in the study. There were no significant differences in antenatal, birth, or infant characteristics between the n-3 LCPUFA and OO groups for those that attended the 5-year follow-up (Tables 1 and 2).
Clinical and Biochemical Outcomes
Measures Only Obtained at 5 Years
Waist circumference was significantly lower by 1.1 cm in the n-3 LCPUFA group compared with the OO group (P = .048; 95% confidence interval [CI]: 0.01 to 2.14), and this remained significant (lower by 0.8 cm) after adjustment for covariates (P = .008; 95% CI: 0.19 to 1.30) (Table 3). No significant differences were observed in arm circumference, SBP, diastolic blood pressure (DBP), HR, skinfolds, fasting glucose, insulin, HOMA-IR, lipids, and hs-CRP between groups at 5 years before and after adjustments (Table 3).
At 5 years, an interaction between sex and intervention group was detected for insulin (Pinteraction = .007) and insulin resistance (Pinteraction = .012) (Table 4). Thus, the between-group differences for insulin concentrations and insulin resistance in the boys is significantly different to the between-group differences in the girls. The boys supplemented with n-3 LCPUFA had significantly lower insulin concentrations (P = .008; ratio: 0.79; 95% CI: 0.66 to 0.94) by 21% and insulin resistance (P = .02; ratio: 0.78; 95% CI: 0.64 to 0.95) by 22% compared with the boys in the OO group. The girls supplemented with n-3 LCPUFA had no significant differences between groups for insulin concentrations (P = .24) and insulin resistance (P = .28).
Analysis of Measures Repeated Over Time
The analyses of measures repeated over time are reported in Table 5. With the randomization, we produced groups that were well balanced as observed through the baseline characteristics at birth (ie, before intervention) including weight, height, BMI, head circumference, adiponectin, and leptin. There were no significant differences in weight, height, BMI, and head circumference between the groups at 2.5 and 5 years for analyses before or after adjusting for covariates and neither was there a significant difference in the change over time between groups (Supplemental Table 8). As expected, offspring anthropometric measures (weight, height, BMI, head circumference) increased from birth to 2.5 and 5 years (P < .001), but there were no significant differences in the change over time between groups (Supplemental Table 8).
Plasma concentrations of adiponectin and leptin at 5 years were not significantly different between n-3 LCPUFA and OO groups, before or after adjustments for covariates (Supplemental Table 8). Adiponectin and leptin concentrations decreased from birth to 5 years of age (P < .001), but there were no significant differences in the change over time between groups (Supplemental Table 8).
Fatty Acid Analysis at 5 Years
At 5 years of age, 239 participants (121 from the n-3 LCPUFA group and 118 from the OO group) provided blood samples for fatty acid analysis. There were no significant differences in EPA and DHA levels between the n-3 LCPUFA and OO groups (Table 6).
With this randomized trial, we suggest that n-3 LCPUFA supplementation in term infants from birth to 6 months of age results in the reduction of waist circumference in 5-year-old children. There were no long-term benefits in other anthropometric measures or any of the cardiometabolic risk factors, including BP and HR, or biochemical markers measured when data for boys and girls are combined. However, boys supplemented with n-3 LCPUFA had significantly lower plasma insulin concentrations and lower insulin resistance compared with the OO group at 5 years of age.
Authors of adult studies have reported changes in certain cardiovascular risk factors at the time of supplementation with n-3 LCPUFA.27 However, nutritional interventions, when applied at critical periods in early life, have the potential to remotely influence or “program” future cardiovascular disease risk, as reported in animal studies. For example, pregnant rodents fed with low-protein diets resulted in offspring with elevated BP that persisted into adulthood compared with rodents in the control group.28,29 A recent study of Sprague Dawley rats revealed that maternal supplementation with conjugated linoleic acid given after a high-fat diet reversed the associated programming of metabolic dysfunction in the offspring.30 Site-specific alteration in gene regulation has been reported in adipose tissue from adult rats fed diets high in either EPA or DHA or with no n-3 LCPUFA.31
Regarding n-3 LCPUFA, such a potential for reduced DBP had been suggested by a randomized controlled trial reported by Forsyth et al.10 Bottle-fed infants received either formula supplemented with DHA and arachidonic acid or standard formula for the first 4 months of life. At 6 years of age, DBP was significantly reduced (mean difference 23.6 mm Hg; P = .018) in the 65 children who had received supplemented formula than in the 71 who had received standard formula. In children with impaired fetal growth, a fish oil supplement taken at 500 mg per day in early childhood prevented intima-media thickening, an early marker of atherosclerosis in adults.6
We observed that waist circumference was attenuated in the long-term in those children supplemented with n-3 LCPUFA from birth to 6 months. In adult studies, plasma n-3 LCPUFA concentrations used as a biomarker of dietary n-3 LCPUFA intake in 124 healthy adults (BMI 20–40) were inversely related to BMI, waist circumference, and hip circumference, suggesting that a higher plasma n-3 LCPUFA status might protect against obesity.32 Waist circumference is a measure of central obesity and, in adults, this measurement correlates with visceral adipose tissue and increased risk of cardiometabolic disorders.33 Authors of a meta-analysis of prospective cohort studies and randomized controlled trials have reported that, for a 1 cm increase in waist circumference, the relative risk of a cardiovascular event increased by 2% (95% CI: 1% to 3%) overall after adjusting for age, cohort year, or treatment.34 The Centers for Disease Control and Prevention reports a mean waist circumference of 54.8 cm in 5-year-old boys and 55.6 cm in 5-year-old girls in the population of the United States.35 These statistics are slightly higher than the observed mean values in our study. In childhood, waist circumference has been strongly associated with subsequent metabolic syndrome characteristics in early adulthood.36
Our observations of decreased insulin concentrations and insulin resistance in boys supplemented with n-3 LCPUFA are in favor with observational studies and randomized controlled trials that reveal beneficial effects of n-3 LCPUFA on insulin sensitivity in children and adults.37 Authors of a study of rats found that the intake of more n-3 LCPUFA during early pregnancy reduced age-dependent insulin resistance in male offspring but not in female offspring.38 Sardinha et al38 proposed that the lower adiposity caused by the increased n-3 LCPUFA during the intrauterine life was responsible for the lower insulin resistance in male offspring. However, we did not observe any gender-specific adiposity changes in our study, and, to our knowledge, the relation between n-3 LCPUFA and glucose and insulin and HOMA-IR in human offspring has not been studied.
We did not find a relationship between n-3 LCPUFA supplementation during infancy and BP and HR at 5 years of age. These findings are in contrast with a 6-year follow-up of a randomized controlled trial in term infants (n = 147) in which authors reported that infant formula supplementation with 0.15 to 0.25 g per 100 g fat of DHA from birth to 4 months of age was associated with a reduction in DBP and mean BP.10 It is noteworthy, however, that the study had a smaller sample size than our study, was of a shorter period of supplementation (4 vs 6 months), and the researchers provided n-3 LCPUFA as a formula rather than as fish oil capsules. In our study, compliance due to the mode of delivery (ie, squirting of capsules into the infant’s mouth) could be used to explain the divergent findings. A more recent study has revealed that HR was reduced after infant formula was supplemented with DHA from birth to 12 months, but the measurements were obtained during the supplementation period.15 The BP findings in our study are more comparable to those reported from randomized controlled trials of n-3 LCPUFA supplemented in formula during varying periods. These studies have consistently failed to reveal the benefits on BP long-term at follow-up ages 5, 8, and 9.8,9,16
n-3 LCPUFA supplementation has the potential to affect structural and functional constituents of membrane phospholipids in tissues, as well as transport, signal transduction, and various enzymatic reactions.39 In our study, n-3 LCPUFA supplementation in infancy did not have residual beneficial effects on fasting blood lipids (cholesterol, triglycerides, HDLC, LDLC), measures of glycaemia (glucose, insulin, HOMA-IR), hs-CRP, or leptin and adiponectin at 5 years of age. Plasma leptin and adiponectin were marked by high concentrations at birth with a subsequent decrease at 5 years, in accordance with previous cohort studies.40 Studies in which authors supplemented n-3 LCPUFA in adults have revealed small changes in plasma total cholesterol, LDLC, and HDLC concentrations (<5%)27 but a consistent 20% to 25% reduction in triglycerides.41 It is difficult to compare the current study with other studies because of the different control oils and/or dietary compositions used in infant studies. Additionally, it is likely that the regulation of lipid metabolism may differ from adults, infants, and children.
The long-term follow-up of this study is a major strength, with repeated measurements of weight, height, head circumference, adiponectin, and leptin, from an established well-characterized group of participants. Our study had several potential limitations. Despite the relatively high dose of n-3 LCPUFA supplementation, the increase in n-3 LCPUFA in erythrocytes at 6 months was modest.17 This could be due, in part, to individual genetic differences in influencing the uptake of n-3 LCPUFA and the bioavailability and/or absorption of the supplements in the ethyl ester form. Although studies have revealed that n-3 LCPUFA used as ethyl esters are bioavailable in adults,42,43 there are no data on their bioavailability in infants. Cohort attrition and varying sample sizes due to missing data are an inevitable consequence of research among infants, but our recall at the 5-year follow-up was considerably large and involved 75% of the original study population.
Supplementing n-3 LCPUFA to infants from birth to 6 months of age may result in a reduction in waist circumference at 5 years of age and may also result in gender-specific effects because boys who received supplementation had a significant decrease in plasma insulin concentrations and insulin resistance at 5 years of age. These findings may have relevance if maintained into adolescence and subsequently into adulthood.
diastolic blood pressure
high-density lipoprotein cholesterol
homeostasis model assessment of insulin resistance
high-sensitivity C-reactive protein
low-density lipoprotein cholesterol
maximum likelihood estimation
- n-3 LCPUFA
omega-3 long-chain polyunsaturated fatty acid
systolic blood pressure
Drs Mori, Prescott, Beilin, and Huang conceptualized, designed, and supervised the study; Ms Burrows conducted the initial analyses; Dr See conducted the study and the initial analyses and drafted the initial manuscript; and all authors reviewed and revised the manuscript and approved the final manuscript as submitted and agree to accountable for all aspects of the work.
This trial has been registered at https://www.anzctr.org.au (identifier ACTRN12606000281594).
FUNDING: Supported by a grant from the National Health and Medical Research Council of Australia (APP1010495). RCH and TAM are supported by NHMRC Fellowships (grant number 1053384 and 1042255, respectively).
We thank the staff and participants who helped in this study. We particularly thank Carlie Dunford for her assistance with recruitment. The study sponsors were not involved in the design, implementation, analysis, presentation, or interpretation of results.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.