Prenatal Docosahexaenoic Acid Supplementation Does Not Affect Nonfasting Serum Lipid and Glucose Concentrations of Offspring at 4 Years of Age in a Follow-Up of a Randomized Controlled Clinical Trial in Mexico^,,

Abstract

Background: Docosahexaenoic acid (DHA) has regulatory effects on lipid and glucose metabolism. Differences in DHA availability during specific developmental windows may program metabolic changes.

Objective: We investigated the effects of maternal DHA supplementation during pregnancy on the nonfasting serum lipid and glucose concentrations of offspring at 4 y of age.

Methods: We used data from the Prenatal Omega-3 Fatty Acid Supplementation, Growth, and Development trial, a double-blind randomized controlled trial conducted in Mexico. Pregnant women were supplemented daily with 400 mg DHA or placebo from 18–22 wk of gestation to delivery. The primary outcomes of the trial were offspring growth and neurological development. Nonfasting blood samples were obtained from the offspring at 4 y of age. We analyzed serum total, HDL, non-HDL, and LDL cholesterol; the total-to–HDL cholesterol ratio; apolipoprotein B (apoB); triglycerides; glucose; and insulin as secondary outcomes and compared their concentrations between treatment groups.

Results: Data from 524 offspring were available. The women were compliant with the intervention based on pill counts and changes in cord blood and breast milk DHA concentrations. None of the between-group differences (DHA compared with placebo), adjusted for maternal height and time since last food intake, were significant (P range 0.27–0.83). Means (95% CIs) were as follows: total cholesterol (TC), 1.73 mg/dL (−2.63, 6.09 mg/dL); HDL cholesterol, 0.66 mg/dL (−1.07, 2.39 mg/dL); non-HDL cholesterol, 1.77 mg/dL (−1.83, 5.37 mg/dL); LDL cholesterol, 1.62 mg/dL (−2.21, 5.45 mg/dL); TC:HDL ratio, 0.01 (−0.09, 0.11); apoB, −0.15 mg/dL (−2.78, 2.48 mg/dL); triglycerides, 0.21 mg/dL (−10.93, 10.52 mg/dL); glucose, −0.67 mg/dL (−2.46, 1.11 mg/dL); and insulin, 0.62 μU/mL (−0.88, 2.11 μU/mL).

Conclusion: Prenatal DHA supplementation does not affect nonfasting serum lipid and glucose concentrations of offspring at 4 y of age. This trial was registered at clinicaltrials.gov as NCT00646360.

Introduction

Prenatal nutrition has long-term implications for metabolic programming, thereby affecting cardiovascular disease risk and metabolic health (1–3). DHA (22:6n–3) plays an important role in cell growth, differentiation, and integration of the cellular response to the metabolic and neuroendocrine environment during pregnancy (4, 5). Hence, DHA might be a key nutrient affecting metabolic programming. Maternal and infant DHA status is influenced by intake during pregnancy. DHA is actively transported through the placenta to the fetus (6). The accretion of maternal, placental, and fetal tissue is especially high during the third trimester; consequently, pregnant women and their developing fetuses have high DHA requirements. Expert consensus recommends that pregnant women have a mean daily intake of ≥200 mg DHA, and supplementation with ≤1 g/d of DHA has been administered in randomized trials without significant adverse effects (7). Supplementation may be especially important in populations with a low dietary intake of DHA.

In rat models, DHA supplementation during pregnancy improves lipid profiles and reduces insulin resistance in the offspring (8–10). Studies in animals and humans suggest that DHA supplementation during pregnancy may decrease BMI and modulate immune function in the offspring, specifically by stimulating the anti-inflammatory pathway (11–13).

In adults, dietary DHA intake is inversely associated with cardiovascular-related morbidity and mortality (14, 15), and DHA supplementation improves lipid profiles (16). However, to our knowledge, the specific effects of prenatal DHA supplementation on markers of cardiometabolic disease risk early in life are not well studied.

We investigated the effect of maternal DHA supplementation during pregnancy on offspring glucose and lipid concentrations at age 4 y in a population with a low dietary intake of DHA.

Methods

Design of the original trial.

This study was a secondary analysis of data collected during the follow-up of offspring whose mothers participated in the Prenatal Omega-3 Fatty Acid Supplementation, Growth, and Development trial. This trial (NCT00646360) was a double-blind randomized controlled trial designed to evaluate the effects of DHA supplementation during pregnancy on offspring growth and neurologic development. A detailed description of the design and methods has been published elsewhere (17). Briefly, pregnant women in Cuernavaca, Mexico, were randomly assigned to receive 400 mg DHA (2 capsules of 200 mg from an algal source) or placebo (2 capsules containing a mixture of corn and soy oils that were similar to the DHA capsules in appearance and taste) daily from midpregnancy to delivery. Eligible women were 18–35 y of age and at 18–22 wk of gestation, and planned to deliver at the Mexican Institute of Social Security General Hospital in Cuernavaca, exclusively or predominantly breastfeed (18) for ≥3 mo, and live in the area for ≥2 y after delivery. Exclusion criteria were a high-risk pregnancy, lipid metabolism or absorption disorders, regular intake of fish oil or DHA supplements, or chronic use of certain medications. A total of 1094 women were randomly assigned. Compliance was evaluated as the total number of capsules actually consumed, expressed as a percentage of the total number expected to be consumed. Offspring were followed since birth.

The study protocol was approved by the Emory University Institutional Review Board and by the National Institute of Public Health Biosafety, Investigation, and Ethics Committees. Written informed consent was obtained from participating mothers after they received a detailed explanation of the study at baseline and during their offspring follow-up.

Follow-up study.

At 4 y of age, the offspring of the supplemented women were studied (n = 524). Nonfasting blood samples were obtained via venipuncture by trained technicians. The samples were centrifuged at 700 × g at 4°C for 15 min. Plasma and erythrocytes were separated; subsequently, the plasma was frozen and stored in aliquots of 200 μl at −70°C.

Samples were analyzed at the Salvador Zubiran National Institute of Medical Sciences and Nutrition in Mexico City. This laboratory is certified by the External Comparative Assessment Program of Laboratories of the American College of Pathologists. Outcomes were analyzed as follows: serum TG, total cholesterol (TC), HDL cholesterol, apoB, glucose, and insulin were measured with the use of commercially available reagents (Synchron; Beckman Coulter Mark); serum insulin was measured with the use of a Microparticle Enzyme Immunoassay (Abbott Laboratories); and glucose was measured with the use of an oxidase glucose test. LDL cholesterol was calculated with the use of the Friedewald formula (19). The interassay CV was <3% for all values.

To categorize abnormal values, we used the following threshold concentrations: glucose, ≥200 mg/dL (20); TGs—boys, ≥157 mg/dL, and girls, ≥155 mg/dL, representing the 97th percentile of nonfasting samples (21); TC, ≥200 mg/dL; HDL cholesterol, <40 mg/dL; non-HDL cholesterol, ≥145 mg/dL; LDL cholesterol, ≥130 mg/dL; apoB, ≥110 mg/dL; and TC:HDL cholesterol ratio, ≥3.5 (22).

Characteristics of the mothers and children.

To evaluate whether the treatment groups were balanced, we evaluated maternal characteristics obtained at baseline and offspring characteristics obtained at birth and at age 4 y.

Highest schooling attained and socioeconomic level were obtained by interview at recruitment. All anthropometric measurements were obtained by trained and standardized technicians, who used standard procedures (23). Maternal weight and height were measured at recruitment. Offspring weight and length at birth were obtained from hospital records. Weight and length were measured at 1 and 3 mo of age. Breastfeeding status at 3 mo of age was assessed by maternal report and categorized as exclusive or predominant breastfeeding or other.

Weight (to the nearest 100 g by a portable balance), height (to the nearest 1 mm with the use of a manual height gauge), and triceps and subscapular skinfold thickness were measured at age 4 y. Overweight in children was defined as a BMI-for-age z score ≥1 SD in accordance with the WHO Child Growth Standards (24). We obtained offspring dietary intake data at age 4 y with the use of a 1-wk FFQ consisting of 73 food items that was administered to the primary caregiver. Physical activity and screen time were assessed by interview. Children who were reported to be physically active for ≥2 h/d were classified as active (25).

Data analysis.

We compared maternal and offspring characteristics of children lost to follow-up or who did not provide a sample to characteristics of children with available data. Among the children included in this analysis, we compared the DHA and control groups with regard to baseline maternal characteristics and offspring characteristics at birth and at age 4 y. We used Student’s t test and the Mann-Whitney U test for normally distributed and skewed continuous variables, respectively. The chi-square test was used to compare categorical variables. Differences were considered to be significant at P < 0.05.

To evaluate the effect of maternal DHA supplementation on nonfasting serum glucose and lipid concentrations, we used multivariable linear regression models, with treatment group as an independent variable. We adjusted for the time elapsed between the blood collection and the last food intake (because most children were in a nonfasting state) and for maternal height (which was significantly different between the treatment groups). We also examined models that were adjusted in addition for sex, maternal and child overweight, gestational age, birth weight, and child activity level. Because these covariates were balanced between groups and inclusion in the model did not alter the coefficient for treatment group, we report the results while adjusting only for maternal height and time elapsed between blood collection and the last food intake.

We calculated the prevalence of abnormal serum glucose and lipid concentrations and we used chi-square analysis to evaluate differences between groups. Power estimations with the use of sample sizes available and actual variances found for each biomarker indicated the ability to detect differences of ≥0.24 SD with a power of 80% and a 2-tailed α level of 0.05%. Analyses were performed with the use of SPSS version 17.

Results

We analyzed data from 524 children (Figure 1), corresponding to 53.9% of the 973 live births. The number of participants lost to follow-up, as well as the causes of loss, did not differ significantly by treatment group. Compliance with supplementation was 88% and did not differ by treatment group. DHA concentrations in maternal plasma at delivery and cord blood were higher in the intervention group than in the control group (26). The analytic sample of children with metabolic results at 4 y of age did not differ in most maternal characteristics at random assignment or in offspring characteristics at birth from those without these data, except that mothers of children included in the analysis had fewer years of formal education than did those who were excluded (P < 0.05) (Supplemental Table 1).

FIGURE 1.

CONSORT diagram through 4 y of follow-up of the offspring of women who received 400 mg DHA/d or placebo from midpregnancy to delivery in a randomized controlled trial.

Maternal characteristics at random assignment and offspring characteristics at birth and at age 4 y are presented by treatment group for the analytic sample in Table 1. Mothers in the placebo group were taller than those in the DHA group (P < 0.05). Approximately 90% of children were in a nonfasting state (<8 h fasting) at the time of blood collection. The time elapsed since the last food intake did not differ between groups. Overall breastfeeding duration was almost 10 mo; almost 80% of children were breastfed at the age of 3 mo. Exclusive and predominant breastfeeding rates were low, and did not differ between the 2 treatment groups.

TABLE 1.

Selected maternal characteristics at random assignment, child characteristics at birth and at 4 y of age, and breastfeeding information for 524 children born to women who participated in a trial of 400 mg DHA/d from midpregnancy to delivery, by treatment group¹

	DHA (n = 276)	Placebo (n = 248)	P2
Maternal characteristics at random assignment
Age, y	26.4 ± 5.1	26.0 ± 4.7	0.42
Gestational age, wk	20.5 ± 1.9	20.5 ± 2.1	0.93
Schooling, y	11.7 ± 3.5	11.8 ± 3.6	0.53
Weight, kg	62.9 ± 12.0	63.6 ± 11.4	0.36
Height, cm	154.6 ± 5.8	155.4 ± 5.5	0.03
BMI, kg/m2	26.3 ± 4.4	26.3 ± 4.5	0.89
Child characteristics at birth
Weight, kg	3.2 ± 0.5	3.2 ± 0.5	0.67
Length, cm	50.3 ± 2.4	50.4 ± 2.2	0.72
Gestational age, wk	38.9 ± 1.94	39.1 ± 1.8	0.94
Breastfeeding duration, mo	9.7 ± 8.0	10.5 ± 9.7	0.15
Exclusive breastfeeding at 3 mo	15.5	14.2	0.91
Predominant breastfeeding at 3 mo	15.1	17.3
Some breastfeeding at 3 mo	53.6	52.4
No breastfeeding at 3 mo	15.9	16.0
Child characteristics at 4 y of age
Age, y	4.2 ± 0.3	4.2 ± 0.3	0.7
Male	57.6	50.4	0.10
Nonfasting blood samples	91.3	92.3	0.67
Time since last meal, h	4.2 ± 3.3	4.1 ± 3.1	0.83
Active >2 h/d	58.4	61.6	0.34
Weight, kg	16.3 ± 2.4	16.2 ± 2.2	0.99
Height, cm	102.4 ± 4.4	102.2 ± 4.1	0.80
BMI, kg/m2	15.5 ± 1.5	15.5 ± 1.4	0.77
Waist circumference, cm	52.8 ± 4.0	52.7 ± 3.9	0.75
Triceps skinfold, mm	8.8 ± 2.2	8.7 ± 2.0	0.99
Subscapular skinfold, mm	6.6 ± 2.1	6.4 ± 1.7	0.49
HAZ	−0.5 ± 0.9	−0.5 ± 0.9	0.99
BAZ	0.1 ± 1.0	0.1 ± 1.0	0.77
Overweight or obese3	16.3	14.1	0.49

¹Values are means ± SDs or percentages. BAZ, BMI-for-age z score; HAZ, height-for-age z score.

²Assessed by the Mann-Whitney test or chi-square test.

³BAZ ≥1 SD.

Offspring anthropometric measurements and indexes were similar across groups. Mean height-for-age z score was lower than the 2006 WHO growth standards (24), whereas BMI-for-age z score was close to the reference. Approximately 60% of children were classified as being active. The consumption of dietary sources of n–3 long-chain PUFAs was low, with ∼40% of children having reported consumption of fresh fish, tuna fish or sardines 1–3 times/mo, with no differences by treatment group.

Between-group differences for those consuming DHA compared with placebo for nonfasting serum insulin, glucose, and lipid concentrations, adjusted for maternal height and time since last food intake, are presented in Table 2. There were no significant differences between groups in any of the metabolic markers measured.

TABLE 2.

Effect of maternal supplementation with 400 mg DHA/d from midpregnancy to delivery on nonfasting serum glucose and lipid concentrations in the offspring at age 4 y¹

	DHA (n = 276)	Placebo (n = 248)	P2	Difference (95% CI)3
Insulin, μU/mL	9.6 ± 9.4	9.0 ± 8.2	0.83	0.62 (−0.88, 2.11)
Glucose, mg/dL	93.2 ± 10.3	94.0 ± 10.5	0.40	−0.67 (−2.46, 1.11)
TGs, mg/dL	120 ± 66.4	121 ± 60.5	0.51	0.21 (−10.93, 10.52)
Total cholesterol, mg/dL	159. ± 23.6	157 ± 27.0	0.34	1.73 (−2.63, 6.09)
HDL cholesterol, mg/dL	52.0 ± 10.4	51.4 ± 9.7	0.49	0.66 (−1.07, 2.39)
Non-HDL cholesterol, mg/dL	106.7 ± 20.4	104.8 ± 21.4	0.31	1.77 (−1.83, 5.37)
LDL cholesterol, mg/dL	83.9 ± 20.4	820.1 ± 23.4	0.27	1.62 (−2.21, 5.45)
apoB, mg/dL	76.9 ± 14.3	76.8 ± 16.3	0.90	−0.15 (−2.78, 2.48)
Total-to–HDL cholesterol ratio	3.1 ± 0.56	3.1 ± 0.57	0.69	0.01 (−0.09, 0.11)

¹Values are means ± SDs, unless otherwise indicated.

²From Mann-Whitney test to compare mean differences.

³Estimates are the differences between the DHA and placebo groups and are derived from multivariable linear regression models, controlling for the time since last food intake and the maternal height.

Regarding abnormal glucose and lipid concentrations, we identified no children with nonfasting glucose ≥200 mg/dL in either group. The proportions of children with high TC, LDL cholesterol, or apoB were low (4.4%, 1.6%, and 1.5% respectively). However, 22% of the children had high serum TG concentrations and 11% had low HDL cholesterol. The proportion of children with high non-HDL cholesterol was low (3.8%), but the prevalence of values for a TC:HDL cholesterol ratio >3.5 was 22%. The prevalence of children with abnormal serum lipid concentrations did not differ by treatment group (P > 0.05) (Table 3).

TABLE 3.

Nonfasting serum glucose and lipid concentrations exceeding threshold values for 524 children at 4 y of age born to women who participated in a trial of 400 mg DHA/d from midpregnancy to delivery, by treatment group¹

	Exceeding threshold values2
	DHA (n = 276)	Placebo (n = 248)	P3
Glucose, mg/dL	0 (0)	0 (0)	—
TGs, mg/dL	22.8 (63)	21.4 (53)	0.69
Total cholesterol, mg/dL	4.4 (11)	4.3 (12)	0.96
HDL cholesterol, mg/dL	11.3 (28)	11.2 (31)	0.98
Non-HDL cholesterol, mg/dL	4.2 (12)	3.2 (8)	0.51
LDL cholesterol, mg/dL	1.9 (5)	1.3 (3)	0.57
apoB, mg/dL	1.6 (4)	1.4 (4)	0.87
Total-to–HDL cholesterol ratio	21.8 (54)	21.7 (60)	0.99

¹Values are n (%).

²Threshold values are as follows: glucose, ≥200 mg/dL (20); TGs: boys, ≥157 mg/dL, and girls, ≥155 mg/dL (21); total cholesterol, ≥200 mg/dL; HDL cholesterol, <40 mg/dL; LDL cholesterol, ≥130 mg/dL; apoB, ≥110 mg/dL; total-to–HDL cholesterol ratio, ≥3.5 (22).

³Differences between groups were evaluated with a chi-square test.

Discussion

In a follow-up of a large double-blind randomized controlled trial of supplementation with 400 mg/d algal DHA from midpregnancy to delivery, we found no overall effects of supplementation on serum lipid, glucose, and insulin concentrations in the offspring at age 4 y. Moreover, we did not find differences between treatment groups in the proportion of children with abnormal concentrations of glucose or lipid markers.

The study was conducted with high fidelity. The women participating in the trial had a low dietary intake of DHA (24) and were highly compliant with the intervention, and the intervention improved cord blood and breast milk DHA concentrations at 1 mo postpartum (27). The field personnel who collected the data and blood samples, the participants, those conducting the laboratory analysis, and the principal author were all blinded to the supplementation groups. The outcome variables were precisely measured, with the interassay CV <3% for all values. The intervention groups remained balanced across many maternal characteristics at baseline and child characteristics at birth and at 4 y of age. Finally, our study sample size provided a statistical power of 80% to detect differences of ≥0.24 SD, which are generally considered to be of small magnitude (28), providing reassurance that large effects would not be missed. Therefore, we conclude that the study was appropriately designed and well conducted.

Nutritional alterations during critical developmental windows may induce metabolic programming (29, 30). n–3 Long-chain PUFAs and their metabolites (such as DHA) are signaling molecules that are involved in appetite; furthermore, these compounds have regulatory effects on the gene expression of enzymes involved in lipid and glucose metabolism (31). Despite the biological plausibility of a link between prenatal DHA supplementation and metabolic programming, we did not find evidence for such a link.

To the best of our knowledge, only one previous human study has evaluated the effect of prenatal DHA supplementation on serum lipid concentrations later in life (32). That study was conducted in Denmark. In that study, a total of 533 pregnant women were randomly assigned to receive four 1-g fish oil capsules, 4 olive oil capsules, or neither from gestational week 30 through delivery. A total of 243 offspring were followed at 18–19 y of age; there was no difference between groups in TC, LDL cholesterol, HDL cholesterol, TGs, apoA, or apoB. However, when the analysis was restricted to the group of offspring whose mothers reported low fish intake (n = 46), a trend toward a healthier lipid profile (less TGs, apoB, and LDL cholesterol) was demonstrated in the fish-oil group, but the difference was not significant, likely because of a lack of power (32). Our study confirms the general null finding and extends it by 1) isolating the potential effect of DHA provided as the sole supplement; 2) adding an earlier follow-up, giving the opportunity to study what happens at an earlier age; 3) increasing by 2-fold the number of offspring studied, thus enhancing study power; and 4) adding a different context, in which maternal DHA intake is consistently low. Because maternal DHA intake in this study population was low, it will be of interest to study these offspring in the future to explore whether a trend toward a healthier lipid profile emerges, as suggested by the offspring of mothers who had a low intake of fish in the Danish study.

We did not find any previous studies on the effect of prenatal DHA supplementation on serum glucose concentrations. Studies have shown that breastfeeding may lower fasting serum glucose concentrations (33). An inverse correlation between fasting serum glucose concentrations and percentage of DHA in muscle phospholipids has been reported in breastfed infants (34). In our study, we found no effect on serum glucose and insulin concentrations in the offspring of women supplemented with algal DHA.

The prevalence of abnormal nonfasting lipid concentrations in our sample was lower than in children aged 1–4 y in northern Mexico (high TGs, 22% compared with 37%; high TC, 4.4% compared with 15%; and low HDL cholesterol, 11.3% compared with 29.7%) (35), and in Chilean children aged 4 y, among whom there was a 20% prevalence of high LDL cholesterol and a 50% prevalence of low HDL cholesterol (compared with 1.6% and 11.3%, respectively, in our sample), even though both of those studies obtained fasting samples (36).

Given the difficulty of measuring fasting glucose in young children, we obtained and analyzed nonfasting blood samples. The distribution of children by fasting state and time since last meal was similar in both groups. In addition, we adjusted the analyses for time since last food intake. For these reasons, the effects of prenatal DHA supplementation on lipid and glucose concentrations found in this study are unlikely to be biased because of differences in fasting conditions between treatment groups.

We used threshold values developed for the fasting state. However, the nonfasting lipid profile is well associated with cardiovascular disease risk, and these markers have been used in both fasting and nonfasting states in previous research (37, 38). Furthermore, HDL cholesterol and apoB concentrations are not affected by fasting or nonfasting status (37). For glucose, the cutoff of 200 mg/dL is widely used to evaluate nonfasting glucose concentrations (20). The cutoff for TGs is above the 97.5th percentile of nonfasting children from 1 to 5 y of age (21). Overall, the proportion of children with abnormal concentrations was low, even allowing for the potential bias resulting from the use of nonfasting samples.

The placebo used a corn-soy blend. We could not find any biological evidence suggesting an effect of prenatal corn and soy oil supplementation on lipid and glucose markers later in life. Furthermore, the amount of oil in the placebo (400 mg) was trivial in relation to normal dietary intake. We conclude that the amount of oil provided to the women in the placebo group was unlikely to affect our results.

In this large randomized controlled trial of supplementation with 400 mg DHA/d from midpregnancy to delivery, we did not observe effects on serum glucose and lipid concentrations in the offspring at age 4 y. Follow-up of these children will provide further information regarding a potential effect later in life.