Maternal High-Dose DHA Supplementation and Neurodevelopment at 18–22 Months of Preterm Children Free

The MOBYDIck trial is a 2-arm, randomized, double-blind, placebo-controlled trial that was conducted in 16 Canadian neonatal ICUs between June 2015 and April 2018. In MOBYDIck, lactating mothers who delivered before 29 weeks’ gestation were randomly assigned to receive a high dose of DHA (ie, 1.2 g DHA daily to achieve ∼1% of DHA in breast milk)^11,12  or placebo capsules within 72 hours of delivery until 36 weeks’ postmenstrual age. Mothers were excluded from the trial if they were taking supplements containing >250 mg per day of DHA in the 3 months before birth or if their neonate had a major congenital or chromosomal anomaly. Mothers were randomized to receive DHA or placebo in a 1:1 ratio, according to a computer-generated randomization list. Because the randomization was performed at the mother level, neonates from multiple birth were assigned to the same group. In the absence of sufficient maternal breast milk available, neonates received donor milk or formula for preterm neonates as per standard clinical care. The previously published trial protocol¹⁰  was approved by Health Canada and the research ethics boards of all sites. The primary outcome for this trial was bronchopulmonary dysplasia (BPD)-free survival at 36 weeks’ postmenstrual age.¹⁰  Written informed consent was obtained from the participating mothers.

Neurodevelopmental outcomes were assessed at 18 to 22 months’ CA with the use of neurologic examination and the Bayley Scales of Infant and Toddler Development third edition (Bayley-III).¹³  Children were assessed by experienced clinicians in neonatal follow-up program clinics affiliated with the Canadian Neonatal Follow-Up Network.¹⁴  All participants, researchers, and clinicians responsible for the neurodevelopmental assessment were blinded to group allocation. Children with at least 1 outcome of interest available at 18 to 22 months’ CA (ie, death, Bayley-III score, cerebral palsy with Gross Motor Function Classification System assessment, and hearing or visual assessment) were included in the analysis.

The primary outcome at 18 to 22 months’ CA was the neurodevelopmental performance in preterm children who survived at 18 to 22 months’ CA, as assessed with the Bayley-III cognitive, language, and motor composite scores. The Bayley-III provides standardized scores with a mean of 100 and a SD of 15. Children who attended neurodevelopmental assessment but were completely untestable because of severe disability were assigned a score of 3 SD below the mean (ie, 55).

The major secondary outcomes at 18 to 22 months’ CA were (1) death or significant neurodevelopmental impairment (NDI), (2) death, and (3) significant NDI. Other secondary outcomes included the Bayley-III in category for each component (cognitive, language, and motor) defined by scores <70 (ie, significant impairment) and <85 (ie, at least mild impairment). Death referred to deaths from any causes between randomization and 18 to 22 months’ CA. Significant NDI was considered present in survivors if the child presented 1 or more of the following:¹⁵  Bayley-III cognitive, language, or motor composite score <70, cerebral palsy with Gross Motor Function Classification System level ≥3, and hearing or visual impairment. Children were considered to have hearing impairment when they required hearing aid or cochlear implant. Visual impairment was defined as bilateral visual impairment with no functional vision. In children with missing data for cerebral palsy classification, hearing or visual assessment at 18 to 22 months’ CA, if a complete Bayley-III (ie, cognitive, language, and motor assessment) was performed with all scores ≥70 and if there was no known hearing or visual impairment, then significant NDI was considered absent. However, for children with an incomplete Bayley-III assessment and no known impairment, significant NDI was considered unknown.

The MOBYDIck trial showed that maternal DHA supplementation did not improve BPD-free survival compared with placebo.¹⁰  At the interim analysis of the MOBYDIck trial, the data and safety monitoring board recommended termination of the recruitment because of concern for harm from DHA supplementation, as also suggested in the N-3 Fatty Acids for Improvement in Respiratory Outcomes trial.^10,16  Thus, the current study includes the children with outcomes available at 18 to 22 month’ CA among the 528 children enrolled in the original trial.

The treatment effect on the primary outcome was measured using mixed linear models and expressed as mean differences with 95% confidence interval (CI). For secondary outcomes, the treatment effect was expressed as a relative risk (RR) with 95% CI, which was measured using log-binomial regression in generalized estimating equation models. All analyses were adjusted for study site and clustering of children because of multiple birth. As previously reported in the original trial, no imputation was performed for missing data.¹⁰ 

Subgroup analysis for primary and major secondary outcomes were prespecified for sex and gestational age (GA) at birth (ie, <27 vs ≥27 weeks’ gestation). An interaction term was examined for each model to describe treatment effect modification by sex and GA at birth on outcomes with P <.1 considered significant. Separate estimate effects of treatment according to the prespecified subgroup for sex and GA were presented with stratified analysis.

Post hoc analyses were conducted to investigate the potential interaction between the treatment and the mode of delivery on Bayley-III scores and major secondary outcomes. As previously reported, there was an imbalance in the mode of delivery between treatment groups. For the primary outcome of BPD-free survival, the effect of DHA was modified by the mode of delivery.¹⁰  In addition, because severe intraventricular hemorrhage (IVH) occurred more rarely in the DHA group compared with the placebo group, post hoc sensitivity analysis excluding participants with severe IVH (defined as grade 3 or 4 IVH)¹⁷  were conducted.

All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc). A significance threshold of 2-sided P ≤ .05 was considered statistically significant.

Among the 528 children (and their 461 mothers) who were enrolled in the original trial, 457 (86.6%) had outcomes available at 18 to 22 months’ CA and were included in the analysis (Fig 1). In total, 234 children (199 mothers) were assigned to the DHA group and 223 children (200 mothers) were assigned to the placebo group. The maternal adherence to the intervention protocol was similar in both groups resulting in a good contrast in the DHA percentage of total fatty acids in expressed breast milk samples on postnatal day 14 between groups (Table 1). Volumes of maternal breast milk received were also similar for both groups (Supplemental Fig 2). There were no significant differences among the baseline characteristics of the children with and without data available at 18 to 22 months’ CA. However, mothers of children with follow-up data available at 18 to 22 months were older, with higher levels of education and a smaller proportion of them smoked during pregnancy in comparison with those without follow-up data available (Supplemental Tables 5 and 6). Neurodevelopmental follow-up assessments were completed between March 2017 and July 2020.

Most of the baseline characteristics of the children with follow-up data at 18 to 22 months’ CA in the DHA and placebo groups were similar (Table 2). However, more mothers delivered via cesarean in the DHA group (68.8%) compared with the placebo group (54.5%). In addition, severe IVH occurred less frequently in children of the DHA group (8.5%) in comparison with children of the placebo group (17.0%).

The Bayley-III cognitive, language, and motor composite scores did not differ between children in the DHA group and the placebo group (Table 3). Because of the lower frequency of severe IVH in the DHA group, a sensitivity analysis was performed excluding participants with severe IVH (Supplemental Table 7). The effect of DHA on Bayley-III scores did not change when excluding these children.

The results for the effect of DHA on death or significant NDI and its components are shown for the children with adequate data for these outcomes in Table 4. Death or significant NDI occurred in 51 of 205 children (24.9%) in the DHA group compared with 60 of 196 children (30.6%) in the placebo group (RR 0.86, 95% CI 0.62–1.20, P = .37). The effect of DHA on each domain of the Bayley-III categorized as significant impairment or at least mild impairment is also presented in Table 4.

Subgroup analysis stratified by sex and GA at birth for the primary and major secondary outcomes are presented in Supplemental Tables 8 and 9. The effect of DHA on Bayley-III language score was different according to GA at birth (interaction P = .07). For neonates born <27 weeks’ gestation, those exposed to DHA had higher language score (mean 89.3, SD 16.8) compared with the placebo group (mean 83.1, SD 17.6, mean difference 5.06, 95% CI 0.08–10.03, P = .05).

Post hoc analyses for the primary and secondary outcomes stratified by mode of delivery are presented in Supplemental Table 10. Results showed a differential effect of DHA according to the mode of delivery on the motor composite score (interaction P = .03) and on death or significant NDI (interaction P = .05). Rates of death or significant NDI were lower among children exposed to DHA and born vaginally (15.8%, n = 9 of 57) in comparison with children in the placebo group and born vaginally (33.3%, n = 29 of 87) (RR 0.48, 95% CI 0.24–0.94, P = .03).

In this randomized, controlled trial of maternal DHA supplementation for breastfed children born before 29 weeks’ gestation, we found no significant difference between groups in the neurodevelopmental scores at 18 to 22 months’ CA. However, in the subgroup of neonates born before 27 weeks’ gestation, a benefit of DHA supplementation on language was observed with a Bayley-III language score higher by 5.1 points, a clinically important difference.

Speech and language disorders are common in very preterm neonates, with significant language impairment at 18 to 24 months’ CA reported in 11.7% of Canadian children born <29 weeks’ gestation.¹⁸  The structural and functional neural networks involved in language abilities are complex and involve different brain regions, including the temporal and prefrontal cortices.¹⁹  These centers for language development are particularly vulnerable during the preterm period,²⁰  and higher DHA levels might have contributed to optimizing their development in the most immature neonates.

Our study emphasizes the need to better delineate the characteristics affecting DHA responsiveness to customize nutritional practices for individual neonates and improve long-term neurodevelopmental outcomes. From preclinical studies, we recognize the key role of DHA in brain development. As such, DHA modifies the basic properties of brain cell membranes²¹  and is implicated in synaptogenesis,²²  regulation of glial cells,²³  and white matter maturation.^24,25  In addition, clinical brain imaging studies have suggested an association between DHA and greater cortical connectivity,²⁶  increased white matter volume,²⁶  improved white matter development,^7,26–28  and markers of improved neuronal integrity on magnetic resonance spectroscopy.²⁶  Yet, not all preterm neonates benefit equally from DHA supplementation.

We and others have identified that the most immature neonates might benefit preferentially from DHA supplementation.^10,29–31  As such, a subgroup analysis from the DHA for the Improvement of Neurodevelopmental Outcome trial²⁹  showed improved cognitive scores with a DHA supplementation only among the smaller preterm neonates (ie, <1250 g). In addition, Hellström et al recently reported a 50% reduction in severe retinopathy of prematurity without significant adverse effect among neonates born <27 weeks’ gestation supplemented with a combination of arachidonic acid and DHA.³¹  Importantly, we and Hellström et al both reported a higher prevalence of BPD among neonates born ≥27 weeks’ gestation after LCPUFA supplementation compared with a placebo.^10,31  Hence, there is evidence to suggest potential DHA benefits for the most immature preterm neonates on several outcomes, whereas for neonates born ≥27 weeks’ gestation, there may not be any benefits from these DHA interventions. As highlighted by a recent Cochrane review⁸  assessing the impact of LCPUFA-supplemented formula on the neurodevelopment of preterm neonates, more studies are warranted to determine whether DHA supplementation might be beneficial in the most extremely preterm neonates.

Although we did not identify a sex-specific response to DHA for neurodevelopmental outcomes in the MOBYDIck cohort, the DHA for the Improvement of Neurodevelopmental Outcome trial suggested an improvement in cognitive outcomes of females only.²⁹  Otherwise, in addition to GA at birth, other factors such as exposure to inflammation and oxidative stress might have modified the effect of DHA. Interestingly, our post hoc analyses suggested a beneficial effect of DHA for neurodevelopmental outcomes of neonates born vaginally. Although these findings are exploratory and might be because of chance, the mode of delivery is influenced by prenatal conditions which may have modulated the response to DHA. In addition, cesarean delivery is associated with increased early oxidative stress,^32–35  along with inflammatory and immune changes, which might influence the treatment response to DHA. Hence, as we aim to optimize nutrition for better outcomes in preterm neonates, a more individualized approach according to different neonatal characteristics should be considered.

Besides the specific characteristics of preterm children, other factors related to the treatment, such as the dose of DHA, mode of administration, and supplementation duration, should be considered. In this trial of maternal DHA supplementation during lactation, approximately half of the neonates were fed exclusively with maternal breast milk at 36 weeks’ postmenstrual age. In addition, the ratio of DHA and arachidonic acid^36,37  might have influenced the effect of DHA on neurodevelopmental outcomes. Indeed, it has been hypothesized that a supplementation with “DHA-only” might have modified the fragile balance between the different LCPUFAs (including arachidonic acid and eicosapentaenoic acid), which could have counterbalanced the expected beneficial effects of high-dose DHA supplementation.³⁶  This rationale might provide a pathophysiological explanation for the relative disappointment of well-designed clinical trials of DHA for neurodevelopment of preterm neonates despite the strong preclinical evidence.

However, with the complexity of brain development and the multiple factors affecting the neurodevelopmental trajectory of preterm neonates, it is unlikely that supplementation with a single nutrient alone will substantially impact a child’s development. Instead, a combination of strategies targeting different aspects of preterm nutrition such as lipid and energy intake, and earlier enteral feeding initiation, along with individualized LCPUFA supplementation, are more likely to be successful.^2,38 

The overall clinical benefits and risks associated with high-dose DHA supplementation should be considered together to determine whether DHA supplementation might represent a clinically applicable nutritional strategy. Although previous studies have identified beneficial effects of DHA on IVH,⁷  growth,³⁹  and retinopathy of prematurity,³¹  the higher rates of BPD^10,16  observed after supplementation are concerning. Indeed, BPD has lifelong consequences with long-term respiratory and neurodevelopmental consequences.⁴⁰  However, both the present report and data from a subset of the N-3 Fatty Acids for Improvement in Respiratory Outcomes trial, a recent trial also assessing the effect of enteral DHA supplementation in preterm neonates, do not suggest worse neurodevelopmental outcomes at 18 to 22 months’ CA with high-dose DHA supplementation.⁴¹  In contrast, a benefit of DHA on language was observed in the most preterm neonates, those at highest risk of BPD.

Because of the early termination of the parent trial, the sample was smaller than planned, which limits study interpretation. Additionally, the primary outcome of the study included 3 distinct outcomes (cognitive, language, and motor scores) and we did not adjust P values for multiple comparisons, which may have inflated type I error risk. Hence, effects on neurodevelopmental outcomes should be interpreted with caution. Nevertheless, the MOBYDIck trial is the largest study to report on the effect of DHA supplementation on the neurodevelopmental performance of preterm neonates born <29 weeks’ gestation. Importantly, our cohort is distinguished by highly educated mothers with high familial income, which might have contributed positively to neurodevelopmental outcomes at 18 to 22 months’ CA.⁴²  Additionally, we also recognize the limitations of the Bayley-III, a standardized global neurodevelopmental assessment designed primarily to screen and identify children at higher risk of developmental delays, which may not be sensitive enough to fully detect neurodevelopmental differences. In addition, the Bayley-III does not include an assessment of behavioral issues and executive functioning impairment, to which preterm children are particularly vulnerable, and in which DHA may be a modifiable contributor as suggested by preclinical and clinical trials (see Cardoso et al⁴³  and Gould et al⁴⁴  for reviews). Lastly, approximately half of the participants in this trial received intravenous DHA-rich lipids as per the standard of clinical care, leading to a higher supplementation of DHA in a subgroup of participants. Yet, because the use of DHA-rich lipids emulsion did not differ between the groups, it is unlikely that their use is implicated in the findings of the study.

There was no significant difference in the neurodevelopmental outcomes at 18 to 22 months’ CA between children born very preterm exposed to maternal high-dose DHA supplementation during the neonatal period and those exposed to placebo.

Bayley-III

Bayley Scales of Infant and Toddler Development third edition

BPD

bronchopulmonary dysplasia

CA

corrected age

CI

confidence interval

DHA

docosahexaenoic acid

GA

gestational age

IVH

intraventricular hemorrhage

LCPUFA

long-chain polyunsaturated fatty acid

MOBYDIck

Maternal Omega-3 Supplementation to Reduce Bronchopulmonary Dysplasia in Very Preterm Infants

NDI

neurodevelopmental impairment

RR

relative risk