Improved Cognitive Development Among Preterm Infants Attributable to Early Supplementation of Human Milk With Docosahexaenoic Acid and Arachidonic Acid
OBJECTIVE. The objective of our study was to evaluate the effect of supplementation with docosahexaenoic acid and arachidonic acid for human milk-fed preterm infants. The primary end point was cognitive development at 6 months of age.
METHODS. The study was a randomized, double-blind, placebo-controlled study among 141 infants with birth weights of <1500 g. The intervention with 32 mg of docosahexaenoic acid and 31 mg of arachidonic acid per 100 mL of human milk started 1 week after birth and lasted until discharge from the hospital (on average, 9 weeks). Cognitive development was evaluated at 6 months of age by using the Ages and Stages Questionnaire and event-related potentials, a measure of brain correlates related to recognition memory.
RESULTS. There was no difference in adverse events or growth between the 2 groups. At the 6-month follow-up evaluation, the intervention group performed better on the problem-solving subscore, compared with the control group (53.4 vs 49.5 points). There was also a nonsignificant higher total score (221 vs 215 points). The event-related potential data revealed that infants in the intervention group had significantly lower responses after the standard image, compared with the control group (8.6 vs 13.2). There was no difference in responses to novel images.
CONCLUSIONS. Supplementation with docosahexaenoic acid and arachidonic acid for very preterm infants fed human milk in the early neonatal period was associated with better recognition memory and higher problem-solving scores at 6 months.
Adequate nutrition during infancy and early childhood is essential for optimal growth, cognitive functioning, and health.1 Preterm infants have increased morbidity and mortality rates in early life and higher prevalence rates of school problems and neurodevelopmental impairments, compared with term infants.2,3 Approximately 50% of preterm infants have psychological complaints, and 25% shows signs of attention-deficit/hyperactivity disorder.4 Low birth weight also is associated with increased risk of cardiovascular diseases and diabetes mellitus in adulthood.5 Some of these disorders may be related to fetal and neonatal nutrient supply.
Present recommendations for preterm infant nutrition are designed to approximate the growth and development of a normal fetus of the same postconceptional age. Docosahexaenoic acid (DHA) and arachidonic acid (AA) are important for growth and neurodevelopment of the fetus and preterm infants.6 During pregnancy, DHA and AA are transferred to the fetus by specific placental proteins and are incorporated into cell membranes of all tissues of the body, particularly those of the retina and central nervous system. Preterm infants are prematurely deprived of this supply. The contents of DHA and AA in human milk vary, depending on the maternal diet.7,8 A previous randomized study showed that preterm infants receiving human milk had higher IQ values than did formula-fed infants at 8 years of age, and this was attributed to the higher content of DHA in human milk.9 Most preterm formulas are now supplemented with DHA and AA to approximately the same levels as found in human milk (0.20%–0.35% of total fatty acids). Several randomized, clinical trials showed beneficial effects on growth, visual function, and cognitive development, but all of those studies were performed with formula-fed infants.10–13
Human milk supplies less DHA and AA than the fetus receives in utero. Even if full enteral intake of human milk is achieved, the calculated intake of DHA is only between 13 and 26 mg/day, which is clearly below the estimated uterine accretion rate of ∼50 mg/day.14 The effect of DHA and AA supplementation for human milk-fed, preterm infants is not known.
One major impediment for further progress in this field is the lack of reliable valid methods for evaluating cognitive development in infants. Global measures of cognitive development, such as the Bayley Scales of Infant Development and the Brunet-Lezine scale, have often been used in clinical trials. These tests were originally designed to identify infants who developed atypically, and they have limited specificity for outcomes related to intake of DHA and AA.15 Assessment of event-related potentials (ERPs) represents another method of investigating cognitive processes among infants.16 This is a noninvasive measure of brain activity derived from standard electroencephalographic (EEG) recordings, which can provide information about changes time-locked to physical or cognitive events. The negative central (NC) component is a much-studied electrophysiological component that is thought to represent important cognitive processes such as attention and recognition memory. The amplitude of the NC component is larger for new, presumably interesting stimuli and decreases as a stimulus is repeated.17,18
The aim of our study was to evaluate the effect of supplementation with DHA and AA (48 mg/kg per day of each) on human milk-fed, very low birth weight (VLBW) infants (birth weight: <1500 g) in the early neonatal period. The primary end point was cognitive development, measured with a global test of development (Ages and Stages Questionnaire), as well as an ERP index of recognition memory.
All VLBW infants born between December 2003 and November 2005 at Rikshospitalet-Radiumhospitalet Medical Center, Akershus University Hospital, Buskerud Hospital, and Vestfold Hospital in Norway were eligible for inclusion. Infants with major congenital abnormalities or cerebral hemorrhage (grade 3 or 4, as determined through ultrasonography) were not included in the study. Written informed consent was obtained from the parents, and the study was approved by the regional ethics committee. The infants were assigned randomly to either the intervention group or the control group by using computer-generated randomization schedules. The randomization was performed separately for each of the neonatal centers, using blocks of 16 participants, without stratification. Included infants were given progressive study numbers corresponding to the number on the bottles of study oil. All personnel recruiting infants, parents, and hospital staff members were blinded to the group allocation.
DHA and AA Supplementation
The infants received human milk (from either the mother or a donor) from the first or second day after birth. As enteral feeding was increased, the milk was fortified with proteins, minerals, vitamins, iron, and folic acid according to the local routines. In addition, the infants received a daily dose of 0.5 mL of study oil per 100 mL of human milk. The intervention group received a study oil with AA and DHA as triacylglycerol (Martek Biosciences, Columbia, MD). The study oils were dispersed in a mixture of soy oil and medium-chain triglyceride oil (Table 1) at the hospital pharmacy, packed, and numbered according to the randomization list. The control group received the same mixture of soy oil and medium-chain triglyceride oil as the study group but without DHA or AA.
The fatty acid compositions of the human milk and the study oils were analyzed, and the results are given in Table 1. The intervention oil contained 6.9% (wt/wt) DHA and 6.9% (wt/wt) AA, providing 32 mg of DHA and 31 mg of AA in 0.5 mL of study oil added to 100 mL of human milk. This supplementation more than doubled the quantities of DHA and AA, compared with unfortified human milk. The added study oil was sonicated into human milk and given to the infants by gavage feeding. The intervention started when the infant received most of his or her nutrients enterally (>100 mL of human milk per kg of body weight per day) and continued until the infant was discharged from the hospital or the bottle of 100 mL of study oil was empty (at 63 days of age, on average). The last week before discharge, when breastfeeding was established, the study oil was given as a fixed dose of 1 mL twice a day. Infants who were not breastfeeding at the time of discharge changed from donor milk to term formula (most commonly, Nan 1 [Nestlé, Sandvika, Norway]) during the last days before discharge.
Maternal characteristics were obtained through interviews. Growth data on infants were obtained from the medical charts each week during hospitalization. Samples of breast milk were collected from each mother 4 weeks after birth and were analyzed for fatty acid patterns through gas-liquid chromatography with flame ionization detection. Nutrient intake was calculated from data on parenteral nutrition, human milk, formulas, and oral supplements obtained from the medical charts, by using a computerized database (Beregn; Department of Nutrition, University of Oslo). All adverse events were recorded on a separate form for each participant, in the medical charts, throughout the study. In addition, all diagnoses attributable to illnesses during hospitalization were obtained from the medical charts.
Blood Sampling and Analyses
Blood samples (venous or capillary) were obtained from the infant (1 mL) at the time of admission to the study and at discharge. The blood samples were collected in EDTA-containing containers at the start and end of the intervention; they were then centrifuged and stored at −80°C until analyses. The plasma samples were analyzed for fatty acid patterns through gas-liquid chromatography with flame ionization detection.19
Ages and Stages Questionnaire
The primary end point was cognitive development at corrected age of 6 months, which was evaluated with the Ages and Stages Questionnaire, a parent-administered standardized questionnaire originally developed in the United States.20 This instrument for measuring mental and motor development includes 30 items designed to assess the infant's development in the areas of communication, gross motor, fine motor, problem-solving, and personal-social skills. The parents or other caregivers are asked whether the child performs the described behavior, with 3 possible responses (yes, sometimes, or not yet). One of the advantages is that the questionnaire requires much less time than instruments that require direct examination. The questionnaire has been translated into Norwegian and validated with Norwegian infants.20,21
We also performed electrophysiological recordings related to recognition memory. A single investigator, who was also blinded to the intervention, tested the infants at a mean age corrected for gestation of 6 months and 4 days. The stimuli included a pseudo-randomized series of colorful images, in which a standard image (a cartoon ball) was shown in 70% of the presentations and novel images (different cartoon toys and animals) were shown in 30% of the presentations. The novel images were never repeated, and 1 novel image did not follow another novel image. This paradigm elicits a negative, long-lasting, ERP deflection that is larger for novel stimuli. This negativity decreases with repetitions, presumably because of recognition. Normal infants quickly recognize the standard cartoon, whereas the novel images continue to elicit large negative amplitudes.17 Therefore, it is possible to infer memory function from the difference between ERPs recorded in response to a unique image and those from standard images.22 Our hypothesis was that the intervention group would show a more-marked decrease of the negative amplitude with repetition of the standard image, compared with the control group, whereas there would be no difference for new images.
During the EEG recordings, the infants were seated on their parents' laps and the stimuli were presented on a 30- × 40-cm computer monitor, in an electrically shielded, sound-insulated, experimental room. The registration lasted for ∼10 minutes. The parents were instructed not to speak and not to direct the child's attention toward the monitor. With the use of an Easycap (Easycap, Herrsching, Germany), 6 active electrodes were attached to standard sites on the head, according to the International 10–20 System (ie, F3, F4, C3, C4, P3, and P4, referenced to the mastoid bones).
EEG traces were recorded by using Neuroscan software, and the signals were amplified through a Neuroscan Nuamps amplifier (Compumedics Neuroscan, El Paso, TX) and digitized at a rate of 500 Hz. Continuous EEG traces were scanned for artifacts, and segments where the EEG signals exceeded 150 μV were excluded from additional analyses. EEG results were then band-pass filtered from 0.5 to 30 Hz, with a 12-dB roll-off. Epochs from −100 to 1500 milliseconds were formed and baseline-corrected by using the prestimulus interval. Averages were calculated separately for the standard and novel stimuli. The mean aggregated amplitude for the frontal and central electrodes in the interval of 400 to 650 milliseconds was used for statistical analyses.
Calculations were performed by using SPSS 14.0 (SPSS Inc, Chicago, IL). Continuous variables are presented as mean and SD and were tested with t tests, whereas categorical variables are presented as percentages and were tested with χ2 tests. Some data (eg, feeding data and duration of assisted ventilation) were not normally distributed; those data are presented as medians and interquartile ranges (25th to 75th percentile values) and were tested with nonparametric methods. For repeated measurements such as weight and blood sample measurements, analyses of variance were used. Power calculations were performed with Ages and Stages Questionnaire scores as the end point. We estimated that 63 participants in each group would be satisfactory to have 80% power to detect a difference of 21 points (corresponding to 0.5 SD), with a mean value of 260 points. Statistical significance was defined as P < .05.
Of a cohort of 222 consecutively born VLBW infants, 141 (64%) were included in the study. Fifty-nine infants did not meet the inclusion criteria, and 22 parents refused to participate. Twelve infants were excluded (6 in each group); therefore, 129 infants completed the intervention. The reasons for these dropouts were possible adverse events (n = 7), prolonged parenteral feeding (n = 2), death (n = 2), congenital abnormalities (n = 1), and parents declined (n = 1) (Fig 1). Of the 129 completing infants, 62 were in the intervention group and 67 in the control group.
There were no significant differences in gender, gestational age, weight, length, or head circumference at birth between the 2 groups at inclusion (Table 2). Furthermore, there was no significant difference in baseline characteristics between completing infants (n = 129) and those who dropped out (n = 12; data not shown).
There was no significant difference in registered adverse events between the 2 groups. There was a trend toward longer duration of nasal continuous positive airway pressure treatment (28 vs 13 days) and oxygen requirement (13 vs 8 days) in the intervention group, compared with the control group, but the differences were not statistically significant (Table 3). Two infants (birth weight: 705 and 830 g) in the control group died during the study, which was not related to feeding protocols. One infant had major congenital malformations (not realized at the time of inclusion), and the other experienced major respiratory failure.
Diet and Growth
Feeding data existed for 127 of the 129 infants for the whole hospital stay (Table 4). There were no significant differences in energy and nutrient intakes between the 2 groups (data not shown), apart from DHA and AA. The intakes of DHA and AA increased in the intervention group after 1 week of age. The nutrient intakes were similar in the 2 groups, apart from DHA (0.86% vs 0.35%) and AA (0.91% vs 0.32%). The mean daily intakes of DHA were 59 mg/kg per day in the intervention group and 32 mg/kg per day in the control group (P < .001), and the intakes of AA were 47 mg/kg per day and 22 mg/kg per day, respectively. There was no significant difference in growth between the 2 groups (Fig 2). Weight gain (mean ± SD) was 23.3 ± 5.2 g/day in the intervention group and 22.8 ± 4.9 g/day in the control group. The mean daily length gain was 1.2 ± 0.5 mm in the intervention group and 1.3 ± 0.7 mm in the control group. The mean gain in head circumference was 1.2 ± 0.7 mm/day in the intervention group and 1.0 ± 0.4 mm/day in the control group.
Plasma Fatty Acid Patterns
At the time of inclusion, there were no significant differences in plasma fatty acid patterns between the intervention group and the control group. During the supplementation, plasma DHA concentrations increased by 12% in the intervention group and decreased by 9% in the control group. There was a significant effect of fatty acid supplementation versus time course (P = .045, analysis of variance). Plasma AA concentrations decreased by 6% in the intervention group and 24% in the control group, and the effect of fatty acid supplementation versus time course was significant (P = .015, analysis of variance) (Table 5).
Mental and motor development was measured for 105 infants with the Ages and Stages Questionnaire at the 6-month follow-up evaluation. The intervention group scored higher on the problem-solving subtest than did the control group (53.4 vs 49.5 points; P = .02) (Table 6). There was also a nonsignificantly higher total score (221 vs 215 points).
ERPs were measured for 98 infants at 6 months of age. Two infants had impaired vision and could not participate in this test. Fifteen recordings had to be discarded because the infant was crying (n = 7) or because of technical failure of the recording equipment (n = 8); therefore, data are presented for 81 infants. Calculations were performed by using the mean amplitude in the interval of 400 to 650 milliseconds after presentations of visual stimuli as standard or novel images (Table 7). Infants in the intervention group had significantly lower (more-negative) amplitudes, compared with the control group, after the standard image (P = .01) (Fig 3A, C, E, and G and Table 7). Presentations of novel images (deviants) did not induce any difference between the 2 groups (P = .14) (Fig 3B, D, F, and H and Table 7). Similar patterns were seen for frontal and central electrodes, as well as left and right hemispherical leads.
Our present study is the first to show a beneficial effect on cognitive function of DHA and AA supplementation (48 mg/kg per day of each) for VLBW infants fed human milk. With a parental questionnaire based on the traditional test paradigm (Ages and Stages Questionnaire), the intervention group obtained a significantly higher problem-solving score than did the control group. There was also an indication toward a higher total score in the intervention group, without reaching statistical significance. We found a clear beneficial effect on recognition memory in the intervention group, as indicated by ERP measurements after visual stimuli.
Earlier studies of DHA and AA supplementation for VLBW infants were conducted with formula-fed or partially formula-fed infants, giving doses of important fatty acids equivalent to the content in human milk. Most studies showed little or only modest effect of DHA supplements on cognitive function,23 which may be explained in part by low dosages and methodologic problems in the earliest studies, such as low statistical power and lack of reliable end points. Another possibility is that the previously observed beneficial effect on IQ was attributable to other differences between human milk and formula (eg, energy, protein, or antioxidants). However, our present study supports the hypothesis that DHA and/or AA are major beneficial components, because these fatty acids represent the main difference between the groups.
Previous studies tested formulas without DHA/AA, compared with different concentrations of these fatty acids.23 In our present study, both groups received human milk containing some DHA and AA, and the intervention group received additional DHA/AA supplementation. Our design is different from that of previous studies, which makes direct comparisons difficult. We found a small partial difference between groups by using a traditional assessment of mental and motor development (Ages and Stages Questionnaire). The Ages and Stages Questionnaire consists of 5 subparts, measuring a wide range of infant development (ie, communication, motor development, problem-solving, and social development), and we did not expect that DHA and AA would have an effect on all of these parameters. There is growing evidence that DHA and AA have specific functions related to memory and problem-solving.24 The result of our present study might be interpreted to support this hypothesis, because we observed a highly significant difference between groups by using an ERP index of recognition memory. Our results add to the evidence that choice of end point measurement is crucial in these types of intervention studies.15
The strength of our present study is the multidisciplinary approach combining aspects of neonatology, nutrition, biochemistry, and neuropsychology. In addition to tests of mental and motor development, we used ERPs to examine processes related to recognition memory. The method is also suitable for young infants, because no complex linguistic or behavioral responses are required. Another advantage is that recognition memory may predict later IQ better than traditional tests.25 The ERP method also makes it possible to obtain specific information about timing and patterns of brain activities. Atypical ERP responses after auditory stimuli were reported previously for preterm infants, and these have been interpreted as signs of cognitive dysfunction.22,26 Atypical ERP responses after visual stimuli also were reported for infants at high risk for neurodevelopmental impairments.27 The present study is the first to compare ERP patterns in infants receiving 2 different nutritional regimens in the neonatal period.
The NC component was shown previously to be larger with novel stimuli, compared with familiar or repeated stimuli.28 The NC component is likely to be affected by an interaction of arousal, attention, and memory mechanisms.28 Ideally, these contributions should be decomposed, which would have called for a much larger number of channels, as well as a more-complex paradigm. This was not feasible. For our purposes, it was not critical to separate components of the modulation of NC amplitude. Instead, we used recognition in this context as a term that encompassed pure memory and attention components. This renders the response to the repeated standards the measure of central interest in our study. Attention and recognition are both essential for learning and information processing. The stability of the present findings needs to be demonstrated, and we intend to monitor the participants later in childhood.
One limitation of our present study was low statistical power for detecting differences by using the Ages and Stages Questionnaire, increasing the probability of type 2 errors. There may be a real difference between groups that we were unable to detect. We estimated that we would need at least 126 participants (63 subjects in each group), but only 105 completed the 6-month questionnaire. Another limitation is the subject's attention with the ERP method. Although the method is noninvasive, the infants need to be relatively calm and awake. A substantial number of participants needed to be excluded because the infants were crying and/or refusing to wear the electrodes. This problem is shared by many tools for measuring cognitive development in infants. Crucially, in the present study, the numbers of records that needed to be rejected were similar in the 2 groups.
The study oil was well tolerated and absorbed. Although we used higher doses of DHA and AA, compared with other studies, we noticed a decline in plasma concentrations of AA in both groups, which suggests that the dose of AA still may be too small. An alternative explanation is that a decline in AA concentrations is normal. We did not notice any negative effect in clinical events during the supplementation. There was a nonsignificant trend toward a longer duration of assisted ventilation and greater oxygen requirements in the intervention group, compared with the control group. This was also noticed in 1 other study,12 but not in others.10,11,13,29,30
The effect of DHA supplementation on infant growth has been controversial. Some early trials among preterm infants reported that supplement-treated infants had lower weight gain and were shorter than non–supplement-treated infants.31,32 This difference was thought to be a consequence of altered proportions between eicosapentaenoic acid and AA but, when both DHA and AA were added to formulas, impaired growth was no longer noted.23 In our study, we used a 50:50 mixture of DHA and AA, and we confirmed earlier studies by not detecting any negative effect of supplementation on weight gain or growth.
Supplementing human milk with DHA and AA led to increased plasma concentrations of DHA and AA in the early neonatal period. At the 6-month follow-up evaluation, infants who received DHA and AA had better problem-solving skills and discriminated better between familiar and unfamiliar objects, compared with the control group. This function is essential for focusing attention, learning, and information processing. It remains to be seen whether this type of intervention may have long-term effects on cognitive function, school performance, and rates of attention-deficit/hyperactivity disorder in later childhood.
Financial support was provided by the Norwegian Foundation for Health and Rehabilitation, the Johan Throne Holst Foundation for Nutrition Research, the Freia Medical Research Foundation, the Research Council of Norway, and the Thematic Program on Perinatal Nutrition, Faculty of Medicine, University of Oslo. Martek Biosciences kindly provided the study oils.
We are grateful to the staff members at the 4 NICUs and to Ane Westerberg for assistance with analysis of the dietary records.
- Accepted September 28, 2007.
- Address correspondence to Christian André Drevon, MD, University of Oslo, Institute of Basic Medical Sciences, Department of Nutrition, PO Box 1046 Blindern, 0316 Oslo, Norway. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
DHA and AA are important for growth and neurodevelopment of the fetus and preterm infants. The average supply of DHA from human milk or formula is <50% of the estimated uterine accretion rate.
What This Study Adds
Preterm infants may benefit from supplementation of human milk with DHA and AA in the first month of life.
- ↵Dewey K, Lutter C. Guiding Principles for Complementary Feeding of the Breastfed Child. Washington, DC: World Health Organization, Division of Health Promotion and Protection, Food and Nutrition Program; 2006
- ↵Mikkola K, Ritari N, Tommiska V, et al. Neurodevelopmental outcome at 5 years of age of a national cohort of extremely low birth weight infants who were born in 1996–1997. Pediatrics.2005;116 (6):1391– 1400
- ↵Indredavik MS, Vik T, Heyerdahl S, Kulseng S, Fayers P, Brubakk AM. Psychiatric symptoms and disorders in adolescents with low birth weight. Arch Dis Child Fetal Neonatal Ed.2004;89 (5):F445– F450
- ↵Harris WS, Connor WE, Lindsey S. Will dietary omega-3 fatty acids change the composition of human milk? Am J Clin Nutr.1984;40 (4):780– 785
- ↵O'Connor DL, Hall R, Adamkin D, et al. Growth and development in preterm infants fed long-chain polyunsaturated fatty acids: a prospective, randomized controlled trial. Pediatrics.2001;108 (2):359– 371
- ↵Thomas K, Casey B. Methods for imaging the developing brain. In: de Haan M, Johnson MH, eds. The Cognitive Neuroscience of Development. Hove, England: Psychology Press; 2003:19– 42
- ↵Richards JE. The development of visual attention and the brain. In: de Haan M, Johnson MH, ed. The Cognitive Neuroscience of Development. Hove, England: Psychology Press; 2003:73– 98
- ↵Fewtrell MS, Morley R, Abbott RA, et al. Double-blind, randomized trial of long-chain polyunsaturated fatty acid supplementation in formula fed to preterm infants. Pediatrics.2002;110 (1):73– 82
- Copyright © 2008 by the American Academy of Pediatrics