The effect of long chain polyunsaturated fatty acid supplementation on intelligence in low birth weight infant during lactation: A meta-analysis

Background Low birth weight infant (LBWIs) are prone to mental and behavioural problems. As an important constituent of the brain and retina, long chain polyunsaturated fatty acids are essential for foetal infant mental and visual development. The effect of lactation supplemented with long chain polyunsaturated fatty acids (LCPUFA) on the improvement of intelligence in low birth weight children requires further validation. Methods In this study, a comprehensive search of multiple databases was performed to identify studies focused the association between intelligence and long chain polyunsaturated fatty acid supplementation in LBWIs. Studies that compared the Bayley Scales of Infant Development (BSID) or the Wechsler Abbreviated Scale of Intelligence for Children (WISC) scores between LBWIs who were supplemented and controls that were not supplemented with LCPUFA during lactation were selected for inclusion in the meta-analysis. Results The main outcome was the mean difference in the mental development index (MDI) and psychomotor development index (PDI) of the BSID and the full scale intelligence quotient (FSIQ), verbal intelligence quotient (VIQ) and performance intelligence quotient (PIQ) of the WISC between LBWIs and controls. Our findings indicated that the mean BSID or WISC scores in LBWIs did not differ between the supplemented groups and controls. Conclusion This meta-analysis does not reveal that LCPUFA supplementation has a significant impact on the level of intelligence in LBWIs.


Methods
In this study, a comprehensive search of multiple databases was performed to identify studies focused the association between intelligence and long chain polyunsaturated fatty acid supplementation in LBWIs. Studies that compared the Bayley Scales of Infant Development (BSID) or the Wechsler Abbreviated Scale of Intelligence for Children (WISC) scores between LBWIs who were supplemented and controls that were not supplemented with LCPUFA during lactation were selected for inclusion in the meta-analysis.

Results
The main outcome was the mean difference in the mental development index (MDI) and psychomotor development index (PDI) of the BSID and the full scale intelligence quotient (FSIQ), verbal intelligence quotient (VIQ) and performance intelligence quotient (PIQ) of the PLOS

Introduction
Low birth weight infants (LBWIs) are born with a birth weight of less than 2500 g, and those with a birth weight less than 1500 g are referred to as very low birth weight infants (VLBWIs) [1]. Worldwide, LBWIs accounted for 15% to 20% of all newborns in 2014; most LBWIs were born in low-or middle-income countries: 28% of LBWIs were born in South Asia, approximately 13% in South Africa, and 9% in Latin America [2]. The mortality rate of 1500-1999 g weights LBWIs is approximately 2.8 times that of 2000-2499 g LBWIs and approximately 8 times that of normal weight newborns (>2500 g) [3]. Low birth weight (LBW) in infants has become a public health problem worldwide. LBW is an important factor in the development of mental and psychomotor intelligence. Cognitive and physical development was reported to be negatively correlated with LBW [4], and LBW may result in cognitive dysfunction, mental retardation or cerebral palsy at early ages [5]. A previous meta-analysis showed that cognitive impairment associated with LBW will continue into adolescence and adulthood [6]. LBW and preterm birth often occur simultaneously, and because the most prolific period of foetal brain development is during the last three months of pregnancy [7], preterm birth can cause developmental immaturity of the neonatal nervous system.
The growth and development of LBWIs are closely related to their nutritional status. First, LBWIs require 110-150 calories daily, with additional milk supplementing that ingested from nursing in order to increase the carbohydrate intake [8]. Second, LBWIs require higher protein intake than normal newborns and need special formulas [9]. Since humans cannot synthesize n-3 and n-6 polyunsaturated fatty acids in vivo, they must get them from their diet, and LBWIs cannot effectively convert the precursor fatty acids, resulting in less capacity for fat storage [10]. LBWIs may lack long chain polyunsaturated fatty acids (LCPUFA) after birth, including docosahexaenoic acid (DHA) and arachidonic acid (AA). DHA and AA are essential for the development of the brain and central nervous system, and they quickly accumulate in the foetal anaphase and affect the development of the nervous system [11]. Previous reports studying whether supplementing breastfeeding with LCPUFA can improve LBWI intelligence were not conclusive. Some reports [12,13] showed that supplementation with DHA and AA can improve infant intelligence. However, other studies [14,15] showed that LBWIs supplemented with DHA and AA led to no significant improvement in neurodevelopment or in levels of intellectual, language, and motor development. The LCPUFA supplement dose, duration, ratio of different fatty acids, supplementation scheme and feeding patterns may impact the nervous system development and intelligence of LBWIs. Therefore, whether LCPUFA supplementation can improve neurodevelopment and intelligence, also, the duration and appropriate dose of LCPUFA supplementation require further investigation.
In this study, we conducted a meta-analysis to explore whether long chain polyunsaturated fatty acid supplementation can improve the intellectual level of LBWIs and to identify the most effective intervention duration.

Sources and methods of data retrieval
We performed a comprehensive literature search that included studies until August 2017; the electronic databases searched included PubMed, Medline, Web of Science, and Google Scholar.
The searches were conducted to identify all published studies that reported data on the mean differences and standard deviations of the mental development index (MDI) and psychomotor development index (PDI) of the Bayley Scales of Infant Development (BSID) or the full scale intelligence quotient (FSIQ) of the Wechsler Abbreviated Scale of Intelligence for Children (WISC) between LBWIs supplemented with DHA and AA and those who were not. The following terms were used for the literature search: very low birth weight infants, low weight birth infants, premature infants, preterm infants, neonatal prematurity, long chain polyunsaturated fatty acids, fish oil, n-3 fatty acid, n-6 fatty acid, eicosapentaenoic acid, docosahexaenoic acid, arachidonic acid, and randomized controlled trial. The term 'OR' was used as the set operator to combine different sets of results. The MDI and PDI of the BSID and FSIQ, verbal intelligence quotient (VIQ) and performance intelligence quotient (PIQ) of the WISC were determined and used in a meta-analysis to understand how intelligence levels differ between LBWIs supplemented with LCPUFA those who were not. The test age of the subjects and the duration of supplementation, as well as other confounding factors, were also considered.

Inclusion criteria
The articles that were included in this meta-analysis matched the following 7 criteria: (1) all study subjects were LBWIs with live birth weights of less than 2500 g; (2) the MDI and PDI of the BSID and the FSIQ, VIQ and PIQ of the WISC were presented as the mean ± standard deviation (SD); (3) LBWIs were randomly assigned to an intervention group that received long chain polyunsaturated fatty acid supplementation directly and a control group that did not; and (4) LBWIs were born without neonatal infection, disease that could have a significant effect on intelligence, or the intervention of LCPUFA could affect the treatment of any preexisting diseases or conditions. (5) We made no restrictions on the feeding mode or supplied dose. (6) Guardians of subjects voluntarily signed the informed consent. (7) We excluded studies that did not provide initial data, studies in animals, in vitro studies, reviews and conference papers. Three investigators independently reviewed and extracted all the potentially eligible studies and discussed the inconsistencies until a consensus was reached (Fig 1). Additionally, the Jadad score was used to assess the quality of studies included in this meta-analysis [16].

Data abstraction
We reviewed all the relevant studies and extracted the following data: (1) lead author, publication year, and sample size; (2) baseline data of the intervention group and the LBWI control group (sample size, sex, weight and gestational age); (3) feeding methods; and (4) the duration of the intervention, the intervention dose and the intelligence test ages (Tables 1 and 2).

Statistical analysis
All statistical analyses were conducted using the statistical software Stata (version 12.0, Stata-Corp LLC, College Station, TX, USA). The mean difference and standard deviation of the intelligence scores of the LBWIs and control group were used for the meta-analysis. We combined the weighted mean difference (WMD) for studies that reported mean and standard  I: low birth weight with long chain polyunsaturated fatty acids supplementation. C: low birth weight without long chain polyunsaturated fatty acids supplementation.
Mo: month. Ã long chain polyunsaturated fatty acids from fish/fungal oil.
# long chain polyunsaturated fatty acids from egg-derived triglyceride/fish oil.  deviation values for the intelligence score of the LBWI and control groups. An inverse variance weighted random-effects model or fixed-effects model was used to determine the WMD and 95% confidence intervals (CIs) and to measure the different intelligence scores of the LBWI and control groups, and the results were used to evaluate the differences in intelligence levels of the LBWIs and controls. In two of the included studies, two treatment groups were examined: one study investigated supplementation with two different doses of long chain polyunsaturated fatty acids, and the other investigated two different sources of long chain polyunsaturated fatty acids [17,18]. To avoid double counting, the controls in each of these 2 studies were split approximately evenly into 2 control groups with the means and standard deviations left unchanged before the results were included in the meta-analysis [19,20].
We used Cochran's Q statistic and the I 2 statistic to assess the statistical heterogeneity of the meta-analysis [21]. If the data were homogeneous (P >0.05), a fixed-effects model meta- analysis was performed; if the data were heterogeneous (P 0.05), a random-effects model meta-analysis was performed. Heterogeneity was considered significant at P <0.05 in the Q test, and the I 2 value was used to evaluate the degree of heterogeneity. We defined low, medium and high heterogeneity at I 2 values of 25%, 50%, and 75%, respectively [22]. Subgroup analyses were performed for the intelligence test age and the duration of supplementation. We used a funnel plot to detect publication bias concerning this meta-analysis, with the symmetry of the funnel plot used to determine whether publication bias occurred. Furthermore, a formal statistical assessment of the funnel plot asymmetry was performed using Egger's regression asymmetry test [23].

Bayley Scales of Infant Development
MDI. For studies that used the BSID to test infants' intelligence before 24 months of age, the results show that the long chain polyunsaturated fatty acids supplemented group did not have significantly higher MDI scores than the group without supplementation (SMD = 0.07, 95% CI = −0.05, 0.19, I 2 = 23.8%, P = 0.222; Fig 2). The study heterogeneity value was I 2 = 23.8%, P = 0.203, showing that the heterogeneity of the studies was small; publication bias results showed that there was no evidence of publication bias (Egger's test: coefficient = 0.036, P = 0.968). The effect of long chain polyunsaturated fatty acid supplementation on intelligence development We conducted subgroup analyses on the intelligence testing age, supplementation duration, mean birth weight, and gestational age. For the intelligence testing age, we divided the data into two subgroups of 0 mo-12 mo old [14,25] and 13 mo-24 mo old [13,14,17,18,[24][25][26]. Because Wezel-Meijler's study tested infants' intelligence at 3 months, 6 months, 12 months and 24 months, the findings were divided into a 0 mo-12 mo old group and a 13 mo-24 mo old group. Subgroup analysis showed that the LCPUFA supplementation groups did not have significantly higher MDI scores than the control groups (Fig 3).
Next, we carried out a subgroup analysis according to the supplementary duration. The results shown are for supplementary duration less than 3 months [13,18,26], 4 months to 6 months [14,25], or more than 7 months [17,24], for this analysis, the MDI scores of the intervention groups were not higher than those of the control groups (Fig 4). Forest plot of the PDI scores by testing age for the long chain polyunsaturated fatty acid supplementation groups vs. the controls; standardized mean differences with the 95% confidence interval and weight percentage are shown. Subtotals are for the testing age less than 12 mo and 13 mo-24 mo. Fish/fungal oil: long chain polyunsaturated fatty acids from fish/fungal oil. Egg/fish oil: long chain polyunsaturated fatty acids from egg-derived triglyceride/fish oil. LCP1: supplements of evening primrose oil (18:3ω6 0.32 mol/100 mol) and a single dose of purified fish oil (LCPω-3 0.38 mol/100 mol). LCP2: supplements of evening primrose oil (18:3ω6 0.32 mol/100 mol) and a double dose of purified fish oil (LCPω-3 0.8 mol/100 mol). https://doi.org/10.1371/journal.pone.0195662.g008 The effect of long chain polyunsaturated fatty acid supplementation on intelligence development We divided the 7 studies into two subgroups of VLBWI [13,14,17,24,[26][27][28][29] and LBWI [18,25] according to the birth weight. The studies in the VLBWI and LBWI subgroups did not show differences in MDI scores between the LCPUFA-supplemented group and the controls (Fig 5).
We carried out further subgroup analysis according to the components of fatty acids, DHA+AA [13,14,17,25], DHA+EPA+DPA [18] or DHA+AA+EPA [24,26]. statistically significant differences in MDI scores were not observed between the long chain polyunsaturated fatty acids supplied group and the controls (Fig 6). The effect of long chain polyunsaturated fatty acid supplementation on intelligence development PDI. The results of these studies indicate that the LCPUFA supplementation group did not have significantly higher PDI scores than the group without supplementation (SMD = −0.01, 95% CI = −0.23, 0.21, I 2 = 60.5%, P = 0.906; Fig 7). The studies' heterogeneity statistics were as follows: I 2 = 60.5%, P = 0.003, and from these data, we consider the studies to show significant heterogeneity; publication bias results showed that there was no evidence of publication bias (Egger's test: coefficient = 0.42, t = 0.35, P = 0.737).
We conducted subgroup analyses according to the intelligence testing age, supplementation duration, mean birth weight, and gestational age. According to the intelligence testing age, we divided into the subjects into two subgroups of 0 mo-12 mo old [14,25] and 13 mo-24 mo old [13,14,17,18,[24][25][26]. Because Wezel-Meijler's study tested infants' intelligence at 3 months, 6 months, 12 months and 24 months, the results were divided into a 0 mo-12 mo-old group and a 13 mo-24 mo-old group. Subgroup analysis showed that the LCPUFA supplementation groups did not have significantly higher PDI scores than the control groups (Fig 8).
Next, we carried out subgroup analyses according to supplementary duration. The results were for supplementary durations less than 3 months [13,18,26], 4 months to 6 months [14,25], or more than 7 months [17,24]; the PDI scores of the intervention groups were not higher than those of the control groups (Fig 9). The effect of long chain polyunsaturated fatty acid supplementation on intelligence development We divided the 7 studies into two subgroups of VLBWI [13,14,17,24,[26][27][28][29] and LBWI [18,25] according to the mean birth weight. The studies in subgroups of VLBWI and LBWI did not show differences in the PDI scores between the LCPUFA group and the controls (Fig 10).
We carried out further subgroup analysis according to the components of fatty acids, DHA+AA [13,17,25], DHA+EPA+DPA [18] or DHA+AA+EPA [24,26]. statistically significant differences in PDI scores were not observed between the LCPUFA supplied group and the controls (Fig 11).

Discussion
Currently, LCPUFA supplements such as DHA or AA are widely used in infant formula, and this is become one of the "selling points" for a lot of brands at least in China, despite there is no conclusive answer to whether these supplements are required or indeed beneficial to infants [30]. Infant formula in China is already a very complicated issue [31,32]; whether LCPUFA supplements had to be added in the infant formula further enhanced the complexity of this issue. Based on the results from this meta-analysis, LCPUFA supplements had no beneficial effect, at least on the intelligence development of low birth infants.
Although previous studies have shown that maternal intake of n-3 polyunsaturated fatty acids during pregnancy and lactation may promote the development of the infant's nervous system and intelligence [12,[33][34][35][36], this meta-analysis could not support this hypothesis. Our analysis showed that there was no statistically significant difference in intelligence scores of LBWIs between LCPUFA supplemented groups and control groups. Subgroup analysis of the duration of interventions or the combinations of LCPUFA (DHA+AA, DHA+EPA+DPA or DHA+AA+EPA) had no effect on the intellectual level of the intervention group compare to that of the control group either. The effect of long chain polyunsaturated fatty acid supplementation on intelligence development Among ten studies we included in this meta-analysis [13,14,17,18,[24][25][26][27][28][29], only 3 studies [18,24,25] fund LCPUFA supplement did help in mental development. And these three studies used PDI or MDI scores which were tested before 2 years of age, but developmental assessment of VLBWIs before 2 years of age might be not reliable for prediction of future mental development since prematurity might affect the test results [37]. In all the long-term follow up studies [27][28][29] that used the WISC scores which were tested at around 7-10 years of age, no beneficial effects of LCPUFA supplements in mental development were found. Taking together, we believe the hypothesis that LCPUFA supplement is required for LBWIs is highly debatable.
There are some limitations in this study. For example, among the studies that include different supplemental doses of LCPUFA, we could not calculate the relationship between the fatty acid supplement dose and infant intelligence. Therefore, future studies would be benefit from using different doses of DHA and AA or a different fatty acid ratio as a supplement for LBWIs, thereby exploring the appropriate LCPUFA supplemental doses and allocation ratios. The other issue we need to point out is that we excluded all the studies with LBWIs with neonatal infection or disease in this meta-analysis, as LBWIs are at high risk of pulmonary diseases [38] or other chronic disorders [39], whether LCPUFA supplements are indeed beneficial for these infants could not be answered by this analysis. Furthermore, there were some weaknesses The effect of long chain polyunsaturated fatty acid supplementation on intelligence development in clinical studies of LCPUFA supplements: relatively short duration of intervention, variance in populations and the limitations of testing methods. So the beneficial effects of LCPUFA which were shown in animal studies might not be revealed in clinical studies due to these shortages. Ultimately, a multi-centre, randomized quality-controlled set of experiments with a large sample size is still needed to prove that LCPUFA supplements improve LBWI intelligence and to provide more reliable evidence.

Conclusions
In summary, although long-chain polyunsaturated fatty acids were reported to be essential for foetal infant mental and visual development, the impact of DHA, AA, EPA or DPA supplementation on level of intelligence of LBWIs could not be proven. Therefore, whether longchain polyunsaturated fatty acids supplements are beneficial for LBWIs has not been shown conclusively.