This study showed a dose response relationship between early breast milk intake and later IQ and, uniquely, with whole brain volume at adolescence. In all subjects, but most clearly in boys, the effects of breast milk were seen more strongly on white than grey matter in the brain. These data support the hypothesis that one or more constituents of mothers’ breast milk promote brain development at a structural level.
Our observation that IQ itself was strongly related to white matter and hence TBV is consistent with the observed links between early breast milk feeding and both later IQ and later WMV. Relationships between white matter and measures of cognition have been reported previously in children (18
). These data support the hypothesis that the beneficial impact of breastfeeding on IQ may be mediated, at least partly, by promoting the development of white matter.
Generalisation of our findings to healthy full term infants is untested but may be supported by the observations that breast feeding in healthy infants is also related to higher subsequent IQ scores and that IQ is related to white matter in children and adults generally. In any case, preterm infants represent a sizeable proportion of the population (e.g. greater than the prevalence of Type I diabetes ), over-represented amongst those with neurodevelopmental problems, so favourable influences on long term brain development are important to identify even within this large subgroup.
While correlations were seen in the whole population, they were always more highly significant in males. Thus, IQ was most clearly related to white matter and hence TBV in males, and expressed breast milk was linked to IQ and to neural volumes only in males. These findings concur with a large body of animal evidence; Smart (19
) for example, examining the effect of early malnutrition on later learning and behavior in 165 animal studies, reported dominant effects in males. A growing literature in humans describes gender differences in brain development and structure (20
) and in relationships between brain structure and cognition (22
). We emphasise that males and females did not differ in mean IQ values but rather in the relationships between IQ and both diet and neural volumes. Percentage EBM accounted for around 25% of the variance in IQ in boys but only 5% in girls. It is interesting to note the recent report by Kesler and colleagues (24
) that measures of WMV differ between preterm males and females.
Correlations were often found for FSIQ, but more detailed analysis showed that VIQ, rather than PIQ, was related to breast feeding and neural volumes (except in one instance). There is growing evidence that early nutrition may have a selective, lasting effect on verbal subscales (25
); in later development, however, diet may selectively affect PIQ (26
). The implication is that the cognitive effects of nutritional intervention may depend on the timetable of brain development.
The mechanism for the observed effects is uncertain. In recent years, long-chain polyunsaturated fatty acids (LCPUFAs), notably docosahexaenoic acid (DHA), have been considered a key explanation for neurodevelopmental benefits of breast milk. A recent meta-analysis of outcome in preterms after early LCPUFA supplementation of infant formula (27
), however, found no conclusive evidence for long-term cognitive effects, suggesting that differences in early LCPUFA intake may not explain our observations. While LCPUFAs remain of interest, it should be remembered that human milk contains a wide variety of other factors including hormones and growth factors, some of which (e.g. thyroxin, nerve growth factor) could plausibly influence cerebral development, although this is not established. During the window of brain development we studied, glial cell production and myelination are key factors related to volume increase (28
). Our study was not designed to examine detailed neuronal development, but a plausible hypothesis for further testing, is that breast feeding promotes glial cell proliferation and, hence, myelination. Interestingly, a recent study showed that dietary manipulation in neonatal rats affected the expression of myelin basic protein more in males than females (29
), consistent with our findings of a greater effect of diet on white matter development in males.
Breast milk contains significant quantities of cholesterol while infant formulas contain little (30
); full term breast fed infants have higher total serum cholesterol at 6 months (31
). Saher and colleagues (32
) reported recently that cholesterol is an indispensable component of myelin membranes in mice and that cholesterol availability in oligodendrocytes is a rate-limiting factor in brain maturation. The massive increase in synapse number post-natally may require large amounts of cholesterol, obtained by neurons from astrocytes (33
). Serum cholesterol has been related to intellectual performance in adults (34
). We suggest, therefore, that dietary cholesterol intake could help explain an impact of breast milk on white matter development and cognition, through enhanced glial production/myelination.
We recognize limitations to the conclusions that can be drawn from this study since it did not have an experimental design, making it impossible to test for causation. We hope, however, to provide an impetus to further research and suggest to cognitive scientists that nutrition as an environmental factor may affect their outcomes. Testing for causation more formally in future intervention studies may be possible but, meanwhile, the strong observed dose-response relationships and the consistency of our findings with previous animal and human research makes causation plausible.
The study sample was restricted to preterm infants who had normal neurological status at 7 years. We are unable to say anything about the relationships we have observed here in those preterms who show evidence of injury or adverse outcome. We do not know whether this group would show the same effects, smaller or larger. It is possible that %EBM might have a greater influence in this vulnerable group but this remains to be determined.
There are also some more specific limitations. Der and colleagues (4
) have recently suggested that the relationship between breastfeeding and IQ is explained by the confounding influence of higher IQ in breast-feeding mothers. While all mothers in the present study had chosen to provide breast milk, we accept that residual confounding might remain, even within this subpopulation, if, for example, higher maternal IQ was associated with greater breast milk production. Lacking maternal IQ scores, we have used level of maternal education as a proxy measure and covaried this in all correlations. Although there was no significant relationship between %EBM and level of maternal education in the study population here, we accept that maternal IQ cannot be ruled out as a factor without directly measuring it. We have noted above that while the main outcome data, cognitive scores and neural volumes, were available for all subjects, data were missing for a control variable we used as a covariate, a weakness of the study.
During the age range studied here, grey matter and white matter develop along different trajectories (16
). Neural volumes were obtained at one time point only which is why we have covaried age at scan in an attempt to crudely control for these differences. Repeated scans would have allowed observations about the effects of breast milk on the course of development. The results show that the %EBM ingested as a child is related to the volume of white matter seen in the brain at the age range studied here but the absolute volumes of white (and grey) matter might differ at different ages without affecting these correlation coefficients.
In summary, the present evidence supports the hypothesis that the effects of constituents of mothers’ milk underlie the repeated observations that breastfeeding is associated with a higher cognitive level, at least in males. This has both scientific and clinical implications, pointing to nutritional practices that could optimise the attainment of cognitive potential. Our study also helps provide a framework for future mechanistic studies on the impact of early nutrition on brain development.