Lactation and Neonatal Nutrition: Defining and Refining the Critical Questions

The financial impact of improving neonatal nutrition in at-risk populations is enormous. One major impact pertains to the consequences of preterm birth. It is estimated that 65,000 very low birth weight infants (birth weight <1,500 g) are born in the United States each year. If the average hospital cost for mother and infant is ~$140,000 (RW Hall, personal communication), the total annual cost in the US is more than $9 billion. If feeding human milk or its important components reduced the incidence of necrotizing enterocolitis (NEC) and other serious medical problems by even 15 %, savings of over $1 billion per year could be realized. A similar economic impact could be realized if the obesity epidemic were reduced. The economic burden of this epidemic is considered to have four elements, namely, increased costs of medical care, costs of lost productivity, transportation costs, and human capital accumulation [10] estimated together to cost at least $250 billion per year. If breastfeeding reduced the incidence of obesity in susceptible individuals by even 10 % then savings of up to $25 billion could be realized. A third area of high impact is that of malnutrition on a global scale. Malnutrition affects 900 million children and their mothers, with an inestimable economic impact. If the balance of feeding human milk and complementary foods were optimized to minimize growth stunting in the first two years in these individuals, the impact would be enormous [8].

Despite national and international initiatives on breastfeeding, we lack an in depth understanding of both the biological and psychosocial aspects of breastfeeding. Studies of the mechanisms by which milk components affect all aspects of infant development are in their infancy, but it is increasingly clear that at-risk infants have special nutritional needs and that, for such infants, the current uniform breastfeeding recommendations are sub-optimal. These deficits seriously impact the information available to health care providers and families alike. Such knowledge could be translated into science-based consumer information and professional “toolkits” to help more women successfully meet their own breastfeeding goals and achieve the best nutrition for their infants.

In this paper we outline the specific questions that arose at this wide-ranging conference as being important for research in the next decade. These questions focus on the composition of milk, the effects of milk components on neonatal development, the regulation of mammary development and milk secretion, and finally problems specific to at-risk populations.

Advantage should be taken of large epidemiological studies to bank samples of human milk at defined intervals postpartum. Analyzing cellular mRNA present in the cytoplasmic crescents of the milk fat globule provides a window into the transcriptome of the mammary alveolar cell in lactating women [18] that can be used to dissect the effects of maternal states such as obesity and stress on mammary cell function. For this reason particular importance should be given to proper sampling and storage of the lipid fraction of milk. It is also essential that the collection and storage of these milk samples be accompanied by systematically recorded detailed information regarding the health, nutrition, immune function and reproductive state of the woman providing the sample as well as the growth and development of her infant.

In general, the variations in milk composition as well as the mechanism of action of many bioactive components of human milk in the infant remain poorly characterized although these factors contribute significantly to infant development [12]. Where certain constituents are low or lacking in the milk of particular mothers, opportunities for supplementation may exist. For infants whose mothers are unable to breastfeed, a greater understanding of bioactive factors may help improve the development of human milk substitutes in relation to infant health. Thus a fuller appreciation of the range of function of the nutritive and non-nutritive components of human milk has significant potential to translate to improved public health.

Most studies have focused on comparing human milk and formula. However, by 6 months of age, most infants are receiving both maternal milk and infant formula. How this combined feeding strategy impacts resistance of the infant to infection and other aspects of development is virtually unknown. As studies are developed they should be designed to include interactions between components of human milk and complementary foods, including formula.

The gastrointestinal (GI) tract undergoes significant postnatal growth and functional adaptation in response to feeding (reviewed in [12, 19]). The transition from predominantly parenteral support via the umbilical cord to enteral support via human milk or formula presents a challenge to the researcher who must come to understand the interaction between GI mucosal surfaces and environmental stimuli such as dietary antigens, commensal and pathogenic microbiota and their products [20, 21]. In addition, hormones and growth factors constitute a class of bioactive proteins and peptides in milk [19], the best-studied of which include insulin-like growth factor-1 (IGF-1), epidermal growth factor (EGF), and transforming growth factor (TGF)-β. Our current knowledge of the mechanisms by which these and other bioactive components act in the neonatal intestinal has been limited by the complexity of human milk, an inadequte understanding of the normal progression of postnatal intestinal development and its underlying regulation, and the limited number of non-invasive methods available to assess the impact of milk components. Recently, Chapkin and colleagues used exfoliated epithelial cells to assess intestinal gene expression, documenting differences between cells from human milk- and formula-fed infants [22]. This non-invasive research model along with studies in neonatal pigs [23] and, potentially, monkeys [24] provides an important system to understand the effects of milk components on GI development.

Colostrum is secreted by lactating women in small quantities (~100 ml per day) for the first 2 days after birth, during which its composition is constantly changing [25]. In species that, unlike the human, do not transfer systemic immune components across the placenta, the IgG in colostrum provides systemic immunity to the offspring while their endogenous immune systems develop [26]. Human colostrum has very little IgG and does not directly provide systemic immunity, but does have high concentrations of protective components such as secretory immunoglobulin A (sIgA), lactoferrin and human milk oligosaccharides (HMO) [27], which provide protection to mucosal surfaces. Human colostrum also contains growth factors and cytokines such as transforming growth factor (TGF)-β, interleukin-(IL)-10, and erythropoietin that can suppress inflammatory responses in the immature neonatal intestine [12]. Detailed knowledge of the time course of secretion of these colostrum components through lactation as well as the mechanisms by which they act to promote intestinal development is likely to lead to optimized protocols for feeding human milk and/or its components to at-risk infants (See Topic IV).

Because conferring resistance to infection is a major function of human milk and is critically important in developing countries, it is imperative to understand which components of milk are most important for this effect and their mechanisms of action. Many milk proteins such as lactoferrin and lysozyme have direct anti-microbial effects [28‐30]. Others such as oligosaccharides may act as decoy receptors to block adhesion of pathogens to epithelial cell surfaces [31]. Human milk nucleotides have been shown to stimulate an increased host immune response [32, 33].

An improved understanding of antimicrobial components in human milk may foster the development of resources to improve the antimicrobial quality of animal milks. For example, elevated lysozyme concentrations in the milk of transgenic goats improved duodenal development and reduced the levels of inflammation and E.coli in the ileum of neonatal pigs [34]. This experiment illustrates both the potential of transgene-expressed protective substances in animal milks and an elegant use of the piglet to study the function of milk components.

Systemic and mucosal immunity begin to develop in utero, and continue to develop from the postnatal period through early childhood. The neonatal mucosal immune system is relatively devoid of IgA producing B cells. Secretory IgA available through human milk may serve a compensatory role as the infant’s IgA producing B cells develop and differentiate following exposure to antigens ex-utero. Cues from the microenvironment including the gut microbiome [35] and nutritional components regulate the development of the immune response both locally in the intestine and systemically. An important example is the observed role of vitamin A metabolites acting via the retinoic acid receptor to suppress development of Th17 cells [36]. Other molecules in milk that have receptors expressed by cells of the immune system include lactoferrin (lactoferrin receptor) [37] and acidic sialic acid-containing carbohydrates that may potentially bind to sialic acid binding receptors, also known as SIGLECs [38]. Data from animal models suggest that allergen exposure via human milk mediates development of regulatory T cells in neonates that, in turn, protect the animal from developing asthma when later challenged with the allergen [39]. Whether these findings translate to humans remains an important question given that the incidence of childhood allergy continues to increase [40].

Studies of the immunological properties of human milk and its effects upon the infant should provide important insights into several issues:Understanding of these issues would afford insight into novel means to impact human health and development.

The infant GI tract is immunologically and microbiologically naive at birth. The early days and weeks following birth mark a period of profound change in the neonatal GI tract as the number of bacteria increase from near-sterility to levels found in adults [41, 42] (Fig. 2). Over the ensuing 12–24 months, the GI microbiota transitions to domination by obligate anaerobes of the phyla Bacteroidetes and Firmicutes. In parallel with this microbial succession, the gut-associated lymphoid tissue (GALT) and other components of the innate and adaptive mucosal immune system reach maturity [13]. In mice the presence of microbial communities in the gut is an essential driver of GALT development [43, 44]. Analogously, the human GI microbiome likely is a critical determinant of infant immune function.

Both diet (breastfeeding vs. formula feeding) and mode of delivery of the infant (vaginal vs. Caesarean) are significant early determinants of the kinds and quantities of microbes that colonize the neonatal intestines [45]. For example, Penders et. al. demonstrated that at 1 month of age, the potentially “beneficial” Bifidobacteria strain was more prevalent and present in higher numbers among 700 exclusively breastfed Dutch infants than in 232 formula-fed infants and 98 infants who received mixed feeding [46]. In contrast, microorganisms such as Clostridium difficile, Bacteroides fragilis and E. coli that are believed to have a higher inflammatory potential were present in higher numbers in the latter group, as were Lactobacilli. Human milk oligosaccharides are fermentable and selectively nourish the growth of specific strains of Bifidobacteria found among the microflora of the breastfed infant [14, 47]. Data from in vitro studies suggest that proteolytic fragments of α-lactalbumin, lactoferrin, and the secretory component of the polyimmunoglobulin receptor (35) can selectively stimulate the growth of Bifidobacteria [48]. An important unknown is the nature of other digestion products of milk proteins that may play a role in specifying the infant microbiome.

Despite the great progress made recently in linking the GI microbiome with nutrition, feeding practices and immunity, many important questions remain unanswered:

Recently it has become clear that a complex microbiome exists in human milk [49] although whether these microflora contribute to the infant microbiome is not clear. Further, the source of these microorganisms and the factors that influence their presence are completely unknown.

In 2006 more than 1 million mothers in the US cited the need to take a wide variety of medications as the reason they elected not to breastfeed [50]. In addition mothers are unfortunately not immune to the effects of recreational drugs and cigarette smoking. Recent articles indicate that codeine and the nicotine derived from cigarette smoking [51] are secreted into the milk with significant effects on the infant. These effects, which can result in infant death [52], emphasize the potential dangers of maternal drug use for the infant. Likewise, beyond the established pronounced effects of alcohol consumption on the unborn fetus, alcohol intake by lactating mothers alters the sensory properties of human milk and infant nursing behavior [53].

Many mothers have conditions that legitimately require consistent drug therapy including psychiatric problems, heart disease, hypertension and rheumatic illnesses. For rheumatic diseases good recommendations about drugs that can and cannot be used during pregnancy and lactation are available [54]. There is considerable information about levels of antidepressants in maternal plasma and human milk, but very little information exists about their effects on infant behavioral outcomes [55]. Caregivers and potential breastfeeding mothers must have appropriate information regarding the effects of specific drugs on breast milk and on infant health in order to make appropriate decisions about breastfeeding.

The dairy industry currently embraces an emerging potential for personalized nutrition and designer milks. These concepts are intertwined. In the context of infant formula, they translate to a design that is (1) humanized, (2) specialized for clinical use, (3) supplemented for optimized neonatal health.

Important questions for future consideration are:

A good deal of current attention has focused on oligosaccharides as likely-important substrates for gut biota [14]. These molecules have been associated with resistance to allergies and infections as they are thought to foster a robust immune system. An important question for future research is to what extent bovine milk oligosaccharides can support “healthy” infant-type growth of Bifidobacteria. The need for further research into the detailed composition of cow and goat milks is important given that its synthesis and post-harvest processing can provide for new, possibly designer molecules that are able to enhance human health.

Specific testing of brain function, especially within the first year of life, has provided mixed, albeit variable, results with respect to a comparison between breastfeeding and formula. An elegant technique known as event-related potentials (ERP) can be used to measure infant responses to stimuli that are potential important in language acquisition. However, experiments using this technique have yielded mixed results with respect to differences between formula- and breastfed infants [75‐77]. The caveat is that all of these ERP studies were conducted at a single age; infants develop rapidly in the first few months of life and acquire processing abilities at slightly different ages, so these differences could reflect the individual developmental trajectories. None-the-less these studies indicate that sophisticated cognitive tests are available to test hypotheses about behavior-modifying effects of specific milk components. It is important to take into account factors such as parent education, IQ, socio-economic status, maternal-infant interactions, and home environment if these studies are to provide reliable information about the effects of milk components.

Compared to the altricial young of rodents, primate infants begin to engage behaviorally with a wide social network including parents, siblings, friends, and other conspecifics soon after birth. Infants depend on mother’s own milk, or an alternative, not only for nutrition but to sustain these social interactions during the post-natal period. Careful studies in primate populations relating variations in milk composition and volume among individual mothers to their nutrition, health, body mass, social condition may identify lasting consequences for neurodevelopment and behavior. Importantly, among primates the early expansion of an primate infant’s behavioral repertoire is co-incident with critical windows of neurodevelopment [78].

A better understanding of the neurobiological underpinnings of behavioral outcomes is needed to extend these findings for translation to human health. Although non-human primates are likely the most similar to humans for investigating how milk influences neuro-psycho-biological development and social behavior, additional animal models particularly pigs and rodents will provide crucial, complementary information. The dairy science literature may also be a source of important information given that milk production is closely linked to the behavior of the cow during milking [79].

Insulin has long been used in culture models of mammary differentiation [82], although the high concentrations likely acted through the IGF1 receptor. Strong evidence now emphasizes that insulin signals through the insulin receptor during lactation [81, 89]. If this is also true of the mammary gland during pregnancy, then the effects of insulin upon epithelial differentiation could substantially influence breastfeeding success. Hypoinsulinemia, as can occur temporarily in Type I diabetes, might interfere with differentiation. The hyperinsulinemia that occurs during obesity and gestational diabetes could hyperstimulate insulin signaling perhaps promoting premature initiation of lactation with subsequent premature involution, since secreted milk could not be removed from the gland. Overexpression of activated Akt, a downstream mediator of insulin signaling had exactly this effect in the mouse mammary gland [90]. Animal studies, possibly in genetically altered mice or larger species more similar to humans such as pigs, along with human studies correlating insulin levels in pregnancy with lactation performance will be necessary to test this hypothesis.

Although glucocorticoids have long been considered important for milk synthesis [82], they are also associated with the suppression of milk synthesis and milk letdown under stressful situations [91]. The complex effects of glucocorticoid action on lactation are illustrated by the observation that betamethasone, a glucorticoid, facilitates milk production by mothers of preterm infants if given antenatally within 2 days of birth, but inhibits milk production when given 3 to 9 days prepartum [92]. In a mouse model of deficient glucocorticoid secretion, both pup growth and the expression of milk proteins were compromised [93] highlighting a clear requirement for corticoids for proper lactation.

An important question is the role of acute or chronic stress on glucocorticoid signaling. Non-human primate models will be valuable for future research, given their translational potential for humans. Rhesus macaques produce singletons, like humans, and show substantial individual variation in milk cortisol concentrations [94]. Marmosets (Callithrix jacchus), a smaller neotropical monkey more distantly related to humans, are emerging as a valuable model of obesity and reproduction [95] and could be particularly useful for understanding metabolic disease and lactational performance.

While the proliferative effects of progesterone on alveolar development in early pregnancy have received considerable attention [96] the mechanism by which progesterone represses secretory activation in late pregnancy has been difficult to study because of the paucity of in vitro model systems for mammary differentiation and the complexity of progesterone and prolactin receptor interactions [97]. New in vitro model systems in which mammary differentiation can be studied [96, 98], coupled with studies of animals with mutated receptors or downstream effectors are an important next step in elucidating this aspect of progesterone action.

Two modes of inherited traits are now recognized: (1) Traits based on DNA sequence are passed genetically to the offspring. (2) Epigenetic alterations in DNA including DNA methylation or other mechanisms can also be passed to the offspring [7]. The relevance of each for human lactation is discussed below.

Mammary development and lactation reflect a collection of quantitative traits that have a continuously distributed variation. It has been clear for nearly a century that variation in these traits cannot be fully understood without studying both genetic and environmental factors [99]. Studies in dairy animals show that genetics can account for as much as 40 % of the variation in lactation-related traits [100]. Litter weight gain in mice, which is an indicator of lactation success, is also heritable with genetics accounting for as much as 33 % of the observed variation [101]. In addition, milk production by dairy animals and rodents can be increased through genetic selection [102, 103]. In parallel, natural genetic variants in humans that alter mammary gland development or produce inadequate lactation are known. For example, isolated prolactin deficiency in women is an autosomal recessive trait known to produce a lactation defect [104]. Finlay-Marks syndrome, thought to involve mutations in the Lef1 gene, is an inherited condition that results in the absence of pregnancy-dependent breast development [105]. Mutations in the gene SLC30A2 that specifies a zinc transporter in the mammary gland have been documented to result in milk with low zinc concentrations [106, 107]. A genetic variation in FADS2 leading to altered fatty acid profiles in human milk was discussed in Topic II [66]. Despite these examples, our understanding of the role of natural genetic variation in determining mammary gland development and lactation in humans and species other than dairy animals is very limited. Work in mice has led to the identification of lactation quantitative trait loci [108] although the specific genomic loci have yet to be established. Lactation success in humans varies by race or ethnicity [109] although again the specific genetic differences are not known.

In the 11 years since the publication of the human genome enormous advances have been made in cataloging the variation that exists both in humans [110] and in animal models such as the mouse [111, 112]. These resources have already been applied to other aspects of human health and disease. A concerted effort is needed to employ these genomic resources toward understanding mammary gland development and lactation in humans and animal models.

An interesting question is whether programming of mammary cells occurs in utero or during lactation in response to maternal nutritional status or other maternal factors. This programming might potentially affect the development and/or lactation performance of the offspring resulting in epigenetic inheritance of lactation traits. Maternal nutrition during pregnancy and lactation can affect the susceptibility of female offspring to mammary tumorigenesis in animals [113, 114]. The nutritional status of gestating sheep produced heritable effects on lactation success in two generations of female offspring [115, 116]. Are such inherited epigenetic changes in mammary development a cause of lactation difficulties in obese women? The answer to this question is particularly important as we seek to reduce the consequences of the obesity epidemic.

Maternal obesity in humans increases the risk of unsuccessful initiation of breastfeeding and impaired ability to sustain lactation once initiated [117, 118]. A similar lactation impairment with obesity was observed in mice fed a “Western” diet, where obese dams produced lipid-poor milk that attenuated litter growth, compared to litters of lean dams on this same diet [9]. As an endocrine disorder obesity has the potential to impact breastfeeding and lactation at several levels including impaired hormonal regulation of mammary gland development [119], altered synthetic and secretory function [9, 120] and disrupted neonatal appetite control mechanisms [121], all impacting higher order integration of infant feeding needs with milk production and ejection [119]. We also need to understand whether the mammary adipose tissue produces locally effective concentrations of estrogen in obese women that could alter mammary gland development and lactation.

Animal models such as rodents (for gene targeting studies) and pigs (for nutrition and endocrine studies) as well as new quantitative approaches in humans that take the mother-child unit into account as an integrated physiological unit need to be developed. The opportunity also exists to incorporate findings from several decades of research on dairy animals into these investigations.

Inadequate lactation performance may reflect deficiencies in mammary gland development and initiation of lactation as well as ineffective milk removal from the breast by the infant. The mechanisms underlying delayed onset of lactation, lactation failure, and low milk production are poorly defined in part because such deficiencies are rarely investigated due to the ready availability of formula.

Mothers of preterm infants [92], obese women, and women with diabetes and other diseases find initiating and sustaining milk production to be difficult [117]. It is known that stress during labor delays the initiation of lactation [122] and epidural anesthesia during labor interferes with milk letdown [123]. As stated earlier, cortisol given during impending preterm labor affects milk production in variable ways depending on the time elapsed between injection and birth of the infant [92]. Do these observations reflect changes in the initiation of lactation or the ability to produce adequate milk having an appropriate composition? Few studies have been designed to differentiate between these two alternatives, which undoubtedly have different mechanistic bases.

Research on the role of maternal obesity and the effects of stress in lactation performance are priority areas. In both cases it is likely that an altered maternal endocrine environment and possibly proinflammatory cytokines are at least partially responsible for lactation impairment. In the case of obesity the roles for insulin and prolactin [124] require detailed investigation. A pumping regimen for breast milk expression [125] should be tested in prospective studies to determine whether a regimen like that used for mothers of preterm infants can improve milk production. Along with studies of such an intervention, ultrasound or MRI measurements of active breast tissue coupled with plasma levels of relevant hormones and proinflammatory cytokines would lay the groundwork for understanding the biological basis of these problems.

A question which often arises is “what is the effect of complementary foods on milk production and composition”. In a 1991 study [25] milk production fell as mothers decreased the number of daily feeds and milk volume was less than 100 ml per day when daily feeds were fewer than two per day. More importantly, at milk volumes below 200 ml per day, significant increases in the concentrations of protein, sodium and potassium and a decrease in lactose concentration were observed. An important question is whether the protein and carbohydrate composition in the presence of low milk volumes is altered in such a way that the concentration of protective substances like lactoferrin, lysozyme, IgA, and oligosaccharides is increased. If so, continuation of breastfeeding as long as possible after introduction of complementary foods may be important for populations in developing countries.

Milk removal has long been known to be a major factor in determining the amount of milk secreted by both humans and animals [80]. Infants with ineffective and/or inefficient sucking patterns do not provide the physical breast stimulation and milk removal needed for continued milk synthesis. Thus, ineffective suck can be a key element in lactation failure in both humans [126] and animals [127].

Infant suck is adversely affected by infant morbidity and prematurity [128]. Late preterm infants are at particularly high risk [129] since these infants are often cared for in the general maternity setting where breastfeeding is managed according to guidelines for healthy term infants. Maternal medication and drug use [130] and infant ankyloglossia [131] are also believed to contribute to ineffective infant suck. A systematic research approach is necessary to define elements of lactation failure related to these problems in the infant.

The developing mammary gland is affected by a variety of environmental factors including xenobiotics such as phytochemicals present in common fruits and vegetables [132, 133]. Semi-purified or processed whey proteins used in infant formulas can have similar effects [134, 135]. Although this research area is relatively new, it is now clear that such substances, whether in medications or in the diet, can act through signaling pathways that regulate mammary development. For example, feeding diets rich in soy or whey proteins can have effects similar to those induced by estrogen [136, 137]. Furthermore, there is widespread use of herbal agents, phytochemicals, off-label pharmaceuticals, and even alcohol by women who have a real or perceived difficulty to lactate, yet the full consequences of these agents are unknown.

Numerous environmental factors likely affect initiation of lactation or milk secretion; most significant among these are pharmaceutical agents taken by the mother. For example, in one study of women initiating breastfeeding the use of SSRI’s (selective serotonin reuptake inhibitors, psychotropic drugs) resulted in a substantial delay in the onset of lactation [138]. No mothers using an SSRI initiated lactation before 72 h postpartum. Drugs can also impair milk ejection, diminish milk volume, and alter milk composition. Animals can be very useful for studying the mechanisms of these effects; for example it has been found that dopaminergic agonists inhibit prolactin release in dairy cows [139] and beagles [140]. Although the effects of many of these compounds on breast cancer have received considerable research attention, studies of their effects on lactation are poorly represented in current research agendas.

Studies assessing the influence of breastfeeding on the risk of childhood overweight or obesity have demonstrated significant protection [150], but the effect of breastfeeding has been smaller and more variable when the outcome is assessed as a difference in mean BMI. These findings led some researchers to conclude that the effect of breastfeeding is small and influenced by bias and confounding. Importantly, in a recent study [151] school entry data on 14,412 children aged 4.5–7 years in southern Germany were broken down by BMI percentile. After adjusting for a large number of potential confounding variables, mean BMI was significantly reduced in children above the 90th BMI percentile who had been breastfed versus the formula-fed children in this BMI group. Interestingly, among children in the lower BMI percentiles (≥30 %), breastfeeding was associated with a low but significant shift towards a higher BMI. These results suggest that breastfeeding prevents overweight and obesity in children with a tendency to these outcomes, but does not affect underweight children in terms of weight reduction. In a very recent study of growth trajectories for offspring of non-diabetic pregnancies slower growth velocity was observed in those infants who were breastfed [152]. The effects of prior breastfeeding were particularly significant at ages 6 to 13 years. Importantly the population in this study was mostly middle class; extending this type of research to lower socioeconomic groups where obesity tends to be more prevalent is critical.

The potential of breastfeeding to reduce the risk for obesity following in utero exposure to overnutrition from a diabetic pregnancy or maternal obesity has great public health significance. Among Pima Indian youth exposed to maternal Type 2 Diabetes or gestational diabetes those who were breastfed for at least 2 months had a lower risk for developing diabetes compared to those who were formula- fed (30.1 vs. 43.6 % risk) [153]. In a recent comparison [154] of middle class children exposed to maternal diabetes in utero a significant decrease in childhood BMI and central adiposity was observed among children who were breastfed according current recommendations. A similar (25 %) reduction in childhood obesity risk was associated with exclusive breastfeeding for 6 months compared to exclusive formula feeding [155]. Children who received a mix of formula and human milk had results similar to the formula-fed group. Similar results were reported from 2 to 14 year old children in the National Longitudinal Survey of Youth [156]. All these results are consistent with the conclusion that children at risk for childhood obesity stemming from exposures in utero benefit significantly from breastfeeding.

The increasing number of observations that breastfeeding alters the growth trajectory of offspring born from both diabetic and non-diabetic pregnancies suggests that early infant diet has an important long-term effect [157]. Understanding the mechanism of these effects is important in determining what potential interventions might be effective. The complexity of the issue is revealed by both human and animal studies. For example, a quantitative relationship between consumption of milk from diabetic mothers and the risk of overweight at 1–5 years of age has been observed [157] suggesting an effect of diabetes on milk composition. When neonatal rats born to lean dams were cross-fostered to obese dams they developed an obese phenotype, including insulin resistance [158]. Maternal obesity in mice was shown to alter milk lipid content leading to decreased neonatal energy intake and decreased weight gain [9].

The question becomes What is the mechanism? Human milk contains more lactose, specific fats and cholesterol and relatively lower protein and mineral contents compared to milk from other animals. It is also possible that specific micronutrients or hydrolysates of macronutrients present in human but not cow’s milk stimulate GI hormones that act on the hypothalamus to regulate appetite. In support of this notion, babies who were given cow’s milk formula with added glutamate or hydrolyzed formula consumed fewer calories than those who drank regular cow’s-milk formula [159]. The influence of the neonatal microbiome on development of appetite regulation has been hinted at in rodent studies, but the mechanism is poorly understood. An answer to the question of whether satiety hormones such as leptin, adiponectin, and ghrelin or pro/anti-inflammation components of human milk have effects on infant adiposity, appetite, and energy expenditure could go a long way toward defining the early-life risk factors that subsequently lead to adiposity.

In summary, breastfeeding may be a powerful strategy to decrease the risk of obesity among offspring of mothers with pregnancy-associated diabetes or obesity during pregnancy. Additional research is needed to confirm this finding, to define compositional differences in human milk from healthy and diabetic mothers, to assess the influence of combination feeding with formula and to find the mechanism(s) involved. In addition, if breast-feeding has a significant effect on childhood obesity, principally for infants born to obese mothers, it will be important to find ways to increase the rate of breastfeeding in this population, particularly given significant evidence that obese women have problems initiating and maintaining lactation [117].

Growth stunting is a serious problem in developing countries where it is associated with poor cognitive and educational performance [57]. This condition, if not addressed prior to the third year of life has deleterious effects on height, strength and intellect in adult life, reducing what Vitora has called “human capital” [8]. The problem is that while breastfeeding is associated with improved survival in this population, it does not prevent stunting.

Progress has been made toward increasing the rates of exclusive breastfeeding for approximately the first 6 months of life in low resource settings and is predicted to have a major impact on infant survival (1). However, recent observations have highlighted faltering of very early postnatal linear growth occurring even during the period of exclusive breastfeeding [160]. Contrary to expectations, in most instances the rate of energy intake is not the growth-limiting factor. A possible causal effect on early growth faltering is micronutrient inadequacy in milk from undernourished mothers particularly on the heels of micronutrient deficiencies acquired by the fetus. Other possible explanations include deleterious bioactive compounds in human milk as a result of a pro-inflammatory environment due to maternal metabolic dysregulation. Additional or alternative factors in the intra-uterine or pre-conceptional environment may contribute to altered infant growth patterns [161]. However, all these possibilities are poorly defined.

After approximately the first 4 to 6 months of near exclusive breastfeeding the infant becomes dependent on other sources to meet its requirements for iron and zinc. These requirements are predictable due to depletion of infant iron stores by this time and the physiologic decline in zinc concentrations in human milk, respectively [162]. Broadly speaking, appropriate options are complementary foods, fortified foods, or supplements [163]. Meat provides both iron and zinc in a bioavailable form. Alternatively, fortified cereal can be an acceptable and efficacious vehicle, although the enteric microbiome and inflammatory responses appear to differ from those achieved on meat diets. In sum, research in this area has not yet found solutions to the problem of appropriate complementary foods in the undernourished population [164].

These observations illustrate the complexity underlying perturbed infant growth patterns in disadvantaged populations, with likely interplay among nutritional, immunostimulatory, and metabolic factors. The choice of interventions must take into account infant nutritional status as well as the nutritional and metabolic status of the mother. Better data that improve our ability to make rational nutritional choices for the undernourished infant will have a profound global impact on long term health.

As stated above, studies in at-risk populations are more likely to yield statistically meaningful results about the benefits of breastfeeding. However, these populations are often difficult to access in a consequential way, particularly for studies with a longitudinal component.

One particularly effective approach has been developed for studying young women in a community that as a whole is at high-risk for triple-negative breast cancer [165]. A longitudinal study of breast cancer initiation in this population was made possible by partnering with community advocates and navigators. Key components of the partnership included

Taken together, these components, as well as a willingness to partner and listen, are essential for collaborating to develop community-partnered studies.

Importantly, breastfeeding rates in this underserved population are low and, for those women who do choose to breastfeed, duration is short with many discontinuing within days after birth [166]. Factors that are hypothesized to influence breastfeeding incidence in this population include aggressive marketing of human milk substitutes, workplaces that discourage breastfeeding, social and personal networks and cultural norms, and individual beliefs about breastfeeding [166]. However, in some populations biological deficiencies in mammary development may underlie lack of successful breastfeeding. Effective studies of this problem will require a community partnership like that developed for the study of breast cancer.