The microbial exposure of the infant from delivery onwards brings risk of early infections, but promotes development of the immune system
On leaving the protected intrauterine environment the newborn meets a world full of microbes, most of which are harmless and some are even protective, but many are potentially dangerous. It is vitally important that this early microbial exposure can be managed by the neonate. The high early-life mortality in poor socio-economic areas clearly illustrates this issue (Stoll, Reference Stoll, Remington, Klein, Wilson and Baker2006).
During passage through the birth canal the offspring becomes exposed to microbes of maternal origin. Being delivered next to the mother's anus the newborn is subsequently colonized by the mother's microbial flora. This microflora is the least threatening, since the mother provides defence against these microbes primarily via breast-feeding, but also to some extent via her transplacentally-transferred IgG antibodies. These antibodies provide tissue defence that is pro-inflammatory by activating the complement system and phagocytes (Hanson, Reference Hanson2004). This form of defence is costly, because the inflammation will consume energy and will cause the common symptoms of disease, such as pain, tiredness, fever and loss of appetite, via the increase in leptin caused by the cytokines IL-1β, TNFα and IL-6 produced by the leucocytes that are activated by exposure to infectious agents. Furthermore, this form of defence, although necessary and vital, is clearly undesirable in the young infant because it can interfere with growth and development. The link between frequent infections in early life and impaired growth is commonly observed in poor socio-economic regions. In contrast, the host defence provided via the milk protects without inducing inflammation (Hanson, Reference Hanson2004); it can even counteract inflammation. The various modes of function of human milk components in relation to host defence will be described in some detail, but first the immune system in early life will be discussed.
The immune system of the neonate has several characteristics. First, it is of limited size, but relatively complete in terms of the presence of various forms of cells (Lewis, Reference Lewis, Remington, Klein, Wilson and Baker2006). The lymphocytes are still of limited number and specificity. The neonate will only begin to enlarge its immune system when it is exposed to the numerous microbes present from delivery onwards. This enlargement process involves an increase in the lymphocytes with receptors that recognize the microbes that are in contact with the mucosal membranes; particularly those of the gastrointestinal tract where about 80% of the immune system is ultimately to be found. This very large number of lymphoid cells can be attributed to the very large surface area of the intestinal mucosa and the fact that it is more exposed to large numbers of microbes of a wider range of species than any other mucosal surface (Brandtzaeg et al. Reference Brandtzaeg, Carlsen, Farstad, Mestecky, Bienenstock, Lamm, Strober, McGhee and Mayer2005). The rate of development of the immune system in early life is comparable with that of the nervous system.
Second, the immune system must be programmed to respond only to threatening microbes and not to the body's own tissues, foods and other harmless material. On the other hand, it must develop tolerance to such factors. The process is not fully understood, but obviously exposure to a normal intestinal microbial flora has a central role in the normal development of oral tolerance and builds on the appearance of regulatory T lymphocytes. There are suggestions, for instance, that the ongoing increase in allergic diseases in recent decades may be linked to inefficient development of oral tolerance. This situation might be a result of changes in delivery strategies that prevent the normal exposure to the mother's microflora, including the greater frequency of Caesarean section deliveries and other changes in the environmental hygiene of the family, with subsequent failure to fully develop oral tolerance. This explanation has been termed the ‘microflora hypothesis’ (Noverr & Huffnagle, Reference Noverr and Huffnagle2005). A recent study in mice clearly demonstrates that exposure to Gram-negative bacteria present in the birth canal induces tolerance to such microbes by eliminating the Toll-like receptor 4 specific for Gram-negative bacteria on the intestinal epithelial cells, presumably to prevent inflammatory intestinal responses to the flora that subsequently colonizes the intestine (Lotz et al. Reference Lotz, Gutle, Walther, Menard, Bogdan and Hornef2006). However, the submucosal macrophages retain their Toll-like receptor 4 so that they can respond, activating the submucosal host defence on exposure to Gram-negative bacteria.
Early breast-feeding promotes the establishment of an optimal microbial gut flora in the infant from delivery onwards, counteracting potential pathogens (Adlerberth et al. Reference Adlerberth, Hanson, Wold, Sanderson and Walker1999). Certain harmless strains, such as Lactobacillus rhamnosus and other lactobacilli, are promoted by human milk (Ahrne et al. Reference Ahrne, Lonnermark, Wold, Aberg, Hesselmar, Saalman, Strannegard, Molin and Adlerberth2005). Some bifidobacteria also seem to be favoured by breast-feeding (Salminen et al. Reference Salminen, Gueimonde and Isolauri2005). The capacity of such micro-organisms to colonize the gut of the infant is promoted by breast-feeding, e.g. via the production of bacterial adhesins, which help the bacteria to attach to the mucosa and remain and multiply in the infant's gut. This process has been demonstrated in a recent study for certain non-pathogenic E. coli (Nowrouzian et al. Reference Nowrouzian, Monstein, Wold and Adlerberth2005).
Components in human milk supporting the infant's host defence
It is apparent, therefore, that early life is a dangerous period, during which the neonate is exposed to the microbial flora of the environment, which may include many potential pathogens. Although the immune system of the neonate is immature, it has a high capacity to adapt and respond to microbial exposure, a process that takes much of the first year of life. After birth and during the first few months of life breast-feeding can contribute to the defence of the infant. Some of the major contributory factors and their functions will now be described.
The major human antibody is secretory IgA (SIgA), which comprises approximately 80% of all human Ig and is present in all exocrine secretions. It prevents microbes from reaching mucosal membranes, so that they cannot cause infections. SIgA was first discovered in, and isolated from, human milk (Hanson, Reference Hanson1961; Hanson & Johansson, Reference Hanson and Johansson1962), and comprises 80–90% of the Ig in colostrum and milk. Its concentration is ⩽12 g/l in colostrum and approximately 0·5–1 g/l in mature milk (Hanson, Reference Hanson2004).
The specificity of SIgA antibodies in milk is based on the incorporation of the microbes into the numerous aggregates of lymphoid cells in the mother's gut (the Peyer's patches). After exposure to these microbes the lymphoid cells migrate from the Peyer's patches to various mucosal membranes and exocrine glands, where they settle and produce the SIgA antibodies (Fig. 1). These antibodies are primarily directed against the microflora in the mother's gut. Thus, the mother's milk will contain SIgA antibodies against the microflora in her gut, i.e. the microbes that normally colonize the infant from delivery onwards; the process is termed the entero-mammaric link. Thus, if there are potential pathogens in the mother's gut flora, protection is provided by the milk SIgA. This process is illustrated by an observation made by Mata et al. (Reference Mata, Urrutia, García and Béhar1969) in the village of Santa Maria Cauque' in Guatemala, where they were investigating the microbial flora of mothers and their infants. It was reported that the breast-fed infants of two mothers with diarrhoea as a result of Shigella did not show any symptoms of diarrhoea, even though the Shigella strain infecting the mothers was also found in their infants' stools. The presence of this homing mechanism has been established in man and in experimental animals (Goldblum et al. Reference Goldblum, Ahlstedt, Carlsson, Hanson, Jodal and Lidin-Janson1975; Roux et al. Reference Roux, McWilliams, Phillips-Quagliata, Weisz-Carrington and Lamm1977) by showing that specific SIgA antibodies appear in the milk soon after oral exposure of the lactating woman or experimental animal.
The major mode of action of SIgA antibodies, which are relatively resistant to enzymic degradation, is to bind micro-organisms, thus preventing them from reaching, attaching to and entering through mucosal membranes. This protective capacity remains in some of the fragments after enzymic degradation of the antibody molecule. Thus, SIgA can prevent activation of defence mechanisms such as the IgM and IgG antibodies in the blood and tissues of the infant, which activate the complement system and establish large numbers of granulocytes and other phagocytes that produce the pro-inflammatory cytokines IL-1β, TNFα, IL-6 and IL-8. The SIgA is more resistant to enzymic degradation than serum IgA; SIgA binds and neutralizes common bacterial enzymes that can degrade serum IgA (Plaut, Reference Plaut1978).
In clinical studies of breast-feeding it has been demonstrated that the levels of SIgA antibodies in the milk relate to protection against infections caused by enterotoxigenic E. coli, Vibrio cholerae, Campylobacter, Shigella and Giardia lamblia (Glass et al. Reference Glass, Svennerholm, Stoll, Khan, Hassain, Huq and Holmgren1983; Cruz et al. Reference Cruz, Gil, Cano, Caceres and Pareja1988; Ruiz-Palacios et al. Reference Ruiz-Palacios, Calva, Pickering, Lopez-Vidal, Volkow, Pezzarossi and West1990; Hayani et al. Reference Hayani, Guerrero, Morrow, Gomez, Winsor, Ruiz-Palacios and Cleary1992; Walterspiel et al. Reference Walterspiel, Morrow, Guerrero, Ruiz-Palacios and Pickering1994; Long et al. Reference Long, Vasquez-Garibay, Mathewson, de la Cabada and DuPont1999).
In addition to the anti-bacterial effects of the milk SIgA, it appears that its carbohydrate side chains may function as promoters of the growth of E. coli with type 1 pili (F Nowrouzian, I Adlerberth, AE Wold and V Friman, unpublished results). These E. coli are generally of low virulence and may compete for space and nutrients with potentially more-pathogenic bacteria in the breast-fed infant's gut. It has also been suggested that human SIgA supports the formation of a thin biofilm on the epithelial surface of the gut, which might promote the normal microbial colonization of the gut (Bollinger et al. Reference Bollinger, Everett, Palestrant, Love, Lin and Parker2003).
Against this background it obviously makes sense to start breast-feeding directly after delivery, thus providing the neonate with SIgA-mediated protection. Indeed, a recent study in Ghana has shown that when breast-feeding is started within 1 h of delivery 22% of neonatal deaths are prevented, while the effect of starting breast-feeding within 1 d of birth is a 16% reduction as compared with starting at 3 d of age (Edmond et al. Reference Edmond, Zandoh, Quigley, Amenga-Etego, Owusu-Agyei and Kirkwood2006). This finding may reflect the protective capacity not only of milk SIgA antibodies, but also of additional protective components in the milk, some of which will be described.
This major milk protein is present in colostrum at a concentration of about 5–7 g/l, which decreases to 1–3 g/l in mature milk (Goldman et al. Reference Goldman, Garza, Nichols and Goldblum1982). It is a member of the transferrin family and consists of two lobes, each of which binds Fe. It is present in all exocrine secretions and also in granulocytes. There is a specific receptor in the gut for lactoferrin and lactoferrin fragments (Kawakami & Lonnerdal, Reference Kawakami and Lonnerdal1991).
Lactoferrin is relatively resistant to degradation by trypsin and chymotrypsin, and stools of breast-fed babies contain considerable amounts of lactoferrin. Both lactoferrin and its fragments are bactericidal for many Gram-positive and Gram-negative bacteria. This effect does not seem to be related to the binding of Fe, which is needed by most bacteria; rather, it is the result of a destabilizing effect on the outer cell membrane of bacteria, making them more sensitive to lysozyme, which is also present in the milk (Ellison, Reference Ellison, Hutchens, Rumballand and Lönnerdal1994; Leitch & Willcox, Reference Leitch and Willcox1998). Lactoferrin promotes growth of the strictly anaerobic Bifidobacteria, but it also cleaves colonization factors on the potential pathogen Haemophilus influenzae and hinders the mucosal adherence of enteropathogenic E. coli (Hendrixson et al. Reference Hendrixson, Qii, Shewry, Fink, Petty, Baker, Plat and St Gem2003; Ochoa et al. Reference Ochoa, Noguera-Obenza, Ebel, Guzman, Gomez and Cleary2003). Furthermore, it has antiviral effects and acts against fungi such as Candida albicans (Nikawa et al. Reference Nikawa, Samaranayake, Tenovuo and Hamada1994). Lactoferrin, like SIgA, seems to have the capacity to decrease the risk of infections caused by bacteria, viruses and possibly fungi without the use of inflammatory mechanisms.
In addition, lactoferrin has the capacity to enter the nuclei of leucocytes and block the transcription factor NF-κB, which otherwise induces the production of the pro-inflammatory cytokines IL-1β, TNFα, IL-6 and IL-8. These cytokines cause inflammation, tenderness, tiredness, fever and loss of appetite by increasing leptin, while giving rise to many more leucocytes at the site of the infection that has initiated this host response. The inflammation associated with dextran sulphate-induced colitis in mice is reduced by human lactoferrin and some of its fragments (Haversen et al. Reference Haversen, Baltzer, Dolphin, Hanson and Mattsby-Baltzer2003). This effect is paralleled by reduced levels of IL-1β in the blood and fewer TNFα-producing cells. It seems advantageous for breast-fed infants to have their host defence supported by a milk protein that helps prevent infections and at the same time may suppress the infant's inflammatory response, which would otherwise disturb their health, appetite and growth (Togawa et al. Reference Togawa, Nagase, Tanaka, Inamori, Umezawa, Nakajima, Naito, Sato, Saito and Sekihara2002).
Lactoferrin and fragments of lactoferrin are taken up by a receptor in the gut and appear in the urine of breast-fed infants (Goldblum et al. Reference Goldblum, Schanler, Garza and Goldman1989). This process may explain why human lactoferrin and certain active fragments of lactoferrin given perorally to mice can protect against experimental urinary tract infections (Haversen et al. Reference Haversen, Engberg, Baltzer, Dolphin, Hanson and Mattsby-Baltzer2000). Clinical studies reported by Mårild et al. (Reference Mårild, Hansson, Jodal, Oden and Svedberg2004) also suggest that breast-feeding protects against urinary tract infections in children. Furthermore, some experimental studies (Edde et al. Reference Edde, Hipolito, Hwang, Headon, Shalwitz and Sherman2001; Gomez et al. Reference Gomez, Herrera-Insua, Siddiqui, Diaz-Gonzalez, Caceres, Newburg and Cleary2001) suggest that lactoferrin may act against intestinal infections caused by E. coli and Shigella flexneri.
Carbohydrate components in milk interfering with microbial attachment to mucosal membranes
Human milk contains substantial amounts of oligosaccharides, glycoproteins and glycolipids. The oligosaccharides constitute the third-largest solid fraction in milk, after lactose and fat (Kunz & Rudloff, Reference Kunz and Rudloff1993). There are more than ninety different oligosaccharides, which are produced in the mammary glands. Only about 1% is absorbed and appears in the urine (Rudloff et al. Reference Rudloff, Pohlentz, Diekmann, Egge and Kunz1996). These oligosaccharides seem to affect the composition of the gut microflora and may partly explain why breast-fed children carry potentially pathogenic E. coli, Klebsiella and other Enterobacteriacae strains less often than non-breast-fed children (Gothefors et al. Reference Gothefors, Olling and Winberg1975; Uauy & Araya, Reference Uauy and Araya2004).
To be able to infect a host via the mucosal membranes, where most infections occur, the microbes must attach to the mucosal cells. This process is specific, with different microbes binding to different carbohydrate structures on the cell surface (Newburg et al. Reference Newburg, Ruiz-Palacios and Morrow2005). The carbohydrate components of human milk function as receptor analogues and can therefore prevent microbes from binding to the carbohydrate moiety on the mucosal epithelium to which they are specifically adapted to bind. Certain bacterial toxins can be blocked in a similar way. Such specific anti-adhesion effects of milk oligosaccharides on the binding of microbes to mucosal epithelium have been demonstrated for: diarrhogenic E. coli, Campylobacter jejuni, V. cholerae and a Salmonella; otitis-causing Streptococcus peumoniae and H. influenzae; HIV-1 and numerous other microbes (Korhonen et al. Reference Korhonen, Valtonen, Parkkinen, Vaisanen-Rhen, Finne, Orskov, Orskov, Svenson and Makela1985; Andersson et al. Reference Andersson, Porras, Hanson, Lagergård and Svanborg1986; Newburg, Reference Newburg1999; Ruiz-Palacios et al. Reference Ruiz-Palacios, Cervantes, Ramos, Chavez-Munguia and Newburg2003; Coppa et al. Reference Coppa, Zampini, Galeazzi, Facinelli, Ferrante, Capretti and Orazio2006).
A similar capacity to prevent microbial attachment to specific mucosal carbohydrate structures has been demonstrated for glycoconjugates in human milk. The milk glycolipid Gb3 prevents binding of Shigella dysenteriae and a Shiga-like toxin from enterohaemorrhagic E. coli (Newburg, Reference Newburg1997). Similarly, the ganglioside GM1 blocks adhesion of V. cholerae and binds to the E. coli, V. cholerae and C. jejuni enterotoxins (Holmgren et al. Reference Holmgren, Svennerholm and Lindblad1983; Otnaess et al. Reference Otnaess, Laegreid and Ertresvag1983; Ruiz-Palacios et al. Reference Ruiz-Palacios, Torres, Torres, Escamilla, Ruiz-Palacios and Tamayo1983). Respiratory syncytial virus (Laegreid et al. Reference Laegreid, Otnaess and Fuglesang1986) and hepatitis A virus (Zajac et al. Reference Zajac, Amphlett, Rowlands and Sangar1991) are each neutralized by a different milk glycoprotein.
Another mechanism is illustrated by a Mac-2 protein, plentiful in milk, that supports the defence system by helping macrophages to bind to microbes as the initial step before engulfing and destroying them (Fornarini et al. Reference Fornarini, Iacobelli, Tinari, Natoli, De Martino and Sabatino1999).
Mucin, and the fat globules in which they are mainly found, block the binding of E. coli to epithelial cells and inhibit replication of rotavirus (Schroten et al. Reference Schroten, Hanisch, Plogmann, Hacker, Uhlenbruck, Nobis-Bosch and Wahn1992). Lactadherin, which is a mucin-associated glycoprotein, inhibits rotavirus and prevents its replication, presumably reducing the risk of this very common infectious agent (Newburg et al. Reference Newburg, Peterson, Ruiz-Palacios, Matson, Morrow and Shults1998). Similarly, casein components seem to reduce cellular binding by Actinomyces and streptococci, whereas κ-casein inhibits mucosal attachment by Helicobacter pylori (Neeser et al. Reference Neeser, Chambaz, Del Vedovo, Prigent and Guggenheim1988; Strömqvist et al. Reference Strömqvist, Falk, Bergström, Hansson, Lönnerdal, Normark and Hernell1995).
This human peptide, which is anti-inflammatory and anti-secretory, is produced in response to bacterial enterotoxins and also as a result of eating a specially-treated cereal (Lange & Lonnroth, Reference Lange and Lönnroth2001). It appears in human milk and has been shown in a placebo-controlled study to prevent clinical mastitis (Svensson et al. Reference Svensson, Lange, Lönnroth, Widström and Hanson2004). In a double-blind placebo-controlled study of acute or prolonged diarrhoea in 240 Pakistani children (Zaman et al. Reference Zaman, Mannan, Lange, Lönnroth and Hanson2007) it was found to be protective. When giving the anti-secretory factor orally together with oral rehydration it was found that recovery within 3 d is attained in an additional 30% of the children.
After a specific reorganization of the molecule a major milk protein, α-lactalbumin, has been found to have a striking effect on human tumour cell lines (Svanborg et al. Reference Svanborg, Ågerstam, Aronson, Bjerkvist, Duringer and Fischer2003) and has been termed ‘human α-lactalbumin made lethal to tumour cells’. In vivo effects on human papilloma and in vitro effects on numerous human tumour cell lines, as well as on a human glioblastoma cell line in an animal model, have been demonstrated (Pettersson et al. Reference Pettersson, Mossberg and Svanborg2006). It is not known whether ‘α-lactalbumin made lethal to tumour cells’ can be the basis for the claims that breast-feeding may reduce childhood leukaemia and maternal breast cancer.
Additional potentially-protective and anti-inflammatory factors and signals
It has been suggested that the milk lysozyme, in cooperation with lactoferrin and SIgA, might be involved in the elimination of E. coli (Adinolfi et al. Reference Adinolfi, Glynn, Lindsay and Milne1966). These claims have yet to be confirmed. Anti-microbial effects of lipid components and a lipase in human milk have also been studied (Gillin et al. Reference Gillin, Reiner and Gault1985; Hernell et al. Reference Hernell, Ward, Blackberg and Pereira1986). The presence of the β-defensin LBD-1 has been demonstrated in human milk (Tunzi et al. Reference Tunzi, Harper, Bar-Oz, Valore, Semple, Watson-MacDonell, Ganz and Ito2000). For all these and several additional factors in milk there is a need for further research in order to define their potential biological role in the breast-fed infant. This position is certainly relevant to the many cytokines, hormones and growth factors present in human milk.
It has been proposed that the presence of IL-7 in milk is linked to the fact that the central organ in the immune system, the thymus, is twice as large in breast-fed infants compared with non-breast-fed infants (Hasselbalch et al. Reference Hasselbalch, Engelmann, Ersboll, Jeppesen and Fleischer-Michaelsen1999; Ngom, Reference Ngom, Collinson, Pido-Lopez, Henson, Prentice and Aspinall2004). Studies in Africa have shown that a smaller thymus at birth predicts a higher infant mortality from infections, independent of other factors known to reduce the size of the thymus such as birth weight and malnutrition (Aaby et al. Reference Aaby, Marx, Trautner, Rudaa, Hasselbalch, Jensen and Lisse2002). IL-7 seems to enhance not only the size of the thymus, but also has a stimulating effect on its output of the Tγδ lymphocytes aggregated in the cryptopatches in the intestinal mucosa (Laky et al. Reference Laky, Lewis, Tigelaar and Puddington2003). Among the many cytokines also present in milk is transforming growth factor-β, which like IL-10 down regulates inflammation (Hvas et al. Reference Hvas, Kelsen, Agnholt, Hollsberg, Tvede, Moller and Dahlerup2007). In a mouse study (Letterio et al. Reference Letterio, Geiser, Kulkarni, Roche, Sporn and Roberts1994) transforming growth factor-β from milk was found to be absorbed in the gut and to depress the inflammation that occurs in mice lacking the gene for transforming growth factor-β. Human milk contains soluble receptors for IL-1β and for TNFα, resulting in blockage of the pro-inflammatory effects of these cytokines (Garofalo & Goldman, Reference Garofalo and Goldman1999). Numerous other milk components are also anti-inflammatory, involving many different mechanisms: lactoferrin, as mentioned earlier, blocks cytokine production via NF-κB in leucocyte nuclei (Haversen et al. Reference Haversen, Baltzer, Dolphin, Hanson and Mattsby-Baltzer2003); complement is inhibited by lysozyme, lactoferrin, α-lactalbumin, soluble complement inhibitors and complement regulatory factors (Ogundele, Reference Ogundele1999); PG in milk inhibit neutrophil enzymes and are cytoprotective; several anti-proteases in milk, such as α1-antitrypsin, α1-antichymotrypsin and elastase inhibitor, block potentially-tissue-damaging enzymes (Garofalo & Goldman, Reference Garofalo and Goldman1999). It should be added that the large amounts of SIgA antibodies in the colostrum and milk also act as an anti-inflammatory agent by binding microbes, thus preventing them from attaching to and invading the mucosal membranes in the respiratory and gastrointestinal tracts, where they would bind to and activate Toll-like receptors on leucocytes in those sites.
Furthermore, there are numerous antioxidants in human milk, such as catalase, lactoferrin, α-tocopherol, β-carotene, l-histidine and ascorbic acid, which prevent hydroxyl radical formation and lipid peroxidation, degrade superoxide and scavenge oxygen radicals (Garofalo & Goldman, Reference Garofalo and Goldman1999).
Colostrum and milk contain high levels of soluble CD14; more than twenty times the concentration in serum (Labeta et al. Reference Labeta, Vidal, Nores, Arias, Vita and Morgan2000). This component helps phagocytes in the gut to be activated, via their Toll-like receptor 4 receptors, by Gram-negative bacteria for which these receptors are specific. Gram-positive bacteria will in a corresponding manner bind to their specific receptor, the Toll-like receptor 2, on phagocytes. Again, this binding is enhanced by the CD14 from the milk (Vidal et al. Reference Vidal, Labeta, Schiffrin and Donnet-Hughes2001). Such a function may help to defend the mammary gland against infections, but on the other hand may bring about clinical and/or subclinical mastitis caused by inflammation in the mammary glands (Filteau, Reference Filteau2003).
Human milk also contains many more signals from the mother to her offspring, such as additional cytokines, chemokines, hormones, growth factors, factors blocking cytokine receptors, complement-inhibiting factors, factors blocking cytokine receptors, maturation factors, soluble Toll-like receptors etc. Little is known about their possible effects in relation to breast-feeding.
Breast-feeding and protection against diseases in the infant
Numerous studies have been performed to determine whether breast-feeding protects against infections and immunological diseases such as asthma and allergy, autoimmune diseases and also tumours. A problem with such studies is that for ethical reasons it is difficult to undertake completely controlled studies, since breast-feeding and non-breast-feeding cannot easily be randomized and blinded. However, there are several studies that provide useful information, although variations in living conditions, dietary intake and exposure to infectious agents in different populations make it important to evaluate the outcomes with caution. The following is a brief review of such studies.
Reduction in infant deaths by breast-feeding
In developing countries that have very high infant mortality and high fertility breast-feeding serves to diminish both factors, which together provide a strong protective effect (Hanson et al. Reference Hanson, Ashraf, Zaman, Karlberg, Khan, Lindblad and Jalil1994; Labbok et al. Reference Labbok, Hight-Laukaran, Peterson, Fletcher, von Hertzen and Van Look1997, Reference Labbok, Clark and Goldman2004). Several studies have demonstrated protection by breast-feeding in early life via the numerous anti-infectious and anti-inflammatory components in the colostrum and mature milk. This effect is illustrated in a study in Brazil (Victora et al. Reference Victora, Smith, Vaughan, Nobre, Lombardi, Teixeira, Fuchs, Moreira, Gigante and Barros1987), in which it was found that exclusive breast-feeding reduces the risk of dying as a result of diarrhoea by 14·2-fold, whereas partial breast-feeding is associated with a reduction of 4·2-fold compared with no breast-feeding. A recent WHO report (Bahl et al. Reference Bahl, Frost, Kirkwood, Edmond, Martines, Bhandari and Arthur2005), based on combined results from Ghana, India and Peru, has shown that there is no significantly different risk of death when comparing infants who are exclusively and predominantly breast-fed. However, the hazard ratios obtained when comparing infants who are not breast-fed with those predominantly breast-fed and those partially breast-fed respectively are 10·5 (P<0·001) and 2·46 (P<0·001). Based on the evident protective capacity of breast-feeding, it has been suggested that not breast-feeding may be the most common immunodeficiency in infancy (Hanson, Reference Hanson1998).
A recent important investigation in Ghana, which was part of the WHO study (Bahl et al. Reference Bahl, Frost, Kirkwood, Edmond, Martines, Bhandari and Arthur2005), has demonstrated, as mentioned earlier, that starting breast-feeding within 1 h of delivery reduces infant mortality by 22% compared with starting after 3 d, which is the most common practice in rural Ghana (Edmond et al. Reference Edmond, Zandoh, Quigley, Amenga-Etego, Owusu-Agyei and Kirkwood2006). Initiating breast-feeding from day 1 was reported to result in a reduction in mortality of 16%, illustrating how supportive the breast-milk defence components are in handling the early microbial colonization in a region of Ghana that has heavy microbial exposure. A careful analysis with robust statistics has claimed that promoting breast-feeding could prevent as many as 720 postneonatal deaths per year in USA (Chen & Rogan, Reference Chen and Rogan2004).
However, a recent critical review of the protective capacity of breast-feeding against early infections in preterm infants indicates that many, if not most, previous studies do not fulfill the strict criteria required for critical evaluation (de Silva et al. Reference de Silva, Jones and Spencer2004).
Breast-feeding and protection against neonatal infections
Several earlier reports have claimed to show protection against neonatal septicaemia and meningitis (Hanson, Reference Hanson, Hale and Hartmann2007). In a planned prospective study Narayanan et al. (Reference Narayanan, Prakash, Prabhakar and Gujral1982) fed expressed breast milk to low-birth-weight babies and noted significant protection compared with feeding formula (P<0·001). Cases of moderate to severe diarrhoea caused by Campylobacter and calicivirus were found to occur less often if the mother's milk is high in 2-linked fucosylated oligosaccharides. This effect may be associated with the capacity of such milk components to function as analogues to the specific mucosal receptors to which these microbes have to bind to achieve contact with the host (Morrow et al. Reference Morrow, Ruiz-Palacios, Altaye, Jiang, Guerrero, Meinzen-Derr, Farkas, Chaturvedi, Pickering and Newburg2004). It has been suggested that breast-feeding may also support host defence in the neonate by preventing hypoglycaemia and hypothermia (Huffman et al. Reference Huffman, Zehner and Victora2001). A recent analysis of late-onset septicaemia in extremely-premature infants has indicated that the lower mortality in Norway compared with studies in other regions may be linked to the practice of very early full enteral feeding with human milk (Ronnestad et al. Reference Ronnestad, Abrahamsen, Medbo, Reigstad, Lossius, Kaaresen, Egeland, Engelund, Irgens and Markestad2005). Without establishing such feeding within the second week of life the adjusted relative risk of late-onset septicaemia was found to be 3·7 (95% CI 2·0, 6·9).
Breast-feeding and necrotizing enterocolitis
This condition presumably illustrates how bacteria colonize the newborn and are not prevented from reaching the intestinal mucosa where the infant has some pro-inflammatory tissue defence, but little or no mucosal defence of its own. Attaching to the mucosa and even infiltrating it because of the lack of an efficient defence, the microbes will initiate an unbalanced inflammatory response that may result in local necrosis and even penetration of the gut mucosa (Lucas & Cole, Reference Lucas and Cole1990; Schanler et al. Reference Schanler, Shulman, Lau, Smith and Heitkemper1999; McGuire & Anthony, Reference McGuire and Anthony2003). This condition has a high mortality, but breast-feeding reduces the risk, presumably by providing mucosal protection and suppression of inflammation. This protection is probably mediated, for example, by the milk SIgA antibodies, which through the enteromammaric link described earlier are directed against the microbes of the mother and her environment, and can bind them to prevent them from attaching to and infiltrating the mucosa (Fig. 1). In addition, numerous other milk components may be protective in this situation, including the many milk oligosaccharides functioning as receptor analogues and thus preventing the microbes from reaching the infant's gut mucosa. The milk contains an enzyme that degrades the pro-inflammatory platelet-activating factor (Furukawa et al. Reference Furukawa, Narahara, Yasuda and Johnston1993). The milk also contains the anti-inflammatory cytokines IL-10 (Fituch et al. Reference Fituch, Palkowetz, Goldman and Schanler2004) and transforming growth factor-β. The milk defensin might be supportive. In addition, breast-feeding seems to promote a probiotic-like microflora that may also be protective by reducing the number of potentially-dangerous microbes by competition for nutrients and space, as discussed earlier.
Breast-feeding and sudden death in infancy
There is evidence that breast-feeding provides protection against this condition, but the effect is weak (Alm et al. Reference Alm, Wennergren, Norvenius, Skjaerven, Lagercrantz, Helweg-Larsen and Irgens2002). Other protective factors are more important, especially not letting the infant sleep on its stomach.
Breast-feeding and diarrhoea
Diarrhoea, as a common cause of death in infancy, is a major contributor to infant mortality, and it is obviously important to determine the protective capacity of breast-feeding. Numerous studies have confirmed that breast-feeding is associated with a marked reduction in morbidity and mortality (Glass & Stoll, Reference Glass and Stoll1989; Howie et al. Reference Howie, Forsyth, Ogston, Clark and Florey1990; Victora, Reference Victora, Atkinson, Hanson and Chandra1990). In a large randomized study in Belarus it was found that breast-feeding protects against gastroenteritis during the first year of life (Kramer et al. Reference Kramer, Chalmers, Hodnett, Sevkovskaya, Dzikovich and Shapiro2001).The study has also shown that exclusive breast-feeding for 6 months results in fewer cases of gastroenteritis than exclusive breast-feeding for 3 months (Kramer et al. Reference Kramer, Guo, Platt, Sevkovskaya, Dzikovich and Collet2003). A UK study (Quigley et al. Reference Quigley, Cumberland, Cowden and Rodrigues2006) has shown, after adjusting for confounders, that breast-feeding protects against gastroenteritis when comparing no milk with any milk (OR 2·74; P<0·005), or not exclusive breast-feeding with exclusive breast-feeding (OR 3·62; P<0·006). The protection reached an OR of 5 in less-privileged areas when comparing no milk with any milk and an OR of 17·66 when comparing no breast-feeding with exclusive breast-feeding.
The protection against diarrhogenic bacteria and their toxins appears to be mediated via the milk SIgA antibodies, according to studies of infections with enterotoxigenic E. coli, Shigella, V. cholerae and G. lamblia (Glass et al. Reference Glass, Svennerholm, Stoll, Khan, Hassain, Huq and Holmgren1983; Cruz et al. Reference Cruz, Gil, Cano, Caceres and Pareja1988; Ruiz-Palacios et al. Reference Ruiz-Palacios, Calva, Pickering, Lopez-Vidal, Volkow, Pezzarossi and West1990; Hayani et al. Reference Hayani, Guerrero, Morrow, Gomez, Winsor, Ruiz-Palacios and Cleary1992; Long et al. Reference Long, Vasquez-Garibay, Mathewson, de la Cabada and DuPont1999). The human peptide anti-secretory factor, which can be induced in human milk (Svensson et al. Reference Svensson, Lange, Lönnroth, Widström and Hanson2004), has recently been shown to give protection against acute diarrhoea as well as prolonged diarrhoea (Zaman et al. Reference Zaman, Mannan, Lange, Lönnroth and Hanson2007). A Mexican study (Morrow et al. Reference Morrow, Reves, West, Guerrero, Ruiz-Palacios and Pickering1992) has shown that breast-feeding is associated with a 5-fold lower risk of diarrhoea caused by G. lamblia compared with no breast-feeding, and a 1·8-fold reduction when comparing partial breast-feeding with no breast-feeding. In relation to protection against rotavirus infections by breast-feeding, a lack of clear protection has been noted or, alternatively, a postponement of the disease or asymptomatic infections (Duffy et al. Reference Duffy, Byers, Riepenhoff-Talty, La Scolea, Zielezny and Ogra1986; Clemens et al. Reference Clemens, Rao, Eng, Ahmed, Ward and Huda1993). A delaying effect on rotavirus appearance in the stool associated with higher levels of SIgA antibodies has been reported (Espinoza et al. Reference Espinoza, Paniagua, Hallander, Svensson and Strannegård1997). However, another study of nosocomial rotavirus infections has demonstrated protection by breast-feeding (Gianino et al. Reference Gianino, Mastretta, Longo, Laccisaglia, Sartore, Russo and Mazzaccara2002).
Breast-feeding and respiratory tract infections
Investigations of the protective capacity of breast-feeding against otitis media have usually shown efficient protection against acute and prolonged infections (Cushing et al. Reference Cushing, Samet, Lambert, Skipper, Hunt, Young and McLaren1998; Dewey et al. Reference Dewey, Heinig and Nommsen-Rivers1995; Duncan et al. Reference Duncan, Ey, Holberg, Wright, Martinez and Taussig1993). It has been suggested that breast-feeding may protect against upper respiratory infections, but this role needs confirmation (Howie et al. Reference Howie, Forsyth, Ogston, Clark and Florey1990). In deprived areas breast-feeding is strongly protective against pneumonia (Victora et al. Reference Victora, Smith, Vaughan, Nobre, Lombardi, Teixeira, Fuchs, Moreira, Gigante and Barros1987; Cesar et al. Reference Cesar, Victora, Barros, Santos and Flores1999). However, the previously-mentioned well-controlled Belarus study (Kramer et al. Reference Kramer, Guo, Platt, Sevkovskaya, Dzikovich and Collet2003) has shown no difference in the prevalence of pneumonia when comparing infants exclusively breast-fed for 3 and 6 months. Other studies from the USA and Australia (Ford & Labbok, Reference Ford and Labbok1993; Oddy et al. Reference Oddy, Sly, de Klerk, Landau, Kendall, Holt and Stanley2003) have found that breast-feeding reduces the risk of developing pneumonia. Another US study (Wright et al. Reference Wright, Holberg, Martinez, Morgan and Taussig1989) has described protection by breast-feeding against wheezing respiratory tract infections during the first 4 months of life. A Norwegian investigation (Nafstad & Jaakola, Reference Nafstad and Jaakola2003) has shown that protection by breast-feeding against respiratory tract infections is more obvious if the mother smokes. A study has been conducted in USA to compare the effects of exclusive breast-feeding for 4 and 6 months on the appearance of pneumonia and otitis media, after adjustment for confounding factors (Chantry et al. Reference Chantry, Howard and Auinger2006). A significant reduction in infections was found in the 6-month group compared with the 4-month group, decreasing the risk of attacks of pneumonia (OR 4·27 (95% CI 1·27, 14·35)) and of having up to three attacks of otitis media (OR 1·95 (95% CI 1·06, 3·59)). A recent investigation indicates that breast-feeding protects against clinical measles infection (Silfverdal & Montgomery, Reference Silfverdal and Montgomery2007).
Breast-feeding and urinary tract infections
In addition to an Italian study (Pisacane et al. Reference Pisacane, Graziano, Mazzarella, Scarpellino and Zona1992), a recent more-extensive prospective case–control study in Sweden has been reported (Mårild et al. Reference Mårild, Hansson, Jodal, Oden and Svedberg2004). Both studies show protection, but the Swedish study has demonstrated that longer duration of breast-feeding reduces the risk of infection among girls, with a similar trend among boys. Breast-feeding until 7 months results in enhanced protection for ⩽2 years of age.
A Spanish group (Talayero et al. Reference Talayero, Liza'n-Garcia, Puime, Muncharaz, Soto, Sanchez-Palomares, Serrano and Rivera2006) has investigated the effect of breast-feeding on hospitalization; cases of perinatal infections were excluded. The study was based on 1385 infants during the years 1996 and 1999. It was estimated that 30% of the seventy-eight hospital admissions caused by infections would have been avoided for each additional month of full breast-feeding. The authors suggest that full breast-feeding at 4 months of age would have prevented 56% of hospital admissions among the infants before 1 year of age.
The American Academy of Pediatrics (Gartner et al. Reference Gartner, Morton, Lawrence, Naylor, O'Hare, Schanler and Eidelman2005) has recently made a policy statement (Breast-feeding and the use of human milk) that summarizes the evidence for the benefits of breast-feeding, including protection against infections.
Long-term effects on the offspring of breast-feeding
It is well known that the transplacentally-transferred maternal IgG antibodies can inhibit the infant's own immune response, e.g. against measles vaccine. For this reason it is preferable, when possible, to start vaccination against measles after 12 months of age.
Breast-feeding, on the other hand, does not inhibit vaccine responses, with one exception; peroral vaccination with live poliovirus will fail if the infant is breast-fed too close before or after the dose is given. The milk antibodies will otherwise neutralize the live vaccine virus (World Health Organization, 1995). Some studies have shown that breast-feeding may enhance vaccine responses. A long-lasting enhancing effect was noted for the IgG2 antibody response to vaccination against Haemophilus influenzae type b (Silfverdal et al. Reference Silfverdal, Bodin, Ulanova, Hahn-Zoric, Hanson and Olcen2002). This result has been confirmed and the study expanded to include similar effects by breast-feeding on the antibody response against pneumococci (Silfverdal et al. Reference Silfverdal, Ekholm and Bodin2007). Exclusive breast-feeding for ≥90 d provides a higher proportion of infants with vaccine responses above protective levels. Previously, it has been shown that breast-feeding may enhance vaccine responses to tetanus, diphtheria, live poliovirus and Haemophilus influenzae type b (Hahn-Zoric et al. Reference Hahn-Zoric, Fulconis, Minoli, Moro, Carlsson, Böttiger, Räihä and Hanson1990; Pabst & Spady, Reference Pabst and Spady1990). However, some studies of vaccine responses to tetanus toxoid and Haemophilus influenzae type b vaccines have shown no such enhancing effects (Stephens et al. Reference Stephens, Kennedy, Lakhani and Brenner1984; Watemberg et al. Reference Watemberg, Dagan, Arbelli, Belmaker, Morag, Hessel, Fritzell, Bajard and Peyron1991; Decker et al. Reference Decker, Edwards, Bradley and Palmer1992). Such differences could be a result of, for example, variations in levels of transplacentally-transferred maternal antibodies against the vaccines used and possibly to differences in the amounts of relevant components in the mothers' milk.
Protection against infections
There are several studies that indicate that breast-feeding may enhance long-term protection against certain infections, e.g. gastroenteritis, respiratory tract infections, skin infections, urinary tract infections and severe complications to measles infections. The data has recently been reviewed (Hanson, Reference Hanson2004, Reference Hanson, Hale and Hartmann2007).
Effects against inflammatory diseases in the infant–child–adult
There have been many recent studies that have investigated the possibility that breast-feeding may affect the risk of developing the many common diseases for which it is known that inflammatory processes play a central role in the pathogenesis. As there are a multitude of components that affect the development and function of the immune system, and various forms of inflammation in the growing individual, many factors in the milk could be involved in such effects. There are very many studies of the possible role of breast-feeding in various forms of allergic diseases, several showing protection, but some the reverse. The literature is very extensive and the results are not easy to summarize briefly; however, they have been the subject of recent reviews (Hanson, Reference Hanson2004, Reference Hanson, Hale and Hartmann2007). Positive effects of breast-feeding on inflammatory autoimmune diseases such as diabetes mellitus types 1 and 2, rheumatoid arthritis, coeliac disease, ulcerative colitis and Crohn's disease have been published and also summarized recently, as indicated earlier. Further studies are needed to confirm many of the results obtained.
Numerous investigators have studied the possibility that breast-feeding may prevent overweight and obesity in children and adults. Protective effects against obesity have been demonstrated in some large European studies (von Kries et al. Reference von Kries, Koletzko, Sauerwald, von Mutius, Barnert, Grunert and von Voss1999; Toschke et al. Reference Toschke, Vignerova, Lhotska, Osancova, Koletzko and Von Kries2002). A dose–dependent protective effect has been noted in meta-analyses (Arenz & von Kries, Reference Arenz and von Kries2005; Harder et al. Reference Harder, Bergmann, Kallischnigg and Plagemann2005). The effect may remain into adolescence (Gillman et al. Reference Gillman, Rifas-Shiman, Berkey, Frazier, Rockett, Camargo, Field and Colditz2006). A similar observation has been reported in a Norwegian study, but with less effect in adulthood (Kvaavik et al. Reference Kvaavik, Tell and Klepp2005). In large investigations from South and North America protective effects have been noted in some population groups, but not in others; it has also been proposed that the effect may be temporary (Bogen et al. Reference Bogen, Hanusa and Whitaker2004; Grummer-Strawn & Mei, Reference Grummer-Strawn and Mei2004; Burke et al. Reference Burke, Beilin, Simmer, Oddy, Blake, Doherty, Kendall, Newnham, Landau and Stanley2005; Kersey et al. Reference Kersey, Lipton, Sanchez-Rosado, Kumar, Thisted and Lantos2005; Araujo et al. Reference Araujo, Victora, Hallal and Gigante2006). Unmeasured confounding factors may explain the positive effects published (Nelson et al. Reference Nelson, Gordon-Larsen and Adair2005). It is possible that breast-feeding can protect against overweight and obesity, but at a rather low level at higher ages, with the effect impaired by many factors (Owen et al. Reference Owen, Martin, Whincup, Davey-Smith, Gillman and Cook2005).
There have been claims that breast-feeding has protective effects against CVD, including hypertension, dyslipidaemia, obesity and/or insulin resistance (Singhal et al. Reference Singhal, Cole, Fewtrell and Lucas2004; Singhal & Lucas, Reference Singhal and Lucas2004). Additional studies support the existence of such effects (Lawlor et al. Reference Lawlor, Najman, Sterne, Williams, Ebrahim and Davey2004; Martin et al. Reference Martin, Gunnell and Smith2005b). These effects may be related to the many components in milk, e.g. leptin, grehlin and insulin-like growth factor, which may affect appetite and metabolism (Savino et al. Reference Savino, Fissore, Grassino, Nanni, Oggero and Silvestro2005; Hanson, Reference Hanson, Hale and Hartmann2007).
A meta-analysis has shown that breast-feeding protects against tumours in childhood such as acute lymphoblastic leukaemia, Hodgkin's disease and neuroblastoma (Martin et al. Reference Martin, Gunnell, Owen and Smith2005a); ⩽5% of childhood acute leukaemia and lymphoma may be prevented by increasing breast-feeding from 50% to 100%.
Tumours in adulthood, e.g. prostate, colo-rectal and gastric cancers are not reduced by breast-feeding, but there is a reduction in premenopausal breast cancer (Martin et al. Reference Martin, Middleton, Gunnell, Owen and Smith2005c). Each pregnancy diminishes the risk of breast cancer by 7% and this risk is reduced by another 4·3 years for each year of breast-feeding (Martin et al. Reference Martin, Middleton, Gunnell, Owen and Smith2005c). The recent discovery of the anti-tumour effect of ‘α-lactalbumin made lethal to tumour cells’, the reorganized milk α-lactalbumin, suggests that it might be involved in some of these anti-tumour effects of human milk (Svanborg et al. Reference Svanborg, Ågerstam, Aronson, Bjerkvist, Duringer and Fischer2003; Pettersson et al. Reference Pettersson, Mossberg and Svanborg2006).
Recent studies suggest that feeding human milk may also support the developmental outcome of extremely-low-birth-weight infants (Vohr et al. Reference Vohr, Poindexter, Dusick, McKinley, Wright, Langer and Poole2006). Other studies have debated whether breast-feeding can enhance IQ development, but a recent extensive investigation has found little or no effect (Der et al. Reference Der, Batty and Deary2006). It has been proposed that breast-feeding may enhance the resilience of children to later psychosocial stress (Montgomery et al. Reference Montgomery, Ehlin and Sacker2006).
Have the mammary glands developed from the innate immune system?
The skin glands that produce milk are unique to mammals. In a recent review their morphological and functional origin have been discussed in relation to their double function of providing the offspring with both nutrients and defence against infectious agents (Vorbach et al. Reference Vorbach, Capecchi and Penninger2006).
Most animal species use innate immune responses for their defence against infectious agents (Kimbrell & Beutler, Reference Kimbrell and Beutler2001). In contrast to the acquired specific immune system based on specific antibodies and lymphocytes primarily directed against infectious agents, the non-specific host defence provides broad defence of lesser specificity (Hanson, Reference Hanson, Hale and Hartmann2007). Innate defence is very complex and includes many different components and mechanisms. Vorbach et al. (Reference Vorbach, Capecchi and Penninger2006) suggest that lactation has evolved in incremental steps from providing protection to also offering nutrients. They highlight that the importance of milk for the offspring both for defence and nutrition is exemplified particularly by the unique evolution of two anti-microbial enzymes, xanthine oxidoreductase and lysozyme, both expressed in and secreted from the lactating mammary epithelium (Shahani et al. Reference Shahani, Herper, Jensen, Parry and Zittle1973). These two components are directly involved in the evolution of the nutritional capacity of milk as well as being antimicrobial.
Vorbach et al. (Reference Vorbach, Capecchi and Penninger2006) thus argue that the mammary gland and its product, the milk, initially developed from the innate immune system and that this origin could explain why there are so many components in the milk that play more than one role for defence and nutrition. Against this background it becomes easier to see how and why so many defence strategies have developed in the mammary glands and the milk, all defending without inducing inflammation in the offspring; some components, like lactoferrin, even actively counteract inflammation. Thus, appetite is not disturbed, as it is when tissue defence is initiated; a process that always induces inflammation, which is consistently costly, particularly for a growing individual, and also affects appetite, i.e. via pro-inflammatory cytokines increasing leptin levels.
Support for research forming the basis of part of this review was obtained from the Swedish International Development Agency, the Bergvall and the Philipsson Foundations and the Royal Society for Arts and Sciences in Göteborg, Sweden.