SEARCH

SEARCH BY CITATION

Keywords:

  • systematic review;
  • milk transforming growth factor-beta;
  • immunological outcomes;
  • infant and young child

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

Oddy WH, Rosales F. A systematic review of the importance of milk TGF-β on immunological outcomes in the infant and young child. Pediatr Allergy Immunol 2010: 21: 47–59. © 2009 Mead Johnson Nutrition Journal compilation © 2009 John Wiley & Sons A/S

Cytokines in milk like transforming growth factor-beta (TGF-β) have been shown to induce oral tolerance in experimental animal studies. However, human studies are less consistent with these findings. The primary objective of this review was to conduct a systematic review of published studies on the association between TGF-β identified in human milk and immunological outcomes in infancy and early childhood. Human prospective clinical studies were identified through MEDLINE, CAB Abstracts, Biological Abstracts and Scopus. Selection criteria included: well described populations of mothers and infants, time of milk sampling, immunological outcome measures and analytical methods of TGF-β determination. We considered a wide range of immunological outcomes in infancy and early childhood, such as wheeze, atopy, eczema and the immunoglobulin switch. Twelve human studies were included in the review and 67% showed a positive association with TGF-β1 or TGF-β2 demonstrating protection against allergy-related outcomes in infancy and early childhood. High variability in concentrations of TGF-β was noted between and within studies, some of it explained by maternal history of atopy or by consumption of probiotics. Human milk TGF-β appears to be essential in developing and maintaining appropriate immune responses in infants and may provide protection against adverse immunological outcomes, corroborating findings from experimental animal studies. Further large clinical studies in diverse human populations are indicated to confirm these results.

It has been little more than a decade since human milk was shown to contain cytokines (1). Cytokines are small soluble glycoproteins that act in an autocrine–paracrine fashion by binding to specific cellular receptors, operating in networks and orchestrating immune system development and function (2–6). Transforming growth factor-beta (TGF-β) is one cytokine identified in human breast milk (7). Human milk contains TGF-β1, TGF-β2 and other isoforms at both mRNA and protein levels with TGF-β2 being the major isoform (95%) (8).

The immunoactive factors in breast milk may influence the development and maturation of the mucosal immune system of the infant (9–15) and growing evidence suggests that TGF-β, a multifunctional polypeptide, may be a key immunoregulatory factor for the establishment of this response, by promoting IgA production as well as induction of oral tolerance (8, 13, 16–25).

Originally, the study of Shull (26) and Kulkarni and Karlsson (27) showed that TGF-β1 was a multifunctional growth factor with profound regulatory effects on many developmental and physiological processes. Shull (26) demonstrated that disruption of the TGF-β1 gene by homologous recombination in murine embryonic stem cells enabled mice to be generated that carry the disrupted allele, and that although animals homozygous for the mutated TGF-β1 allele showed no gross developmental abnormalities, about 20 days after birth they succumbed to a wasting syndrome accompanied by a multifocal, mixed inflammatory cell response and tissue necrosis, leading to organ failure and death. Subsequently, Letterio et al. (13) observed that TGF-β deficient mice (i.e. TGF-β1 gene knockout) survived for the first few weeks of life while breastfeeding, indicating that maternal sources of TGF-β1 via both placental transfer and milk are essential for the normal development and postnatal survival of these mice.

Since these studies were published there have been numerous reviews related to TGF-β regulation of immune responses (12, 28, 29) and other animal studies support the importance of this milk cytokine (30–32). Although the various mechanistic pathways by which TGF-β modulates the development and maintenance of the immune system, its role regulating tolerance and immunity has previously not been fully recognized, TGF-β appears to play an important immune-regulatory role on targets in autoimmunity, immunodeficiencies and other immune mediated-pathologies. In addition, oral administration of TGF-βin vivo in animal studies has been shown to result in biological activity sufficient to promote oral tolerance (33). Thus, this role would suggest that in an infant prone to having cow’s milk allergy (CMA), an increased TGF-β content of mother’s milk may be beneficial by promoting IgG–IgA antibody production and inhibiting IgE- and cell-mediated reactions to cow’s milk (25, 34). Verhasselt et al. (35) examined the role of milk-borne TGF-β by exposing lactating mice to an air-borne allergen and assessing the development of asthma in the progeny. They found that breastfeeding-induced tolerance relies on the presence of TGF-β during lactation. This effect was mediated by CD4+ T lymphocytes and depended on TGF-β signalling in T cells, and they concluded that breast milk-mediated transfer of an antigen and TGF-β to the neonate resulted in oral tolerance induction and antigen-specific protection from allergic airway disease.

Overall, these and other studies suggest that this particular cytokine in milk may influence the development of atopic disease, but to our knowledge the relationship of TGF-β in milk with the occurrence of illness in infancy has been little studied in humans, and this relationship requires further investigation. The objective of our systematic review was to obtain an unbiased perspective of the importance of TGF-β in maternal milk in relation to immunological outcomes in the infant and young child. The review was based on results from published human studies only, to answer the specific question ‘is TGF-β in maternal milk associated with immunological outcomes in the infant and young child?’.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

Literature sources and search terms

The studies considered for inclusion in this systematic review were identified by multiple pass searches of computerized bibliographic databases conducted during March–October 2007. These bibliographic databases included searches of MEDLINE (National Library of Medicine, Bethesda, MD), CAB Abstracts (applied life sciences search site), Biological Abstracts and Scopus. For each of the multiple searches, all articles that included the words TGF-β, cytokines, growth factors, human milk, immunological outcomes, infants, asthma, allergy, atopy or infections in the title or in the key words were selected. The lists of papers from the various searches were merged, and the resulting articles were then examined for inclusion or exclusion. The bibliographic citations of each of the articles ultimately selected for inclusion in the analyses were also examined to identify any other acceptable studies that were not captured by the bibliographic data searches.

Study selection

Studies were considered acceptable for inclusion if they met the following criteria: (i) the study was a clinical study of a mother–infant dyad, (ii) the study had an epidemiological design, i.e. pregnancy cohort study, birth cohort study, human prospective study or randomized controlled trial during pregnancy or lactation, (iii) the study provided information on sample size, analytical methods of measuring TGF-β in human milk, well-defined clinical and biochemical immune outcomes/end-points, complete description of follow-up and appropriate statistical analysis, (iv) the study population included ‘clinically healthy’ mothers and infants and provided information on milk collection and sampling and (v) the study provided sufficient information to assess its applicability and relevance in answering this review’s objectives.

When studies identified during the bibliographic search were excluded from the review, a record was maintained of these studies and the reason for exclusion was categorized. Reasons for exclusion were due to the objectives of this study not pertaining to this review, i.e. only determinations of TGF-β concentrations in human milk or the TGF-β functions was not related to human milk, i.e. cancer, in vitro studies of TGF-β or determinations of TGF-β in adult serum. Although many studies were identified and excluded, they were not all noted because their list was extensive (>11,600 hits).

Data abstraction

The suitability of each study for inclusion in the systematic review was determined on a minimum of three phases, and followed an intensive examination of the reported data. The data from each study were extracted and tabulated and included author and year of publication, descriptive information concerning the study design, country and setting, baseline characteristics of the study sample, exposure measure such as milk sample collection, source and how TGF-β was determined (bioassay, ELISA), isoform and type of TGF-β (free, latent, active), age outcome measured in the child and the outcome measured (clinical and biochemical), statistical test and the effect measured, comment and an overall effect (+, positive; −, negative; NS, not significant). If multivariable results were reported these were also extracted. Four scientists were contacted for additional information not available within the published documents during June 2007 and these additional results were included where appropriate.

Outcome variables tested

The primary immunological outcome variables included but not limited to immunoglobulin titres such as IgA, IgG or IgE measured in infants, were those related to clinical symptoms of allergy like eczema, wheezing and asthma measured at differing ages up to 5 yr.

TGF-β concentration in milk

An overall estimate of TGF-β concentration in non-allergic mothers was calculated by applying a formula (36) that estimated the mean using the values of the median (m), low and high end of the range (a and b, respectively) and n (the sample size) from studies that did not report means. The means and mean estimates of concentration of TGF-β were multiplied by the number of mothers in each study group to obtain an overall means (from all studies combined) for TGF-β1 and TGF-β2 separately in colostrum and mature milk (Table 2).

Table 2.   Human milk TGF-β concentrations (pg/ml unless otherwise indicated) in studies collecting milk samples at specific times during lactation and whether mothers were not allergic or allergic*
Author and yearNot allergic or not specifiedAllergy specified
ColostrumMature milk (weeks 4–6)ColostrumMature milk
TGF-β1TGF-β2TGF-β1TGF-β2TGF-β1TGF-β2TGF-β1TGF-β2
  1. Not available, not enough studies provided data to enable calculation of an estimate.

  2. *The concentrations are given for latent TGF-β1 and TGF-β2 concentration in full term infants – data for premature infants not given.

  3. †Saito study excluded from calculating the overall mean concentration of TGF-β because this study used a different method (bioassay) to measure TGF-β. The overall mean was estimated using the values of the median (m), low and high end of the range (a and b, respectively) and n (the sample size) from studies that did not report means. The mean concentration of TGF-β was multiplied by the number of mothers in each study group to obtain an overall means (from all studies combined) for TGF-β1 and TGF-β2 separately in colostrum and mature milk.

Saito, 1993 (39) 1365.7 ± 242.9 ng/ml (mean ± s.d., n = 21) 952.5 ± 212.6 ng/ml at 1 month (n = 21)     
Srivastava, 19963280 (mean, n = 2)19200 (n = 2)328 ± 212 (mean ± s.d., n = 9)5310 ± 6151 (n = 23)    
Kalliomaki, 1999 (8)140, 67–186 (median and IQR, n = 43)3325, 1376–5394 (n = 43)83, 17–114 (n = 38)1644, 592–2697 (n = 38)    
Hawkes, 1999 (18)391 ± 54, 43–1950 (mean ± s.e.m., range n = 49)3048 ± 339, 208–10257 (n = 49)471 ± 168, 91–7108 at week 6 (n = 42)2723 ± 804, 431–34033 (n = 42)    
Saarinen, 1999 (34)807, 684–953 (geometric mean, 95% CI, n = 255)   683, 485–963 (n = 60)   
Bottcher, 2000 (41)325, 125–1165 (median and range, n = 49)11,31,357–16,784 (n = 49)261, 125–616 (n = 48)787, 250–11696 (n = 45)    
Laiho, 2003 (45)   539, 255–21,094 at 2–3 months (median and range, n = 37)   420, 153–42,117 (n = 37)
Oddy, 2003 (43)787, 731–844 at 11 days (geometric mean, 95% CI, n = 142)       
Rautava, 2002 (40)  226, I 118–335 at 3 months receiving probiotics (mean, 95% CI, n = 30) 178, 122–233 (n = 32) receiving placebo2885, 1624–4146 (n = 30) 1340, 978–170 (n = 32)   5085, 1818–8352 (n = 10) 1136, 532–1740 (n = 9)
Ogawa, 2004 (21)515 ± 59, 73–1768 (mean ± s.e.m. and range, n = 55)565 ± 76, 39–2240 (n = 55)      
Savilahti, 2005 (22)447, 376–531 (geometric mean, 95% CI, n = 118)3162, 2529–3954 (n = 94)  417, 347–500 with at least one symptom of atopy (n = 109) 537, 363–525 with at least 2 or more symptoms (n = 84)3090, 2574–3710 (n = 89) 3236, 2660–3936 (n = 73)  
Rigotti, 2006 (46)1200, 0–4700 (median, range, n = 19) 1059, 0–6250 at 1 month (n = 19) 330, 0–3400 (n = 20) 215, 0–2400 (n = 20) 
Snijders, 2006 (42)  221.1 ± 140.1 (mean ± s.d., at 1 month, n = 125)   216.9 ± 155.1 (n = 182) 
Overall mean (pg/ml)†710.72956.9364.63089.0556.7Not available255.5Not available

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

Description of studies

Of 82 hits identified with the Medline search and 45 additional hits identified with the other searches combined, 127 relevant papers were screened. Of these, 15 that met the selection criteria were reviewed and 12 were included in the systematic review. All the studies were identified with cross-referencing. Two studies were excluded as they were not directly relevant (37, 38) and one because TGF-β in human milk was examined in isolation from any immunological outcome (18). Tables 1 and 2 summarize the data from these human studies. Although the study by Saito et al. (39) was the first to demonstrate the presence, isotype and type (latent vs. active) of TGF-βs, it was not included in the review because the bioactivity assay was used to measure TGF-β concentration in human milk and not its immune-reactivity. Of the 12 human studies assessing the association between TGF-β in human milk and immunological outcomes (Table 1), 10 were human prospective studies, one was a nested case–control study (34) and one was a double-blind placebo controlled trial (40). Five of the 12 studies were carried out in Finland, two in Sweden and one each in the USA, Japan, the Netherlands, Italy and the Czech Republic. All 12 studies reported to have included only ‘clinically healthy’ ambulatory populations of mothers and infants who had a risk of developing allergic disease based on history of atopy. Together a total of 739 atopic mothers and 906 healthy mothers were included from the 12 selected studies.

Table 1.   Human studies – characteristics of the studies that examined the association of TGF-β in human milk and immune outcomes in children or their lactating mothers and met the inclusion criteria for the systematic review
Author and yearStudy designCountry & settingSampleExposure measured:Outcomes measured and AgeEffect measureComment Overall effect*
  1. NS, not significant effect; +, positive effect.

  2. *Overall effect of TGF-β (TGF-β1 or TGF-β2) as a binary indicator.

Kalliomaki, 1999 (8)Human prospective of infants with high risk of atopic diseaseFinland47 families with newborn infantsColostrum  TGF-β1 & TGF-β2IgA-secreting cells, 3 month Atopic dermatitisColostrum TGF-β2 was significantly higher in infants who had specific IgA-secreting cells [1840 (627–2779) median (IQR)] concentrations at age, 3 months, in response to at least one of the dietary antigens tested (α-lactoglobulin, casein, gliadin or ovalbumin) compared with those who did not have such cells [1215, (307–2087)] (p = 0.048).Higher concentrations of TGF-β1 & β2 in the colostrum of mothers of infants with post-weaning onset of atopic disease compared with those with no and pre-weaning onset of disease+
Saarinen, 1999 (34)Nested case–control in a prospective studyFinland, population study of 6209 infants118 cases 207 controls with milk samplesColostrum  TGF-β1IgA cows milk allergy in child at 6 monthsCorrelation coefficient of TGF-β transformed values and immuno-biochemical outcomes R = 0.204 p = 0.04β-coefficients are provided in Fig. 1 (data generously provided by authors)+
Böttcher, 2000 (41)Human prospectiveSwedenAtopic history (n = 24) and non-atopic mothers with newborn infants (n = 25)Atopic status of mothers prior during lactationIFN-γ, IL-4, IL-5, IL-6, IL-10 and IL-13 concentration in milk from allergic and non-allergic mothers at 1, 2 & 3 monthsColostrum and mature milk TGF-β1 and -β2 correlated significantly (correlation coefficient, ρ = 0.21–0.88, p < 0.05) with IL-6 and moderately with IL-10 concentrationsAtopic disease based on history of symptoms provided by mothers, 79% confirmed. Human milk TGF-β1 and -β2 an effect on stimulation of IgA indirectly through their association with IL-10 or IL-6 in human milk+
Rautava, 2002 (40)Double-blind placebo controlled trialFinland159 atopic mothers randomized to probiotic (n = 30) or control n = 32)Probiotics enhancing the production of TGF-βMature milk collected at 3 m TGF-β1 & TGF-β2 and association with atopic eczema in child 2 yrOdds ratio of atopic eczema based on probiotic suppl. 95% CI: 0.32, 95% CI 0.12–0.85 p = 0.0098Probiotics significantly impacted on (TGF)-β2 levels (p = 0.018) milk with more TGF-β showed less atopic eczema+
Bottcher, 2003 (41)Human prospectiveSweden53 mother/infant pairs (24 with allergy; 29 no allergy)Mothers atopic status during pregnancyColostrum & mature milk at 1 month TGF-β1 & β2 and other cytokines compared between infants with atopic disease and skin positive test vs. no atopyMedian differences between infants with vs. without allergic disease at 2 yr, and Spearman’s correlation of TGF-β concentration and immuno-biochemical outcomes p = 0.6 at 3 monthsSmall sample size with limited statistical power and small number of infants with positive skin tests. No differences on other milks cytokines (IL-4, IFN-γ, IL-6, IL-10, IL-13 and IL-5) between infants with vs. without allergic disease. No control for maternal history of allergy.NS
Laiho, 2003 (45)Human prospectiveFinlandAtopic (n = 43) and non-atopic (n = 51) mothers with newborn infants (n = 49)Maternal atopic dermatitis (AD)Colostrum TGF-β2Median difference in TGF-β conc by allergic disease (AD), (n = 37; range: 153–42,117) vs. no AD (n = 38; range: 255–21,094) p-value = 0.031Allergic disease based on doctor’s diagnosis. No association between milk polyunsaturated fatty acids and TGF-β concentrations. No differences infants anthropometric measures between allergic and non-allergic mothers+
Oddy, 2003 (43)Human prospectiveUSA Infant Immune Study243 mothers (21% with asthma)Colostrum TGF-β1 concentrationsWheeze ever in child (40% with wheeze) at 1 yrOdds ratio wheezing and dose of TGF-β, 0.22 (0.05–0.89) p = 0.034Dose = breastfeeding duration by TGF-β concentration+
Ogawa, 2004 (21)Human prospectiveJapan55 healthy mothersColostrum TGF-β1 & TGF-β2IgA serum of infants at birth and 1 monthLinear regression analysis (95% CI: not given) of TGF-β1 & TGF-β2 and serum IgA of infants at 1 month Correlation coefficient, TGF-β1: R = 0.38, p = 0.005, TGF-β2: R = 0.45, p = 0.0005Random selection of non-allergic, lactating mothers. TGF-β may be starter for IgA production in newborns+
Savilahti, 2005 (22)Human prospectiveFinland from a cohort of 46764 groups selected depending on atopic family history and breastfeeding durationColostrum TGF-β1 & TGF-β2Atopy in child 4 yrMean differences of TGF-β conc. 3162 (2529–3954) vs. 3090 (2574–3710) p = 0.10No atopic symptoms compared to at least one atopic symptomNS
Snijders, 2006 (42)Human prospectivethe Netherlands315 lactating mothers provided milk samples and (TGF)-β1 was sampled in 307. 182 (61%) with allergy and 125 (39%) no allergyMature milk TGF-β1 concentrationsEczema, wheeze sensitization at 2 yrOdds ratio 1.07 (0.56–2.00) for eczema 1.17 (0.59–2.58) for wheezeAdjusted odds ratios reanalysed as med/low compared to high combined. Dose response not considered: effect of low (2.0–166.9 pg/ml), middle (166.9–248.4 pg/ml), high (248.5–1536.8 pg/ml) TGF-β (concentration only) was tested on outcomesNS
Rigotti, 2006 (46)Human prospectiveItaly22 mothers and newborns (9 with allergy; 13 with no allergy)Atopic status of mothers prior to deliveryTGF-β1 in Colostrum and mature at 1 month Atopic status of infants at 6 monthsMedian differences in mature milk TGF-β1, 215 (n = 13; range 0–2400 pg/ml) allergic vs. non-allergic 1059 (n = 9; range 0–6250 pg/ml) p = 0.015. All, six infants with atopic dermatitis at 6 months were in moms with allergic disease statusAtopic status verified in mothers by skin prick tests, infants assessed for symptoms of allergic disease. Non-significant difference in IL-10.+
Prokešová, 2006 (44)Human prospectiveCzech Republic21 healthy mothers, 21 allergic mothersAtopic status of mothers prior to deliveryColostrum Mature milk TGF-β isotype not specified and other cytokinesMedian differences between healthy vs. allergic 471 (n = 21; 25th–75th %ile 219–821) vs. non-allergic moms 300 (n = 18; 70–734) p = 0.3749Allergic mothers diagnosed by specialist. TGF-β isoform not specified. No difference in other milk cytokines, IL-4, IL-5, IL-6, IL-10, IL-13, IFN-γ between groupsNS
Overall effect      8/12 = + 4/12 = NS*Overall 67% +

Overall 67% of studies in the review (8/12) reported an association of immunological outcomes in children with TGF-β1 or TGF-β2 concentration or dose in colostrum or mature milk, advantageous to immune development and expression (Table 1).

Examples of the type of immune responses examined were reported by Saarinen et al. (34) in infants with CMA. These investigators demonstrated that high TGF-β1 concentration in human milk colostrum was proportionally associated with immunological responses that reduced the likelihood of allergy in infants with CMA. The β-coefficients and 95% confidence limits of the association between TGF-β1 concentration and immunological outcomes are shown in Fig. 1. High concentration of TGF-β1 was associated with high IgG titres to α-casein (B = 0.27; 95% CI: 0.03–0.51; p = 0.014) and whole cows milk formula (B = 0.28; 95% CI: 0.03–0.52; p = 0.028). Although the association with IgA or IgE titres were not significant, these show a trend of high IgA titres and conversely low IgE titres with increasing TGF-β1. Consequently, a negative correlation between TGF-β1 concentration and skin prick test to cows milk (B = −0.32; 95% CI: −0.62 to −0.02; p = 0.035) was observed. In this study, other human milk cytokines [interleukin (IL)-6 and interferon-γ] were assessed, but only TGF-β1 was shown to be associated with immunological outcomes indicative of reduced atopy. Similarly, Ogawa et al. (21) demonstrated a significant correlation among colostral TGF-β1 and TGF–β2 and serum IgA in 1-month infants, but not between colostral cytokines, IL-6 or IL-10. Other investigators have compared TGF-β concentrations from mothers whose infants had specific IgA-secreting cells at 3 months of age to those who did not and found significantly higher TGF-β concentration in those with the specific IgA-secreting cells (8). Although the studies assessed TGF-β1 or TGF-β2, a difference between TGF-β1 and TGF-β2 in inducing an immune response was not observed overall (Table 1).

image

Figure 1.  TGF-β1 and immunological outcomes in linear regression analysis from one study (Saarinen 34).

Download figure to PowerPoint

Among the clinical outcomes measured in our review, three studies assessed the association of human milk TGF-β1 or TGF-β2 concentration and risk of developing atopic eczema (40–42) and two the risk of wheezing (42, 43). These symptoms were categorized and clustered under allergic responses. All studies in the review other than the study by Savilahti et al. (22) assessed allergic responses in infants or young children (≤2 yrs, and only the studies by Oddy et al. (43), Rautava et al. (40) and Snijders et al. (42) provided sufficient information to calculate an odds ratio (OR: 0.58; 95% CI: 0.36–0.94).

Table 1 demonstrates that the studies in the review differed in the effect measure used in assessing the association between human milk TGF-β concentrations and immunological outcomes. Six studies used Mann–Whitney U-test to compare milk TGF-β concentrations across atopic or non-atopic (healthy) groups. Three studies gave odds ratios (40, 42, 43), one reported linear regression analyses (21) and the remaining reported correlation coefficients (34). Two studies had information on relevant confounders other than maternal atopic status (42, 43). Two studies reported a dose–response effect of milk (breastfeeding duration multiplied by TGF-β concentration) (40, 43), whereas 10 studies reported concentration of TGF-β in milk only. These differences reflected the complex study designs. In some cases, the main comparison was made between the mother’s history of allergy and the concentrations of cytokines in milk and their association with IgA production in infants. In others, the mother’s history of allergy was used to select cases and controls; however, the final comparison was made among infants based on their development of atopy (41).

TGF-β concentration in milk

Colostrum and mature milk TGF-β1 and TGF-β2 concentrations from mothers with a history of specified allergy or with no history of specified allergy are reported in Table 2. Of the 12 studies included in the review 10 measured TGF-β concentrations in milk using standardized procedures for milk collection and sample storage; however, only two studies provided information on the duration of lactation enabling calculation of a dose (40, 43). Most studies used a commercially available sandwich ELISA kit (R & D Systems, Mineapolis, MN, USA) to measure TGF-β concentrations, and one study used a bioassay (39). One study reported TGF-β concentration but did not differentiate isoform type although it reported mature human milk TGF-β concentrations beyond 1 month of lactation (44). Three other studies were added to Table 2 because they reported cytokine concentrations in milk only in non-allergic mothers (2, 18, 39). The two major TGF-β isoforms reported in human milk were β1 and β2 (Table 2). Three studies (2, 7, 39) found that TGF-β1 and TGF-β2 were present mainly in a latent form in colostrum or in mature milk, i.e. non-activate TGF-β. Moreover, the distributions of TGF-β1 and TGF-β2 concentrations in colostrum or mature milk show high variability within studies as indicated by the spread of the distributions (e.g. high coefficient of variation) and between studies (e.g. high difference between measures of central tendency). In most studies, human milk TGF-β concentrations were skewed to positive values and thus, studies reported mainly medians or geometric means.

The concentrations of TGF-β1 and TGF-β2 in colostral samples from mothers classified with or without asthma or atopic symptoms did not differ significantly; however, none of the concentrations of IgA, IgA specific to cow milk or casein antibody differ as well (22). On the other hand, Laiho et al. (45) found that TGF-β2 concentrations in mature milk were significantly lower among mothers with allergic disease (i.e. mother’s report of doctor-diagnosed atopic eczema or asthma) compared to mothers with no allergic disease. Also, Rigotti et al. (46) demonstrated significantly lower TGF-β1 concentrations in mature milk of mothers with verified atopic status (i.e. skin prick test) compared to milk of mothers with no atopy. A similar but non-significant trend was observed in colostrum. The studies represented in Table 2 show that maternal colostrum or mature milk was lower in TGF-β1 and TGF-β2 in mothers who were allergic compared to mothers who were not allergic in some studies (34, 45, 46). However, the largest study with more than 300 mothers (22) did not show any difference between non-allergic mothers or mother with allergic symptoms.

Moreover, it is apparent among non-allergic mothers that TGF-β1 decreases from colostrum to mature milk whereas TGF-β2 remains constant or increases slightly from colostrum to mature milk (overall mean, Table 2).

In the present study, two factors were identified that may be associated with high TGF-β concentrations as follows: no history of allergy and consumption of probiotics. The higher TGF-β concentrations may be associated with a reduction of the likelihood of allergy (Tables 1 and 2). For example, in the study by Oddy et al. (43) the estimated concentration of TGF-β1 that was protective against wheeze was >895 pg/ml and as high as 2040 pg/ml. The estimated concentration of TGF-β2 in the Rautava et al. (40) study that was protective against atopic eczema was 2723 ± 804 pg/ml.

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

This systematic review of the associations between TGF-β in human milk and immunological outcomes in infants and children provides an overall measure of effect based on the literature currently available. Our primary objective was to examine the published studies that considered TGF-β in association with biochemical, immunological and clinical outcomes related to allergy in infancy and early childhood. Overall 67% of the human studies in the review showed a positive association between high TGF-β1 or TGF-β2 concentrations and a reduction in immunological outcomes of allergy.

There was a high level of heterogeneity among the published studies, which made it difficult to pool effects of published data in a meta-analysis. Factors which could explain this heterogeneity included differences in study populations, design and methodology as well as differences in milk source, TGF-ß isoform type, age at which outcome was measured and lack of uniformity in the immunological outcome measured (Table 1).

Two studies considered clinical immunologic outcomes related to allergy and both gave an absolute measure of TGF-β (high dose vs. low dose). The study by Oddy et al. (43) was a prospective pregnancy cohort study that measured TGF-β1 from milk at 11 days and dose was calculated according to duration of breastfeeding. In this study, an estimated dose was calculated by multiplying the anti-logged values of age-adjusted TGF-β concentration by the duration of breastfeeding in weeks. This study provides information on the concentration of TGF-β1, duration of breastfeeding and clinical outcomes of allergy including wheezing showing a significant reduction in wheeze in children with an increasing dose of TGF-β. The study by Rautava et al. (40) was a double-blind placebo controlled trial of probiotics given to mothers at risk of atopy, antenatally and during lactation for at least 3 months. This study (40) provided TGF-β1 and TGF-β2 concentrations measured at 3 months and confirmation of breastfeeding for at least 3 months of lactation, which allowed the calculation of an exposure dose of (TGF)-β. Because mothers in the intervention group had significantly higher TGF-β2 levels than the placebo group (2885 pg/ml 95% CI: 1624–4146 vs. 1340 pg/ml 95% CI: 978–1702; p = 0.018) made the effect of both the trial and the placebo groups appropriate for comparison for TGF-β2 (but not TGF-β1) because a high dose compared to a lower dose of TGF-β was possible. The atopic outcome was measured in the infants at 2 yr which was an appropriate outcome age for this study as it agreed with age measured in other studies. Rautava’s study was an intervention study using a probiotic strain in the intervention arm and finding a 50% reduction of early infantile eczema (47). Although the mechanism mediating the increase in TGF-β was not assessed in this study, studies have demonstrated that probiotic induction of oral tolerance may be mediated through activation of TGF-β-dependent regulatory T-cell (Th3-cells), which also secrete TGF-β (48).

Among the studies that found non-significant associations of TGF-β on immunological outcomes in children, an important finding was the observed low concentrations of TGF-β in human milk among non-allergic mothers, suggesting that low concentrations may impact on observed atopic associations. Prokešová et al. (44) reported median colostral total TGF-β1 (assumed, isoform type not reported) of 471 pg/ml in healthy mothers, Böttcher et al. (41), median TGF-β1 c. 300 pg/ml, or Snijders et al. (42) mean TGF-β1 of 221 pg/ml whereas the TGF-β1 concentrations reported in studies showing a significant association were higher. Saarinen et al. (34) reported colostral TGF-β1 mean concentration of 807 pg/ml, Oddy et al. (43) reported geometric mean of 787 pg/ml and Ogawa et al. (21) reported a mean of 515 pg/ml. The higher concentrations in these studies as well as the immunological response assessed may have influenced the statistical power of the observed results (Table 1). The results from Saarinen et al. (34) indicate that TGF-β may participate in class switching of immunoglobulins such as increasing IgA and inhibiting IgE.

Mechanism

There is sufficient evidence from animal studies to show that milk TGF-β survives digestion and has functional effects in neonates. Letterio (12, 13, 49) showed that when TGF-ß was fed to TGF-ß null mice, the cytokine was found distributed intact to lung tissue. TGF-β appears to be rapidly taken up by the neonatal intestine suggesting it may influence immunity beyond the gut (23). Letterio’s study was ground breaking as it substantiated the important postnatal immunoregulatory effect of milk TGF-β. TGF-β null newborn mice were able to survive and develop normally only if TGF-β1 was present in maternal milk (13) indicating that immunosuppressive properties of TGF-β both conserved after ingestion are vital for immune regulation. Further, Penttila et al. (31) has demonstrated that oral supplementation with TGF-β induces oral tolerance in allergy prone rats and Ando (33) has demonstrated that oral supplementation with (TGF-β) encourages ova-induced tolerance in mice. Several other studies have examined and demonstrated the pathways by which TGF-β in milk induces oral tolerance (50–52), stimulation of IgA isotype switching in B cells (53) and maintenance of intestinal epithelium barrier (54) all necessary for immunological health. Recently, Verhasselt et al. (35) assessed the role of milk-borne TGF-β and found that air-borne antigens were efficiently transferred from the mother to the neonate through milk and that tolerance induction did not require the transfer of immunoglobulins, but rather breastfeeding-induced tolerance relied solely on the presence of milk TGF-β.

The evidence from the present systematic review supports the result from animal studies at various levels. Among the included studies, human milk TGF-β concentration was proportionally associated with the level of IgA β-lactoglobulin antibodies and other immunological responses of oral tolerance (Table 1). In studies where other milk cytokines were measured (41, 43), the association with immunological outcomes of oral tolerance or of reduced allergy was found to be specific to the concentrations of milk TGF-β, whereas the other milk cytokines were not. Finally, the concentrations of TGF-β1 and TGF-β2 in colostrum or mature milk from mothers of infants with allergy were lower than in milk from mothers of infants with no allergy (Table 2). These findings suggest that human milk TGF-β may play an important role in determining the intensity and type of immune response to specific allergic antigens.

Some important findings were revealed by the measure of milk exposure. TGF-β2 isoform is the predominant cytokine in human milk, it tends to increase throughout lactation, and most of it is present in a latent form; however, there is considerable variability in the concentrations of TGF-β between and within studies (Table 2). This variability has important implications for the interpretation of the concentration of TGF-β associated with immune benefits in human infants.

Although there is no apparent explanation for the variation in TGF-β, the variation has been demonstrated by Hawkes et al. (55) and may not be related to plasma levels of TGF-β. Dietary factors like fatty acid composition of the maternal diet may influence breast milk TGF-β2 concentration. Laiho et al. (45) showed a positive association between TGF-β2 and the proportion of polyunsaturated fatty acids and a negative association between TGF-β2 and the proportion of saturated fatty acids. On the other hand, Hawkes et al. (56) showed no correlation between TGF-β and fatty acid content.

It is also possible that local infection within the mammary gland may influence the levels of TGF-β in maternal milk. TGF-β1 and TGF-β2 are cytokines that regulate mammary gland development and have a role in mediating inflammation; thus, infection elicits host production of a range of TGF-β isoforms (57). Familial or genetic factors may control the endogenous production by epithelial cells in the mammary gland and may be a determining factor for levels of both TGF-β1 and TGF-β2 in breast milk. Administration of probiotics may increase human milk TGF-β concentration but this may depend on the probiotic strain as a recent study showed inverse effects with Lactobacillus reuteri (53). Therefore, concentrations of human milk TGF-β may be critical in determining immune function. Laiho et al. (45) showed a lower concentration of TGF-β in breast milk in mothers with atopic dermatitis compared to those without. Furthermore, their results imply that regulatory circuits of fatty acid, eicosanoid and cytokine metabolism may be connected in joint action. More studies are needed to uncover the factors that determine TGF-β concentrations in breast milk and to establish whether these factors are amenable to therapeutic or preventative exploitation (8).

Although the mammary gland has substantial TGF-β expression, and high levels of TGF-β are found in mouse mammary epithelial tissues with little expression in fat cells and connective tissues, the form of TGF-β found in human, bovine or rat milk is associated with its ‘Latency-Associated-Peptide’ or LAP protein, the form commonly known as latent TGF-β. This type of TGF-β is not active and thus, has not been activated or utilized by the mammary tissue per se. This type of TGF-β is mainly secretory (58) and requires activation by hydrolytic proteolysis or enzymatic action which is highest at a low pH (59). The human studies identified in our systematic review show that 90% of all human milk TGF-β is in a latent form, and will subsequently be activated in the gastrointestinal tract of the infant. Once activated, the TGF-β, lacking the LAP protein, will be absorbed to a receptor in the intestinal epithelium. Studies have demonstrated that the availability of these receptors in the intestinal epithelium show an ontogenic development that parallels the presence of milk TGF β (31, 32, 60).

Strengths and limitations of the systematic review

There were certain limitations within the studies included in the review as follows. The time when milk samples were collected varied among the studies especially when mature milk TGF-β collections were reported (Table 2). Most studies collected a sample of colostrum as well as one sample of mature milk but many studies collected only one sample at one time-point. Few studies collected data on breastfeeding duration; therefore an exposure dose could not be calculated. Moreover, few studies measured clinical outcomes and compared them using measures of risk.

Several strengths were associated with this review. Prospective assessment clarified the temporal relations between exposure to the cytokine in milk and later wheeze and atopic eczema. Moreover, because randomized controlled trials that assign participants to either breastfeeding or formula feeding were not feasible, observational data formed a crucial part of the evidence for the long-term health effects of different infant feeding approaches. The findings of the 12 published observational studies that assessed TGF-β in milk with atopic risk were broadly consistent despite the widely differing nature of the populations studied (i.e. relatively different methods in diagnosing allergy), from different ethnic and geographic backgrounds, and studies with widely different outcomes (i.e. Table 1, Effect Measure). In this regard, the review provides a wider scope of the effects of human milk TGF-β and of its ontogeny compared with results from individual studies. Individual studies have noted that human milk TGF-β2 does not change from colostrum to mature milk or decreases, and that these changes reflect an ontogenic response in which exogenous milk TGF-β1 and TGF-β2 declines with lactation as the infant’s own endogenous production of TGF-β increases. Administration of probiotics may enhance TGF-β2 production in pregnancy or in early postnatal life during lactation. When all published studies are assessed together and their population characteristics and methodologies are evaluated (Tables 1 and 2), the trend is for TGF-β1 to decline whilst TGF-β2 concentrations rise or remain constant. These changes may reflect the high variability observed among studies rather than possible ontogenic changes.

Conclusion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

Transforming growth factor-beta from human milk is an important family of growth factors involved in maintaining homeostasis in the intestine, regulating inflammation and allergy development, as well as promoting oral tolerance development. The dose of human milk TGF-β1 and TGF-β2 may potentially modulate or regulate the immunological response of the infant in early postnatal life.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

Mr Ron Stoner, Principal Information Scientist at MJ Nutritionals, was consulted on the selection of search engines.

Disclosure

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References

The research was funded by MJ Nutritionals through Bristol Myers Squibb. The authors responsibilities were as follows: WHO wrote the paper, had full access to the manuscripts discussed in this review and takes full responsibility for the integrity and accuracy of the results presented here. FJR had an essential role in the ideation and planning of the study design and in reviewing drafts of the manuscript as well as a significant contribution to writing of the manuscript. Both authors edited the final manuscript. Neither of the authors report a conflicts of interest.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgment
  8. Disclosure
  9. References