The composition of gut microbiota plays an important role in the health of infants. Breast-fed infants have a gut microbiota that is dominated by bifidobacteria, whereas formula-fed infants have a more heterogeneous composition, with comparatively lower levels of bifidobacteria (1,2). The prevalence of bifidobacteria in the gastrointestinal (GI) tract of breast-fed infants has been associated with reduced infection rates compared with formula-fed infants (1,3). Additionally, breast-fed infants consistently have fewer Clostridia species compared with formula-fed infants (4,5). Well-controlled clinical studies have also shown an association between the occurrence of allergies and gut microbiota composition (6). For example, differences in the bifidobacteria composition of infants' microbiota have been associated with an increased predisposition of formula-fed infants to develop allergies compared with breast-fed infants. Children with allergies were found to harbor different gut microbiota compared with healthy infants, the number of bifidobacteria being lower and the number of Clostridia being higher in children with allergies (6,7). This observation was confirmed in studies of children with atopic dermatitis, which showed that these infants had fewer bifidobacteria counts compared with healthy infants (8), but not in studies on respiratory allergies (9). Priming of the gut mucosal immune system seems of critical importance. Mice raised in a sterile environment fail to develop oral tolerance to innocuous feeding antigens (10). The overreacting TH2-dependent antibody response can be corrected by the reconstitution of gut microbiota, but only if this occurs within the neonatal period (10). This effect may be mediated via the transforming growth factor-β producing CD25(+) CD4 (+) T intestinal cells (11).
Several approaches have been used to promote a higher number of bifidobacterial counts in the infant gut, including supplementing infant formulas with oligosaccharides that preferentially stimulate bifidobacterial growth, or directly adding Bifidobacterium species to formulas. An alternative approach has been to modify concentrations of existing components so that infant formulas closer resemble breast milk, which is characterized by the predominance of whey protein, low phosphorus content, high lactose content, and low buffering capacity. This contrasts with the standard cow's-milk–based infant formulas, which have relatively high protein content, primarily composed of casein protein and high phosphate content (12). These distinctions are believed to contribute to the differences in bifidogenic effect (an increase in the total number of bifidobacteria). Studies have suggested an effect of whey-predominant formulas on the development of infant gut microbiota in a manner that is similar to that of breast-fed infants (13,14).
In the present study we aimed to determine whether a whey-predominant (70% whey, 30% casein) infant formula, with low protein and phosphate concentrations, stimulates bifidobacteria colonization of the newborn infant gut and compared it with the effects of a similar bifidobacteria-supplemented formula. We also compared the bifidobacterial counts of infants fed these formulas with those of infants fed a reference formula (high in protein and casein-predominant) and breast-fed infants. We then compared the overall microbiota profiles. Finally, we assessed the effect of the different feeding regimens on mucosal immunity.
PATIENTS AND METHODS
The trial population consisted of healthy full-term infants whose mothers had chosen to feed their infants exclusively with either breast milk or with formula starting from day 7 of birth until they were 4 months old. To be included in the study, infants had to be ≤7 days old at enrollment, weigh between 2500 and 4500 g, and be singletons. Infants were excluded from the study if they had congenital illnesses or malformations that could affect normal growth; significant prenatal and/or postnatal disease; been rehospitalized for more than 2 days in the first 14 days of life (with the exception of infants that were rehospitalized because of jaundice); received antibiotics prenatally, at the time of enrollment, or during the 5 days preceding enrollment; received probiotic- or prebiotic-containing formulas at the time of enrollment; and caregivers who were not expected to comply with the study requirements. Infants participating in a different study were excluded.
This study was conducted in accordance with the principles and rules of the Declaration of Helsinki and adhered to the Good Clinical Practice guidelines of the International Conference on Harmonization. It was approved on May 19, 2006 by the Lorraine ethics committee (Comité consultatif de protection des personnes se prêtant à des recherches biomédicales, number 06.05.02). Written informed consents were signed by the infants' legal representatives before infants received any of the study formulas.
This was a single-center, randomized, double-blind, controlled trial in which infants were fed 1 of 3 infant formulas. A group of exclusively breast-fed infants served as a reference. The trial was set in the Maternité Régionale Universitaire of Nancy, France.
Healthy newborn infants were randomly assigned at enrollment to receive the study formula, the study formula containing Bifidobactrium (B) longum BL999, or a control formula, and were exclusively fed the assigned formula ad libitum from the age of 4 days to 4 months. Visits to the hospital occurred when infants were 2 weeks (±3 days), 1 month (±3 days), 2 months (±3 days), and 4 months (±3 days). For 3 days before each visit, caregivers kept records of feeding and digestive tolerance. Compliance was evaluated based on the amount of formula consumption, the number of cans of formula dispensed, the number of cans remaining at each visit, and compliance with the dietary restrictions, as recorded in the 3-day diary before each visit.
At the time of enrollment, baseline characteristics and any medical history were recorded. Thereafter, anthropometric measurements (weight, length, and head circumference), evaluations of current health status and medical history since previous visit, digestive tolerance, any adverse events (AEs), and any concomitant medications were recorded at each visit. At the 1- and 2-month visits, stool samples were collected for bacteriological and IgA analyses.
Formula and Diet
All 3 formulas in the study contained proteins, carbohydrates, fats, minerals, and vitamins in quantities that provided full nutritional support for infants from birth up to 4 months of age. The study formula had low protein and phosphate content, high lactose and contained predominantly whey protein compared with the control formula (Table 1). Among others, the threonine (Thr), methionine (Met), and phenylalanine (Phe) concentrations in the study formula reflected the amino acid profile in human milk better than did the control formula 100 mg Thr, 45 mg Met, 80 mg Phe per 100 kcal in the low protein formula versus 120 mg Thr, 70 mg Met, 120 mg Phe per 100 kcal in the control formula versus 77 mg Thr, 23 mg Met, 83 mg Phe per 100 kcal in human milk (source: EC Directive 2006 on infant formulae and follow-on formulae). The third formula was the same as the study formula but was supplemented with 2 × 107 colony-forming units (CFU)/g of B longum, strain BL999 (ATCC: BAA-999 designation BB536, Morinaga, Japan; NCC3001). Formulas were manufactured and blinded by the sponsor (Nestlé, Konolfingen, Switzerland), which was the only party with knowledge of the product codes.
No complementary foods, including preparations containing probiotics, were allowed during the study. Any deviations were recorded in case report forms.
The primary outcome was bifidobacterial counts at the 2-month visit. Secondary outcomes were non-Bifidobacteria bacterial counts, weight, length, head circumference, digestive tolerance, stool IgA, and the occurrence of AEs.
All of the infants were weighed on the same calibrated electronic scale and weights recorded to the nearest 10 g. Recumbent length and head circumference were measured to the nearest 1 mm using a standardized length board and a standard nonelastic measuring tape, respectively. Digestive tolerance was assessed based on stool frequency (occurrence/day) and characteristics, spitting up and vomiting, and infant's behavior recorded by the caregiver during the 3 days before each visit. Stool characteristics consisted of color (brown, green, yellow, other), consistency (hard/lumps, formed/normal, soft/creamy, or liquid/watery), odor (normal or unusual), and the occurrence of flatulence (less than usual, the same as usual, or more than the usual). Spitting up was assessed based on the volume of spit (small <5 mL, moderate 5–25 mL, or large >25 mL). The frequencies of vomiting, crying, fussing, and colic were quantified in counts per day.
AEs were defined as illnesses or signs and symptoms of illnesses (including abnormal laboratory measurements) that occurred or worsened during the course of the study. All of the AEs were recorded and evaluated by the investigators for causality, severity, and seriousness. AEs were assessed as serious if they were life-threatening, caused permanent harm, resulted in hospitalization or extension of in-patient hospital treatment, or were considered to be medically relevant by the investigator. All of the other AEs were categorized as nonserious.
Stool Sample Collection
Approximately 5 g of fresh stool (collected within 30 minutes of emission) was obtained during the scheduled visits at the study center. For determination of lactobacilli, enterobacteria, Clostridium, Bacteriodes, and bifidobacteria counts, 1 g was immediately stored at −80°C. For B longum BL999 quantification, 1 g of stool was suspended in 10 mL of sterile Ringers solution (Oxoid) containing 10% glycerol, homogenized, and stored at −80°C until further analysis.
For the analysis of IgA, 0.5 to 1.0 g of stool was suspended in 0.5 mL of extraction buffer (phosphate-buffered saline containing 3 mmol EDTA and 10.5 mg soybean trypsin), homogenized, and centrifuged at 13,000 rpm for 15 minutes at 4°C. The supernatant was stored at −80°C until further analysis. A commercial enzyme-linked immunosorbant assay (ELISA) was used (Immuno-Tek, Zeptometrix, Buffalo, NY) to determine the IgA stool levels.
The presence of total bacterial counts and lactobacilli, enterobacteria, clostridia, Bacteriodes, and bifidobacteria were determined using fluorescence in situ hybridization (FISH). The total number of bacteria present in the fecal samples was determined with the EUB 338 probe, which targets all bacteria (15). FISH analysis was essentially performed as described previously (16). Analysis was performed at Microscreen (Groningen, the Netherlands). Quantification of B longum BL999 was performed by Advanced Analytical Technologies (Piacenza, Italy) using a method specifically developed by the Nestlé Research Center and described elsewhere (17).
Sample size calculation was based on the primary outcome of demonstrating equivalence in the 2-month bifidobacteria counts in infants fed the study formula with or without B longum BL999 and a difference in the 2-month bifidobacterial counts between infants fed the study formula and those fed the control formula. The standard deviation (SD) of bifidobacteria counts quantified by FISH in a previous study was 0.7 log CFU/g (14). To demonstrate equivalence with delta = 0.7 log CFU/g, statistical power = 90%, significance level = 2.5% following Bonferroni adjustment, and an equivalence margin of 0.7 to −0.7 log CFU/mL, n = 32 per group would be required. After accounting for a 20% drop-out rate, n = 39 infants per group were expected. Stratified randomization was performed using SAS version 8.2 (SAS Institute, Cary, NC). Stratification was based on sex and mode of delivery (cesarean section vs vaginal birth).
Outcome measurements were calculated for the intention-to-treat (ITT) population unless otherwise indicated. Primary outcomes were analyzed using nonparametric methods. Bacterial counts were log-transformed and the 25th, 50th, and 75th percentiles were reported. The difference in bacterial count between the study formula group and the study formula + B longum BL999 group and the 90% confidence interval (CI) were calculated using the Hodges-Lehmann method. Once equivalence was demonstrated, bacterial counts between these groups (separately and combined) and the control formula group were compared by the Wilcoxon 2-sample test. All P values were adjusted by the Hommel method.
Analysis of the mean changes in body weight, length, and head circumference between 2 weeks and 4 months were performed using a mixed model and compared between groups using ANCOVA, correcting for sex. A noninferiority test was used to compare weight, and equivalence testing was used for length and head circumference. All P values were adjusted using the Dunnett test. Absolute stool immunoglobulin A (IgA) concentrations were compared using the equivalence testing Wilcoxon 2-sample test, with P values adjusted according to Hommel. Equivalence test was used to compare mean stool counts/day using ANOVA. The mean proportion of specific stool characteristics (consistency, color, and odor) and behavior (flatulence, spitting up, vomiting, crying, fussing, and colic) were analyzed by equivalence testing between groups using logistic regression. Significance level was adjusted for multiplicity using the Bonferroni method. Statistical analyses were performed using SAS version 8.2.
One hundred ninety healthy infants were enrolled in this study, and all of those in the formula groups received their assigned formula at least once (ITT population). There were no significant differences in baseline characteristics among the different groups (Table 2). Forty-nine infants (26%) were withdrawn before the end of the study, but data from these infants were used when available. The reasons for the withdrawals were AEs (n = 9), consent withdrawal by caregiver (n = 10), loss to follow-up (n = 16), and other (n = 16) (Fig. 1). Seven infants were withdrawn after 2 months; therefore, they were included in the per-protocol (PP) analysis. Six infants who completed the study were protocol violators and were excluded from the PP analysis.
Comparison of Bacterial Counts
At the 2-month time point, bifidobacteria were detected in the stools of 83% of infants in the study formula group, 79% of infants in the study formula + BL999 group, 67% of infants in the control formula group, and 88.6% of infants in the breast-fed group. Bifidobacteria also made up a relatively large percentage of the gut microbiota of infants fed either study formulas or breast milk, unlike infants fed the control formula (Table 3). The median bifidobacterial counts in the study formula and study formula + BL999 groups were similar (Table 4). The difference in the median counts between the 2 groups was 0.16 log CFU/g, and the 90% CI of the difference was 0 to 0.4 log CFU/g in the ITT population and 0.14 log CFU/g (90% CI [0–0.4 log CFU/g]) in the PP population. These differences were within the prespecified equivalence margin of −0.7 to 0.7 log CFU/g.
There was a significant difference in the bifidobacteria counts in the study formula groups (combined) and the control formula group (Wilcoxon-Mann-Whitney test, P = 0.0001). The differences were also significant with individual pairwise comparisons of the different groups (Wilcoxon-Mann-Whitney test, P = 0.0002 for the study formula vs control formula group and P = 0.007 for the study formula + BL999 vs control formula group). By contrast, there was no significant difference in the bifidobacteria counts between either study formula group and the breast-fed group (Wilcoxon-Mann-Whitney test, P > 0.05; Table 3).
No significant differences were observed in the prevalence of lactobacilli, enterobacteria, Clostridia, and Bacteroides among the study formula groups and the control group at 2 months (Wilcoxon-Mann-Whitney test, P > 0.05); however, Clostridia appeared to be less prevalent among breast-fed infants (54.3%) than in the formula groups (>85% in all 3 formula groups). Total bacterial counts as well as lactobacilli, enterobacteria, Clostridia, and Bacteroides counts at 2 months were not significantly different between the 2 study formula groups and between either study formula group and the control formula or the breast-fed group (Wilcoxon-Mann-Whitney test, P > 0.05; Table 4). Overall microbiota profiles are shown on Table 3.
At the 1-month time point, B longum BL999 was detected in infants fed the study formula containing this strain, but it was no longer detectable at the 2-month time point.
A comparison of the mean weight gain between 2 weeks and 4 months in the study formula and control groups showed that both study formulas support growth at least as well as the control formula (Table 5). The difference in the mean weight gain was 0.23 g/day (90% CI [−2.42 to 2.87]) between the study formula and the control formula groups and 0.02 g/day (90% CI [−2.6 to 2.63]) between the study formula + BL999 and the control formula groups (ITT population). In both cases, the lower limit of the 90% CI of the differences was above the lower limit of −0.3 g/day set by the American Association of Pediatrics (18).
Gain in body length and head circumference between 2 weeks and 4 months was not significantly different between the study formula groups and the control formula group.
At the 1-month time point, there was a significant difference in the IgA concentration between the study formula group and the control group (P < 0.05). At 2 months, there were no significant differences between either of the study formula groups and the control group, although the trend persisted (P = 0.07 for both comparisons). Comparison with the breast-fed group at 2 months of age yielded no significant difference (P = 0.05 for the comparison with study formula and P = 0.1 for the comparison with study formula + BL999). However, a significant difference was observed between the control formula and the breast-fed groups (P = 0.001). Although the differences in concentration were not statistically significant, the IgA concentrations in the 2 study formula groups (42 and 33 μg/mL in the study formula with and without BL999, respectively) were intermediate between the control formula (22 μg/mL) and the breast-fed groups (53 μg/mL).
Digestive Tolerance and AEs
There was no difference in stool frequency between the 2 study formula groups and the control formula group; however, stool frequency was significantly higher in the breast-fed infants compared with those fed either study formula (mean frequency difference 2.03 and 1.66 for the study formula with or without BL999, respectively; P < 0.001 for both). Compared with infants in the control formula group, those in the 2 study formula groups were significantly more likely (P < 0.001) to have green stools (odds ratio [OR] 3.45, 95% CI [1.23–9.67] for study formula alone and OR 4.44, 95% CI [1.58–12.46] for study formula + BL999) and significantly less likely (P < 0.001) to have yellow stools (OR 4.44, 95% CI [1.58–12.46] for study formula and OR 0.22, 95% CI [0.08–0.6] for study formula + BL999). Infants in the study formula + BL999 group were significantly more likely to have soft stools (OR 2.80, 95% CI [1.05–7.47]). Similar comparisons of statistical significance regarding stool consistency and color also applied to infants in the study formula groups versus breast-fed infants.
Other symptoms of digestive tolerance (frequency of vomiting, spitting up, crying and being fussy, colic, and flatulence) were not significantly different among groups (data not shown).
Fifty-three AEs were reported in 43 infants (Table 6). The total number of infants experiencing AEs in each group was not significantly different. In all of the groups of infants, the most common AEs were ones that affected the GI or upper respiratory tract.
The ability of breast milk to modulate the composition of the infant gut microbiota contributes to its beneficial health effects. Among the properties of breast milk affecting the development of the infant gut microbiota are the type and quantity of protein. Protein composition in breast milk is predominantly whey early in lactation (19–21), and this property, along with low phosphate and buffering capacity, is thought to promote growth of bifidobacteria in the infant gut (13). In the present study we tested whether the bifidogenic properties of a whey-predominant infant formula, low in protein and phosphate content and high in lactose, were sufficiently optimized in its bifidogenic properties by comparing it with the same formula, which had been supplemented with 2 × 107 CFU/g BL999.
We showed that the study formula had bifidogenic properties similar to the formula supplemented with BL999. Both the absolute bifidobacterial counts and the relative proportion of bifidobacteria in the microbiota were significantly higher in infants fed the study formulas compared with those who were fed the control formula. Noteworthy is that the protein content of the tested control formula was within the recommended compositional requirement values proposed by the European Society for Pediatric Gastroenterology, Hepatology, and Nutrition (22). Together with the observation that the microbiota composition of infants fed the study formulas is similar to that of breast-fed infants, the outcome of the present trial suggests that the study formula is able to modulate the gut microbiota, at least for the duration of feeding with the study formulas. The observation that the study formula supplemented with BL999 did not induce colonization of the infants' gut with higher numbers of bifidobacteria compared with the unsupplemented formula suggests that the bifidogenic properties of the latter are as good as adding live bifidobacteria to formulas. It also suggests that there is no synergistic effect to adding bifidobacteria to the study formula. It was not possible to measure specific components in the study formula that influence the infant microflora because the study formula itself optimizes the microbiota; however, an effect of BL999 on other physiological parameters cannot be excluded. At the 2-month time point, BL999 was no longer detected, implying that the study formula preferentially stimulates growth of the endogenous bifidobacteria, which out-compete the added BL999 by the 2-month time point (the study did not aim to measure colonization by the probiotic bacteria, but to assess its action in the GI tract). Alternatively, the plating technique used to quantify BL999 may not have been sensitive enough, and it will be interesting to confirm these results by FISH when probes become available.
In addition to bifidobacteria analysis, enterobacteria and clostridia appeared to constitute a higher percentage of the gut microbiota in infants fed the control formula compared with the study formula and the breast-fed groups. Furthermore, apart from Clostridia prevalence, there was no significant difference in counts of the bacteria in the study formula and breast-fed groups, suggesting modulation of different components of the microbiota resulting in a microbiota profile (and not only bifidobacteria) similar to breast-fed infants (Table 3).
The benefits of the increased bifidobacterial counts in infants who were fed the study formula were not clear in this short-term study. Although there were signs of better digestive tolerance in the study formula groups, these were not statistically significant, but the study was not powered for it. For example, hard and formed stools were less likely to occur in the study formula groups compared with the control group, but this was statistically significant only for the comparison of the control group with the study formula + BL999 group.
By contrast, comparison with the breast-fed infants pointed to significant differences in some of the indicators for digestive tolerance (eg, the higher frequency of liquid and soft stools in the breast-fed infants); however, there were no differences in the occurrence of AEs (including GI AEs) between different groups, including the breast-fed group. As such, it could be argued that the bifidogenic effect may be a pure microbiologic effect, make link between a bifidogenic effect and other physiological parameters cannot be excluded.
Conversely, there was an indication that the intestinal immunity in infants fed the study formulas was improved. Although differences in IgA concentrations were not statistically significant between the various groups, infants fed the study formulas (with or without added bifidobacteria) consistently had higher stool IgA concentrations compared with infants in the control group, although these were lower than in the breast-fed group. This is consistent with the role of commensal bacteria in the development of the intestinal immune system (23), and the important role of IgA in this process.
The study formulas promoted growth of infants similar to that of breast-fed infants. Additionally, infants fed the study formula had weight gain within the preset margin of equivalence compared with the control formula (a formula approved for normal nutrition of infants), indicating its safety.
In conclusion, this study shows that an infant formula with a composition closer to that of human milk is more bifidogenic than a reference formula and is equivalent to the same study formula supplemented with a probiotic such as B longum. This property also leads to an intestinal microbiota profile closer to breast-fed infants than a reference formula. Furthermore, this study also suggests that the formula with a composition closer to human milk tended to enhance the development of a better mucosal immune response compared with a control formula. Further studies are needed to determine the potential clinical benefits of this kind of formula.
We thank Marie-Christine Buchweiller and Sylvie Voirin (Nancy) and Corina Boschat and Françoise Chauffard (Vevey) for the monitoring of the study and the follow-up of the children. We also thank Bernard Berger (Nestlé Research Center) for the probiotic identification probe and Prof Fares Namour (CHU-Nancy) for the IgA processing and analysis. We thank OmniScience SA, which provided medical writing services on behalf of Nestlé Nutrition.
1. Harmsen HJ, Wildeboer-Veloo AC, Raangs GC, et al. Analysis of intestinal flora development in breast-fed and formula-fed infants by using molecular identification and detection methods. J Pediatr Gastroenterol Nutr 2000; 30:61–67.
2. Wharton BA, Balmer SE, Scott PH. Sorrento studies of diet and fecal flora in the newborn. Acta Paediatr Jpn 1994; 36:579–584.
3. Newburg DS. Oligosaccharides in human milk and bacterial colonization. J Pediatr Gastroenterol Nutr 2000; 30(suppl 2):S8–S17.
4. Mackie RI, Sghir A, Gaskins HR. Developmental microbial ecology of the neonatal gastrointestinal tract. Am J Clin Nutr 1999; 69:1035S–1045S.
5. Tannock GW. The acquisition of normal microflora of the gastrintestinal tract. In: Gibson SAW, editor. Human Health: The Contribution of Microorganisms. London: Springer-Verlag; 1994. pp. 1–16.
6. Kalliomaki M, Kirjavainen P, Eerola E, et al. Distinct patterns of neonatal gut microflora in infants in whom atopy was and was not developing. J Allergy Clin Immunol 2001; 107:129–134.
7. Sepp E, Julge K, Mikelsaar M, et al. Intestinal microbiota and immunoglobulin E responses in 5-year-old Estonian children. Clin Exp Allergy 2005; 35:1141–1146.
8. Ouwehand AC, Isolauri E, He F, et al. Differences in Bifidobacterium flora composition in allergic and healthy infants. J Allergy Clin Immunol 2001; 108:144–145.
9. Murray CS, Tannock GW, Simon MA, et al. Fecal microbiota in sensitized wheezy and non-sensitized non-wheezy children: a nested case-control study. Clin Exp Allergy 2005; 35:741–745.
10. Sudo N, Sawamura S, Tanaka K, et al. The requirement of intestinal bacterial flora for the development of an IgE production system fully susceptible to oral tolerance induction. J Immunol 1997; 159:1739–1745.
11. Ishikawa H, Tanaka K, Maeda Y, et al. Effect of intestinal microbiota on the induction of regulatory CD25+ CD4+ T cells. Clin Exp Immunol 2008; 153:127–135.
12. Heavey PM, Rowland IR. The gut microflora of the developing infant: microbiology and metabolism. Microbial Ecol Health Dis 1999; 11:75–83.
13. Langhendries JP, Detry J, Van Hees J, et al. Effect of a fermented infant formula containing viable bifidobacteria on the fecal flora composition and pH of healthy full-term infants. J Pediatr Gastroenterol Nutr 1995; 21:177–181.
14. Rochat F, Cherbut C, Barclay D, et al. A whey-predominant formula induces fecal microbiota similar to that found in breast-fed infants. Nutr Res 2007; 27:735–740.
15. Amann RI, Binder BJ, Olson RJ, et al. Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Appl Environ Microbiol 1990; 56:1919–1925.
16. Langendijk PS, Schut F, Jansen GJ, et al. Quantitative fluorescence in situ hybridization of Bifidobacterium spp. with genus-specific 16S rRNA-targeted probes and its application in fecal samples. Appl Environ Microbiol 1995; 61:3069–3075.
17. Rougé C, Goldenberg O, Ferraris L, et al. Investigation of the intestinal microbiota in preterm infants using different methods. Anaerobe 2010; 16:362–370.
19. Kunz C, Lonnerdal B. Human-milk proteins: analysis of casein and casein subunits by anion-exchange chromatography, gel electrophoresis, and specific staining methods. Am J Clin Nutr 1990; 51:37–46.
20. Kunz C, Lonnerdal B. Re-evaluation of the whey protein/casein ratio of human milk. Acta Paediatr 1992; 81:107–112.
21. Lonnerdal B. Nutritional and physiologic significance of human milk proteins. Am J Clin Nutr 2003; 77:1537S–1543S.
22. Koletzko B, Baker S, Cleghorn G, et al. Global standard for the composition of infant formula: recommendations of an ESPGHAN coordinated international expert group. J Pediatr Gastroenterol Nutr 2005; 41:584–599.
23. Forchielli ML, Walker WA. The role of gut-associated lymphoid tissues and mucosal defence. Br J Nutr 2005; 93(suppl 1):S41–S48.