Factors Influencing the Composition of the Intestinal Microbiota in Early Infancy

Subjects and Study Design

The KOALA Birth Cohort Study is a prospective birth cohort in the Netherlands. The design of this study was described in detail elsewhere.¹² Briefly, from October 2000 until December 2002, we recruited participants with diverse lifestyles, at 34 weeks of gestation. Pregnant women with a conventional lifestyle (N = 2343) were recruited from an ongoing prospective cohort study on pregnancy-related pelvic girdle pain in the Netherlands.¹³ Pregnant women with an alternative lifestyle (N = 491) were recruited through organic food shops, anthroposophic doctors and midwives, Steiner schools, and magazines.

Beginning halfway during recruitment of the cohort (subjects recruited from January 2002 onward), fecal samples were collected from infants (N = 1176) at the age of 1 month. Subjects received a feces tube with spoon (Sarstedt, Nümbrecht, Germany), together with a sanitary napkin, an instruction form, and a brief questionnaire (feces questionnaire). Parents collected a fecal sample by placing a sanitary napkin in the diaper (to prevent absorption of the feces by the diaper) and collecting the feces out of the napkin into the collection tube, and they sent the sample as soon as possible (the same day) to the Department of Medical Microbiology at the University Hospital of Maastricht, by mail. Exclusion criteria were insufficient amount of feces (<1 g), feces collected before the age of 3 weeks or after the age of 6 weeks, and missing feces questionnaire.

Information on Potential Determinants

During pregnancy and the first months of the infant's life, information on perinatal determinants of the child's health, as well as hygiene, infections, nutrition, child rearing, and other lifestyle characteristics, was collected for all members of the cohort with repeated questionnaires. On the basis of these questionnaires, the following variables were selected as potential determinants of gut microbiotic composition: maternal education (lower education, vocational education, higher general secondary/preuniversity education, or higher vocational/academic education); maternal diet, defined as (1) regular diet (>50% of meat, eggs, vegetables, fruit, and milk of regular origin), (2) vegetarian diet (no meat but other products mainly of regular origin), (3) organic/biodynamic diet (>50% of meat, eggs, vegetables, fruit, and milk of organic or biodynamic origin), or (4) organic/biodynamic vegetarian diet; maternal probiotic use during pregnancy (frequency of consumption of dairy products with additional lactic acid-producing bacteria); maternal antibiotic use during last month of pregnancy (yes or no); prolonged rupture of membranes (<24 or >24 hours before delivery); place and mode of delivery (vaginal delivery at home, vaginal delivery in hospital, artificial delivery [forceps delivery and vacuum extraction] in hospital, or cesarean section in hospital); hospitalization of the infant after birth (days of hospitalization immediately after birth); infant gender (male or female); gestational age (<37, 37–41, or >41 weeks); birth weight (<2500, 2500–4500, or >4500 g); birth season (winter, spring, summer, or autumn); type of infant feeding during the first 1 month of life (exclusively breastfed, exclusively formula fed, or a combination); brand of infant formula (exclusive consumption of brand A, brand B with locust bean gum, brand C, or brand D with oligosaccharides during the first 1 month of life); antibiotic/antifungal use by the infant during the first 1 month of life (yes or no); fever of the infant in the first 1 month of life (yes or no); number of siblings (0, 1 or 2, or >2); living on a farm (yes or no); and having furry pets (none, cat, dog, other, or a combination).

DNA Purification From Feces

At the laboratory, fecal samples were 10-fold diluted in peptone/water (Oxoid CM0009) containing 20% (vol/vol) glycerol (Merck, Darmstadt, Germany) and were stored at −20°C until analysis. For DNA isolation, 0.2 mL of the diluted feces was added to a 2-mL vial containing ∼300 mg of glass beads (diameter: 0.1 mm) and 1.4 mL of ASL buffer from the QIAamp DNA stool minikit (Qiagen, Hilden, Germany), and the samples were disrupted in a mini-bead beater (Biospec Products, Bartlesville, OK) at 5000 rpm for 3 minutes. Subsequently, the bacterial DNA was isolated from the samples with the QIAamp DNA stool mini kit, according to the instructions provided by the manufacturer. The DNA was eluted in a final volume of 200 μL.

Microbial Analysis With Real-Time PCR Assays

DNA from all fecal samples was subjected to real-time PCR assays for bifidobacteria, E coli, C difficile, Bacteroides fragilis group, lactobacilli, and total bacteria based on 16S rDNA gene sequences (primers and probes are listed in Table 1). Development and validation of the real-time PCR assays were described in detail elsewhere.⁶,¹⁴–¹⁷ The 5′-nuclease technique was used for detection of bifidobacteria, E coli, C difficile, and members of the B fragilis group. For detection of bifidobacteria, amplifications were conducted in a total volume of 50 μL, containing 1× TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA), 300 nmol/L levels of both primers, 150 nmol/L TaqMan probe, and 20 μL of purified target DNA (see above). For E coli, C difficile, and B fragilis group, amplifications were conducted in a total volume of 25 μL, containing 1× TaqMan Universal PCR Master Mix (Applied Biosystems), 900 nmol/L levels of both primers, 200 nmol/L TaqMan probe, and 10 μL of purified target DNA. The amplification (2 minutes at 50°C, 10 minutes at 95°C, and 42 cycles of 15 seconds at 95°C and 1 minute at 60°C) and detection were conducted with an Applied Biosystems Prism 7000 sequence detection system (Applied Biosystems).

For quantification of lactobacilli and total bacterial load, real-time detection of PCR products was conducted with SYBR Green I (Bio-Rad Laboratories, Hercules, CA). The PCR for lactobacilli was conducted in a total volume of 25 μL, containing 1× iQ SYBR Green Supermix (Bio-Rad), 500 nmol/L levels of both primers, and 5 μL of purified target DNA. The amplification was conducted as follows: 5 minutes at 95°C, followed by 35 cycles consisting of 15 seconds at 95°C, 20 seconds at 58°C, and 45 seconds at 72°C, with a final extension step at 72°C for 5 minutes. After amplification, melting curve analysis was performed from 60°C to 95°C, with increments of 0.5°C per 10 seconds. For quantification of the total bacterial load, amplifications were conducted in a total volume of 25 μL, containing 1× iQ SYBR Green Supermix (Bio-Rad), 300 nmol/L levels of both primers, and 5 μL of purified target DNA. The amplification consisted of 4 minutes at 95°C and 30 seconds at 60°C, followed by 35 cycles of 30 seconds at 60°C, 15 seconds at 95°C, and 30 seconds at 60°C. Finally, melting curve analysis was performed from 60°C to 95°C, with increments of 0.5°C per 10 seconds. Amplification, melting curve analysis, and detection were conducted with the MyiQ single-color, real-time PCR detection system (Bio-Rad).

Statistical Analyses

Values of log₁₀ colony-forming units (CFU) per gram for the bacterial groups and species were calculated for each stool sample from the threshold cycle values by using the constructed standard curves. The prevalence of colonization was expressed as the percentage of infants colonized with a specific bacterial group or species. To determine the unadjusted overall effects of the individual determinants on the prevalence and counts of the bacteria simultaneously, the Mann-Whitney rank-sum test was used.

To analyze the effects of the individual determinants, with adjustment for all other determinants, 2 multivariate approaches were used. First, linear regression analyses were used to examine the effects of the determinants on the bacterial counts. In the linear regression analyses, only infants who were colonized with the bacterial group or species were included. Total bacterial count was added as an additional independent variable to account for differences in the consistency of fecal samples. Second, logistic regression analyses were used to examine the effect of the determinants on the prevalence of colonization (colonized compared with not colonized). Both linear and logistic regression models included all of the determinants under study as independent variables and 1 of the bacterial groups or species as the dependent variable at a time. Length of hospitalization (in days) was included as a continuous variable in the regression models, whereas all other independent variables were incorporated as categorical variables.

To limit the chance of falsely rejecting the null hypothesis (no association) as a result of multiple testing, we chose to set α at .01 (2-sided), instead of the usual .05. Therefore, we present 99% confidence intervals (CIs) for the odds ratios (ORs) from logistic regression analyses.

Ethical Considerations

The KOALA Study was approved by the ethics committee of the University Hospital of Maastricht, and all parents signed informed consent for the study.

Study Samples

Fecal samples from a total of 1176 infants were collected. After exclusion of samples that were of insufficient amount (n = 65), samples that were collected before the age of 3 weeks or after the age of 6 weeks (n = 54), and samples for which the feces questionnaire was missing (n = 25), fecal samples from 1032 infants were included for analysis.

Almost all infants were colonized with bifidobacteria, and these bacteria outnumbered all other bacterial groups and species (Table 2). The majority of infants were also colonized with E coli and members of the B fragilis group, whereas both the prevalence and counts of lactobacilli and C difficile were much lower.

Table 3 shows the prevalence and counts of the bacterial groups and species for the individual determinants under study. In Table 4, the associations between these determinants and the prevalence and counts of the fecal bacteria under study, as determined in the multivariate analyses, are presented. More-detailed information on these associations, including regression coefficients and ORs from the linear and logistic regression analyses, is shown in Table 5.

Maternal Education, Diet, Antibiotic Use, and Probiotic Use

More than one half of the women in this population (n = 572) had a higher vocational or academic degree. Maternal education was not associated, however, with the infants' gut microbiotic composition.

Infants whose mothers consumed an organic or biodynamic diet seemed to have slightly lower numbers of E coli in their stools than did infants whose mothers consumed a regular diet (Table 3); however, this association was not observed in the multivariate analysis (data not shown). Mothers who consumed an organic diet more often (95%) breastfed their infants exclusively, compared with mothers who consumed a regular diet (59%), which suggests that the type of infant feeding was the underlying cause of the association between maternal diet and infant's microbiotic composition in the univariate analysis. Probiotic use and antibiotic use by the mother during pregnancy had no influence on the infant's gut microbiotic composition.

Delivery and Birth Characteristics

Amniotic membranes ruptured >24 hours before delivery for only 36 women. We found no association between prolonged rupture of membranes among women and the gut microbiotic composition of their infants.

More than one half of the infants were born in the hospital, and ∼10% of all infants were born through cesarean section. In comparison with vaginal delivery at home, cesarean section resulted in lower colonization rates and counts of bifidobacteria and B fragilis-group species, whereas prevalence and counts of C difficile and counts of E coli were higher. The most-pronounced differences in colonization were seen for the B fragilis group and C difficile; compared with infants born vaginally at home, the median counts of B fragilis-group bacteria and C difficile were ∼100-fold lower and ∼100-fold higher, respectively, for infants born through cesarean section (Table 3). After adjustment for the other determinants, cesarean section was still associated with lower counts of bifidobacteria, a lower colonization rate and counts of B fragilis-group species, and higher counts of C difficile. In contrast, counts and colonization rate of E coli were no longer associated with cesarean section in the adjusted analyses. This is illustrated in Table 5 by a regression coefficient for counts close to 0 and an OR for prevalence close to 1.0. Although vaginal delivery and artificial delivery in the hospital seemed to be associated with higher counts and prevalences of E coli and C difficile, compared with home birth (Table 3), these associations were present only in the unadjusted analyses.

Hospitalization after birth was associated only with a higher C difficile colonization rate after controlling for other determinants. The prevalence of C difficile increased ∼13% per day of hospitalization, compared with nonhospitalized infants (OR: 1.13; 99% CI: 1.01–1.25).

Only 11 infants in this study population were born premature (<37 weeks). These infants were colonized more often with C difficile (64%, as opposed to 23%), with considerably higher counts (7.12 log₁₀ CFU/g feces, compared with 5.06 log₁₀ CFU/g feces), compared with term infants (Table 3). With adjustment for other determinants, the counts were still statistically significantly higher (Tables 4 and 5). The OR was still consistent with a higher colonization rate (OR: 4.47) but was not statistically significant, probably because of the low power caused by the small number of infants. No association was found between the gut microbiota and birth weight, birth season, or gender of the infant.

Infant Feeding

Most infants (n = 700) were breastfed exclusively up to the first 1 month of life, whereas 232 infants were formula fed exclusively and 98 infants received a combination of breastfeeding and formula feeding. Exclusively formula-fed infants were more often colonized with E coli, C difficile, B fragilis group, and lactobacilli than were their exclusively breastfed counterparts, as shown in both the unadjusted (Table 3) and adjusted (Tables 4 and 5) analyses. The counts of E coli, C difficile, B fragilis group, and lactobacilli were also significantly higher for formula-fed infants, compared with breastfed infants, in the unadjusted analyses. In the adjusted analyses, only counts of C difficile were still significantly higher for formula-fed infants, whereas counts of both E coli (P = .03) and B fragilis (P = .027) still tended to be higher for formula-fed infants (not reaching the level of significance of P < .01) (Table 5).

In our population, 4 brands of formula were used frequently. Brand B contained locust bean gum and brand D was enriched with oligosaccharides, whereas the others were not. Infants fed exclusively with 1 of these 4 formulas were compared. As shown in Table 3, infants fed the oligosaccharide-enriched formula (brand D) harbored greater numbers of bifidobacteria in their stools. After adjustment for the other determinants under study, counts of bifidobacteria (coefficient: 0.60; P = .04) and also counts of lactobacilli (coefficient: 0.75; P = .02) tended to be higher for infants fed formula D, compared with reference formula A (data not shown).

Antibiotics, Antimycotic Agents, and Fever

Oral use of antibiotics (mainly amoxicillin) by the infant during the first 1 month of life resulted in decreased numbers of bifidobacteria and B fragilis-group species. Lower counts of bifidobacteria were also observed after oral administration of the antimycotic miconazole (Tables 3 and 4). Infants who experienced a fever in their first 1 month of life did not have different gut microbiotic composition, compared with infants without a fever.

Home Environment

Infants with older siblings had lower total bacterial counts per gram of feces than did infants without siblings, as seen in the unadjusted analysis (Table 3). After adjustment for these differences in total bacterial counts and other potential differences between infants with and without siblings, a greater proportion of bifidobacteria was found for infants with older siblings, compared with infants without siblings (Tables 4 and 5). Infants' gut microbiota was not associated with farm residence or the presence of furry pets in the home, as determined in both the unadjusted (Table 3) and multivariate (data not shown) analyses.