Conditions of Bifidobacterial Colonization in Preterm Infants: A Prospective Analysis

Abstract

Background: 

Premature birth results in a delayed and abnormal qualitative pattern of gut colonization. This abnormal pattern is thought to affect intestinal development and contribute to a higher risk of gastrointestinal infectious diseases such as neonatal necrotizing enterocolitis (NEC). In particular, bifidobacteria are thought to play a major role. We therefore studied bifidobacterial colonization in preterm infants during the first month of life.

Patients and Methods: 

Fecal samples were prospectively analyzed in 52 infants born at a gestational age ranging from 30 to 35 weeks fed with a preterm formula alone and, in 18, with their mother's milk. Fecal samples were collected twice per week during the hospital stay. Bifidobacterial colonization was analyzed with culture and a molecular method.

Results: 

Bifidobacterial colonization occurred in 18 infants at a median age of 11 days, always greater than the corrected mean gestational age of 35.4 weeks (SD, 0.9) and greater than 34 weeks for 16 of 18. Colonization by bifidobacteria was affected by neither birthweight nor mode of delivery nor antibiotics given to the mother or infant. In contrast, birth gestational age had a significant impact on colonization by bifidobacteria (P < 0.05), which always occurred in children born at a birth gestational age greater than 32.9 weeks (P < 0.05).

Conclusions: 

Birth gestational age seems to act as a major determinant of bifidobacterial colonization in the premature infant, suggesting the role of gut maturation, a finding that should probably be taken into account in manipulations of the gut flora aimed at reducing NEC.

INTRODUCTION

Colonization of the gastrointestinal tract of newborn infants starts immediately after birth, from maternal and environmental sources, until a complex bacterial community with a dominance of strictly anaerobic genera is established. Mode of delivery and type of feeding are known to strongly influence the pattern of bacterial colonization. However, little is known about gut bacterial establishment in preterm infants, except for its delayed appearance, particularly for beneficial bacteria like bifidobacteria, normally dominant in full-term infants ^(1,2). In contrast, the number of potentially pathogenic bacteria is high. This impaired intestinal colonization process may affect the intestinal development. Indeed, resident bacteria are able to exert important functions on the host's physiology, including metabolic activities, trophic effect on the intestinal epithelium, and barrier effect against potential pathogens ⁽³⁾, leading to a higher risk of gastrointestinal diseases such as necrotizing enterocolitis (NEC) ⁽⁴⁾. The importance of these host–microbe interactions is more vital in the neonatal period during maturation of the developing intestine ⁽⁵⁾. Among the gut microbiota, specific strains, particularly bifidobacteria, are thought to play a major role. Bifidobacterial supplementation was demonstrated to inhibit NEC-like lesions in 2 NEC animal models ^(6,7), and 2 recent studies of prophylactic administration of probiotics containing bifidobacteria in neonates with very low birth weight reported a reduction in the incidence and severity of NEC ^(8,9). We thus aimed at studying bifidobacterial colonization in preterm infants during the first weeks of life with use of conventional culture and a culture-independent method.

PATIENTS AND METHODS

Study Design and Data Collection

This study, conducted between December 2002 and February 2005 in 2 French hospitals, was approved by a local ethics committee, and informed written consent was obtained from the parents of the patients involved in the study. Included infants were preterm infants born at a gestational age ranging from 30 to 35 weeks. All were hospitalized in the neonatal intensive care unit. Exclusion criteria were any deformities, chromosomal abnormalities, or inappropriate weight for gestational age.

Fecal samples were collected twice per week during the hospital stay. The fecal samples were placed in 2 sterile tubes, 1 containing 0.5 mL of brain heart infusion with glycerol 15% as a cryoprotective agent. All of the samples were immediately frozen at −80°C until analyses were performed.

Microbiological Analysis of Fecal Samples

After thawing the samples collected in brain heart infusion, serial dilutions from 10⁻² to 10⁻⁶ were performed and spread using an automatic spiral system (AES Laboratoire, Bruz, France) on Wilkins-Chalgren agar modified for bifidobacteria as previously described ⁽⁶⁾. This medium consists of Wilkins-Chalgren agar base supplemented with D-glucose 10 g/L, L-cysteine 0.5 g/L, Tween 80 0.5% vol/vol, and kanamycin 7.5 mg/L. Plates were incubated for 5 days in an anaerobic cabinet (N₂:CO₂:H₂, 80:10:10; AES Laboratoire). The count threshold was 10³ CFU (colony-forming units)/g of feces and bacterial counts were expressed as log₁₀ CFU/g of feces. Genus determination was performed according to Gram stain, morphological characteristics, presence of fructose-6-phosphate-phosphoketolase, and polymerase chain reaction (PCR) using bifidobacterial specific primers (ie, S-G-Bif-164-a-S-18 [GGG TGG TAA TGC CGG ATG] for the forward primer and S-G-Bif-662-a-A-18 [CCA CCG TTA CAC CGG GAA] for the reverse primer) ⁽¹⁰⁾. Species were determined using a multiplex PCR as described by Mullie et al ⁽¹¹⁾.

Extraction and Purification of Total DNA

For the culture-independent method, fecal DNA was extracted from the stools samples collected without the cryoprotective agent using a bead-beating method, as previously described ⁽¹²⁾.

PCR Amplification and Bifidobacterial Electrophoresis

To amplify the 16S RNA gene of bifidobacterial strains, the same specific primers were used—S-G-Bif-164-a-S-18 and S-G-Bif-662-a-A-18. A GC-rich sequence (5′ CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GCA CGG GGG G 3′) was attached to the 5′ end of the primer S-G-Bif-662-a-A-18. PCR studies were performed using the Taq DNA polymerase (AmpliTaq Gold; Perkin-Elmer, Foster City, CA) as previously described ⁽¹²⁾. PCR amplicons were separated by temporal temperature gel electrophoresis (TTGE) using the Decode universal mutation detection system (Bio-Rad, Hercules, CA) ⁽¹²⁾. In addition, known bifidobacterial strains were loaded to allow standardization of band migration and gel curvature among different gels. Furthermore, these standards allowed their specific distinction and band identification by comparison of the distance migration. These bifidobacterial strains consisted of Bifidobacterium adolescentis (CIP64.59T), B angulatum (CIP104167T), B animalis subspecies animalis (CIP105419T), B animalis subspecies lactis (CIP105265), B bifidum (CIP56.7T), B breve (CIP64.69T), B dentium (CIP104176T), B gallicum (CIP 103417T), B longum biovar longum (CIP 64.62T), B longum biovar infantis (CIP 64.67), B longum biovar longum (CIP 64.63), and B pseudocatenulatum (CIP104168T). They were included in ladders (M1, M2; Fig. 1). When identification could not be performed with the multiplex PCR or the TTGE profile, bands were excised from TTGE gels, cloned, and sequenced as previously described ⁽¹³⁾.

FIG. 1: Bifidobacterial TTGE profiles of fecal samples of the 18 preterm infants with colonization by bifidobacteria. Lanes are marked with the code of the infants, followed by the day of sampling. Ladders are mixture of strains. M1 comprised B adolescentis (1), B angulatum (2), B longum (3), B longum CIP 64.63 (4), B pseudocatenulatum (5), B animalis subspecies animalis (6). M2 comprised B bifidum (7), B dentium (8), B animalis subspecies lactis (9).

Statistical Analyses

Parametric results were reported as means with SD. Statistical comparisons between infants with versus without colonization by bifidobacteria were performed with the Fisher exact test for infants' characteristics and the Student t test for continuous data (ie, birth weight, gestational age, and days of life). P < 0.05 was considered significant.

RESULTS

Fifty-two preterm infants, born with a birth weight ranging from 990 to 2750 g at a mean gestational age of 33.4 weeks (SD, 1.2; range, 30–35.9 weeks), were included (Table 1). They were delivered vaginally (n = 28) or by cesarean section (n = 24). All were fed with a standard preterm formula or fermented preterm formula and 18 were also fed with their mother's milk, which was first administered from day 5 to day 14. The fermented formula is a standard formula for preterm infants that includes in its manufacturing process a fermentation step with 2 probiotic strains (ie, B breve C50 and Streptococcus thermophilus), followed by a thermic stage to kill bacteria.

TABLE 1: Characteristics of preterm infants

The follow-up lasted from 4 to 34 days (18 ± 7 days), with an average of 5 fecal samples per infant. All of the infants were followed at least 2 weeks, except 8 who were followed between 11 and 13 days, 1 who was followed for 9 days, and 1 who underwent only 1 sample collection at day 4.

Thirty-four of the preterm infants (66%) still had no colonization at 18.7 ± 7.4 days (range, 7–34 days). Eighteen of the 52 infants (34%) had colonization between the fourth and the 23rd days of life (11 ± 6 days; Table 2; Fig. 1). Both methods—culture and PCR-TTGE—were in accordance for the detection of bifidobacterial species except in 3 infants (infants H, I, and J). However, these 3 infants had only 1 positive fecal sample, and for 1 infant culture could not be performed the same day because of an insufficient quantity of stool. In these 3 cases, identification of the species was performed with sequencing. In the other cases, bifidobacteria were detected at a level ranging from 4.1 to 10.0 log₁₀ CFU/g feces (7.2 ± 1.7 log₁₀ CFU/g; Table 2).

TABLE 2: Characteristics of bifidobacterial colonization in the 18 colonized preterm infants

Most of the infants harbored only 1 species and only 5 infants harbored 2 to 3 species. The most frequent isolated species were B longum (6 infants), B bifidum (5 infants), and B breve (5 infants). B animalis subspecies lactis and B pseudocatenulatum were isolated in 4 infants each and B adolescentis and B dentium in 1 infant each.

Gut bifidobacterial colonization appeared to be transitory in 3 infants (only 1 positive fecal sample), persistent and remaining at a low level in 3 infants (4.7 ± 1.7 log₁₀ CFU/g), and persistent and reaching a high level in 6 infants (8.4 ± 0.9 log₁₀ CFU/g; Table 2). In the 6 remaining infants the status of the bifidobacterial colonization could not be determined because of the lack of a fecal sample after the only positive sample.

Bifidobacterial colonization occurred at a corrected gestational age of 35.4 ± 0.9 weeks in the 2 study centers, at later than 34 weeks for 16 of 18 infants. One infant with colonization at this threshold age (34 weeks) exhibited only a transient colonization (<3 days), and the only infant with colonization before this threshold (33.6 weeks) had only 1 fecal sample collected, not allowing a follow-up.

Colonization by bifidobacteria was affected neither by birth weight, mode of delivery, antibiotics given to the mother or infant, nor type of feeding (ie, type of formula and mother's milk supplementation) (Table 1). In contrast, gestational age at birth was a significant condition for colonization by bifidobacteria (P < 0.05), which always significantly occurred in children born at greater than a threshold birth gestational age of 32.9 weeks (P < 0.05; Fig. 2). In the only infant with colonization who was born before this threshold (32.7 weeks) colonization occurred at day 9 at a high level (7.6 log₁₀ CFU/g) and decreased 3 days later to a low level (3.6 log₁₀ CFU/g), suggesting a short transient colonization. Moreover, the colonization in the 2 infants born at the threshold gestational age (32.9 weeks) was also transient (only 1 positive sample at day 23) and at a low level (<3.0 log₁₀ CFU/g).

FIG. 2: A significant threshold birth gestational age of 32.9 weeks between infants without colonization (n = 34) and those with colonization (n = 18) by bifidobacteria (P < 0.05). Solid lines indicate median gestational age.

DISCUSSION

To our knowledge, our study suggests for the first time that a threshold birth age of 32.9 weeks may at least partially condition gut colonization by bifidobacteria, thus further documenting previous findings of delayed bifidobacterial establishment in premature infants. Sakata et al ⁽¹⁴⁾ and Blakey et al ⁽¹⁵⁾ reported the appearance of bifidobacteria delayed until the third week of life in infants with low birth weight. However, all of the infants underwent colonization during the first month of life in these studies conducted more than 20 years ago. Recent studies in industrialized countries reported delayed gut colonization in newborns compared with studies conducted in the 1980s, a phenomenon likely related to deliveries with better hygienic procedures, such as wide use of vaginal antiseptics or per-partum antibiotic prophylaxis ⁽¹⁶⁾. In the study of Gewolb and colleagues in 1999 ⁽¹⁷⁾, only 1 of 29 preterm infants born at a mean gestational age of 26.2 weeks was colonized by bifidobacteria at 1 month of life. This delay was confirmed in the study of Schwiertz et al, who analyzed gut bacterial succession during the first month of life using a molecular method, denaturing gradient gel electrophoresis ⁽¹⁸⁾. In this latter study no bands in the denaturing gradient gel electrophoresis gels could be attributed to bifidobacteria in the 29 preterm infants delivered between weeks 24 and 37. In a recent study investigating the effects of a bifidobacterial supplementation, all 32 preterm infants in the placebo group were colonized by bifidobacteria during the first week of life at a mean level of 4.82 log₁₀ CFU/g ⁽¹⁹⁾. However, no indication on their gestational age was given. In the present study 66% of the preterm infants still had no colonization at 18.3 ± 7.4 days.

Little is known about bifidobacterial diversity in infants, and no information was available concerning the premature infants. In recent studies in full-term infants B longum is the most commonly found species ^(20–22). B bifidum and B breve isolation was almost as frequent as that of B longum, which was not the case in the other study. Interestingly, we found B animalis subspecies lactis in 4 infants, which is a species rarely described in human fecal microbiota and never described in feces of newborns. This species, which exhibits an elevated oxygen tolerance, has been isolated from fermented milk, and some strains of this species are applied in probiotic dairy products, food supplements, and pharmaceutical preparations ⁽²³⁾. This fact can explain the presence of this species in the surroundings of the preterm infants. Moreover, B animalis subspecies lactis was isolated only in infants from the same neonatal intensive care unit.

Favier et al, studying the succession of bifidobacterial colonization in 5 full-term infants, observed that primary colonization by a combination of B longum subspecies infantis and B pseudocatenulatum leads to a stable bifidobacterial population, in contrast with the combination of B breve and B scardovii⁽²⁴⁾. We did not observe such consequences of the first colonizing species, as in the 9 infants with several fecal samples with bifidobacteria, the bifidobacterial composition was stable except in 1 infant.

Factors influencing gut microbiota colonization are still elusive. Our study strongly suggests that bifidobacterial implantation may depend, at least in part, on gestational age, according to the following 2 observations. First, bifidobacterial implantation occurred always after a corrected gestational age of circa 34 weeks, and second, a birth at a gestational age of less than circa 33 weeks seems to seriously impair the ability of the gut to be colonized. Gordon et al recently provided new insights into the commensal bacteria–host relationships in the intestine: gnotobiotic models suggest that gut epithelial glycans laid down during the perinatal period act as sign posts that direct early colonizers to a particular place along the developing intestine ⁽²⁵⁾. These early colonizers are also able to direct synthesis of epithelial glycans by regulating expression of glycosyltransferases. In that respect, our data support the ideas that a sufficient developmental maturation of the gut is mandatory for bifidobacterial implantation and that a premature birth is likely to durably impair this colonization, likely through an alteration of the relationship between bifidobacteria and intestinal cells required for their implantation.

Bifidobacteria, normally dominant in full-term infants' microbiota, is thought to be responsible for many beneficial effects, such as resistance to pathogen colonization and stimulation of the gut associated lymphoid tissue. In particular, bifidobacterial isolates were shown to inhibit pathogens in vitro ⁽²⁶⁾ and in vivo ⁽⁶⁾, noticeably clostridia, which are involved in NEC ⁽²⁷⁾. Thus, the lack of bifidobacterial colonization may predispose extremely preterm infants to significant changes in their gut colonization pattern, making them more susceptible to gut infection and/or to diseases such as NEC, the incidence of which increases with decreasing gestational age. Strategies aiming at reducing NEC during prematurity through manipulation of gut microbiota are appealing, and should probably take into account this impaired bifidobacterial colonization in children born before 33 weeks of gestation.