Impact of β-lactam antibiotics and the delivery method on Bidobacteria and Bacteroides populations in early infancy: A Japanese cohort study

Background Intestinal ora changes dynamically in early infancy. Association between colonisation of Bidobacterium and Bacteroides, the dominant bacterial genera in the intestinal ora 6 months after birth, and development of intestinal immunity has been suggested. Antibiotic administration to the mother is an important factor in this colonisation, but few studies have considered antibiotics taken immediately before delivery, including before Caesarean section. The association of the gut ora with the delivery method is also uncertain. Here, we performed a cohort study to determine the inuence of antibiotics taken immediately before delivery on the intestinal ora in early infancy, with a focus on Bidobacterium and Bacteroides, in 130 healthy Japanese infants. Faecal samples were collected from infants at 1, 3, and 6 months after birth (383 samples) in a hospital in Japan. The samples were analysed using a next-generation sequencer. Two β-lactam antibiotics were administered to mothers immediately before delivery: Cefazolin was used before Caesarean sections and ampicillin was used for cases with premature rupture of the membranes and in Group B Streptococcus positive cases. At all ages, Bidobacterium and Bacteroides were dominant, and the mean combined occupancy was 60–70%. Bidobacterium was the most dominant at all ages but its abundance was signicantly lower in infants who were exposed to antibiotics at delivery, and 1 and 3 months after birth, whereas no difference was found between the delivery methods. Bacteroides were the second or third dominant bacteria at all ages but their abundance was signicantly lower after Caesarean section compared with that after vaginal delivery, whereas there was no difference due to antibiotic exposure. These effects were conrmed in a linear mixed-effect model. In addition, occupancy by Bidobacterium at 1 and 3 months and by Bacteroides at 3 months differed between infants with and without siblings.


Background
The human body is inhabited by 100-1000 trillion bacteria in the oral cavity, skin, and intestine, in uencing the biological health of the host. [1] , [2] There are large differences in the composition of human intestinal ora depending on race, country, and lifestyle. [3] , [4] , [5] Although Bi dobacteria population in intestinal ora in infancy is known to affect Japanese children in the subsequent school period, studies on the intestinal ora of Japanese infants are very few. [6] , [7] , [5] , [8] , [9] , [10] Many studies have indicated an association between intestinal ora and disease. [11,12] , [13] Colonisation by intestinal bacteria in early infancy is known to have a major effect on intestinal immunity [14] , [15] , [16] and changes in the intestinal ora. Especially the dominant bacteria, Bi dobacteria and Bacteroides [17] , [18] , [19] , [20] , [21] , [22] , [23] resulting from the start of food ingestion at 6 months after birth, have been linked to subsequent development of diseases, such as allergies. [24] , [25] , [26] Various factors are known to in uence the bacterial colonisation, including maternal exposure to antibiotics administered as intrapartum antimicrobial prophylaxis (IAP) in Group B Streptococcus (GBS)-positive mothers [27] , [28] , [29] and those administered in the late perinatal period. [30] These studies, however, have reported differences in the administration and screening period, thereby resulting in the effects being inconclusive. The in uence of the delivery method on the Bi dobacteria and Bacteroides populations in infants has also been extensively studied. [31] , [32] , [16] , [33] It is a widely accepted hypothesis that since Caesarean section does not require the infant to pass through the birth canal, the infant does not come in contact with the Japanese children in early infancy, and especially on the dominant bacterial genera, Bi dobacteria and Bacteroides, samples were collected from infants until 6 months after birth, and the in uence of AED and other factors in the infant population were investigated.

Subjects
This prospective cohort study was performed as a part of a study investigating the association of allergic diseases with time-course changes observed in the microbiome in infants. The subjects were 142 infants and their mothers, who gave consent to registration before delivery or 2 weeks after delivery at Iwate Prefectural Iwai Hospital between February 2018 and March 2019. This hospital is in the Tohoku (Northeast) region of Japan and handles about 800 deliveries each year as a core perinatal medical care centre in the southern Iwate prefecture.
Inclusion criteria were as follows: Infants born by full-term natural delivery or Caesarean section, whose mothers had not been exposed to antibiotics for one month before delivery except for the antimicrobial use just before the delivery. Faecal samples were collected at 1, 3, and 6 months after birth by their parents.
Exclusion criteria were as follows: Premature babies born before 37 weeks of gestation; the samples were collected at 6 months after birth from infants, whose mothers had received antibiotics, the samples collected were handled as dropouts. If the sample collection was missed, the infant was considered as a dropout at that age and after that. If a sample was inappropriate for MiSeq analysis and could not be analysed, data for this infant was excluded from the analysis only at this time-point, while the analytical results for the infant at all other ages were used.
During IAP at delivery, a single dose of cefazolin (CEZ) (1 g) was given systemically to mothers for all Caesarean sections just before the surgery. Ampicillin (ABPC) (2 g) was given at least 4 h before delivery, followed by intermittent administration every 6 h until delivery in GBS-positive and PROM cases. Antibiotics were administered to all subjects at the dose and time de ned in the clinical protocol determined by the hospital board.
Containers for collection of faecal samples at 1 month after birth were handed to mothers during hospitalisation for delivery or at the health examination of infants 2 weeks after delivery. The container for the collection of faecal samples at 3 months was sent by mail to the address of each registered infant from Core Technology Laboratories, Asahi Group Holdings, and the container for 6 months was sent by mail from the Department of Microbiome Research, Juntendo University. Each faecal sample collected by the infants' parents was transferred to a test tube (Techno Suruga Laboratory, Shizuoka, Japan) containing 0.001% bromothymol and 100 mM Tris-HCl (pH 9), 40 mM EDTA, 4 M guanidine thiocyanate, and was mixed well as described in a previous study. [44] Mixed faecal samples were delivered to a laboratory of Asahi Group Holdings (Sagamihara, Kanagawa, Japan) and were stored at − 80 °C until processing for DNA extraction.

DNA extraction
The processed samples were subjected to DNA extraction, as described previously. [43] Brie y, the samples (2 mL) were transferred to plastic tubes, centrifuged at 14,000 × g for 3 min, washed in 1.0 mL of phosphate-buffered saline, and centrifuged at 14,000 × g. Pellets were re-suspended in 500 µL of extraction buffer (166 mM Tris/HCl, 66 mM EDTA, 8.3% sodium dodecyl sulphate, pH 9.0) and 500 µL of TE buffer-saturated phenol. Next, 300 mg of zirconium beads (0.1 mm diameter) was added to the suspension, and the mixture was vortexed vigorously for 60 s × 3 times using a Multi-Beads Shocker (Yasui Kikai Corp., Osaka, Japan). After centrifugation at 14,000 × g for 5 min, 400 µL of the supernatant was puri ed using a Maxwell Instrument (Promega KK, Tokyo, Japan).
Sequencing and data processing 16S rRNA gene sequencing was performed using a MiSeq V3 kit as per the manufacturer's protocol (Illumina, CA, USA). Brie y, the V3-V4 region of the bacterial 16S rDNA was ampli ed by PCR with forward and reverse primers (5'-TCG GCA GCG TCA GAT GTG TAT AAG AGA CAG CCT ACG GGA GGC WGC AG-3' and 5'-GTC TCG TGG GCT CGG AGA TGT GTA TAA GAG ACA GGA CTA CHV GGG TAT CTA ATC C-3'), using the TaKaRa Ex Taq HS Kit (TaKaRa Bio, Shiga, Japan) to amplify 5 ng of DNA from a faecal sample. After the PCR products were puri ed with Agencourt AMPure XP (Beckman Coulter, CA, USA), the products were ampli ed using a Nextera XT Index Kit v2 (Illumina, CA, USA). After the second round of PCR, the products were again puri ed using Agencourt AMPure XP. The library was quanti ed, normalised, and pooled in equimolar amounts. Sequencing was conducted using a paired-end 2 × 300-bp cycle run on an Illumina MiSeq system with a MiSeq Reagent Kit v.3 (600 cycles).
16S rRNA-based taxonomic and diversity analysis QIIME2 (Quantitative Insights into Microbial Ecology, http://qiime2.org/) v.2019.4.0. was used for the analysis of the sequences [45]. The quality of the sequences was checked and ltered using QIIME2 plugin DADA2 [46], and chimeric sequences were removed. The primers were trimmed, and the remaining forward and reverse sequences were truncated to a nal length of 280 bp. As the reference sequence, "gg-13-8-99-nb-classi er.qza" from the greengenes database was used, and operational taxonomic unit were identi ed at the genus level. A phylogenetic tree was then created using FastTree [47], and the beta diversity distance was estimated. Beta diversity was visualised by principal coordinate analysis.

Data collection
The following data were collected from medical records at Iwate Prefectural Iwai Hospital: Delivery method, gender, body weight at birth, perinatal history, records of hospital visits and treatments received by the infant up to 6 months after birth including the use of antibiotics after birth. Additional information related to age (days) at sample collection, feeding method (exclusive breastfeed or added top feed), and siblings were obtained from a questionnaire completed by the mothers. The following data for mothers were also collected from medical records and a questionnaire: Age, delivery method, history of allergies (food allergy, bronchial asthma, atopic dermatitis, allergic rhinitis), abnormal ndings at delivery (including PROM and GBS-positive status), antimicrobial use during late pregnancy, and systemic antibiotics (including types) given at delivery.

Statistical analysis
The signi cance of the difference between the two groups was analysed using a non-parametric ANOSIM (analysis of similarities) test based on unweighted UniFrac distances within QIIME2 (https://qiime2.org/). Acquired 16S rRNA gene data for bacteria were analysed using SAS® 9.4 (SAS Institute Inc., Cary, NC, USA). Background factors of the mother and child were compared using the Mann-Whitney U-test for continuous variables and the Pearson's chi-square test for categorical variables. The in uence of each background factor on occupancies by higher-rank dominant bacterial genera in intestinal ora at each age was examined by logistic regression analysis. The dependence of occupancy of each bacterial genus on factors, for which a signi cant association was found in diversity analysis and logistic regression analysis, was examined by Mann-Whitney U-test for between-group comparison and by Kruskal-Wallis test with a Bonferroni correction for multiple group comparison. Continuous comparative changes of dominant bacterial genera due to different factors were analysed using a linear mixed-effect model (random intercept and rst-order autoregression model). The signi cance level was set at p < 0.05 in all analyses.

Results
Subject data A total of 142 mother and infant pairs were registered in the study, and 424 samples were obtained. Among these, 130, 127, and 126 samples collected at 1, 3, and 6 months, respectively, adhered to the inclusion criteria of the study. Exclusion of subjects occurred due to the following reasons: Three premature babies dropped out by 1 month of age, 1 infant received antibiotic administration for fever in the neonatal period, and there were eight cases with non-analysable samples at 1 month. While dropouts from 1 to 3 months were due to antibiotic administration in eight cases, no samples were received in one case, and a nonanalysable sample was received at 3 months in one case, dropouts from 3 to 6 months were due to antibiotic administration in one case and a non-analysable sample at 6 months in one case.
The background information of the mothers and infants at 1 month, which were in uencing factors, are shown in Table 1. Since the dropout cases resulted in no major change in the mean value of the background factors, the data at 3 and 6 months are shown in Supplementary Table S1. Antibiotics were used immediately before delivery in about 55% of cases at each age. Caesarean section, GBS-positive, and PROM cases accounted for about 20%, 15%, and 15% of all cases at each age, respectively. The PROM cases included one emergency and 6 GBS-positive Caesarean sections. These seven cases were included in the Caesarean section group because CEZ was administered as an antibiotic. In 5 PROM cases, delivery rapidly progressed, and no antibiotic was used. Of the infants born by Caesarean section, the Apgar score at birth was low in two cases, oxygen was administered after birth for mild neonatal respiratory disorder in one case, phototherapy was performed for neonatal jaundice in one case, the mothers had diabetes in two cases, the mothers had hyperthyroidism in two cases, and one infant was admitted for respiratory syncytial virus (RSV) infection during the observation period and discharged after symptomatic treatment.
All these infants were con rmed to be healthy via a health examination at each age by physicians of the paediatrics and neonatology department.
The top 20 bacterial genera constituting intestinal ora at each age are shown in Fig. 1. At 1 month, based on the occupancy, Bi dobacteria population was found to be overwhelmingly dominant (49.7% ± 34.1%), while the Bacteroides population was found to be the third most dominant (7.7% ± 12.6%), but the occupancy was almost equivalent to the second most dominant bacteria, Streptococcus (7.8% ± 12.6%). Bi dobacteria population was also the most dominant at 3 months (61.7% ± 28.0%), while the Bacteroides population was the second most dominant (6.5% ± 10.9%). The Bi dobacteria population continued to be the most dominant at 6 months (66.2% ± 21.6%), followed by the Bacteroides population (5.7% ± 9.2%).  The bacterial genera are shown from left to right in the order of higher occupancy (mean). The occupancy was classi ed into two groups based on the median for each genus, and the odds ratio and 95% con dence interval were calculated by logistic regression analysis. The signi cance level was set at 5%. *p < 0.05, **p < 0.01, ***p < 0.001.

Analysis at three months
The effects of background factors on the ve most dominant bacterial genera and other high-ranking genera in the intestinal ora of 3-month-old infants are shown in Supplementary Table S2. Bi dobacteria occupancy was signi cantly dependent on the exposure to AED (non-AED: OR, 0.3; 95% CI, 0.09-0.9) and the existence of siblings (no sibling: OR, 3.73; 95% CI, 1.1-12.6), similar to that seen at 1 month (p < 0.05), whereas that of Bacteroides was signi cantly dependent on the delivery method (vaginal delivery: OR, 0.14; 95% CI, 0.03-0.63), also similar to that at 1 month (p < 0.05). The fth dominant genus, Ruminococcus (2.9% ± 8.5%), showed a signi cant dependence on feeding methods (exclusive breastfeeding: OR, 2.5; 95% CI, 1.13-5.62).

Analysis at six months
The effects of background factors on the ve most dominant bacterial genera in the intestinal ora in 6-month-old infants and on other high-ranking genera are shown in Supplementary

Effects of AED
Infants at each age were divided based on the AED status and delivery method, and background factors of the mother and child were compared. The results are shown in Supplementary Table S3. In infants born via Caesarean section, the existence of siblings and age of the mother were signi cantly higher due to the in uence of the previous Caesarean section, and the gestational age and birth weight were signi cantly lower because the date of Caesarean delivery was decided beforehand unless performed as an emergency. This tendency also existed between the two types of delivery methods in the AED group, but a subgroup analysis of the vaginal delivery group, excluding the in uence of Caesarean section, showed no difference in background factors between the AED and non-AED groups. The effects of background factors that in uenced the dominant bacterial genera at each age (AED, delivery method, siblings) and the occupancies by Bi dobacteria and Bacteroides were then analysed.
AED had a signi cant effect on the diversity of overall intestinal ora at 1 and 3 months (Fig. 2a). Bi dobacteria occupancy was signi cantly lower in AED cases than in non-AED cases in 1-month-old infants, regardless of the use of ABPC or CEZ (p < 0.001). In contrast, in 3-month-old infants, occupancy was not affected by AED. Bacteroides population was markedly lower in the CEZ group at both 1 and 3 months compared with that of the non-AED group (p < 0.001) and this tendency was also noted in the ABPC group. A signi cant difference was also found between AED and the non-AED groups (p < 0.05) (Fig. 2b).
In a sub-group analysis of AED in vaginal delivery cases without CEZ administration (i.e., excluding infants born by Caesarean section), there was a signi cant difference in diversity (p = 0.03) (Fig. 3a). Bi dobacteria occupancy in 1-month-old infants was signi cantly lower in the AED group (all were included in the ABPC group) (p < 0.001), and a signi cant difference was also noted at 3 months (p < 0.05). In contrast, occupancy of Bacteroides did not differ between these two groups (Fig. 3b). In the AED group, the exposure to antibiotics did not affect the Bi dobacteria and Bacteroides populations in the PROM and GBS-positive groups (Supplementary Figure S1).

Effects of the delivery method
The delivery method signi cantly in uenced the diversity of the intestinal ora at 1 month (Fig. 4a). The occupancy of Bi dobacteria did not differ with age, whereas that of Bacteroides was signi cantly lower in 1-and 3-month-old infants born via Caesarean section (p < 0.001) (Fig. 4b). A comparison of delivery methods within the AED group gave similar ndings (Fig. 5a,b).

Effects of siblings
The presence of siblings signi cantly changed the diversity of the intestinal ora at 1 and 3 months (Fig. 6a). Bi dobacteria occupancy was signi cantly higher in 1-month-old (p = 0.001) and 3-month-old (p < 0.001) infants with siblings. Occupancy of Bacteroides did not differ at 1 month but was signi cantly lower in infants with siblings at 3 months (p < 0.05). At 6 months, there was no signi cant difference in occupancy for either genus (Fig. 6b). Sub-group analysis within the AED group also showed a signi cant change in the diversity of the intestinal ora at 1 and 3 months (Fig. 5a), and bi dobacterial occupancy was signi cantly higher in 1-and 3-month-old infants with siblings (p < 0.01 and p < 0.001, respectively) (Fig. 5b).

Time-course changes in Bi dobacteria and Bacteroides (linear mixed-effect model)
Comparison of bi dobacterial occupancy in AED and non-AED infants using a linear mixed-effects model (Fig. 7a) showed a signi cant difference over 6 months after birth, with lower occupancy in the AED group, but the difference in occupancy decreased with time. There was, however, no difference seen due to the delivery method. Occupancy appeared to be higher in infants with siblings, but there was no signi cant difference by the 6th month. Occupancy of Bacteroides did not differ signi cantly between the AED and non-AED groups (Fig. 7b), but the occupancy was lower in the Caesarean section group and infants with siblings.

Discussion
The results of this study indicated that the diversity of intestinal ora was in uenced by AED, delivery method, and siblings, with signi cant effects on Bi dobacteria and Bacteroides populations, which remained dominant at a combined occupancy of 60-70% in the bacterial ora of infants aged up to 6 months. These ndings con rm the results of an earlier pilot study [43] in 1-month-old infants.
AED to β-lactamase antibiotics has a major in uence on Bi dobacteria population in early infants regardless of the use of ABPC or CEZ with the in uence being especially marked in 1-month-old infants. The in uence of ABPC persisted until 3 months, but then gradually weakened and mostly disappeared by the 6th month. These results also comply with those of the pilot study [43]. Several previous studies have suggested a minor effect of IAP on the Bi dobacteria occupancy. However, these studies were limited to use of antibiotics for speci c reasons, such as for GBS-positive mothers [48] , [49] or in the late stage of delivery [30], and the study design and screening timeline of the intestinal ora were inconsistent. A study on the antibiotic administration, including Caesarean section, has been reported [49]. However, to the best of our knowledge, the present study is the rst of its kind to evaluate the effects of antibiotic administration to mothers immediately before delivery, including the cases of Caesarean section. The study comprised of a statistically appropriate number of samples, thereby allowing sub-group analyses and the effects were evaluated until 6 months after birth using a 16S rRNA-targeting next-generation sequencer.
The delivery method (vaginal vs Caesarean section) did not affect the Bi dobacteria occupancy in 1-month-old infants, but there was a signi cant difference in the CEZ and non-AED groups. The delivery method did not show any difference in the effect on the Bi dobacteria population between the ABPC and CEZ groups. Thus, Bi dobacteria occupancy was signi cantly in uenced in the Caesarean section and CEZ groups compared to non-AED infants, but this effect did not change by AED in the vaginal delivery group ( i.e., the ABPC group). These results suggest that β-lactam antibiotics may directly in uence bi dobacterial occupancy. In uence of delivery method on Bi dobacteria population from immediately after birth to early infancy was shown previously, [39,[50][51][52][53] but the in uence of antibiotics administered before Caesarean section [36,37,54] was not considered in these studies, and this effect may have been simultaneously observed. The present study is not capable of judging whether the observed effect is due to the infant not passing through the birth canal in Caesarean section or CEZ. Further studies with the use of the same antibiotics for vaginal delivery and Caesarean section are therefore warranted.
Bi dobacteria colonisation was also in uenced by the presence or absence of siblings. Bi dobacteria occupancy in AED infants was signi cantly higher in those with siblings. The results suggested that at least until 3 months after birth, the presence of an elder sibling promoted the colonisation of Bi dobacteria, even in infants exposed to antibiotics at delivery, i.e., there may be mutual interference of the microbiome between siblings. This may explain the maintenance of high Bi dobacteria occupancy in infants, even in the AED group. In a previous pilot study, [43] the presence of siblings was suggested to in uence IAP, and the present study con rmed this effect. Previous studies on the effects of siblings have reported that Bi dobacteria colonisation occurs more easily in infants with siblings, [19,39,40]. However, this effect has been scarcely studied as compared to studies on factors such as AED and the delivery method, particularly in Japanese infants. Thus, the present study is signi cant in showing the effect of siblings in a large cohort study. This effect was con rmed by a sub-group analysis within the AED group to allow the interpretation in the context of AED. The association of this effect with the intestinal ora of siblings requires a continuous study, including siblings living together.
The in uence of Caesarean section was more substantial than that of AED in Bacteroides. This effect was rmly maintained at 3 months and persisted at least until 6 months after birth. The same tendency was observed in the sub-analysis of the AED group. This con rmed that birth via Caesarean section is an important factor in the occupancy of Bacteroides, compared to AED. The persistence of the in uence of Caesarean section on Bacteroides until the weaning period and thereafter has been pointed out, [34, 39, 53, 55, 56] similar to our results. However, as described above, all Caesarean section cases received preoperative CEZ. Thus, although the Bacteroides population was not in uenced by ABPC, it may have been markedly in uenced by CEZ. This possibility is outstanding and was not examined in previous studies.
Bacteroides population was not affected by siblings at 1 month, but a signi cant effect was seen at 3 months, with no in uence at 6 months. In contrast to Bi dobacteria, siblings negatively in uenced the occupancy of Bacteroides. The in uence of siblings on Bacteroides population has been shown before, [39] with the occupancy by Bacteroides at 18 months after birth being higher in infants with siblings, in contrast to the ndings of this study until 6 months. [57] Similar to Bi dobacteria, the effect on Bacteroides until 3 months may re ect mutual interference among siblings.
Many studies have examined the clinical signi cance of colonisation with Bi dobacteria and Bacteroides in intestinal ora in early infancy. [17][18][19][20][21][22] The present study evaluated the in uence of AED on early infants in terms of the effects on their subsequent health. The samples from this study were also used to assess the relation of allergies with changes in the intestinal ora. This study was performed as a part of a cohort study evaluating the clinical signi cance of changes in the intestinal ora during early infancy in healthy Japanese infants and their effect on the intestinal ora later in life. Data at 1, 3, and 6 months after birth were used in this study, but data related to longer-term time-course changes in the same individuals are needed for a complete and comprehensive investigation. The composition of intestinal ora shows substantial similarities at different time-points in the same individual, [58] but external factors in uencing the intestinal ora increase with growth, such as the increase in baby food intake, interaction with siblings and other infants in group nursing, and further use of antibiotics for various diseases. However, this study is signi cant as a cohort study with a statistically signi cant large number of samples performed on intestinal ora in infants in the rst 6 months of life, which is a crucial period for the development of the immune system since this is a time at which external factors have the least in uence in the entire lifetime [24][25][26].
The present study is not without limitations despite the undeniable signi cance. This study was conducted in a single hospital, which does not allow for extrapolation of the obtained results to the population from other regions within and without Japan, with differences in ethnicity or geography. Although the protocol for the use of antimicrobial agents before the delivery was similar to that in previous studies on IAP [27] [29], multi-centre studies are needed for validation of such extrapolation. Also, while this study focused on several factors that could have a signi cant impact on the gut microbiota of infants, it did not account for some factors such as feeding methods (only exclusively breast-fed infants included), which could affect colonisation characteristics of some bacteria in the intestinal microbiota of infants. Inclusion of another group of infants that were milk-fed for the most part along with top feed is imperative to understand the factors in uencing gut microbiota in infants.

Conclusions
This prospective cohort study con rmed the ndings from a previous pilot study indicating that β-lactam antibiotics administered to the mother immediately before delivery have a signi cant in uence on the intestinal ora of healthy Japanese infants at 1 and 3 months after birth. This effect is more important than the effect of the delivery method for Bi dobacterium, the dominant bacterial genus. The evaluation of the in uence of AED requires the inclusion of all antibiotics used immediately before delivery, including before the Caesarean section. The presence of siblings also affects Bi dobacteria colonisation, and this effect persists until 3 months and increases with time. In contrast, for Bacteroides, the in uence of the delivery method is greater than that of AED. However, it is unclear whether this is an in uence of delivery via Caesarean section or CEZ administration. These results provide a new perspective on essential factors in uencing the intestinal ora in early infancy, which is vital for the development of intestinal immunity. The clinical signi cance of the results in the later life of the infants requires further long-term studies.

Declarations
Funding None.

Consent for publications
All mothers agreed to publication of the data and written informed consent was obtained for this purpose. Aoyagi, Kano, and Morita performed storage, processing, and analysis using MiSeq of samples. Imoto, Aoyagi, Kano, and Nojiri performed data management and statistical analysis.
Imoto drafted the manuscript.
Watanabe, Hashiguchi, Amanuma, and Maruyama corrected and reviewed the manuscript.
All authors read and approved the content of the manuscript, including the accuracy of the data, ethical legitimacy, and validity of the results. Figure 1 Mean       Time-course changes in bi dobacterial (a) and Bacteroides (b) occupancies (from 1 to 6 months after birth) based on AED or non-AED, delivery method, and the presence or absence of siblings, using a linear mixed-effect model. The analysis set at 1 month included 130 infants. Dropouts at 3 and 6 months were handled as missing values.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.