[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.
Lifetime Antibiotic Use in Relation to Duration of Breastfeeding
Lifetime Antibiotic Use in Relation to Duration of Breastfeeding

Lifetime antibiotic use in relation to the duration of breastfeeding in 113 children with and 113 children without antibiotic use during breastfeeding or within 4 months after weaning. Shaded areas indicate the 95% confidence interval of each trend line.

Figure 2.
Body Mass Index (BMI) z Score in Relation to Duration of Breastfeeding
Body Mass Index (BMI) z Score in Relation to Duration of Breastfeeding

The BMI (calculated as weight in kilograms divided by height in meters squared) z score at ages 2 to 6 years in relation to the duration of breastfeeding in 113 children with and 113 children without antibiotic use during breastfeeding or within 4 months after weaning. Shaded areas indicate the 95% confidence interval of each trend line.

Figure 3.
Global Intestinal Microbiota Composition in the Subcohort of 42 Children With Fecal Microbiota Composition Analysis
Global Intestinal Microbiota Composition in the Subcohort of 42 Children With Fecal Microbiota Composition Analysis

A, Principal coordinates analysis of the intestinal microbiota with Bray-Curtis dissimilarity, categorized by the duration of breastfeeding (short, 0-6 months; long, 8-16 months) and antibiotic use during breastfeeding or shortly after weaning (no early-life antibiotic use, AB−; early-life antibiotic use, AB+). Each circle is the microbiota of a child and the number in each circle indicates the age (in years) of the child at the time of fecal sample collection. B, Box and whisker plot of Actinobacteria to Firmicutes relative abundance ratio by group. For the taxa, the unit is the number of DNA sequencing reads assigned to that taxon relative to the total number of reads for the sample; the ratio is therefore computed as the following: (Actinobacteria reads/total reads)/(Firmicutes reads/total reads). The horizontal line in the middle of each box indicates median; top and bottom borders of each box, 75th and 25th percentiles, respectively; whiskers above and below each box, maximally 1.5 times the length of the box; and points beyond the whiskers, outliers beyond 1.5 times the length of the box.

aP < .001.

bP < .01.

Figure 4.
Differences in Abundance of the Intestinal Microbiota Associated With Breastfeeding and Antibiotic Use
Differences in Abundance of the Intestinal Microbiota Associated With Breastfeeding and Antibiotic Use

Significant differences between the mean relative abundance of the indicated genera (phyla shown at the top of panels) in the group with long breastfeeding duration (8-16 months) with early-life antibiotic use (long/AB+) and the group with short breastfeeding duration (0-6 months) and no early-life antibiotic use (short/AB−) compared with the group with long breastfeeding duration and no early-life antibiotic use (long/AB−). Error bars indicate standard error.

aFalse discovery rate–corrected P < .10, based on negative binomial models.

Table.  
Characteristics of the Cohort
Characteristics of the Cohort
1.
Hörnell  A, Lagström  H, Lande  B, Thorsdottir  I.  Breastfeeding, introduction of other foods and effects on health: a systematic literature review for the 5th Nordic Nutrition Recommendations.  Food Nutr Res. 2013;57.PubMed
2.
Woo  JG, Martin  LJ.  Does breastfeeding protect against childhood obesity? moving beyond observational evidence.  Curr Obes Rep. 2015;4(2):207-216.PubMedArticle
3.
Penders  J, Thijs  C, Vink  C,  et al.  Factors influencing the composition of the intestinal microbiota in early infancy.  Pediatrics. 2006;118(2):511-521.PubMedArticle
4.
Ward  RE, Niñonuevo  M, Mills  DA, Lebrilla  CB, German  JB.  In vitro fermentation of breast milk oligosaccharides by Bifidobacterium infantis and Lactobacillus gasseri Appl Environ Microbiol. 2006;72(6):4497-4499.PubMedArticle
5.
Bäckhed  F, Roswall  J, Peng  Y,  et al.  Dynamics and stabilization of the human gut microbiome during the first year of life.  Cell Host Microbe. 2015;17(5):690-703.PubMedArticle
6.
Ridaura  VK, Faith  JJ, Rey  FE,  et al.  Gut microbiota from twins discordant for obesity modulate metabolism in mice.  Science. 2013;341(6150):1241214.PubMedArticle
7.
Cox  LM, Yamanishi  S, Sohn  J,  et al.  Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences.  Cell. 2014;158(4):705-721.PubMedArticle
8.
Cox  LM, Blaser  MJ.  Pathways in microbe-induced obesity.  Cell Metab. 2013;17(6):883-894.PubMedArticle
9.
Kalliomäki  M, Collado  MC, Salminen  S, Isolauri  E.  Early differences in fecal microbiota composition in children may predict overweight.  Am J Clin Nutr. 2008;87(3):534-538.PubMed
10.
Dogra  S, Sakwinska  O, Soh  SE,  et al; GUSTO Study Group.  Dynamics of infant gut microbiota are influenced by delivery mode and gestational duration and are associated with subsequent adiposity.  MBio. 2015;6(1):e02419-14.PubMedArticle
11.
Saari  A, Virta  LJ, Sankilampi  U, Dunkel  L, Saxen  H.  Antibiotic exposure in infancy and risk of being overweight in the first 24 months of life.  Pediatrics. 2015;135(4):617-626.PubMedArticle
12.
Grönlund  MM, Arvilommi  H, Kero  P, Lehtonen  OP, Isolauri  E.  Importance of intestinal colonisation in the maturation of humoral immunity in early infancy: a prospective follow up study of healthy infants aged 0-6 months.  Arch Dis Child Fetal Neonatal Ed. 2000;83(3):F186-F192.PubMedArticle
13.
Sjögren  YM, Tomicic  S, Lundberg  A,  et al.  Influence of early gut microbiota on the maturation of childhood mucosal and systemic immune responses.  Clin Exp Allergy. 2009;39(12):1842-1851.PubMedArticle
14.
Kumpu  M, Kekkonen  RA, Kautiainen  H,  et al.  Milk containing probiotic Lactobacillus rhamnosus GG and respiratory illness in children: a randomized, double-blind, placebo-controlled trial.  Eur J Clin Nutr. 2012;66(9):1020-1023.PubMedArticle
15.
Korpela  K, Salonen  A, Virta  LJ,  et al.  Intestinal microbiome is related to lifetime antibiotic use in Finnish pre-school children.  Nat Commun. 2016;7:10410.PubMedArticle
16.
Edgar  RC, Haas  BJ, Clemente  JC, Quince  C, Knight  R.  UCHIME improves sensitivity and speed of chimera detection.  Bioinformatics. 2011;27(16):2194-2200.PubMedArticle
17.
Caporaso  JG, Kuczynski  J, Stombaugh  J,  et al.  QIIME allows analysis of high-throughput community sequencing data.  Nat Methods. 2010;7(5):335-336.PubMedArticle
18.
Korpela  K, Flint  HJ, Johnstone  AM,  et al.  Gut microbiota signatures predict host and microbiota responses to dietary interventions in obese individuals.  PLoS One. 2014;9(6):e90702.PubMedArticle
19.
Venables  W, Ripley  B.  Modern Applied Statistics With S. 4th ed. New York, NY: Springer; 2002.
20.
Oksanen  J, Blanchet  FG, Kindt  R,  et al.  vegan: Community Ecology Package: R Package Version 2.0-6. Vienna, Austria: R Foundation for Statistical Computing; 2013.
21.
Owen  CG, Martin  RM, Whincup  PH, Davey-Smith  G, Gillman  MW, Cook  DG.  The effect of breastfeeding on mean body mass index throughout life: a quantitative review of published and unpublished observational evidence.  Am J Clin Nutr. 2005;82(6):1298-1307.PubMed
22.
Brown  A, Lee  M.  Breastfeeding during the first year promotes satiety responsiveness in children aged 18-24 months.  Pediatr Obes. 2012;7(5):382-390.PubMedArticle
23.
Cani  PD, Neyrinck  AM, Fava  F,  et al.  Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia.  Diabetologia. 2007;50(11):2374-2383.PubMedArticle
24.
Lin  HV, Frassetto  A, Kowalik  EJ  Jr,  et al.  Butyrate and propionate protect against diet-induced obesity and regulate gut hormones via free fatty acid receptor 3-independent mechanisms.  PLoS One. 2012;7(4):e35240.PubMedArticle
25.
Remely  M, Aumueller  E, Merold  C,  et al.  Effects of short chain fatty acid producing bacteria on epigenetic regulation of FFAR3 in type 2 diabetes and obesity.  Gene. 2014;537(1):85-92.PubMedArticle
26.
Fiorucci  S, Mencarelli  A, Palladino  G, Cipriani  S.  Bile-acid-activated receptors: targeting TGR5 and farnesoid-X-receptor in lipid and glucose disorders.  Trends Pharmacol Sci. 2009;30(11):570-580.PubMedArticle
27.
Joyce  SA, MacSharry  J, Casey  PG,  et al.  Regulation of host weight gain and lipid metabolism by bacterial bile acid modification in the gut.  Proc Natl Acad Sci U S A. 2014;111(20):7421-7426.PubMedArticle
28.
Zivkovic  AM, German  JB, Lebrilla  CB, Mills  DA.  Human milk glycobiome and its impact on the infant gastrointestinal microbiota.  Proc Natl Acad Sci U S A. 2011;108(suppl 1):4653-4658.PubMedArticle
29.
Ottman  NA. Host Immunostimulation and Substrate Utilization of the Gut Symbiont Akkermansia muciniphila [PhD thesis]. Wageningen, the Netherlands: Wageningen University; 2015.
30.
Everard  A, Belzer  C, Geurts  L,  et al.  Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity.  Proc Natl Acad Sci U S A. 2013;110(22):9066-9071.PubMedArticle
31.
Safavi  M, Farajian  S, Kelishadi  R, Mirlohi  M, Hashemipour  M.  The effects of synbiotic supplementation on some cardio-metabolic risk factors in overweight and obese children: a randomized triple-masked controlled trial.  Int J Food Sci Nutr. 2013;64(6):687-693.PubMedArticle
32.
Santacruz  A, Collado  MC, García-Valdés  L,  et al.  Gut microbiota composition is associated with body weight, weight gain and biochemical parameters in pregnant women.  Br J Nutr. 2010;104(1):83-92.PubMedArticle
33.
Karlsson  CL, Onnerfält  J, Xu  J, Molin  G, Ahrné  S, Thorngren-Jerneck  K.  The microbiota of the gut in preschool children with normal and excessive body weight.  Obesity (Silver Spring). 2012;20(11):2257-2261.PubMedArticle
34.
Jones  BV, Begley  M, Hill  C, Gahan  CG, Marchesi  JR.  Functional and comparative metagenomic analysis of bile salt hydrolase activity in the human gut microbiome.  Proc Natl Acad Sci U S A. 2008;105(36):13580-13585.PubMedArticle
35.
Bergström  A, Skov  TH, Bahl  MI,  et al.  Establishment of intestinal microbiota during early life: a longitudinal, explorative study of a large cohort of Danish infants.  Appl Environ Microbiol. 2014;80(9):2889-2900.PubMedArticle
36.
Balamurugan  R, George  G, Kabeerdoss  J, Hepsiba  J, Chandragunasekaran  AM, Ramakrishna  BS.  Quantitative differences in intestinal Faecalibacterium prausnitzii in obese Indian children.  Br J Nutr. 2010;103(3):335-338.PubMedArticle
37.
Murphy  EF, Cotter  PD, Healy  S,  et al.  Composition and energy harvesting capacity of the gut microbiota: relationship to diet, obesity and time in mouse models.  Gut. 2010;59(12):1635-1642.PubMedArticle
Original Investigation
August 2016

Association of Early-Life Antibiotic Use and Protective Effects of BreastfeedingRole of the Intestinal Microbiota

Author Affiliations
  • 1Immunobiology Research Programme, Department of Bacteriology and Immunology, University of Helsinki, Helsinki, Finland
  • 2Research Department, Social Insurance Institution, Turku, Finland
  • 3Research and Development, Valio Ltd, Helsinki, Finland
  • 4Laboratory of Microbiology, Wageningen University, Wageningen, the Netherlands
JAMA Pediatr. 2016;170(8):750-757. doi:10.1001/jamapediatrics.2016.0585
Abstract

Importance  Long duration of breastfeeding is known to reduce the frequency of infections and the risk of overweight, both of which are prevalent health problems among children, but the mechanisms are unclear.

Objectives  To test whether early-life antibiotic use in children prevents the beneficial long-term effects of breastfeeding on weight development and lifetime antibiotic use, and to investigate whether the duration of breastfeeding is associated with long-term microbiota development.

Design, Setting, and Participants  Retrospective cohort study, conducted from June 2015 to December 2015, of the association between the duration of breastfeeding and lifetime antibiotic use by children as well as body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) z score in a cohort of 226 healthy children aged 2 to 6 years attending day care at the study area in northern Finland and participating in a probiotic trial from October 1, 2009, through April 30, 2010. Fecal microbiota composition analysis was performed in a subcohort of 42 of these children.

Exposures  Duration of breastfeeding and the number of different antibiotic courses purchased for the child.

Main Outcomes and Measures  The BMI z score, lifetime antibiotic use after weaning, and fecal microbiota composition.

Results  A total of 226 children (mean [SD] age, 55 [1.4] months; 54% male) were included in the study. Among the 113 children with no antibiotics before weaning, each month of breastfeeding decreased the mean number of postweaning antibiotic courses by 5% (95% CI, 2% to 8%; P = .001) and mean BMI z scores by 0.08 unit (95% CI, 0.04 to 0.11; P < .001). Among the 113 early-life antibiotic users, the effect of breastfeeding on postweaning antibiotic use was borderline significant (estimated 4% decrease per month; 95% CI, 0% to 7%; P = .04) and the effect on BMI z score disappeared (estimated 1% increase; 95% CI, −3% to 5%; P = .50). In the subcohort of 42 children with fecal microbiota composition analysis, the children with short breastfeeding duration (0-6 months) and no early-life antibiotic use or with long breastfeeding duration (8-16 months) and early-life use of antibiotics had a significantly lower abundance of Bifidobacterium (by 55%; 95% CI, 43% to 87%; P = .006; and 39%, 95% CI, 30% to 68%; P < .001, respectively) and Akkermansia (by 71%; 95% CI, 28% to 87%; P = .008; and 69%; 95% CI, 22% to 90%; P = .02, respectively) compared with those with long duration of breastfeeding and no early-life antibiotics.

Conclusions and Relevance  Antibiotic use in a child during breastfeeding may weaken the beneficial effects of long breastfeeding duration. The results suggest that particularly the long-term metabolic benefits of breastfeeding are conveyed by the intestinal microbiota.

Introduction

Breastfeeding is known to protect against infections in early life, and long duration of breastfeeding has also been shown to reduce the risk of overweight in children.1 The benefits of breastfeeding are likely related to the early development of the intestinal microbiota,2 which is strongly dependent on the infant’s diet.3 Human milk contains considerable amounts of complex fucosylated and sialilated glycans that are nonnutritious to the host but used as energy sources by Bifidobacterium species,4 which are abundant in the breastfed infant’s intestine.5

Mouse studies indicate a causal role for the intestinal microbiota regulating the development of growth, energy metabolism, fat accumulation, and susceptibility to diet-induced adiposity,6,7 a phenomenon termed microbiota-induced obesity.8 Intestinal microbiota composition in infancy9,10 as well as frequent antibiotic use in early childhood11 have been repeatedly associated with subsequent excessive weight development in children. The early-life microbiota composition also influences immune system development,12,13 offering a clear link between breastfeeding and the metabolic and immunological development of the infant. Most important, antibiotic use may be a modifying factor in the relationship between breastfeeding and infant weight development, which may explain inconsistencies in previous studies.

We hypothesize that the protective effects of breastfeeding are partly due to the influence of breastfeeding on the intestinal microbiota, which may be disrupted by antibiotic use. We tested this hypothesis in a retrospective cohort study of 226 children aged 2 to 6 years. Furthermore, as it is currently unknown whether the duration of breastfeeding has any long-term effects on the microbiota, we investigated this issue by analyzing the microbiota of a subcohort of these children in relation to the duration of breastfeeding and antibiotic use by the child during breastfeeding.

Box Section Ref ID

Key Points

  • Question Does antibiotic use in children during breastfeeding influence the long-term health benefits of breastfeeding?

  • Findings Breastfeeding decreased the number of postweaning antibiotic courses and decreased body mass index in later childhood. The protective effect of breastfeeding against high body mass index in later childhood was evident only in the children with no antibiotic use during the breastfeeding period.

  • Meaning The results suggest that the metabolic benefits of breastfeeding are largely conveyed by the intestinal microbiota, which is disturbed by antibiotic treatment.

Methods

The study cohort consisted of 226 children who participated in a probiotic trial from October 1, 2009, through April 30, 2010.14 The original trial was approved by the Ethics Committee of Joint Authority of Kainuu Region (clinicaltrials.gov identifier NCT01014676). The parents provided written informed consent. The original trial involved a larger cohort (501 children), from which a subset of 226 children with full records of breastfeeding duration, weight, height, and consent to retrieve antibiotic purchase information was selected for the current study, conducted from June 2015 to December 2015. Of the 226 children selected for the study, 142 children provided fecal samples for microbiota analysis. This retrospective study involving antibiotic purchase records received approval from the ethical committee of the Hospital District of Helsinki and Uusimaa.

Exclusion criteria in the original study were milk allergy, lactose intolerance, congenital heart disease requiring regular medication, malignant diseases, cytostatic treatment, use of biological rheumatic medication, continuous microbial medication, regular use of oral corticosteroids, diabetes, and simultaneous participation in other clinical trials. All children attended day care and resided in the same region in northern Finland at the time of the study. They were therefore exposed to a shared environment and diet. They were also ethnically highly homogeneous, almost exclusively of Finnish background.

Information on the total duration of breastfeeding was collected in a questionnaire completed by the mothers at the beginning of the trial, when the children were aged 2 to 6 years. Information on the duration of exclusive breastfeeding was not available. In the beginning of the trial, the children were measured for weight and height. These were used to generate body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) z scores according to LMS parameters obtained from the Centers for Disease Control and Prevention; z scores higher than 2 indicated overweight, and z scores lower than −2 indicated underweight. Lifetime antibiotic use records, a proxy for bacterial infections, for the study children were obtained from the Social Insurance Institution of Finland, which maintains a database of all reimbursed prescription drug purchases and eligibility for special reimbursement due to physician-diagnosed chronic disease, in this cohort asthma or allergic dermatitis.

The Mann-Whitney test was used to analyze the difference in antibiotic use per month per child during breastfeeding and during the year after weaning. Antibiotic use during the first year of life and lifetime antibiotic use (calculated as antibiotic use after 4 months following weaning) were modeled using generalized linear models, assuming the negative binomial distribution. In the model for lifetime antibiotic use, the age of the child was used as an offset. For the analyses of lifetime antibiotic use and BMI z score, the children were divided into 2 equal-sized groups based on their antibiotic use before or immediately after weaning: children who did not receive antibiotics during breastfeeding through 4 months after weaning (n = 113; no early-life antibiotics) and those who received antibiotics during breastfeeding through 4 months after weaning (n = 113; early-life antibiotic users). Groupwise characteristics are presented in the Table. In the lifetime antibiotic use and BMI z score models, the interaction between early-life antibiotic use and the duration of breastfeeding was assessed, after which the effect of breastfeeding duration was analyzed separately in both antibiotic use groups. We further subdivided the early-life antibiotic users into those receiving penicillin-type antibiotics (mainly amoxicillin, n = 78) during early life and those receiving other antibiotics (macrolides, cephalosporins, or sulfonamide-trimethoprim, n = 35), and we tested for the interaction between antibiotic type and breastfeeding duration. No adjustment variables were included in the univariate models, as we did not have information on potentially important background variables such as maternal BMI, birth mode, or socioeconomic status.

We have previously analyzed the fecal microbiota composition in the cohort and found that antibiotic use is an extremely strong driver of microbiota composition.15 In addition, BMI and asthma diagnosis were associated with the microbiota composition. Hence, to eliminate the overriding effect of different antibiotic use histories, BMI, and asthma or atopy, for the analysis of the association between duration of breastfeeding and the microbiota we selected a subcohort consisting of 42 children. We excluded all children who had used antibiotics within 6 months prior to sample donation, had received more than 10 antibiotic courses in their lifetime, had never used antibiotics (2 cases excluded to avoid different antibiotic use backgrounds in the different breastfeeding groups), had asthma or atopy, had a BMI z score less than −2 (underweight) or greater than 2 (overweight), or had fewer than 6610 DNA sequencing reads. If both fecal samples collected from a child fulfilled the criteria, we calculated the average microbiota composition over the 2 samples. Children with intermediate breastfeeding duration (7 or 8 months) were excluded to classify the children into distinct categories: short duration of breastfeeding (0-6 months) with antibiotic use by the child during breastfeeding (n = 0); short duration of breastfeeding (0-6 months) without antibiotic use by the child during breastfeeding (n = 19); long duration of breastfeeding (8-16 months) with antibiotic use by the child during breastfeeding (n = 11); or long duration of breastfeeding (8-16 months) without antibiotic use by the child during breastfeeding (n = 12). We compared the microbiota composition of those with long breastfeeding duration and no antibiotic use (12 children) with that of the composition in the group with short breastfeeding duration and no antibiotic use (19 children) and the group with long breastfeeding duration and antibiotic use (11 children) to analyze the influence of the duration of breastfeeding and antibiotic use during breastfeeding on the microbiota.

The fecal samples were collected at home and transported immediately to the study center for storage at −70°C or were initially stored in the home freezer for 0.5 to 78 hours (mean [SD], 19 [11] hours) before being transported to the study center. The DNA was extracted using the Promega Genomic Wizard DNA Purification Kit, followed by 454 Titanium pyrosequencing on a GS FLX instrument (Roche Diagnostics) of the V4 to V6 region of the 16S ribosomal RNA gene (primers S-D-Bact-0564-a-S-15/S and Univ-1100-a-A-15). The sequences were filtered for chimeras with the UCHIME program.16 Reads shorter than 501 base pairs were filtered out. After preprocessing, we had a mean (SD) of 8801 (1704) reads/sample (range, 1469-14653 reads/sample). De novo operational taxonomic unit picking was done using QIIME software.17 To avoid batch effects, we normalized the data following a method we developed earlier.18 The sequencing reads are publicly available at the European Nucleotide Archive (http://www.ebi.ac.uk/ena/data/view/PRJEB11685). The associations between the relative abundance of bacterial groups (at the class and genus levels) and breastfeeding group were analyzed using negative binomial models. Principal coordinates analyses were conducted using Bray-Curtis dissimilarities calculated based on the genus-level data. All statistical analyses were conducted with R version 3.2.1 statistical software (R Foundation for Statistical Computing), using the packages MASS19 and vegan.20

Results
Protective Effects of Breastfeeding Are Weakened by Antibiotic Use

We analyzed the associations between the duration of breastfeeding, antibiotic use, and BMI z scores in a cohort of 226 Finnish children aged 2 to 6 years (mean [SD] age, 55 [1.4] months; 54% male) (Table). Almost all children (97%) were breastfed for at least 1 month. The mean (SD) duration of breastfeeding was 8 (4.4) months (range, 0-18 months; median, 8 months; interquartile range, 5-12 months). By ages 1 and 2 years, 57% and 88% of the children had received at least 1 antibiotic course, respectively. Among the 226 children included in the study, 113 had no antibiotic use before weaning and 113 had early-life antibiotic use.

Breastfeeding was associated with a reduction in antibiotic use, ie, a reduction in bacterial infections. The frequency of antibiotic use before weaning (mean [SD], 0.06 [0.1] course/month, or 0.7 [1.3] course/year) was significantly lower than during the year after weaning (mean [SD], 0.21 [0.2] course/month, or 2.5 [2.7] courses/year) (odds ratio of antibiotic use after weaning vs before weaning, 7.5; 95% CI, 4.9-11.2; P < .001). Each month of breastfeeding was associated with a 6% reduction in the number of antibiotic courses during the first year of life (95% CI, 2%-10%; P = .003), amounting to a 72% reduction in antibiotic courses in infants breastfed for the whole first year compared with those not breastfed.

Lifetime antibiotic use, defined as the number of antibiotic courses purchased for the child after 4 months following weaning, was negatively associated with the duration of breastfeeding, and the association was stronger among the children with no early-life antibiotic use. In this group, each month of breastfeeding was associated with a 5% reduction in lifetime antibiotic use (95% CI, 2%-8%; P = .001), translating to a 60% reduction in lifetime antibiotic use in children breastfed for 12 months (Figure 1). Among the children who had received antibiotics before weaning, the relationship was only borderline significant (P = .04) (Figure 1), and each month of breastfeeding was associated with a 4% (95% CI, 0%-7%) reduction in antibiotic use. There was no evidence for an interaction between breastfeeding and antibiotic group (P = .53), as the same negative trend was observed in both groups. Breastfeeding was estimated to have a 1.6% weaker effect on antibiotic use in the early-life antibiotic users (95% CI, 6% weaker effect to 3% stronger effect). Similarly, among the early-life antibiotic users, the antibiotic type did not significantly modify the relationship between breastfeeding and lifetime antibiotic use. The effect of penicillin use was estimated to be 5% weaker than the use of other types of antibiotics (95% CI, 14% weaker to 3% stronger; P = .27).

The duration of breastfeeding was negatively associated with BMI z scores among the children who did not receive antibiotics during early life: each month of breastfeeding was associated with a 0.08-unit reduction in BMI z scores (95% CI, 0.04 to 0.11; P < .001) (Figure 2), translating to a whole z score unit reduction in children breastfed for 12 months compared with the nonbreastfed children. However, among the children who received antibiotics during early life, the duration of breastfeeding showed no association with BMI z scores (estimated 1% increase; 95% CI, −3% to 5%; P = .50) (Figure 2). Each month of breastfeeding was estimated to have a 9% stronger reduction in BMI z scores among the children with no early-life antibiotics compared with those with early-life antibiotics (95% CI, 4% to 14% stronger reduction; P = .002), supporting the modifying role of early antibiotic use in the relationship between breastfeeding and BMI z score. The antibiotic type received in early life did not significantly modify the relationship between breastfeeding and BMI z score; the duration of breastfeeding was estimated to have a 6% weaker effect on BMI z scores in the children with early-life penicillin courses compared with children with other antibiotic courses in early life (95% CI, 16% weaker effect to 4% stronger effect; P = .22).

Duration of Breastfeeding Is Associated With the Development of the Microbiota

Fecal microbiota composition was analyzed in a selected subcohort of 42 children: 12 children with long breastfeeding duration and no antibiotic use; 11 children with long breastfeeding duration and antibiotic use; and 19 children with short breastfeeding duration and no antibiotic use. The 3 groups did not differ in terms of age distribution (mean [SD] ages, 61 [22.0], 59 [12.4], and 60 [18.3] months, respectively; analysis of variance P = .99), lifetime antibiotic use (mean [SD] lifetime antibiotic courses, 3.3 [2.5], 4.6 [1.8], and 4.8 [2.4], respectively; generalized linear model P = .15), BMI z score (mean [SD] BMI z score, 0.37 [0.88], 0.75 [0.75], and 0.17 [0.93], respectively; analysis of variance P = .23), or number of DNA sequencing reads (mean [SD] number of DNA sequencing reads, 8779 [2305], 9381 [1641], and 8447 [1297], respectively; generalized linear model P = .36). The distributions were also fairly similar between the groups (eFigure 1 in the Supplement).

As is evident from the clustering of the different groups by colors (Figure 3A), the major driver of interindividual differences in microbiota composition was the early-life antibiotic use and duration of breastfeeding, regardless of age. The children with long breastfeeding duration with no antibiotic use during breastfeeding separated from the other groups, especially clearly from the group with long breastfeeding duration and antibiotic use. Principal coordinates analysis component 1, which explained 26% of the variance, mostly differentiated the groups and correlated negatively with the abundance of Akkermansia (r = 0.51; 95% CI, −0.71 to −0.25; P < .001) and Bifidobacterium (r = −0.43; 95% CI, −0.65 to −0.15; P = .004) species. The group with long breastfeeding duration and no early-life antibiotic use had a significantly higher relative abundance of Actinobacteria in relation to Firmicutes compared with the other groups (Figure 3B and eFigure 2 in the Supplement). The ratio of Actinobacteria to Firmicutes was 76% lower in the children with long duration of breastfeeding and early-life antibiotics (95% CI, 117% to 36% lower; P < .001) and 57% smaller in the children with short duration of breastfeeding (95% CI, 93% to 20% smaller; P = .003) compared with the group with long duration of breastfeeding and no antibiotics. The ratio of Verrucomicrobia to Firmicutes was 89% lower in the children with long duration of breastfeeding and early-life antibiotics (95% CI, 152% to 26% lower; P = .007) and 84% smaller in the children with short duration of breastfeeding (95% CI, 140% to 28% smaller; P = .004) compared with the group with long duration of breastfeeding and no antibiotics.

In more detailed analysis it was evident that children with short breastfeeding duration (0-6 months) and no early-life antibiotic use or with long breastfeeding duration (8-16 months) and early-life use of antibiotics had a significantly lower abundance of Bifidobacterium (by 55%; 95% CI, 43% to 87%; P = .006; and 39%; 95% CI, 30% to 68%; P < .001, respectively) and Akkermansia (by 71%; 95% CI, 28% to 87%; P = .008; and 69%; 95% CI, 22% to 90%; P = .02, respectively) and a higher abundance of several groups of Clostridia, particularly Roseburia (5-fold higher; 95% CI, 4- to 40-fold; P = .004; and 12-fold higher; 95% CI, 2- to 17-fold; P < .001, respectively), Faecalibacterium (2-fold; 95% CI, 1.2- to 3.6-fold; P = .01; and 2.8-fold; 95% CI, 1.6- to 5.2-fold; P < .001, respectively), unidentified members of Ruminococcaceae (1.8-fold; 95% CI, 1.2- to 2.6-fold; P = .003; and 1.8-fold; 95% CI, 1.1- to 2.7-fold; P = .01, respectively), and unidentified members of Lachnospiraceae (2.5-fold; 95% CI, 1.5- to 4.0-fold; P < .001; and 2.2-fold; 95% CI, 1.3- to 3.8-fold; P = .006, respectively) (Figure 4). The abundance of Bacteroides species was significantly higher in the group with early-life antibiotic use compared with both groups without early-life antibiotic use (4.2-fold higher compared with the children with long breastfeeding duration and no early-life antibiotic use; 95% CI, 2.0- to 9.2-fold; P < .001) (Figure 4). To confirm that these patterns were not restricted to the subcohort, we tested for group differences in the identified organisms among the 142 children. The same patterns were evident (eFigure 2 in the Supplement), although they were mostly not significant owing to the various confounding factors such as differences in antibiotic use history, BMI z score, and asthma or atopy.

Discussion

We combined a detailed data set of a cohort of children aged 2 to 6 years by including questionnaire information, weight and height measurements, register information on drug purchases, and next-generation sequencing–based analysis of fecal microbiota composition. This data set enabled a thorough and controlled analysis of the associations between breastfeeding duration, microbiota, and health indicators. Our results show that the duration of breastfeeding was associated with long-term development of the intestinal microbiota, lifetime antibiotic use, and risk of overweight, and the association persisted at least until age 7 years. A major finding was that the protective effects of breastfeeding against infections and overweight were weakened or completely eliminated by early-life antibiotic use. Furthermore, the apparent promotion of a protective microbiota by long breastfeeding was also eliminated by antibiotic use during breastfeeding. These results suggest that particularly the long-term metabolic benefits of breastfeeding are conveyed by the beneficial effects of breast milk on the intestinal microbiota, which are disrupted by antibiotic use.

Both early-life antibiotic use11 and short duration of breastfeeding1 have independently been shown to increase weight gain in children but, to our knowledge, their interactive effects have not been assessed. We found that the weight-increasing effect of early-life antibiotic use overshadows the protection against overweight endowed by long breastfeeding. This may explain some of the inconsistent results regarding the protection against overweight in previous large studies where antibiotic use was not analyzed.21

Breastfed children have better satiety and appetite regulation than formula-fed children,22 which could be partly due to the differences in microbiota, as the short-chain fatty acids produced by bacteria in the intestine influence gut hormones related to satiety and energy homeostasis.23,24 In addition, short-chain fatty acids are involved in the epigenetic programming of metabolism-regulating genes.25 Bacterial metabolism of bile acids has a significant role in the metabolic regulation of the host.26 Bile acid modification by the intestinal bacteria influences hepatic lipid and sugar metabolism and regulates the host’s energy homeostasis and weight.27 The secondary bile acids produced by bacteria are strong activators of TGR5, which increases basal energy expenditure.26

Bifidobacterium species and Akkermansia muciniphila, both of which were reduced in relative abundance in the children with short duration of breastfeeding or early-life antibiotic use, are known to specialize in host glycan use and are capable of growing in breast milk.28,29 Both Bifidobacterium and Akkermansia have been associated with protection against diet-induced obesity and related metabolic effects in rodents,23,30 and supplementation with a bifidogenic synbiotic has been shown to reduce weight in overweight children.31 In humans, obesity and related metabolic markers have been shown to correlate negatively with the abundance of Bifidobacterium and Akkermansia.32,33 The beneficial metabolic effects of these species are thought to arise at least partly from improved gut barrier function, which reduces metabolic endotoxemia.23,30 Furthermore, many Bifidobacterium species carry bile salt hydrolase genes (bsh) and are able to perform the first step in bacterial bile acid metabolism.34

Several bacteria belonging to the large group of Clostridia, whose relative abundance was increased in the children with early weaning or early-life antibiotic use, specialize in the degradation of plant-derived polysaccharides. In infants, the increased abundance of the polysaccharide-fermenting Firmicutes and decreased abundance of bifidobacteria after weaning have been associated with increased weight gain and adiposity.10,35 Increased abundance of Faecalibacterium prausnitzii has been observed in obese children aged 10 to 15 years compared with lean children aged 10 to 15 years.36 A reduced Actinobacteria to Firmicutes ratio has been observed in response to genetic and diet-induced obesity in mice.37 Overall, the results suggest that a prematurely adult-like microbiota in childhood, consisting largely of Clostridium clusters XIVa and IV, may promote weight gain. A microbiota dominated by human milk oligosaccharide–using bifidobacteria and Akkermansia may be associated with enhanced gut barrier development, reduced circulating lipopolysaccaride and inflammation, and increased bile acid metabolism, leading to increased amounts of secondary bile acids, increased energy expenditure, and possibly slower growth.

Long breastfeeding duration has been shown to protect against infections in early life.1 According to our results, the protection was still evident at age 6 years, especially in the children with no antibiotic use during early life. The relationship appeared weaker among the early-life antibiotic users, suggesting that microbiota may contribute to but not fully explain the protective effects of breastfeeding against infections in later life. However, early-life antibiotic use did not statistically significantly modify the association between breastfeeding duration and lifetime antibiotic use.

We cannot exclude the possibility that some of the observed effects of breastfeeding may have been due to potential confounding effects of birth mode, infant birth weight, lifestyle, diet, maternal education, or maternal BMI. Furthermore, if the mothers of the infants who needed antibiotics in early life reported the breastfeeding duration less accurately than those of healthy infants, some of the results may have been caused by recall bias. We cannot exclude the possibility of selection bias affecting the results. We were unable to include the entire cohort of 501 children owing to missing information and denial of consent to retrieve the drug purchase data. Furthermore, the exclusion criteria in the original study as well as the selection of samples for the analyses may reduce the generalizability of the results. The microbiota results apply only to children with no recent and moderate lifetime use of antibiotics, normal BMI, and no asthma or atopy.

Conclusions

Our results indicate that short duration of breastfeeding and antibiotic use before weaning are associated with microbiota development toward a composition that may increase weight gain. Promoting the natural development of the microbiota may be an effective way to improve long-term health in children.

Back to top
Article Information

Corresponding Author: Katri Korpela, PhD, Immunobiology Research Programme, Department of Bacteriology and Immunology, University of Helsinki, Haartmaninkatu 3, PO Box 21, FI-00014 Helsinki, Finland (katri.korpela@helsinki.fi).

Accepted for Publication: March 3, 2016.

Published Online: June 13, 2016. doi:10.1001/jamapediatrics.2016.0585.

Author Contributions: Dr Korpela had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Korpela, Kekkonen, de Vos.

Acquisition, analysis, or interpretation of data: Korpela, Salonen, Virta, de Vos.

Drafting of the manuscript: Korpela.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Korpela, Virta.

Obtained funding: de Vos.

Administrative, technical, or material support: Virta, Kekkonen, de Vos.

Study supervision: Salonen, Virta, de Vos.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported in part by grants 141140 from the Academy of Finland and 329/31/2015 from Tekes.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

References
1.
Hörnell  A, Lagström  H, Lande  B, Thorsdottir  I.  Breastfeeding, introduction of other foods and effects on health: a systematic literature review for the 5th Nordic Nutrition Recommendations.  Food Nutr Res. 2013;57.PubMed
2.
Woo  JG, Martin  LJ.  Does breastfeeding protect against childhood obesity? moving beyond observational evidence.  Curr Obes Rep. 2015;4(2):207-216.PubMedArticle
3.
Penders  J, Thijs  C, Vink  C,  et al.  Factors influencing the composition of the intestinal microbiota in early infancy.  Pediatrics. 2006;118(2):511-521.PubMedArticle
4.
Ward  RE, Niñonuevo  M, Mills  DA, Lebrilla  CB, German  JB.  In vitro fermentation of breast milk oligosaccharides by Bifidobacterium infantis and Lactobacillus gasseri Appl Environ Microbiol. 2006;72(6):4497-4499.PubMedArticle
5.
Bäckhed  F, Roswall  J, Peng  Y,  et al.  Dynamics and stabilization of the human gut microbiome during the first year of life.  Cell Host Microbe. 2015;17(5):690-703.PubMedArticle
6.
Ridaura  VK, Faith  JJ, Rey  FE,  et al.  Gut microbiota from twins discordant for obesity modulate metabolism in mice.  Science. 2013;341(6150):1241214.PubMedArticle
7.
Cox  LM, Yamanishi  S, Sohn  J,  et al.  Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences.  Cell. 2014;158(4):705-721.PubMedArticle
8.
Cox  LM, Blaser  MJ.  Pathways in microbe-induced obesity.  Cell Metab. 2013;17(6):883-894.PubMedArticle
9.
Kalliomäki  M, Collado  MC, Salminen  S, Isolauri  E.  Early differences in fecal microbiota composition in children may predict overweight.  Am J Clin Nutr. 2008;87(3):534-538.PubMed
10.
Dogra  S, Sakwinska  O, Soh  SE,  et al; GUSTO Study Group.  Dynamics of infant gut microbiota are influenced by delivery mode and gestational duration and are associated with subsequent adiposity.  MBio. 2015;6(1):e02419-14.PubMedArticle
11.
Saari  A, Virta  LJ, Sankilampi  U, Dunkel  L, Saxen  H.  Antibiotic exposure in infancy and risk of being overweight in the first 24 months of life.  Pediatrics. 2015;135(4):617-626.PubMedArticle
12.
Grönlund  MM, Arvilommi  H, Kero  P, Lehtonen  OP, Isolauri  E.  Importance of intestinal colonisation in the maturation of humoral immunity in early infancy: a prospective follow up study of healthy infants aged 0-6 months.  Arch Dis Child Fetal Neonatal Ed. 2000;83(3):F186-F192.PubMedArticle
13.
Sjögren  YM, Tomicic  S, Lundberg  A,  et al.  Influence of early gut microbiota on the maturation of childhood mucosal and systemic immune responses.  Clin Exp Allergy. 2009;39(12):1842-1851.PubMedArticle
14.
Kumpu  M, Kekkonen  RA, Kautiainen  H,  et al.  Milk containing probiotic Lactobacillus rhamnosus GG and respiratory illness in children: a randomized, double-blind, placebo-controlled trial.  Eur J Clin Nutr. 2012;66(9):1020-1023.PubMedArticle
15.
Korpela  K, Salonen  A, Virta  LJ,  et al.  Intestinal microbiome is related to lifetime antibiotic use in Finnish pre-school children.  Nat Commun. 2016;7:10410.PubMedArticle
16.
Edgar  RC, Haas  BJ, Clemente  JC, Quince  C, Knight  R.  UCHIME improves sensitivity and speed of chimera detection.  Bioinformatics. 2011;27(16):2194-2200.PubMedArticle
17.
Caporaso  JG, Kuczynski  J, Stombaugh  J,  et al.  QIIME allows analysis of high-throughput community sequencing data.  Nat Methods. 2010;7(5):335-336.PubMedArticle
18.
Korpela  K, Flint  HJ, Johnstone  AM,  et al.  Gut microbiota signatures predict host and microbiota responses to dietary interventions in obese individuals.  PLoS One. 2014;9(6):e90702.PubMedArticle
19.
Venables  W, Ripley  B.  Modern Applied Statistics With S. 4th ed. New York, NY: Springer; 2002.
20.
Oksanen  J, Blanchet  FG, Kindt  R,  et al.  vegan: Community Ecology Package: R Package Version 2.0-6. Vienna, Austria: R Foundation for Statistical Computing; 2013.
21.
Owen  CG, Martin  RM, Whincup  PH, Davey-Smith  G, Gillman  MW, Cook  DG.  The effect of breastfeeding on mean body mass index throughout life: a quantitative review of published and unpublished observational evidence.  Am J Clin Nutr. 2005;82(6):1298-1307.PubMed
22.
Brown  A, Lee  M.  Breastfeeding during the first year promotes satiety responsiveness in children aged 18-24 months.  Pediatr Obes. 2012;7(5):382-390.PubMedArticle
23.
Cani  PD, Neyrinck  AM, Fava  F,  et al.  Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia.  Diabetologia. 2007;50(11):2374-2383.PubMedArticle
24.
Lin  HV, Frassetto  A, Kowalik  EJ  Jr,  et al.  Butyrate and propionate protect against diet-induced obesity and regulate gut hormones via free fatty acid receptor 3-independent mechanisms.  PLoS One. 2012;7(4):e35240.PubMedArticle
25.
Remely  M, Aumueller  E, Merold  C,  et al.  Effects of short chain fatty acid producing bacteria on epigenetic regulation of FFAR3 in type 2 diabetes and obesity.  Gene. 2014;537(1):85-92.PubMedArticle
26.
Fiorucci  S, Mencarelli  A, Palladino  G, Cipriani  S.  Bile-acid-activated receptors: targeting TGR5 and farnesoid-X-receptor in lipid and glucose disorders.  Trends Pharmacol Sci. 2009;30(11):570-580.PubMedArticle
27.
Joyce  SA, MacSharry  J, Casey  PG,  et al.  Regulation of host weight gain and lipid metabolism by bacterial bile acid modification in the gut.  Proc Natl Acad Sci U S A. 2014;111(20):7421-7426.PubMedArticle
28.
Zivkovic  AM, German  JB, Lebrilla  CB, Mills  DA.  Human milk glycobiome and its impact on the infant gastrointestinal microbiota.  Proc Natl Acad Sci U S A. 2011;108(suppl 1):4653-4658.PubMedArticle
29.
Ottman  NA. Host Immunostimulation and Substrate Utilization of the Gut Symbiont Akkermansia muciniphila [PhD thesis]. Wageningen, the Netherlands: Wageningen University; 2015.
30.
Everard  A, Belzer  C, Geurts  L,  et al.  Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity.  Proc Natl Acad Sci U S A. 2013;110(22):9066-9071.PubMedArticle
31.
Safavi  M, Farajian  S, Kelishadi  R, Mirlohi  M, Hashemipour  M.  The effects of synbiotic supplementation on some cardio-metabolic risk factors in overweight and obese children: a randomized triple-masked controlled trial.  Int J Food Sci Nutr. 2013;64(6):687-693.PubMedArticle
32.
Santacruz  A, Collado  MC, García-Valdés  L,  et al.  Gut microbiota composition is associated with body weight, weight gain and biochemical parameters in pregnant women.  Br J Nutr. 2010;104(1):83-92.PubMedArticle
33.
Karlsson  CL, Onnerfält  J, Xu  J, Molin  G, Ahrné  S, Thorngren-Jerneck  K.  The microbiota of the gut in preschool children with normal and excessive body weight.  Obesity (Silver Spring). 2012;20(11):2257-2261.PubMedArticle
34.
Jones  BV, Begley  M, Hill  C, Gahan  CG, Marchesi  JR.  Functional and comparative metagenomic analysis of bile salt hydrolase activity in the human gut microbiome.  Proc Natl Acad Sci U S A. 2008;105(36):13580-13585.PubMedArticle
35.
Bergström  A, Skov  TH, Bahl  MI,  et al.  Establishment of intestinal microbiota during early life: a longitudinal, explorative study of a large cohort of Danish infants.  Appl Environ Microbiol. 2014;80(9):2889-2900.PubMedArticle
36.
Balamurugan  R, George  G, Kabeerdoss  J, Hepsiba  J, Chandragunasekaran  AM, Ramakrishna  BS.  Quantitative differences in intestinal Faecalibacterium prausnitzii in obese Indian children.  Br J Nutr. 2010;103(3):335-338.PubMedArticle
37.
Murphy  EF, Cotter  PD, Healy  S,  et al.  Composition and energy harvesting capacity of the gut microbiota: relationship to diet, obesity and time in mouse models.  Gut. 2010;59(12):1635-1642.PubMedArticle
×