Table 1.
Study characteristic and main results
| Study participants; age at stool sample collection | Technique of microbiota determination* | Respiratory outcome; age at outcome evaluation; follow-up n/N (%) | Method of evaluation; outcome definition | Participants in each outcome group |
Children with respiratory disease versus no respiratory disease |
Adjustments for diversity; adjustments for relative abundance | Study quality | ||
|---|---|---|---|---|---|---|---|---|---|
| Diversity and relative abundance of taxa | Validation | ||||||||
| Arrieta et al (2015);27 nested case-control study | 319 healthy, more than 35 weeks' gestation, singleton newborn babies; 3 months and 1 year | 16S rRNA V3, Ilumina HiSeq, and Greengenes database (2006); validation with qPCR and 16S rRNA V3 amplicon | Atopic wheeze (as a proxy for later asthma);† 1 year; NA (loss to follow-up does not apply in case-control study designs) | Parental questionnaires at 3 months, 6 months, and 1 year of age; study clinical assessment at age 1 year; four phenotypes: atopy (positive prick test at age 1 year), wheeze (≥1 wheezing episodes in first year of life), atopy and wheeze, and control individuals (no asthma or atopy at age 1 year) | 87 (27%) in atopy group; 136 (43%) in wheeze group; 22 (7%) in atopy and wheeze group; 74 (23%) control individuals | No difference in α-diversity (Shannon index) between the four phenotypes at ages 3 months or 1 year; relative abundance of bacteria taxa (genera): atopy and wheeze (compared with control individuals) showed lower FLRV and Peptostreptococcus (p>0·05) at age 3 months, atopy and wheeze (compared with control individuals) showed lower Oscillospira at age 1 year (p>0·05), and no differences between participants with wheeze and participants without wheeze | Validation with qPCR (for FLRV): individuals with atopy and wheeze have lower relative abundance of FLRV genera at age 3 months than control individuals (p<0.0001) | Unadjusted; unadjusted | Poor |
| Laursen et al (2015);28 cohort study | 114 healthy, singleton children; 9 months and 18 months | 16S rRNA V3, Ion Torrent, OneTouch, and Ion Personal Genome Machine systems; CLC Genomic workbench, Ribosomal database project classifier | Asthmatic bronchitis (cumulative prevalence); 3 years; at 3 years: 104/114 (91·2%) | Parental interviews when child was age 9 months, 18 months, and 3 years; cumulative prevalence (age 0–3 years) of diagnosed asthmatic bronchitis defined as squeaky and wheezing breathing in connection with cold or other viral infections in the respiratory system | 19/104 (18%) had asthmatic bronchitis | No differences in α-diversity (Shannon Index) at ages 9 months or 18 months; relative abundance of bacteria taxa (genera): no differences at ages 9 months or 18 months (p>0·05 after correction for multiple testing) | NA | Unadjusted; unadjusted | Poor |
| Fujimura et al (2016);29 cohort study | 298 children (all births); 130 at 1 month (neonatal) or‡ 168 at 6 months (infants) | 16S rRNA V4, Illumina MiSeq, Greengenes database (2013), and Ribosomal Database Project classifier; fungal internal transcribed spacer region 2, UNITE database V6 | Asthma; 4 years; at 4 years: 111/130 (85%, neonatal samples), 168/168 (100%, infant samples) | Parental interviews at months 1, 6, 12, 24, and 48, and clinical study visit at age 2 years; asthma according to parental-reported doctor diagnosis of asthma | 39/279 (14%) had asthma; 17/111 (15%) were neonatal, 22/168 (13%) were infants | α-diversity not reported relative to asthma risk; bacterial β-diversity (PERMANOVA; R2=0·09, p<0·001) and fungal β-diversity (Bray–Curtis; PERMANOVA, R2=0·037, p=0·068) differed between clusters; relative abundance of bacteria taxa (genera): no differences in infant group (age 6 months), newborn babies with high risk of asthma at age 4 years (OR 1·09–10·3; p<0·05) had lower Bifidobacterium, Lactobacillus Faecalibacterium, and Akkermansia compared with low-risk asthma participants (adjusting for false discovery rate: B-H q<0·05); relative abundance of fungal taxa: newborn babies with high risk of asthma had lower Malassezia and higher Candida and Rhodotorula compared with low-risk asthma participants (adjusting for false discovery rate: B-H q<0·20) | NA | Unadjusted; unadjusted (data reported and individually adjusted for current breastfeeding, detectable cat allergen, detectable dog allergen, ever breastfed, female, first-born, maternal age, maternal education, maternal history allergic disease, birth delivery method, pets, and race) | Poor |
| Stiemsma et al (2016);30 nested case-control study | 76 healthy, term (>35 weeks), singleton newborn babies; 3 months and 1 year | 16S rRNA V3, Hiseq Ilumina, and Greengenes database (2006); validation with qPCR and 16S rRNA V3 amplicon | Asthma; 4 years; NA | Multiple questionnaires and clinical assessments by study physicians at ages 1 year and 3 years; cases defined as physician diagnosis of asthma by age 4 years or prescribed inhaled medications from ages 3–4 years and controls defined as negative for asthma or inhaled medication, wheezing, and atopy (based on standardised allergen skin prick testing at ages 1 year and 3 years) | 39 (51%) with asthma; 37 (49%) matched healthy controls | No differences in α-diversity (Shannon index) or β-diversity at ages 3 months or 1 year; relative abundance of bacteria taxa: individuals with asthma showed lower Clostridiales (class; log2FC 1–2; p=0·035) and Lachnospira (genera; log2FC 1–2; p=0·098), higher Clostridium neonatale (species; log2FC 1–2; p=0·076), Clostridiaceae (family; p=0·005), and Firmicutes (phylum; log2FC 1–3; p=0·035) than did control individuals at age 3 months, individuals with asthma showed higher Lachnospiraceae (family; p=0·032) and Rothia (genera; p=0·003) than did control individuals at age 1 year | Validation with qPCR: individuals with asthma showed lower relative abundance of Lachnospira at age 3 months (p=0·008) and lower Clostridium neonatale at age 1 year (p=0·02) | Matched on gender, birth delivery method, feeding practices, antibiotic exposure; matched on gender, birth delivery method, feeding practices, antibiotic exposure | Good |
| Arrieta et al (2018);31 nested case-control study | 97 healthy newborn babies; 3 months | 16S rRNA V4, Miseq Ilumina, Greengenes database (2006), 18S rRNA V4 (fungi), SILVA database (2013); validation with qPCR | Atopic wheeze; 5 years; NA | Parental interviews; cases defined as maternally reported wheeze in the previous 12 months at age 5 years and positive skin prick test response and controls defined as random sample of children with no previous history of wheeze and no evidence of atopy at age 5 years | 27 (28%) with atopic wheeze, 70 (72%) healthy controls | No differences in α-diversity (Chao1 index) or β-diversity, bacterial or fungal; relative abundance of bacteria taxa (genera): children with atopic wheeze had low Bifidobacterium (log2FC>4; p<0·001) and high Streptoccocus (log2FC 2–4; p=0·044) and Veillonella (log2FC 2–4; p=0·031), adjusted for false discovery rate at age 3 months; relative abundance of fungal taxa (genera): children with atopic wheeze had higher Pichia Kudriavzevii at age 3 months (log2FC 2–6; p<0·01) | Validated increase in Pichia kudriavzevii species using qPCR at 3 months in children with atopic wheeze compared with healthy controls | Unadjusted; adjusted for antibiotic use during pregnancy or during first year of life, antibiotic duration, birth delivery method, household potable water, number of respiratory tract infections during first year of life, eosinophilia at age 7 months, number of diarrhoeal episodes during first year of life | Fair |
| Stockholm et al (2018);32 cohort study | 690 newborn babies (all births); 1 week, 1 month, and 1 year | 16S rRNA V4, Illumina Miseq, and Greengenes database (2013) | Asthma; 5 years; at 5 years: 648/690 (94%) | Clinical visits at age 1 week, ages 1, 3, 6, 12, 18, 24, 30, and 36 months, and yearly thereafter (including during acute respiratory episodes); asthma diagnosis if all of 5 episodes of lung symptoms within 6 months (at least for 3 consecutive days), exercise-induced symptoms, prolonged nocturnal cough, or persistent cough outside of common colds, need for intermittent rescue use β2-agonist, and response to 3 months of inhaled steroids and relapse after stopping | 60/648 (9%) had asthma | No differences in α-diversity (Shannon and Chao1 indices) at any timepoint; no differences in β-diversity at ages 1 week and 1 month; β-diversity at 1 year in individuals with asthma vs individuals without asthma (PERMANOVA; F=3·4, R2=0·6%, p=0·003); relative abundance of bacteria taxa (genera): no differences at ages 1 week or 1 month; for the 20 most abundant bacterial genera, individuals with asthma had lower Roseburia (median relative abundance 0·27% vs 0·66%; p=0·042), Alistipes (median relative abundance 0·04% vs 0·35%; p=0·002), and Flavonifractor (0·05% vs 0·07%; p=0·002) and higher Veillonella (0·94% vs 0·29%; p=0·035) at age 1 year than had controls; diversity and relative abundance differences were influenced by children born to mothers with asthma (effect modifier; p=0·011) | NA | Adjusted for birth delivery method, duration of exclusive breastfeeding, older sibling hospitalisation after birth, and antibiotic use; unadjusted | Good |
| Reyman et al (2019);33 cohort study | 120 healthy, term (>37 weeks) newborn babies; 1 week | 16S rRNA V4, llumina MiSeq, Naive Bayesian Ribosomal Database Project classifier (version 2.2), and SILVA database (2012); validation with qPCR | Respiratory infection (cumulative incidence); 1 year; at 1 year: 118/120 (98%) | Structured interview and questionnaire; respiratory infection defined as fever (>38·0°C) and any of cough, wheezing, dyspnoea, earache, or malaise (events in the first year of life were summed up to a cumulative number and categorised into 0–2 vs 3–7) | 77 (65%) 3–7 respiratory infections; 41 (35%) with 0–2 respiratory infections (considered healthy) | Diversity not reported relative to respiratory infection; relative abundance of bacterial taxa (genera): children with more respiratory infections (those with 3–7 compared with those with 0–2) had lower Bifidobacterium (log2FC 2·1; p=0·049) and higher Klebsiella (log2FC 3·2; p=0·007) and Enterococcus (log2FC 2·8; p=0·009) at age 1 week | Validation with qPCR: children with 3–7 respiratory infections had higher relative abundance of Enterococcus spp (p=0·015) than did children with 0–2 respiratory infections | NA; adjusted for birth delivery method | Good |
| Galazzo et al (2020);34 cohort study | 440 healthy, term newborn babies and parents with atopic disease; 5 weeks, 3·3 months, 5·3 months, and 7·8 months | 16S rRNA V3, Illumina MiSeq, and Greengenes database (2011) | Asthma; 6–11 years; at 6–11 years: 292/440 (66%) | Clinical visit; parent-reported doctor diagnosis of asthma in combination with any indicative symptoms in the past 12 months (wheezing, shortness of breath, nocturnal awakening due to symptoms; included lung function testing) | Not reported | No differences in α-diversity (Shannon index); relative abundance of bacteria taxa (genera): children with asthma showed, at all ages, lower Lachnobacterium, Lachnospira, and Dialister (p<0·001) than children without asthma | NA | Adjusted for breastfeeding, age at introduction to solid food, birth delivery method, treatment (placebo vs probiotic), birthweight, sex, mother or father with atopic dermatitis, >2 older siblings, and pets; adjusted for breastfeeding, age at introduction to solid food, birth delivery method, treatment (placebo vs probiotic), birthweight, sex, mother or father with atopic dermatitis, >2 older siblings, and pets | Good |
| Boutin et al (2020);35 cohort study | 837 healthy, term (>35 weeks), singleton newborn babies; 3 months | 16S rRNA V4, HiSeq Ilumina, and Greengenes database (2013) | Recurrent wheeze and atopic wheeze; 1 year; 659/837 (79%) after 1 year | Questionnaires and clinical assessments by study physicians at ages 1, 3, and 5 years; recurrent wheezing defined as ≥2 episodes of wheezing in the first year of life, healthy control individuals did not have asthma, recurrent wheeze, atopic dermatitis, atopy, or allergic sensitisation at ages 1, 3, and 5 years | 142/659 (21%) had recurrent wheeze; 45/659 (6%) had atopic wheeze; 16/659 (2%) had both (included in recurrent wheeze and atopic wheeze groups) | Children with recurrent wheeze and atopic wheeze had lower α-diversity at age 3 months than individuals without; increased α-diversity at age 3 months was protective of recurrent wheeze (OR 0·75 [0·6–0·95]; p=0·007) and atopic wheeze (OR 0·55 [0·1–0·90]; p=0·016); relative abundance of bacterial taxa (analyses aimed at identifying only decreased bacteria genus): children with recurrent wheeze had lower Faecalibacterium, Lachnospira, Coprococcus, and Oscillospira at age 3 months than did healthy children and children with atopic wheeze showed lower Faecalibacterium, Lachnospira, Coprococcus, Roseburia, Blautia, Parabacteroides, and Ruminococcus at age 3 months than did healthy children | NA | Unadjusted; unadjusted (machine learning predictive methods) | Poor |
| Patrick et al (2020);36 cohort study | 917 healthy, term (>35 weeks), singleton newborn babies; 3 months and 1 year | 16S rRNA V4, Hiseq Ilumina, and Greengenes database (2013) | Asthma; 5 years; outcome evaluated in those with stool sample processed at 1 year: 570/917 (62%) | Multiple questionnaires and clinical assessments by study physicians at ages 1, 3, and 5 years; diagnosis of asthma based on questionnaire data, clinical history, and medical examination | 63/570 (11%) had asthma | At age 1 year, children with asthma showed decreased α-diversity (Chao1 index) compared with children without asthma; having increased α-diversity at age 1 year protected from asthma (OR 0·68 [0·46–0·99]; p=0·046); relative abundance of bacterial taxa (only summarising those with at least 1·5 log2FC): at age 1 year, children with asthma had lower Faecalibacterium prausnitzii (log2FC −1·57 to 1·77), Ruminococcus bromii (log2FC −2·07), and Rikenellaceae (family) (log2FC −2·59) and higher Dialister (genus) (log2FC 2·04; false discovery rate p<0·05) than did children without asthma§ | NA | Adjusted for antibiotic use at age 1 year, race, birth delivery method, older siblings, sex, birthweight, prenatal atopy, breastfeeding at age 6 months, tobacco smoke exposure in first year of life, NO2 exposure during first year of life, season of birth, living area (urban or rural); adjusted for antibiotic use at age 1 year, race, birth delivery method, older siblings, sex, birthweight, prenatal atopy, breastfeeding at age 6 months, tobacco smoke exposure in first year of life, NO2 exposure during first year of life, season of birth, living area (urban or rural), sequencing batch, and study centre | Good |
| Depner et al (2020);37 cohort study | 720 newborn babies; 2 months and 1 year | 16S rRNA V4, Illumina MiSeq, and Greengenes database (2013); fungal internal transcribed spacer region 1, UNITE dynamic database (2010) | Asthma; 6 years; at 6 years: 626/720 (87%) | Parent-reported doctor diagnosis of asthma at least once or recurrent diagnoses of obstructive bronchitis or asthmatic bronchitis; atopic and non-atopic asthma defined as the presence or absence of concomitant sensitisation to inhalant allergens (seasonal or perennial) with specific IgE concentrations higher than 0·7 IU/mL–1 at age 6 years (lung function measured by spirometry at age 6 years) | 53/626 (9%) had asthma | Reported EMA¶ as a proxy for α-diversity (positive correlation [r=0·70] for number of different bacteria genera); children with asthma had a difference in EMA at age 2 months and lower EMA at age 1 year than did children without asthma; relative abundance of bacteria taxa (genera): children with asthma had, at age 2 months, lower Bacteroides and Parabacteroides and higher Enterococcus than did children without asthma; having a high relative abundance of Bacteroides and Parabacteroides and low relative abundance of Enterococcus at age 2 months protected from both atopic and non-atopic asthma (OR 0·68 [0·49–0·95]; p=0·024); children with asthma showed, at age 1 year, lower Roseburia, Ruminococcus, and Faecalibacterium than did children without asthma; having higher relative abundance of Roseburia, Ruminococcus, and Faecalibacterium at age 1 year protected from non-atopic asthma (OR 0·62 [0·39–1·00]; p=0·048; association fully explained by EMA) | NA | Unclear if adjusted (collected clinical variables including birth delivery method, breastfeeding, antibiotic use, gestation age, birthweight, Apgar score, parental history of atopy, farm exposure, pets, smoke exposure, number of siblings, and parental education) | Good |
API= Asthma Predictive Index. B-H=Benjamini-Hochberg. EMA=estimated microbiome age. FLRV=Faecalibacterium, Lachnospira, Rothia, Veillonella. Log2FC=log2 fold change. NA=not available. OR=odds ratio. qPCR=quantitative PCR.
*
Target, sequencing platform, pipeline, and database.
†
Children were followed up for 3 years. At age 3 years children had a clinical visit to predict asthma development between ages 6 years and 11 years using the API. The atopy and wheezing group at age 1 year were 5·4 times more likely to have a positive API score at age 3 years compared with the wheeze only group at 1 year.
‡
All other studies used and instead of or (participants were followed up and samples were collected at different timepoints).
§
Findings extracted from supplementary materials. Findings reported in main paper were that patients with asthma and antibiotic use at age 1 year had relative decreased abundance of Faecalibacterium prausnitzii, Roseburia (genera), and Ruminococcus bromii and increased abundance of Clostridium perfringens (false discovery rate p<0·05).
¶
Random forest analysis (machine learning) was used to estimate the healthy age of gut microbiota sampled at ages 2 months and 1 year in 133 healthy individuals (no diarrhoea, wheezing, or asthma in the first year of life).