New research finds that one in four US babies lack key gut bacteria, raising their risk for allergies, eczema, and asthma, and spotlighting the microbiome as a powerful early-life health determinant.
Study: Bifidobacterium deficit in United States infants drives prevalent gut dysbiosis. Image Credit: Corona Borealis Studio / Shutterstock
In a recent study published in the journal Communications Biology, a group of researchers determined how the absence of infant Bifidobacterium shapes gut microbiome composition, function, and two-year health outcomes in a demographically representative cohort of United States (US) infants.
Background
One in four US babies now carry no detectable Bifidobacterium, a genus that once dominated the guts of breast-fed babies worldwide. This deficit emerges as noncommunicable disease (NCD) rates have surpassed infectious illnesses, and allergy diagnoses in toddlers have doubled over two decades. Birth by cesarean section, antibiotic use, and formula feeding disrupt early microbial assembly, yet the precise ecological void created when Bifidobacterium disappears remains poorly defined. Understanding that the void matters because infant microbiota guides immune development, vaccine response, and metabolic programming. Accurate population-level data are scarce; therefore, further research is needed to clarify the long-term medical consequences of this microbial shift.
About the study
Investigators utilized the Institutional Review Board (IRB)-approved My Baby Biome study to recruit 412 healthy infants aged 1 to 3 months from 48 US states, matching national race, birth mode, and feeding statistics reported by the Centers for Disease Control and Prevention (CDC). Caregivers collected stool into deoxyribonucleic acid /ribonucleic acid (DNA/RNA) Shield tubes or on ice; laboratory staff recorded color, consistency, and pH before freezing phosphate-buffered saline (PBS) aliquots.
Whole-genome sequencing libraries were prepared and run on an Illumina platform; reads were filtered and taxonomically classified with a custom Kraken2/Bracken pipeline that used expectation maximization (EM) to apportion multi-mapped reads. Researchers annotated functional genes using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Carbohydrate-Active enZymes (CAZy) databases, and identified antimicrobial resistance (AMR) and virulence factor (VF) markers through Double Index Alignment of Next-generation sequencing Data (DIAMOND). Targeted metabolomics quantified 79 metabolites across 109 breast-fed samples. The team clustered microbiomes via a Dirichlet Multinomial Mixture (DMM) model and visualized compositional distances through Principal Coordinate Analysis (PCoA).
Statistical tests included analysis of covariance, generalized linear models (GLM), and Recursive Feature Elimination (RFE)-regularized logistic regression. Two-year caregiver surveys documented antibiotic exposures and pediatrician-diagnosed eczema, allergies, or asthma in 210 participants. It is important to note that these atopic outcomes were determined by caregiver report rather than direct clinical assessment.
Study results
Sequencing detected a total of 559 distinct species across all samples, with an average of just 12.1 species per infant, showing nascent guts are less diverse than adult microbiomes. However, Bifidobacterium showed a pronounced bimodal pattern: reads were entirely absent in 24% of infants, including 35% delivered by cesarean section and 19% born vaginally. When present, four species are identified, namely Bifidobacterium breve, Bifidobacterium bifidum, Bifidobacterium longum subsp. longum, and Bifidobacterium longum subsp. infantis dominated; still, B. infantis appeared in only 8% of samples.
DMM modeling of species counts separated the microbiomes into three community types, including Cluster 1 (C1, 24%), Cluster 2 (C2, 37%), and Cluster 3 (C3, 39%). C1, typical of breast-fed vaginally delivered infants, carried abundant B. breve and broad human milk oligosaccharide (HMO) gene sets. C2 displayed higher B. longum and phylum Bacteroidota with intermediate HMO potential. C3, most common after cesarean delivery, lacked Bifidobacterium almost entirely and instead harbored microbes such as Clostridium perfringens and other pathobionts. A random-forest model ranked the absence of B. breve and B. longum as the strongest cluster discriminators.
Functional profiling revealed pronounced metabolic divergence. C3 possessed fewer genes from human milk oligosaccharide gene cluster 4 (H4) and gene cluster 5 (H5) yet more urease loci, which are often associated with pathogenesis when found outside of Bifidobacterium, plus elevated AMR and VF genes; cesarean-born infants showed the highest VF density, and exclusive breastfeeding paradoxically magnified VF signatures when Bifidobacterium was absent, as it appeared to support the growth of other, potentially pathogenic HMO-consuming microbes in infants born by C-section. AMR gene load was negatively correlated with Bifidobacterium abundance (Spearman ρ = -0.71). Notably, the study also found that AMR gene abundance was unexpectedly higher in vaginally born infants, a surprising result that warrants further investigation. These functional differences forecast alterations in nutrient processing, epithelial barrier integrity, and inflammatory tone.
Metabolomics echoed these genomic gaps. C3 stools contained less thiamine and indole-3-lactate (ILA) (metabolites that guide immune maturation) and displayed a butyrate-skewed short-chain fatty-acid profile as well as altered bile-acid pools. Network analysis revealed a positive co-occurrence among Bifidobacterium, ILA, and thiamine, and negative links to opportunistic pathogens, underscoring the tight coupling between community membership and metabolic output. Similar metabolic patterns predict allergic sensitization in other infant cohorts.
Prospective health tracking highlighted clinical relevance. By the age of two, 30% of surveyed children had at least one physician-diagnosed atopic condition. After adjusting for postnatal antibiotics, infants assigned to C2 and C3 were 3.2 and 3.0 times more likely, respectively, than C1 peers to develop eczema, allergy, or asthma. High Bifidobacterium abundance cut the relative risk by 3.1-fold, and B. breve alone reduced it by 4.8-fold. Machine-learning importance scores further implicated phage repressor (originating from Proteobacteria) and lipopolysaccharide (LPS) biosynthesis gene clusters (unusually, dominated by genes from Firmicutes) as predictors of adverse outcomes, indicating that both microbial membership and functional traits shape early immune trajectories.
Conclusions
To summarize, among US infants, the early-life gut often lacks Bifidobacterium and instead hosts human milk oligosaccharide-scavenging or pathogenic taxa, especially after cesarean delivery. This imbalance was associated with restricted HMO metabolism, an increase in AMR and VF burdens, a shift in metabolites away from immunomodulatory ILA, and was linked to a nearly three-fold increase in the two-year risk of eczema, allergy, or asthma. Infant-type Bifidobacterium, particularly B. breve, appears to reduce these effects and thus represents a prime target for nutritional or probiotic interventions.
The study's observational design precludes causal inference. Ongoing longitudinal work will reveal whether restoring these microbes durably improves child health, but current evidence underscores the microbiome’s important role in early NCD trajectories.