Infant Skin Microbiome Patterns Differ by Atopic Phenotype

Key Takeaways

• In the VITALITY infant cohort, distinct skin microbiome signatures were reported across atopic phenotypes rather than a single shared profile.
• No taxa distinguished infants with food allergy alone or food sensitization alone from healthy controls; differences sat within atopic dermatitis and varied by whether FA or FS co-occurred.
• Skin dysbiosis characterized by enrichment of Staphylococcus species at 2-3 months, before diagnosis, was associated with later AD accompanied by FS or FA, but not with later AD alone.
• Among infants with AD, FLG mutation carriers showed additional microbial shifts at 12 months but not at 2-3 months, while a microbial genome-wide association analysis identified strain-level variants correlating with AD severity in both directions.

Infants with atopic dermatitis, food sensitization, and food allergy often cluster clinically, but their overlap does not necessarily reflect one underlying biologic pattern. Early skin colonization and host barrier genetics may help explain why some infants develop dermatitis alone while others follow a broader atopic trajectory. That uncertainty has focused attention on skin sampled before diagnosis and again after disease emerged, with comparisons across co-occurring atopic phenotypes.

Investigators reported a deep shotgun metagenomic analysis of 1078 skin swabs from the cheek and antecubital fossa of 429 infants, sampled at 2-3 months before diagnosis and at 12 months after diagnosis. The work was a nested, phenotype-enriched analysis within VITALITY, a randomized trial of postnatal vitamin D supplementation in healthy term infants in metropolitan Melbourne, Australia. The cohort comprised 95 infants with IgE-mediated food allergy (73 with AD), 55 with food sensitization (31 with AD), 171 with AD alone, and 108 healthy controls. Food allergy was diagnosed on oral food challenge, skin prick testing, and parent-reported reaction history; food sensitization was positive skin prick test and/or food-specific IgE with clinical tolerance.

At 12 months, no taxa differed between infants with FA alone or FS alone and healthy controls at either site. Differences were confined to AD and varied by co-occurring allergy. Staphylococcus epidermidis was enriched in the AD-affected antecubital fossa of infants with AD alone. AD with FA showed decreased Staphylococcus hominis and increased Dermacoccus nishinomiyaensis at the affected site, increased Malassezia slooffiae regardless of affected status, and reduced Lactococcus species at both the cheek and AD-affected antecubital fossa, with L. cremoris the only taxon correlating negatively with SCORAD. AD with FS instead showed reduced Prevotella melaninogenica and Veillonella species. Staphylococcus aureus did not differ across AD subgroups at 12 months, which the authors attribute to the predominance of mild AD.

At 2-3 months, five species were enriched in infants who later developed AD with FA, including S. aureus, and S. saprophyticus was elevated in those who later developed AD with FS. No species preceded AD alone. Among infants with AD, 17 species, most prominently Streptococcus and Granulicatella, were decreased at the AD-affected antecubital fossa of FLG mutation carriers, while M. slooffiae was increased at the cheek and unaffected fossa; none of these differences appeared at 2-3 months. The microbial GWAS found an S. aureus fusA variant associated with lower SCORAD and S. thermophilus prs mutations with higher SCORAD, though the authors note fusA may reflect prior antibiotic exposure. Strain tracking showed mother-infant overlap across all six tested species regardless of infant disease status.

The authors name several limitations. Sample sizes for the FA-only and FS groups and for the pre-diagnosis timepoint were limited, so null findings should not be read as absence of effect. Only 70 infants contributed AD-affected antecubital fossa samples at 12 months, 8 of them with AD and FS. Most AD was mild (median SCORAD 7.9). Antibiotic exposure data were unavailable, vitamin D effects cannot be excluded while assignment remains blinded, and shotgun profiling yields relative abundance rather than absolute biomass. The cohort was entirely Australian-born.

The authors concluded that infant atopic phenotypes are associated with distinct, phenotype-specific skin microbiome features emerging before and after disease onset, highlighting the microbiome as a potential target for early risk stratification.

Clinician Questions

Which infant atopic phenotypes were compared in this skin microbiome analysis?

The analysis compared AD alone, AD with FS, AD with FA, FA alone, and FS alone against healthy controls. Infants with food allergy or sensitization but without AD showed no taxa differing from controls at either site, so all reported signatures were within AD. The cohort was phenotype-enriched within a supplementation trial, so phenotype proportions do not represent population prevalence.

How was FLG mutation status related to the skin microbiome in infants with atopic dermatitis?

Among infants with AD, FLG null mutation carriers showed 17 species decreased at the AD-affected antecubital fossa and increased M. slooffiae at the cheek and unaffected fossa. No taxa differed by FLG status among healthy infants or at 2-3 months. Genotype was inferred from metagenomic sequence rather than conventional genotyping.

Were microbiome differences reported before infant atopic dermatitis was diagnosed? Yes. At 2-3 months, specific species were enriched in infants who later developed AD with FA or with FS, but none preceded AD alone. Only a subset of infants was sampled at this timepoint, and limited power means the AD-alone null should not be read as absence of effect. The timing suggests an early association pattern rather than a validated predictive test.