Prenatal Probiotics: Eczema Hype, GDM Evidence

The infant-gut theory has a problem

Maternal probiotics have a plausible route to shaping a baby's immune system. The popular version of the story says they improve the infant gut microbiome, lower eczema risk, and perhaps blunt the progression from eczema to asthma or food allergy. The longest follow-up available does not support that clinical promise, and a small mechanistic trial found no evidence that maternal probiotics remodeled infant gut microbes in the proposed protective direction.

That does not prove probiotics have no biological effects during pregnancy. They may affect maternal metabolism, immune signaling, and breast milk composition. But it does mean that "take probiotics in pregnancy to prevent your child's eczema" runs well ahead of the evidence.

Long-term allergy outcomes: no benefit in the SPRING follow-up

The strongest practical test is whether children actually develop less allergic disease. In the 2024 OFFSPRING follow-up study, researchers contacted families from the SPRING randomized trial. In SPRING, 411 overweight or obese pregnant women received either Lactobacillus rhamnosus GG plus Bifidobacterium lactis BB12 or placebo, starting in the second trimester.

At ages 3 to 7 years, 107 children had follow-up data. Eczema occurred in 26.9% of children in the probiotic group and 23.6% in the placebo group (p=0.696). Wheeze or asthma occurred in 15.3% versus 16.3% (p=0.89), and food allergy in 13.4% versus 12.7% (p=0.91). None of those differences was statistically significant.

The pattern matters more than any one endpoint. The trial did not find less eczema, wheeze/asthma, or food allergy at school age. Earlier studies and pooled analyses have reported lower eczema rates before age 3 with mixed Lactobacillus/Bifidobacterium products, including a 2021 meta-analysis. But that analysis pooled maternal and infant supplementation routes. It cannot tell us whether a prenatal maternal intervention caused the apparent benefit, or whether infant-directed treatment accounted for it.

The OFFSPRING result also has limits. Only 107 of 379 contacted families responded, a 28.2% response rate. Responders were more likely to have tertiary education and less likely to smoke, which could narrow behavioral differences between the trial groups. SPRING also enrolled women with BMI above 25, mainly in a gestational-diabetes-prevention context. Its allergy findings may not apply neatly to normal-weight women or families selected for high atopic risk. Still, this is the longest available follow-up, and its three null outcomes deserve more weight than a short-term signal with an uncertain route of action.

The proposed microbiome mechanism did not appear

A 2025 Finnish mechanistic sub-study asked a narrower question: if maternal probiotics reduce eczema risk, do they do it by reshaping the infant gut microbiome? The study followed 46 infants for two years after a perinatal maternal intervention using either LPR plus BL999, ST11 plus BL999, or placebo. Nineteen of the 46 children developed eczema.

At 6 months, infants who later developed eczema had higher relative abundance of Clostridia than healthy controls. That is an association with eczema risk, not proof that Clostridia caused eczema. More important for the intervention question, probiotic assignment did not shift the infant microbiota in a protective direction. Microbial diversity also did not differ significantly between groups at 1 or 6 months.

This finding weakens the claim that maternal probiotics prevent eczema by installing a better infant microbiome. The study was small, with 19 eczema cases spread across three probiotic arms and one placebo arm, so it could miss modest compositional effects. It cannot rule out every microbiome-mediated pathway. It does show that the central mechanism has not appeared when tested directly.

Breast milk and maternal immune signaling remain hypotheses

If direct infant colonization is not the route, maternal immune signaling is the next obvious candidate. Probiotics could alter maternal short-chain fatty acids such as acetate, propionate, and butyrate; lipopolysaccharide signaling; or extracellular vesicles. A 2025 systematic review of 56 studies describes these as plausible pathways for maternal gut-derived signals to influence fetal immune programming.

Breast milk could also carry immune signals after birth. Transforming growth factor beta 1 and 2 and secretory IgA are plausible candidates because they may support neonatal oral tolerance and regulatory T-cell development. Plausibility, however, is not a measured effect. The only meta-analysis of randomized trials that directly examined maternal prebiotic or probiotic effects on breast-milk TGF-β1, TGF-β2, and IgA found contradictory results across studies.

Differences in strain, dose, intervention timing, breastfeeding duration, and product composition likely contribute to that inconsistency. Some trials also used synbiotics, which makes it difficult to separate probiotic effects from prebiotic effects. A broader 2025 review of maternal diet and immune programming supports the general idea that prenatal diet can influence allergic predisposition, but it does not establish that a probiotic capsule recreates the effect of a whole dietary pattern.

Infant-type bifidobacteria add useful biology, but not a shortcut around this gap. A 2025 review describes how postnatal infant-directed bifidobacteria can produce short-chain fatty acids and aromatic lactic acids with immunomodulatory properties. Those pathways depend on colonization in the infant gut. Evidence for infant supplementation cannot simply be reassigned to supplements taken by the mother during pregnancy.

Gestational diabetes has a more coherent evidence base

The metabolic case is different. In women with gestational diabetes mellitus (GDM), randomized-trial meta-analyses report improvements in glucose and lipid markers with probiotic or synbiotic supplementation. A 2023 meta-analysis identified gut dysbiosis as a feature of GDM and found metabolic improvements across randomized trials. A 2024 meta-analysis of 11 randomized trials involving 713 participants reported that a median probiotic dose of 6 × 10⁹ CFU per day reduced fasting glucose by 4.16 mg/dL; the supplied summary does not provide the confidence interval for that estimate.

A multicenter, randomized, double-blind, placebo-controlled 2024 trial also tested oral probiotics in women with established GDM, focusing on fasting and postprandial glycemic control. That design provides more direct evidence for glycemic management than observational microbiome studies, although glucose markers remain intermediate outcomes rather than proof of fewer pregnancy complications.

Timing may matter for prevention. In a 2025 randomized trial of overweight or obese women enrolled at 6 to 12 weeks of pregnancy, probiotic and dietary-fiber fermented milk produced GDM in 18.42% of participants (21 of 114), compared with 33.33% (39 of 117) receiving standard care. The reported odds ratio was 0.45, with a 95% confidence interval of 0.25 to 0.83. The intervention included fermented milk and dietary fiber, so the result does not isolate a probiotic-only effect. It does support testing gut-directed metabolic interventions before dysregulation is entrenched.

Diet may matter alongside supplements. In women with GDM, a randomized diet study found that a higher-complex-carbohydrate diet increased maternal bifidobacteria and altered early infant microbiome acquisition. That finding supports maternal diet as a modifiable influence on microbial transmission; it does not demonstrate that a probiotic supplement will produce the same result.

What to do with this evidence

Do not treat prenatal probiotics as a reliable eczema-prevention strategy. The best long-term follow-up found no reduction in eczema, wheeze/asthma, or food allergy, while the direct test of infant microbiome remodeling found no protective shift. Breast-milk and maternal immune pathways remain biologically credible but quantitatively unsettled.

For GDM or elevated metabolic risk, probiotics have a stronger, though still product-specific, evidence base for improving glucose and lipid markers. They belong in a clinician-guided GDM plan, not in place of glucose monitoring, dietary treatment, insulin, or metformin when those treatments are indicated. Pregnancy is an unusually poor time to let a mechanism story outrun the endpoint data.