A baby's brain isn't built solely from what it inherits at conception. A study published Friday in Cell Press Blue by researchers at the Chinese University of Hong Kong shows that the epigenetic settings present at birth (molecular switches that control how genes are expressed without altering the DNA sequence) shape which bacteria colonize an infant's gut during the first year of life. And critically, those gut bacteria can then modify the risk of autism spectrum disorder and ADHD that was written into those epigenetic patterns before the child was born.
The finding, from a birth cohort study tracking 571 infants from cord blood through age three, reframes neurodevelopmental risk as a conversation between two biological systems rather than a fixed outcome. It also names the specific bacteria that may be able to override that risk, at least partially.
"This, to our knowledge, represents the first longitudinal study with multiple sample types to depict the intimate interplay between perinatal exposures, epigenetic hallmarks, and gut microbiome development and neurodevelopmental outcomes within the first three years of life," the authors write.
The mechanistic finding
Francis Ka Leung Chan, a gastroenterologist at CUHK and co-senior author, has spent years studying the gut microbiome and its influence on health. What his team found is that a baby's epigenetic profile at birth predicts the diversity of its gut microbiome by age one. Higher rates of DNA methylation in the major histocompatibility complex, a region of the genome involved in immune recognition, correlated with less diverse microbiomes at twelve months. The epigenome, in other words, sets the conditions for which bacteria can take hold.
The causal arrow also runs the other direction. When the children were assessed at thirty-six months using a behavioral questionnaire, two specific bacteria stood out as protective: Lachnospira pectinoschiza was associated with lower rates of autism signs, and Parabacteroides distasonis with lower rates of ADHD signs, even among infants whose epigenetic profiles at birth put them at elevated risk.
"We discovered a kind of conversation happening: a baby's epigenetic setting at birth can influence their risk for neurodevelopmental disorders, but the presence of certain 'good' bacteria in their gut can step in and modify the risk," said co-senior author Hein Min Tun, a public health researcher at CUHK.
The authors call this integrated host-microbiome system the "holo-epigenome." They argue it should be read together, not as separate contributors to neurodevelopment.
Why birth mode keeps appearing
Cesarean section delivery was associated with distinct DNA methylation changes in genes involved in immune responses and brain development. Those changes, in turn, correlated with differences in early gut colonization. The finding adds mechanistic detail to a body of epidemiological work linking C-section to elevated risks of immune-mediated and neurodevelopmental disorders.
"Infants who were born by Caesarean section showed different patterns of DNA methylation for several genes involved in immune responses and brain development," the authors note. "Some of the changes in methylations of immune- and nervous-system-related genes, associated with CS delivery, are linked to neurodevelopmental outcomes."
This does not mean C-sections cause autism or ADHD. Birth mode was one of several factors associated with epigenetic patterns, alongside length of gestation, whether a child had older siblings, and maternal allergies. The researchers explicitly caution against reading causation into the association.
The first author: Siew Chien Ng
The first author is Siew Chien Ng, a gastroenterologist and director of the Microbiota I-Center at CUHK, also a New Cornerstone Investigator. Ng has spent much of her career building toward clinical translation of microbiome science: she led the development of the MOMmy cohort, an ongoing birth study tracking mother-infant microbiota transmission since 2019, which supplied the data for this analysis. Her lab has been systematically working toward live biotherapeutics: defined microbial treatments rather than vague probiotic formulations.
"The ultimate goal is to develop safe, non-intrusive early interventions such as specific probiotics or live biotherapeutics, that could help nurture a healthy gut microbiome and potentially reduce the risk of neurodevelopmental challenges," Ng said.
The same group published a separate paper this week in Cell Host & Microbe introducing a reference database for infant gut microbiomes. That work is related but reports different findings.
What this means for intervention
The study identifies a biologically plausible window for intervention: the first year of life, when the microbiome is still plastic and the immune system is still learning which organisms belong. By twelve months, the microbial community is largely established. The paper proposes that restoring or supporting specific bacteria during this window could mitigate neurodevelopmental risk seeded by epigenetic patterns at birth.
But several important caveats apply. The associations between specific bacteria and reduced ASD or ADHD signs were observed in a Chinese birth cohort and have not yet been replicated in other populations. Laboratory experiments are needed to confirm the mechanisms, as the authors themselves note. Behavioral questionnaires at thirty-six months are screening tools, not clinical diagnoses. And neurodevelopmental disorders like autism and ADHD are profoundly heterogeneous; no single biological pathway explains all cases.
What the paper argues it adds
The authors position their contribution as establishing the bidirectional relationship between epigenome and microbiome as a combined risk factor for neurodevelopmental outcomes. Previous studies have examined each system separately. This one looks at how the two interact over time and finds that their interplay rather than either system alone predicts which children show signs of ASD or ADHD at age three.
The phrase "holo-epigenome" appears twice in the paper and frames the argument: the host epigenome cannot be understood in isolation from its associated microbiome. Interventions targeting either system alone may be leaving the other half of the equation unaddressed.
The CUHK team is continuing to follow the children in the MOMmy cohort. Whether the thirty-six-month behavioral signals persist and predict clinical diagnoses will take years to answer.
Paper: Epigenome-microbiome interplay in early life associates with infants' neurodevelopmental outcomes. Cell Press Blue. DOI: 10.1016/j.cpblue.2026.100009.
