Prenatal Maternal Stress and Infant Gut Microbiome in Human and Animal Studies: A Comprehensive Systematic Review

ABSTRACT

The infant gut microbiome plays a fundamental role in immune maturation, metabolic programming, and neurodevelopment, with early microbial colonization representing a sensitive period for long‐term health outcomes. Maternal factors during pregnancy powerfully modulate infant microbiome assembly; however, the specific effects of prenatal stress exposure (PSE) on offspring gut microbial communities remain incompletely understood. Here, we present a cross‐species systematic review, the first of this scale, examining associations between prenatal stress and offspring gut microbiome composition. Searches of PubMed (MEDLINE), Scopus, and PsycINFO identified 33 review‐eligible articles published over the past two decades. Across studies, we evaluated stressor characteristics, offspring age at microbiome assessment, and methodological issues relevant to result interpretation. Although the literature is heterogeneous and reports mixed findings, several recurring patterns emerge. Most studies report an increased relative abundance of Proteobacteria and reduced levels of early‐life beneficial taxa from the Actinobacteria and Firmicutes phyla following PSE, driving a negative shift toward a pro‐inflammatory profile and early‐life dysbiosis. At the genus level, decreases in Bifidobacterium, Lactobacillus, Muribaculum, and Parabacteroides are frequently observed, alongside increases in Klebsiella, Enterobacter, and Sutterella, taxonomic shifts that disrupt gut homeostasis and are associated with increased lifelong risks for metabolic, inflammatory, and neurodevelopmental disorders. Differences between human cohorts and animal models regarding alpha diversity patterns and beta diversity clustering are identified. Finally, we explore methodological limitations in the existing literature and discuss recent advances and recommendations to support the continued development of research on prenatal stress, the gut microbiome, and developmental outcomes. Altogether, our study demonstrates that while PSE acts as a powerful upstream ecological modifier of the early‐life gut microbiota across species, its long‐term biological embedding is heavily moderated by sex, developmental timing, and specific environmental context.

1. Introduction

The early establishment of the gut microbiome represents a critical developmental window with far‐reaching implications for long‐term health (Gensollen et al. 2016; Bliss and Whiteside 2018; Sarkar et al. 2021; Borrego‐Ruiz and Borrego 2025). Emerging evidence indicates that the infant gut microbiome plays a central role in immune system maturation (H. Zhang et al. 2022; Gershon and Margolis 2021), metabolic programming (Mulligan and Friedman 2017), and neurodevelopment (Warner, 2019; Maiuolo et al. 2021). Although gut microbial colonization is generally thought to begin at birth (Milani et al. 2017; Korpela and de Vos 2018; Kennedy et al. 2021), with some evidence suggesting prenatal microbial exposure may occur in utero (Aagaard et al. 2014; Rautava et al. 2012; Koren et al. 2012), the assembly of this pioneer microbial community is highly susceptible to maternal influences. In addition to physiological factors such as maternal diet, pre‐pregnancy body mass index, delivery mode, and intrapartum antibiotic use, prenatal stress exposure (PSE)—defined as maternal exposure to psychological, physical, or environmental stressors during pregnancy—has emerged as a potentially powerful, upstream maternal determinant of infant microbiome development. Understanding how PSE shapes the composition and function of the infant gut microbiome may therefore provide critical insight into the developmental origins of health and disease, helping to identify early‐life biomarkers of risk and inform targeted interventions aimed at mitigating stress‐related vulnerabilities before chronic conditions become established.

Pregnancy represents a period of heightened developmental vulnerability during which maternal neuroendocrine and immune responses to stress can substantially alter the intrauterine environment.

A growing body of evidence from both human cohort studies and animal models indicates that PSE has lasting consequences for offspring development. Individuals exposed to prenatal stress show increased behavioral reactivity and physiological stress sensitivity (Kapoor et al. 2008; Mueller and Bale 2008; Hartman et al. 2023; Tung et al. 2024), as well as elevated risks for neurodevelopmental and psychiatric disorders later in life, including attention‐deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD) (O'Donnell et al. 2014; Manzari et al. 2019; Dickerson and Dickerson 2023), schizophrenia (Khashan et al. 2008; Lipner et al. 2019) and cognitive impairment (Morley‐Fletcher et al. 2013). However, the effects of PSE extend beyond the brain. Findings from both human and animal studies demonstrate that prenatal stress can disrupt the development of multiple peripheral physiological systems. Reported consequences include impaired fetal growth and skeletal development, such as intrauterine growth restriction and reduced bone mineral density (Amugongo and Hlusko 2014; Pappalardo et al. 2023), as well as alterations in respiratory (van de Loo et al. 2016; de los Ángeles Aldirico et al. 2023), cardiovascular (J. Gu and Guan 2021), and immune (Veru et al. 2014; Andersson et al. 2016) functions. Emerging evidence also implicates PSE in the disruption of intestinal microbiota development (Mepham et al. 2023; Graf et al. 2025). Collectively, these findings suggest that prenatal stress acts as a broad developmental risk factor, shaping trajectories across multiple organ systems and increasing susceptibility to adverse health outcomes throughout the lifespan.

Stress‐induced disruption of the intestinal environment can trigger downstream effects across the gut–brain axis (GBA), a complex bidirectional communication network linking the gastrointestinal tract and the central nervous system. More recently, this framework has expanded into the microbiota–gut–brain axis (MGBA), which highlights the central role of the gut microbiome in shaping host physiology and behavior (Yassin et al. 2025; Nie et al. 2025). Communication within the MGBA occurs through multiple pathways, including vagal signaling, neuroendocrine and immune mediators, and microbially derived metabolites capable of crossing the blood–brain barrier (Hyland and Cryan 2016; Strandwitz 2018; Y. Yu et al. 2020; Yoo et al. 2020). For example, short‐chain fatty acids regulate microglial and astrocytic development and function (Carabotti et al. 2015; Lynch and Pedersen 2016; Gareau, 2016), thereby influencing neuronal differentiation and myelination (Erny et al. 2015).

Although PSE is widely recognized as a contributor to adverse offspring outcomes through disruption of GBA‐related pathways (Jasarevic et al. 2015; Jasarevic et al. 2018; De Cillis et al. 2025), its effects on offspring gut microbial communities remain incompletely understood. One major challenge is the absence of a universally accepted definition of a “healthy” gut microbiome, reflecting substantial interindividual variation in microbial composition (Lloyd‐Price et al. 2016). Rather than representing a fixed set of taxa, the healthy microbiome is increasingly viewed as a dynamic and individualized ecological equilibrium (Gillingham et al. 2025). This challenge is particularly pronounced during infancy, when the microbiota undergoes rapid developmental succession (Bäckhed et al. 2015). Early microbial communities are characterized by low diversity and enrichment of Bifidobacterium species, including B. infantis, B. breve, and B. longum (Milani et al. 2017), which support nutrient metabolism, gut barrier integrity, and immune maturation (Lawson et al. 2020). As diet diversifies, the microbiome gradually transitions toward a more complex adult‐like configuration (Stewart et al. 2018).

Despite growing interest in early‐life programming, evidence linking maternal stress to offspring microbiome remains fragmented. Previous reviews have focused predominantly on human cohorts and generally report associations between maternal psychological distress and infant dysbiosis, often characterized by increased abundance of Proteobacteria and reduced levels of beneficial Bifidobacteria (Grech et al. 2021; Mepham et al. 2023; Agusti et al. 2023; Ryan et al. 2025). At the same time, these reviews consistently emphasize substantial methodological challenges, including confounding influences of metabolic factors, postpartum exposures, and broader postnatal environments, which complicate the identification of a consistent microbial signature of prenatal stress. While Yeramilli and colleagues (2023) incorporated both human and animal evidence, their synthesis was narrative rather than systematic. More recently, Graf and colleagues (2025) conducted the first cross‐species systematic review using 19 empirical studies and confirmed that prenatal stress alters microbial beta diversity across species. However, the available evidence remained limited primarily to human and rodent studies, with only one primate study (Bailey et al. 2004) and minimal representation of physically induced stress paradigms (Zheng et al. 2020).

To address these limitations, we conduct a comprehensive cross‐species systematic review comprising 33 eligible studies. This review expands the taxonomic and methodological scope of the literature by integrating evidence from humans, rodents, non‐human primates, and, for the first time, porcine models. The inclusion of pigs is particularly valuable because porcine gut ecosystems share approximately 87% of bacterial genera and over 95% of functional metabolic pathways with humans, making them a highly relevant translational model for gut physiology and microbiota‐host interactions (Xiao et al. 2016; Heinritz et al. 2013; Menneson et al. 2019; S.‐J. Wang et al. 2026). In addition, this review systematically evaluates the effects of both psychological and physical prenatal stressors. By comparing studies that examine psychological stress alone with those involving physical stress exposure, we assess whether distinct stressor classes produce different microbiome outcomes, potentially reflecting divergent neuroendocrine, inflammatory, metabolic, or ischemic mechanisms. Finally, our synthesis considers key methodological moderators, including stressor chronicity, offspring age at sampling, biological sample type, and study design characteristics. Together, this approach provides a more comprehensive and translationally relevant framework for understanding how prenatal stress shapes early‐life microbiome development and helps explain inconsistencies across the existing literature.

2. Method

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analysis (PRISMA) guidelines (Page et al. 2021). The protocol for this study was registered on PROSPERO under the number CRD42024527717.

2.1. Search Strategy and Eligibility Criteria

An initial search was conducted to identify articles of interest using the following databases: PubMed (MEDLINE), PsycINFO, and Scopus. Two different sets of search queries were applied:

• −For PubMed: ((“maternally”[All Fields] OR “maternities”[All Fields] OR “maternity”[All Fields] OR “mothers”[MeSH Terms] OR “Maternal Exposure”[MeSH Terms] OR “mother*”[All Fields] OR “maternal”[All Fields] OR “Pregnant Women”[MeSH Terms] OR “prenatal”[All Fields] OR “pregnant” [All Fields]) AND (“stress, psychological”[MeSH Terms] OR “stress, physiological”[MeSH Terms] OR “stress”[All Fields] OR “stressed”[All Fields] OR “stresses”[All Fields] OR “stressful”[All Fields] OR “stressfulness”[All Fields] OR “stressing”[All Fields])) AND (“Infant”[MeSH Terms] OR “Infant”[All Fields] OR “infants”[All Fields] OR “infant s”[All Fields] OR “infant, newborn”[MeSH Terms] OR “Infant”[MeSH Terms] OR “animals, newborn”[MeSH Terms]) AND (“Gastrointestinal Microbiome”[MeSH Terms] OR “Microbiota”[MeSH Terms] OR “Dysbiosis”[MeSH Terms]).
• −For Scopus, PsycINFO: ((perinatal OR pregnant* OR prenatal OR maternal OR mother OR antenatal OR dam) AND (stress OR anxiety OR distress)) AND (infant OR newborn OR neonate OR baby OR babies OR pup) AND (“gut microbio*” OR “gut flora” OR microbiome OR dysbiosis)

The final queries were determined after testing multiple combinations of terms and evaluating the relevance of the articles retrieved from the databases. The selected combination was assessed and compared with alternative options to ensure it was the most comprehensive and accurate in addressing the research question. Searches were performed in December 2023 and then updated in January 2026; therefore, no studies published after January 2026 are included in the review.

2.2. Inclusion and Exclusion Criteria

Studies were eligible for inclusion if they met the following criteria: (1) were original empirical studies; (2) involved human or animal subjects; (3) reported data on offspring microbiome composition; (4) examined prenatal exposure to physical, psychological, or combined stress; and (5) in animal studies, included an appropriate control group.

2.2.1 Concept of Stress

Rooted in classical adaptation models and modern allostatic load theory, stress is a broad, neutral response to environmental demands that challenge homeostasis. Within an organism's coping capacity, stress can facilitate adaptation and resilience. Distress, in contrast, represents a maladaptive state that arises when the intensity or duration of stress exceeds available coping resources, often manifesting as internalized conditions such as anxiety and depression. In this review, prenatal stress in humans was conceptualized as a multidimensional construct encompassing both objective environmental stressors and maternal perceptions of those stressors during pregnancy. Accordingly, eligible exposures included perceived stress, symptoms of maternal depression and anxiety, intimate partner violence (IPV), food insecurity, low social support, social disadvantage, experiences of discrimination, and biological indicators of stress such as cortisol concentrations. Although anxiety and depression are typically classified as forms of psychological distress, they were analyzed alongside broader psychosocial stressors because they reflect related pathways of maternal stress exposure. In contrast, studies examining only physical exposures, such as environmental toxins, drugs, or medications, were excluded. For animal models, a broader definition was employed to capture experimentally induced and naturally occurring forms of maternal stress. Eligible stressors included physical challenges, such as temperature extremes, high humidity, and food or water restriction, as well as environmental exposures when accompanied by a psychological stress component. Studies of both laboratory manipulations and natural environmental adversity in wild populations were included. However, studies using only toxic substances, drugs, and medications as stressors were excluded. Although such exposures can adversely affect fetal development, they do not necessarily elicit the maternal neuroendocrine stress response that was central to our conceptualization of prenatal stress.

2.2.2 Age limit

In human studies, offspring age was restricted to the first 2 years of life. Although the gut microbiome continues to develop throughout early childhood, the most pronounced compositional changes occur during the first ∼28 months, after which the microbial community begins to stabilize (Stewart et al. 2018). This age range was selected for several reasons. First, it corresponds to the “first 1,000 days” framework, widely recognized as a critical window for developmental programming (Indrio et al. 2023) and long‐term health trajectories. Second, by approximately 2 years of age, most children have completed the transition from milk‐based feeding to solid foods (Moore and Townsend 2019). As a result, microbiome composition beyond this point is increasingly shaped by postnatal dietary and environmental exposures, potentially obscuring the effects of prenatal influences. Restricting eligibility to ages 0–2 years therefore allows us to examine the gut microbiome during its most dynamic and developmentally sensitive period, when susceptibility to prenatal programming is likely greatest.

2.2.3 Exclusion Criteria

Studies were excluded if they met any of the following criteria: (1) were not primary research articles (e.g., systematic reviews, scoping reviews, meta‐analyses, opinion pieces, or protocols); (2) did not include a stress‐related exposure or measure; (3) did not report offspring microbiome outcomes; (4) involved genetically modified animals; (5) included human offspring older than 2 years of age; (6) involved mothers with gastrointestinal symptoms or disorders; or (7) were published in a language other than English, Russian, Arabic, French, or Spanish. Studies in which all infants received microbiome supplementation were also excluded. When supplemented and non‐supplemented groups were both included, only data from the non‐supplemented groups were extracted.

2.3. Selection of Studies

All records were retrieved from online databases and imported into the PICO Portal (New York, NY, USA, available at www.picoportal.org), a platform designed to support the management of systematic reviews and other evidence‐synthesis projects. Duplicate records were identified and removed within the platform. Initial title and abstract screening were conducted independently by three researchers, with disagreements resolved through discussion and consensus. Full‐text articles were then assessed against a predefined set of inclusion and exclusion criteria by two independent reviewers. Any discrepancies were adjudicated by a third researcher.

2.4. Data Extraction

Data extraction was conducted independently by three researchers, and all extracted information was subsequently cross‐checked by a fourth researcher to ensure accuracy. A common set of variables was collected for all included studies, including species (human or animal), maternal and offspring sample size, offspring sex, offspring age at specimen collection, microbiome sample source (e.g., fecal sample or colon tissue), microbiome assessment method, and reported changes in microbial composition and relative bacterial abundance. Additional information was extracted for animal studies, including the number of male and female offspring, experimental and control conditions and their respective sample sizes, the type of maternal stressor, and the timing of stress exposure. For human studies, supplementary variables included recruitment location, cesarean section status, antibiotic exposure, the name and construct of the psychological assessment used, timing of assessment, subjective stress measures, pregnancy‐related stressors and their classification, and the type of biological sample used to assess stress.

2.5. Assessment of Risk of Bias

Study quality and risk of bias were assessed using the Risk of Bias in Non‐randomized Studies—of Exposure (ROBINS‐E) tool (Higgins et al. 2024; available at https://www.riskofbias.info/welcome/robins‐e‐tool). All eligible studies were evaluated independently by two researchers with disagreements resolved by a third reviewer. Risk of bias was assessed across the following domains: confounding, exposure measurement, post‐exposure interventions, missing data, outcome measurement, and selection of reported results. Each domain, as well as the overall risk of bias, was rated as “low risk,” “some concerns,” or “high risk.” Studies were classified as having an overall high risk of bias when at least two domains received a “high risk” rating.

2.6. Strategy for Data Analysis and Analysis of Subgroups

Owing to the heterogeneity of the literature, findings were synthesized narratively, focusing on the associations reported in individual studies. Studies were categorized by species (human vs. animal), and relevant variables included stressor timing, stressor chronicity, timing of specimen collection, and specimen type.

3. Results

3.1. Search Results

The study selection process is summarized in Figure 1. The initial database search yielded 462 publications after duplicates were removed, and no non‐English publications were identified. Following title and abstract screening, 114 publications remained for full‐text review. Full‐text screening excluded 88 publications, resulting in 26 eligible studies. Reference screening of these studies identified an additional seven publications, yielding a final sample of 33 studies included in the review.

FIGURE 1.

PRISMA flow diagram depicting the selection of studies.

Of these, 14 involved human subjects and 19 involved animal subjects; results are presented separately for each group. The publication dates ranged from 2004 to 2025.

3.2. Included Studies

3.2.1. Study Characteristics and Methodologies

Nineteen animal studies were included (Table 1; Supporting Information File), comprising research in pigs (n = 1), macaques (n = 2), rats (n = 7), and mice (n = 9). Across these studies, the reported number of dams was 10 for pigs, 30 for macaques (excluding one study that did not report the number; Bailey et al. 2004), 137 for rats (excluding one study that did not report the number; Hua et al. 2023), and 335 for mice (excluding one study that did not report the number; Gur et al. 2019). The total number of offspring reported was 10 for pigs; 54 for macaques, including 23 females (Bailey et al. 2004; Anza et al. 2023); 323 for rats, including 90 females (excluding Golubeva et al. 2015); and 283 for mice, including 108 females (excluding studies by Jasarevic and colleagues: Jasarevic et al. 2015, 2017, 2018). Two studies analyzed only male offspring (n = 29, Gur et al. 2019; number is not reported, Golubeva et al. 2015), whereas another included only female offspring, n = 32 (Gur et al. 2017).

TABLE 1.

Characteristics of included animal studies.

Species	Type of stress	Timing of the stress	Stressor	Sample collection timing	Reference
Pig	Chronic physical stress	E85 to farrowing	Heat stress	Meconium	He et al. (2020)
Macaque	Elevated stress of unspecified nature	Various periods of pregnancy	Food insecurity	Infant (< 1 year)	Anza et al. (2023)
	Psychological maternal stress	E50 to E92 for the “early stress” group, E105 to E147 for the “late stress” group; five times per week	Random noise in the darkened room	PND 2, 14, 56, 112, 168	Bailey et al. (2004)
Rat	Chronic physical stress	5 h a day during the entirety of gestation	High‐temperature housing	PND 7, 14, 22, 28	Adebiyi et al. (2022)
		E14 till birth	Cold‐temperature housing	PND 21	Zheng et al. (2020)
	Physical and psychological stress	Three times a day on the last week of pregnancy	Restraint and bright light	PND120 (4 months)	Golubeva et al. (2015)
		E5 to E21	Forced swimming, restraint, food deprivation, damp bedding	PND2	Pawluski et al. (2023)
		Lead exposure from 2 weeks before gestation and throughout pregnancy; restraint stress throughout pregnancy	Restraint; lead (Pb) exposure in drinking water	PND21	Hua et al. (2023)
		3 days before mating to the end of pregnancy, once a day	Food and water deprivation, overcrowding, high humid environment, tail clip, shake, swimming, restraint, heat stress	PND20	Zhao et al. (2021)
		E14 to E21, three times a day for 45 min	Restraint and bright light	PND84	De Cillis et al. (2025)
Mice	Physical and psychological stress	E0.5 to E19.5, 2 h a day	Immobilization	PND48–56 (6–8 weeks)	Z. Zhang et al. (2021)
		E7 to E20, 2 h a day every day	Restraint and bright light	PND14	Brawner et al. (2020)
		E10 to E16, 2 h every day	Restraint	Not specified	Gur et al. (2017)
		E10 to E16, 2 h every day	Restraint	PND70	Gur et al. (2019)
		E0 to E7	Set of seven stressorsa	PND2	Jasarevic et al. (2015)
		E1 to E7	Set of seven stressorsa	PND 2, 6, 28	Jasarevic et al. (2017)
		E1 to E7	Set of seven stressorsa	PND2	Jasarevic et al. (2018)
		DEL exposure from E2 till birth, 3 h a day; nest restriction from E13 till birth	Diesel exhaust particles (DEL) exposure and nest restriction	PND45	Smith et al. (2023)
		E10 till birth	Restraint, damp bedding overnight, lights overnight, white noise, water avoidance	PND21, 56	Sun et al. (2021)

Abbreviations: E—embryonic day; PND—postnatal day.

^aFox odor, restraint, constant light, novel noise overnight, three cage changes throughout the light cycle, saturated bedding overnight, and novel object exposure overnight.

For animal studies, offspring counts included only pups designated for microbiome analyses when experimental litters were subdivided for multiple research purposes. Reporting practices varied considerably, with some studies presenting the number of individual animals and others reporting the number of litters. Sample sizes were extracted from the main text, figure legends, and supplementary materials. Additional inconsistencies arose because some studies included only male or only female offspring, whereas others did not report sex‐specific information. Reported control and experimental offspring numbers were five and five for pigs, 154 and 157 for rats (excluding Golubeva et al. 2015, and Hua et al. 2023), 104 and 147 for mice not including studies from Jasarevic's research group (2015, 2017, 2018) and Brawner (2020), and nine and 45 for macaques.

Experimental conditions for pregnant dams varied considerably across studies. Fourteen of the 19 articles employed both physical and psychological stressors, whereas three used chronic physical stress alone and one used only psychological stress. One additional study examined the effects of naturally occurring elevated maternal stress rather than experimentally induced stress in a controlled setting. Maternal stressors included heat and cold exposure, food and water deprivation, overcrowding, high humidity, tail clipping, forced swimming, shaking, restraint, predator odor exposure, constant light, loud noise, darkness, exposure to novel objects, frequent cage changes during light cycles, toxic substance exposure, bright light, nest restriction, damp or saturated bedding, food insecurity, or combinations of these factors. Although nutritional stress was represented by food deprivation and food insecurity, no studies examined overnutrition. More than half of the eligible studies (11 of 19) employed multiple stressors. Stress exposure either spanned the entire gestational period (3 of 19 studies) or was restricted to a specific portion of gestation (16 of 19 studies).

The source, timing, and frequency of microbiome sample collection also varied substantially across studies. Offspring microbiota samples were obtained from feces (11 studies), gut content (1 study), various digestive tract tissues including the ileum, colon, and duodenum (7 studies), and rectal swabs (1 study). The earliest samples were collected immediately after birth as meconium (1 study on pigs), whereas the latest were collected nearly 6 months postpartum (2 studies in macaques). Among mice, stool samples were mostly collected two days after birth (3 of 9 studies), whereas for rats they were most frequently collected 20–22 days postpartum (4 of 7 studies). A range of approaches was used to characterize the microbiome. Most studies (17 of 19) employed 16S rRNA gene sequencing, while one study used quantitative polymerase chain reaction (PCR) with 16S rRNA‐targeted primers for selected intestinal bacteria, and another relied on a culture‐based method.

Fourteen human studies met inclusion criteria, representing a total sample of 2480 mother–infant dyads (Table 2; Supporting Information File). Among the 11 studies that reported infant sex (78.5% of the total sample), 676 infants (44.7%) were female. Participants were recruited from diverse geographical regions, including China (n = 875), Singapore (n = 450), Finland (n = 399), the United States (n = 293), the Netherlands (n = 207), Switzerland and Germany (n = 100), South Africa (n = 84), The Republic of the Congo (n = 47), and San Cristobal Island in the Galapagos, (n = 25).

TABLE 2.

Characteristics of included human studies.

Psychological assessment	Timing of assessment	Construct	Sample collection timing	Reference
1) EPDS 2) Symptom Checklist 90, anxiety subscale 3) Daily Hassles (Korpela), negative scale 4) PRAQ‐R2	Gestational weeks 14, 24, and 34	1) Depression 2) Anxiety 3) Negative affect 4) Pregnancy‐related stress	2.5 months old	Aatsinki et al. (2020)
1) EPDS 2) State/Trait Anxiety Inventory 3) Perceived Stress Scale	Gestational weeks 34–36	1) Depression 2) State and trait anxiety 3) Stress broadly	24 h postpartum	Deflorin et al. (2024)
1) Violence Trauma and Pregnancy Trauma Questionnaires 2) Early Trauma Inventory Self Report Short Form, General Trauma and Sexual Trauma subsections 3) Perceived Stress Scale 4) Impact of Event Scale to measure PTSD, intrusion subscale 5) EPDS 6) State/Trait Anxiety Inventory, 20 state anxiety questions	Within 1 day of delivery	1) Experiences of violence and pregnancy‐related trauma 2) Early trauma 3) Stress broadly 4) Trauma intrusions 5) Depression 6) State anxiety	6 weeks, 3 months, 6 months old	Dutton et al. (2023)
1) Overall Anxiety Severity and Impairment Scale 2) PHQ‐9 3) Perceived Stress Scale	Gestational weeks 8–16, 20–26, 30–36	1) Anxiety 2) Depression 3) Stress broadly	4–8 weeks old, 5–7 and 11–13 months old	Galley et al. (2023)
1) EPDS 2) State/Trait Anxiety Inventory 3) PRAQ‐R 4) Perceived Stress Scale 5) Psychiatric Epidemiology Research Interview Life Events	Second trimester	1) Depression 2) State and trait anxiety 3) Pregnancy‐related stress 4) Stress broadly 5) Universal life events	48 h postpartum	Hu et al. (2019)
1) Latin American and Caribbean Food Security Scale 2) Perceived Stress Scale 3) PHQ‐8 4) Perceived Social Support‐Family and Friends	Gestational weeks 34–36	1) Food insecurity 2) Stress broadly 3) Depression 4) Family support	2 months old	Jahnke et al. (2021)
State Trait Anxiety Inventory, Form Y	Gestational weeks 26–28	State and trait anxiety	2 years old	Querdasi et al. (2023)
1) EPDS 2) Perceived Stress Scale 3) Everyday Discrimination Scale 4) Social Disadvantage Score	Each trimester	1) Depression 2) Stress broadly 3) Discrimination 4) Social disadvantage	4 months old	Warner et al. (2023)
1) Center for Epidemiologic Studies Depression Scale 2) Self‐Rating Anxiety Scale	Gestational weeks 32–36	1) Depression 2) Anxiety	24 h postpartum	Wei et al. (2022)
Perceived Stress Scale	Third trimester	Stress broadly	1 month old	Weiss and Hamidi (2023)
1) State‐Trait Anxiety Inventory, state anxiety subscale 2) PRAQ‐R 3) Daily Hassles 4) Pregnancy Experience Scale	Third trimester	1) State anxiety 2) Pregnancy‐related stress 3) General stress + level of bother 4) Pregnancy‐related uplifts and hassles	9 time points from birth to 3.7 months old	Zijlmans et al. (2015)
1) Beck Depression Inventory 2) The modified PTSD Symptom Scale 3) The IPV questionnaire 4) SRQ‐20 questionnaire	Gestational weeks 27.4 ± 4.2	1) Depression 2) PTSD 3) Lifetime and recent (past‐year) exposure to emotional, physical, and sexual IPV 4) Psychological distress	24 h postpartum	Naudé et al. (2020)
1) EPDS 2) Perceived Stress Scale‐10 3) State‐Trait Anxiety Inventory 4) PRAQ‐R2	Gestational weeks 18 and 32	1) Depression 2) Stress broadly 3) State anxiety 4) Pregnancy‐related stress	2, 6, and 12 weeks and 8 months old	Eckermann et al. (2025)
1) EPDS 2) Generalized Anxiety Disorder 7‐ item Scale 3) PRAQ 4) Perceived Stress Scale‐4 5) The Fear of Birth Scale 6) Prenatal Life Events Scale 7) Adverse Childhood Experience Questionnaire‐Revised	Gestational weeks 12 and 20	1) Depression 2) State anxiety 3) Pregnancy‐related stress 4) Stress broadly 5) Fear of birth 6) Life events during pregnancy 7) Exposure to adverse childhood experiences	48 h postpartum	X. Zhang et al. (2025)

Abbreviations: EPDS—Edinburgh postnatal depression scale; IPV—intimate partner violence; PHQ‐9—Patient Health Questionnaire‐9; PRAQ‐R—Pregnancy‐Related Anxiety Questionnaire–Revised; PTSD—post‐traumatic stress disorder; SRQ‐20—Self‐Reporting Questionnaire‐20.

Across the human studies, maternal psychological stress was the most commonly examined exposure. Eight studies used broad self‐report measures of stress, while depressive symptoms were assessed in 11 studies, most commonly using the Edinburgh Postnatal Depression Scale, Patient Health Questionnaire, or the Beck Depression Inventory. Anxiety was examined in eight studies, typically using the State‐Trait Anxiety Inventory, and three of these studies additionally assessed pregnancy‐related anxiety (i.e., Pregnancy‐Related Anxiety Questionnaire–Revised). Additionally, trauma exposure and post‐traumatic stress disorder (PTSD) were evaluated in two studies.

The assessment windows for the self‐reported stress generally captured symptoms experienced during the preceding week to month. Studies focusing exclusively on lifetime stress or pre‐pregnancy stress exposure were excluded, with the exception of one study examining adverse childhood experience (Z. Zhang et al. 2021). Alternative indicators of stress were less commonly investigated. For example, one study examined food insecurity during pregnancy (Jahnke et al. 2021), while another evaluated exposure to violence and trauma during pregnancy (Dutton et al. 2023). In the latter study, stress indicators were combined into a composite score rather than analyzed individually. Six studies measured cortisol as a physiological indicator of stress, with samples obtained from saliva (n = 3), hair (n = 2), or blood (n = 1).

Considerable variability was observed in the timing of both maternal stress assessment and infant fecal sample collection. Maternal stress was most frequently assessed during the second (n = 7) and third (n = 12) trimesters, while two studies included first‐trimester measurements. Three studies measured maternal stress at multiple gestational time points. Infant fecal samples were collected from birth (meconium) through 2 years of age, and four studies included repeated sampling across this period. Most studies (n = 13) characterized the microbiome using 16S rRNA sequencing, whereas one study employed whole‐genome metagenomic sequencing (Eckermann et al. 2025).

3.2.2. Risk of Bias

All studies were judged to have a low risk of bias in the following domains: bias arising from measurement of the exposure, bias due to post‐exposure interventions, bias in measurement of the outcome, and bias in selection of the reported results. No major factors likely to introduce bias were identified in these domains. For example, although outcome assessment in the human studies relied largely on self‐reported measures of stress, the questionnaires used are widely accepted as reliable and valid tools. Only a small number of studies supplemented self‐reports with biological measures of stress, such as cortisol levels (Zijlmans et al. 2015; Aatsinki et al. 2020; Jahnke et al. 2021; Galley et al. 2023; Deflorin et al. 2024; Eckermann et al. 2025).

Comments related to the selection of reported results primarily concerned studies that reported outcomes for only female or only male offspring (Gur et al. 2017, 2019), although the authors indicated that results for the other sex were presented in separate publications. Four studies were rated as raising some concerns regarding bias due to confounding because they did not report information on antibiotic treatment (Jahnke et al. 2021; Wei et al. 2022; Querdasi et al. 2023; Galley et al. 2023; X. Zhang et al. 2025), a potentially important covariate in microbiome research. The study by Naude and colleagues (2020) was rated as having some concerns for bias due to missing data because of substantial participant attrition over the course of the study. After evaluating all seven bias domains, overall risk‐of‐bias judgments were assigned.

Despite the minor concerns noted above, all 33 papers were ultimately judged to be at low overall risk of bias.

3.3. Main Findings: Microbiome Changes Associated With Maternal Prenatal Stress

Assessment of the gut microbiome commonly focuses on bacterial composition and function, which are evaluated using measures of alpha diversity, beta diversity, and taxonomic abundance. Definitions of these key metrics, including the diversity indices referenced throughout this review, are provided in Table 3. Figure 2 presents a visual summary of the patterns of infant gut microbiome alterations observed across the reviewed studies.

TABLE 3.

Microbiome metrics.

Measure	Definition and interpretation
Alpha diversity	A within‐sample measure that reflects the variety of species present in local, homogeneous habitats (e.g., gut). A higher value indicates greater within‐individual microbial diversity
Richness	The number of distinct bacterial species present within a single sample
Evenness	A measure of how equally abundant the different bacterial species are within a sample
Chao1 index	A measure of species richness. It accounts for species that may be present in a sample but were not detected (due to insufficient sensitivity of the method), thus providing a more accurate estimate of true community richness. A higher value reflects a larger number of bacterial species
Sobs index	A measure of species richness. It estimates the number of species based on the number of observed sequences. A higher value reflects a larger number of bacterial species
Faith's phylogenetic diversity index	A measure of alpha diversity that calculates the total evolutionary history found within a single sample (evolutionary richness). A higher score means that a sample contains bacteria from many different branches of the evolutionary tree, rather than just many similar species from the same family
Shannon index	A key metric to measure both the richness and evenness of bacterial species present in a sample. A higher value indicates a greater number of species that are relatively evenly distributed
Simpson index	A metric to measure both the richness and evenness of bacterial species present in a sample. For evenness, it gives more weight to highly abundant, dominant species. A higher value indicates a community with high species richness but strong dominance by a few species
Pielou evenness	A metric that quantifies how evenly individual bacteria are distributed across species in a sample, ranging from 0 (no evenness) to 1 (complete evenness). Pielou evenness is calculated using the Shannon index.
Beta diversity	A measure that evaluates differences in microbiota composition between two (or more) individuals. Higher values indicate that two communities are less similar
Bray–Curtis dissimilarity	A key metric to assess beta diversity. It quantifies the differences in microbial communities across different samples. Higher values indicate less level of similarity
Weighted UniFrac distance	A metric of similarity/dissimilarity between microbial communities that incorporates both relative abundance data and phylogenetic relationships. Higher weighted UniFrac distances mean two microbial communities are less similar, looking at both the abundance of the bacteria and how closely related they are
Jaccard distance	A measure of dissimilarity between two distinct samples. It evaluates the data on a binary scale: presence or absence (whether a bacterial species exists in the samples or not). Higher Jaccard distance means two microbial communities are less similar

FIGURE 2.

Heat map reflecting gut microbiome changes in offspring exposed to prenatal stress. Red markers indicate a relative decrease in quantity, green markers indicate an increase, and orange markers indicate a mixed signal (e.g., a decrease in females and an increase in males). For articles that studied taxa higher than the genus level, we marked all the genera belonging to the family/order/class/phylum for which changes were reported. The heat map includes only those bacteria that were mentioned in two or more studies. For the full list of bacteria, see Tables 4 and 5. V = Verrucomicrobia.

When describing the gut microbiome alterations, p‐values are reported exactly as presented in the original publications and are not rounded to a consistent number of decimal places.

3.3.1. Animal Studies

3.3.1.1. Alpha Diversity

The effects of PSE on offspring alpha diversity are highly variable across mammalian models (see Table 4). Rather than producing a consistent directional change, PSE appears to induce dynamic and often transient alterations that depend on multiple factors, including species, the timing of gestational exposure, type of maternal stressor, specimen source, offspring sex, and age at sampling. Evidence suggests that the timing of stress during gestation is particularly important. In wild Assamese macaques (Macaca assamensis), naturally elevated maternal glucocorticoid levels during early gestation are associated with a significant long‐term reduction in infant fecal bacterial richness as measured by Faith's Phylogenetic Diversity index (p = 0.018; Anza et al. 2023). In contrast, elevated maternal stress during late gestation has no detectable effect on offspring alpha diversity (Anza et al. 2023). Similarly, in rats exposed to combined physical and psychological stress throughout most or all of pregnancy, reduced fecal alpha diversity is observed during the juvenile period (PND 20–21), as reflected by lower Sobs, Chao1, and Simpson indices (p < 0.05; Zhao et al. 2021; Hua et al. 2023). However, comparable effects are not detected at earlier neonatal stages (PND2) or at later developmental timepoints (PND84 and PND120), particularly when stress is restricted to late gestation (Golubeva et al. 2015; De Cillis et al. 2025; Pawluski et al. 2023). A broadly similar pattern has been reported in mice.

TABLE 4.

Microbial changes associated with prenatal maternal stress in animal studies.

Reference	Species	Changes in microbial diversity	PSE group
Jasarevic et al. (2015)	Mice	—	(CS) ↓Firmicutes: Lactobacillus ↑Firmicutes: Clostridium ↑Bacteroidetes: Bacteroides
Jasarevic et al. (2017)		Alpha diversity changes were dynamic, shifting in a sex‐specific and developmental stage‐dependent manner (Shannon index, Chao1 index); beta diversity significantly differed between stressed/control males at PND28 (UniFrac distance, p < 0.01)	(CS) (PND2)↓Firmicutes: Lactobacillus, Streptococcus (PND28) ↑Desulfovibrio (PND28) ↓ Proteobacteria: Flexispira (PND28) ↑Firmicutes: Dehalobacterium, Lachnospiraceae (m), Clostridiales (m)
Jasarevic et al. (2018)		Alpha diversity (Shannon index) did not show uniform direct shifts from prenatal stress alone; beta diversity significantly differed between stressed/control animals (UniFrac distance, p = 0.001)	(CS) ↓Proteobacteria: Escherichia coli ↓Firmicutes: Streptococcus acidominimus, Streptococcus thoraltensis ↑Firmicutes: Peptococcaceae
Gur et al. (2017)		No significant difference in alpha diversity (Shannon index, Chao1 index); beta diversity significantly differed between stressed/control females (t (30) = 3.83, p < 0.001), (males not studied)	(R) ↓Bacteroidetes ↓Bacteroidetes: Rikenellaceae, S24‐7 ↓Firmicutes ↓Actinobacteria: Bifidobacteriaceae
Gur et al. (2019)		No significant difference in the alpha diversity (Shannon index, Chao1 index); beta diversity significantly differed between stressed/control males (UniFrac distance, p < 0.05), (females not studied)	(R) ↓Bacteroidetes; Bacteroides, Parabacteroides
Brawner et al. (2020)		Alpha diversity was greater in stressed females, but not in males (observer features, Shannon index, p = 0.0016); beta diversity significantly differed between stressed/control females, but not males (UniFrac distance, p = 0.0016)	(R+L) ↑Firmicutes: Lachnospiraceae (f), Clostridiales (f)
Z. Zhang et al. (2021)		Both alpha (Chao1, Shannon indices, Abundance‐based Coverage Estimator, p < 0.05) and beta diversity (weighted UniFrac distances, p < 0.001) significantly differed between stressed/control animals	(Ι): ↑Bacteroidetes: Prevotellaceae, Bacteroidaceae, Bacteroides, Alloprevotella ↑Firmicutes: Lactobacillus ↓Bacteroidetes: Muribaculaceae
Sun et al. (2021)		No significant difference in the alpha diversity was observed; beta diversity significantly differed between stressed and control groups	(CS): ↑Desulfovibrio ↑Firmicutes: Streptococcus, Enterococcus ↓Firmicutes: Blautia, Robinsoniella ↓Actinobacteria: Bifidobacterium
Smith et al. (2023)		Community evenness was significantly increased in stressed males, but not in females (Pielou's evenness, p = 0.02); beta diversity significantly differed between stressed/control males (weighted UniFrac dissimilarity, p = 0.03), but not in females	—
Golubeva et al. (2015)	Rats	There were no differences in the alpha diversity between groups; beta diversity differed significantly in males (p < 0.05), (females not studied)	(R+L): ↓Firmicutes: Lactobacillus, Streptococcaceae ↑Firmicutes: Oscillibacter, Anaerotruncus, Peptococcus
Zheng et al. (2020)		No statistically significant differences in baseline alpha diversity metrics were found (Chao1, Shannon indices, p > 0.05); beta diversity significantly differed between groups (p < 0.05)	(C): ↑ Firmicutes: Lactobacillus (f), Lactobacillus_gasseri (f), ↑Bacteroidetes: Bacteroides (f), Bacteroides‐acidifaciens (f) ↓Firmicutes: Lachnospiraceae (m), ↓Bacteroidetes: Prevotellaceae (m)
Zhao et al. (2021)		Alpha diversity was decreased in the stressed group (Chao1, Shannon indices, p < 0.05); beta diversity significantly differed between stressed and control groups (UniFrac, p < 0.05)	(CS): ↑Verrucomicrobia ↓Firmicutes: Lacobacillus
Adebiyi et al. (2022)		—	(Η) ↑Firmicutes
Pawluski et al. (2023)		There were no differences in alpha diversity between groups (Chao1, Shannon indices); beta diversity significantly differed between groups (Bray–Curtis distance, UniFrac distance, p < 0.05)	(CS): ↓Bacteroidetes ↓Actinobacteria
Hua et al. (2023)		Alpha diversity was decreased in a group that received Pb + psychological stress, comparing with control group and a group that received Pb only (Sobs, Chao1, and Simpson indices were lower, Shannon index was higher, p < 0.05); beta diversity significantly differed between control group and a group that received only psychological stress (Bray–Cruis and weighted UniFrac distances, p < 0.05), beta diversity between the Pb and Pb + psychological stress groups showed overlap	↓Bacteroidetes (R, R+Pb, Pb) ↑Odoribacter (R+Pb) ↑Bacteroides (R+Pb) ↑Firmicutes: Lactobacillus (R, R+Pb, Pb) ↓Firmicutes: Turicibacter (R+Pb), Allobaculum (R+Pb), Veillonella (R+Pb), Clostridium (R+Pb), Oscillospira (R+Pb) ↑Actinobacteria: Bifidobacterium (R+Pb) ↓Actinobacteria: Corynebacterium (R+Pb) ↑Proteobacteria: Helicobacter (R, R+Pb, Pb) ↓ Proteobacteria: Oligella (R+Pb), Sutterella (R+Pb), Escherichia (R+Pb), Psychrobacter (R+Pb), Flexispira (R+Pb) ↓Mycoplasma (R+Pb)
De Cillis et al. (2025)		There were no differences in the alpha diversity between groups in colonic crypts; in luminal content, beta diversity was increased in stressed animals (UniFrac and Jaccard distances, p ≤ 0.029).	(R+L): ↓Verrucomicrobiota ↓Akkermansia (m) ↓Firmicutes: Anaerostipes (m), Lachnospiraceae (m), Clostridium sensu stricto 1, Roseburia (f), Eubacterium coprostanoligenes (f), ↑Firmicutes: Anaerotruncus, Anaerovorax (m), Auricoccus‐Abyssicoccus (m), Defluviitaleaceae (m), Eubacterium coprostanoligenes (m), Erysipelatoclostridium (f), Anaerostipes (f), Butyricicoccus (f), Incertae Sedis (f), Lachnoclostridium (f) ↓Actinobacteria: Cutibacterium (m) ↑Actinobacteria: Corynebacterium (m) ↓Proteobacteria: Escherichia‐Shigella (f) ↓Bacteroidetes: Bacteroides (m), Muribaculum (f) ↑Patescibacteria ↑Candidatus Saccharimonas (m) ↑Desulfobacterota (f)
Bailey et al. (2004)	Macaques	—	(N): ↓Firmicutes: Lactobacillus ↓Actinobacteria: Bifidobacteria ↓aerobes ↓facultative anaerobes Shigella flexneri
Anza et al. (2023)		Reduced gut‐bacterial richness (Faith's index, p = 0.018) was associated with elevated maternal stress during early gestation; beta diversity was reshaped by early prenatal stress (p = 0.021). Elevated stress levels during late pregnancy did not have a significant effect on either alpha or beta diversity	(FI): ↓Firmicutes/Bacteroidetes ratio
He et al. (2020)	Pigs	Beta diversity, but not alpha diversity, significantly differed between stressed/control animals (Bray–Curtis and weighted UniFrac distances, p < 0.05)	(H): ↑Proteobacteria: Acinetobacter, Klebsiella, Stenotrophomonas, Comamonas, Escherichia‐Shigella ↓Firmicutes: Clostridium sensu stricto 1, Romboutsia, Turicibacter

Note: – data on alpha/beta diversity are not reported in the paper; (C)—cold; (CS)—complex of stressors; for details, see the Supporting Information File; (f) demonstrates that this effect was noticed for females, but not for males; (FI)—food insecurity; (H)—heat; (I)—immobilization; (L)—light; (m) demonstrates that this effect was noticed for males, but not for females; (N)—noise in a dark room; (Pb)—lead; (R)—restraint; PND—post‐natal day; ↑ positive association with prenatal stress exposure; ↓ negative association with prenatal stress exposure.

Studies restricting prenatal stress exposure to either early (Jasarevic et al. 2015, 2018) or late gestation (Gur et al. 2017, 2019; Sun et al. 2021) generally found no significant differences in alpha‐diversity indices measured across developmental stages ranging from PND2 to PND70. However, these findings contrast with longitudinal evidence from a single study showing that early‐gestation stress can produce highly dynamic and non‐linear changes in alpha diversity, with effects varying by both offspring sex and developmental stage (Shannon index, Chao1 index, Jasarevic et al. 2017). A similar lack of effect is observed when physical stressors are examined in isolation. Baseline alpha‐diversity metrics remain unchanged following prenatal thermal stress in both swine and rodent models, including maternal heat stress during late gestation in pigs (He et al. 2020) and cold‐temperature exposure from embryonic day 14 until birth in rats (Zheng et al. 2020). In contrast, studies combining physical and psychological or chemical stressors reveal more complex interactions. For example, the combined effect of prenatal lead exposure and restraint stress was associated with a lower Simpson index than lead exposure alone (p < 0.05), while simultaneously producing a higher Shannon index (p < 0.05) relative to the lead‐only group (Hua et al. 2023). Alpha‐diversity outcomes also vary when analyses are stratified by offspring sex or the anatomical source of the microbiome sample. In mice exposed to gestational restraint and bright light, female offspring exhibited greater microbial richness than controls, as reflected by Shannon indices (p = 0.0016), whereas no significant changes were detected in males (Brawner et al. 2020). The opposite pattern was reported in a model combining diesel exhaust exposure with nest restriction, where no significant changes were detected in males (p = 0.02, Pielou's evenness), while females remained unaffected (Smith et al. 2023). Anatomical sampling location further contributes to this variability. In rats exposed to chronic physical and psychological stressors, alpha‐diversity measures were unchanged within microbial communities residing in the colonic crypts but differed in luminal samples collected from the same animals (De Cillis et al. 2025). Together, these findings indicate that the effects of prenatal stress on alpha diversity are highly context‐dependent and influenced by factors including stressor type, developmental timing, offspring sex, and the microbial niche being examined.

3.3.1.2. Beta Diversity

The effects of maternal stress on offspring microbiome most consistently manifest as alterations in overall community composition rather than within‐sample diversity. Among the 16 studies reporting beta‐diversity outcomes, all observed distinct microbial community clustering according to the experimental group, regardless of the distance metric applied. Evidence from directly comparing different stressor modalities further suggests that these compositional changes may depend on the nature of preliminary exposure. For example, beta diversity differed significantly between control animals and offspring exposed exclusively to psychological stress (p < 0.05, Bray–Curtis and weighted UniFrac distances), whereas substantial overlap was observed between microbial communities from the lead‐only and lead‐plus‐restraint groups (Hua et al. 2023). This finding suggests that the addition of psychological stress to prenatal lead exposure may not further alter overall community structure, despite affecting other aspects of microbial diversity. Notably, several murine studies indicate that beta‐diversity responses may be sex dependent (Brawner et al. 2020; Smith et al. 2023), although such effects were not detected in six other mouse studies. Interestingly, the two studies reporting sex‐specific differences identified opposite patterns. Smith and colleagues (2023) observed significant compositional divergence between stressed and control males but not females, whereas Brawner et al. (2020) detected significant community‐level shifts exclusively in female offspring. Although both studies employed combined physical and psychological stress paradigms extending across much of gestation, they differed substantially in sampling strategy. Brawner et al. (2020) analyzed fecal samples collected during early postnatal life (PND14), while Smith et al. (2023) examined tissue‐associated and luminal microbial communities from the duodenum, ileum, and colon at a later juvenile stage (PND45). These methodological differences suggest that sex‐dependent effects may vary across development and between intestinal niches. Gestational timing may also influence beta‐diversity outcomes. In a non‐human primate model, maternal stress exposure during early gestation, but not late gestation, produced significant alterations in offspring microbial community structure (Anza et al. 2023). Taken together, these findings indicate that prenatal stress reliably alters microbial community composition, while the magnitude and direction of these effects appear to be shaped by factors such as stressor type, gestational timing, offspring sex, developmental stage, and the anatomical site sampled.

3.3.1.3. Taxonomy

Given substantial species‐specific differences in baseline gut microbial composition, taxonomic findings are best interpreted separately for each animal model.

Murine models account for nearly half of the included studies (9 of 19) and provide the most consistent evidence that PSE reshapes offspring gut microbial compositions. Across studies, PSE is frequently associated with shifts in the relative abundance of the dominant phyla Firmicutes and Bacteroidetes, often resulting in an altered Firmicutes/Bacteroidetes ratio. Notably, neither the gestational timing of stress exposure, nor the biological source of the microbiome sample appears to consistently determine the direction of these changes. Instead, developmental stage emerges as a more important factor in explaining variation across studies. Several taxa exhibit age‐dependent responses to prenatal stress. The clearest examples are the genera Lactobacillus and Streptococcus. Studies examining neonatal offspring at PND2 consistently reported reductions in the relative abundance of both genera following prenatal stress (Jasarevic and colleagues: p < 0.05, 2015; p < 0.05, 2017; p < 0.01, 2018). In contrast, assessments conducted at later developmental stages (PND21–PND56) generally revealed increased abundances of these taxa in stress‐exposed offspring (Jasarevic et al. 2017, p < 0.05; Z. Zhang et al. 2021, p < 0.05; Sun et al. 2021, p < 0.05). Similar developmental effects have been observed for other bacterial groups. For example, Z. Zhang et al. (2021) reported increase abundance of Bacteroidaceae and Prevotellaceae (p < 0.01) alongside reduced abundance of Muribaculaceae, formerly known as the S24‐7 family (p < 0.01), one of the dominant bacterial clades in the murine gut (Chung et al. 2020). Consistent alterations within the Bacteroidetes phylum have also been reported in adult mice. Gur et al. (2017) demonstrated pronounced reductions in the abundance of the S24‐7 family (p < 0.0001), Bifidobacteriaceae (p = 0.005), and overall Bacteroidetes levels (p < 0.001) in adult female offspring. A similar pattern was later observed in adult males, with parental stress reducing the abundance of the genera Bacteroides (p = 0.003) and Parabacteroides (p = 0.001) following prenatal stress (Gur et al. 2019). Together, these findings suggest that reductions in Bacteroidetes‐related taxa may represent one of the more reproducible signatures of prenatal stress in mice. Sex‐specific effects further contribute to variability in murine taxonomic outcomes. Interestingly, studies reporting alterations in Lachnospiraceae (p = 0.0093) and Clostridiales (p = 0.0044) have produced opposing sex‐dependent patterns. Brawner and colleagues (2020) observed increased abundances of both taxa exclusively in female offspring at PND14, whereas Jasarevic et al. (2017) detected comparable increases only in male offspring at PND28 (p < 0.05). These discrepancies likely reflect interactions among sex, developmental stage, and microbial maturation rather than fundamentally conflicting biological effects.

In contrast to the relatively consistent findings in mice, taxonomic changes reported in rat studies are considerably more heterogeneous. No patterns emerge when studies are grouped according to gestational timing, stress paradigm, or microbiome sample source. At the phylum level, several studies reported reductions in Bacteroidetes during early life, including at PND2 (p = 0.00154, Pawluski et al. 2023) and PND21 (p = 0.027, Hua et al. 2023), accompanied by increases in PND15 (p = 0.0057, Adebiyi et al. 2022) and PND21 (p = 0.023, Hua et al. 2023). However, these findings are not universal. Zhao and colleagues (2021), for example, reported reduced Firmicutes abundance in stressed animals (p < 0.05) at a similar developmental stage. The genus Lactobacillus illustrates this inconsistency particularly well. Golubeva and colleagues (2015; p = 0.054) and Zhao and colleagues (2021; p < 0.05) reported substantial reductions in Lactobacillus abundance following prenatal stress, whereas other researchers observed significant increases (p < 0.05, Zheng et al. 2020; p < 0.01, Hua et al. 2023). Notably, both Hua et al. (2023) and Zhao et al. (2021) employed nearly identical multimodal stress paradigms spanning gestation, highlighting the difficulty of identifying robust taxonomic signatures across rat studies. Comparable discrepancies are evident among other major taxa. At PND21, Hua et al. (2023) reported decreased Clostridium and increased Bacteroides count (p < 0.05), while De Cillis et al. (2025), observed the opposite pattern (i.e., decreased Bacteroides count, p < 0.05, along with an increase in the Clostridium number, p < 0.05) in adulthood (PND84). These findings suggest that some prenatal stress‐induced alterations may change direction across development rather than persist uniformly throughout life. Within the Actinobacteria phylum, parental stress effects also appear highly dynamic. Pawluski et al. (2023) reported reduced Actinobacteria abundance at PND2 (p = 0.02775), whereas later developmental stages revealed genus‐specific, sex‐dependent, and age‐restricted responses. For example, Hua et al. (2023) documented increased Bifidobacterium and reduced Corynebacterium abundance at PND21. By PND84, however, Corynebacterium abundance increased selectively in males (De Cillis et al. 2025). Similar sex‐dependent divergences were observed for Eubacterium coprostanoligenes and Anaerostripes, which increased in female offspring (p ≤ 0.047) while decreasing (p ≤ 0.028) in males exposed to prenatal stress (De Cillis et al. 2025). Direct comparisons between stressor types further indicate that taxonomic outcomes depend not only on developmental factors but also on the nature of the prenatal exposure. Hua et al. (2023) found that restraint stress alone reduced the abundance of Lactobacillus and Helicobacter, whereas combined lead exposure and restraint stress induced broader enrichment of the class Bacilli, order Lactobacillales, and family Lactobacillaceae (p < 0.001). These findings suggest that mixed physical and psychological stress paradigms may produce microbial alterations that differ qualitatively from those induced by individual stressors.

Evidence from long‐lived animal models remains scarce. In rhesus monkey, Bailey and colleagues (2004) reported reduced abundances of Lactobacilus (p < 0.05) and Bifidobacterium (p < 0.05) in infants born from stressed mothers. Although both early‐ and late‐gestation stress reduced Lactobacilus abundance, offspring exposed to stress during early gestation retained higher Bifidobacterium levels (p < 0.05) than those exposed during late pregnancy. Early gestational stress was also associated with a markedly higher prevalence (43% in early compared to 13% in late stress and 0% in control groups) of Shigella flexneri colonization (Bailey et al. 2004), suggesting increased susceptibility to opportunistic pathogens. Importantly, Lactobacillus abundance was significantly lower in colonized animals, indicating a potential link between prenatal stress, reduced colonization resistance, and pathogen vulnerability. More recently, Anza et al. (2023) extended these observations to wild Assamese macaques, providing greater ecological validity through the assessment of naturally occurring stressors reflected by maternal glucocorticoid concentrations. As in captive rhesus macaques, gestational timing influenced microbiome outcomes. Early prenatal stress produced stronger and more persistent effects, whereas changes associated with late gestational stress were weaker and tended to diminish with age. Across stressed offspring, a reduced Firmicutes/Bacteroidetes ratio and increased representation of potentially pro‐inflammatory taxa were consistently observed (p < 0.05). The only available swine study similarly identified evidence of microbial dysbiosis. Maternal heat stress during late gestation increased the abundance of several Proteobacteria‐associated genera, including Acinetobacter, Klebsiella, Stenotrophomonas, Comamonas, and the Escherichia‐Shigella group (p < 0.05), while reducing Clostridium sensu stricto 1, Romboutsia, and Turicibacter (p < 0.05; He et al. 2020). These changes resulted in an elevated Proteobacteria/Firmicutes ratio (p < 0.05) and were accompanied by an expansion of opportunistic pathogens alongside modest reductions in commensal bacteria (p < 0.05).

Overall, the available evidence indicates that prenatal stress consistently alters offspring gut microbial composition, but the specific taxonomic changes observed depend strongly on species, developmental stage, offspring sex, and stressor characteristics. While broad shifts involving Firmicutes, Bacteroidetes, and key commensal genera such as Lactobacillus recur across studies, considerable heterogeneity remains at finer taxonomic levels. The substantial methodological and biological heterogeneity across the animal literature also precludes meaningful meta‐analysis. Studies differ considerably with respect to stressor type (physical, psychological, or combined), timing and duration of exposure, offspring sex and age at sampling, anatomical source of microbiome specimens, sequencing approaches, and taxonomic reporting levels. These sources of variation limit direct comparability between studies and violate key assumptions required for quantitative evidence synthesis. Consequently, a systematic review provides the most appropriate framework for integrating the current evidence, allowing a structured and transparent assessment of common patterns while preserving important biological context.

3.3.2. Human Studies

Prenatal maternal psychological stress, assessed either broadly or through specific domains such as depression and anxiety, appears to be associated with both alpha and beta diversity of the infant fecal microbiota. However, the direction of these associations varies across studies (see Table 5).

TABLE 5.

Microbial changes associated with prenatal maternal stress in human studies.

Reference	Changes in microbial diversity	PSE Group
Aatsinki et al. (2020)	The prenatal psychological distress symptoms were not associated with either alpha or beta diversity in the infant fecal microbiota	↑ Proteobacteria: Erwinia (dh), Haemophilus (dh, a), Serratia (dh, a, d), Citrobacter (a, d), ↑ Firmicutes: Veillonella (ppd), Finegoldia (ppd), Dialister (ppd), Dorea (ppd), Coprococcus (ppd), Staphylococcus (a), ↑ Bacteroidetes: Butyricimonas (d), Prevotella (d), ↑ Actinobacteria: Actinomyces (ppd), Rothia (ppd), ↑Campylobacter (a), ↓ Firmicutes: Megasphaera (ppd), Eubacterium (ppd), Pseudoramibacter (ppd), Epulopiscium (ppd), Anaerotruncus (ppd), Staphylococcus (d), Pseudoramibacter_Eubacterium (ppd), ↓ Bacteroidetes: Megamonas (ppd), Paraprevotella (ppd), Parabacteroides (ppd), Odoribacter (ppd) ↓ Actinobacteria: Slackia (ppd), Actinobaculum (ppd), Propionibacterium (ppd) ↓Desulfovibrio (d), ↓Phaslarctobacterium (ppd), ↓Akkermansia (ppd)
Deflorin et al. (2024)	Higher maternal depressive symptoms at 34–36 weeks’ gestation were associated with lower alpha diversity (Shannon index, τ = 0.15, p = 0.04); no significant associations were found between beta diversity and prenatal psychological markers	↑Proteobacteria (css)
Dutton et al. (2023)a	Alpha diversity (Shannon index) was significantly lower in six‐months old infants born to high‐stress moms; beta diversity (Bray–Curtis distance) was lower for PSE infants at 3 months and earlier, but not at 6 months	↓ Firmicutes: Lactobacillus gasseri (css)
Galley et al. (2023)	Reduced alpha diversity linked to maternal depression/anxiety Faith's Phylogenetic Diversity Index (H = 4.37, p < 0.05/ H = 2.88, p = 0.089); beta diversity is not reported	↓Actinobacteria: Bifidobacterium dentium (a, d, pss), Eggerthella lenta (d) ↓Firmicutes: Lactobacillus rhamnosus (a, d), Streptococcus salivarius (d)
Hu et al. (2019)	Trend for increase in alpha diversity (p = 0.07); beta diversity increased (p = 0.001) with stress measured through the Pregnancy‐Related Anxiety Questionnaire (total score)	↓ Proteobacteria: Enterobacteriaceae (a)
Jahnke et al. (2021)	No significant differences in alpha and beta diversity	↑ Proteobacteria: Enterobacteriaceae (fi, lfs) ↑ Actinobacteria: Bifidobacterium (lrs) ↓ Firmicutes: Lachnospiraceae (fi)
Querdasi et al. (2023)	Prenatal adversity was linked to differences in alpha diversity (Pielou evenness, β = 0.13, 95% CI = (0.01, 0.25), ΔR2 = 0.02); no effect was found for beta diversity	↑ Firmicutes: Streptococcus (ad) ↓ Firmicutes: Ruminococcus (ad)
Warner et al. (2023)	Both alpha and beta diversities (Bray–Curtis distance, unweighted UniFrac, Aitchison distance) positively correlated with prenatal high‐social disadvantage and psychosocial distress	↑Proteobacteria: Enterobacter nimipressuralis (sd), Klebsiella pneumoniae (sd), Sutterella sp. (css) ↑Firmicutes: Faecalicatena gnavus (css)
Wei et al. (2022)	Higher alpha diversity (richness measured through Chao1 index and observed species; evenness measured through Shannon index, Simpson index) in prenatally stress‐exposed neonates; beta diversity significantly differed between stressed and non‐stressed infants (Bray–Curtis, p = 0.047; unweighted UniFrac distance, p = 0.024)	↑ Proteobacteria (css) ↓ Actinobacteria (css) ↑ Firmicutes: Lactobacillus (css) ↑ Proteobacteria: Ralstonia (css), Burkholderia (css)
Weiss and Hamidi (2023)	Alpha diversity was increased in a stressed group (Simpson index, β = 0.30, p = 0.02; Shannon index, β = 0.25, p = 0.06); beta diversity is not reported	↑ Firmicutes: Lactobacillus, Lactococcus (pss) ↑ Actinobacteria: Bifidobacterium (pss) ↓ Proteobacteria: Enterobacteriaceae (pss) ↓Bacteroides (pss) ↓ Firmicutes: Erysipelotrichaceae (pss), Enterococcus (pss), Ruminococcus (pss) ↓ Actinobacteria: Eggerthella (pss)
Zijlmans et al. (2015)	No significant impact on richness or evenness was shown; the overall diversity was higher in high‐stress infants	↑ Proteobacteria: Escherichia (css), Enterobacter (css), Serratia (css), Haemophilus (css) ↓ Firmicutes: Lactobacillus (css), Lactococcus (css), Aerococcus (css) ↓Actinobacteria: Bifidobacterium (css), Collinsella (css), Eggerthella (css) ↓Akkermansia (css)
Naudé et al. (2020)a	Alpha diversity is not measured; no significant associations found between maternal prenatal psychological measures and infant fecal bacterial diversity indices	↑Proteobacteria: Citrobacter (ipv)
Eckermann et al. (2025)	No significant differences in alpha and beta diversity	No evidence for a positive association between any of the stress variables and Proteobacteria or for a negative association with Lactobacillus or Bifidobacterium
X. Zhang et al. (2025)	Maternal high negative emotions and depression during pregnancy were associated with significantly reduced alpha diversity (Chao1 index, p = 0.001); beta diversity differed significantly between groups (p = 0.001)	↑Bacteroidetes (css) ↑Bacteroidales (css) ↑ Firmicutes: Lachnospiraceae (css), Clostridiales (css), Oscillospira (css), Ruminococcus (css) ↑ Bacteroidetes: Muribaculaceae (css)

Note: (a)—anxiety; (ad)—adversity; (css)—composite stress score; (d)—depression; (dh)—daily hassles; (fi)—food insecure; (ipv)—intimate partner violence; (lfs)—low friend support; (lrs)—low relative support; (ppd)—prenatal psychological distress; (pss)—perceived stress scale; (sd)—social disadvantage; ↑ positive association with prenatal stress exposure; ↓ negative association with prenatal stress exposure.

^aAdditional data on the relative abundance alterations of microbial species depending on age can be found in the works of Dutton et al. 2023; Naudé et al. 2020. In the table, we mention only those that are non‐age‐specific positive.

3.3.2.1. Alpha Diversity

Findings regarding the association between maternal psychological distress and infant microbial alpha diversity are highly heterogeneous and often vary according to the diversity of metrics used and the timing of stress exposure during pregnancy. Several studies have reported no significant association between prenatal psychological measures and infant fecal microbiota alpha diversity, including studies assessing distress longitudinally across pregnancy (gestational weeks 14, 24, and 34; Aatsinki et al. 2020), during mid‐to‐late gestation (weeks 18 and 32; Eckermann et al. 2025), and in late pregnancy (third trimester; Zijlmans et al. 2015; weeks 34–36; Jahnke et al. 2021). When significant associations have been observed, reduced alpha diversity has most commonly been linked to maternal depressive symptoms or broader psychological distress. For example, lower alpha diversity was associated with depressive symptoms measured at gestational weeks 34–36 (Shannon index, p = 0.04; Deflorin et al. 2024), and at weeks 12 and 20 (Chao1 index, p = 0.001; X. Zhang et al. 2025). Similarly, Galley et al. (2023) reported lower alpha diversity as assessed by Faith's Phylogenetic Diversity Index, in relation to both depression (p < 0.05) and anxiety (p = 0.089) assessed during early pregnancy (weeks 8–16). Beyond prenatal assessments, Dutton and colleagues (2023) observed that acute maternal stress measured within 24 h of delivery using a composite stress index predicted lower Shannon diversity in infants at 6 months of age. In contrast, several studies have reported positive associations between prenatal distress and infant alpha diversity. Focusing on third‐trimester stress, Weiss and Hamidi (2023) found higher alpha diversity among infants exposed to elevated prenatal stress, as measured by the Simpson index (p = 0.02), with a similar trend in the Shannon index (p = 0.06). Warner and colleagues (2023) likewise reported positive associations between alpha diversity and prenatal social disadvantage or psychosocial distress assessed across pregnancy. Consistent with these findings, Wei and colleagues (2022) found that distress during weeks 32–36 was associated with greater alpha diversity across multiple richness (Chao1, observed species) and evenness (Shannon, Simpson) metrics. Some evidence further suggests that associations may depend on the specific dimension of alpha diversity being examined. Querdasi et al. (2023) reported that prenatal adversity assessed during the late second trimester was associated specifically with Pielou evenness (p < 0.05), whereas other diversity measures were unaffected. Similarly, Hu et al. (2019) observed a marginal increase in alpha diversity (p = 0.07) in relation to second‐trimester pregnancy‐related anxiety. Collectively, these findings indicate that the relationship between prenatal psychological distress and infant microbial alpha diversity remains inconsistent, with observed effects varying according to the timing and type of maternal distress as well as the diversity metric employed.

3.3.2.2. Beta Diversity

Findings regarding infant gut microbial beta diversity are similarly mixed, with studies reporting both null and significant associations depending on the timing, chronicity, and type of maternal psychological distress assessed. Several studies found no associations between prenatal psychological measures and infant microbial community composition, regardless of whether stress was evaluated longitudinally across pregnancy (gestational weeks 14, 24, and 34; Aatsinki et al. 2020), during the second trimester (weeks 26–28; Querdasi et al. 2023), in mid‐to‐late pregnancy (weeks 18 and 32; Eckermann et al. 2025, 27.4 ± 4.2 weeks, Naudé et al. 2020), or during late‐gestation (weeks 34–36; Jahnke et al. 2021; Deflorin et al. 2024). In contrast, several studies have reported significant alterations in beta diversity among stressed cohorts. Focusing on the second trimester, Hu et al. (2019) found that beta diversity increased significantly with pregnancy‐related anxiety (p = 0.001). Consistent with this observation, Warner et al. (2023) reported significant associations between overall community composition—measured using Bray–Curtis, unweighted UniFrac, and Aitchison distance—and cumulative prenatal stress, including social disadvantage, depression, perceived stress, and racial discrimination assessed throughout pregnancy. Similarly, maternal stress during late pregnancy (weeks 32–36) was associated with significant differences in neonatal microbial community structure across both Bray–Curtis (p = 0.047) and unweighted UniFrac (p = 0.024) distances (Wei et al. 2022), a finding later replicated by X. Zhang et al. (2025; p = 0.001). Zijlmans et al. (2015) likewise reported greater overall community diversity among infants exposed to high prenatal stress, despite observing no differences in richness or evenness. Notably, not all significant findings pointed in the same direction. Whereas most studies reported greater community dissimilarity in association with prenatal stress, Dutton et al. (2023) found that infants exposed to prenatal stress exhibited lower beta diversity, as measured by Bray–Curtis distance, during the first 3 months of life. However, these differences were no longer evident by 6 months, suggesting that stress‐related effects on microbial community composition may attenuate over time.

3.3.2.3. Taxonomy

Across studies examining broad socio‐environmental adversity, daily stressors, and composite measures of prenatal psychosocial distress, alterations have been reported across multiple major bacterial taxa. However, findings remain inconsistent, with the direction and magnitude of effects varying according to the type, timing, and chronicity of maternal stress exposure. Within the Firmicutes phylum, prenatal stress is frequently associated with a reduction in beneficial lactic acid bacteria. For example, acute maternal stress assessed within one day of delivery was associated with a longitudinal decrease in Lactobacillus gasseri across all infant sampling timepoints (6 weeks, 3 months, and 6 months) among infants born to highly stressed mothers (Dutton et al. 2023). Similarly, stress measured during the late second trimester (weeks 26–28) was linked to increased Streptococcus abundance alongside a reduction in Ruminococcus (p < 0.05, Querdasi et al. 2023). Comparable patterns have been observed within the Actinobacteria phylum, where broad psychosocial distress was associated with persistent reductions in protective Bifidobacterium species, including Bifidobacterium pseudocatenulatum, through 3 months postpartum (Dutton et al. 2023). Notably, more specific psychosocial stressors do not always follow this pattern. Low family support during late pregnancy was associated with a greater relative abundance of Bifidobacterium (p < 0.01; Jahnke et al. 2021). Likewise, perceived stress during the third trimester was linked to increased infant abundances of Bifidobacterium (p < 0.01), Lactobacillus (p < 0.001), and Lactococcus (p < 0.01) (Weiss and Hamidi 2023). These findings contrast with studies using broader composite stress measures. For example, Zijlmans et al. (2015) reported that a cumulative stress index incorporating anxiety, pregnancy‐related depression, and daily hassles was associated with increased abundances of Proteobacteria genera (Escherichia and Enterobacter) and reduced levels of several putatively beneficial taxa, including Lactobacillus, Lactococcus, Aerococcus, Bifidobacterium, Collinsella, Eggerthella, and Akkermansia (p < 0.05 with certain taxa reaching p < 0.01). Evidence linking prenatal stress to shifts in Proteobacteria is comparatively consistent. Daily environmental stressors measured longitudinally across gestation (weeks 14, 24, and 34) were associated with increased abundances of the gram‐negative Proteobacteria genera Erwinia, Haemophilus, and Serratia (False Discovery Rate, FDR < 0.01, Aatsinki et al. 2020). Similarly, chronic psychological distress and social‐environmental adversity have been associated with enrichment of potentially pathogenic taxa. Warner et al. (2023) observed increased abundances of Enterobacter nimipressuralis (p = 1.2E‐05), Klebsiella pneumoniae (p = 9.4E‐04), Sutterella sp. (p = 0.021), and Faecalicatena gnavus (p = 0.017) among infants exposed to high maternal social disadvantage and distress. Comparable findings have been reported for food insecurity during late gestation (weeks 34–36), which was associated with increased relative abundances of Proteobacteria (p = 0.05) and the Enterobacteriaceae family (p = 0.05), alongside lower abundances of Lachnospiraceae (p = 0.01; Jahnke et al. 2021). Likewise, prenatal exposure to intimate partner violence was linked to expansion of the opportunistic genus Citrobacter (p = 0.002, Naudé et al. 2020). Nevertheless, even within the same phyla, associations are not uniformly directional. Aatsinki et al. (2020) found that chronic psychological distress exhibited bidirectional relationships across microbial taxa, enriching some members of the Firmicutes and Actinobacteria phyla while negatively correlating with others across Firmicutes, Actinobacteria, and Bacteroidetes (FDR < 0.01; see Table 5). Such findings suggest that stress‐related microbiome alterations may be highly taxon‐specific rather than reflecting broad phylum‐wide disruptions. Maternal depression shows a relatively consistent association with depletion of beneficial symbionts and a shift toward a more inflammatory microbial profile, although the affected taxa vary by gestational timing. Depressive symptoms during early pregnancy (weeks 8–16) were associated with reduced abundances of Bifidobacterium dentium, Lactobacillus rhamnosus, and Streptococcus salivarius (p < 0.05, Galley et al. 2023). Broader depressive symptomatology measured during the first and second trimesters (weeks 12 and 20) was linked to increases in Lachnospiraceae, Clostridiales, Ruminococcus, and Oscillospira (p = 0.001, X. Zhang et al. 2025). Longitudinally elevated depression scores across pregnancy (weeks 14, 24, and 34) were additionally associated with reductions in Desulfovibrio and Staphylococcus (FDR < 0.01, Aatsinki et al. 2020). A recurring finding across studies is an enrichment of Proteobacteria under conditions of prenatal depression. Elevated depression during late gestation (approximately weeks 32–36) was associated with increased Proteobacteria abundance (Wei et al. 2022; Deflorin et al. 2024), while longitudinal depressive symptoms predicted grater abundances of the Citrobacter and Serratia (FDR < 0.01, Aatsinki et al. 2020). Additional changes have been observed within the Bacteroidetes phylum, including enrichment of Muribaculaceae in response to early‐to‐mid‐pregnancy depressive symptoms (p = 0.001, X. Zhang et al. 2025) and increased abundances of Butyricimonas and Prevotella under longitudinal gestational distress (FDR < 0.01, Aatsinki et al. 2020).

Pregnancy‐related anxiety yields similarly mixed microbial signatures. Anxiety during the second and third trimesters was associated with a reduction in Enterobacteriaceae (p = 0.002, Hu et al. 2019) but increases in the Proteobacterial genera Citrobacter, Serratia, and Haemophilus (FDR < 0.01, Aatsinki et al. 2020). Early‐pregnancy anxiety was also linked to reduced Lactobacillus rhamnosus abundance (p < 0.0001, Galley et al. 2023), whereas Staphylococcus and Campylobacter demonstrated positive associations with maternal anxiety scores later in pregnancy (FDR < 0.01, Aatsinki et al. 2020). Taken together, the literature suggests that prenatal stress, depression, and anxiety frequently coincide with reductions in beneficial commensals and enrichment of Proteobacteria and other potentially inflammatory taxa. However, substantial heterogeneity remains across studies, with contradictory findings reported even for hallmark genera such as Lactobacillus and Bifidobacterium. This lack of convergence underscores broader concerns regarding reproducibility in the field. Indeed, Eckermann et al. (2025) reported entirely null results, finding no positive associations with Lactobacillus or Bifidobacterium, highlighting the need for greater methodological consistency and replication.

3.3.2.4. Cortisol Levels

Neither of the two studies examining associations between prenatal cortisol levels and infant microbial alpha diversity identified a significant relationship (Deflorin et al. 2024; Jahnke et al. 2021). In contrast, more consistent associations emerged for beta diversity. Jahnke et al. (2021) reported that maternal salivary cortisol awakening response measured at 34–36 weeks of gestation was associated with differences in infant beta diversity at 2 months of age when assessed using weighted UniFrac distances (p = 0.02), although no association was observed for Bray–Curtis dissimilarity. Similarly, Deflorin et al. (2024) found that maternal cortisol exposure at 38 weeks, indexed by mean AUCg, was associated with meconium beta diversity measured using weighted UniFrac distances (p < 0.003), but not Bray–Curtis dissimilarity. Notably, cortisol levels assessed earlier in late gestation (34–36 weeks) were not associated with meconium beta diversity in the same study, suggesting that timing of cortisol exposure may influence microbiome outcomes. At the taxonomic level, both Actinobacteria and Firmicutes demonstrated consistent negative associations with maternal cortisol levels across studies. Higher maternal cortisol concentrations, measured in either hair or saliva, were generally associated with reduced abundances of taxa belonging to these phyla (Aatsinki et al. 2020; Zijlmans et al. 2015). Associations within the Bacteroidetes phylum were less consistent. Aatsinki et al. (2020) reported negative relationships between maternal hair cortisol concentrations and specific Bacteroidetes taxa, including Paraprevotella and Butricimonas (FDR < 0.01). In contrast, Jahnke and colleagues (2021) reported that lower cortisol awakening response levels were associated with lower abundances of Bacteroides (p = 0.01), implying a positive association between cortisol and this genus. Further highlighting this heterogeneity, Eckermann et al. (2025) observed negative associations between maternal hair cortisol concentrations and Bacteroides ylanisolvens and Bacteroides thetaiotaomicron, but a positive association with Bacteroides fragilis (FDR ≤ 0.1). Findings regarding Proteobacteria were similarly mixed. Several studies reported inverse associations between maternal cortisol levels and Proteobacteria abundance (Aatsinki et al. 2020, FDR < 0.01; Deflorin et al. 2024, p < 0.001), suggesting that elevated cortisol may be linked to reduced representation of this phylum. However, Zijlmans et al. (2015) reported the opposite pattern, with higher maternal cortisol concentrations associated with a relative enrichment of Proteobacteria. Collectively, these findings indicate that while cortisol‐related alterations in infant gut microbiota may be detectable at the community level, taxon‐specific associations remain heterogeneous and highly dependent on the timing and method of cortisol assessment.

4. Discussion

We discuss the observed patterns in the literature along the following dimensions: (1) overall patterns of microbiome disruption following PSE; (2) patterns of animal studies that provide evidence for microbiome restructuring following PSE; (3) patterns from human studies suggesting that stress type matters more than timing; (4) the loss of beneficial taxa as a recurring microbial signature of PSE; (5) shifts toward dysbiosis and a pro‐inflammatory microbial profile; (6) the value of animal models for understanding PSE‐related microbiome changes; (7) sources of heterogeneity across studies; (8) methodological limitations and opportunities for improvement; (9) potential mechanisms linking prenatal stress to offspring microbiome development; and (10) future directions for research.

4.1. Overall Patterns of Microbiome Disruption Following Prenatal Stress Exposure

Empirical research examining the effects of PSE on the offspring gut microbiome remains limited, and available findings are highly heterogeneous. Consistent with previous reviews (Grech et al. 2021; Mepham et al. 2023; Agusti et al. 2023; Yeramilli et al. 2023; Ryan et al. 2025; Graf et al. 2025), we found no reproducible associations between PSE and offspring alpha diversity. Evidence regarding beta diversity has also been mixed. However, the first cross‐species systematic review concluded that PSE generally induces structural alterations in the offspring microbial communities (Graf et al. 2025). Our findings partly support this conclusion. In animal models, PSE was consistently associated with beta‐diversity differences, suggesting stress‐related restructuring of the gut microbiome. In contrast, human studies showed considerably less agreement, mirroring the broader literature and revealing substantial outcome inconsistency (Grech et al. 2021; Mepham et al. 2023; Agusti et al. 2023; Yeramilli et al. 2023; Ryan et al. 2025). Notably, whereas Graf et al. (2025) synthesized six human studies reporting beta diversity outcomes, the present review included 12 such studies, potentially capturing a broader spectrum of variability and contributing to the less consistent picture observed here. Despite these inconsistencies, a central finding emerges across both human and animal research: offspring exposed to elevated maternal stress during pregnancy frequently exhibit alterations in gut microbial composition. Given the pivotal role of the microbiome in metabolic, immune, and neurodevelopmental processes, synthesizing these scattered findings is essential for understanding potential mechanisms of MGBA programming.

4.2. Animal Studies Provide Stronger Evidence for Microbiome Restructuring

In contrast to human cohorts, all animal studies examining beta diversity reported significant PSE‐related effects. This consistency likely reflects the controlled nature of preclinical models. Animal studies are conducted under highly standardized housing conditions, minimizing many of the environmental and behavioral confounders that characterize human populations. At the same time, they revealed an important temporal pattern. Significant microbiome alterations observed during early developmental stages frequently weakened or disappeared with age. This is yet another noncompliance with Graf et al. (2025) who concluded the absence of associations between beta diversity and the timing of gut microbiome sampling. Similar age‐dependent effects were observed for alpha diversity in rodent models (Golubeva et al. 2015; Zhao et al. 2021; Pawluski et al. 2023; De Cillis et al. 2025), suggesting that the consequences of prenatal stress may be most pronounced during early‐life microbial assembly. Notably, non‐human primate studies demonstrated temporal changes in both alpha and beta diversity and identified early gestation as a particularly sensitive developmental window (Anza et al. 2023). Together, these findings suggest that PSE may influence not only microbial composition but also the developmental trajectory of microbiome maturation.

4.3. Human Studies: Stress Type Matters More Than Timing

In human cohorts, the gestational timing of stress exposure does not appear to produce uniform microbiome outcomes. Instead, more consistent patterns emerge when studies are grouped according to the nature of the maternal stressor. Specifically, maternal depression, anxiety, or combined affective distress were more consistently associated with reduced infant alpha diversity than broader influences, regardless of the trimester during which symptoms were assessed (Galley et al. 2023; Deflorin et al. 2024; X. Zhang et al. 2025). This pattern suggests that affective symptoms may exert more direct or biologically salient influences on microbial colonization than broader environmental stress measures. Interestingly, this relative consistency was not observed for beta diversity findings, with no clear pattern emerging according to either gestational timing or specific stress measure. One possible explanation is that maternal distress may reduce overall microbial richness and evenness while leaving the identity of surviving taxa highly individualized. Factors such as maternal baseline microbiota, diet, medication use, delivery mode, feeding practices, and environmental exposures could produce distinct microbial assemblages across infants despite similar reductions in alpha diversity (Lu et al. 2024). Consequently, two infants exposed to maternal depression may both exhibit reduced microbial diversity while harboring substantially different bacterial communities. Because beta diversity captures differences in taxonomic composition between individuals rather than overall richness, such individualized responses could obscure consistent group‐level effects. Additional sources of variation—including cesarean delivery, antibiotic exposure, and infant feeding practices—likely further obscure detectable beta‐diversity signals. For example, cesarean delivery is known to disrupt the natural colonization process of the infant gut, resulting in microbial communities that differ substantially from those of vaginally delivered infants (Inchingolo et al. 2024; Sassin et al. 2022). Supporting the importance of maternal microbial transmission, experimental exposure of newborns to maternal vaginal secretions has been shown to partially restore microbiome composition and increase microbial diversity in both human and animal studies (Dominguez‐Bello et al. 2016; Jasarevic et al. 2018). However, vaginal microbiota transfer alone may not be sufficient to overcome the effect of prenatal stress. It was documented that inoculating prenatally stressed pups with vaginal fluid from non‐stressed mothers failed to normalize their gut microbiota, suggesting that factors beyond maternal microbial transmission contribute to the establishment of pioneer microbial communities in the neonate gut (Jasarevic et al. 2018). Together, these findings highlight the complex interplay between prenatal and postnatal influences on microbiome development and illustrate how multiple sources of variability can obscure consistent beta‐diversity patterns in human cohorts. The apparent consistency of depression‐ and anxiety‐related reductions in alpha diversity should also be interpreted cautiously. Many studies combine anxiety and depression with broader psychosocial and environmental stressors into composite stress indices (Zijlmans et al. 2015; Dutton et al. 2023; Warner et al. 2023), making it difficult to isolate the independent contribution of specific psychological symptoms from those of more general prenatal adversity.

4.4. A Recurring Microbial Signature: Loss of Beneficial Taxa

Although taxonomic findings were heterogeneous, several microbial groups emerged repeatedly across studies. Most notably, reductions in Bifidobacterium and Lactobacillus represented the most consistent taxonomic PSE signature across species. These genera are among the earliest and most important colonizers of the neonatal gut and play fundamental roles in immune development, epithelial barrier maintenance, pathogen resistance, and metabolic homeostasis (Rodriguez et al. 2015; Kumar et al. 2020). Across both human and animal studies, PSE was more often associated with decreased abundances of these taxa than with increases, particularly following stress exposure during early gestation (Bailey et al. 2004; Jasarevic et al. 2015; Galley et al. 2023). The functional implications of these changes may be substantial. Reduced Lactobacillus has been associated with increased susceptibility to infection, impaired immune regulation, altered nutrient metabolism, and adverse neurodevelopmental outcomes (Lu et al. 2022; Shah et al. 2024). Similarly, diminished Bifidobacterium has been associated with allergic, metabolic, and neurological disorders (Di Gioia et al. 2014), as well as compromised intestinal barrier integrity, potentially leading to chronic inflammatory conditions (Duca et al. 2013). Importantly, beyond their direct physiological functions, these taxa also are pioneer colonizers of the neonatal gut. Through oxygen consumption, Lactobacillus and Streptococcus help establish the anaerobic environment necessary for subsequent colonization by obligate anaerobes such as Bifidobacterium and Bacteroides (Favier et al. 2002; Penders et al. 2006). Consequently, disruptions affecting these early colonizers may trigger cascading on microbial succession, altering the developmental trajectory of the gut ecosystem. Such disruptions may be particularly consequential for Bifidobacterium, a dominant representative of the Actinobacteria phylum, and a key contributor to gut homeostasis (Binda et al. 2018; Zafar and Saier 2021). As a result, a reduction in this phylum‐level count significantly expands disease susceptibility. Specifically, a reduction in Bifidobacterium abundance often translates into broader declines in Actinobacteria, a relatively small but functionally important phylum. Depletion of Actinobacteria has been associated with a range of adverse health outcomes, including metabolic disorders such as gestational diabetes (Mousavi et al. 2025), poorer baseline cognitive performance and less favorable cognitive trajectories in humans (Kolobaric et al. 2024), and impaired immune regulation resulting from disrupted T‐cell modulation showed in murine models (O'Mahony et al. 2008; Lyons et al. 2010). Together, these findings suggest that PSE‐related reductions in early‐life colonizers may have consequences that extend beyond microbial composition, influencing metabolic, immunological, and neurodevelopmental outcomes later in life. Interestingly, decreases in Lactobacillus abundance do not necessarily persist in adulthood, suggesting the existence of developmental‐stage–specific or compensatory microbial adaptations over time (Z. Zhang et al. 2021). Nevertheless, transient disturbances during early colonization may still have long‐term consequences by altering the trajectory of microbiome development. While findings within the Bacteroidetes phylum are less consistent overall, several taxa show recurrent patterns across studies. In particular, Muribaculum and Parabacteroides frequently decrease following PSE in a range of mammalian species. Muribaculum species maintain cooperative, cross‐feeding relationships with beneficial genera such as Bifidobacterium and Lactobacillus (Y. Zhu et al. 2024), making their parallel decline biologically plausible. Members of the Muribaculaceae family contribute to intestinal barrier integrity, anti‐inflammatory signaling, and propionate production, all of which support gut health (Barouei et al. 2017; Strong et al. 2016; Harrison et al. 2019; Y. Zhu et al. 2024). Moreover, lower Muribaculaceae abundance has been linked to increased obesity risk (H. Zhu and Hou 2024). Similarly, Parabacteroides abundance is negatively associated with obesity‐related indices, including body weight gain and liver weight in animal models (H. Zhu and Hou 2024). Another notable finding is the frequent reduction of Akkermansia, the most extensively studied genus within the Verrucomicrobia phylum. Decreased Akkermansia abundance was reported in most human studies and several animal models following PSE. Such reductions have been associated with obesity (Zhou et al. 2020; Liu et al. 2021), irritable bowel syndrome (Cruz‐Aguilar et al. 2019), and ulcerative colitis (Earley et al. 2019). Collectively, these patterns point to the potential functional significance of PSE‐related microbiome alterations. By disrupting taxa that support intestinal homeostasis, immune regulation, and metabolic health, prenatal stress may increase vulnerability to inflammatory, metabolic, cognitive, and neuropsychiatric disorders later in life through microbiota–gut–brain axis mechanisms.

4.5. A Shift Toward Dysbiosis and Pro‐Inflammatory Microbial Profile

The taxonomic reorganization associated with PSE frequently manifests as bidirectional changes within the Firmicutes phylum in both human (Aatsinki et al. 2020; Querdasi et al. 2023; Weiss and Hamidi 2023) and animal studies (Golubeva et al. 2015; Jasarevic et al. 2015; Jasarevic et al. 2018; Zheng et al. 2020; Sun et al.2021; Hua et al. 2023; De Cillis et al. 2025). Depletion of stress‐sensitive taxa such as Lactobacillus may create ecological opportunities for compensatory expansion of neighboring Firmicutes, potentially preserving phylum‐level equilibrium with Proteobacteria. However, a broader pattern emerging across studies is a shift from beneficial commensals toward taxa associated with inflammation and dysbiosis. In particular, an altered Firmicutes‐to‐Proteobacteria balance has been linked to oxidative stress and disruptions in carbohydrate and amino acid metabolism (Morgan et al. 2012). Increased abundance of Proteobacteria has been associated with inflammatory bowel disease and metabolic syndrome in humans (Bradley et al. 2017), as well as neonatal porcine diarrhea (Hermann‐Bank et al. 2015). Elevated Proteobacteria abundance has also been reported in cognitively impaired rat pups, suggesting a potential link between gastrointestinal inflammation and cognitive development (Lu et al. 2022). Several Proteobacterial genera repeatedly emerge in association with adverse health outcomes. For example, Sutterella species are enriched in children with autism and co‐occurring gastrointestinal symptoms (Williams et al. 2012), while their increased abundance has also been associated with metabolic syndrome (Lim et al. 2017), Down syndrome (Biagi et al. 2014), autism (L. Wang et al. 2013), and inflammatory bowel disease (Lavelle et al. 2015). Likewise, other genera within the phylum Proteobacteria—Enterobacter, Escherichia, and Klebsiella—although common constituents of the mammalian gut microbiome (Marques et al. 2019; W. Gu et al. 2019; X. Wang et al. 2022), include opportunistic pathogens capable of promoting systemic inflammation (Ding et al. 2026). Notably, these genera were enriched following prenatal social disadvantage and food insecurity in humans (Warner et al. 2023) and in heat‐stressed piglets (He et al. 2020). Together, these findings support a recurring pattern of reduced beneficial bacteria and increased abundance of potentially pathogenic taxa following PSE (Grech et al. 2021; Agusti et al. 2023; Mepham et al. 2023; Ryan et al. 2025; Graf et al. 2025).

4.6. The Value of Animal Models for Understanding PSE‐Related Microbiome Changes

Although findings from animal studies cannot be directly extrapolated to humans, preclinical models remain crucial for understanding the biological consequences of PSE. Unlike human studies, animal experiments permit simultaneous examination of the gut, brain, immune system, and other physiological systems, allowing researchers to investigate potential mechanisms underlying stress‐induced microbiome changes. Animal studies also benefit from standardized housing conditions and precise experimental manipulation of stress exposure, minimizing many of the environmental and behavioral confounders that contribute to variability in human cohorts. At the same time, substantial diversity exists among experimental paradigms, including differences in stressor type, duration, timing, and intensity. Greater standardization of animal models would improve reproducibility and facilitate comparison across studies.

4.7. Sources of Heterogeneity Across Studies

A major challenge in the field is the substantial heterogeneity of reported findings. One important source of variation is the timing of microbiome sampling. Across the literature, offspring were sampled from birth to 1 year of age in animal studies and up to 2 years in human cohorts. Because the neonatal gut microbiome undergoes rapid and dynamic maturation (Bäckhed et al. 2015; Stewart et al. 2018; D. W. Chen and Garud 2022), even small differences in sampling age can yield markedly different microbial profiles. Similar age‐dependent microbial restructuring has been documented across rodents, non‐human primates, and humans (Inoue and Ushida 2003; Jasarevic et al. 2017; Pandey and Aich 2023). Although some studies employed longitudinal sampling designs (Bailey et al. 2004; Jasarevic et al. 2017; Naude et al. 2020; Adebiyi et al. 2022; Dutton et al. 2023), developmental trajectories of alpha and beta diversity remained inconsistent, making it difficult to identify discrete windows of microbiome vulnerability. Sex represents another underappreciated source of variability. While most studies included both males and females, some analyzed sexes jointly (Adebiyi et al. 2022; Hua et al. 2023), potentially masking sex‐specific effects. Studies that examined males and females separately frequently reported sexually dimorphic microbiome responses to PSE, consistent with broader evidence for sex‐dependent prenatal programming (Weinstock 2011; Sutherland and Brunwasser 2018; Musillo et al. 2023). These findings suggest that sex should be systematically accounted for in future experimental and analytical designs. Methodological differences further complicate comparisons between studies. Although fecal samples were most commonly analyzed, approximately one‐third of animal studies characterized microbial communities from intestinal tissue, most often the distal colon. Previous work has demonstrated substantial differences between fecal, luminal, and mucosal microbial communities across humans (Eckburg et al. 2005), rodents (Ouwehand et al. 2004), pigs (C. Chen et al. 2018), and macaques (Yasuda et al. 2015). Even within the same experimental cohort, microbial profiles may differ across intestinal regions (De Cillis et al. 2025), suggesting that sample origin alone may contribute substantially to outcome variability and underscoring the need for methodological harmonization in future studies.

4.8. Methodological Limitations and Opportunities for Improvement

Most studies relied on 16S rRNA gene sequencing. Although cost‐effective and widely used, this approach suffers from limited taxonomic resolution and biases introduced by variation in rRNA gene copy number among bacterial species (Case et al. 2007; Sun et al. 2013; Regueira‐Iglesias et al. 2023; Weinroth et al. 2022). Bioinformatic correction tools such as PICRUSt (Langille et al. 2013), CopyRighter (Angly et al. 2014), and PAPRICA (Bowman and Ducklow 2015) partially address these limitations, but their performance depends heavily on database completeness (Nearing et al. 2021). Furthermore, 16S rRNA sequencing provides relative abundance estimates rather than absolute bacterial quantities. Consequently, meaningful interpretation often requires complementary quantitative techniques such as quantitative polymerase chain reaction, qPCR (Gloor et al. 2017; Ganda et al. 2021). Notably, only one study included in this review incorporated quantitative measurements (Jasarevic et al. 2015). Encouragingly, recent studies have begun utilizing whole‐genome metagenomic sequencing (Eckermann et al. 2025), an approach that provides improved taxonomic resolution and more accurate abundance estimates. Nevertheless, the field continues to be limited by its heavy reliance on statistical significance testing in the absence of effect size reporting. Consequently, the biological significance of observed microbiome alterations often remains unclear. The routine inclusion of effect size measures would facilitate cross‐study comparisons and improve the robustness of future meta‐analyses. Alongside methodological advances, the continued development of bioinformatics tools and the growing number of sequenced bacterial genomes available in reference databases are likely to further improve the accuracy of microbiome analyses.

4.9. Potential Mechanisms Linking Prenatal Stress to Offspring Microbiome Development

The mechanisms by which maternal stress influences offspring microbiome development remain poorly understood. Most studies have focused on cortisol or corticosterone because of their central role in HPA‐axis signaling (Carabotti et al. 2015; Sze, Brunton 2024; Rusch et al. 2023). Elevated maternal cortisol has been consistently associated with a reduction in Actinobacteria and Firmicutes in both humans (Aatsinki et al. 2020; Zijlmans et al. 2015) and non‐human primates (Bailey et al. 2004). However, emerging evidence suggests that neurodevelopmental and immunological pathways may also contribute. Prenatal stress has been associated with mitochondrial damage, astrogliosis, neuroinflammation, altered cytokine profiles, reduced myelin proteins, decreased expression of Iba1 and doublecortin, and impaired neuronal development in animal models (Adebiyi et al. 2022; Hua et al. 2023; Pawluski et al. 2023; Smith et al. 2023; Z. Zhang et al. 2021). These findings indicate that prenatal stress may affect the developing nervous system through multiple converging pathways. Evidence from animal models highlights the diversity of these mechanisms. Using a rat model of combined physical and psychological prenatal stress, Hua et al. (2023) observed mitochondria swelling vacuolar degeneration, dissolution of mitochondrial cristae within hippocampal neurons, and mild astrogliosis that was most pronounced in the combined‐stress group. Similarly, gestational heat stress in rats was associated with astrocytic alterations, elevated inflammatory cytokines TNF‐a and IL‐10 levels, and reduced myelin basic protein expression (Abediyi et al. 2022). PSE also reduced Iba1 and doublecortin expression in neonatal rodent brains, suggesting impairments in neuronal migration and microglia development (Pawluski et al. 2023; Smith et al. 2023). Notably, although both studies incorporated physical and psychological stressors, Smith et al. (2023) used diesel exhaust particle inhalation as a physical component. Whereas psychological stress primarily activates HPA‐axis signaling, chemical stressors may additionally trigger neuroinflammatory and metabolic pathways, suggesting that similar neurodevelopmental outcomes can arise through different upstream mechanisms. Finally, maternal immobilization stress has been shown to reduce neuronal production in the embryonic and newborn mouse cortex (Z. Zhang et al. 2021). Collectively, these findings point to widespread disruptions of neurodevelopment under conditions of prenatal stress. Given that the GBA is a bidirectional communication network, altered fetal neurodevelopment may itself influence microbiome assembly. Brawner et al. (2020) reported changes in the proportion of IgA‐bound bacteria in offspring born to mothers exposed to restraint and bright‐light stress during pregnancy. These offspring also exhibited more severe colonic tissue damage in a necrotizing enterocolitis‐like injury model, implicating immune dysregulation as a potential mediator (Brawner et al. 2020). Inflammatory processes in the neonatal intestine may alter nutrient availability within the gut lumen, thereby shaping microbial community development. Moreover, because the intestinal epithelium plays a crucial role in recognizing microbial signals (Henderson et al. 2011; Peterson and Artis 2014), impairments in epithelial integrity may influence microbial colonization from the earliest stages of life (Cho and Blaser 2012). This vulnerability is particularly relevant during birth, when primary colonization of the gut occurs (Milani et al. 2017; Korpela and de Vos 2018; Kennedy et al. 2021). Prenatal stress has been shown to alter maternal vaginal microbial communities (Jasarevic et al. 2015, 2017; Gur et al. 2017), potentially modifying the microbial inoculum transferred to the infant during delivery. Colonization by these stress‐altered microbial communities may subsequently shape the intestinal niche in ways that favor taxa associated with increased disease risk later in life. Although the available literature remains limited and findings are heterogeneous, current evidence suggests that prenatal stress influences offspring microbiota through multiple interconnected pathways involving neuroendocrine, neurodevelopmental, immunological, and microbial processes. These alterations may initiate self‐reinforcing cycles of dysfunction that persist throughout the lifespan. Importantly, this growing body of evidence raises the possibility of translating mechanistic insights into preventive strategies. Maternal hair or salivary cortisol assessments could be incorporated alongside standardized psychometric inventories (such as the Edinburgh postnatal depression scale or Pregnancy‐Related Anxiety Questionnaire–Revised) during routine obstetric care to identify pregnancies characterized by elevated stress exposure. Longitudinal monitoring throughout gestation could provide an early warning signal before neonatal dysbiosis emerges. For high‐risk pregnancies, targeted interventions might be developed to restore beneficial taxa commonly depleted following prenatal stress, including Bifidobacterium and Lactobacillus. At the population level, public health policies could increasingly recognize social determinants such as socioeconomic adversity and food insecurity as biological risk factors during pregnancy. Investing in integrated behavioral health services and social support programs during the prenatal period may ultimately help reduce the long‐term burden of inflammatory and neurodevelopmental disorders in future generations.

4.10. Future Directions: Toward Mechanistic and Translational Research

Despite growing evidence that PSE influences offspring microbiome development, several methodological and conceptual challenges continue to limit progress. Although many studies report statistically significant differences in microbial diversity or taxonomic abundance following PSE, few quantify the magnitude of these effects. Consequently, the biological significance of many reported findings remains difficult to evaluate, particularly given the substantial interindividual variability characteristic of the gut microbiome. Small statistically significant changes may have limited functional consequences, whereas potentially meaningful effects may be obscured by differences in analytical approaches. The absence of standardized effect size reporting further restricts cross‐study comparisons and complicates efforts to assess the developmental relevance of PSE‐associated microbiome alterations. Routine reporting of effect sizes alongside significance testing would improve the interpretability of findings and strengthen future meta‐analytic efforts. Another important limitation is the heavy reliance on rodent models. A comprehensive search conducted for this review identified only a small number of studies in non‐rodent species, including one study in pigs and two in macaques. Because these species have longer lifespans, extended gestational periods, more complex diets, and physiological characteristics that more closely resemble those of humans (Carlsson et al. 2004; Roura et al. 2016), expanding research in these models may enhance translational relevance and improve understanding of the long‐term developmental consequences of PSE. Beyond methodological improvements, future research should focus on clarifying the pathways that connect maternal stress‐related microbiome alterations to offspring outcomes. Research examining the effects of stress on maternal vaginal (Amabebe and Anumba 2018; Jasarevic et al. 2015, 2018; Sudhakaran et al. 2026), gut (Gur et al. 2017; Jasarevic et al. 2017; Matsunaga et al. 2024; Sudhakaran et al. 2026; J. Yu et al. 2024), and breast milk microbiomes (Juncker et al. 2025; J. Yu et al. 2024; Ziomkiewicz et al. 2021) has been expanding in recent years. However, comparatively less attention has been devoted to systematically characterizing downstream consequences for the offspring microbiome. Although existing studies suggest that prenatal stress can alter microbial composition during early life, the nature and consistency of these changes have not yet been comprehensively integrated across studies. Addressing this gap will require not only additional research but also greater methodological standardization to facilitate future quantitative syntheses. Longitudinal study designs with repeated sampling across developmental stages may be particularly valuable for identifying when stress‐related microbiome alterations emerge and how they evolve over time. Larger cohorts and standardized protocols for sample collection, sequencing, and statistical analysis would further improve reproducibility and comparability across studies. Additionally, while some investigations have examined maternal and infant microbiomes simultaneously (He et al. 2020; Pawluski et al. 2023; Warner et al. 2023), recent work has begun incorporating paternal factors as well (Dubois et al. 2024; Veerus et al. 2024; Simonyté et al. 2025). These studies suggest that both parents may contribute to infant microbiota establishment, opening new avenues for understanding intergenerational influences on microbiome development. Finally, advances in sequencing technologies offer promising opportunities for overcoming several limitations that have characterized earlier microbiome research. While most studies have relied on 16S rRNA sequencing, recent investigations have begun adopting whole‐genome metagenomic approaches (Eckermann et al. 2025). Compared with 16S‐based methods, metagenomic sequencing provides improved taxonomic resolution and may reduce errors related to microbial classification and relative abundance estimation. Combined with ongoing advances in bioinformatics and expanding microbial reference databases, these approaches have the potential to substantially improve the accuracy and interpretability of future studies examining the impact of prenatal stress on offspring microbiome development.

5. Conclusion

Despite the rapidly expanding literature on PSE, understanding of its effects on offspring gut microbiome development remains fragmented. By systematizing evidence from both human cohorts and animal models, this review identifies the gut microbiota as a plausible pathway linking maternal stress during pregnancy to offspring developmental outcomes. Although no single, uniform microbial signature emerged, the collective evidence suggests that PSE acts as a powerful ecological modifier of the developing gut ecosystem, altering microbial colonization trajectories during critical developmental windows.

Several cross‐species patterns were evident. Most notably, PSE was repeatedly associated with reductions in beneficial early‐life taxa, particularly Bifidobacterium and Lactobacillus, alongside a relative enrichment of potentially inflammatory and opportunistic taxa, including members of the Proteobacteria phylum. These alterations are consistent with a shift toward a more dysbiotic microbial profile and may have important implications for immune maturation, metabolic regulation, and microbiota–gut–brain axis functioning across the lifespan. At the same time, substantial inconsistencies emerged across studies. Animal studies consistently detected PSE‐related changes in beta diversity, suggesting that prenatal stress can restructure microbial community composition under controlled experimental conditions. In contrast, human cohorts showed little evidence of a reproducible association between PSE and microbial diversity, with findings for both alpha and beta diversity varying considerably across populations and stress measures. These discrepancies likely reflect the methodological and environmental heterogeneity that characterizes human microbiome research, including differences in stress assessment, sampling age, delivery mode, infant feeding practices, antibiotic exposure, sequencing approaches, and analytical methods. Taken together, the current evidence suggests that the challenge facing the field is no longer simply to demonstrate that prenatal stress influences the offspring microbiome, but rather to determine when, how, and under which conditions these effects occur. Addressing this question will require a shift from largely descriptive studies toward mechanistically informed, longitudinal research designs.

Several priorities should guide future work. In animal research, greater emphasis should be placed on translationally relevant species, particularly underrepresented long‐lived omnivores such as pigs and non‐human primates. Comparative studies examining psychological and physical stressors independently and in combination are especially needed to clarify whether distinct stress modalities converge on shared biological pathways. Standardization of intestinal sampling sites would be helpful to secure reproducible outcomes. In human cohorts, maternal psychological distress should be assessed longitudinally beginning in early pregnancy, with anxiety, depression, and socio‐environmental adversity examined as distinct constructs rather than combined into broad composite stress indices. Key postnatal determinants of microbiome development—including delivery mode, infant feeding practices, and antibiotic exposure—should be incorporated as primary covariates to better isolate prenatal effects. Longitudinal microbiome sampling across infancy is also needed to distinguish transient disruptions from persistent developmental alterations. Standardization of sequencing protocols, taxonomic reporting, and diversity analyses would further improve reproducibility and facilitate future quantitative syntheses and meta‐analyses.

Ultimately, advancing the field will require integrating microbiome data with neuroendocrine, immune, metabolic, and neurodevelopmental measures to elucidate the mechanisms through which maternal stress becomes biologically embedded in offspring health. Such efforts will be critical for determining whether microbial alterations represent merely biomarkers of prenatal adversity or active mediators of developmental programming. Resolving this question has important implications not only for understanding the biology of prenatal stress, but also for developing targeted interventions aimed at promoting lifelong health from the earliest stages of life.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material: dev70199‐sup‐0001‐SuppMat.xlsx

Data Availability Statement

The primary sources evaluated in this study are publicly available on PubMed, PsycINFO, and Scopus. The extracted study characteristics analyzed during this systematic review are included in the Supporting Information File.