The fetal exposome and Preterm Birth: a systematic synthesis of environmental exposures and multi-omics evidence

Study design

This study employed an integrative synthesis methodology to accommodate heterogeneous data types across epidemiologic, clinical, molecular, and translational domains. This design supported a systems-level interrogation of how the fetal exposome influences preterm birth (PTB) risk. Core elements of the PRISMA 2020 guidelines were incorporated to enhance transparency and reproducibility during study identification, screening, and reporting. The overall selection workflow is shown in Figure 9.

Figure 9:

PRISMA-inspired flow diagram of literature selection process. This PRISMA-style flow diagram illustrates the systematic literature selection process conducted for the integrative review. A total of 1,472 records were retrieved from four databases (PubMed, Scopus, Web of science, and EMBASE) spanning January 2000 to March 2025. After removing 314 duplicates, 1,158 records underwent title and abstract screening. From these, 224 full-text articles were assessed for eligibility based on predefined inclusion and exclusion criteria. Ultimately, 95 studies were included in the final synthesis, capturing a diverse range of environmental exposures – including chemical, psychosocial, nutritional, and structural domains – and employing multi-omic technologies such as epigenomics, transcriptomics, metabolomics, proteomics, and microbiomics.

Literature search strategy

A comprehensive search was conducted in PubMed/MEDLINE, Scopus, Web of Science, and EMBASE for articles published from January 2000 through March 2025. The search strategy combined controlled vocabulary and Boolean operators and included keywords such as “fetal exposome”, “preterm birth”, “multi-omics”, “epigenomics”, “proteomics”, “metabolomics”, “placenta”, “microbiome”, and “environmental exposures”. Searches were restricted to peer-reviewed, English-language human studies. To ensure completeness, reference lists of key publications were manually screened to identify additional eligible studies.

Eligibility criteria

Studies were included if they examined prenatal environmental exposures such as chemical, nutritional, psychosocial, microbial, or occupational factors and evaluated gestational age or PTB outcomes. Eligible studies applied at least one omics platform, including DNA methylation, transcriptomics, proteomics, metabolomics, or microbiome analysis, in maternal, fetal, or placental samples. Both original research and systematic reviews were considered when they provided mechanistic insights relevant to PTB. Studies were excluded if they focused only on postnatal outcomes, used animal or in vitro models without translational linkage, or were editorials, commentaries, or non–peer-reviewed sources.

Screening and selection

Article screening was performed in three stages: title review, abstract review, and full-text assessment. Two independent reviewers evaluated all records for eligibility, and any disagreements were resolved by consensus or adjudicated by a third reviewer. Inter-rater agreement was quantified using Cohen’s kappa (κ=0.88), indicating strong concordance. After duplicate removal, 1,214 records were screened, resulting in 95 studies retained for final synthesis. The selection process is illustrated in Figure 9.

Data extraction and quality assessment

A standardized data extraction matrix was used to collect study attributes, including author and publication year, population characteristics, geographic setting, exposure type and timing, biological sample used, omic platform applied, key molecular findings, and reported limitations. Studies were categorized according to exposure domain and omic focus. The Newcastle–Ottawa Scale was applied to assess study quality, emphasizing sample size, diversity of the population studied, reliability of exposure assessment, and reproducibility of the omic platform. These risk-of-bias assessments were integrated into the interpretation of findings to account for variability in study design and reporting quality.

Thematic synthesis approach

Because of heterogeneity in exposure definitions, outcome classifications, and omics platforms, a formal meta-analysis was not feasible. Instead, a narrative thematic synthesis was conducted. Studies were clustered by exposure domain such as air pollution, endocrine-disrupting chemicals, psychosocial stress, nutritional patterns, occupational hazards, climate stress, and microbiome alterations. They were further compared according to pregnancy period (early, mid, or late gestation) and outcome subtype (spontaneous vs. indicated PTB). These groupings enabled identification of convergent molecular themes, including oxidative stress, immune activation, endocrine disruption, and placental aging, that were shared across exposure and omic layers. The main environmental exposures and their biological effects are summarized in Table 1, while omic methodologies and biological compartments analyzed are detailed in Tables 2–5. Integrative models of exposure–omics interactions are illustrated in Figure 6, and a trimester-based vulnerability timeline is provided in Figure 1. Findings from this synthesis informed the development of a translational roadmap (Figures 7 and 9) and a proposed framework for clinical implementation (Table 6) for early detection and prevention of PTB.

Study selection and characteristics

Across four databases (PubMed, Scopus, Web of Science, and EMBASE), 1,472 records were retrieved between January 2000 and March 2025. After removing 314 duplicates, 1,158 titles and abstracts were screened, resulting in 224 full-text reviews. Of these, 95 studies met the inclusion criteria. These studies examined exposures to environmental chemicals, psychosocial stressors, nutritional deficiencies, structural inequities, and microbiota alterations, applying omics platforms to maternal, fetal, or placental samples. When reported, we distinguished between spontaneous and indicated PTB, although most studies aggregated these outcomes, which we note as a limitation. The geographic scope spanned high- and low-income settings on five continents. Omics approaches included epigenomics, transcriptomics, proteomics, metabolomics, and microbiomics, with study designs ranging from birth cohorts and case-control analyses to meta-analyses and translational models. Figure 9 summarizes the selection process, while Table 1 includes a formal Newcastle–Ottawa Scale (NOS) quality assessment score for each included study.

Air pollution and multi-omic disruption

Exposure to fine particulate matter (PM₂.₅) was consistently associated with an increased risk of preterm birth and disruption of multiple molecular pathways, with adjusted odds ratios typically in the range of 1.2–1.4 per 10 μg/m³ increase. DNA methylation studies revealed alterations in immune and vascular regulatory loci, including genes involved in endothelial stability and inflammatory signaling [2], 16], 48]. Metabolomic studies demonstrated disturbances in amino acid and lipid metabolism, with early-to mid-gestation exposure linked to disrupted carnitine and phospholipid pathways [8], 12], 64]. Transcriptomic analyses identified activation of hypoxia-inducible factor signaling (HIF1A) and downstream responses associated with placental oxidative stress [4], 39]. Inflammatory biomarkers in maternal and cord blood further confirmed immune activation [18], 48]. Homocysteine elevation and endothelial dysfunction were observed as intermediate phenotypes mediating exposure effects [64]. Figure 5 illustrates the biological cascade linking maternal air pollution exposure to systemic inflammation, endocrine disruption, and placental dysfunction, while Table 2 summarizes key studies by sample type, omics platform, and pathway relevance.

Endocrine-disrupting chemicals and fetal programming

Prenatal exposure to phthalates, bisphenol A, and per- and polyfluoroalkyl substances (PFAS) consistently altered placental hormone signaling and immune tolerance pathways, with relative risks generally 1.3–1.5 in high-exposure groups [10], 20], 21], 25], 53]. Epigenetic analyses identified differential methylation at hormone and growth-related loci, including ESR1, NR3C1, and IGF2 [2], 8], 52], 57]. Transcriptomic data revealed disrupted trophoblast invasion and immune regulation, while proteomic findings showed altered extracellular matrix and angiogenic proteins [20], 53]. Metabolomic studies demonstrated shifts in bile acid metabolism and lipidomic profiles, particularly in nutritionally vulnerable populations [53], 93]. Table 3 highlights representative studies integrating exposure and omics data across environmental domains, while Figure 2 illustrates convergent molecular response patterns.

Psychosocial stress and epigenomic remodeling

Maternal stress, anxiety, and chronic socioeconomic strain were associated with epigenetic modifications in hypothalamic–pituitary–adrenal (HPA) axis regulatory genes, notably FKBP5 and NR3C1, with odds of PTB typically 1.3–1.6 across cohorts [5], 14], 24], 49]. These effects were most pronounced during second-trimester exposures [49], 101]. Transcriptomic evidence revealed increased expression of inflammatory cytokines and reduced expression of immune-modulating genes [6], 36]. Experimental models demonstrated that pharmacologic inhibition of FKBP51 reversed stress-induced PTB [32]. Reduced placental 11β-HSD2 activity increased fetal glucocorticoid exposure and impaired stress regulation [24], 49]. These findings are visualized in Figure 4, which maps hormonal pathways and epigenetic remodeling linked to immune dysregulation and preterm labor.

Social determinants and omic disparities

Structural inequities, including racism, poverty, and environmental injustice, were linked to accelerated epigenetic aging, immune exhaustion, and mitochondrial dysfunction within the placenta [9], 35], 43], 49], 72], 74], [81], [82], [83]. Methylation changes in stress-responsive genes were observed in racially stratified cohorts, particularly among Black and Indigenous populations [49], 83]. Allostatic load was evidenced by telomere shortening, altered cortisol metabolism, and systemic inflammation [43], 74], 82]. Metabolomic analyses revealed micronutrient deficiencies and altered lipid homeostasis [30], 53], while transcriptomic profiles indicated downregulation of vascular and growth-related genes [39], 95]. Table 5 integrates omic evidence showing how social determinants shape molecular risk profiles associated with PTB.

Convergent mechanisms across omic layers

Despite varied exposures, recurring biological mechanisms emerged, including inflammation, oxidative stress, endocrine dysregulation, immune imbalance, and mitochondrial dysfunction [2], 4], 6], 10], 28], 42], 58], 65], 83]. Epigenetic shifts in vascular and immune loci were replicated across multiple cohorts [17], 24], 46], 97]. Transcriptomic and proteomic studies confirmed extracellular matrix remodeling and cytokine network activation [6], 42], 57], 100], while metabolomics highlighted disruptions in amino acid and lipid metabolism [11], 53], 56], 93]. Integrated systems biology analyses identified mechanistic hubs common to multiple exposures, as depicted in Figure 7, which synthesizes exposure timing and molecular embedding along the maternal–fetal axis. Table 4 summarizes key molecular pathways, including signaling cascades involving IL6, TNF-α, and glucocorticoid receptors.

Integration of multi-omics into clinical framework

Advances in exposomic and omic science have yielded candidate biomarkers for early PTB risk detection. DNA methylation signatures in NR3C1 and FKBP5 [5], 14], 24], metabolomic shifts in lipid and carnitine pathways [8], 12], 56], and microbiome changes linked to maternal dysbiosis [36], 41], 78] demonstrated potential for early risk stratification. These molecular features, identified in maternal blood, placental tissue, or cord blood, may enable risk categorization well before clinical symptoms manifest.

Omic-informed prenatal care frameworks could be deployed across pregnancy, such as first-trimester cfDNA methylation screening for placental signaling dysfunction [24], 49] and second-trimester metabolomics to detect oxidative stress and inflammatory profiles associated with spontaneous PTB [20], 53], 93]. Table 6 presents a proposed implementation model integrating these omic tools into prenatal care, while Figure 8 visually maps how environmental exposures translate into molecular changes and clinical risk profiles.

Translation to clinical practice will require overcoming barriers such as cost-effectiveness, assay standardization, bioinformatic integration, and population validation [43], 52], 85]. Publication bias toward positive findings may also inflate biomarker performance, underscoring the need for pre-registration and open-access datasets. Nevertheless, the integration of exposomic and multi-omic approaches provides a biologically grounded, temporally sensitive, and equity-informed path toward precision-guided perinatal medicine.

Toward a systems biology of Preterm Birth: the multi-omic exposome frontier

Preterm birth (PTB) is now recognized not as a single pathological endpoint but as a complex, multifactorial syndrome shaped by interactions among genetic, environmental, biological, and social determinants. Advances in multi-omic technologies – including epigenomics, transcriptomics, proteomics, metabolomics, and microbiomics – have revealed how prenatal exposures are molecularly embedded, transforming PTB from a downstream obstetric complication into a systemic, temporally dynamic, and socially contextualized phenotype [6], 17], 20], 49], 52]. These platforms demonstrate how maternal exposures are transcribed into fetal cellular programs, altering developmental trajectories and influencing lifelong health risks [2], 8], 17], 41], 50]. This systems approach integrates molecular and social layers, enabling identification of early, modifiable biomarkers and establishing a foundation for precision-guided perinatal care [10], 33], 54], 95].

Quality assessment and study reliability

A major strength of this synthesis is its explicit evaluation of study quality using the Newcastle–Ottawa Scale (NOS), as reflected in Table 1, which provides quality grades for all key studies. Most included studies achieved moderate-to-high quality scores (≥6), indicating adequate control for confounding and robust exposure assessment, though variability in cohort designs, biospecimen selection, and omic platforms remains an important limitation. The integration of these NOS ratings improves confidence in reported findings and addresses prior reviewer concerns about quality assessment transparency.

The omic architecture of environmental and chemical insults

Airborne toxicants, especially PM2.5, NO₂, and black carbon, perturb placental homeostasis via oxidative stress, lipid peroxidation, and inflammatory priming [1], 3], 4], 7], [11], [12], [13, 18], 48], 56], 64], 72]. Epigenomic analyses (e.g., HELIX cohort) show reproducible methylation changes in angiogenesis, immune, and metabolic loci [2], 8], 17], reflecting systemic inflammation and altered transcriptomic signatures linked to hypoxia and vascular remodeling [4], 11], 39]. Reported effect sizes for PM2.5 consistently show 1.2–1.4 adjusted odds ratios for PTB per 10 μg/m³ increase, highlighting exposure consistency despite heterogeneous study designs.

Endocrine-disrupting chemicals (EDCs) such as phthalates, PFAS, and bisphenols similarly disrupt hormonal and immune pathways, affecting trophoblast invasion and extracellular matrix stability [10], 20], 21], 52], 53], 66], 91], 100]. Effect estimates for EDC exposure are smaller (odds ratios ∼1.1–1.3 per IQR change) but biologically consistent, particularly for glucocorticoid signaling and WNT pathway disruption. Nanomaterials, including nanoplastics, are emerging as high-priority research targets, with initial evidence linking them to ferroptosis and early placental senescence [69], 87].

Psychosocial stress and epigenomic imprinting

Maternal psychosocial stress – from trauma and discrimination to chronic socioeconomic adversity – produces persistent epigenetic modifications in key HPA axis and immune regulatory loci (NR3C1, FKBP5, IL6) [5], 14], 24], 32], 49]. These molecular signatures parallel physiological consequences including altered cortisol dynamics, inflammatory cytokine profiles, and endothelial dysfunction [23], 32], 84]. Mid-gestation exposures appear most sensitive, with sex-specific fetal responses suggesting differential vulnerability [14], 32]. Effect size estimates for stress-related exposures are smaller (risk ratios ∼1.1–1.2), but findings remain consistent across birth cohort and epigenome-wide association study designs.

Notably, protective factors such as social support and resilience buffer epigenomic risk, yet few studies stratify analyses by these moderators [9], 74]. Preclinical evidence indicates that targeting stress signaling, including FKBP51 antagonism, can reverse PTB phenotypes [32], 37], representing a potential translational pathway for intervention trials.

Social determinants as biological determinants

Structural inequities – including systemic racism, occupational hazards, environmental injustice, and limited healthcare access – leave measurable molecular imprints. Multi-omic studies reveal accelerated epigenetic aging, telomere attrition, and immune exhaustion among racially and socioeconomically marginalized populations [35], 43], 49], 73], [81], [82], [83, 97]. The NuMoM2b, JECS, and EPIC cohorts demonstrate how maternal disadvantage predicts placental inflammation and altered metabolic regulation, confirming social determinants as upstream biological drivers of PTB risk [43], 49], 73].

Figure 7 and Table 5 illustrate how gradients of adversity translate into convergent molecular signatures spanning DNA methylation, transcriptomic stress responses, proteomic disruption, and metabolomic depletion. These findings underscore the ethical imperative to embed health equity into exposome–omics research and to contextualize biomarker interpretation within social frameworks.

Microbiome–immune–neuroendocrine interactions

Microbiome dysbiosis influences uterine tone, cervical remodeling, and systemic immunity through metabolites such as short-chain fatty acids and lipopolysaccharides [34], 41], 76], [78], [79], [80, 83]. NLRP3 inflammasome activation provides a mechanistic link between dysbiosis and inflammatory labor initiation [62], 78], 92]. The microbiome also modulates nutrient and methyl donor availability, intersecting with epigenetic fetal programming [80], 84]. Interventions such as probiotics and dietary fiber have shown early promise [76], 80], though translation is limited by strain specificity and lack of longitudinal omics-integrated clinical trials.

Clinical implications, cost-effectiveness, and publication bias

Multi-omic biomarkers – including DNA methylation (NR3C1, FKBP5), metabolomic lipid and carnitine profiles, and microbiome signatures – hold promise for early PTB risk stratification [5], 8], 12], 24], 36], 41], 56], 78]. Predictive modeling shows AUC values ranging from 0.70–0.85, demonstrating reasonable accuracy [33], 54], 85], 95]. However, cost-effectiveness remains a major translational barrier: high-throughput sequencing and metabolomics currently exceed the budget of most routine obstetric care settings. Moreover, publication bias favoring positive findings may overestimate biomarker readiness. Future research should include economic evaluations and preregistered, replication-focused omics studies to ensure realistic and equitable clinical adoption.

Future interventions and research priorities

The findings highlight actionable pathways for prevention trials, including environmental remediation (e.g., air quality interventions), stress mitigation (e.g., social support programs, FKBP51 antagonists), and microbiome modulation (e.g., probiotics). Future directions include deploying single-cell spatial omics to localize placental dysregulation, integrating causal inference models to distinguish spontaneous from indicated PTB mechanisms, and linking omics data to real-world interventions through federated data networks [17], 39], 44], 46], 54].

Figure 9 and Table 6 outline how exposomic data can inform stratified prenatal care models, emphasizing the need for trial-based validation to move from biological plausibility to clinical utility.

Key messages for clinicians and policymakers

1. Preterm birth (PTB) is driven by integrated environmental, social, and biological exposures beginning early in gestation.

2. The fetal exposome framework, when combined with multi-omic data, provides unique mechanistic insights into PTB etiology.

3. Omic biomarkers (e.g., DNA methylation, lipid and inflammatory signatures) show potential for early risk stratification and personalized prenatal care.

4. Psychosocial adversity and structural inequities produce measurable biological imprints, necessitating socially informed clinical strategies.

5. Advancing exposomic science requires investment in omics infrastructure, cost-effectiveness analyses, and global harmonization for clinical translation.

Strengths

This review uniquely integrates literature across chemical, biological, and social domains within a unified multi-omic framework. It synthesizes findings from diverse global cohorts, spanning air pollution, endocrine-disrupting chemicals, psychosocial stress, nutrition, occupational hazards, and microbiome alterations, and links these exposures to molecular mechanisms of preterm birth (PTB). The analysis leverages emerging omics technologies, including epigenomics, transcriptomics, proteomics, and metabolomics, to illuminate convergent pathways and biological signatures. By combining these data within a fetal exposome conceptual model, this review advances the field toward precision-guided prenatal risk stratification and intervention design. Figures 1–9 and Tables 1–6 collectively illustrate this integrated approach, providing clinicians and researchers with a comprehensive resource for both mechanistic insight and translational innovation.

Limitations

Despite its scope, this review faces inherent challenges. Cohort heterogeneity – including differences in population characteristics, environmental exposures, omic platforms, and analytical pipelines – limits direct comparability and meta-analytic synthesis. Most studies lack preconceptional exposure data and standardized sampling protocols, restricting causal inference. Marginalized populations remain underrepresented in omics studies, raising concerns about generalizability and equity. Technical barriers, such as data interoperability, batch effects, and limited cross-platform validation, further constrain clinical translation. Additionally, while omic technologies can identify associations and biomarkers, their mechanistic interpretation often remains incomplete, and temporal directionality is frequently ambiguous.

Future directions

Several priorities emerge from this synthesis. Multi-omic causal models should be developed to resolve temporal and directional uncertainties, enabling mechanistic inference rather than correlative interpretation. Single-cell and spatial omics offer an opportunity to localize cellular sources of dysregulation within the placenta and fetal membranes, capturing microenvironmental heterogeneity. Expanding open-access biorepositories and longitudinal data networks will allow harmonized analysis across global populations. Community-based participatory research designs should be prioritized to ensure contextual relevance and ethical integrity, particularly in historically marginalized groups. Finally, integrating machine learning with exposomic and omic data will accelerate biomarker discovery and enable predictive tools for early risk stratification, transforming prenatal care from reactive to proactive.