Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2026 Feb 13;60(8):6048–6056. doi: 10.1021/acs.est.5c14467

Prenatal Exposure to Artificial Light at Night and the Offspring’s First 1000-Day Growth: A Prospective Metabolomic and Gene-Environment Interaction Study

Wen Jiang †, Azhu Han †, Yun Huang †, Zhichao Yuan †, Jian Xu ‡,*, Jun Zhang †,‡,*
PMCID: PMC12961921  PMID: 41686999

Abstract

Our previous exposome study suggested associations between prenatal exposure to artificial light at night (ALAN) and offspring weight and fat growth during the first 1000 days. This study aims to further explore the underlying biomechanism and modifying effects of maternal genetic regulators. Among 1944 mother-child pairs from the Shanghai Birth Cohort, meet-in-the-middle, mediation, and enrichment analyses were combined with the early pregnancy untargeted metabolome to explore potential biological links between prenatal ALAN exposure and early life growth. Maternal polygenetic risk scores (PRS) for glucose- and lipid-metabolism-related phenotypes were constructed and dichotomized (high/low) to examine gene-environment interactions. We identified 35 mediating metabolites between prenatal ALAN exposure and growth outcomes, which were enriched in neural signal transduction-related pathways and associated with the circadian rhythm. The effects of ALAN exposure on fetal and child growth were more pronounced among mothers with high PRS levels for fasting glucose, HbA1c, and triglycerides. To sum up, prenatal ALAN exposure may be associated with reduced weight and adiposity gains during the first 1000 days. These associations may be partly explained by disturbances in maternal circadian rhythms and appeared more pronounced among mothers with high genetic predispositions to glucose and triglyceride levels.

Keywords: artificial light at night, fetal growth, child growth trajectory, metabolomics, gene-environment interaction

graphic file with name es5c14467_0006.jpg

graphic file with name es5c14467_0004.jpg

1. Introduction

The first 1000 days of a child’s life, spanning from conception until 2 years old, are a unique period for establishing optimal lifelong health and for environmental susceptibility as well. Improper growth in the period has been shown to profoundly influence a child’s ability to learn and thrive, and to be associated with increased risks of prevalent chronic diseases, including obesity, diabetes, and cardiovascular diseases. As a result, exploring early life growth-associated environmental factors and proposing appropriate interventions are the first steps toward protecting and promoting human health.

Following the rapid urbanization of the past few decades, exposure to artificial light at night (ALAN) has become increasingly widespread globally. It was estimated that ALAN has covered 49.5% of the Earth’s land surface between 55° S and 59° N, affecting more than 83% of the population, and the radiation area continues to grow by 2.2% per year. , Given the key role of natural light-dark cycles in regulating many human behaviors and physiological functions, exposure to ALAN has been posited to be harmful to human health and well-being in recent years. Accumulating evidence has suggested potential links between ALAN exposure and the incidence of cancers, circadian rhythm disorders, cardiometabolic diseases, mental disorders, and infectious diseases.

Despite growing concerns about the health effects of ALAN exposure, the association between prenatal ALAN exposure and early life growth remains underexplored. To date, only two studies are available, , one of which observed a positive association between ALAN and low birth weight. In contrast, the other found no significant effect of ALAN on birth weight. Neither study included other anthropometric measurements, such as body length and weight-for-length, to achieve a more comprehensive evaluation on growth status, nor incorporated the corresponding longitudinal growth trend indexes, which are comparatively more important in predicting long-term health. , Moreover, the biological mechanisms linking prenatal ALAN exposure to early life growth remain unclear, reducing the strength of the potential causal relationship. More importantly, although offspring’s growth has been proven to be affected by maternal genetics that regulate metabolism, and ALAN exposure has been observed to lead to the disturbance of glucose and lipid metabolism, − there is a lack of exploration on the potential interactions of maternal glucose and lipid metabolism genetic predispositions with ALAN exposure on early life growth, which would facilitate discovering subgroups of susceptible populations.

In our recent exposome study involving 70 prenatal environmental factors, we found that ALAN was associated with decreased neonatal sex-specific weight-for-age (WAZ) and weight-for-length (WLZ) z-scores, increased risk of slow WAZ growth trajectory, and decreased risk of rapid WLZ growth trajectory during 0–2 years of age, suggesting prenatal exposure to ALAN could continuously affect weight and adiposity gains in early life. In this study, we aimed to further quantify the associations of prenatal ALAN exposure with these growth outcomes in the same study population. We then incorporated untargeted metabolomics, a promising technology for detecting endogenous changes in response to environmental stimuli, to measure serum samples from early pregnancy and explore the underlying biological mechanisms. Ultimately, we performed genome-wide gene-environment interaction analysis to assess whether the associations between ALAN exposure and early life growth were modified by genetic susceptibility to maternal glucose and lipid metabolism.

2. Methods

2.1. Study Population

The Shanghai Birth Cohort (SBC) is a multicenter cohort study that began in 2013 and aims to examine environmental, behavioral, and genetic factors affecting fecundability, pregnancy, child health, and disease risk. The cohort enrolled 4127 couples from six participating hospitals in Shanghai, China, from the launch to 2016, with continuous and periodic follow-up, and is still ongoing. A comprehensive description of the cohort profile can be found elsewhere. In this study, we excluded those lost to follow-up (n = 303), miscarriages and stillbirths (n = 132), and multiple pregnancies (n = 51), as well as those without any child body measurement during the first two-year follow-up (n = 348) and without genome-wide genotyping data (n = 1349), leaving 1944 mother-child pairs for analysis (Figure S1).

The study was approved by the Ethics Review Board of all participating hospitals (XHEC-C-2013-001). All participating women signed informed consent for themselves and their children.

2.2. Exposure Assessment

Data on ALAN (nW/cm2/sr) were derived from the high-quality global night-time light product produced by the Earth Observation Group, which has a monthly temporal resolution and a 500 × 500 m spatial resolution at the equator. Specifically, the data set originates from high-quality night-time Earth images collected by the Visible and Infrared Imaging Suite (VIIRS) Day Night Band (DNB) aboard the Joint Polar-orbiting Satellite System (JPSS) satellites. All maps have undergone preprocessing to exclude background noise, solar and lunar contamination, data degraded by cloud cover, and features unrelated to electric lighting (e.g., fires, flares, and volcanoes). We downloaded ALAN maps from 2013 to 2016 from https://payneinstitute.mines.edu/ and converted the maternal residential addresses during pregnancy into longitude and latitude using Baidu map (https://map.baidu.com). The geocoded addresses were then used to extract ALAN data from the map using ArcGIS 10.8 (ESRI, Redlands, CA).

We averaged the ALAN data across the pregnancy to use as the exposure metric in the exposure-outcome association and gene-environment interaction analyses. Given that the maternal metabolome was detected using the early pregnancy serum samples, we also calculated the average ALAN in early pregnancy (0–12 gestational weeks) for metabolic analysis to ensure the time sequence of exposure, metabolome, and outcomes (see below). Additionally, fine particulate matter (PM2.5) data (μg/m3) in daily and 1 km temporospatial resolution from a validated random forest product, as well as daily mean temperature (°C) and relative humidity (%) of the nearest national meteorological monitoring station (https://data.cma.cn/) from residential address during pregnancy, were used to estimate average PM2.5, temperature, and relative humidity exposure levels across the entire pregnancy and early pregnancy.

2.3. Outcome Assessment

Based on the findings from our previous exposome study, outcomes of interest include neonatal WAZ and WLZ, as well as the risks of slow WAZ growth trajectory and rapid WLZ growth trajectory between 0 and 2 years, compared with the normal trajectories. Details of the outcome assessment have been illustrated earlier. In brief, WAZ and WLZ at birth and at each follow-up visit (postnatal 42 days, 6 months, 12 months, and 24 months) were calculated from original weight and length measured using the World Health Organization (WHO) Child Growth Standards (version 3.2.2). For preterm births (n = 86), the age was corrected before generating z-scores. The growth trajectories of WAZ and WLZ were constructed using growth mixture models (GMM), in which the ones with the lowest Bayesian information criterion (BIC) and posterior group probabilities greater than 0.7 for all classes were identified as the best-fitted.

2.4. Serum Metabolome Profiling

Untargeted metabolomics profiling of early pregnancy serum (9–16 gestational weeks) was conducted in 1723 out of 1944 mothers included using ultrahigh-performance liquid chromatography–mass spectrometry (UPLC-MS). Details on the profiling process are given in Supporting Information Text 1. After quality control, 413 annotated metabolites were left for analysis. Missingness of the metabolites was imputed using their medians from semiquantifications, and concentrations were log2-transformed to reduce skewness before analysis.

2.5. Genotyping

Maternal DNA was extracted from venous blood during pregnancy with the TGuide Large Volume Blood Genomic DNA kit (Tiangen, OSR-M104), followed by genotyping with the Illumina Global Screening Array v.3.0 (Illumina; San Diego, CA) that originally contains 7,30,059 variants. Quality control and single-nucleotide polymorphism (SNP) imputation were then performed, as detailed elsewhere. The processes ultimately resulted in 81,99,139 SNPs for analysis.

2.6. Polygenetic Risk Score Construction

To reflect maternal genetic predisposition to glucose and lipid metabolism, we constructed five PRS of glucose metabolism-related phenotypes, including type 2 diabetes (T2D), fasting glucose (FG), fasting insulin (FI), two-hour oral glucose tolerance test (OGTT2h), and HbA1c, and five PRS of lipid metabolism-related phenotypes, including total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), and nonhigh-density lipoprotein cholesterol (NHDL), using PRSice 2.0 software. More details on the selection, validation, and calculation processes are provided in Supporting Information Text 2, and information on the SNPs used to calculate the PRS is summarized in Table S1.

2.7. Covariates

In our previous exposome study, the same set of covariates was adjusted for all environmental factors, involving maternal age (continuous, years), preconceptional body mass index (BMI) (continuous, kg/m2), weight gain during pregnancy (continuous, kg), parity (nulliparous/multiparous), education levels (high school or lower/college/master’s degree or above), self-reported financial status (very good/good/relatively poor), the season of conception (spring/summer/autumn/winter), the year of conception (2013/2014/2015), neonatal sex (male/female), and feeding mode before 6 months of age (breastfeeding/artificial feeding, only for child growth trajectory). We further included PM2.5, temperature, and relative humidity as covariates in this study. The selection of these covariates was determined by existing evidence and a directed acyclic graph (Figure S2).

2.8. Statistical Analysis

For descriptive analysis, continuous variables were presented as mean ± standard deviation (SD) or median (interquartile range, IQR), and categorical variables were presented as frequency (proportion) [N (%)]. The missingness in ALAN and covariates was imputed using the chained equations method, where variables correlated at levels between 0.2 and 0.8 in absolute value, or with missingness correlated ≤ 0.4 in absolute value, were used to predict each other. The process created 10 data sets with imputed missing data, of which the first was used for subsequent analyses to reduce computational burden.

Multivariable linear and logistic regression models were used to quantify associations between ALAN exposures during the entire and early pregnancy periods and growth outcomes, with ALAN treated as both a continuous variable and a categorical variable (tertiles). For continuous ALAN, the results were presented as regression coefficients (β) or risk ratios (RRs) with 95% confidence intervals (CIs) for a SD increase in exposure levels. For categorical ALAN, linear trend tests were conducted using the medians within each tertile as continuous variables. The restricted cubic spline (RCS) function was employed to visualize the exposure-response relationships and assess the potential nonlinearity. Specifically, RCS models with three to six knots were fitted successively, among which the ones with the smallest Akaike Information Criterion (AIC) were selected (see Table S2 to obtain the exact number of knots selected for different RCS models). The medians of ALAN in different exposure windows were used as reference points. Additionally, stratified analyses were performed to assess the effect modification by maternal age (≤28/>28 years, based on the mean of the distribution), preconceptional BMI (<18.5/18.5–24.0/>24.0), self-reported financial status, neonatal sex, and the season of conception on entire-pregnancy ALAN exposure. Heterogeneity across strata was evaluated using likelihood ratio tests.

The metabolome-wide association study (MWAS), the meet-in-the-middle (MITM) approach, mediation analysis, and functional enrichment analysis were combined to explore the underlying mechanisms linking prenatal ALAN exposure to early life growth. Specifically, the MWAS, adjusting for all of the aforementioned covariates except weight gain during pregnancy, was first used to screen for early pregnancy serum metabolites significantly associated with first-trimester ALAN and neonatal WAZ/WLZ (P < 0.05). The MITM approach and mediation analysis were then combined to identify metabolites associated with both the exposure and outcomes, as well as those with significant mediation effects (P < 0.05). Lastly, a functional enrichment analysis of the mediating metabolites was performed using the online MetaboAnalyst 6.0 software (https://www.metaboanalyst.ca/) to identify potential biological pathways (false discovery rate, FDR < 0.05), from which the top 10 were selected for visualization.

To assess the gene-environment interactions of prenatal ALAN exposure with maternal glucose and lipid-metabolic genetic predispositions, a product term of entire-pregnancy ALAN and each PRS of metabolism-related phenotypes was successively included in multivariable linear and logistic regression models, where PRS were dichotomized by the median of the distributions in the study population (high/low). In addition to basic covariates, the top 10 genetic principal components were included in all models to account for the residual population structure. A P for interaction <0.10 was considered statistically significant.

Several sensitivity analyses were performed to test the robustness of the results. First, we further adjusted for metabolic equivalent tasks (METs), derived from the International Physical Activity Questionnaire administered in early pregnancy, in the exposure-outcome regression models to account for potential confounding by physical activity intensity. In addition, we averaged ALAN levels during the three months preceding conception as an alternative exposure indicator to replicate the metabolomics analyses. All statistical analyses were performed using R software (version 4.2.1, “mice,” “stat,” “rms,” and “mediation” packages).

3. Results

3.1. Descriptive Statistics

The baseline characteristics of the included 1944 mother-child pairs and the growth trajectories of WAZ and WLZ have been described in detail in our previous study and are shown in Table S3 and Figure S3. The medians (IQRs) of ALAN, PM2.5, temperature, and relative humidity were 30.94 (11.45) nW/cm2/sr, 50.68 (7.11) μg/m3, 17.54 (4.13) °C, and 72.93 (2.77) % in the entire pregnancy and were 30.43 (12.30) nW/cm2/sr, 51.14 (22.36) μg/m3, 14.58 (13.61) °C, and 71.00 (5.94) % in early pregnancy, respectively (Table S4).

3.2. ALAN-Growth Outcome Associations

Table presents the associations between continuous and tertile levels of ALAN during the entire and early pregnancy periods and growth outcomes. Each SD increase in the entire pregnancy ALAN was associated with 0.06 (95% CI: −0.10, −0.01) and 0.08 (95% CI: −0.14, −0.02) decreased neonatal WAZ and WLZ, as well as 14% elevated slow WAZ growth trajectory risks (RR = 1.14, 95% CI: 1.01, 1.29) and 10% decreased rapid WLZ growth trajectory risks (RR = 0.90, 95% CI: 0.81, 0.99), respectively. When entire-pregnancy ALAN was modeled as tertile categories, the higher tertiles showed consistent and inverse associations with neonatal WAZ (P trend = 0.08), WLZ (P trend = 0.03), and rapid WLZ growth trajectory risk (P trend = 0.08), and were positively associated with slow WAZ growth trajectory risk (P trend = 0.06), compared to the lowest tertile. Consistent and similar association patterns with growth outcomes were also observed for early pregnancy ALAN exposure. Furthermore, the best-fitted RCS regression models did not show significant nonlinear relationships between ALAN exposure and growth outcomes (P for nonlinearity >0.05), although the exposure-response curves tended to be flatter at higher levels (Figures and S4). No significant differences in the associations were observed when stratified by maternal age, preconceptional BMI, self-reported financial status, neonatal sex, and the season of conception (Table S5).

1. Associations of ALAN Exposure during the Entire Pregnancy and the First Trimester with Fetal and Child Growth Outcomes ,

  neonatal WAZ
 neonatal WLZ
 slow WAZ trajectory risk
 rapid WLZ trajectory risk
  β (95% CI) RR (95% CI)
entire pregnancy
continuous –0.06 (−0.10, −0.01) –0.08 (−0.14, −0.02) 1.14 (1.01, 1.29) 0.90 (0.81, 0.99)
tertile 1 0.00 (ref) 0.00 (ref) 1.00 (ref) 1.00 (ref)
tertile 2 –0.04 (−0.14, 0.06) –0.04 (−0.17, 0.10) 1.19 (0.90, 1.58) 0.69 (0.54, 0.88)
tertile 3 –0.09 (−0.19, −0.01) –0.15 (−0.29, −0.01) 1.32 (1.01, 1.76) 0.81 (0.63, 1.04)
P trend 0.08 0.03 0.06 0.11
first trimester
continuous –0.06 (−0.10, −0.01) –0.07 (−0.13, −0.01) 1.12 (1.01, 1.27) 0.93 (0.84, 1.04)
tertile 1 0.00 (ref) 0.00 (ref) 1.00 (ref) 1.00 (ref)
tertile 2 –0.13 (−0.23, −0.03) –0.16 (−0.30, −0.03) 1.08 (0.81, 1.43) 0.84 (0.66, 1.08)
tertile 3 –0.13 (−0.23, −0.03) –0.19 (−0.32, −0.05) 1.34 (1.01, 1.79) 0.85 (0.66, 1.10)
P trend 0.02 0.01 0.04 0.22
Open in a new tab
a

Note: ALAN, artificial light at night; RR, risk ratio; CI, confidence interval; WAZ, weight-for-age z score; WLZ, weight-for-length z score.

b

All models were adjusted for maternal age, preconceptional BMI, weight gain during pregnancy, parity, education levels, self-reported financial status, the season and year of conception, neonatal sex, and averages of PM2.5, temperature, and relative humidity during the entire pregnancy or the first trimester. Logistic regression models were also adjusted for feeding mode before 6 months of age.

c

Multivariable linear regression models.

d

Multivariable logistic regression models.

1.

1

Open in a new tab

Restricted cubic spline curves of entire-pregnancy ALAN exposure levels and (A) neonatal WAZ, (B) neonatal WLZ, (C) slow WAZ growth trajectory risk, and (D) rapid WLZ growth trajectory risk. WAZ, weight-for-age z score; WLZ, weight-for-length z score. All models were adjusted for maternal age, preconceptional BMI, weight gain during pregnancy, parity, education levels, self-reported financial status, the season and year of conception, neonatal sex, and averages of PM2.5, temperature, and relative humidity during the entire pregnancy.

3.3. Metabolomic Analyses

Among the 413 metabolites included, 406 had missing rates of less than 10% (Table S6). In the ALAN MWAS, 123 out of 413 metabolites were found to be associated with the early pregnancy ALAN exposure (P < 0.05, Figure S5 and Table S6), with 80 showing negative associations and 43 showing positive associations, which mainly included fatty acids (FAs), acylcarnitines (CARs), lysophosphatidylcholines (LPCs), phosphatidylcholines (PCs), amino acids (AAs) and derivatives, steroids hormones and metabolites (e.g., androsterone sulfate, estrone sulfate, and pregnanediol 3-O-glucuronid), and N-acylethanolamines (NEAs, e.g., linoleoyl ethanolamide (LEA), oleoylethanolamide (OEA), and palmitoylethanolamide (PEA)). The outcome MWAS identified 108 and 70 metabolites to be associated with neonatal WAZ and WLZ, respectively (P < 0.05, Figure S5 and Table S6), predominated by FAs, CARs, PCs, and AAs and derivatives, in which most CARs, FAs, and LPCs showed negative associations, but most PCs showed positive associations.

Combining MITM and mediation analysis, 35 metabolites were found to significantly mediate the association between ALAN and neonatal WAZ/WLZ, with mediation proportions ranging from 5 to 12% in absolute value, within CARs and AAs and derivatives were the most, followed by NEAs, amides, and a few FAs, LPCs, and other metabolites, such as PC and androsterone sulfate (Figure A and Table S7). The top 10 enriched pathways based on these mediating metabolites are shown in Figure B and Table S8, involving neurotransmitter release cycle, metabolism of proteins, transmission across chemical synapses, neuronal system, SLC transporter disorders, transcription/translation, amino acid transport defects, disorders of transmembrane transporters, transport of inorganic cations/anions and amino acids/oligopeptides, and G α (q) signaling events, all of which have FDR < 0.05.

2.

2

Open in a new tab

Summary of the MITM, mediation, and enrichment analysis results. (A) Mediation proportion of the metabolites mediating ALAN-fetal growth outcome associations. (B) Top ten functional enrichment pathways.

3.4. Gene-Environment Interactions on Early Life Growth

Figure A and Table S9 show the interactions between the entire-pregnancy ALAN and the maternal glucose metabolic PRS on growth outcomes. The effects of ALAN exposure on the risks of slow WAZ and rapid WLZ growth trajectories were significantly more pronounced in the high FG PRS group, with corresponding RRs (95% CI; P for interaction) of 1.29 (1.10, 1.53; P for interaction = 0.023) and 0.83 (0.71, 0.95; P for interaction = 0.033) for each SD increase in ALAN, respectively. Besides, significant interactions between ALAN and high HbA1c PRS on the risk of rapid WLZ growth trajectory were also observed (RR = 0.81, 95% CI: 0.70, 0.95; P for interaction = 0.025).

3.

3

Open in a new tab

Forest plots of the interactions of prenatal exposure to ALAN with maternal (A) glucose metabolism-related PRS and (B) lipid metabolism-related PRS on fetal and child growth outcomes. WAZ, weight-for-age z score; WLZ, weight-for-length z score; T2D, type 2 diabetes; FG, fasting glucose; OGTT2h, 2-h oral glucose tolerance test; FI, fast insulin; TC, total cholesterol; TG, triglyceride; HDL, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol; NHDL, nonhigh-density lipoprotein cholesterol; PRS, polygenetic risk score. All models were adjusted for maternal age, preconceptional BMI, weight gain during pregnancy, parity, education levels, self-reported financial status, the season and year of conception, neonatal sex, the averages of PM2.5, temperature, and relative humidity during the entire pregnancy, and the top 10 genetic principal components.

In the interaction analysis on ALAN and the lipid-metabolic PRS (Figure B and Table S10), the associations of ALAN with neonatal WAZ and WLZ were consistently more pronounced in the high TG PRS group, with corresponding β (95% CI; P for interaction) of −0.10 (−0.16, −0.04; P for interaction = 0.027) and −0.13 (−0.21, −0.05; P for interaction = 0.087), respectively, for each SD increase in exposure in the subgroup. No significant interaction between ALAN and other glucose- and lipid metabolism-related PRS was observed.

3.5. Sensitivity Analysis

After further adjustment for METs in the exposure-outcome regression models, the associations of the entire and early pregnancy ALAN exposures with growth outcomes remained stable (Table S11). When using preconceptional ALAN as the exposure indicator, 113 statistically significant metabolites were observed in ALAN MWAS (P < 0.05), 107 of which overlapped with metabolites associated with early pregnancy ALAN exposure (Figure S6A and Table S12). Thirty-three of 113 metabolites showed significant mediation effects in the association between preconceptional ALAN exposure and growth outcomes. Among these, 28 overlapped with those using early pregnancy ALAN as the exposure indicator (Figure S6B and Table S13), leaving five unique mediating metabolites: phenylalanine, triphenylphosphine oxide, Amide 22:0, L-Theanine, sphingosine-1-phosphate, PC 32:1, and LPC 16:1 (Figure S6C). Functional enrichment analysis based on the 33 mediating metabolites also reproduced six of the top ten previously identified pathways (Figure S6D and Table S14).

4. Discussion

Based on findings from our previous exposome study that prenatal exposure to ALAN was associated with reduced weight and adiposity gains during the first 1000 days, we further integrated the maternal metabolome and genome to investigate potential mechanisms and the effect of modification by maternal genetic predisposition in glucose and lipid metabolism. We identified 35 metabolites that mediate the associations between ALAN and neonatal WAZ and WLZ, including CARs, AAs, and NEAs. These metabolites were mainly enriched in neural signal transmission-related pathways, suggesting that prenatal ALAN exposure may affect early life growth by interfering with maternal neural signal transmission. We also found that ALAN interacted with maternal glucose- and lipid-metabolic genetic predisposition in early life growth, as evidenced by more pronounced associations of ALAN in high PRS groups for FG, HbA1c, and TG.

To date, evidence on the association between prenatal ALAN exposure and early life growth is scarce and inconclusive, with most studies focusing on birth outcomes. Findings of this study indicate that exposure to ALAN during pregnancy is associated not only with restricted fetal growth but also retarded postnatal weight and body fat growth, which is partly supported by a recent study that reported 0.42% (95%: 0.36, 0.46) increases in the low birth weight rate with an interquartile range increase in ALAN. By contrast, nonsignificant associations between ALAN exposure and birth weight were observed in another birth cohort study from China. The discrepancies among studies may be due to several factors, including heterogeneity in study populations, ALAN exposure levels, and the confounders included in the analyses, which calls for more related studies to validate.

Our findings of this study are biologically plausible. It has been established that ALAN exposure interferes with circadian rhythms and accompanying melatonin secretion. Experimental data overwhelmingly suggest that melatonin is an effective radical scavenger and indirect antioxidant against oxidative stress in the placenta, fetus, and mother, thereby alleviating placental dysfunction and ensuring proper fetal growth. In addition, circadian rhythms have been shown to play vital roles in maintaining hormone, glucose, and lipid metabolism homeostasis; − dysregulation of these rhythms could lead to adverse growth outcomes. As a result, we hypothesize that prenatal exposure to ALAN may inhibit fetal growth by exacerbating oxidative stress and inflammation at the mother-fetal interface and inducing hormonal and metabolic dysfunction via disruption of the maternal circadian rhythm. These changes further lead to lasting physiological alterations in the offspring, thereby retarding postnatal longitudinal growth.

To a certain extent, our metabolomic analysis supports the hypothesis. As shown in Figure B, the enrichment analysis provided initial clues that interference with maternal neural signal transmission might be the biological linkage between ALAN exposure and early life growth. Interestingly, we noticed that NEAs, including PEA and SEA, were the main metabolites enriched in the related pathways. NEAs are a family of ubiquitous bioactive lipid molecules consisting of a fatty acid linked to an ethanolamine by an amide bond, which have recently been shown to interact with the circadian rhythm. Specifically, animal studies have reported diurnal fluctuations in endogenous NEAs in the brain and gastrointestinal tract. − Moreover, this group of lipid molecules can modulate the transcription of BMAL1 and REV-ERBα, two core clock genes, by activating PPARα, their main receptor, thereby regulating circadian rhythms. PPARα is also the direct target gene of the heterodimer CLOCK–BMAL1, a key molecular clock component, which drives rhythms in clock-controlled genes. , Therefore, by combining metabolites enriched in these pathways with existing evidence, our metabolomic analysis further suggests that prenatal ALAN exposure may affect early life growth by influencing maternal circadian rhythms.

In addition, ALAN was found to be negatively associated with cortisol, estrone sulfate (the reserve form of estrogen), and pregnanediol 3-O-glucuronide (the final product of progesterone metabolism) and positively associated with androsterone sulfate (sulfated dehydroepiandrosterone, an androgen precursor), of which androsterone sulfate showed a positive mediation effect on the ALAN-neonatal WAZ association, suggesting ALAN exposure might affect offspring’s growth via interference with steroid hormones. Under normal physiological conditions, cortisol is synthesized from cholesterol under the catalysis of hydroxysteroid dehydrogenases (HSDs) and cytochrome P450s (CYPs), with progesterone as an intermediate. Simultaneously, dehydroepiandrosterone can be translated to estrogen by HSDs and CYPs. Both progesterone and estrogen play critical roles in the placental development and functional maturity, decreases in which have been associated with increased risks of multiple adverse birth outcomes. Of note, the transcriptional activity of CYPs and HSDs is rhythmic and has been increasingly shown to be regulated by core clock genes. − For example, knocking out BMAL1 in rats decreased the activity of CYPs and HSDs, further inhibiting the secretion of progesterone and estrogen. Put together, we speculate that the observed associations among ALAN, steroid hormones, and neonatal WAZ are the downstream of the disturbance of circadian rhythm; that is, the disrupted circadian rhythm by ALAN downregulates the expression of CYPs and HSDs, which further reduces the synthesis of progesterone, as well as the transformation from dehydroepiandrosterone to estrogen, ultimately adversely affecting early life growth.

Based on maternal glucose- and lipid-metabolic PRS, our gene-environment interaction analysis showed that ALAN exposure had more pronounced effects on early life growth among mothers with high genetic susceptibility to FG, HbA1c, and TG. Hyperglycemic status during pregnancy has been proven to lead to less fetal insulin secretion or insulin resistance, which could further exert lasting effects on postnatal growth. Additionally, higher levels of gestational TG have been shown to increase oxidized LDL (oxLDL) and dense LDL particles, which can impair placental nutrient and oxygen exchange by initiating endothelial dysfunction and aggravating oxidative stress, thereby impeding intrauterine growth. On the one hand, some previous epidemiological and experimental studies have shown that ALAN exposure was associated with impaired glucose metabolism (e.g., elevated fasting glucose, HbA1c, and HOMA-IR) and increased risks of T2D, ,, which is believed to be achieved by reducing melatonin levels and affecting the hypothalamic-pituitary-adrenal (HPA) axis via disturbing circadian rhythm. On the other hand, a lipogenesis effect of ALAN, involving the upregulation of core clock gene-dependent enzymes (e.g., FAS, ACC, and SREBP1), has been demonstrated in animal models. , As a result, the observed synergistic effects of ALAN with high PRS for FG, HbA1c, and TG may be explained by these factors acting synergistically to upregulate maternal glucose and lipid metabolism during pregnancy.

To the best of our knowledge, this epidemiological study is the first to explore potential biological mechanisms linking prenatal ALAN exposure to early life growth and its interactions with maternal genetic predispositions. The combination of multiple rather than single neonatal anthropometric measurements and corresponding growth trajectories enabled a comprehensive evaluation of the growth status and exploration of the lasting effects of ALAN exposure. However, this study is also subject to several limitations. First, the assessment of ALAN exposure levels relied on remote sensing data and did not account for curtain usage, indoor light exposure, or outdoor time. Therefore, exposure misclassification is inevitable. However, such bias has been suggested to be nondifferential across the study population and to underestimate the effect sizes. Second, in this observational exploratory study, we neither performed multiple-comparison adjustment in MWAS to include as many metabolites as possible for the subsequent enrichment analysis nor set a strict cutoff to define statistically significant gene-environment interactions, given that such an analysis has been well acknowledged to be underpowered. As a result, the results may be subject to a higher incidence of type I errors, which needs to be interpreted with caution and validated in large-scale future studies. Third, although we applied a sensitive UPLC-MS method in metabolite detection, it cannot fully cover all of the metabolites. Simultaneously, because untargeted metabolome profiling was performed only on early pregnancy serum samples, we were unable to assess potential metabolic changes in later pregnancy in relation to ALAN exposure, which may have led to missed important biomechanical clues and should be addressed in future studies with the appropriate data. The median imputation approach used to handle metabolite missingness may have also influenced the precision of the MWAS results. Nevertheless, the method has demonstrated consistently low normalized root mean squared error (NRMSE) across sample types when metabolomics data exhibit low missingness rates. Thus, we do not think that the imputation process would have substantially biased our findings. Furthermore, the PRS used in this study were derived from GWAS summary data of the general population rather than pregnant women, primarily due to data availability. Although there is evidence of strong genetic correlations between gestational and nonpregnant metabolic phenotypes, , and such an analysis strategy has been applied in previous studies , and validated in ours (Figure S7), bias remains possible, as these PRS did not account for potential distinct genetic and physiological mechanisms regulating glucose and lipid metabolism in pregnancy. Therefore, more gene-environment studies are encouraged, particularly when GWAS summary data from pregnant women become available. Lastly, although we attempted to adjust for covariates as much as possible throughout the analyses, residual confounding from unmeasured factors (e.g., maternal nutritional status) is possible and should be taken into account in future validations.

In summary, our study suggested an association between prenatal exposure to ALAN and slower gains in weight and adiposity during the first 1000 days of life. These observed associations may be partly explained by disruptions to maternal circadian rhythms and appeared more pronounced among mothers with higher genetic predispositions to fasting glucose, HbA1c, and triglycerides. In the context of rapid urbanization and excessive use of ALAN today, these findings highlight a potential public health concern that warrants further verification and investigation using large multicenter cohorts and experimental studies that incorporate multiomics. If corroborated, they could support public health strategies, such as mindful urban lighting planning, to support healthy fetal and child growth, which is the basis of lifelong health, and clinical attention might be particularly warranted for populations at high genetic susceptibility.

Supplementary Material

es5c14467_si_001.pdf (1MB, pdf)
es5c14467_si_002.xlsx (86.3KB, xlsx)

Acknowledgments

Funding: This work was supported in part by the National Key R&D Program of China (2023YFC39005203) and the National Natural Science Foundation of China (82574101, 82473657).

The data used in this study are available from the corresponding authors upon request via email.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c14467.

  • Additional study details, materials, methods, and results, including baseline characteristics of the study population, the growth trajectories of weight-for-age and weight-for-length z-scores, (PDF)

  • The number of knots of the best-fitted restricted cubic spline models, the source of GWAS summary data used to construct polygenetic risk scores, and sensitivity analysis results (XLSX)

All authors read and approved the final manuscript. J.Z. and J.X. have full access to all study data and are responsible for the integrity and accuracy of the data analysis. Concept and design: W.J. and J.Z. Acquisition, analysis, or interpretation of data: W.J., A.H., and Y.H. Drafting of the manuscript: W.J. Critical revision of the manuscript for important intellectual content: W.J., A.H., Y.H., Z.Y., J.X., and J.Z. Supervision: J.X. and J.Z.

The authors declare no competing financial interest.

References

  1. Indrio F., Dargenio V. N., Marchese F., Giardino I., Vural M., Carrasco-Sanz A., Pietrobelli A., Pettoello-Mantovani M.. The Importance of Strengthening Mother and Child Health Services during the First 1000 Days of Life: The Foundation of Optimum Health, Growth and Development. J. Pediatr. 2022;245:254–256 e250. doi: 10.1016/j.jpeds.2022.03.001. [DOI] [PubMed] [Google Scholar]
  2. Hoffman D. J., Powell T. L., Barrett E. S., Hardy D. B.. Developmental origins of metabolic diseases. Physiol Rev. 2021;101(3):739–795. doi: 10.1152/physrev.00002.2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Kyba C. C. M., Kuester T., Sánchez de Miguel A., Baugh K., Jechow A., Hölker F., Bennie J., Elvidge C. D., Gaston K. J., Guanter L.. Artificially lit surface of Earth at night increasing in radiance and extent. Sci. Adv. 2017;3(11):e1701528. doi: 10.1126/sciadv.1701528. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Costello H. M., Gumz M. L.. Circadian Rhythm, Clock Genes, and Hypertension: Recent Advances in Hypertension. Hypertension. 2021;78(5):1185–1196. doi: 10.1161/HYPERTENSIONAHA.121.14519. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Wang T., Kaida N., Kaida K.. Effects of outdoor artificial light at night on human health and behavior: A literature review. Environ. Pollut. 2023;323:121321. doi: 10.1016/j.envpol.2023.121321. [DOI] [PubMed] [Google Scholar]
  6. Bożejko M., Tarski I., Małodobra-Mazur M.. Outdoor artificial light at night and human health: A review of epidemiological studies. Environ. Res. 2023;218:115049. doi: 10.1016/j.envres.2022.115049. [DOI] [PubMed] [Google Scholar]
  7. Boakye K., Iyanda A., Asante E., Oppong J.. Association between outdoor artificial light at night and low birth weight in California. Int. J. Environ. Health Res. 2025;35:1883–1893. doi: 10.1080/09603123.2024.2412137. [DOI] [PubMed] [Google Scholar]
  8. Zhang L., Yin W., Yu W., Wang P., Wang H., Zhang X., Zhu P.. Environmental exposure to outdoor artificial light at night during pregnancy and fetal size: A prospective cohort study. Sci. Total Environ. 2023;883:163521. doi: 10.1016/j.scitotenv.2023.163521. [DOI] [PubMed] [Google Scholar]
  9. Sicard M., Nusinovici S., Hanf M., Muller J. B., Guellec I., Ancel P. Y., Gascoin G., Rozé J. C., Flamant C.. Fetal and Postnatal Head Circumference Growth: Synergetic Factors for Neurodevelopmental Outcome at 2 Years of Age for Preterm Infants. Neonatology. 2017;112(2):122–129. doi: 10.1159/000464272. [DOI] [PubMed] [Google Scholar]
  10. Zhang T., Xu J., Li S., Bazzano L. A., He J., Whelton P. K., Chen W.. Trajectories of childhood BMI and adult diabetes: the Bogalusa Heart Study. Diabetologia. 2019;62(1):70–77. doi: 10.1007/s00125-018-4753-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Juliusdottir T., Steinthorsdottir V., Stefansdottir L., Sveinbjornsson G., Ivarsdottir E. V., Thorolfsdottir R. B., Sigurdsson J. K., Tragante V., Hjorleifsson K. E., Helgadottir A.. et al. Distinction between the effects of parental and fetal genomes on fetal growth. Nat. Genet. 2021;53(8):1135–1142. doi: 10.1038/s41588-021-00896-x. [DOI] [PubMed] [Google Scholar]
  12. Rao F., Xue T.. Circadian-independent light regulation of mammalian metabolism. Nat. Metab. 2024;6(6):1000–1007. doi: 10.1038/s42255-024-01051-6. [DOI] [PubMed] [Google Scholar]
  13. Zheng R., Xin Z., Li M., Wang T., Xu M., Lu J., Dai M., Zhang D., Chen Y., Wang S.. et al. Outdoor light at night in relation to glucose homoeostasis and diabetes in Chinese adults: a national and cross-sectional study of 98,658 participants from 162 study sites. Diabetologia. 2023;66(2):336–345. doi: 10.1007/s00125-022-05819-x. [DOI] [PubMed] [Google Scholar]
  14. Han B., Deng J., Huang X., Yin X., Liang Y.. Effects of exposure to outdoor light at night on blood lipids in China: a nationwide cross-sectional study. Sci. Rep. 2024;14(1):28263. doi: 10.1038/s41598-024-79484-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Jiang W., He X., Han A., Yin S., Tang W., Jin H., Zeng X., Gan Y., Wang X., Huang Y.. et al. Incorporating maternal metabolome and genome to understand the effects of prenatal phenol exposure on the first 1000-day growth in an exposome-based study. Environ. Int. 2025;204:109816. doi: 10.1016/j.envint.2025.109816. [DOI] [PubMed] [Google Scholar]
  16. Betts, K. ; Sawyer, K. . Use of Metabolomics to Advance Research on Environmental Exposures and the Human Exposome. 2016. 10.17226/23414. [DOI] [PubMed]
  17. Zhang J., Tian Y., Wang W., Ouyang F., Xu J., Yu X., Luo Z., Jiang F., Huang H., Shen X.. et al. Cohort profile: the Shanghai Birth Cohort. Int. J. Epidemiol. 2019;48(1):21–21g. doi: 10.1093/ije/dyy277. [DOI] [PubMed] [Google Scholar]
  18. Elvidge C. D., Baugh K., Zhizhin M., Hsu F.-C., Tilottama G.. VIIRS night-time lights. Int. J. Remote Sens. 2017;38:1–20. doi: 10.1080/01431161.2017.1342050. [DOI] [Google Scholar]
  19. Shi S., Wang W., Li X., Hang Y., Lei J., Kan H., Meng X.. Optimizing modeling windows to better capture the long-term variation of PM2.5 concentrations in China during 2005–2019. Sci. Total Environ. 2023;854:158624. doi: 10.1016/j.scitotenv.2022.158624. [DOI] [PubMed] [Google Scholar]
  20. de Onis M., de Onis M.. WHO Child Growth Standards based on length/height, weight and age. Acta Paediatr. 2006;95(S450):76–85. doi: 10.1111/j.1651-2227.2006.tb02378.x. [DOI] [PubMed] [Google Scholar]
  21. Jiang, W. ; Huang, Y. ; Jin, H. ; Gan, Y. ; Zhang, Q. ; He, X. ; Tian, Y. ; Zhang, J. ; The Shanghai Birth, C. . A prospective exposome-based gene-environment interaction study on the effects of prenatal environmental exposure on fetal growth in the Shanghai Birth Cohort Environ. Health Perspect. 2025. 10.1289/ehp15902. [DOI] [PubMed]
  22. White I. R., Royston P., Wood A. M.. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med. 2011;30(4):377–399. doi: 10.1002/sim.4067. [DOI] [PubMed] [Google Scholar]
  23. Kim Y., Park I., Kang M.. Convergent validity of the international physical activity questionnaire (IPAQ): meta-analysis. Public Health Nutr. 2013;16(3):440–452. doi: 10.1017/S1368980012002996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Reiter R. J., Tan D. X., Korkmaz A., Rosales-Corral S. A.. Melatonin and stable circadian rhythms optimize maternal, placental and fetal physiology. Hum. Reprod. Update. 2014;20(2):293–307. doi: 10.1093/humupd/dmt054. [DOI] [PubMed] [Google Scholar]
  25. Ishida A., Mutoh T., Ueyama T., Bando H., Masubuchi S., Nakahara D., Tsujimoto G., Okamura H.. Light activates the adrenal gland: timing of gene expression and glucocorticoid release. Cell Metab. 2005;2(5):297–307. doi: 10.1016/j.cmet.2005.09.009. [DOI] [PubMed] [Google Scholar]
  26. Park Y. M., White A. J., Jackson C. L., Weinberg C. R., Sandler D. P.. Association of Exposure to Artificial Light at Night While Sleeping With Risk of Obesity in Women. JAMA Intern Med. 2019;179(8):1061–1071. doi: 10.1001/jamainternmed.2019.0571. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Mason I. C., Grimaldi D., Reid K. J., Warlick C. D., Malkani R. G., Abbott S. M., Zee P. C.. Light exposure during sleep impairs cardiometabolic function. Proc. Natl. Acad. Sci. U. S. A. 2022;119(12):e2113290119. doi: 10.1073/pnas.2113290119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Gómez-Boronat M., Isorna E., Armirotti A., Delgado M. J., Piomelli D., de Pedro N.. Diurnal Profiles of N-Acylethanolamines in Goldfish Brain and Gastrointestinal Tract: Possible Role of Feeding. Front. Neurosci. 2019;13:450. doi: 10.3389/fnins.2019.00450. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Liedhegner E. S., Sasman A., Hillard C. J.. Brain region-specific changes in N-acylethanolamine contents with time of day. J. Neurochem. 2014;128(4):491–506. doi: 10.1111/jnc.12495. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Murillo-Rodriguez E., Desarnaud F., Prospero-Garcia O.. Diurnal variation of arachidonoylethanolamine, palmitoylethanolamide and oleoylethanolamide in the brain of the rat. Life Sci. 2006;79(1):30–37. doi: 10.1016/j.lfs.2005.12.028. [DOI] [PubMed] [Google Scholar]
  31. Charoensuksai P., Xu W.. PPARs in Rhythmic Metabolic Regulation and Implications in Health and Disease. PPAR Res. 2010;2010:243643. doi: 10.1155/2010/243643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Chen L., Yang G.. PPARs Integrate the Mammalian Clock and Energy Metabolism. PPAR Res. 2014;2014:653017. doi: 10.1155/2014/653017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Miller W. L., Auchus R. J.. The molecular biology, biochemistry, and physiology of human steroidogenesis and its disorders. Endocr. Rev. 2011;32(1):81–151. doi: 10.1210/er.2010-0013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Vidal M. S. Jr., Lintao R. C. V., Severino M. E. L., Tantengco O. A. G., Menon R.. Spontaneous preterm birth: Involvement of multiple feto-maternal tissues and organ systems, differing mechanisms, and pathways. Front. Endocrinol. (Lausanne) 2022;13:1015622. doi: 10.3389/fendo.2022.1015622. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Chen H., Zhao L., Kumazawa M., Yamauchi N., Shigeyoshi Y., Hashimoto S., Hattori M. A.. Downregulation of core clock gene Bmal1 attenuates expression of progesterone and prostaglandin biosynthesis-related genes in rat luteinizing granulosa cells. Am. J. Physiol.: Cell Physiol. 2013;304(12):C1131–C1140. doi: 10.1152/ajpcell.00008.2013. [DOI] [PubMed] [Google Scholar]
  36. Shimizu T., Hirai Y., Murayama C., Miyamoto A., Miyazaki H., Miyazaki K.. Circadian Clock genes Per2 and clock regulate steroid production, cell proliferation, and luteinizing hormone receptor transcription in ovarian granulosa cells. Biochem. Biophys. Res. Commun. 2011;412(1):132–135. doi: 10.1016/j.bbrc.2011.07.058. [DOI] [PubMed] [Google Scholar]
  37. Shao S., Zhao H., Lu Z., Lei X., Zhang Y.. Circadian Rhythms Within the Female HPG Axis: From Physiology to Etiology. Endocrinology. 2021;162(8):bqab117. doi: 10.1210/endocr/bqab117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Hughes A. E., Hattersley A. T., Flanagan S. E., Freathy R. M.. Two decades since the fetal insulin hypothesis: what have we learned from genetics? Diabetologia. 2021;64(4):717–726. doi: 10.1007/s00125-021-05386-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Echeverria C., Eltit F., Santibanez J. F., Gatica S., Cabello-Verrugio C., Simon F.. Endothelial dysfunction in pregnancy metabolic disorders. Biochim. Biophys. Acta, Mol. Basis Dis. 2020;1866(2):165414. doi: 10.1016/j.bbadis.2019.02.009. [DOI] [PubMed] [Google Scholar]
  40. Zhang L., Wang H., Zu P., Li X., Ma S., Zhu Y., Xie T., Tao F., Zhu D. M., Zhu P.. Association between exposure to outdoor artificial light at night during pregnancy and glucose homeostasis: A prospective cohort study. Environ. Res. 2024;247:118178. doi: 10.1016/j.envres.2024.118178. [DOI] [PubMed] [Google Scholar]
  41. Borck P. C., Rickli S., Vettorazzi J. F., Batista T. M., Boschero A. C., Vieira E., Carneiro E. M.. Effect of nighttime light exposure on glucose metabolism in protein-restricted mice. J. Endocrinol. 2022;252(2):143–154. doi: 10.1530/JOE-21-0133. [DOI] [PubMed] [Google Scholar]
  42. Mezhnina V., Ebeigbe O. P., Velingkaar N., Poe A., Sandlers Y., Kondratov R. V.. Circadian clock controls rhythms in ketogenesis by interfering with PPARalpha transcriptional network. Proc. Natl. Acad. Sci. U. S. A. 2022;119(40):e2205755119. doi: 10.1073/pnas.2205755119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Luo Y., Yang J., Kang J., Chen K., Jin X., Gonzalez F. J., Liu A.. PPARalpha mediates night neon light-induced weight gain: role of lipid homeostasis. Theranostics. 2020;10(25):11497–11506. doi: 10.7150/thno.50953. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Di Guida R., Engel J., Allwood J. W., Weber R. J. M., Jones M. R., Sommer U., Viant M. R., Dunn W. B.. Non-targeted UHPLC-MS metabolomic data processing methods: a comparative investigation of normalisation, missing value imputation, transformation and scaling. Metabolomics. 2016;12(5):93. doi: 10.1007/s11306-016-1030-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Elliott A., Walters R. K., Pirinen M., Kurki M., Junna N., Goldstein J. I., Reeve M. P., Siirtola H., Lemmelä S. M., Turley P.. et al. Distinct and shared genetic architectures of gestational diabetes mellitus and type 2 diabetes. Nat. Genet. 2024;56(3):377–382. doi: 10.1038/s41588-023-01607-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Gu Y., Zheng H., Wang P., Liu Y., Guo X., Wei Y., Yang Z., Cheng S., Chen Y., Hu L.. et al. Genetic architecture and risk prediction of gestational diabetes mellitus in Chinese pregnancies. Nat. Commun. 2025;16(1):4178. doi: 10.1038/s41467-025-59442-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Tekola-Ayele F., Chen J., Bacelis J., Sole-Navais P., Srivastava A., Juodakis J., Rouse A., Hallman M., Teramo K., Melbye M.. et al. Dissecting maternal and fetal genetic effects underlying the associations between maternal phenotypes, birth outcomes, and adult phenotypes: A mendelian-randomization and haplotype-based genetic score analysis in 10,734 mother–infant pairs. PLoS Med. 2020;17(8):e1003305. doi: 10.1371/journal.pmed.1003305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Tyrrell J., Richmond R. C., Palmer T. M., Feenstra B., Rangarajan J., Metrustry S., Cavadino A., Paternoster L., Armstrong L. L., De Silva N. M.. et al. Genetic Evidence for Causal Relationships Between Maternal Obesity-Related Traits and Birth Weight. JAMA. 2016;315(11):1129–1140. doi: 10.1001/jama.2016.1975. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es5c14467_si_001.pdf (1MB, pdf)
es5c14467_si_002.xlsx (86.3KB, xlsx)

Data Availability Statement

The data used in this study are available from the corresponding authors upon request via email.

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES