Prenatal antifungal exposure disrupts fetal steroidogenesis and is associated with persistent effects in children

Abstract

Background:

Exposure to endocrine-disrupting chemicals (EDCs) during fetal life is associated with reproductive disorders in humans. While research has largely focused on environmental EDCs and developmental exposure, some pharmaceuticals, including commonly used antifungal agents, also possess endocrine-disrupting properties. We investigate whether prenatal exposure to over-the-counter antifungal drugs is associated with altered fetal and postnatal steroidogenesis and reproductive development.

Methods:

We combined epidemiological, experimental, and translational approaches using data from a prospective mother-child cohort (n = 589), ex vivo human fetal adrenal cultures, and in vivo rat models. Maternal antifungal use during pregnancy was assessed prospectively, and associations with fetal growth, anogenital distance (AGD), penile measurements, and circulating steroid hormone levels during infancy were evaluated. Experimental studies examined the effects of selected antifungal agents on steroidogenesis in cultured human fetal adrenals and gestationally exposed Sprague-Dawley rats.

Results:

Early-pregnancy antifungal exposure is associated with reduced circulating steroid hormone levels and indicators of diminished androgen action, including decreased penile width at infancy. In contrast, exposure during mid-to-late pregnancy is associated with elevated steroid hormone levels and longer AGD in both female fetuses and male infants. Consistent with these observations, the antifungal agent clotrimazole reduces progesterone levels in human fetal adrenal cultures and in gestationally exposed rats, whereas miconazole increases AGD in rats.

Conclusions:

These translational findings suggest that commonly used antifungal agents alter fetal adrenal steroidogenesis and influence reproductive development, with effects persisting into postnatal life. Given their widespread use and over-the-counter availability, the safety of topical azole use during pregnancy may warrant re-evaluation.

Plain language summary

Prenatal exposure to some chemicals can affect reproductive development before birth. While most research has focused on environmental chemicals, certain medicines may also influence hormone production. We investigated whether antifungal medications used during pregnancy affect fetal hormone levels and reproductive development. We combined data from a mother-child cohort with laboratory studies using human fetal tissue and animal models. We examined associations between antifungal exposure, hormone levels, anogenital distance (AGD), and genital development in infants. Early-pregnancy exposure was associated with lower steroid hormone levels and reduced androgen-related development in boys, while later exposure was associated with higher hormone levels and longer AGD. Experimental studies supported these findings. These results suggest that commonly used antifungal medications may influence fetal development and warrant further safety evaluation during pregnancy.

Mola, Draskau, et al. investigate whether prenatal exposure to commonly used antifungal medications affects fetal hormone production and reproductive development. Antifungal exposure is associated with altered steroid hormone levels and changes in reproductive development in both humans and experimental models.

Introduction

Exposure to endocrine-disrupting chemicals (EDCs) during critical periods of development is linked to various reproductive disorders, including genital malformations, hormone-related cancers, and subfertility^1,2. Fetal life – particularly the late first trimester and the transition into the second trimester - represents a critical window of susceptibility, during which exogenous chemicals can disrupt sexual development by interfering with adrenal and gonadal differentiation or steroidogenic function³.

Over-the-counter medicines, such as antifungal agents, provide easy and convenient first-line treatment for minor and common health problems, eliminating the need for a prescription and reducing the burden on healthcare systems⁴. However, some over-the-counter medications harbour endocrine-disrupting properties, posing risks of adversely affecting fetal development if used during pregnancy. Medication use during pregnancy is common, and although many treatments are considered safe and clinically necessary, prenatal exposure to certain pharmaceuticals has been associated with adverse developmental outcomes more generally, highlighting the need to evaluate potential fetal effects of widely used medication^5–7. In this context, the widespread use of antifungal agents as well as mild analgesics has raised concerns^8–10. Topical azoles are first-line treatment for vulvovaginal candidiasis, which is highly prevalent among pregnant women¹¹. The common mode of action for these agents is inhibition of the cytochrome P450 family 51 (CYP51) of the fungal ergosterol synthesis pathway, effectively compromising fungal cell membrane integrity. However, many azoles also interfere with other CYP enzymes involved in steroid hormone biosynthesis in both animals and humans^12–14.

Human adrenal and gonadal development is initiated early in the first trimester of pregnancy¹⁵. Early testis development is characterized by the differentiation of Sertoli cells (from approximately gestational week (GW) 7), followed shortly thereafter by fetal Leydig cells, after which the testes begin producing testosterone required for masculinization of the male fetus. A masculinization programming window (MPW, GW 8-14) has been identified as a critical period during which human male reproductive organs develop in response to androgen signaling¹⁶. From GW 15 onward, the fetal zone in the adrenal cortex undergoes rapid proliferation, driving adrenal glandular enlargement. In parallel, the expression of key steroidogenic enzymes such as CYP11A1, CYP17A1, CYP21A2, and CYP11B1/2 is markedly increased¹⁷.

In rats, the MPW occurs later in gestation relative to humans, around gestational day (GD) 16–20 (with parturition normally at GD23), with peak androgen surge around GD 19¹⁶. Notably, rat fetal adrenals produce markedly lower levels of androgens compared to humans because expression of key enzymes such as CYP17A1 is absent or very limited¹⁸.

Although topical antifungal agents are generally considered safe due to limited systemic absorption and rapid metabolism¹⁹, several findings have prompted re-evaluation. Endocrine-disrupting effects have been observed in fetal rats at internal systemic concentrations relevant to human use²⁰. In vitro studies suggest that clotrimazole may disrupt human steroidogenesis at lower concentrations than those measured in serum after vaginal application¹². A recent human cohort study reported associations between prenatal exposure and altered steroid hormone concentrations, as well as altered anogenital distance (AGD) in children, suggesting potential effects on androgen-dependent masculinization of male fetuses²¹. These observations raise concern that even azoles with low systemic availability may alter fetal steroidogenesis during critical developmental windows.

In the present study, we investigate whether prenatal exposure is associated with altered steroid hormone profiles and growth in children at 3 months of age. To address causality, we combine human cohort analysis with complementary experimental models, including in vivo rat studies and an ex vivo study of human fetal adrenal tissue. We find that early-pregnancy exposure is associated with reduced circulating steroid hormone levels and indicators of diminished androgen action, whereas exposure during mid-to-late pregnancy is associated with elevated hormone levels and increased AGD. These findings are supported by experimental models demonstrating that selected antifungal agents directly disrupt fetal steroidogenesis.

Methods

Human Cohort

The Copenhagen Analgesic Study (COPANA) (ClinicalTrials.gov ID: NCT04369222, registered 1^st March 2020) is a single-centre, prospective, observational pregnancy and birth cohort from Copenhagen University Hospital - Rigshospitalet, Denmark (2020–2022). As part of routine clinical practice, all pregnant women completed a clinical web-based questionnaire at approximately GW 10 when booking their first-trimester ultrasound scan. This information was used for antenatal counseling and treatment. The questionnaire captured current medical use. For the latter, participants were specifically asked: “Do you currently take any medication (including painkillers)? If yes, please specify which medication and frequency”. This provided information on current medication use. A midwife identified eligible women in compliance with the inclusion criteria (Caucasian descent, no endocrine disorders, BMI between 18–35 kg/m2) and invited them and their partners to participate in the study. We excluded participants with multiple pregnancies, gestational diabetes, thyroid dysfunction, pre- or post-term deliveries (before GW 37 or beyond GW 42), or serious infant illness. In total, 685 couples consented (20% of 3425 eligible couples). Enrolled women subsequently completed electronic questionnaires every two weeks, reporting all medication use, including antifungals and mild analgesics; mild analgesic brand names available in Denmark were listed. Obstetrical outcomes were retrieved from the mothers’ medical records.

Third-trimester ultrasound scan

A study-specific ultrasound scan was performed in the third trimester (GW 29–34), including fetal biometry (head and abdominal circumference, femur length) used to estimate fetal weight via Hadlock’s equation²², and measurement of anogenital distance (AGD). AGD was measured in the axial plane from the centre of the anus to the posterior fourchette in females and to the posterior base of the scrotum in males. In total, 657 women attended the scan.

Child examination

Children were examined at approximately three months postpartum, coinciding with minipuberty; a transient activation of the hypothalamic-pituitary-gonadal axis^23,24. A total of 589 families (302 girls, 287 boys) completed this follow-up. Blood samples were obtained from 518/589 infants (88%). During the visit (girls: mean (SD) age 3.5 ± 0.8 months; boys: 3.1 ± 0.4 months), infants underwent anthropometric assessment (weight, length), and AGD was measured following TIDES methodology with the infant supine and legs elevated in a frog-leg position²⁵.

Serum samples

Reproductive hormones were measured in serum from blood drawn from the antecubital vein (269 girls, 247 boys). Non-fasting human venous blood was collected before oral sucrose and topical anaesthetic (EMLA), clotted, centrifuged, and the serum was stored at −20 °C until analysis. Sixteen steroid metabolites were quantified: three progestins (progesterone (PROG), 17-hydroxypregnenolone (17-OHPreg), 17-hydroxyprogesterone (17-OHP)), three mineralocorticoids (11-deoxycorticosterone (DOC), corticosterone (CORT), aldosterone (ALDO)), three glucocorticoids (11-deoxycortisol (11-DOC), cortisol, cortisone), four androgens (dehydroepiandrosterone sulfate (DHEAS), androstenedione (Adione), testosterone (T), dihydrotestosterone (DHT)), and three estrogens (estrone (E1), estradiol (E2), estrone sulfate (E1-S)) using two different in-house isotope-dilution online TurboFlow LC–MS/MS methods^26,27. Calibrators and triplicate controls were included in all batches. Undetectable values (DHT, E2, E1-S in some samples) were assigned 0.5× LOD. Free T was calculated using the Vermeulen formula²⁷. Steroid hormone concentrations in human infant serum, fetal rat serum, and conditioned media from human fetal adrenal ex vivo cultures were measured in the same laboratory. Anti-Müllerian hormone (AMH), follicle-stimulating hormone (FSH), luteinizing hormone (LH), and inhibin B were measured by immunoassays. Additional details are provided in the cohort profile²⁸. Limited serum volumes occasionally prevented full hormone analysis.

Circulating steroid hormone concentrations were converted to age- and sex-specific standard deviation (SD) scores using reference curves generated with generalized additive models for location, scale, and shape (GAMLSS) implemented in the GAMLSS R package. The L (skewness), M (median), and S (coefficient of variation) curves were derived from a longitudinal cohort of healthy Danish infants (Copenhagen Minipuberty Study; ClinicalTrials.gov NCT01411527)²⁹. SD scores were calculated as: SD score = ((X/M)^L − 1)/(L × S), where X is the observed concentration and L ≠ 0. Scores were computed separately for boys and girls. SD scores were not generated for DHEA and E1-S (both sexes) or DHT (girls) owing to undetectable values.

Human data evaluation and statistical analyses

Children were stratified according to fetal exposure to antifungal medications, accounting for developmental changes in adrenal morphology and steroid enzyme activity. Exposure was classified as early fetal life (GW < 15; 32 girls, 13 boys), mid/late fetal life (GW ≥ 15; 24 girls, 32 boys), or unexposed controls (246 girls, 242 boys). The early exposure group includes children whose mothers may also have used antifungals later in pregnancy. Subgroup analyses by specific antifungal types were not performed due to limited statistical power.

Primary outcomes were predefined as circulating steroid hormone levels and AGD, measured prenatally and postnatally. Outcomes in each exposure group were compared with unexposed controls using Student’s t test, with additional sex-specific analyses. Associations between maternal antifungal use and fetal sex were evaluated with the Chi-squared test. Univariate linear regression assessed relationships between exposure windows and outcomes, with model assumptions checked using residual and Q–Q plots.

Multiple linear regression models were adjusted for maternal parity (nulliparous: yes/no), pre-pregnancy BMI, and paracetamol use during pregnancy. Sensitivity analyses tested robustness by sequentially excluding infants based on corticosteroid use or emergency cesarean delivery. Emergency caesarean section was tested a priori, as the mode of delivery has been suggested to affect neonatal steroid hormone levels. For steroid hormones below the limit of detection, associations with undetectable levels were assessed using Chi-squared or Fisher’s exact tests, as appropriate.

Analyses were based on available data, and participants with missing values for exposure, outcome, or covariates were excluded from the relevant analyses.

Statistical significance was defined as a two-sided p ≤ 0.05, with p values of 0.05–0.1 interpreted as trends. Analyses were performed using IBM SPSS v29.0.1.0³⁰ and RStudio v2024.09.1³¹.

Rat in vivo study

Twenty-four time-mated, nulliparous Sprague-Dawley rats (Crl:CD(SD); Charles River, Germany via Scanbur, Denmark) weighing 200–250 g were delivered on gestational day (GD) 3, with the day following overnight mating designated GD1. A sample size of eight litters was selected based on historical data from comparable in-house rat in vivo experiments and the expected variability of the endpoints. Two cohorts were divided into three groups of eight dams each on GD4, balanced by body weight. Dams were housed in pairs until GD17, then singly housed until GD21 in High Temperature polysulfone cages with wood chip bedding, nesting material, and a wooden shelter (Tapvei, Denmark). Standard conditions included a 12 h light/dark cycle, 21 °C ± 1, 55% ±5 humidity, and 50–60 air changes per h. Animals received ad libitum Altromin 1314 feed (soy- and alfalfa-free; Altromin GmbH, Germany) and tap water in Bisphenol A-free bottles.

Dams were exposed to clotrimazole (25 or 75 mg/kg bw/day) or miconazole (35 or 70 mg/kg bw/day, purity ≥98%; Merck, Supelco®, Germany) via daily oral gavage in corn oil from GD7 to GD21, following previously described protocols²⁰. Dams were acclimatized for four days prior to exposure. Dosing was initiated first at GD7 to avoid implantation loss, yet early enough to cover early fetal developmental stages and throughout the sensitive periods for reproductive development. Body weights were recorded daily, and animals were monitored twice daily for clinical signs. Doses were selected based on two main considerations: (i) avoiding significant maternal toxicity, guided by published literature, and (ii) achieving fetal exposure levels relevant to those observed in humans. For clotrimazole, dose selection is described in our previous publication²⁰, in which physiologically based kinetic (PBK) modeling predicted that the chosen doses would yield fetal plasma concentrations overlapping with those measured in women. Miconazole doses were selected using a similar rationale, guided by available literature and with the aim of achieving comparable exposure conditions without inducing maternal toxicity.

On GD17, selected dams were gavaged 1 h ± 15 min prior to decapitation under CO2/O2 anaesthesia. The dam and fetal adrenals were excised, fixed in 10% formalin and processed using an Excelsior AS Tissue Processor (Thermo Scientific™) before paraffin embedding.

On GD21, dams were gavaged 1 h ± 15 min before decapitation under CO2/O2 anaesthesia. Maternal trunk blood was collected in heparinized vials. Uteri were excised and weighed, and the number of resorptions, implantations, live fetuses, and fetal positions was recorded. Fetal body weights and anomalies were noted, and AGD was measured under a stereomicroscope by an experienced technician blinded to exposure. Fetal trunk blood was pooled by sex within each litter, kept on ice, centrifuged at 4000 rpm for 10 min at 4 °C, and plasma was stored at −80 °C.

Fetal adrenal glands were dissected, preserved in RNA later, and stored at −80 °C, or fixed in 10% formalin and processed on an Excelsior AS Tissue Processor (Thermo Scientific™) before paraffin embedding.

Steroid metabolites in rat plasma

Pooled fetal rat plasma was analysed for steroid metabolites by the above-mentioned LC-MS/MS methods (see under serum samples).

Histopathological assessment of adrenal glands from animals exposed to Miconazole

Mayer’s Hematoxylin and Eosin (H&E)-stained adrenal gland sections (5 µm, one section per animal) from control and high-dose animals were assessed for histopathological changes, including hypertrophy, vacuolation, and infiltration of inflammatory cells. Adrenals from GD17 dams and male fetuses, as well as GD21 dams and male fetuses, were assessed.

Statistical analyses of rat in vivo data

Data from the in vivo rat study, including steroid hormones, were assessed for normality and homogeneity of variance using residual statistics. Non-normally distributed data were log-transformed and reassessed. Normally distributed data were analyzed by one-way ANOVA with Dunnett’s post hoc test. Non-normally distributed data were analyzed by Kruskal-Wallis with Dunn’s multiple comparison test (GraphPad Prism 10).

AGD, AGD index (AGDi; AGD/cube root of body weight), and fetal body weights at GD21 were analyzed using a General Linear Model with Dunnett’s post hoc test (SAS Enterprise Guide 8.2). AGD and AGDi analyses included litter as a random, nested factor and body weight as a covariate. Fetal body weights were analyzed with litter as the statistical unit and number of offspring per litter as a covariate.

Rat fetal adrenal transcriptomics

Total RNA was extracted from GD21 fetal adrenal glands using the RNeasy Microkit (Qiagen) with on-column DNase I digestion. Eight samples per group (male controls, clotrimazole-exposed males, female controls) were included; female-exposed samples were not included because of an unforeseen error. RNA quality and quantity were assessed using the Agilent RNA 6000 Nano Kit on a Bioanalyzer 2100, with all samples exhibiting RIN ≥ 8.

Bulk RNA Barcoding and sequencing (BRB-seq) was performed as previously described^32,33. Briefly, 4 µL total RNA (2.5 ng/µL) underwent reverse transcription and template switching, followed by ds-cDNA synthesis and tagmentation using the Illumina Nextera XT Kit. Libraries were sequenced on a NovaSeq platform (IntegraGen, France), and adapter dimers were removed with DimerRemover. (https://sourceforge.net/projects/dimerremover/).

Preprocessing and quality control were performed by SciLicium (Rennes, France). Read 1 included a 6-bp sample barcode and 10-bp unique molecular identifier (UMI); reads with base quality <10 were excluded. Read 2 was aligned to the rat reference transcriptome (UCSC) using BWA v0.7.4.4 (−l 24), and multi-mapping reads were discarded. A gene count matrix was generated from unique UMIs and normalized using the rlog transformation in DESeq2³⁴. Raw and processed data are deposited at GEO [accession GSE317878] and TOXsIgN³⁵. The resulting transcriptomic signatures were also deposited at the TOXsIgN repository (https://toxsign.genouest.org/)³⁶.

Differential gene expression

Genes with at least three counts and rlog-transformed expression above background (median = 0.659) and ≥1.2-fold change were selected. Statistical significance was determined with LIMMA, using FDR-adjusted p ≤ 0.05³⁷. Comparisons included male vs female controls and male controls vs clotrimazole-exposed males. Differentially expressed genes were partitioned into four gene clusters based on supervised K-means clustering (n = 4). Functional enrichment of gene clusters was analyzed by Gene Ontology in the AMEN suite of tools³⁸, with FDR-adjusted p ≤ 0.05 considered significant.

Human DNA methylation analysis

DNA was purified from infant blood samples using the automated Maxwell 16 system (SEV AS1010, Promaga, Madison, WI, USA) and quantified by NanoDrop (Life Technologies Europe BV, Naerum, Denmark). DNA methylation was analyzed in a nested COPANA cohort including all available samples from children exposed to antifungal drugs in early fetal life (GW < 15; 9 boys, 22 girls) and randomly selected age- and sex-matched unexposed controls (46 boys, 198 girls).

Bisulfite-treated DNA was hybridized to Infinium MethylationEPIC v2.0 BeadChips (Illumina), scanned on the iScan system, and probe intensities extracted in GenomeStudio to generate IDAT files. IDATs were imported into RStudio (R “Darwin” v23.5.0) using minfi package (v1.50)³⁹. Data quality was assessed with ShinyMethyl (v1.40)⁴⁰ and normalized using Subset-quantile Within Array Normalization⁴¹. CpGs and single-nucleotide extension probes containing SNPs were removed. To adjust for batch effects and variation in blood cell composition, surrogate variables were estimated using SmartSVA (v0.1.3)⁴². Differentially methylated regions (DMRs) were identified with the DMRcate package⁴³, which ranks DMRs across the genome based on a tunable kernel smoothing method. Default settings with a bandwidth of 1000 nt (lambda = 1000) and a scaling factor of 2 (C = 2) were applied, and the results adjusted for surrogate variables and multiple testing. DMRs with a smoothed FDR < 0.05 were plotted on karyoplots using the karyoploteR package⁴⁴, and ENSEMBL tracks were plotted using the DMR.plot function in the DMRcate package. Enrichment analysis was performed with the enricher function in the clusterProfiler package⁴⁵ using DSigDB⁴⁶ and the Elsevier Pathway Collection⁴⁷. Annotation was done according to hg38.

Human fetal adrenal ex vivo study

Fetal adrenal tissues (GW 7–12) were obtained following elective surgical termination of pregnancy at Copenhagen University Hospital, Hvidovre Hospital, and Herlev Hospital. Women provided written and oral informed consent, and none of the terminations were due to known fetal abnormalities or pregnancy pathologies. Samples were stored at 4 °C during transport, and fetal age was determined by crown-rump and foot length measurements⁴⁸. A total of 23 intact adrenal glands from 23 fetuses (11 male, 12 female) were included.

Adrenal glands were dissected in ice-cold PBS and cultured ex vivo as 1 mm³ tissue fragments in 40 µL of media at 37 °C under 5% CO2 using a hanging drop setup, as previously described⁴⁹. Each gland was halved; fragments from one half (1–9 pieces) were cultured in vehicle control (0.1% DMSO), and fragments from the other half in media containing 1 µM miconazole (n = 11) or clotrimazole (n = 12). Each half was considered one biological replicate to preserve comparable cell population distribution.

Culture media consisted of MEMα supplemented with MEM non-essential amino acids, 2 mM sodium pyruvate, 2 mM L-glutamine, ITS supplement, 1× Penicillin/Streptomycin, and 10% FBS (Gibco), with ITS from Sigma-Aldrich. Media was changed every 48 h, and conditioned media from all replicates of a sample and treatment were pooled to reduce cellularity-based variation. At the end of the 14-day culture, fragments were either snap frozen at –80 °C for analysis or incubated with BrdU for 6 h before formalin fixation. Miconazole nitrate, clotrimazole, and DMSO were obtained from Sigma-Aldrich.

Steroid metabolites in human fetal adrenal culture media

Conditioned media from ex vivo human fetal adrenal cultures were analysed for ten steroid metabolites (DHEAS, PROG, 17-OHP, Adione, T, CORT, 11-DOC, cortisol, cortisone, E1-S) by one of the above-mentioned LC-MS/MS methods (see under serum samples)²⁷. Samples were analysed in a single batch, including calibration material, prepared cell media and triplicates of spiked controls.

Human fetal adrenal DNA and RNA extraction, and RT-qPCR analysis

Total DNA and RNA were extracted from fetal adrenal tissue using the AllPrep DNA/RNA Micro Kit (Qiagen) and quantified with a Nanodrop-1000 spectrophotometer. cDNA was synthesized from 500 ng RNA using the Omniscript RT Kit (Qiagen). RT-qPCR was performed in duplicate on a QuantStudio 7 Flex system (Applied Biosystems) in 20 µL reactions containing TaqMan Fast Universal Master Mix, 3 µL of 1:20 diluted cDNA, and gene-specific TaqMan assays: CYP17A1 (Hs01124136_m1), CYP11A1 (Hs00167984_m1), CYP11B1 (Hs01596404_m1), CYP11B2 (Hs01597732_m1), CYP21A2 (Hs00365734_g1), STAR (Hs00986559_g1), HSD3B2 (Hs00605123_m1), SULT2A1 (Hs00234219_m1). Relative expression was normalized using the geometric mean of RPS20 and ACTB, and analyzed by the comparative Ct method.

Fetal sex was determined by SRY genotyping using primers 5’-GAATATTCCCGCTCTCCGGA-3’ (forward) and 5’-GCTGGTGCTCCATTC TTGAG-3’ (reverse). PCR was performed with AmpliTag DNA polymerase (Life Technologies) under the following conditions: 95 °C for 2 min; 34 cycles of 94 °C for 30 s, 54 °C for 45 s, 72 °C for 1 min; followed by 72 °C for 7 min. PCR products were separated on a 1% agarose gel (SeaKem GTG, Bionordika) containing ethidium bromide alongside a 100 bp DNA ladder (Thermo Scientific).

Statistical analyses of ex vivo adrenal data

Data from human fetal adrenals were assessed for normality and homogeneity of variance using residual statistics. Non-normally distributed data were log-transformed and reassessed. Normally distributed data were analyzed by a two-tailed paired t-test comparing exposed fragments to vehicle controls from the same fetus. Non-normally distributed data were analyzed by the two-tailed Wilcoxon signed-rank test (GraphPad Prism 10).

Ethics statement

The Copenhagen Analgesic Study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Capital Region of Denmark (No. H-19044036) and the Danish Data Protection Agency (No. P-2019-330). Written informed consent was obtained from all participating parents on behalf of the study participants in accordance with Group C regulations under the Danish Committee Act (§19). Animal experiments were approved by the Danish Animal Experiments Inspectorate (license number 2020-15-0201-00539) and were conducted in accordance with national legislation, EU Directive 2010/63/EU, and relevant institutional guidelines for the care and use of laboratory animals. The study adhered to the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines, where applicable. All efforts were made to minimize animal suffering and to reduce the number of animals used. Collection of human fetal material and ex vivo culture experiments was approved by the regional ethics committee (permit number H-1-2012-007).

Results

Study design

In a mother-child cohort, we assessed associations between maternal azole use and steroid hormone levels in the infant offspring. To explore potential mechanisms, we conducted an epigenome-wide DNA methylation study in infant peripheral blood leukocytes and evaluated direct effects on fetal endocrine development in rats, with adrenal transcriptome analyses revealing exposure-related changes in steroidogenic pathways. Finally, hormone levels and expression of key steroidogenic enzymes were assessed in an ex vivo human fetal adrenal tissue model (Fig. 1).

Fig. 1. The three study designs.

The human cohort, the human fetal adrenal ex vivo study, and the rat in vivo study. Timelines indicate the period of exposure (blue) during fetal life for each study. The lollipop markers denote the time of outcome assessments (end of study).

Associations between maternal use of antifungal agents and clinical outcomes in infants

To contextualize the subsequent outcome analyses, maternal and infant characteristics across exposure groups are summarized in Table 1. The proportion of nulliparous women and maternal use of paracetamol during pregnancy, as well as pre-pregnancy body mass index (BMI), differed between the exposure groups.

Table 1.

Maternal characteristics stratified by timing of exposure: Early fetal life (GA < 15 weeks), mid/late fetal life (GA ≥ 15 weeks), and unexposed children as the reference group, in total 287 boys and 302 girls

Maternal characteristics	n	Total Boys	n	Unexposed	n	Early	n	Mid/late
Age at delivery (years)	287	33.02 (4.58)	242	33.05 (4.64)	13	33.92 (4.63)	32	32.47 (4.13)
Gestational age at delivery (days)	287	281.85 (7.76)	242	281.47 (8.00)	13	283.85 (5.64)	32	284.00 (6.14)b
Pre-pregnancy BMI (kg/m²)	282	22.54 (2.83)	237	22.61 (2.80)	13	20.93 (1.07)a	32	22.69 (3.36)
Nulliparous (yes)	287	215 (74.9%)	242	184 (76.0%)	13	10 (76.9%)	32	21 (65.6%)
Use of nicotin products in first trimester (yes)	287	13 (4.5%)	242	11 (4.5%)	13	2 (15.4%)	32	0 (0.0%)
Alcohol consumption in first trimester (yes)	285	89 (31.2%)	240	77 (32.1%)	13	3 (23.1%)	32	9 (28.1%)
Use of paracetamol before GA 15 weeks (yes)	287	80 (27.9%)	242	65 (26.9%)	13	6 (46.2%)	32	9 (28.1%)
Emergency caesarean section (yes)	272	32 (11.8%)	228	31 (13.6%)	12	0 (0.0%)	32	1 (3.1%)
Polyendocrine metabolic ovarian syndrome (yes)	279	11 (3.9%)	234	11 (4.7%)	13	0 (0.0%)	32	0 (0.0%)
		Girls
Age at delivery (years)	302	33.59 (5.16)	246	33.50 (5.37)	32	33.66 (4.07)	24	34.42 (4.29)
Gestational age at delivery (days)	302	282.11 (8.11)	246	281.66 (8.24)	32	283.62 (7.65)	24	284.67 (6.88)b
Pre-pregnancy BMI (kg/m²)	294	22.36 (2.76)	238	22.38 (2.78)	32	22.00 (2.50)	24	22.72 (2.97)
Nulliparous (yes)	302	228 (75.5%)	246	194 (78.9%)	32	18 (56.2%)a	24	16 (66.7%)
Use of nicotin products in first trimester (yes)	302	9 (3.0%)	246	8 (3.3%)	32	0 (0.0%)	24	1 (4.2%)
Alcohol consumption in first trimester (yes)	301	102 (33.9%)	245	87 (35.5%)	32	10 (31.2%)	24	5 (20.8%)
Use of paracetamol before GA 15 weeks (yes)	302	84 (27.8%)	246	69 (28.0%)	32	3 (9.4%)a	24	12 (50.0%)a
Emergency caesarean section (yes)	289	29 (10.0%)	233	26 (11.2%)	32	1 (3.1%)	24	2 (8.3%)
Polyendocrine metabolic ovarian syndrome (yes)	298	22 (7.4%)	242	16 (6.6%)	32	2 (6.2%)	24	4 (16.7%)b

P values were derived from Student’s t test (equal variances) for continuous variables (mean (SD)) and Chi-square test (Yates’ correction) for categorical variables (%).

GA Gestational age, BMI body mass index.

^aIndicates p ≤ 0.05.

^bIndicates 0.05 < p ≤ 0.1.

Maternal use of antifungal medication during pregnancy

We characterized the pattern and timing of antifungal use during pregnancy. Pregnant women reported use of the following antifungal agents: clotrimazole (n = 79) as vaginal suppositories and cream; miconazole (n = 24), as vaginal suppositories and cream; fluconazole (n = 3) as oral tablets; and ketoconazole (n = 5) as shampoo. Dosage and the exact amount of vaginal cream applied were not consistently recorded. In early gestation, the use of clotrimazole was lower in pregnancies with male fetuses than in those with females (8/287 = 2.8% vs. 26/302 = 8.6%, p = 0.002). In mid/late gestation, no differences between sexes were observed, and the use of other antifungal agents was too infrequent to permit stratified analysis.

Anthropometrics and AGD

We subsequently assessed the associations between antifungal exposure and anthropometric measures. For both boys and girls, exposure to antifungal medication in early fetal life was associated with reduced estimated fetal weight (EFW) standard deviation score (SDS), adjusted for parity, maternal pre-pregnancy BMI, and maternal use of paracetamol: −0.28 (−0.49; −0.07), p = 0.010 (all results refer to adjusted estimates by multiple regression analyses) (Table 2). Stratification by sex revealed that this association was primarily driven by the boys (Tables 3 and 4). In girls, exposure during mid/late fetal life was also associated with a reduced EFW: −0.37 SDS (−0.65; −0.08), p = 0.012 (Table 4).

Table 2.

Associations between maternal antifungal medication use and steroid hormones (SDS) and weight parameters in boys and girls

Outcome		Early fetal life (GA < 15 weeks)				Mid/late fetal life (GA ≥ 15 weeks)
		Univariate		Multiple		Univariate		Multiple
Clinical outcomes	n		n		n		n
EFW (g)	45	−69 (−172; 34)	45	−68 (−173; 36)	56	6 (−87; 99)	56	−3 (−98; 91)
EFW (SDS)	45	−0.30 (−0.50; −0.09)a	45	−0.28 (−0.49; −0.07)a	56	−0.14 (−0.33; 0.05)	56	−0.16 (−0.35; 0.03)b
BW (g)	45	−37 (−173; 98)	45	−51 (−185; 83)	55	35 (−89; 159)	55	14 (−108; 136)
BW (SDS)	45	−0.22 (−0.52; 0.09)	45	−0.25 (−0.54; 0.05)b	55	−0.08 (−0.35; 0.20)	55	−0.13 (−0.40; 0.13)
Steroid hormones
PROG (SDS)	36	−0.46 (−0.83; −0.10)a	36	−0.45 (−0.82; −0.08)a	53	0.24 (−0.06; 0.55)	53	0.31 (−0.00; 0.62)b
17−OHPreg (SDS)	36	−0.41 (−0.78; −0.05)a	36	−0.40 (−0.77; −0.04)a	53	0.22 (−0.08; 0.53)	53	0.28 (−0.03; 0.58)b
17−OHP (SDS)	36	−0.35 (−0.69; −0.01)a	36	−0.34 (−0.68; −0.01)a	53	0.20 (−0.08; 0.48)	53	0.25 (−0.04; 0.53)b
DOC (SDS)	36	−0.38 (−0.74; −0.02)a	36	−0.38 (−0.74; −0.03)a	53	0.15 (−0.15; 0.45)	53	0.20 (−0.10; 0.50)
Cortisol (SDS)	36	−0.25 (−0.56; 0.05)	36	−0.25 (−0.55; 0.06)	53	0.05 (−0.20; 0.31)	53	0.08 (−0.18; 0.33)
Cortisone (SDS)	36	−0.18 (−0.49; 0.13)	36	−0.16 (−0.47; 0.15)	53	−0.12 (−0.38; 0.14)	53	−0.08 (−0.34; 0.19)
ALDO (SDS)	36	−0.13 (−0.42; 0.16)	36	−0.15 (−0.44; 0.15)	53	−0.03 (−0.27; 0.22)	53	−0.04 (−0.28; 0.21)
CORT (SDS)	36	−0.30 (−0.64; 0.05)b	36	−0.30 (−0.64; 0.05)b	53	0.04 (−0.25; 0.32)	53	0.06 (−0.23; 0.35)
11−DOC (SDS)	36	−0.42 (−0.79; −0.06)a	36	−0.42 (−0.78; −0.06)a	53	0.12 (−0.18; 0.43)	53	0.17 (−0.13; 0.48)
T (SDS)	36	−0.12 (−0.42; 0.19)	36	−0.10 (−0.41; 0.21)	53	0.24 (−0.02; 0.50)b	53	0.28 (0.02; 0.54)a
Adione (SDS)	36	−0.34 (−0.74; 0.06)b	36	−0.32 (−0.72; 0.09)	53	0.27 (−0.07; 0.60)	53	0.30 (−0.04; 0.64)b
DHEAS (SDS)	36	−0.18 (−0.66; 0.30)	36	−0.14 (−0.63; 0.35)	53	0.14 (−0.27; 0.54)	53	0.19 (−0.22; 0.60)

Analyses were stratified by timing of exposure: early fetal life (GA < 15 weeks) and mid/late fetal life (GA ≥ 15 weeks), with unexposed children as the reference group (univariate: n = 421−487, multiple regression: n = 409−474).

Univariate and multiple linear regression models, β (95% CI). Multiple models were adjusted for maternal pre-pregnancy BMI, parity, and maternal paracetamol use during pregnancy.

SDS Standard deviation score, GA Gestational age, EFW estimated fetal weight, BW birth weight, PROG progesterone, 17-OHPreg 17-hydroxypregnenolone, 17-OHP 17-hydroxyprogesterone, DOC 11-deoxycorticosterone, ALDO aldosterone, CORT corticosterone, 11-DOC 11-deoxycortisol, E1 estrone, E2 estradiol, T testosterone, Adione androstenedione, DHEAS dehydroepiandrosterone sulfate.

^aIndicates associations with a p value ≤ 0.05.

^bIndicates a p value 0.05 < p ≤ 0.1.

Table 3.

Associations between maternal use of antifungal agents and key outcomes in boys

Outcome		Early fetal life (GA < 15 weeks)				Mid/late fetal life (GA ≥ 15 weeks)
		Univariate		Multiple		Univariate		Multiple
Clinical outcomes	n		n		n		n
EFW (g)	13	−192 (−375; −10)a	13	−190 (−376; −4)a	32	−72 (−193; 49)	32	−75 (−197; 47)
EFW (SDS)	13	−0.52 (−0.89; −0.14)a	13	−0.48 (−0.85; −0.10)a	32	−0.01 (−0.25; 0.24)	32	−0.02 (−0.27; 0.23)
BW (g)	13	−139 (−378; 100)	13	−105 (−342; 133)	31	76 (−84; 236)	31	55 (−103; 213)
BW (SDS)	13	−0.48 (−1.00; 0.034)b	13	−0.39 (−0.90; 0.12)	31	0.04 (−0.3; 0.38)	31	−0.02 (−0.36; 0.32)
Fetal AGD (mm)	13	−0.1 (−1.8; 1.6)	13	0.2 (−1.5; 1.9)	32	−0.0 (−1.2; 1.1)	32	−0.1 (−1.2; 1.0)
AGDas (mm)	13	1.9 (−1.0; 4.9)	13	2.2 (−0.8; 5.2)	32	2.3 (0.3; 4.2)a	32	2.2 (0.3; 4.2)a
AGDap (mm)	13	0.9 (−3.0; 4.8)	13	1.8 (−2.2; 5.7)	32	1.5 (−1.1; 4.1)	32	1.6 (−1.0; 4.1)
PW (mm)	13	−1.2 (−2.1; −0.3)a	13	−1.1 (−2.0; −0.2)a	32	0.2 (−0.4; 0.8)	32	0.2 (−0.4; 0.8)
Testis volume (mm3)	12	−38.0 (−97.2; 21.3)	12	−26.5 (−85.5; 32.5)	32	−12.1 (−49.9; 25.6)	32	−7.8 (−44.9; 29.4)
Steroid hormones
PROG (SDS)	11	−0.67 (−1.31; −0.02)a	11	−0.63 (−1.29; 0.03)b	31	0.46 (0.06; 0.87)a	31	0.50 (0.09; 0.90)a
17-OHPreg (SDS)	11	−0.58 (−1.26; 0.09)b	11	−0.57 (−1.25; 0.12)	31	0.42 (0.00; 0.85)a	31	0.46 (0.03; 0.88)a
17-OHP (SDS)	11	−0.46 (−1.01; 0.09)	11	−0.44 (−1.00; 0.12)	31	0.36 (0.02; 0.71)a	31	0.39 (0.05; 0.74)a
DOC (SDS)	11	−0.45 (−1.06; 0.15)	11	−0.52 (−1.13; 0.10)	31	0.37 (−0.00; 0.75)b	31	0.37 (−0.01; 0.75)b
Cortisol (SDS)	11	−0.63 (−1.17; −0.10)a	11	−0.65 (−1.19; −0.11)a	31	0.19 (−0.15; 0.52)	31	0.18 (−0.16; 0.51)
Cortisone (SDS)	11	0.10 (−0.47; 0.66)	11	0.12 (−0.46; 0.70)	31	−0.08 (−0.43; 0.27)	31	−0.05 (−0.40; 0.31)
ALDO (SDS)	11	−0.24 (−0.73; 0.25)	11	−0.22 (−0.73; 0.28)	31	−0.07 (−0.37; 0.24)	31	−0.10 (−0.41; 0.21)
CORT (SDS)	11	−0.66 (−1.26; −0.07)a	11	−0.71 (−1.31; −0.11)a	31	0.18 (−0.18; 0.55)	31	0.17 (−0.20; 0.54)
11−DOC (SDS)	11	−0.41 (−1.08; 0.25)	11	−0.40 (−1.07; 0.28)	31	0.33 (−0.09; 0.74)	31	0.35 (−0.06; 0.77)b
T (SDS)	11	0.03 (−0.50; 0.56)	11	−0.03 (−0.58; 0.51)	31	0.26 (−0.07; 0.59)	31	0.29 (−0.04; 0.63)b
DHT (SDS)	11	−0.02 (−0.47; 0.43)	11	−0.09 (−0.55; 0.37)	31	0.20 (−0.08; 0.48)	31	0.22 (−0.06; 0.50)
Adione (SDS)	11	−0.57 (−1.23; 0.10)b	11	−0.47 (−1.16; 0.22)	31	0.23 (−0.19; 0.64)	31	0.27 (−0.15; 0.69)
DHEAS (SDS)	11	−0.05 (−0.91; 0.81)	11	0.11 (−0.76; 1.00)	31	−0.11 (−0.64; 0.43)	31	−0.07 (−0.61; 0.47)
Free T (pmol/L)	11	−3.96 (−11.01; 3.09)	11	−3.75 (−10.94; 3.43)	31	−0.30 (−4.76; 4.16)	31	−0.14 (−4.63; 4.34)
Pituitary hormones and related markers
LH (SDS)	10	0.22 (−0.24; 0.68)	10	0.22 (−0.26; 0.69)	31	0.23 (−0.05; 0.50)	31	0.21 (−0.07; 0.49)
FSH (SDS)	10	−0.58 (−1.25; 0.08)b	10	−0.51 (−1.20; 0.17)	31	0.00 (−0.39; 0.40)	31	−0.03 (−0.43; 0.37)
Inhibin B (SDS)	10	−0.20 (−0.80; 0.40)	10	−0.18 (−0.79; 0.44)	31	−0.04 (−0.39; 0.32)	31	0.01 (−0.35; 0.37)
SHBG (SDS)	11	0.57 (−0.02; 1.16)b	11	0.49 (−0.12; 1.10)	30	0.44 (0.06; 0.81)a	30	0.46 (0.08; 0.84)a

Analyses were stratified by timing of exposure: early fetal life (GA < 15 weeks) and mid/late fetal life (GA ≥ 15 weeks), with unexposed children as the reference group (univariate: n = 203–242, multiple regression: n = 198–237).

Univariate and multiple linear regression models, β (95% CI). Multiple models were adjusted for maternal pre-pregnancy BMI, parity, and maternal paracetamol use during pregnancy.

SDS Standard deviation score, GA Gestational age, EFW estimated fetal weight, BW birth weight, AGD anogenital distance, AGDas ano-scrotal anogenital distance, AGDap ano-penile anogenital distance, all AGD measures adjusted for EFW/BW in the regression model, PW penile width, PROG progesterone, 17-OHPreg 17-hydroxypregnenolone, 17-OHP 17-hydroxyprogesterone, DOC 11-deoxycorticosterone, ALDO aldosterone, CORT corticosterone, 11-DOC 11-deoxycortisol, E1 estrone, E2 estradiol, T testosterone, Adione androstenedione, DHEAS dehydroepiandrosterone sulfate, LH luteinizing hormone, FSH follicle stimulating hormone, SHBG sex hormone binding globulin.

^aIndicates associations with a p value ≤ 0.05.

^bIndicates a p value 0.05 < p ≤ 0.1.

Table 4.

Associations between maternal antifungal medication use and key outcomes in girls

Outcome		Early fetal life (GA < 15 weeks)				Mid/late fetal life (GA ≥ 15 weeks)
		Univariate		Multiple		Univariate		Multiple
Clinical outcomes	n		n		n		n
EFW (g)	32	11 (−116; 137)	32	4 (−127; 135)	24	92 (−52; 236)	24	82 (−65; 229)
EFW (SDS)	32	−0.15 (−0.40; 0.09)	32	−0.15 (−0.40; 0.11)	24	−0.35 (−0.63; −0.07)a	24	−0.37 (−0.65; −0.08)a
BW (g)	32	55 (−108; 219)	32	22 (−143; 187)	24	−45 (−231; 140)	24	−64 (−249; 121)
BW (SDS)	32	0.02 (−0.34; 0.39)	32	−0.07 (−0.43; 0.30)	24	−0.29 (−0.71; 0.12)	24	−0.35 (−0.76; 0.06)b
Fetal AGD (mm)	32	0.1 (−0.7; 0.9)	32	0.1 (−0.7; 0.9)	23	0.9 (0.1; 1.8)	23	1.0 (0.1; 1.8)a
AGDaf (mm)	32	0.1 (−1.1; 1.3)	32	0.0 (−1.2; 1.2)	23	0.7 (−0.7; 2.0)	23	0.5 (−1.0; 1.9)
AGDac (mm)	32	1.0 (−0.6; 2.7)	32	1.0 (−0.7; 2.8)	24	0.9 (−1.1; 2.9)	24	0.9 (−1.1; 2.9)
Steroid hormones
PROG (SDS)	25	−0.39 (−0.84; 0.07)b	25	−0.39 (−0.85; 0.07)b	22	−0.06 (−0.54; 0.42)	22	0.05 (−0.44; 0.54)
17-OHPreg (SDS)	25	−0.33 (−0.75; 0.10)	25	−0.30 (−0.73; 0.14)	22	−0.07 (−0.52; 0.38)	22	−0.01 (−0.47; 0.45)
17-OHP (SDS)	25	−0.26 (−0.70; 0.18)	25	−0.23 (−0.67; 0.22)	22	−0.07 (−0.53; 0.40)	22	−0.01 (−0.49; 0.46)
DOC (SDS)	25	−0.28 (−0.73; 0.17)	25	−0.29 (−0.74; 0.16)	22	−0.22 (−0.70; 0.26)	22	−0.08 (−0.55; 0.40)
Cortisol (SDS)	25	−0.04 (−0.41; 0.33)	25	−0.02 (−0.40; 0.36)	22	−0.18 (−0.57; 0.21)	22	−0.11 (−0.51; 0.30)
Cortisone (SDS)	25	−0.24 (−0.59; 0.12)	25	−0.15 (−0.51; 0.21)	22	−0.25 (−0.63; 0.13)	22	−0.20 (−0.58; 0.18)
ALDO (SDS)	25	−0.01 (−0.38; 0.35)	25	−0.05 (−0.43; 0.32)	22	−0.04 (−0.43; 0.35)	22	−0.03 (−0.42; 0.37)
CORT (SDS)	25	−0.08 (−0.50; 0.34)	25	−0.08 (−0.51; 0.35)	22	−0.22 (−0.67; 0.22)	22	−0.13 (−0.58; 0.33)
11-DOC (SDS)	25	−0.42 (−0.85; 0.01)b	25	−0.40 (−0.83; 0.04)b	22	−0.17 (−0.63; 0.29)	22	−0.08 (−0.55; 0.38)
E1 (SDS)	26	0.05 (−0.25; 0.35)	26	0.05 (−0.25; 0.35)	22	0.25 (−0.07; 0.57)	22	0.25 (−0.07; 0.57)
E2 (SDS)	26	0.08 (−0.43; 0.59)	26	0.08 (−0.43; 0.59)	22	0.38 (−0.18; 0.93)	22	0.38 (−0.18; 0.93)
T (SDS)	25	−0.140 (−0.525; 0.245)	25	−0.11 (−0.51; 0.29)	22	0.17 (−0.24; 0.58)	22	0.23 (−0.20; 0.65)
Adione (SDS)	25	−0.18 (−0.69; 0.33)	25	−0.18 (−0.71; 0.34)	22	0.26 (−0.28; 0.81)	22	0.31 (−0.25; 0.867)
DHEAS (SDS)	25	−0.23 (−0.82; 0.36)	25	−0.14 (−0.75; 0.46)	22	0.47 (−0.15; 1.09)	22	0.55 (−0.09; 1.19)b

Analyses were stratified by timing of exposure: early fetal life (GA < 15 weeks) and mid/late fetal life (GA ≥ 15 weeks), with unexposed children as the reference group (univariate: n = 218–245, multiple regression: n = 211–237).

Univariate and multiple linear regression models, β (95% CI). Multiple models were adjusted for maternal pre-pregnancy BMI, parity, and maternal paracetamol use during pregnancy.

SDS Standard deviation score, GA Gestational age, EFW estimated fetal weight, BW birth weight, AGD anogenital distance, AGDaf ano-fourchettal anogenital distance, AGDac ano-clitoral anogenital distance, all AGD measures adjusted for EFW/BW in the regression model, PW penile width, PROG progesterone, 17-OHPreg 17-hydroxypregnenolone, 17-OHP 17-hydroxyprogesterone, DOC 11-deoxycorticosterone, ALDO aldosterone, CORT corticosterone, 11-DOC 11-deoxycortisol, E1 estrone, E2 estradiol, T testosterone, Adione androstenedione, DHEAS dehydroepiandrosterone sulfate.

^aIndicates associations with a p value ≤ 0.05.

^bIndicates a p value 0.05 < p ≤ 0.1.

Beyond fetal growth, timing-specific associations were also observed for penile size. Exposure in early fetal life was associated with decreased penile width (Table 3), whereas exposure in mid/late fetal life was associated with increased anoscrotal anogenital distance (AGDas) in infant boys (Table 3) and increased fetal AGD in girls (Table 4).

Circulating levels of steroid hormones in infancy

Given the established role of steroid hormones in fetal growth and sexual differentiation, we assessed circulating hormone profiles during infancy. For both boys and girls, exposure to antifungal agents in early fetal life was associated with reduced circulating levels of progesterone, 17-hydroxypregnenolone, 17-hydroxyprogesterone, 11-deoxycorticosterone and 11-deoxycortisol in infancy compared to the unexposed group; e.g. progesterone: −0.45 SDS (−0.82; −0.08), p = 0.016) (Table 2, Fig. 2 and Supplementary Fig. 1). These associations were primarily driven by boys (Tables 3 and 4). In contrast, exposure in mid/late fetal life was associated with higher circulating levels of progesterone, 17-hydroxypregnenolone, and 17-hydroxyprogesterone and trends towards higher levels of 11-deoxycorticosterone, 11-deoxycortisol, as well as testosterone and sex hormone binding globulin (SHBG) in boys (Table 3).

Fig. 2. Associations between prenatal exposure to antifungal agents and circulating levels of steroid hormones at infancy, stratified by timing of exposure and by sex.

Analyses were stratified by exposure timing: Early gestation (GA < 15 weeks, in total n = 36, boys n = 11, girls n = 25) and mid/late gestation (GA ≥ 15 weeks, in total n = 53, boys n = 31, girls n = 22), compared with unexposed infants (in total n = 409, boys n = 198, girls n = 211). Analyses were performed in boys and girls combined (black) and further stratified in boys (purple) and girls (green). Effect estimates are derived from multiple linear regression models adjusted for maternal pre-pregnancy BMI, parity, and maternal paracetamol use during pregnancy. Progesterone (PROG), 17-hydroxypregnenolone (17-OHPreg), 17-hydroxyprogesterone (17-OHP), 11-deoxycorticosterone (DOC), corticosterone (CORT), aldosterone (ALDO), 11-deoxycortisol (11-DOC), dehydroepiandrosterone sulfate (DHEAS), androstenedione (Adione), testosterone (T), dihydrotestosterone (DHT), estrone (E1), estradiol (E2). Dots represent adjusted effect estimates, and horizontal lines indicate 95% confidence intervals. All p values are two-sided. *indicates associations with a p value ≤ 0.05 and †indicates a p value 0.05 < p ≤ 0.1.

Sensitivity analyses

Finally, to assess the robustness of the observed associations, we conducted sensitivity analyses excluding (a) infants treated with corticosteroid-containing medication and (b) deliveries by emergency caesarean section. Across these analyses, associations between antifungal exposure in early or mid/late fetal life and the primary outcomes remained consistent (Supplementary Tables 1–2).

Animal study

Toxicity and litter parameters

Our next objective was to conduct a rat in vivo study to investigate potential causal relationships between azole exposure and effects observed in human cohorts, allowing controlled single-compound exposure. Pregnant rat dams were exposed to miconazole at low (35 mg/kg bw/day) and high (70 mg/kg bw/day) doses from GD 7–21, which includes the MPW. No external signs of maternal toxicity were observed, and maternal body weights at termination and gestational body weight gain were unaffected (Supplementary Table 3). Dam liver weights were slightly increased in both exposure groups (11–13%), reaching statistical significance for relative liver weight (11%) in the high-dose group (Supplementary Table 3). No statistically significant effects were seen on litter size, post-implantation loss, or fetal weights. In male fetuses, AGD was longer in both low- and high-dose exposure groups compared with controls (5%), with statistically significant differences observed for the AGD index (AGD divided by body weight cube root). Female AGD was not significantly affected (Supplementary Table 3).

Histopathological changes in adrenal glands

No treatment-related histopathological changes were observed in the adrenal glands of GD17 or GD21 dams or male fetuses. One high-dose GD17 dam showed zona glomerulosa hypertrophy (Supplementary Fig. 2). Mild angiectasis occurred in the adrenal glands of one GD21 dam at 70 mg/kg (Supplementary Fig. 3), while minimal angiectasis was observed in adrenal glands of both control (2/5 GD17 dams) and high-dose animals (3/6 GD17 and 5/7 GD21 dams). As the angiectasis distribution and severity were within control levels, it was not considered treatment-related (Supplementary Table 4).

Steroid hormone concentrations in fetal rat plasma were affected by miconazole and clotrimazole exposure

As summarized in Fig. 3, exposure to both 35 and 70 mg/kg bw/day miconazole significantly increased 17-hydroxypregnenolone levels in the GD21 rat fetuses, independent of sex. Androstenedione levels were also increased in both male and female fetuses in a dose-dependent manner, reaching statistical significance at 70 mg/kg bw/day miconazole. However, testosterone, estradiol, and estrone levels were unaffected by miconazole exposure, and while aldosterone levels were higher in high-dose animals, this was not statistically significant and there were no observed effects on other measured steroid metabolites (17-hydroxyprogesterone, 11-deoxycorticosterone, corticosterone, aldosterone) (Supplementary Fig. 4). Notably, clotrimazole exposure significantly decreased progesterone and estrone levels in the blood of both male and female fetuses at 75 mg/kg bw/day, as previously reported²⁰, with fold-changes shown in Fig. 3. In male fetuses only, estradiol levels were also reduced in a dose-dependent manner, reaching statistical significance in the high dose group. Other steroid metabolites (corticosterone, androstenedione, testosterone) were not affected by clotrimazole exposure.

Fig. 3. Hormone concentrations in plasma from control and exposed fetal rats.

Hormone concentrations at gestational day (GD) 21 following exposure to miconazole (35 or 70 mg/kg bw/day) or clotrimazole (25 or 75 mg/kg bw/day) GD7-21. Data are presented as box and whiskers plots (controls in gray, exposed in white) showing the 2.5th, 25th, 50th (median), 75th, and 97.5th percentiles, expressed as fold change relative to the control mean of pooled samples for each sex within each litter. Progesterone (PROG), 17-hydroxypregnenolone (17-OHPreg), and corticosterone (CORT) are shown as combined data for males and females (n = 12–16), whereas other hormones are presented separately by sex (n = 5–8). NA = not assessed. Statistical analyses were performed on raw hormone concentrations (nM) (see supplementary Fig. 4). Mean control concentrations in the miconazole study were 4.6 nM for 17-OHPreg, 24.7 nM for PROG, 801 nM for CORT, 1.9 nM (male) or 1.3 nM (female) for androstenedione (Adione), 1.4 nM (male) or 0.3 nM (female) for testosterone (T), 8.34 nM (male) or 8.30 nM (female) for estrone (E1), and 0.199 nM (male) or 0.125 nM (female) for estradiol (E2). Hormone concentration data for clotrimazole have been published previously (Draskau et al 2021). Normally distributed data were analyzed by one-way ANOVA with Dunnett’s post hoc test, and non-normally distributed data using Kruskal-Wallis with Dunn’s multiple comparison test.*p ≤ 0.05.

Fetal rat adrenal transcriptomes

To further characterize adrenal involvement, we analyzed transcriptomes from GD21 fetal rat adrenal glands in clotrimazole-exposed male rats and unexposed control male and female rats (Fig. 4). A total of 358 statistically significantly differentially expressed genes (DEGs) were identified across these groups (Supplementary Data 1). A set of 342 genes was found to be differentially expressed between female and male adrenal glands in late gestation (Fig. 4a). Clotrimazole exposure in males led to 79 DEGs (27 expressed at higher levels and 52 lower expressed) compared to control males (Fig. 4a).

Fig. 4. Analysis of fetal rat adrenal transcriptomes.

a BRB-profiling detected 358 differentially expressed genes (DEGs) comparing control (Ctrl) male, ctrl female and clotrimazole (cloz)-exposed males (full list can be found in Supplementary Data 1) by applying three filtration steps: the detectable cutoff (≥0.659), the fold change (FC) cutoff (≥1.2), and the adjusted F-value (≤0.05). The listed numbers of DEGs refer to unique genes within each analysis, so that genes shared between groups are not counted repeatedly. b Heatmap representation of DEGs in rat adrenal glands from gestational day (GD) 21. The DEGs could be divided into four expression patterns P1-4. Red- and blue-colored bars indicate up (red) or down (blue) regulated genes. c Selected significantly enriched biological process terms (full list can be found in Supplementary Data 2) are given, followed by the total number of genes associated with the term and the adjusted (adj.) p values. N = 8 samples/group. d DEGs from selected enriched terms in P2 and P3; only unique genes are listed.

The 358 DEGs were further partitioned into four expression patterns using K-means clustering (P1-4, Fig. 4b and Supplementary Data 1). These patterns were then associated with enriched biological processes through functional enrichment analysis (Fig. 4c and Supplementary Data 2). P1 comprised 80 DEGs with higher expression in male versus female adrenal glands, significantly associated with the functional term “bounding membrane of organelle” (n = 20 genes, p value = 0.0446) (Fig. 4c). P2 comprised 138 DEGs expressed at lower levels in control male adrenal glands than in females, significantly associated with 41 enriched biological process terms such as “anatomical structure development” (n = 73 genes, p value = 0.0431), “cell differentiation” (n = 60 genes, p value = 0.0120), “sex differentiation” (n = 14 genes, p value = 0.0120), “reproductive system development” (n = 14 genes, p value: 0.0135), and “male sex differentiation” (n = 10 genes, p value = 0.0225) (Fig. 4c, d and Supplementary Data 2).

P3 and P4 comprised 140 DEGs showing distinct changes in response to clotrimazole in the male adrenal glands, in addition to sex differences (Fig. 4b). P3 included 69 DEGs with decreased expression after clotrimazole exposure, significantly associated with 14 enriched biological process terms including “metabolic pathways” (n = 19 genes, p value = 0.0024), “steroid hormone biosynthetic process” (n = 5 genes, p value = 0.0073), “response to gonadotropin” (n = 5 genes, p value = 0.0330), “cortisol synthesis and secretion” (n = 5 genes, p value = 0.0024), “response to fungicide” (n = 4 genes, p value = 0.0072), “Leydig cell differentiation” (n = 4 genes, p value = 0.0072), and ‘androgen biosynthetic process’ (n = 3 genes, p value = 0.0160) (Fig. 4b–d and Supplementary Data 2). P4 included 71 DEGs with increased expression after clotrimazole exposure, but without significant association with any biological process terms (Fig. 4b, c).

Overall, the gene expression profile of the clotrimazole-exposed males was intermediate to the male and female controls (Fig. 4b).

DNA methylation analysis of peripheral blood leucocytes in infants

To assess whether in utero exposure to antifungal agents leaves detectable molecular signatures, we performed an epigenome‑wide association study (EWAS) of DNA methylation in peripheral blood leukocytes from 275 infants using the Illumina EPIC array. Based on reduced steroid levels in infants exposed during early gestation, we compared DNA methylation profiles between those exposed in early fetal life (n = 31) and unexposed infants (n = 244).

We identified 262 differentially methylated regions (DMRs) with a smoothed FDR < 0.05 (Fig. 5a), corresponding to 237 putatively affected genes (Supplementary Data 3). Two genes, GPC3 and EMD, were also differentially expressed in clotrimazole‑exposed rats (Fig. 4; Supplementary Data 1), and both DMRs showed significantly lower methylation levels in exposed infants (Fig. 5c, d). GPC3 is implicated in the regulation of Wnt/β‑catenin and Yap signaling pathways⁵⁰, whereas EMD (encoding Emerin) is an inner nuclear membrane protein with diverse regulatory and structural roles⁵¹.

Fig. 5. Differential DNA methylation in infancy following in utero exposure to fungicides.

DNA methylation measured in peripheral blood collected during infancy was analysed according to fungicide exposure early in utero. a A total of 262 differentially methylated regions (DMRs) were identified using a smoothed threshold of FDR < 0.05, and were distributed across all chromosomes as indicated by purple ticks in the karyogram. b These DMRs mapped to 237 putatively affected genes, including GPC3 and EMD, which also overlapped with genes differentially expressed in clotrimazole-exposed versus control rats. c, d Genomic regions containing the DMRs associated with GPC3 and EMD, respectively, are shown with ENSEMBL genome tracks, CpG sites indicated by green ticks, and the DMRs by purple shading. Smoothed methylation levels for infants exposed early in utero (purple) and unexposed infants (green) are shown below with shaded confidence intervals (CI). e Enrichment analyses of the 237 putatively affected genes using the Drug Signature Database (left) and the Elsevier Pathway Collection (right). The color of points indicates the adjusted P values, and the dot size reflects the number of genes associated with each term. The q-score indicates the -log of the adjusted P value.

Functional enrichment analyses using the Drug Signature Database (DSigDB) revealed a significant drug signature (adjusted P = 2.8 × 10⁻⁵; Fig. 5e and Supplementary Data 4) matching Oligomycin, a polyketide with potent antifungal activity⁵². This enrichment was driven by DMRs in RB1, ABCG5, ABCG8, and SOD2. Pathway enrichment analysis using the Elsevier Pathway Collection further revealed a modest but significant overrepresentation (adjusted P < 0.02; Supplementary Data 5) of endocrine‑related pathways involving corticosteroids, androgens, luteinizing hormone (LH), and follicle‑stimulating hormone (FSH), all converging on a shared DMR within GNAS (Fig. 5e).

Human fetal adrenal ex vivo cultures

Miconazole and clotrimazole exposure alter hormone production in human fetal ex vivo cultured adrenals

To corroborate observations in infant cohorts and the rat in vivo study, we analysed hormone synthesis and expression of key steroidogenic enzymes in primary cultures of human fetal adrenal tissue. Exposure to miconazole (1 µM) increased several adrenal steroids (Supplementary Fig. 5), with progesterone and 17-hydroxyprogesterone elevated 3.1-fold and 1.8-fold, respectively, compared to vehicle controls (Fig. 6). Miconazole also increased 11-deoxycorticosterone (2.1-fold) and testosterone (7.1-fold) (Fig. 6). In contrast, clotrimazole (1 µM) markedly reduced glucocorticoid production (Supplementary Fig. 5), with corticosterone decreased to 0.02-fold, cortisone to 0.2-fold, cortisol to 0.06-fold, and dehydro-epiandrosterone sulfate (DHEAS) to 0.2-fold of control levels (Fig. 6). Conversely, clotrimazole increased 11-deoxycorticosterone (4.3-fold), androstenedione (4.6-fold) and testosterone (3.5-fold) production compared to vehicle control treated cultures (Fig. 6). The reduced cortisol:11-deoxycorticosterone ratio suggests inhibition of CYP11B1 activity (analysis of hormone ratios are shown in Supplementary Fig. 6). Despite these steroid changes, expression of steroidogenesis genes (CYP17A1, CYP11A1, CYP11B1, CYP11B2, STAR, CYP21A2, HSD3B2, SULT2A1) was unchanged (Supplementary Fig. 7).

Fig. 6. Hormone production from ex vivo human fetal adrenals.

Hormone production measured in medium from control and miconazole (1 µM, miz) or clotrimazole (1 µM, cloz) exposed human fetal adrenals cultured ex vivo. Data presented in box and whiskers plots showing the 2.5th, 25th, 50th (median), 75th, and 97.5th percentile expressed as fold change compared to control mean. N = 11 samples/group. Statistical analyses were performed on raw hormone concentrations (nM) (see Supplementary Fig. 5). Progesterone (PROG), corticosterone (CORT), 17-hydroxyprogesterone (17-OHP), 11-deoxycortisol (11-DOC), dehydroepiandrosterone sulfate (DHEAS), androstenedione (Adione), testosterone (T), and estrone sulfate (E1-S). Normally distributed data were analyzed by a two-tailed paired t-test comparing exposed fragments to vehicle controls from the same fetus. Non-normally distributed data were analyzed by the two-tailed Wilcoxon signed-rank test. *p ≤ 0.05.

Overall, the suppressive effects of clotrimazole on fetal rat and ex vivo human fetal adrenal steroid hormone production are consistent with reduced circulating steroid hormone levels observed in infants exposed in early fetal life. Similarly, the effects of miconazole exposure observed in ex vivo human fetal adrenals are largely in line with the findings in fetal rat plasma as well as in children exposed in mid/late fetal life. Finally, it also aligns with previous studies showing little effects in testis transcriptome despite changes to hormone output, suggesting enzyme-level interference and not direct gene regulation in target tissues⁵³.

Discussion

In this translational study, we integrated data from a human mother-child cohort with evidence from in vivo animal studies and an ex vivo human fetal adrenal model to demonstrate that prenatal exposure to topical antifungal agents persistently alters circulating steroid hormone levels in infants. Effects across models support a causal link between azole exposure and disrupted steroid hormone regulation. Importantly, the rat in vivo exposures resulted in fetal concentrations within the human-relevant range, and placental transfer of azoles has been demonstrated both in our in vivo study²⁰ and in a human term placental perfusion model⁵⁴. Topical antifungals, including vaginal-antimycotic agents, have not previously been associated with adverse outcomes in humans^55,56. Leveraging a human cohort with well-characterized exposure timing and state-of-the-art LC-MS/MS measurements of multiple steroid hormones, we uncover associations that, to the best of our knowledge, have not previously been evaluated.

In the human cohort, associations differed markedly depending on the timing of exposure during pregnancy. Exposure in early fetal life, when sex differentiation and the MPW occur, was associated with reduced steroid hormone output and indicators of diminished androgen action, including smaller penile width. In contrast, exposure in mid/late fetal life, when growth and maturation of reproductive organs and tissues occur, was associated with elevated steroid levels and longer AGD in both male infants and female fetuses. Given the major developmental transition of the human fetal adrenal around GW 15¹⁷, some temporal differences were expected; however, the reversal in effect direction across time points is particularly notable and might be linked to sex development. A recent human cohort study reported a similar pattern, with exposure prior to GW 19 associated with reduced androgen precursors and shorter male AGD, whereas later exposure tended to be associated with longer AGD²¹. Because AGD is sensitive to fetal androgen activity⁵⁷, one plausible mechanism of action is suggested by our ex vivo model, where clotrimazole and miconazole increased androgen production in human fetal adrenal tissue. However, in early human male fetal life, testes contribute substantially to circulating androgen levels, and we speculate that the effects observed on adrenal steroidogenesis are similarly present in the gonads. The design of the present study does not allow evaluation of direct effects on the human fetal gonads, but early gestation is a critical window for sex differentiation and testis-derived testosterone production in human male fetuses. Direct interference with key CYP enzymes involved in androgen synthesis (e.g. CYP11A1, CYP17A1) could alter circulating androgen levels and affect masculinization of the male fetus. Supporting this, our rat model demonstrated exposure-related changes in sex steroid concentrations. Given the near absence of CYP17A1 expression, which is normally suppressed in rat adrenals by a DNA methylation mechanism⁵⁸, these effects likely originate from the testis, highlighting an important species difference. Although prenatal exposure was not associated with infant levels of pituitary or testis-derived hormones in the present human cohort study, we cannot rule out effects on the human fetal gonads.

The epidemiological associations persisted after adjusting for relevant covariates. While experimental data support a causal effect of antifungal medication, residual confounding cannot be excluded. Causality is supported by consistent findings across the three models. In the human cohort, clotrimazole was the most used antifungal agent, and early fetal exposure was associated with reduced progesterone levels. Correspondingly, clotrimazole reduced progesterone in cultured human fetal adrenal tissues and in plasma from exposed fetal rats. Associations between early fetal exposure and reduced glucocorticoid levels were supported by decreased corticosterone, cortisol and cortisone production in cultured human fetal adrenal tissues after clotrimazole exposure, as well as changes in the DNA methylation of infants that associate with hormone pathways. Mid/late fetal exposure was associated with elevated androgens and increased AGD, consistent with findings from the rat in vivo study. Different effects of clotrimazole and miconazole on specific steroids highlight the complexity of azole-induced endocrine disruption. The limited number of exposed children precluded compound-specific subgroup analyses in the cohort, highlighting the need for larger studies.

The integrative analyses of epigenetic profiles in blood cells of human infants, as well as transcriptional data of adrenal tissue from the rat study, demonstrate that early fetal exposure to antifungal agents is associated with persistent transcriptional changes in pathways central to steroid hormone production, providing a plausible mechanistic explanation for the clinically observed reduction in steroid hormone levels at infancy. The overlap between the rat transcriptomics study and the human EWAS was limited (two genes: GPC3 and EMD) and, as such, does not suggest a strong association. However, there is not always a one-to-one relationship between a change in DNA methylation and a subsequent change in transcript levels. Furthermore, the substantial species differences, differences in exposure windows and the different tissues investigated make it unlikely that the overlap should be big. Therefore, we believe that the cross-species overlap of GPC3 and EMD still represents a biologically interesting and potentially conserved effect of azole exposure, that strengthen the causal inference and reduces the risk that the findings reflect species - or cohort-specific effects. Both genes are implicated in developmental and cellular structural processes, which could suggest that azole exposure disrupts organ differentiation or integrity, leading to compromised steroid hormone synthesis regulation.

Functional enrichment analyses indicated an overrepresentation of differentially methylated loci in pathways related to mitochondrial function and enzymatic activity. These findings may be compatible with the pharmacological properties of antifungal agents, including effects on CYP enzyme activity. The observed enrichment of LH and FSH signaling and steroidogenesis pathways suggests a biologically plausible link between early fetal antifungal exposure and the observed alterations in postnatal steroid hormone profiles. Together, these results support a model in which early fetal antifungal exposure may influence the transcriptional regulation of key steroidogenic networks, potentially leading to longer-term functional consequences detectable during infancy.

In the human cohort, associations between prenatal exposure and altered gluco- and mineralocorticoid hormone levels were more pronounced in males than in females. While sex-specific effects were not observed in the experimental models, sex differences in adrenal development as well as androgen- and estrogen-induced regulation of adrenal function have been suggested⁵⁹. We speculate that male fetuses have inherently higher adrenal activity, rendering them more sensitive to disruption.

Interestingly, only one-third of women carrying male fetuses used antifungal medication in early pregnancy compared to those carrying female fetuses. Studies suggest that the sex of the fetus affects circulating androgens in pregnant women⁶⁰. Vulvovaginal candidiasis is strongly influenced by estrogen levels⁶¹, and we speculate that sex-specific hormone production in early fetal life affects maternal susceptibility to fungal infection.

A few additional findings of interest emerged in the human cohort; prenatal exposure to antifungal medication was associated with reduced estimated fetal weight. The associations depended on sex and the timing of exposure. Similar effects were not observed in the rat study, nor have they been reported in previous human studies⁶², possibly due to the lack of detailed information on exposure timing.

Study limitations include the modest number of exposed infants, limited power for compound-specific analyses, and incomplete data on medication dosing. Although exposure information was collected prospectively during pregnancy, some degree of exposure misclassification cannot be excluded. Residual confounding from unmeasured maternal characteristics or environmental exposures may also have influenced the observed associations despite adjustment for predefined confounders. Experimental models suggest that different azoles have distinct effects on CYP enzyme activity and steroid hormone levels, emphasizing the need for larger, long-term human cohort studies specifically designed to assess individual azoles. Given the developmental and species-specific ontogeny of steroid hormone production, direct comparison across models should be interpreted with caution. In the animal study, fetuses were exposed during the entire period of adrenal development, preventing comparative analyses of effects on prenatal timing of exposure. Rat fetuses were examined in late gestation, not postnatally as in the human study. In the experimental models, effects of clotrimazole and miconazole were assessed separately; however, the lack of power did not allow similar subgroup analyses in the human cohort. Antifungal medications pass the placenta and enter the fetal circulation; however, it cannot be ruled out that part of the observed effects in the offspring may be secondary to maternal physiological responses.

Conclusion

This study demonstrates that prenatal exposure to commonly used topical antifungal agents is associated with an altered steroid hormone profile in infancy. The effects seem dependent on the timing of exposure, and causality is supported by experimental models.

Given the widespread over-the-counter availability of topical antifungal agents, these findings stress the need for larger studies with extended follow-up that can distinguish long-term and compound-specific effects.

Supplementary information

Transparent Peer Review file ^{(752.6KB, pdf)}

Supplementary Tables and Figures PDF ^{(2.8MB, pdf)}

43856_2026_1815_MOESM3_ESM.pdf ^{(118.4KB, pdf)}

Description of Additional Supplementary Files

Supplementary Data 1–8 ^{(681.3KB, xlsx)}

Author contributions

G.M., M.K.D., A.J., T.S. and C.P.H. conceptualized and designed the study. G.M., M.K.D., T.S. and C.P.H. wrote the original draft of the manuscript. All authors contributed to the review and editing, and they approved the final version of the manuscript. G.M., A.J.O., A.J., M.B.F., T.S. and C.P.H. acquired funding. G.M., M.K.D., A.J.O., C.M., B.E., K.M. and M.B.F. performed the experimental investigations. G.M., M.K.D., F.C., K.A., K.M., T.K.H. and M.B.F. conducted the formal analysis. T.S., C.P.H., H.H., F.C., A.A., A.J. and A.J.O. provided resources. G.M., M.K.D., H.F., F.C. and K.A. curated the data. M.K.D., G.M., F.C., K.A., A.J.O., T.S., and C.P.H. contributed to data visualization. T.S., C.P.H., C.E.F., A.J., A.A. and A.J.O. supervised the project.

Peer review

Peer review information

Communications Medicine the anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was funded by the Danish Environmental Protection Agency as a project under the Danish Centre for Endocrine Disrupters (CeHoS). Rigshospitalets Research Council supported the research under grant (E-22778-02). This study is part of EDMaRC, an International Centre for Research and Research Training in Endocrine Disruption of Male Reproduction and Child Health at the Department of Growth and Reproduction at Rigshospitalet. The publication was completed as part of the MERLON project⁶³ under grant agreement No. 101137411. The MERLON project is funded by the European Union. Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union or the European Health and Digital Executive Agency (HADEA). Neither the European Union nor HADEA can be held responsible for them.

Data availability

The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request. The source data for Fig. 2 is in Supplementary Data 6, source data for Fig. 3 is in Supplementary Data 7, and source data for Fig. 6 is in Supplementary Data 8. Raw and processed RNA sequencing data are deposited at GEO [accession GSE317878] and the TOXsIgN repository (https://toxsign.genouest.org/)^35,36. The DNA methylation data have been submitted to the EMBL-EBI BioStudies repository (https://www.ebi.ac.uk/biostudies/) under the accession E-MTAB-17259.

Code availability

No novel algorithms or custom software were developed for this study. Statistical and bioinformatic analyses were performed using standard, publicly available software and packages. Custom scripts used for data processing, statistical analyses, and figure generation are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no conflicts of interest.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s43856-026-01815-z.