Diethylstilbestrol and autism.

- Psychiatric effects of estrogens and progestins in in-utero exposed children were analyzed using data from a French cohort. Comparisons of groups exposed to different synthetic progestins or estrogens (DES, EE) revealed that the frequency of psychiatric disorders (schizophrenia, bipolarity, eating disorders) is the same whatever the drug. - A study on the molecular mechanism of estrogen action in the brain of in-utero exposed children showed that psychosis was correlated with specific methylome modifications (hypermethylation) in two genes implicated in the neurodevelopment, thus highlighting an epigenetic mechanism. - Progestin may contribute to ASD development via an epigenetic mechanism (hypermethylation of the ERβ gene promoter in the amygdala). A large Chinese epidemiological study showed that many mothers of children with ASD were treated or contaminated by progestins and by oral contraceptive components during pregnancy. - Analysis of the amniotic fluid in a Danish cohort of boys that included also children with ASD showed that prenatal exposure to estrogens, or other prenatal sex steroids (e.g. progestins) contributes to autism. - A multigenerational and possibly transgenerational effect of DES on neurodevelopment and psychiatric disorders had been observed, especially for ASD.

Diethylstilbestrol (DES), a synthetic diphenol with powerful estrogenic properties, is considered the culprit of many somatic disorders in children and grandchildren of mothers treated during pregnancy ( 1 ). It has been estimated that worldwide, ~10 million people have been exposed to DES ( 2 ). However, very few studies investigated DES effects on neurodevelopment and psychiatric disorder occurrence, especially Autism Spectrum Disorders (ASD), in exposed children and their descendants. A previous review that included a small number of epidemiological studies reported inconclusive results on the risk of psychiatric disorders in DES-exposed individuals ( 3 ). On the other hand, O’Reilly et al. ( 4 ) analyzed data from a large US epidemiological study (the Nurses’ Health Study), on 76,240 American women among whom 1,612 were exposed to DES in utero . They found that the occurrence of major depressive and anxiety disorders was significantly increased (by a factor of 1.47) in exposed women. Two other studies reported psychiatric effects, particularly schizophrenia, bipolar disorders, depression, eating disorders, suicide attempts, and suicide, in adolescent/adults who were in-utero exposed to DES ( 5 , 6 ) and to progestins ( 7 ). In humans, ASD, a neurodevelopmental disorder, is characterized by impaired social interaction, language and communication and also stereotypical behavior ( 8 ). In the last years, several studies have brought insights into ASD genetics ( 9 )and epigenetics ( 10 ); however, very few works have investigated the relationship between ASD and in-utero exposure to synthetic estrogens and/or progestins. Specifically, Zou et al. ( 11 ) demonstrated that in rats, prenatal exposure to levonorgestrel (LNG; a synthetic progestin used in oral contraceptive pills) inhibits estrogen receptor ERβ expression in the amygdala through increased methylation of its promoter. ERβ downregulation led to decreased expression of superoxide dismutase 2 and estrogen-related receptor α, and subsequently triggered damage in amygdala tissue through oxidative stress and dysfunction of mitochondria and fatty acid metabolism. These effects contributed to the autism-like behavior observed in the in-utero exposed offspring. This suggests that upon LNG exposure, ERβ downregulation in the amygdala during neurodevelopment may contribute to autism-like behavior. Similarly, Li et al. ( 12 ) in a large study on a human population found that prenatal progestin exposure was associated with ASD (see section 1.3). According to Perrotti ( 13 ), ERs and progesterone receptors (PRs) function as ligand-dependent transcription factors and their distinctive transcriptional activities are initiated when they bind to their respective ligands, like other members of the nuclear hormone receptor superfamily. The ER and PR signaling pathways are modulated through epigenetic mechanisms. In recent decades, ASD prevalence has significantly increased. This indicates that besides heritability and genetic background, other factors might be implicated in these disorders, particularly environmental factors that influence the central nervous system (CNS) development. Several toxic compounds have been linked to ASD risk, such as heavy metals, air pollutants, pesticides, polycyclic aromatic hydrocarbons and polybrominated diphenyl ethers. In this review, we present studies suggesting that exposure to DES and other synthetic sexual hormones also might contribute to the increasing ASD prevalence.

A systematic review of the literature was carried out by searching the PubMed and Google Scholar databases for the period from 2010 to 2022 following the PRISMA guidelines. The chosen starting year was 2010 because it was the year of publication of the study by Titus-Ernstoff et al. ( 14 ) showing for the first time DES multigenerational effect in humans, although it only concerned somatic disorders. Only the study by Colborn published in 2004 ( 15 ) was added, as a tribute. Studies were identified using the following keywords: diethylstilbestrol, diethylstilbestrol and ASD, progestins, progestins and ASD, psychosis and endocrine disrupting compounds, estrogens and epigenetics. Publications that were not articles (e.g. conference abstracts) and publications that were not in English were excluded. Although DES has not been administered to pregnant women since 1971 (USA) and 1977 (France), it continues to wreak havoc in the subsequent generations. Moreover, it was replaced by ethinylestradiol (EE), which cannot be used during pregnancy since 1980, and by synthetic progestins that are still marketed with EE, particularly in contraceptive pills. Synthetic hormones, such as DES, EE, synthetic progestins, were prescribed often as a cocktail not only to women who had a miscarriage or were at risk of miscarriage, but also to those in discomfort or even as a “morning after pill”. The database of the families of the French patient association Hhorages-France allowed assessing the causal link between in-utero exposure to synthetic hormones, particularly DES, and the occurrence of severe psychotic disorders after adolescence in exposed children. This database is based on the responses to a detailed questionnaire written by physicians and researchers and approved by the French national commission for data protection and freedom (CNIL) ( 7 ). In 2016, we analyzed data on 1,182 pregnancies from 529 mothers from the Hhorages-France cohort ( Table 1 ). We found that 603/720 children exposed to DES in-utero and 16 post-DES children (i.e. born after a previous exposed pregnancy but without in utero exposure) had psychiatric disorders, but none of the 180 children born before the pregnancy with DES treatment. Compared with the general population, the prevalence of psychiatric disorders in this group was significantly increased ( Table 1b ) ( 5 ). Other synthetic hormones, such as EE and progestogens, also are implicated. For instance, in-utero exposure to estrogens or progestins increases the risk of bipolar disorders/depression and schizophrenia ( Figure 1 ) ( 16 , 17 ). Patients with bipolar disorder or depression often present also eating disorders (bulimia, anorexia) ( Table 1 ). In the Hhorages cohort ( 5 ), 30% of children in-utero exposed to DES had eating disorders (n =83 vs n=257 with bipolar depression, depression, or anxiety disorder alone). This high rate suggests a possible endocrine cause for this pathology the vulnerability factors of which are still incompletely known ( 18 ). Similar observations where reported also in other studies (in the general population) where eating disorders were often associated with major depressive disorders ( 19 , 20 ). Specifically, Liu et al. ( 21 ) carried out a genome-wide association study in which they compared genomic data on 184 patients with bipolar disorder comorbid with eating disorders, 1,370 healthy controls, and 2,006 patients with bipolar disorder only. They found a genome-wide association of eating disorders and bipolar disorders and confirmed the association of the SOX2-OT gene with bipolar disorders comorbid with eating disorders. Regions containing genes involved in neurodevelopment were also associated with this group. Moreover, Remnelius et al. ( 22 ), Carpita et al. ( 23 ), Saure et al. ( 24 ), Westwood et al. ( 25 ) reported a link between ASD and anorexia nervosa, and suggested a specific genetic link between anorexia nervosa and ASD. Effects of synthetic estrogens (DES, EE) on psychiatric disorders in-utero exposed children (2 nd generation). (a) Total number ofpsychological/psychiatric disorders among the 982 DES/EE-exposed and post-DES unexposed children. The first unexposed children (n= 180) were used as intrafamilial control. With permission by Taylor & Francis Ltd ( www.tandfonline.com ). (b) Psychiatric disorder rate among the 982DES/EE in-utero exposed and post-DES children and in the general population. Note the particularly high number of suicides and suicideattempts. With permission by Taylor & Francis Ltd ( www.tandfonline.com ). Psychiatric disorder rates in boys in-utero exposed to DES/EE (A) or to progestins (B) and in girls exposed to DES/EE (C) or to progestins (D) . © Hhorages-France data. Data and families from the Hhorages-France cohort were used also to study the molecular basis of the causal link between in-utero exposure to synthetic estrogens and the appearance of psychotic disorders, such as schizophrenia and bipolar disorders ( 26 ). From this cohort, 69 siblings (n=37 in-utero exposed to DES: 12 boys and 25 girls; and n=32 unexposed: 12 boys and 20 girls) from 30 families were selected for DNA methylation analysis in peripheral blood samples. This analysis highlighted specific differences in DNA methylation at the ZFP57 and ADAMTS9 genes between exposed and unexposed participants that were correlated with psychotic disorders, particularly schizophrenia, depression and bipolar disorders. In this pioneering work, the authors analyzed the methylation variations of 411,947 CpG sites/per genome in peripheral blood samples, because it was previously reported that the methylation profiles in brain and blood samples are rather similar ( 27 ). Controls were unexposed elders (“informative family”) and another cohort of unexposed adolescents who became schizophrenic and were followed at St. Anne Hospital, Paris ( 28 ). On the basis of these results, the authors suggested that in exposed individuals, psychosis is associated with specific methylome modifications (i.e. hyper methylation) that modify the gene activity and consequently could influence neurodevelopment and neuroplasticity. One of the genes in which methylation changes have been observed is ADAMTS9 that encodes a disintegrin and metalloproteinase with thrombospondin motifs 9 (ADAMTS9). Members of the ADAMTS family of proteoglycanases have been implicated in proteoglycan cleavage, organ shape control during development, and angiogenesis inhibition ( 29 , 30 ). Specifically, sexual organ shape alteration, which is often observed in children in-utero exposed to DES, is caused by deregulation of procollagen N-proteinases of the ADAMTS family that regulate collagen fibril assembly and of aggrecanases, other ADAMTS members that regulate the cleavage of extracellular matrix proteoglycans, thus modifying the extracellular matrix structure and function. In addition, the aberrant activation of ADAMTS9 could induce premature rupture of gestational membranes before the end of gestation ( 31 ). Moreover, Tokmak et al. ( 32 ) and Wan et al. ( 33 ) suggested a link between ADAMTS9 and endometriosis. Therefore, deregulated expression of ADAMTS proteoglycanases might influence fetal neuroplasticity by affecting the extracellular matrix structure and function in nervous system cells ( 34 , 35 ). Baron-Cohen et al. ( 36 ) analyzed by liquid chromatography and mass spectrometry analysis amniotic fluid samples from 128 boys with ASD from the Danish Historic Birth Cohort. They found ( 36 ) that the fetal concentrations of sex hormones (especially progesterone and 17-α-hydroxyprogesterone) were elevated compared with matched control and provided the first evidence of increased fetal steroidogenic activity in autism. Using the same cohort and methods, the authors analyzed the prenatal levels of estrogens and progesterone in the amniotic fluid of boys with (n=98) and without autism (n=177). They observed that the prenatal estrogen and progesterone concentrations were higher in amniotic fluid of boys who developed autism. They concluded that prenatal estrogen alterations influence brain development and sexual differentiation and also contribute to the risk of autism. Biochemical and molecular studies using limbic tissues from the offspring of progestin-injected pregnant rats ( 11 ) showed that prenatal progestin exposure induces autism-like behavior in the offspring through estrogen receptor β (ERβ) suppression in the brain, suggesting that progestin may be implicated in ASD development. Specifically, pregnant rats received LNG alone, LNG plus EE, or EE alone from day 1 to day 21 of gestation. Half of the offspring was used for in vivo behavioral studies and the other half for cytochemical, immunological and biochemical studies of the dissected limbic compartment (amygdala, hippocampus and hypothalamus). The authors found that both LNG and LNG/EE significantly increased DNA methylation at the ERβ promoter in amygdala neurons from 10-week-old male and female offspring. These results in an animal model were then confirmed in a large case-control epidemiological study ( Table 2 ) showing the association of prenatal progestin exposure with ASD ( 12 ). The authors investigated a large population in Hainan, an island of 8 million people in southern China. Among 37,863 children (0–6 years of age), they selected 235 children with ASD and 682 matched controls. They observed that the following mother-related factors were strongly associated with ASD prevalence: use of progestin to prevent abortion (15.3%), use of progestin contraceptives at conception time (11%), and prenatal consumption of progestin-contaminated seafood during the first trimester of pregnancy (all mothers). The association with the consumption of contaminated seafood by mothers is particularly striking. Indeed, according to the authors, currently in China, combined oral contraceptives are widely used by the seafood industry for pregnancy prevention in fish and shrimps because they can grow faster and become fatter if they do not lay eggs. This may partly explain the increasing incidence of ASD in China (and in other countries). Therefore, pregnant women must be especially careful and must avoid eating oral contraceptive-contaminated seafood. To confirm these findings, the authors gave to pregnant rats synthetic progestin (norethindrone)-treated zebrafish as food, and found that their offspring showed autism-like behavior. The authors concluded that prenatal progestin exposure may be associated with ASD development. Progestin exposure and ASD: An epidemiological study in a large Chinese population. *, indicates significant differences; #, during the first trimester of pregnancy; df, degrees of freedom (a) Potential impact factors of progestin exposurefor ASD in a Chinese population-based case-control study. The three major exposure sources during pregnancy are indicated with black arrows.©Li et al. ( 12 ). (b) Multivariate regression analysis of the impact factors on ASD prevalence after progestin in-utero exposure. Note the high oddsratio for the three major progestin exposure sources (black arrows). ©Li et al. ( 12 ). With permission. In a large multi-center study on prenatal DES exposure, Titus-Ernstoff et al. ( 14 ) observed somatic defects among 4,029 sons (genitourinary anomalies, risk of infertility and testicular cancer) and 3,808 daughters (infertility, reproductive tract anomalies, pregnancy loss, premature delivery, vaginal clear cell adenocarcinoma, risk of breast and cervical cancer) who were in-utero exposed. Moreover, for the first time, they reported birth defects in the offspring of prenatally exposed and unexposed children. They also reported the first molecular studies in female mice showing that exposure to DES results in epigenetic alterations. In 2011, using data on 529 families from the Hhorages-France cohort, Kalfa et al. ( 37 ) found that the percentage of children with hypospadias was more than doubled in the grandsons compared with the sons of women who took DES during pregnancy (8.2% vs 3.5%), showing DES multigenerational effect on genital malformations. Rivollier et al. ( 26 ) also proposed a possible transgenerational epigenetic effect. These findings confirm the work by Skinner’s group ( 38 ) on the epigenetic transmission of the alterations caused by endocrine disruptors and on the epigenetic transgenerational inheritance of diseases. More recently, in a cohort study on DES-linked neurodevelopmental effects that included 47,540 women spanning three generations, Kioumourtzoglou et al. ( 39 ) reported attention-deficit/hyperactivity disorder (ADHD) in the third generation of children. ADHD etiology involves a combination of genetic and environmental risk factors. The authors concluded that “DES exposure is associated with multigenerational neurodevelopmental disorders”. Similarly, Gaspari et al. ( 40 , 41 ) described DES effects on somatic disorders in grandchildren (i.e. primary clear cell carcinoma of the cervix in an 8-year-old granddaughter and endometriosis in granddaughters and possibly one great-granddaughter). Recently, we described in an informative family of 11 children (the elder not exposed was the control), the impact of DES prescribed to suppress lactation for 3 months on the children born after such treatment and their progeny (genealogic tree of this informative family in Figure 2 ), by focusing on psychiatric disorders (bipolar disorders, suicide attempts and suicide, eating disorders) ( 42 , 43 ). These psychiatric disorders were associated with somatic disorders, such as endometriosis and hypospadias ( Table 3 ). Importantly, in the third generation, 10/19 DES-exposed grandchildren had psychiatric disorders, especially ASD, Asperger syndrome or ASD without Asperger syndrome (in boys) and bipolar disorders for girls, associated with dyspraxia and learning disabilities, mood and behavioral disorders and eating disorders, as well as somatic comorbidities. In the fourth generation (n=7 DES-exposed great-grandchildren, aged between 0 and 18 years), one child had dyspraxia and ASD. In the family of the non-exposed daughter, no psychiatric or somatic disorders were reported ( Figure 2 and Table 3 ). Pedigree of an informative family where the mother (I-2) was treated with DES (30 mg/day) for 3 months after each delivery to inhibit lactation. Only the first child (II-1) was not exposed to DES in utero . Daughter II-1 and her descendants do not have any psychiatric disorder. No history of psychiatric disorders was reported for the maternal and paternal sides. red = bipolar disorder, yellow = psychosis borderline, blue = attempted suicide(s), black = suicide, orange = autism spectrum disorder, light blue = dyspraxia and learning disabilities, light pink = mood and sleep disorders, green = behavioral disorders, fuchsia = eating disorders. © Soyer-Gobillard et al. ( 42 ) IJERPH, 18, 9965, MDPI. Psychiatric disorders (columns 2 and 3) and comorbidities (column 4) in the four generations of an informative family (column 1) whose mother (first generation) took DES for 3 months after each delivery (see Figure 2 ). There are very few studies on the multigenerational (and likely transgenerational) effects of DES in human diseases. It has been hypothesized that ASD originate in utero due to molecular perturbations of the developing brain. Ladd-Acosta et al. ( 44 ) in a pilot post-mortem study found that ZFP57 is more methylated in cerebellar tissue samples of women with autism than unrelated controls. In Sprague-Dawley rats, Zou et al. ( 11 ) detected DNA hypermethylation on the promoter of the gene encoding ERβ in the amygdala of the offspring in-utero exposed to LNG alone or the LNG/EE combination. Interestingly, this effect was less important in female than male rats, as observed also in humans, in agreement with our own recent observations in four generations ( 42 ) ( Figure 2 ; Table 3 ). In a post-mortem study, Corley et al. ( 45 ) showed significant DNA methylation defects in the brain, supporting the hypothesis of an early developmental origin of ASD. In our recent works ( 42 , 46 ), we confirmed the findings by Kiourmoutzoglou et al. ( 39 ) on the multi-generational and very probably trans-generational impact of DES on neurodevelopment. This strengthens the hypothesis that in-utero exposure to DES (and to other endocrine disruptors) contributes to the pathogenesis of psychiatric disorders, particularly autistic disorders ( 47 , 48 ). According to the theory proposed by Nilsson and Skinner ( 49 ), epigenetic alterations (epimutations) induced by environmental insults (e.g. exposure to toxicants) and their transgenerational (germline-mediated) inheritance increase disease susceptibility. Indeed, besides the pregnant mother (F0 generation) who is directly exposed to the insult, also the fetus (F1) and its germline are exposed ( Figure 1 ) ( 50 ). These exposed germ cells will give rise to the F2 generation (grandchildren). Therefore, the first generation without direct environmental exposure is the F3 generation (great-grandchildren). The F3 generation may exhibit transgenerational inheritance of disease susceptibility. Several epimutations can be implicated in this non-genetic inheritance ( 48 ): DNA and RNA methylation changes, histone modification alterations, non-coding RNAs and extracellular vesicles. It has been already shown that upon DES exposure, the ZFP57 and ADAMTS9 genes become hypermethylated. More studies are needed to identify other epimutations linked to in-utero exposure to DES. Concomitant neurodevelopmental impairment and male genital malformations (disorders of sex development, DSD) have been observed in children exposed in utero to DES. Chen et al. ( 51 ) reported that boys with cryptorchidism were more likely to receive a diagnosis of ASD (HR 1.24). Similarly, using nation-wide Israeli healthcare data, Rotem et al. ( 52 ) highlighted a correlation between male genital malformations and ASD (odds ratio of 1.62) ( 52 ). The authors suggested that prenatal androgen defects could be implicated in ASD etiology. We identified several girls and boys with concomitant ASD and DSD in the Hhorages-France cohort (preliminary data) ( Table 3 ). Moreover, Bodur et al. ( 53 ) recently reported that among 432 boys with ASD, 4.4% had a genital abnormality. Similarly, Pham et al. ( 54 ) reviewed data on 447 male children with ASD. They detected genital malformations in 68.5% of them, a much higher rate than previously reported.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

All these results, particularly those reported by Zou et al. ( 11 ) and Li et al. ( 12 ), support the hypothesis formulated by Kim Strifert ( 55 , 56 ) on the link between oral contraceptives and higher ASD risk in children (in 2015, 1 in 68 ( 56 ) vs in 2018, 1 in 44 children ( 57 ) were diagnosed with ASD in the USA. Strifert hypothesized that the estrogen and progestogen used in oral contraceptives act as endocrine disruptors. Specifically, they may affect germ cells/oocytes and give rise to a potent risk factor that helps to explain the recent increase in ASD prevalence. Moreover, such synthetic hormones might cause epigenetic changes. For instance, EE, a known endocrine disruptor, may trigger DNA methylation of the ERβ gene, leading to its downregulation and impaired estrogen signaling in the progeny brain. In 2015, Strifert hypothesized also that the deleterious effects of estrogens and progestogens could be transgenerational. Therefore, in the absence of large-scale epidemiological studies, the principle of precaution must be imperatively retained for the protection of the future generations: pregnant or not, women should be extremely vigilant about any kind of synthetic hormone therapy. In 2004, the great biologist and epidemiologist T.E. Colborn wrote ( 15 ): “The fetus can be protected from endocrine disruptors, whatever they are, only at the zero dose”.

M-OS-G is a researcher, president of the HHORAGES-France Association and a mother concerned with DES and other synthetic hormones. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. HHORAGES-France Association is financed exclusively by subscriptions and donations.

This work could not continue without the daily ongoing support of the Hhorages-France families and board especially Mrs. Mauricette P. and Yette B. Authors warmly thank Mr Simon Ball, Independent Autism/ADHD researcher, for his support and encouragements during the writing of the chapter concerning progestogen damages (9). Authors also thank Elisabetta Eandermarcher for corrections and relevant suggestions in the manuscript as well as Professor Paul Yao for invitation, encouragements and permission to reuse some of his team’s results in Table 2 .

The authors received specific funding for this work from Hhorages-France Association.

M-OS-G. drafting the manuscript; LG, PC and CS: revising the manuscript critically for intellectual content. All authors contributed to the article and approved the submitted version.