Fetal Valproate Spectrum Disorder

Fetal valproate spectrum disorder (FVSD)

Literature and regulatory information updated: 17 October 2026.

0. Initial Reference Table

1. Introduction

1.1. Definition and scope

Fetal valproate spectrum disorder (FVSD) is the set of structural, dysmorphic and neurodevelopmental consequences of prenatal exposure to valproic acid and its salts, which Orphanet defines as a rare teratogenic disorder characterised by a distinctive facial dysmorphism, congenital anomalies and developmental delay, chiefly of language and communication [1]. The term "spectrum" was proposed by the consensus of the European Reference Network for congenital malformations and intellectual disability (ERN-ITHACA) in 2019, by analogy with the fetal alcohol spectrum, because neurodevelopmental involvement occurs without malformations and without facial dysmorphism; cognitive deficits are more frequent than malformations, and the studies that excluded children with a major malformation still show the reduction in intelligence quotient (IQ) and the excess of ASD [12]. Orphanet adopted the designation FVSD for the entity ORPHA:1906, which forms part of the group ORPHA:370068 of the fetal anticonvulsant syndromes (FACS), alongside the embryopathies caused by hydantoin, carbamazepine and trimethadione [1, 2]. The Oxford Desk Reference uses FACS as a term common to the effects of valproate, phenytoin, carbamazepine and phenobarbital [3]. Exposure to other antiseizure medications (ASM) and hydantoin embryopathy enter here only as risk comparators and differential diagnoses. This article also covers the question of paternal exposure to valproate during spermatogenesis, which the European Medicines Agency (EMA) assessed between 2023 and 2026.

1.2. Historical overview

The suspicion that anticonvulsants were teratogenic was put forward by Meadow in 1968, in a letter to the Lancet, with the epidemiological evidence of Fedrick and of Monson and colleagues in 1973, at a time before valproate was marketed [23, 24]; in the Oxford Record Linkage Study, Fedrick found congenital defects in 13.8% of the live births of 168 women with epilepsy (223 infants), against 5.6% in controls (p<0.0005) [25]. Valproate was authorised in Europe in the mid-1970s [12]; in the United States of America (USA), valproic acid (Depakene®) was approved by the Food and Drug Administration (FDA) on 28 February 1978 and semisodium valproate or divalproex sodium (Depakote®) on 10 March 1983 [26]. In 1982, Robert and Guibaud, from the malformation register of the Rhône-Alpes region (Lyon), published in the Lancet the association between maternal valproic acid and neural tube defects [27, 28]; the ERN-ITHACA consensus attributes the first epidemiological alert from the same region to Bjerkedal and colleagues, also in 1982 [12]. In 1984, DiLiberti and colleagues described seven children with a broader malformative and dysmorphic pattern, which they named "fetal valproate syndrome" [12, 29], and Hanson and colleagues reported similar observations in the same year, in a conference abstract that is not indexed in PubMed [28]. In 1988, Ardinger and colleagues verified the phenotype in 19 exposed children and described as features characteristic of valproate the prominent metopic ridge, bifrontal narrowing with deficiency of the outer orbital ridge, tracheomalacia, talipes with an intact spine and lumbosacral meningomyelocele; Smith cites this work with the date 1986, but the article is from 1988 [28, 30]. In 2001, Kozma reviewed 69 cases published between 1978 and 2000 and described two affected siblings, documenting recurrence in successive exposed pregnancies [4]; in the same year, Phiel and colleagues and Göttlicher and colleagues identified histone deacetylase (HDAC) inhibition as a direct action of valproate [31, 32].

1.3. From malformative risk to neurodevelopmental risk

In the 2000s, the prospective pregnancy registers (the international prospective EURAP register, with 42 countries; the North American NAAPR register, North American Antiepileptic Drug Pregnancy Registry, also designated NAAED; and the United Kingdom register) established that valproate has the highest malformation rate among the ASM and a dose-dependent effect [7, 33]. The prospective NEAD study (Neurodevelopmental Effects of Antiepileptic Drugs), in 25 centres in the USA and the United Kingdom, showed a lower IQ at 3 and at 6 years after valproate exposure, and Danish population studies quantified the excess of ASD and of attention deficit hyperactivity disorder (ADHD) [14, 34–36]. This evidence gave rise to successive regulatory measures: the FDA communication of 2013, which contraindicated valproate in migraine prophylaxis in pregnant women [15]; the European referral of 2014, requested by the British Medicines and Healthcare products Regulatory Agency (MHRA) [11]; that of 2018, requested by the French agency ANSM (Agence nationale de sécurité du médicament), which created the PPP [18]; the British measures of 2023–2025, which require two specialists to start valproate before 55 years of age [37]; and the precautionary measures of 2024 for treated men, maintained by the Pharmacovigilance Risk Assessment Committee (PRAC) in June 2026 despite subsequent evidence being inconsistent [38, 39].

2. Molecular Genetics and Pathophysiology

2.1. Absence of a causal gene

FVSD is a teratogenic, not a monogenic, entity: there is no causal gene, no mendelian inheritance pattern and no variants to be standardised according to Human Genome Variation Society (HGVS) nomenclature [1]. Genetics intervenes in three ways. First, the risk in a given pregnancy depends on the drug, on maternal and fetal metabolism and on a genetic susceptibility that the sources describe as probable but do not characterise [3]. OMIM classifies the entity as susceptibility to valproate embryopathy, with no associated gene, and supports that hypothesis with two observations: Malm and colleagues (2002) described three families in which all exposed children were affected (three in one sibship and two in each of the other two), and Faiella and colleagues (2000) showed, in the mouse, that parental factors condition fetal susceptibility to valproate-induced malformations, many of them homeotic transformations, and that valproate alters the expression of HOX genes in pluripotent human embryonal carcinoma cells [5]. Second, maternal epilepsy may have a genetic cause of its own, which the ERN-ITHACA consensus recommends considering and investigating, because it conditions the risk of seizures in the child and the differential diagnosis of developmental delay [12]. Third, the genetic assessment of the exposed child must exclude monogenic and chromosomal causes of delay and dysmorphism, including microdeletions and microduplications that predispose to epilepsy and that may have been inherited from the mother (15q11.2, 15q13.3, 16p13.11) [3]. Mitochondrial disease due to POLG variants is a contraindication to valproate because of the risk of hepatic failure, but it bears no relation to FVSD [40].

2.2. Histone deacetylase inhibition

Valproate was for years regarded as a folate antagonist, but it is now known as a potent inhibitor of histone deacetylases (HDAC), the mechanism proposed for neural tube defects and for the other defects [41]. Phiel and colleagues showed that valproic acid activates Wnt-dependent gene expression by a pathway distinct from that of lithium, through direct HDAC inhibition (50% inhibitory concentration for HDAC1 of 0.4 mM); at therapeutic levels it mimics trichostatin A in histone hyperacetylation, valproate and trichostatin A have similar teratogenic effects in vertebrate embryos, and non-teratogenic analogues of valproate neither inhibit HDAC nor activate transcription [31]. Göttlicher and colleagues confirmed that valproate probably binds to the catalytic centre of the enzyme, relieves HDAC-dependent transcriptional repression and causes histone hyperacetylation in vitro and in vivo [32]. Oxford attributes the HDAC mechanism to Vassar and Fuchs (1991), but the article listed in its bibliography deals with transforming growth factor alpha (TGF-alpha) in the epidermis of transgenic mice and mentions neither valproate nor HDAC; the primary sources are those of 2001 [31, 32, 41]. The dissociation between teratogenicity and folate antagonism explains why folic acid supplementation has no demonstrated efficacy in preventing valproate-induced malformations [40, 42].

2.3. Dose relationship and windows of exposure

Malformations are dose-related in the first trimester: in EURAP, the rate of MCM increased with the dose at conception for valproate (p<0.0001), and even doses ≤650 mg/day carried a higher risk than levetiracetam (odds ratio, OR, 2.43; 95% confidence interval, CI, 1.30–4.55) [7]. In the Australian register, the mean valproate dose was 2081 mg in pregnancies with congenital defects and 1149 mg in the remainder (p<0.0001) [3, 43]. The EURAP dose-response analysis of 2011 (42 countries; dose at conception; MCM assessed up to 12 months; 1010 pregnancies exposed to valproic acid, 1402 to carbamazepine, 1280 to lamotrigine and 217 to phenobarbital) showed an increase in malformations with dose for all four drugs; the lowest rates occurred with lamotrigine <300 mg/day (2.0%; 95% CI 1.19–3.24) and carbamazepine <400 mg/day (3.4%; 1.11–7.71), and valproate and phenobarbital carried a significantly greater risk at all doses than lamotrigine <300 mg/day; a parental history of MCM was associated with an OR of 4.4 (2.06–9.23) [44]. Figure 6.2 of Oxford reproduces this analysis, with the valproate strata <700 mg/day (n=431), 700–1499 mg/day (n=480) and ≥1500 mg/day (n=99) and rising rates of about 6%, 11% and 25% on reading the graph, which are not values published in the abstract; the book's legend attributes the figure to a 2006 EURAP article that deals with seizure control in pregnancy and not with malformations [3, 44, 45]. The ERN-ITHACA consensus reports a malformation risk of about 11%, reaching 24% above 1500 mg/day [12]. The 2019 task force of the International League Against Epilepsy (ILAE) summarises that all registers show an increasing risk with valproate dose, with higher-risk thresholds of 500 mg/day in NAAPR, 600 mg/day in the UK and Ireland register and 650 mg/day in EURAP, and that, in EURAP, low-dose valproate combined with another drug was accompanied by fewer MCM than high-dose valproate in monotherapy [9]. In the South Korean national cohort of 2013–2021, valproate monotherapy in the first trimester was associated with an OR of 1.46 (1.11–1.91) for congenital anomalies and, from 500 mg/day upwards, with an OR of 1.57 (1.12–2.19) [46]. The FDA concludes that no threshold dose below which risk is absent can be established, and the European SmPC text says the same [40, 47]. The windows differ by outcome: neural tube defects result from exposure in the first weeks, because primary neurulation takes place between days 18 and 26 after fertilisation, open spina bifida results from failure of closure of the caudal neural folds on days 24–26, and the neural tube is normally closed by day 28 [41, 48]; limb reduction defects result from exposure between 5 and 11 weeks of gestation [49], whereas the effect on IQ is not attributed to a specific period, because in NEAD exposure continued throughout pregnancy; the FDA considers that exposure at any time in pregnancy may reduce IQ, and the European SmPC states that the period of risk is uncertain and that a risk throughout pregnancy cannot be excluded [15, 40, 47]. In the French cohort of the ANSM and the National Health Insurance Fund (CNAM), the risk of mental and behavioural disorders was greater with prolonged exposure (hazard ratio, HR, 5.5; 4.0–7.4) and with exposure confined to the 2nd or 3rd trimester (HR 3.0; 1.3–6.6) than with exposure confined to the 1st trimester (HR 1.0; 0.4–2.6), a result that does not exclude a risk with exposure limited to the 1st trimester [50]. Oxford stresses that severity is unpredictable even with the same exposure, because it depends on the pharmacometabolism of the fetoplacental complex, in part genetically determined: a woman may remain on the same valproate dose in three pregnancies and have only one affected child [51]. In NEAD, the valproate dose correlated negatively with IQ at 6 years (r=−0.56; p<0.0001), with verbal and non-verbal abilities, memory and executive function, with no correlation for the other ASM [14]. High daily doses late in pregnancy increase the likelihood of neonatal withdrawal symptoms [3]. The French PNDS of 2017 summarises the windows in weeks of gestation: exposure confined to the embryonic phase (before 12 weeks) is associated with dysmorphism and malformations, with a variable intellectual effect; exposure beginning only in the fetal phase (after 12 weeks) causes no malformation and little or no dysmorphism, but retains the psycho-intellectual risk [22]. The same protocol relates the severity of the phenotype to the number of antiseizure medications combined with valproate, to the dose and to the period of exposure, and recalls that the dose relationship is not linear in the individual: effects occur at low doses and children without anomalies occur after high doses [22].

2.4. Tissue mechanisms

Valproate-induced neural tube defects predominantly affect the lower lumbar and sacral regions [41]. Valproic acid acts on limb morphogenesis, as do thalidomide, warfarin and phenytoin, by a mechanism distinct from vascular disruption (misoprostol, chorionic villus sampling) [49]. In animal models (mouse, rat, rabbit, monkey) valproate causes skeletal, cardiac and urogenital malformations and, in the mouse, neural tube defects correlated with peak maternal concentrations, as well as cognitive, locomotor and social deficits in the offspring [40]; occult neural tube defects have been described in animal models [12]. The hypothesis of IKr current blockade (hERG channel) with embryonic arrhythmia and hypoxic injury applies to phenytoin, phenobarbital and carbamazepine, and not to valproate, which does not figure among the IKr blockers studied by Danielsson and colleagues [23, 52].

2.5. Epigenetic signature

A consortium from France, the United Kingdom, Spain and Canada identified a DNA methylation episignature in the peripheral blood of individuals with prenatal valproate exposure, present in most of those affected, with a classifier that distinguishes exposed individuals from controls and from a wide range of syndromes with a known episignature, and with enrichment in cell adhesion genes of the cadherin superfamily; it is the first peripheral blood epigenetic biomarker for a prenatal teratogenic disorder [17]. The abstract gives neither the number of individuals nor sensitivity and specificity values, and one of the authors declares being a shareholder in the company that markets the EpiSign test [17].

2.6. Paternal exposure: plausibility and confounding

For paternal exposure, the biological hypothesis is an effect on spermatogenesis, and studies define the window of exposure as valproate use in the 3 months before conception [38, 53]. Valproate causes male infertility, at times reversible after discontinuation or dose reduction, and testicular toxicity in animals [37, 40]. The Norwegian MoBa cohort showed that fathers treated with valproate had higher polygenic risk scores for epilepsy than those treated with lamotrigine or levetiracetam (mean difference 0.66; 95% CI 0.21–1.11), and that those scores share 428 genes in neurodevelopmental pathways with ADHD and ASD, which supports genetic confounding by indication in the association observed in some studies [54].

3. Clinical Spectrum and Manifestations by System

3.1. Overview

FVSD has no treatment that modifies established injury, so the natural history described corresponds to the course of the exposure and to symptomatic interventions (surgery for malformations, educational support, therapies). The spectrum ranges from children with multiple malformations and evident dysmorphism to children without malformation and without dysmorphism, whose only effect is cognitive or behavioural; cognitive and neuropsychological deficits frequently occur without the physical features [12, 28]. In Kozma's review of 69 cases exposed to valproate alone (1978–2000), a clinical series biased towards severe cases, musculoskeletal anomalies arose in 62%, minor skin defects in 30%, cardiovascular defects in 26%, genital defects in 22% and pulmonary defects in 16%, with neural tube defects in 3%; 12% died in childhood and 29% of survivors had developmental delay or intellectual disability, with growth restriction in 15% and overgrowth in 9% [4]. In the series of Ardinger and colleagues (19 exposed children), growth was not consistently altered in monotherapy, and developmental delay or neurological abnormality affected 71% of children on monotherapy and 90% of those exposed to polytherapy [30]. In the population-based retrospective study from Grampian (Scotland), with 293 children of 149 mothers treated with ASM between 1976 and 2000 and with all drugs pooled, exposed children had neonatal withdrawal in 20%, malformations in 14% (5% in unexposed siblings), developmental delay in 24% (11%), facial dysmorphism in 52% (25%), medical problems in childhood in 31% (13%) and behavioural disorders in 20% (5%) [55]. The French PNDS (French National Diagnosis and Care Protocol) summary for the attending physician groups the consequences of exposure into five types: disturbances of fetal growth (growth restriction or macrosomia) and a withdrawal syndrome in the first 12–48 hours; malformations in about 10% of children; dysmorphism almost always present, of variable intensity and not always persistent in the adult; neurodevelopmental and learning disorders in 30–40%; and progressive organic disease, such as inguinal and umbilical hernias, tracheomalacia, joint hyperlaxity, very frequent scoliosis, visual and hearing deficits and frequent ear, nose and throat infections in early childhood [22]. The protocol stresses the dose effect and the potentiation produced by combining several antiseizure medications [22].

3.2. Major congenital malformations: frequency

Prospective registers are the best source for absolute risk. The table brings together the series with a known denominator; the figures are not directly comparable because they differ in period, age at assessment (5 days to 1 year), definition of malformation and dose.

The FDA summarises that the malformation rate with valproate is about four times that observed with other ASM in monotherapy and, in the NAAED register, 9–11% at a mean dose of 1000 mg/day, up to five times that of other ASM; polytherapy including valproate increases the risk further [40]. The background risk in the USA is 2–4% for major malformations [40], and 2.4–2.9% in the synthesis of the guideline of the American Academy of Neurology (AAN), the American Epilepsy Society (AES) and the Society for Maternal-Fetal Medicine (SMFM) [42]. The network meta-analysis cited by the ILAE gives valproate an OR for MCM of 2.93 (2.36–3.69) against unexposed pregnancies, the highest of the drugs compared (topiramate 1.90; phenobarbital 1.83; phenytoin 1.67; carbamazepine 1.37; lamotrigine 0.96; levetiracetam 0.72), and the 2016 Cochrane review an RR of 5.69 (3.33–9.73) against children of women without epilepsy [9]. In EURAP, the overall prevalence of MCM fell by 27% over 14 years, in parallel with the reduced use of valproate and carbamazepine [9]. The Korean estimate, obtained with administrative codes, is lower than those of the prospective registers; in another Korean analysis, continuous valproate exposure in dual therapy was associated with an adjusted RR of 3.40 (1.36–8.48) for malformations and of 5.50 (1.29–23.51) for cardiac malformations, with a greater risk with prolonged exposure than with exposure limited to the 1st trimester [46, 57]. The French PNDS of 2017 places the overall risk of MCM between 5 and 11%, 2 to 7 times that associated with another antiseizure medication (carbamazepine, lamotrigine, phenytoin, phenobarbital) and 3 to 4 times that of the general population, and gives rates of 5.6% below 700 mg/day, 10.4% between 700 and 1500 mg/day and 24.2% above 1500 mg/day [22]. The dose limits are those of the three EURAP strata of 2011 described in section 2.3, and the rates are compatible with the graphical reading of Figure 6.2 of Oxford [3, 44].

3.3. Specific malformations

The EUROCAT study by Jentink and colleagues combined a meta-analysis of 8 cohorts (1565 exposed pregnancies, 118 MCM), which identified 14 candidate malformations, with a case-control study in European population registers that included 180 records of exposure to valproate in monotherapy [13].

The neural tube defects of valproate are predominantly lumbosacral [41]. The frequency of neural tube defects differs according to denominator and definition: about 5% in Oxford, 1–2% for spina bifida in the FDA label and in Yerby's review, 1.4% in the AAN synthesis and 3% in Kozma's series [3, 4, 40, 42, 58]. Oxford reports a relative increase of about tenfold; the source it cites gives the prevalences (1–2% with valproate, 0.5% with carbamazepine) and not that relative risk [41, 58]. The early literature associated valproate with cleft lip [61], but the ERN-ITHACA consensus, on the basis of the EUROCAT data, concludes that cleft lip is not increased and that the risk is concentrated in isolated cleft palate [3, 12]. The cardiac defects described range from septal defects (ostium secundum atrial septal defect, perimembranous ventricular septal defect) to coarctation of the aorta, hypoplastic left heart, aortic valve stenosis, interrupted aortic arch and pulmonary atresia with intact ventricular septum [28]. For the anticonvulsant class as a whole, Oxford estimates a risk of congenital heart disease of 1.8%, three times the background figure [62]. In the French cohort of 2007–2014 (14,322 exposed pregnant women and 8701 exposed live births), valproate used for epilepsy was significantly associated with spina bifida, atrial or ventricular septal defect, hypoplasia of the left ventricle, pulmonary atresia, cleft palate, anorectal atresia, hypospadias and preaxial polydactyly [59]. In the pooled analysis of 32 cohorts cited by the ILAE, valproate showed a high prevalence of neural tube defects and hypospadias, with no specific association for lamotrigine, levetiracetam, carbamazepine, oxcarbazepine or gabapentin [9]. The French PNDS of 2017 takes up the odds ratios of the EUROCAT study (hypospadias 4.8; craniosynostosis, chiefly trigonocephaly, 6.8; cleft palate 5.2; polydactyly 2.2) and gives for spina bifida a risk of 1–3%, about 30 times that of the general population, with anencephaly exceptional [22] [13]. The risk of congenital heart disease is 2 to 3 times that of the general population, with a predominance of ventricular and atrial septal defects, aortic and pulmonary stenoses and patent ductus arteriosus [22]. In the limbs, the protocol includes a short or absent radius, absent or triphalangeal thumb, preaxial polydactyly and split hand; it further reports pulmonary, laryngeal (hypoplasia), cerebral and renal malformations, and rare ocular malformations (iris or retinal colobomas, microphthalmia, congenital cataract); the summary for the attending physician adds omphalocele [22]. The PNDS warns that some malformations escape prenatal and neonatal diagnosis [22]. The three estimates of the relative risk of spina bifida (about 30 times in the PNDS, about 20 in the ERN-ITHACA consensus and OR 12.7 in EUROCAT) reflect different denominators and comparators [22] [12, 13].

3.4. Craniofacial dysmorphism

The face is the most recognisable feature and is more marked with valproate than with the other ASM [3]. It includes a narrow bifrontal diameter with a metopic ridge and, at times, trigonocephaly; a high forehead; epicanthic folds continuing into an infraorbital groove; hypertelorism or telecanthus; a broad, low nasal root and a short nose with anteverted nares; midface hypoplasia; a long, smooth philtrum; a thin upper lip; a small mouth with downturned corners and an everted lower lip; micrognathia; and arched eyebrows [3, 12, 28]. Kozma's series adds small ears and a long upper lip [4], and Ardinger's the deficiency of the outer orbital ridge with bifrontal narrowing [30]. The features vary with age and become attenuated in the young adult, although they remain recognisable, so the consensus recommends reviewing photographs from younger ages [12]. Figure 2 of the ERN-ITHACA consensus shows three exposed individuals of increasing age (toddler, school-age child, young adult woman) with anteverted nares, a small mouth, a thin upper lip, an everted lower lip, a flattened philtrum and prominence of the frontal midline, less marked in the adult (https://pmc.ncbi.nlm.nih.gov/articles/PMC6642533/figure/Fig2/) [12]. Antiseizure medication taken by the child herself may coarsen the features and cause gingival hypertrophy, a postnatal effect that must not be confused with the dysmorphism of FVSD [63, 64]. The French PNDS adds to the list eyebrows that are arched and laterally displaced, palpebral fissures at times downslanting, nares that are everted, anteverted or horizontal, an effaced columella, a flat and broad or at times narrow nasal bridge, a drooping lower lip, a horizontal mouth, microretrognathia and dysplastic, often low-set ears [22]. It distinguishes two evolving forms: a moderate dysmorphism, in which affected individuals resemble one another more than they differ from the general population (homomorphy) and retain resemblances to their parents, with "disappearance of the morphotype" and finer, non-discriminating features as growth proceeds; and a marked dysmorphism, which persists or worsens and shapes the morphology of the older child and the adult [22]. Dysmorphism is best recognised between 1 and 5 years, varies within the same sibship exposed to equal doses and between twins, appears more evident in young girls, and value should be placed on the departure from the parental morphotype (the children resemble their parents less than they resemble other affected children) [22]. The face associated with carbamazepine or phenytoin has finer and more vertical features, that of fetal alcohol spectrum disorder has overlapping aspects, and the protocol accepts the use of facial analysis software and recommends the opinion of a dysmorphologist [22]. Appendix 3 of the PNDS includes a clinical form for suspected in utero exposure, which begins by requesting childhood photographs and lists the signs to look for, from low birth weight (<3rd centile) to craniofacial, ocular, cardiac, laryngeal, genitourinary and limb anomalies [22].

3.5. Limbs and musculoskeletal system

In Kozma's review, according to Oxford, contractures of small joints with long overlapping fingers occurred in 36%, foot deformity in 30%, thumb anomalies in 17% and radial defects in 16%; the article's abstract gives only the overall proportion of musculoskeletal anomalies (62%) [3, 4]. Long, slender fingers and toes, hyperconvex nails, nail hypoplasia, talipes equinovarus, postaxial polydactyly, triphalangeal thumbs, preaxial defects of the feet and radial ray defects ranging from a flat thenar eminence to radial aplasia are described [12, 28]. Oxford attributes radial hypoplasia to carbamazepine and alcohol, through a vasodilator action, which makes the radial defect a shared rather than a valproate-exclusive finding [65]. The combination of craniosynostosis with radial aplasia or hypoplasia reproduces the phenotype of Baller-Gerold syndrome [66]: Santos de Oliveira and colleagues described three valproate-exposed patients with metopic synostosis and upper limb malformations compatible with that phenotype, treated by frontocranial reconstruction, and, reviewing 32 published patients with a diagnosis of Baller-Gerold syndrome, concluded that the phenotype is heterogeneous [67]. Joint laxity is frequent: Moore and colleagues, in a sample recruited through a support group, found hypermobility in 24 of 34 children exposed to valproate in monotherapy and in 10 of 12 exposed to polytherapy, with hernia in 4 and frequent hypotonia; hypermobility translates into leg pain, a need for orthoses, difficulty with stairs and with handwriting, and fatigue [12]. Figure 3 of the consensus shows hypoplastic, overlapping toes and flattening of the plantar arches from laxity (https://pmc.ncbi.nlm.nih.gov/articles/PMC6642533/figure/Fig3/) [12]. Talipes is often postural, and overlapping or hypoplastic toes do not usually require treatment [12]. Beyond malformations, the French PNDS describes, in varying combinations, brachydactyly that may or may not be harmonious, shortening of the 4th and 5th metacarpals or metatarsals, arachnodactyly, usually moderate syndactyly, a thumb that is digitalised in the plane of the hand or hypoplastic, hyperlax fingers, moderate nail hypoplasia, abnormal palmar creases, clinodactyly and flat feet, to be documented with a podoscope [22]. In the axial skeleton and the large joints it adds pectus excavatum or carinatum, a narrow chest, scoliosis and kyphosis, pelvic tilt, hip dysplasia, genu valgum and hyperlaxity of the wrists, hands and large joints; the summary for the attending physician classes scoliosis as very frequent [22].

3.6. Eyes, hearing and airway

Glover and colleagues studied 46 children with confirmed fetal anticonvulsant syndrome (37 exposed to valproate, 29 in monotherapy): 67% had ocular anomalies, 41% refractive errors, 24% astigmatism, 20% strabismus and 11% anisometropia, with myopia in 14 of 28 (50%) of those exposed to valproate in monotherapy and one case of bilateral congenital cataract; the authors recommended preschool vision screening of all children exposed to ASM [68]. The ERN-ITHACA consensus, citing the same group, reports myopia greater than −4 dioptres in 28.6% and stresses that myopia often goes unnoticed [12]. Smith and Oxford further describe iris and chorioretinal coloboma, microphthalmia, iris defects, cataract, corneal opacities, nystagmus, esotropia, lacrimal duct anomalies and, more rarely, septo-optic dysplasia and optic nerve hypoplasia; coloboma, which Oxford links to Glover's work, is not mentioned in that article's abstract [3, 28, 68–70]. In an analysis of the VigiBase pharmacovigilance database, with 676 reports of congenital ocular anomalies after in utero exposure to medicines, valproic acid figured among the suspect drugs, in a descriptive design that does not establish causality [71]. Serous otitis media was reported in 33% in Oxford and in 44% of valproate-exposed children in a support group sample, but the prospective Liverpool and Manchester cohort found no difference in otitis or hearing, so the consensus does not regard it as specific; hyperacusis is reported by parents without formal study, and hearing loss and deafness have been described [3, 12, 40]. Laryngomalacia, tracheomalacia and stridor are increased, and laryngeal hypoplasia and subglottic stenosis have been described [12, 28, 30]. The French PNDS describes recurrent ear, nose and throat infections, with frequent recourse to tympanostomy tubes, tonsillectomy and adenoidectomy, hearing disturbances and strabismus, and recommends that an iris coloboma should prompt consideration of other diagnoses according to the associated signs (cat eye syndrome, CHARGE syndrome) [22].

3.7. Genitourinary system and other anomalies

Besides hypospadias, renal hypoplasia, an abnormal collecting system, hydronephrosis, poor bladder control and enuresis are described [12, 28]. In the Liverpool and Manchester cohort, functional bladder problems at 6 years were of similar frequency in exposed children (12/196; 6.1%) and in controls (14/256; 5.4%) [12]. Other anomalies described include inguinal and umbilical hernias, omphalocele, supernumerary nipples, haemangiomas, pigmentary anomalies, scalp defects, a broad chest, bifid rib, pulmonary hypoplasia, cerebral atrophy and a cavum septi pellucidi cyst [12, 28]. The European SmPC adds, in the treated woman, amenorrhoea, polycystic ovaries and raised testosterone, and in men a possible impairment of fertility, reversible after discontinuation according to case reports [47]. The French PNDS extends the list to other organs: supporting tissues (linea alba, inguinal and umbilical hernias, bursitis); growth (short stature, but also tall stature, macrocephaly and overweight); skin (aplasia cutis of the vertex, haemangiomas, café-au-lait or achromic patches, striae, naevi); the renourinary tract (enuresis, vesicoureteric reflux and infections); and, among others, encopresis and epilepsy [22].

3.8. Neonatal period

Most exposed newborns have a normal neonatal period [12]. High doses late in pregnancy are associated with withdrawal, with irritability, tremor, poor feeding and neonatal seizures [3, 72]. The data are divergent: Thisted and colleagues described withdrawal or hypoglycaemia in 13 of 22 exposed infants, but a prospective study of 227 women with epilepsy and 315 controls found no difference in neonatal problems or in admission to a neonatal intensive care unit, and a Norwegian study with 215 exposed infants found no increase in hypoglycaemia [12]. NEAD found no difference in Apgar score, but two larger studies showed almost twice the risk of a low Apgar score [12]. The FDA describes neonatal hypoglycaemia, fatal hepatic failure in infants exposed in utero and, in the pregnant woman, thrombocytopenia and hypofibrinogenaemia with a decrease in clotting factors, which may cause neonatal haemorrhage, including death [40]. The European SmPC describes in the newborn a very rare haemorrhagic syndrome (thrombocytopenia, hypofibrinogenaemia or a decrease in other factors; afibrinogenaemia may be fatal), distinct from the deficiency of vitamin K-dependent factors caused by enzyme inducers, and recommends investigating in the newborn the platelet count, fibrinogen, coagulation tests and factor levels; it further describes neonatal hypoglycaemia after exposure in the 3rd trimester, neonatal hypothyroidism and a withdrawal syndrome with exposure in the last trimester (agitation, irritability, hyperexcitability, tremor, hyperkinesia, changes in tone, seizures and feeding difficulties) [47]. Valproate is not an enzyme inducer, so haemorrhagic disease of the newborn from vitamin K deficiency is not increased unless inducers are used concomitantly [12]. Birth weight does not differ in the Liverpool and Manchester cohort [12], but the AAN/AES/SMFM guideline considers an excess of newborns small for gestational age with valproate or topiramate possible [42]. In EURAP, with 15,893 exposed singleton live births between 1999 and 2023, valproic acid (n=1829) was associated with a birth weight centile 2.54 points lower than that of lamotrigine (95% CI −4.50 to −0.58), an effect smaller than that of topiramate (−11.93) or phenobarbital (−8.08); the combination of lamotrigine and valproate (n=258) did not differ from lamotrigine in monotherapy [73]. In Scotland, in more than 900,000 pregnancies conceived between 2010 and 2023 with matched comparators, valproate was associated with pregnancy loss (adjusted OR 1.92; 1.50–2.47), with congenital conditions (1.85; 1.06–3.21) and with developmental concerns (1.43; 1.01–2.03) [74]. The leaflet of the UK Teratology Information Service (UKTIS) reports a study in which doses above 750 mg/day may increase spontaneous miscarriage, against a background risk of 1 in 5 pregnancies [75]. The French PNDS places withdrawal in the first 12–48 hours, with irritability, hypotonia, seizures and feeding difficulties, and relates it to the doses taken in the 3rd trimester [22]. It attributes neonatal hypoglycaemia to two mechanisms (reduced endogenous glycogen production and glycolytic dysfunction) and reports rare cases of neonatal hyperammonaemia in children of mothers treated with high doses, with no other cause found [22]. In women with epilepsy, intrauterine growth restriction and prematurity are more frequent, the Apgar score is on the whole lower, and good adaptation to extrauterine life occurs in 92–96% of newborns, against 98% in the general population [22]. Epilepsy in the child herself is not part of the picture of the embryofetopathy and should prompt a search for familial epilepsy of genetic cause [22].

3.9. Neurodevelopment

The ERN-ITHACA consensus estimates developmental delay in about 30% of exposed children, with a greater risk when physical signs of embryopathy are present, but also present without malformations and dependent on dose [12]. The EMA based the 2014 measures on developmental problems in up to 30–40% of children of preschool age (delayed walking and speech, deficits in memory, speech and language, lower intellectual ability) [11]; the resulting SmPC adds an IQ at 6 years on average 7–10 points lower than that of children exposed to other ASM, a risk of intellectual disability possibly independent of maternal IQ, ASD about 3 times and childhood autism about 5 times more frequent than in the study populations, and limited data in favour of more ADHD symptoms [47]. The deficits change with age: in early childhood, cognitive, speech and motor delay predominate; at school age they affect IQ, language, memory, attention and executive function [28]. The discrepancy between verbal IQ, which is lower, and non-verbal IQ is the most characteristic cognitive feature, with a deficit of auditory working memory [12, 28].

The adjusted estimates collected by the FDA vary between studies: ASD from 1.6 (95% CI 0.98–2.5) to 6.4 (3.5–11.5), intellectual disability from 1.6 (0.9–2.9) to 9.6 (3.5–26.2) and ADHD from 0.7 (0.4–1.1) to 2.2 (1.3–3.5), with limitations of confounding by indication, residual confounding, short follow-up and exposure inferred from prescriptions [40]. The difference between the aHR for ASD in the general population and that observed in children of mothers with epilepsy, in the Danish study, illustrates the weight of the indication [35]. In the French cohort, the risk increased with the mean daily dose, with HRs of 2.0 (1.0–4.0), 2.2 (1.2–4.2) and 6.5 (4.6–9.2) in the tertiles ≤730, 731–1100 and >1100 mg/day, and with the cumulative dose (up to HR 7.6; 5.4–10.8); lamotrigine carried a risk about 3 times lower than valproate, and its association (HR 1.6) disappeared in children of mothers without psychiatric disease [50]. In the Cochrane review by Bromley and colleagues (22 prospective cohorts and 6 register studies, search to May 2014), the IQ of children exposed to valproate was lower than that of children of women without epilepsy (mean difference −8.94; −11.96 to −5.92; 76 against 552 children) and than that of children of women with untreated epilepsy (−8.17; −12.80 to −3.55), and lower than that of children exposed to carbamazepine (8.69 points; 5.51–11.87), to lamotrigine (10.80; 7.17–14.42) and to phenytoin (9.25; 4.78–13.72); the IQ of those exposed to carbamazepine did not differ from that of controls (−0.03), and six studies showed a dose effect, with worse results from 800–1000 mg/day upwards [84]. In 2020, Cochrane added an editorial note to this review for non-compliance with its conflict of interest policy, with an update planned [84]. According to the ERN-ITHACA consensus, 20–30% of exposed children fall below average [3, 12]; the ILAE, at school age, reports an IQ reduction of 7–11 points, 20–40% of children below average and a need for educational support in 29–48%, dose-dependent, reaching 74% in children with clinically diagnosed embryopathy, and indicates that doses below 400 mg/day have already been associated with lower verbal IQ and a greater need for educational support [9]. The prevalence of ASD was 6.3% in a prospective Manchester study and 8.9% in a Scottish population study, and autistic traits reached 7.7% in monotherapy and 46.7% in polytherapy in a small group; many children who do not meet ASD criteria have social communication difficulties [12]. Special educational needs are increased 3.4-fold, and about eightfold with doses above 800 mg/day; in Denmark, exposed children had lower marks in Danish and mathematics [12]. A meta-analysis of 17 studies (854,142 participants) estimated for valproate an RR for educational difficulties of 2.38 (1.25–4.53), against 1.3 (1.01–1.69) for any ASM [85]. Lower motor and adaptive functioning, poorer social skills and behavioural problems are also described [28, 76]. Polytherapy including valproate is associated with lower cognitive abilities than monotherapy [28]. NEAD found right-handedness less frequent in those exposed to valproate (79%) than in the normative sample (93%) [14]; in the North American MONEAD study, with little valproate exposure, there was no overall difference in handedness at 6 years between 351 children of women with epilepsy and 105 of healthy women [86]. The results from Taiwan and Australia diverge from those of NEAD, the Nordic cohorts and the French cohort: the sibling design has low power and is sensitive to differences in exposure between siblings, and the Australian sample of valproate-exposed children is small [82, 83]. The French PNDS of 2017 estimates disorders of intellectual development, learning and behaviour in 30–40% of exposed children and adults, probably underestimated, with a developmental quotient or IQ on average 10 points lower, a dose effect not observed with carbamazepine, phenytoin or topiramate, a greater risk from 800 mg/day upwards and identical results when the mother was treated for a condition other than epilepsy [22]. It describes two profiles: in severe cases, global delay of acquisitions; in less severe cases, an IQ in the low normal range (70–90), with verbal skills below the remainder, difficulties of comprehension and expression and, in some children, a profile close to dysphasia; gross motor function and attention are also frequently affected [22]. Associated malformations are accompanied by a lower IQ, and growth restriction and early hypoglycaemia are risk cofactors; early correction of joint hyperlaxity and visual deficits reduces their impact on development [22]. ADHD is frequently described, but the efficacy of drug treatment is not established in this disorder [22]. As to autism, the PNDS states that it has no epidemiological data, although the 2014 EMA communication reproduced in its appendix 5 reports a risk of ASD about 3 times and of childhood autism about 5 times that of the general population [22] [11, 35]. The symptoms that allow diagnosis appear at around 30 months, but parents notice earlier signs of altered maturation, between 12 and 18 months: loss of previously regular sleep, stagnation of vocabulary after the first words, any regression, difficulties of interaction and sensory intolerances; the protocol recommends starting intervention without waiting for confirmation, which is made in an autism resource centre [22]. Anxiety, depression, angry outbursts and minor difficulties of social integration are frequent and are often associated with language disorders, but have a good prognosis; neuropsychological assessment should be repeated from childhood to adulthood [22].

3.10. Growth, general health and adulthood

Growth retardation and microcephaly are uncommon, although microcephaly may occur with some polytherapies [3]. In NEAD, microcephaly (below the 3rd centile) was more frequent with carbamazepine and valproate at birth and normalised by 2 years, and in Swedish data head circumference was 0.2 standard deviations smaller with valproate than with lamotrigine [9]. In the Liverpool and Manchester cohort (57 exposed to valproate in monotherapy and 283 controls), there was no difference in height, weight or head circumference at 6 years, but medical consultations for asthma, eczema, respiratory infections and otitis were more frequent in exposed children (35% in monotherapy, 46% in polytherapy, 23% in controls); the consensus group observed weight gain from puberty onwards and reports of precocious puberty that have not been studied [12]. In a cross-sectional survey of the European ConcePTION project among carers in the United Kingdom, Ireland, New Zealand, the USA and Australia, people exposed to valproate in monotherapy (n=88) had physical health problems in a median of 6 of 12 systems with doses ≥1000 mg/day and of 5 with doses <1000 mg/day, against 1 in unexposed people (n=18), with worse health, pain and limitation of activities (p<0.001) that persisted into adulthood [87]. In adulthood, Bromley and colleagues assessed 18 individuals over 16 years of age with a confirmed diagnosis and found an excess of intellectual disability (IQ<70), poor verbal comprehension and reasoning and impairment of auditory working memory and processing speed; the deficits persist and affect independence, employment, mental health and relationships, with no known adult-specific medical disease, and parents identify the transition from 16 to 20 years as the most difficult period [12]. In a qualitative study with 13 young adults with FVSD (21–37 years) and 17 parents, participants described lifelong physical, mental and developmental challenges, difficulties in employment, independent living and relationships, and limited or absent adult services [88].

3.11. Population dimension in France

The ANSM estimated in April 2017, according to the journal Prescrire, that between 1967 and 2016 some 41,200–75,300 children exposed to valproate or its derivatives were born in France and that 2150–4100 had at least one major congenital malformation (1900–3800 children of mothers with epilepsy, with a risk about 4 times the background figure, and 250–300 of mothers with bipolar disorder, with a risk about 2 times); the estimate excludes fetal losses, terminations of pregnancy and neurodevelopmental disorders without malformation [59]. In the French cohort of 2007–2014, exposed pregnancies per year fell from 2316 to 1333; the indication was epilepsy in 57% and bipolar disorder in 43%; 85% of exposures occurred in the first two months, with a mean duration of 117 days in epilepsy and 44 days in bipolar disorder [59]. The estimate of 16,600–30,400 children with neurodevelopmental disorders between 1967 and 2016 was publicised by the specialist press in June 2018 and does not appear in the technical report of the ANSM and the CNAM of that month, which presents relative risks and not absolute numbers; it is a modelled estimate and not a count of cases [50, 89]. Since the French action plan, the number of women of childbearing age treated with valproate has fallen by 45% and the number of children born of exposed pregnancies by 66%, according to the National Health Data System (SNDS) [90].

4. Clinical Recognition and Warning Signs

4.1. The starting point is the exposure history

Diagnosis is possible only when exposure is asked about. Oxford recommends, faced with a child with dysmorphism and delay, a three-generation pedigree, the exact type and dose of each antiseizure medication in pregnancy (which may require the obstetric record), other teratogenic exposures (alcohol, diabetes or other maternal disease), maternal seizures in pregnancy, at delivery and in the perinatal period, birth weight, neonatal withdrawal symptoms (tremor, seizures, poor feeding) and a detailed developmental and behavioural history [3]. Antiepileptic drugs are among the teratogenic medicines most prescribed to women of childbearing age, and anticonvulsants are increasingly used for migraine and bipolar disorder, so exposure is not limited to mothers with epilepsy [3]. Since most pregnancies in people with epilepsy are unplanned (more than 65% in the American Epilepsy Birth Control Registry, n=1114), the regimen at conception is often the regimen already in use [42]. The French PNDS identifies three contexts of discovery (prenatal diagnosis, a malformation discovered in the neonatal period and a child followed for a neurodevelopmental disorder) and details the history: the nature of the maternal epilepsy or bipolar disorder, its history, any familial character, monitoring in pregnancy and the number of seizures; the antiseizure medications taken (number, doses, combinations, generic and brand name), adherence, dose variations and interruptions; the ultrasound findings, possibly completed by fetal MRI or fetal echocardiography; the mode of delivery; and the pedigree of both parental lines and of the sibship [22]. Suspicion rests on five criteria: exposure to valproate during all or part of the pregnancy, alone or with other anticonvulsants; facial, skeletal or extremity morphological signs; malformations; cognitive or behavioural disorders; and variable organ involvement [22]. The summary for the attending physician accepts the diagnosis on a convergence of arguments, even when another cause is possible, and calls for a search for associated pathology (prematurity, a genetic anomaly, complications of care) [22].

4.2. Contexts of suspicion by age

4.3. Targeted physical examination

The examination includes height, weight and head circumference; metopic ridge, hypertelorism or telecanthus, epicanthic folds and infraorbital groove; short nose, length and smoothness of the philtrum, thin upper lip, eversion of the lower lip and cleft lip or palate; in the hands and feet, joint laxity, nail hypoplasia, stiffness of the interphalangeal joints, long overlapping fingers, hypoplastic or overlapping toes and talipes; cardiac auscultation; the genitalia (hypospadias); and the back, looking for signs of spina bifida occulta [3]. The consensus adds inspection of the forearms and hands in the first year (hypoplastic thumb, flat thenar eminence) and the Beighton score where there is hypermobility [12].

4.4. Urgent signs

Urgent signs are respiratory difficulty or stridor from laryngomalacia or tracheomalacia, which require assessment by an airway surgeon; omphalocele, which requires immediate surgery; neonatal haemorrhage or jaundice in the newborn of a mother with a coagulation or liver function abnormality, which the European SmPC directs should be investigated with platelet count, fibrinogen, coagulation tests and factor assays; hypoglycaemia; and neonatal seizures from withdrawal [3, 12, 40, 47]. In the treated pregnant woman, abrupt discontinuation of valproate is a warning sign in itself, because of the risk of status epilepticus with maternal and fetal hypoxia [40]. There is no newborn screening for FVSD; recognition depends on information about the exposure being passed to the neonatal team [12].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. ERN-ITHACA diagnostic criteria

There is no confirmatory test and no validated specific biomarker; the diagnosis is clinical and requires documented exposure [3, 12]. Oxford states the general principle that a teratogenic syndrome is diagnosed once a genetic or other cause of the anomaly has been excluded, with additional caution when exposure data are scanty or when the phenotype rests on few reports with limited genetic investigation [51]. The ERN-ITHACA consensus of 2019, which took up earlier criteria of Dean and colleagues for fetal anticonvulsant syndrome, established that the diagnosis of FVSD requires all the essential criteria and two suggestive criteria, or one suggestive criterion and a supportive score equal to or greater than 3 [12].

The criteria require neither dysmorphism nor malformation, which allows the exclusively neurodevelopmental form to be diagnosed, provided that the cognitive profile or ASD is accompanied by another criterion [12]. The figure of 6–15% for ASD and the risk of about 20 times for spina bifida are those of the consensus; the EUROCAT study gives an OR of 12.7 for spina bifida [12, 13]. The criteria were formulated by expert consensus (16 members, with patient representation; agreement graded as +++ >75%, ++ 50–74%, + 25–49%; evidence A when supported by the literature or by full consensus, B when it is good practice without specific evidence) and have no estimated sensitivity or specificity [12]. The French PNDS of 2017, which predates the consensus, states that no diagnostic score exists: the diagnosis rests on the exposure and on the presence of all or some of the morphological, malformative, neurodevelopmental and organ signs, and no maternal or parental test is validated [22]. Instead of differential diagnosis, the protocol speaks of "overlapping" diagnoses or comorbidities: an identified genetic disease does not exclude the embryofetopathy, because the exposure may aggravate that disease or mimic it [22]. This position differs from that of the ERN-ITHACA consensus, which requires as an essential criterion the absence of another diagnosis that explains the phenotype [12].

5.2. Investigations and their limits

The consensus requires a normal chromosomal microarray and normal fragile X testing and the exclusion of other syndromes (agreement +++, evidence A) [12]. Oxford adds urinary amino acids and organic acids and other investigations according to the clinical picture, and draws attention to the 15q11.2, 15q13.3 and 16p13.11 microdeletions, which predispose to epilepsy and may be present in the mother and the child [3]. The microarray detects neither single nucleotide variants nor small insertions or deletions, so a normal result does not exclude a monogenic cause of delay, epilepsy or dysmorphism; when the phenotype is atypical, when the mother has epilepsy of probable genetic cause or when there is a family history, the consensus recommends considering genetic study of seizure predisposition and Oxford recommends further investigation [3, 12]. An alternative genetic diagnosis does not of itself exclude a superimposed valproate effect, but the essential criterion "no other diagnosis that explains the phenotype" is then no longer met. Neuropsychological or clinical psychological assessment forms part of the diagnosis, except where a severe deficit is evident: in the infant, cognition, speech and motor function are assessed; at school age, IQ, language, memory, attention and executive function (agreement ++, evidence A) [12]. Echocardiography, renal ultrasound and limb radiography document the supportive criteria [12]. The DNA methylation episignature in peripheral blood distinguishes most affected exposed individuals from controls and from other syndromes with an episignature, but its numerical performance was not published in the abstract consulted and the test does not form part of the 2019 criteria [17]. The French PNDS does not define a standard work-up and adjusts the initial work-up to age and examination; Table 5.2a compares it with that of the European consensus [22] [12].

Table 5.2a. Initial work-up according to the French PNDS (2017) and the ERN-ITHACA consensus (2019) [12, 22]

5.3. Prenatal diagnosis

Prenatal diagnosis addresses malformations and does not detect the neurodevelopmental effect. Faced with an exposure in an ongoing pregnancy, Oxford recommends recording the drug, the dose, the gestational age of exposure, maternal health and the pedigree, and performing a detailed anomaly scan directed at the malformations expected for the nature and timing of the exposure [51]. Ultrasound detects anencephaly at 11–12 weeks and large open defects from about 13 weeks [41]. In the EUROCAT study by Boyd and colleagues (18 registers from 11 countries, 670,766 deliveries), the prenatal sensitivity of ultrasound was 98% for isolated anencephaly (n=166) and 75% for isolated spina bifida (68% correctly identified; n=252), with variation between centres from 33% to 100%; targeted ultrasound may have greater sensitivity [41, 95]. Maternal serum alpha-fetoprotein (AFP), with a threshold of 2.5 multiples of the median (MoM) at 16 weeks, detects 82% of cases of spina bifida with 1.9% false positives, does not detect closed neural tube defects and is less sensitive in women treated with valproate, so it should be combined with a detailed anomaly scan [41]; according to the Royal College of Obstetricians and Gynaecologists (RCOG), the combination of serum AFP and ultrasound raises the detection of neural tube defects to 94–100% [96].

The sources differ on the schedule. The ERN-ITHACA consensus recommends ultrasound at about 13 and 20 weeks, performed by a sonographer informed of the exposure and of the anomalies to look for, and does not require additional examinations if the heart is well visualised at 20 weeks; cleft palate and some cardiac defects may escape detection [12]. The RCOG recommends the detailed ultrasound of the UK fetal anomaly screening programme between 18+0 and 20+6 weeks (grade D) and found no studies on routine fetal echocardiography [96]; the ILAE recommends a detailed anomaly scan at 18–22 weeks by a specialist, fetal echocardiography only on clinical indication and serial scans where growth restriction is suspected [9]; the UKTIS leaflet accepts the anomaly scan from about 16 weeks when exposure occurred in the 1st trimester [75]. The AAN/AES/SMFM guideline recommends a detailed anatomical scan for any ASM (level B) and monitoring of fetal growth throughout pregnancy with valproate or topiramate (level B) [42]. The European SmPC directs that specialised prenatal monitoring for neural tube defects and other malformations be instituted when pregnancy proceeds on valproate [97]. The FDA recommends offering prenatal diagnosis of neural tube defects and other defects [40]. In the Brazilian source consulted, the diagnosis of neural tube closure defects should be prenatal: pelvic or transvaginal ultrasound from 12 weeks in the 1st trimester, nuchal translucency between 11 and 14 weeks presented as a screening test, AFP in amniotic fluid by amniocentesis at 14–16 weeks and, in the 2nd and 3rd trimesters, two-dimensional and three-dimensional ultrasound of the fetal spine and of the signs of the Arnold-Chiari malformation [98]. Maternal obesity reduces image quality, as the case in section 10 illustrates, in which neither the radial defect nor the cardiac malformation was seen at 20 weeks [91]. Amniocentesis is little used for neural tube defects, given the efficacy of ultrasound combined with serum screening [41]. Faced with a diagnosed neural tube defect, Oxford recommends defining the type (open or closed) and whether it is isolated, by examination or autopsy: additional anomalies, present in more than 20% of cases, should prompt reconsideration of a syndrome or a chromosomal disorder, karyotyping is not requested routinely in isolated neural tube defect, and genomic analysis (microarray, panel, exome or genome) is reserved for cases with syndromic signs [41]. Preimplantation genetic diagnosis has no application, because there is no causal variant.

5.4. Practical algorithm

1. Document the exposure (drug, daily dose, period of pregnancy, polytherapy, plasma levels where available), from obstetric or prescribing records [3, 12].
2. Dysmorphological and systemic examination by a clinician experienced in FVSD, with review of photographs from earlier ages [12].
3. Chromosomal microarray and fragile X testing; exclusion of fetal alcohol spectrum disorder and other teratogenic disorders; further genetic investigation if the phenotype is atypical or there is a family history [3, 12].
4. Developmental or neuropsychological assessment appropriate to age [12].
5. Echocardiogram and renal ultrasound (recommended systematically in the exposed newborn), ophthalmological and hearing assessment [12].
6. Apply the criteria (all the essential ones + 2 suggestive, or 1 suggestive + ≥3 supportive points) and record the diagnosis as FVSD [12].

6. Differential Diagnosis

6.1. Comparative table

6.2. Main differentials

The first differential is exposure to other antiseizure medications or to polytherapy, including hydantoin embryopathy: the face is less marked and the risk of neural tube defects and of a neurodevelopmental effect is lower; with carbamazepine the risk of neural tube defects is about 1%, against about 5% with valproate, and nail hypoplasia is more characteristic [3]. In polytherapy including valproate, the phenotype is generally attributed to valproate, but the total risk is greater [28, 81]. The second is fetal alcohol spectrum disorder, whose exclusion is an essential criterion because the smooth philtrum and thin upper lip are common to both exposures [12, 102]. The third is developmental delay of genetic cause, chiefly the copy number variants associated with maternal epilepsy and fragile X, which justify microarray and FMR1 testing in all cases [3, 12]. In the fetus or newborn with a radial defect and a facial anomaly, Kikuchi and colleagues propose distinguishing by the associated facial anomaly: a flat nasal bridge or saddle nose suggests FVSD; micrognathia, trisomy 18, Cornelia de Lange, acrofacial dysostosis, Baller-Gerold or TAR; cleft lip or palate, trisomy 13, Roberts or Aase; microphthalmia, Fanconi anaemia; hypertelorism, Holt-Oram [91]. This guidance comes from a case report and narrative review, and has not been validated prospectively.

Within the FACS group, the distinction is made by the distribution of the findings. Smith describes a similar "anticonvulsant facies" with carbamazepine, hydantoin, primidone and phenobarbital, an increased risk of meningomyelocele with carbamazepine, of oral clefts with the hydantoins and carbamazepine and of cardiac defects with phenobarbital and carbamazepine, and a greater risk with polytherapy [23]. In practice, hypoplastic nails and distal phalanges and poor growth favour hydantoin, and neural tube defects, isolated cleft palate, radial defects and a face with trigonocephaly and infraorbital grooves favour valproate [23, 28]. Orphanet describes the face of fetal carbamazepine syndrome and that of trimethadione syndrome as in part similar to that of valproate, and associates combined exposure to carbamazepine and valproate with significant developmental delay, chiefly of verbal intelligence [100, 101]. The FACS group (ORPHA:370068) has no definition of its own in Orphanet [2].

In Smith's appendix of differential diagnosis by anomaly, fetal valproate syndrome figures as a frequent finding in frontal bossing or prominent central forehead, lateral displacement of the inner canthi with short palpebral fissures, inner epicanthic folds, low nasal root and broad nasal root, short nose with anteverted nares, maxillary hypoplasia with a narrow or high-arched palate, cleft lip with or without cleft palate (a category in which Smith retains the syndrome, although the EUROCAT data show no increase in cleft lip [12]), long philtrum, microstomia, hyperconvex nails and cardiac malformation; and as an occasional finding in deafness, meningomyelocele, lacrimal defects, strabismus, nystagmus, microphthalmia, corneal opacity, cataract, micrognathia, cleft palate or bifid uvula without cleft lip, laryngeal anomalies, supernumerary nipples, bifid rib, hypoplasia or aplasia of the thumb and radius, triphalangeal thumb, polydactyly, limitation of joint extension, altered skin pigmentation, inguinal or umbilical hernias, omphalocele, renal malformation and ambiguous genitalia or hypospadias [107]. These category lists allow the dominant finding to be cross-referenced with the other syndromes that the same appendix associates with each anomaly [107].

The French PNDS organises the differential by groups: other fetopathies (phenytoin, carbamazepine, fetal alcohol spectrum disorder); chromosomal anomalies, with microarray in the presence of ASD, intellectual disability, atypical autism or a set of dysmorphic or malformative signs, and fragile X testing in the presence of intellectual disability or ASD, chiefly in boys; and monogenic syndromes suggested by a particular combination of signs, among which it cites Baller-Gerold syndrome (coronal craniosynostosis and radial ray anomalies), Fanconi anaemia (thumb hypoplasia, growth restriction, pancytopenia) and Smith-Lemli-Opitz syndrome (hypospadias, growth restriction, extremity anomalies) [22]. If a microrearrangement is identified, whether inherited or not, the protocol directs that the respective role of the two causes be assessed, and the finding may justify studies in the parents and the family with information on specific risks [22].

Faced with a neural tube defect, Oxford recommends asking about maternal epilepsy and the use of valproate, carbamazepine, phenytoin, phenobarbital or primidone in the first trimester [41]. The MTHFR C677T and A1298C polymorphisms (historical nomenclature, = NM_005957.5:c.665C>T p.(Ala222Val) and NM_005957.5:c.1286A>C p.(Glu429Ala)) have been associated with a risk of neural tube defects about 1.8 times higher only in non-Hispanic populations, and genes predisposing to neural tube defects are not tested in clinical practice [41]. The empirical recurrence risks of the multifactorial inheritance of neural tube defects do not apply to valproate-related neural tube defects, whose risk depends on the exposure (section 8.1) [41]. Two pitfalls arise from the child's own medication: facial features may become coarse, and a child with poorly controlled seizures on high doses of ASM may appear to show developmental regression [64, 106]. In Noonan syndrome and other RASopathies, Oxford recommends asking about prenatal exposure to alcohol and anticonvulsants, because of the overlap of dysmorphic features [105].

7. Treatments and Risk Management Strategies

7.1. Principles

There is no specific drug treatment for FVSD [12], and no intervention trials with that aim in the searches made in ClinicalTrials.gov in October 2026 [108]. Management has three levels: primary prevention, which consists of avoiding fetal exposure to valproate through the therapeutic choice and through regulatory measures; management of the already exposed pregnancy, in which the priority is control of maternal seizures; and symptomatic management and surveillance of the child and the adult with FVSD, according to the ERN-ITHACA consensus. At all levels, the decision on valproate belongs to the specialist who treats the maternal condition (epilepsy, bipolar disorder or migraine), in liaison with obstetrics and genetics.

7.2. Valproate: substances, indications and posology

Valproate medicines include valproic acid, sodium valproate, magnesium valproate, semisodium valproate (divalproex sodium, divalproex) and valpromide, with national authorisations in all Member States of the European Union (EU), in Norway and in Iceland [18]. In the EU they are indicated in epilepsy and bipolar disorder and, in some countries, in migraine prophylaxis; the EMA list includes, among others, the brands Depakine®, Depakin®, Depakote®, Diplexil®, Epilim®, Convulex®, Episenta®, Ergenyl® and Orfiril® [18]. In France, the PNDS of 2017 cites Dépakine®, Dépakine Chrono®, Micropakine®, Dépakote®, Dépamide® (valpromide) and generics, and valproate is not authorised in France for migraine prevention [22]. In the United Kingdom, Belvo®, Convulex®, Depakote®, Dyzantil®, Epilim®, Epilim Chrono®/Chronosphere®, Episenta®, Epival® and Syonell® are marketed [37]. In the USA, valproic acid is marketed as Depakene® and divalproex sodium as Depakote® (delayed-release tablets), Depakote® Sprinkle (capsules) and Depakote® ER (extended release) [26]. The current North American indications are manic episodes of bipolar disorder, monotherapy and adjunctive therapy of complex partial seizures and of simple and complex absence seizures, adjunctive therapy in multiple seizure types that include absences, and migraine prophylaxis [40]. In Portugal, the INFARMED (National Authority of Medicines and Health Products) Infomed database listed in 2024 Ácido Valpróico Generis, Ácido Valpróico Ratiopharm 300 mg and 500 mg, Depakine®, Depakine Chrono® 300 and 500, Depakine Chronosphere®, Diplexil® (150, 300, 500 and 1000), Diplexil-R®, Epixival® and Valproato de sódio Altan [109]. In France, Dépakine®, Micropakine®, Dépakote®, Dépamide®, Divalcote® and generics are marketed [110]; in the USA, MotherToBaby also lists Stavzor® and Depacon® [111]. The sources consulted do not identify the brands authorised in Brazil.

Valproate is here the causal agent and not a treatment for FVSD; the posology matters because the fetal risk depends on the dose. The table summarises the FDA label for Depakote® and the harmonised text of sections 4.2 to 4.6 of the European SmPC defined by the 2018 referral; the harmonised text does not set numerical doses, and the national SmPCs (for example, Depakine® and Diplexil® in Portugal) were not read individually.

The serious adverse effects that give rise to the boxed warning of the North American label are hepatotoxicity (greater risk below 2 years of age and in mitochondrial disease due to POLG variants), fetal risk and pancreatitis; thrombocytopenia increases with trough concentrations above 110 µg/mL in women [40]. For dose and fetal risk, the thresholds cited vary with the source and should not be merged: the 2016 MHRA guidance reproduced in Oxford recommended avoiding doses ≥700 mg/day; the Cochrane review associated doses >800 mg/day with greater cognitive effects; Oxford reports more malformations above 900 mg/day; Kikuchi's case report mentions a greater risk above 1000 mg/day or levels of 70 µg/mL; the ERN-ITHACA consensus, up to 24% malformations above 1500 mg/day; EURAP showed an increased risk already at ≤650 mg/day; the ILAE places the higher-risk thresholds at 500–650 mg/day according to the register; in the French ANSM cohort, the risk of neurodevelopmental disorders trebled in the tertile >1100 mg/day; and UKTIS associates doses >1000 mg/day with the highest risk [3, 7, 9, 12, 50, 75, 91]. The FDA and the European SmPC conclude that there is no dose without risk [40, 47]. Division of the daily dose and extended release, which the EMA recommends in order to avoid plasma peaks, have no clinical data showing that they reduce the teratogenic risk [9, 97].

7.3. Comparative regulatory status

Sources for the table: [3, 11, 12, 15, 18, 19, 26, 37–40, 109, 112–115].

7.4. European pregnancy prevention programme

The 2014 referral, started in October 2013 at the request of the MHRA under Article 31 of Directive 2001/83/EC, ended with a consensus position of the Coordination Group for Mutual Recognition and Decentralised Procedures (CMDh): do not prescribe valproate for epilepsy or bipolar disorder to pregnant women, to women who may become pregnant or to girls, unless alternatives are ineffective or not tolerated; effective contraception; initiation and supervision by an experienced physician; and, in migraine prophylaxis, do not use in pregnancy, exclude pregnancy before starting and do not prescribe without contraception [11]. As women continued not to receive adequate information, the ANSM requested a further referral on 9 March 2017; the CMDh endorsed the measures by majority on 21 March 2018 and the European Commission adopted the legally binding decision on 31 May 2018 [18]. The procedure included an EMA public hearing in September 2017, with contributions from affected families [18].

The PPP comprises assessment of each patient's childbearing potential; pregnancy tests before starting and during treatment as required; counselling on the risks and on the need for effective contraception throughout treatment; review of the treatment by a specialist at least once a year; and an annual risk acknowledgement form signed by the patient and the prescriber [18]. The packaging now carries a visual warning, with an attached patient card that the pharmacist reviews at each dispensing, and the educational materials for patients and professionals were updated [18]. For women planning a pregnancy, an alternative must be decided with specialist consultation, while recognising that there is a small number of women with epilepsy without a suitable alternative, who must be supported and counselled [18]. Marketing authorisation holders were required to carry out studies on the nature and extent of the risks, to monitor use, to conduct surveys on the effectiveness of the measures, to use registers to characterise fetal anticonvulsant syndrome with valproate in comparison with other ASM, and to conduct a retrospective observational study, from which the paternal exposure study assessed in 2023–2024 arose [18, 38]. Patients must not stop valproate without consulting their doctor [18].

The harmonised SmPC text of 2018 gives effect to these conditions [97]. In epilepsy, valproate is contraindicated in pregnancy unless there is no suitable alternative and, in women of childbearing potential, unless the PPP is complied with; in bipolar disorder and in migraine prophylaxis, where that indication exists, it is contraindicated in pregnancy without exception. Pregnancy is excluded before initiation by a negative plasma pregnancy test confirmed by a healthcare professional, and tests are repeated during treatment as required. Contraception requires at least one effective method, preferably user-independent (an intrauterine device or implant), or two complementary methods including a barrier method, even in the case of amenorrhoea; the conditions also apply to women with no current sexual activity, unless there are compelling reasons, in the prescriber's judgement, for the absence of risk. In girls, parents or carers must contact the specialist at menarche, full information is given after menarche, the need for treatment is reassessed every year and a change to an alternative should be attempted before adulthood. In epilepsy, every effort must be made to change to an alternative before conception and before stopping contraception; in bipolar disorder and in migraine, valproate is discontinued and replaced before conception. Faced with a pregnancy, the patient is referred immediately to the specialist, and the patient and her partner to a specialist in teratology. The pharmacist hands over the patient card at each dispensing and advises her not to stop treatment and to contact the specialist. The educational materials are the professional's guide, the patient's guide, the patient card and the annual risk acknowledgement form [97]. The SmPC recalls that tonic-clonic seizures and status epilepticus with hypoxia may entail a risk of death for the mother and the fetus [97].

In Portugal, INFARMED transposed the CMDh position by Information Circular no. 054/CD/550.20.001, of 29 March 2018, which prohibited valproate in migraine and in bipolar disorder during pregnancy, limited its use in epilepsy during pregnancy to the absence of an alternative and its use in women of childbearing age to the conditions of the PPP, including girls before puberty, with a visual alert on the packaging and a patient card [19]. In June 2017, an INFARMED communication, accompanied by an infographic, already indicated that valproate should be stopped immediately in the event of pregnancy only in migraine prevention and should not be stopped in epilepsy and bipolar disorder without contacting the doctor as a matter of the greatest urgency [116]. The measures for men reached professionals through Information Circular no. 006/CD/550.20.001, of 19 January 2024, known in the sources consulted through its circulation by the Regional Health Directorate of Madeira [115]. The Pharmacovigilance Bulletin of September and October 2024 summarises the nine prescriber measures in women of childbearing age (assess childbearing potential; explain the risks to the patient and her partner; negative pregnancy test by blood analysis before and during treatment; effective contraception; plan the pregnancy; urgent consultation in the event of pregnancy; at least annual review; Patient Guide; annual risk acknowledgement form) and the measures in men (initiation and supervision by a specialist; effective contraception, including by the partner, during treatment and for at least 3 months afterwards; no sperm donation; regular review; consider alternatives if conception is planned; Patient Guide), and reports that the educational materials were published in Infomed on 23 September 2024, addressed to neurology, psychiatry, general and family medicine, child psychiatry and neuropaediatrics [109]. The INFARMED Medicines Information Centre (CIMI; cimi@infarmed.pt; Medicines Line 800 222 444) answers requests for information [19].

7.5. National measures in the United Kingdom and in France

The United Kingdom adopted its own rules, more restrictive than those of the EU. Since January 2024, valproate should be prescribed to patients of either sex under 55 years of age only when two specialists agree that the condition does not respond to other treatments and that the benefits outweigh the risks [37]. In women and girls of childbearing potential, the PPP and the annual form are mandatory, with countersignature by a second specialist at initiation and the signature of a single specialist at subsequent reviews; men under 55 years who start valproate complete a male risk acknowledgement form [37]. The chronology includes the national patient safety alert of November 2023, entry into force in January 2024, the precautionary recommendation of contraception for men and their partners in September 2024, the limitation, in February 2025, of two-specialist review in men to those starting treatment, and new educational materials in June 2025 [37]. The second specialist may be a consultant neurologist or psychiatrist, a paediatrician with an interest in epilepsy, a specialist nurse or a specialist pharmacist, according to local determination, and valproate is always dispensed in the full pack with the warnings [37]. The MHRA communicates to patients a risk of about 1 in 9 (11%) of a birth defect and of 3 to 4 in 10 of early developmental problems, which it describes as seriously disabling and permanent [37]. Prescribing is monitored by six-monthly reports from the CPRD Aurum primary care database [37]. In a time series of 4,879,978 prescriptions in England between January 2022 and April 2025, the 2023 national patient safety alert was associated with a change in trend of −85.2 patients per month (95% CI −122.9 to −47.5), with no effect on the prescribing of lamotrigine or levetiracetam [117]. The NICE (National Institute for Health and Care Excellence) guideline NG217, updated on 5 August 2026, incorporates the MHRA rules: do not start valproate below 55 years of age, in either sex, unless two specialists independently agree and document that there is no other effective and tolerated treatment or that there are compelling reasons why the reproductive risks do not apply; in women and girls of childbearing potential, only after other options have failed, with discussion of the risks and the PPP; in boys and men, effective contraception (condom and partner contraception) during treatment and for 3 months afterwards, and information on fertility and options for those planning to have children in the coming year [118].

In France, according to the EPI-PHARE communication of November 2025 on paternal exposure (section 8.3), the initial prescription of valproate and its derivatives in adolescents and men liable to father children has been reserved, since 6 January 2025, to neurologists, psychiatrists and paediatricians, and dispensing now requires an attestation of shared information signed by the doctor and the patient (for initiations since 6 January 2025 and for renewals since 30 June 2025) [114]; according to the association Épilepsie-France, the restriction of January 2025 also covered carbamazepine and topiramate [110]. In French paediatrics, valproate use fell by 37% between 2014 and 2023 (from 58,845 to 37,014 patients) and by 62% in girls (from 23,480 to 8975) [119]. The French rules in force when the PNDS was published, in March 2017, were as follows: do not prescribe valproate to girls, adolescents, women of childbearing potential or pregnant women, unless other treatments have failed; mandatory effective contraception in treated adolescents and women; prompt consultation with the specialist in the face of a pregnancy plan or a pregnancy; initial prescription by a neurologist, psychiatrist or paediatrician, repeated at least once a year, with the signing of a care agreement by the patient; renewal by any doctor [22]. The protocol records that France triggered, on 20 February 2017, a European referral intended to contraindicate valproate in psychiatric indications during pregnancy and in women without effective contraception; the EMA page dates the ANSM's formal request to 9 March 2017 (section 7.4) [22] [18]. Therapeutic education of the young woman with epilepsy, before any pregnancy, includes in the PNDS annual reassessment of the epilepsy, dose adjustment and surveillance of complications, a change of antiseizure medication and optimisation of doses before pregnancy, contraception with a planned pregnancy, and the handing over of the ANSM information leaflet and patient card [22].

7.6. Clinical guidelines

The AAN/AES/SMFM guideline of 2024, based on a systematic review to August 2022 and replacing the AAN practice parameter of 2009, is the most recent; levels A ("must") and B ("should") convey the strength of the recommendation [42].

Recommendation 2B is based on EURAP, in which generalised tonic-clonic seizures or focal seizures with bilateral evolution in pregnancy were twice as likely when valproate was withdrawn or replaced as when it was maintained [42]. The task force of the Commission on European Affairs of the International League Against Epilepsy (CEA-ILAE) and the European Academy of Neurology (EAN), in 2015, after the 2014 referral, recommended avoiding valproate wherever possible in women of childbearing potential; deciding in a shared manner with assessment of risk and benefit by seizure type; discussing alternatives when valproate is the most effective option; not using it first line in focal epilepsy; accepting it first line in the syndromes in which it is the most effective, including the idiopathic generalised epilepsies with tonic-clonic seizures, and when pregnancy is very unlikely (significant intellectual or physical disability); and maintaining regular follow-up [120]. This guidance predates the 2018 PPP, which made valproate contraindicated in women of childbearing potential outside the programme; the first-line acceptance proposed in 2015 must today fit within the rules of the PPP [18, 120]. The ILAE task force on women and pregnancy published in 2019 a review of the risks of seizures and of teratogenicity, of pharmacokinetics in pregnancy and of preconception optimisation, with recommendations based on expert opinion [9]. The table compares the recommendations of the European and international guidelines that complement the North American one.

The ILAE illustrates counselling with two cases. In a woman with juvenile myoclonic epilepsy controlled on valproate 800 mg/day after failure of lamotrigine and levetiracetam, the information to be given includes an MCM risk of 7–10%, an IQ reduction of 7–10 points and a risk of autism about 5 times greater; the alternatives zonisamide and topiramate have limited data (MCM about 4.5% with topiramate), and one option is to reduce valproate and add lamotrigine. In a pregnant woman at 7 weeks on valproate 900 mg/day, the period of greatest malformative risk has already passed, the neurodevelopmental risk depends on the dose (it increases above 800–1000 mg/day, but exists below 800 mg/day), and a change during pregnancy entails 7–8 weeks of polytherapy and a risk of tonic-clonic seizures, so the decision is shared [9]. The 2nd edition of the RCOG guideline was in preparation in October 2026 [96]. A systematic review of 14 international guidelines (appraisal with the AGREE II instrument, Appraisal of Guidelines for Research and Evaluation) found consensus on preconception counselling, folic acid, vaginal delivery, breastfeeding and avoiding valproate, and divergence on the choice of ASM, on monitoring of levels and on the dose and duration of folic acid [121]. Oxford, on the basis of the 2016 MHRA guidance, adds reassessing the indication and considering gradual withdrawal in women in remission with a low risk of relapse who accept the risk, aiming at monotherapy at the minimum effective dose, measuring the plasma concentration once the dose is optimised, and monitoring the levels of lamotrigine and possibly of oxcarbazepine and levetiracetam during pregnancy [3]. The divergences between the sources concern the acceptance of valproate first line (ILAE/EAN 2015 versus the PPP), the number of specialists required (one in the EU, two in the United Kingdom), the measures for men (EU, United Kingdom and France versus the USA), monitoring of serum levels (AAN and Oxford versus RCOG), neonatal vitamin K (all newborns for the ILAE, only those exposed to inducers for the RCOG) and the dose of folic acid (section 7.7).

The French PNDS of 2017 differs from the more recent guidelines on several points, summarised in Table 7.6a [22].

Table 7.6a. Divergences between the French PNDS (2017) and the other sources cited [9, 12, 22, 47, 96, 118]

7.7. Folic acid

The dose of folic acid is the point of greatest divergence between the sources, and its efficacy against the effects of valproate is not demonstrated.

The Medical Research Council (MRC) trial of 1991, which excluded women with epilepsy, showed that 4 mg/day of folic acid protected 72% of women with a previous neural tube defect from a further occurrence (RR 0.28; 95% CI 0.12–0.71); it is the basis of the high doses for recurrence prevention, and not a trial in pregnant women on ASM [41, 122]. In NEAD, periconceptional folate was associated with a higher IQ at 6 years (108 against 101) in all drug groups, a secondary, non-randomised finding [14, 15]; in MONEAD, periconceptional folate was associated with better non-verbal outcomes at 6 years (coefficient 2.54; 95% CI 0.93–4.15) [86]. As the teratogenic mechanism of valproate does not depend exclusively on folate antagonism, supplementation does not replace the choice of an alternative drug [41, 42, 97]. The childhood cancer signal with doses ≥1 mg/day, observed in a single study, is the North American guideline's argument for not using high doses routinely, whereas the RCOG, UKTIS and the ERN-ITHACA consensus accept 4–5 mg/day [12, 42, 75, 96]. The 5 mg dose in the United Kingdom comes from the RCOG and UKTIS, and not from NG217 or the MHRA guidance, which give no dose; the NICE Clinical Knowledge Summaries entry with the folate dose is accessible only from the United Kingdom and was not read [75, 96, 118]. Practice follows this divergence: in a survey of 52 international experts, 73.1% recommended 1–2 mg/day, 15.4% 0.4–0.8 mg/day and 36.5% higher doses with valproate [123]; in a Dublin maternity hospital, over 10 years and 250 pregnant women on ASM, 75.2% took preconception folate (65.6% at the high dose of 5 mg, 9.6% at 0.4 mg), but compliance with the Irish national guideline of 5 mg was 30.8% in the 16 pregnant women on valproate, against 76.0% in the remainder [124]. If a neural tube defect occurs despite high-dose folate, Oxford recommends reconsidering a syndromic association and maintains the high dose in the next pregnancy, with little expectation of benefit [41].

7.8. The already exposed pregnancy

Faced with an unexpected pregnancy on valproate, the woman must not stop the medication without speaking to her neurologist; abrupt discontinuation may cause status epilepticus with maternal and fetal hypoxia [12, 40]. The risks of seizures include accidents, social consequences such as loss of the driving licence, and sudden unexpected death in epilepsy (SUDEP), which figures among the causes of maternal death; for the fetus, trauma from a fall, status epilepticus and prolonged seizures in labour, which occur in 1–2% of women with epilepsy [3]. Frequent maternal tonic-clonic seizures in pregnancy may also have fetal neurodevelopmental consequences [125]. As a rule, withdrawal or change of ASM once pregnancy is established carries more risks than gains [3, 42]. In a meta-analysis of 9 studies (3 in the quantitative synthesis), the 277 women who discontinued or changed valproate before or during pregnancy had more recurrence of tonic-clonic seizures in pregnancy than the 2206 who continued it (OR 1.73; 95% CI 1.06–2.84), and 7.9–72.2% restarted valproate [126]. In a centre in Kerala (India), 10 of 18 women seizure-free for at least 1 year who changed drug (55.6%) had a recurrence within 12 months [127]. In the Australian register (2403 pregnancies), valproate was the drug with the best maintenance of seizure freedom, and levetiracetam the one with the best overall balance between seizure control and fetal outcome [128]. The European SmPC directs that the woman and her partner be referred to a specialist in teratology [97], and the ILAE recommends screening for depression and anxiety in each trimester and postpartum [9]. Management includes a joint obstetric and neurology clinic, ultrasound at about 13 and 20 weeks with a sonographer who has been informed, monitoring of fetal growth, the offer of prenatal diagnosis and, in the presence of a major malformation, planning delivery in a hospital with suitable neonatal care [12, 40, 42]. The FDA recommends monitoring maternal coagulation (platelets, fibrinogen, factors) and, if abnormal, that of the newborn, and monitoring liver function [40]. In pregnancy, the total serum valproate concentration may fall by up to 23% without a fall in the free fraction, so the ILAE considers monitoring optional and, if performed, preferably of the free fraction; the RCOG does not recommend routine monitoring of levels [9, 96]. Intrauterine growth retardation and caesarean section were not significantly increased in the Liverpool and Manchester cohort [12]. Every exposure should be recorded in a pregnancy register (NAAED in the USA, with enrolment by the patient herself on 1-888-233-2334; EURAP; UK Medicines and Pregnancy Registry) and suspected fetal effects reported to the pharmacovigilance system [3, 37, 40]. The French PNDS insists on avoiding generalised tonic-clonic seizures in pregnancy, because they cause fetal hypoxia and are associated with more spontaneous miscarriages, intrauterine fetal deaths, intracranial haemorrhages and developmental disorders, and recommends clinical balance with the minimum dose of antiseizure medication; plasma level measurement is recommended, but clinical balance takes precedence [22]. The protocol gives a fetal mortality of about 3% with valproate, equal to that of the other anticonvulsants, and a significant increase in pre-eclampsia, placental abruption and haemorrhage [22]. Faced with ultrasound findings, fetal MRI or fetal echocardiography define the level of a spina bifida, whether a cleft is isolated, or a suspected cardiac defect [22].

7.9. Neonatal period and breastfeeding

The consensus recommends that the newborn be examined by a professional familiar with FVSD, with direct visualisation of the palate looking for a submucous cleft, inspection of the radial ray, of the spine (sacral dimples >5 mm, high-placed or with a dermal sinus) and of the metopic ridge; postnatal renal ultrasound in all exposed infants (agreement +++, evidence A); a single echocardiogram (+++, B); surveillance for hypoglycaemia; treatment of signs of withdrawal; assessment by the airway surgeon in the presence of respiratory symptoms, with rigid endoscopy under general anaesthesia if necessary; physiotherapy or orthopaedics for talipes; and coordination of paediatric follow-up before discharge [12]. Suspicion of dysraphism (dermal sinus, a large or high-placed dimple, a spinal cord anomaly on ultrasound) leads to imaging and to paediatric neurosurgery; small sacral dimples may be disregarded [12]. The European SmPC recommends investigating in the newborn the platelet count, fibrinogen, coagulation tests and clotting factors, because of the very rare risk of a haemorrhagic syndrome, and describes neonatal hypoglycaemia and hypothyroidism [47]. Maternal oral vitamin K is not necessary, because valproate is not an enzyme inducer; many countries give 1 mg of vitamin K intramuscularly to all newborns, as the ILAE recommends, while the RCOG directs it at the children of women on enzyme-inducing drugs [9, 12, 96]. UKTIS recommends neonatal observation for signs of withdrawal [75]. An epilepsy nurse should review the mother after delivery with regard to seizures, medication, care of the child and contraception [12].

Breastfeeding should be encouraged according to the consensus (+++, A), the ILAE and NICE [9, 12, 118]. The European SmPC, which reports excretion into milk of 1–10% of maternal serum levels and haematological abnormalities described in breastfed infants, leaves the decision between breastfeeding and treatment to individual assessment, with a different emphasis from that of the guidelines [47]; the RCOG describes minimal transfer into milk (milk/serum ratio 0.01–0.1) [96], and MotherToBaby reports levels that are generally undetectable in the infant [111]. Valproate passes into milk at concentrations of 0.4–3.9 µg/mL (1–10% of maternal serum levels), and the infant's serum reaches 0.7–4 µg/mL (1–6% of the maternal level) [40]. In NEAD, the adjusted IQ of breastfed children was not lower: 93 (n=11) against 90 (n=24) at 3 years and 106 (n=11) against 94 (n=25) at 6 years (p=0.04), and a Norwegian study found no impact at 6 and 18 months [12, 40]. The infant should be monitored for jaundice, bruising or bleeding [40]. The French PNDS reproduces on breastfeeding the text of the SmPC (excretion of 1–10% of maternal serum levels, haematological abnormalities described, individual decision) and recommends for vitamin K a scheme of its own with three doses (Table 7.6a) [22]. It proposes enrolling the child in a follow-up register and systematic follow-up, with early screening, in a network for vulnerable newborns linked with consultations at an early medical and social action centre (CAMSP) [22]. Mothers with epilepsy who care for the baby alone are advised to change the nappy on the floor and not to bath the child when they are alone [22].

7.10. Management of the child and the adult with FVSD

Management is symptomatic and multidisciplinary, coordinated by a paediatrician and later by the family doctor, with the intervals of the surveillance table in section 11. Malformations are treated according to their own indications: surgery for spina bifida, for the cardiac defect, for cleft palate, for hypospadias and for craniosynostosis (with referral to the craniofacial unit in the presence of altered skull shape or growth), and orthopaedics for the radial defect and talipes [12]. Language is the priority domain: the consensus recommends referral to speech and language therapy by 2.5 years where there are concerns and, if necessary, schooling with training in speech, language and communication needs or a specialist language unit [12]. Neuropsychological assessment before or during the first school year guides an individualised education plan, and educators should be informed; in adolescence, extra time or scribe support in examinations is recommended [12]. Hypermobility with leg pain is assessed by the Beighton score; above 6, paediatric physiotherapy and a graded exercise programme (swimming, soft play), with podiatry and orthoses for foot deformities and occupational therapy when hand function is affected [12]. The risk of seizures in the child is increased when the mother has epilepsy, and the consensus recommends considering a genetic test of predisposition [12]. In open spina bifida, postnatal surgical repair is the basis of treatment and ventricular shunting is frequent; caesarean section was advocated and later questioned; the fetal surgery of the MOMS trial (Management of Myelomeningocele Study) halved shunting for hydrocephalus, reduced hindbrain herniation and improved motor function, at the cost of more prematurity and maternal uterine dehiscence [41, 48]. The Brazilian source consulted describes protection of the placode with gauze moistened in warm saline and a plastic covering, prone or lateral positioning, a latex-free environment, prophylactic antibiotics (ampicillin and gentamicin), closure on the 1st day of life and ventriculoperitoneal shunting if there is hydrocephalus, and recommends elective caesarean section [98]. In the adult, an annual health check for people with learning difficulties, annual weight and diet review, annual optometry, podiatry, physiotherapy and occupational therapy or workplace ergonomics are recommended [12].

The French PNDS sets out the aims of treatment: specific neonatal surveillance; surgical correction of malformations; ear, nose and throat care and growth surveillance; early assessment and stimulation of psychomotor development; speech and language care; prevention and treatment of behavioural and psychiatric disorders; support for schooling; and socio-occupational integration in the adult [22]. In ear, nose and throat care, otitis media with effusion is looked for at every consultation and treated to avoid hearing loss; recurrent otitis without complete resolution justifies tympanostomy tubes, with swimming allowed using earplugs; repeated rhinosinusitis is treated with nasal clearance once or more daily in early childhood; and adenoidectomy or tonsillectomy may improve language when combined with speech and language therapy, by decision of a paediatric ear, nose and throat team after multidisciplinary discussion [22]. Speech and language therapy begins in the newborn if there are swallowing difficulties and, for language, before the first year of life, with parental guidance [22]. In cleft palate or cleft lip and palate, the usual scheme is veloplasty at about 6 months, combined with cheilorhinoplasty in cleft lip and palate, and closure of the bony cleft between 12 and 18 months; velopharyngeal insufficiency may require velopharyngoplasty [22]. All children are seen in ophthalmology from 6 months and once a year thereafter, to screen for myopia, refractive errors, strabismus and amblyopia; in the presence of a coloboma, microphthalmia and glaucoma are looked for [22]. Urogenital malformations, craniosynostosis, diaphragmatic hernia and umbilical and inguinal hernias are treated in paediatric surgery or neurosurgery; spina bifida is treated by neurosurgery in the prenatal or neonatal period, late vertebral deformities by orthopaedics, and the spine is monitored throughout growth, chiefly in adolescence [22]. In congenital heart disease, the PNDS refers to the French recommendations on antibiotic prescribing in dentistry: endocarditis prophylaxis in unoperated cyanotic heart disease, in operated disease with a residual shunt, in operated disease with prosthetic material if there is a residual shunt or in the 6 months after placement, and in bearers of a valve prosthesis [22]. Care is completed by oral hygiene with fluoride and orthodontics; physiotherapy, functional rehabilitation, orthoses and surgery for deformities and hyperlaxity; psychomotor therapy from the first developmental lag, often in the first year; occupational therapy; and psychological support for the patient, the parents and the siblings, guided by a neuropsychological assessment [22]. Medicines for behavioural or psychiatric disorders in the adolescent and the adult are prescribed within an overall approach and after multidisciplinary discussion; the symptomatic drugs cited in the PNDS are summarised in Table 7.10a [22].

Table 7.10a. Symptomatic treatments cited in the French PNDS (2017) and comparison with the labels [22, 129–133]

7.11. Anaesthesia and the perioperative period

Preoperative assessment includes cardiology and ear, nose and throat (ENT) assessment of the upper airway; micrognathia, a small mouth and cleft palate predispose to a difficult airway; subglottic stenosis should be suspected, with ENT present if necessary; and newborns and infants should be monitored for postoperative apnoea (all with agreement +++, evidence B) [12].

7.12. Evidence: studies supporting the decisions

There are no randomised intervention trials in FVSD or on the prevention of its effects; the ClinicalTrials.gov searches of October 2026 on valproate with pregnancy, fetal or paternal exposure returned 17 intervention studies with no relevance (oncology, Wolfram syndrome, kidney injury), and a negative search does not demonstrate non-existence [108]. Across all study types, the relevant records are observational: NEAD (NCT00021866, completed, 331 participants), MONEAD (NCT01730170, completed), NCT01097720 (development after lamotrigine, valproate or carbamazepine, 298 participants, status unknown), NCT06499779 (Strasbourg, qualitative study on the prescribing of valproate in women of childbearing age, 2024–2026) and NCT01231646 (autoantibodies against the folate receptor on valproate, completed, 38 participants) [108]. The decisions rest on observational cohorts.

These studies do not permit indirect comparisons with one another, because they differ in period, comparators, definition of outcomes and age at assessment. Relative risks must be read together with the absolute rate: an aHR for ASD of 2.9 corresponds to 4.42% in exposed children against 1.53% in the population [35], and the aHR of 1.50 in the paternal study to about 5% against 3% [38]. The indication (maternal or paternal epilepsy and its genetics) is the main confounder: in Denmark, restriction to the children of mothers with epilepsy reduced the aHR for ASD from 2.9 to 1.7, with a CI including 1 [35]. The paternal exposure studies subsequent to the PASS were for the most part negative, with the exception of the French EPI-PHARE study, and enter the table for the same reason as the positive ones (section 8.3). The recent Asian cohorts (Korea, Taiwan) use administrative codes and give estimates lower than those of the prospective European and North American registers, which does not allow a conclusion of lower risk in those populations [46, 82].

7.13. Recent developments (July 2025 to October 2026)

7.14. Transition

The consensus recommends an integrated transition from paediatric to adult services with the patient's participation (+++, A), between 16 and 17 years of age, with transfer to the family doctor accompanied by a summary and an adult checklist, because parents identify the period from 16 to 20 years as the most difficult [12]. For the young woman with FVSD who herself takes antiseizure medication, the same prescribing rules and the same PPP apply. The French PNDS proposes a "pivot" team to ensure continuity of care throughout life, because those involved change with age, and calls for support at the critical periods: starting school, learning, adolescence and the passage to adulthood [22]. Between 16 and 18 years it provides for medical and, if necessary, institutional transition; in the adult, an annual general medical or specialist consultation that includes gynaecological follow-up with prevention of pregnancy complications, fertility follow-up in both sexes and adaptation of health conditions to the working environment [22].

8. Reproductive Options and Family Counselling

8.1. Nature of the risk and recurrence

FVSD is not heritable: the risk arises from the exposure and recurs in every exposed pregnancy. Kozma described two affected siblings from exposure to valproate in monotherapy in successive pregnancies [4, 28]. Oxford attributes to Dean and colleagues a sibling recurrence risk of 39–55% if the mother continues the same drug at the same dose [3], but the article does not contain that range [55]. In the Grampian cohort, with all ASM pooled and in an era of frequent polytherapy, 68 of 149 families had a child with a major malformation or developmental delay; counting the first affected child as the proband, 19 of the 65 exposed siblings were also affected, a crude recurrence of 29% (26%, 14 of 54, excluding families with a history of developmental delay); by Weinberg's proband method, which corrects for incomplete ascertainment, recurrence was 43%, and 39% excluding the positive family history [55]. The outcome is major malformation or developmental delay, and not major malformation alone, and the figure of 55% does not appear in the article [55]. Oxford adds that the risk is high if the mother continues the same medication and dose, but that severity is unpredictable and may vary between siblings exposed to the same dose [51]. According to the ILAE, recurrence of MCM is high in the pregnancies that follow a child with MCM, and in EURAP a parental history of MCM trebled the risk; in the EURAP dose-response analysis of 2011, the OR was 4.4 (2.06–9.23) [9, 44]. The empirical risks of the multifactorial inheritance of neural tube defects (for example, 2–5% after an affected sibling, depending on the population incidence) and the recurrence of 1.9% after a child with a neural tube defect cited by Smith serve only to explain the difference between the multifactorial and the teratogenic neural tube defect, whose risk depends on the exposure [41, 48]. For a first exposed pregnancy, the reference risks are those of the exposed population: about 10% MCM, 1–2% spina bifida and 30–40% developmental problems at preschool age [7, 11, 40]. The children of a person with FVSD are at no increased risk on that account, because the injury is not transmitted; that person's reproductive risk depends on the medication she herself takes.

8.2. Counselling before a further pregnancy

After the birth of an affected child, the mother should be referred to the neurologist to assess withdrawal of the drug (if she has been seizure-free for several years), simplification to monotherapy or dose reduction, always without stopping or changing the medication without speaking to the specialist [3]; Oxford insists that the medication review be carried out before a further pregnancy, with the attending physician [51]. The ERN-ITHACA consensus recommends that valproate be supervised by a specialist according to EMA and national guidance, that the genetic cause of the maternal seizures be considered and investigated, that the risk of recurrence of seizures in the child be assessed and communicated, and that folic acid be started 2–3 months before conception [12]. The AAN/AES/SMFM guideline recommends choosing the ASM and the dose as early as possible before conception, preferably lamotrigine, levetiracetam or oxcarbazepine, with MCM rates of 3.1%, 3.5% and 3.1% in the series with more than 1000 exposures [42]. The older data cited by Oxford point in the same direction: in the international lamotrigine register, 3 of 168 exposed in monotherapy (1.8%; 95% CI 0.5–5.5) had malformations, against 5 of 50 (10%) in polytherapy including valproate [148]; in the UK and Ireland register, 2 of 304 exposed to levetiracetam in monotherapy (0.70%; 0.19–2.51) [149]; the figure of 2.1% in 390 exposed to lamotrigine that Oxford attributes to the UK register does not appear in the abstract of the 2006 analysis [3, 56]. Effective contraception forms part of the PPP for as long as treatment lasts [18].

The French PNDS assigns five aims to the genetics clinic: to establish the diagnosis and the recurrence risk; to study comorbidities; to assess the risks linked to familial epilepsies and to the possible transmission of the maternal disease that includes epilepsy; to advise on prevention of recurrence by modifying or stopping the treatment; and to request genetic tests in the child with intellectual disability, with or without malformations, beginning with fragile X and the microarray [22].

8.3. Paternal exposure

In March 2023, the EMA started, under Article 107p of Directive 2001/83/EC, the assessment of the post-authorisation safety study (PASS) EUPAS34201, an obligation of the 2018 procedure: a retrospective register-based study in Denmark, Norway and Sweden that compared the children of men treated with valproate in the 3 months before conception with those of men treated with lamotrigine or levetiracetam [38]. Before its conclusion, the PRAC identified limitations in the definition of neurodevelopmental disorder (NDD) and in the type of epilepsy, and the marketing authorisation holders reported errors in the Norwegian database, which required reanalysis [150]. The meta-analysis of the three countries gave an adjusted HR of 1.50 (95% CI 1.09–2.07) for neurodevelopmental disorders (autism spectrum, intellectual disability, communication disorders, ADHD and movement disorders), with an adjusted cumulative risk of about 5% against about 3%, with no difference in congenital malformations; the study did not assess fathers who stopped valproate more than 3 months before conception [38]. In the publication, neurodevelopmental disorders after weighting occurred in 5.6% against 3.2% in Denmark, 4.0% against 2.3% in Norway and 5.6% against 2.5% in Sweden; for malformations (Denmark and Norway), the unadjusted OR was 0.81 (0.48–1.36); the authors called for caution because of heterogeneity, and the duration of follow-up was longer in the valproate group [53, 151]. The PRAC acknowledged that causality could not be established and that specific types of neurodevelopmental disorder could not be identified, and considered the measures precautionary [38].

According to the direct healthcare professional communication (DHPC), agreed by the PRAC on 11 January 2024 and by the CMDh on 25 January and circulated in all Member States of the EU and the European Economic Area on 19 February 2024, the composite outcome of the PASS comprised ASD, intellectual disability, communication disorders, ADHD and movement disorders up to 11 years of age, mean follow-up was 5.0–9.2 years in the valproate group and 4.8–6.6 years in the comparator group, the adjusted cumulative risk was 4.0–5.6% against 2.3–3.2%, and the study was not powered for the subtypes and suffered from confounding by indication and from differing duration of follow-up [112]. The measures recommended in January 2024 are: initiation and supervision of valproate in male patients by a specialist in epilepsy, bipolar disorder or migraine; information on the potential risk; discussion of effective contraception for the patient and his partner during treatment and for at least 3 months after stopping it; regular reviews, chiefly if the patient plans to conceive and before contraception is stopped; no sperm donation during treatment and for at least 3 months afterwards; a male patient guide and card; and a direct healthcare professional communication [38]. The patient must not stop treatment without consulting his doctor, and if his partner becomes pregnant with paternal exposure in the preceding 3 months she should speak to her doctor; the EMA does not recommend termination of pregnancy in that situation [38]. The MHRA communicates a risk of about 5 in 100 children with a neurodevelopmental disorder, of uncertain causality, and recalls male infertility, possibly reversible [37]. In Portugal, the same measures were circulated by Information Circular no. 006/CD/550.20.001 and by the Pharmacovigilance Bulletin of 2024, with the Guide for the male patient [109, 115]. The FDA label of March 2026 contains no recommendations on paternal exposure [40].

The pattern of the studies is consistent with confounding by indication: the crude estimates against the population are high and approach 1 when the comparator is another ASM or when the analysis is restricted to fathers with epilepsy, and fathers treated with valproate have a greater polygenic burden of epilepsy, shared with ADHD and ASD [54, 142, 145]. The living meta-analysis did not include the PASS among the eligible studies [138]. In the opposite direction, the French EPI-PHARE report of November 2025, read in full, found an association [136]. On the basis of the EPI-MÈRES register of the SNDS, it included 2,832,850 live births from 2010–2015 with father and mother identified, excluding cerebral malformations, medically assisted reproduction and maternal exposure to valproic acid, valpromide, carbamazepine or topiramate; exposure was at least one paternal dispensing of valproate for epilepsy in the 4 months before conception (the press release speaks of 3 months), with 4773 exposed children and 3115 children of fathers treated with lamotrigine or levetiracetam, followed for a median of 11.9 and 11.2 years [114, 136]. At least one neurodevelopmental disorder occurred in 583 (12.2%) against 310 (10.0%); after propensity score weighting, the incidence rate was 11.0 (10.1–11.9) against 9.2 (8.1–10.2) per 1000 child-years (HR 1.24; 1.07–1.44), and disorders of intellectual development had an HR of 2.12 (1.02–4.30), with an incidence difference of 0.35 per 1000 child-years (0.01–1.03) and about 3.5 additional cumulative cases per 1000 children; ADHD (HR 1.24; 0.92–1.65), ASD (1.24; 0.82–1.87), communication disorders (1.23; 1.00–1.52) and learning disorders (1.08; 0.83–1.41) showed no significant increase in the main analysis [136]. In the sensitivity analyses, the HR for disorders of intellectual development remained above 2 when restricted to fathers with identified epilepsy (2.62; 1.07–6.41), when children with epilepsy (2.50; 1.07–5.83) or fathers with mental disorder (2.64; 1.17–5.96) were excluded, and against fathers with epilepsy not taking these drugs (2.10; 1.17–3.76), but not against fathers who had stopped valproate (2.32; 0.94–5.76) [136]. The authors present it as the largest study to date, more than 8 times the size of the Swedish one and 5 times that of the Danish one; it is a technical report of a public agency, which the PRAC included in the 2026 assessment [113, 136]. Comments published in 2024 regarded the 2024 measures as insufficiently founded [152, 153]. On 11 June 2026, the PRAC concluded that the evidence is inconsistent and that the causal role of valproate is uncertain, being unable to distinguish the effect of the drug from that of the father's condition; it maintained paternal exposure as an important potential risk and the 2024 measures until the final report of the TANGO PASS, a register study with father-mother-child linkage in Denmark, Finland, Germany, Norway and Sweden, with a planned start on 30 September 2026 and a final report planned for 30 March 2028 [39, 113, 147]. The recommendation, published on 6 July 2026 (EMA/PRAC/124080/2026), asks the marketing authorisation holders for a variation within 2 months to amend sections 4.4 and 4.6 of the SmPC and the package leaflet: the new text retains the HR of 1.50 and the risks of 4.0–5.6% against 2.3–3.2% from the PASS, adds that other population studies have shown no increase in risk in the children of men treated with valproate in monotherapy in the 3–4 months before conception, that differences in design and in control of confounding may explain the discordance and that factors other than valproate, including the father's disease, may contribute to the association, and maintains effective contraception, including by the partner, for at least 3 months after discontinuation and the prohibition on sperm donation [113]. In the package leaflet, the information to the patient becomes about 5 in 100 children against about 3 in 100, without certainty that the cause is valproate [113]. For nationally authorised medicines, the recommendation was sent to the CMDh for information; the CMDh position and the date of entry into force in the national SmPCs were not found in the sources consulted [113]. The UKTIS leaflet (November 2025) reports that studies indicate no increase in malformations and no good evidence of learning or behavioural problems after paternal exposure, recommending medication review only as a precaution, and MotherToBaby describes the results as mixed [75, 111]. In counselling a man treated with valproate, the current information is that the risk, if it exists, is lower than that of maternal exposure, that the European and British measures remain in force and that the decision to change drug should be taken with the specialist, without discontinuation on the patient's own initiative [37–39].

8.4. Prenatal diagnosis and options in the exposed pregnancy

There is no prenatal or preimplantation genetic test, because there is no causal variant. Prenatal diagnosis is by ultrasound (section 5.3) and addresses malformations; a normal result does not exclude the neurodevelopmental effect, which is the main consequence and has no prenatal marker [12, 40]. The options in the presence of a severe malformation depend on the legislation of each country and are discussed in a fetal medicine clinic; the Brazilian source consulted reports that Brazilian legislation permits termination of pregnancy in anencephaly and advocates prenatal diagnosis of risk in pregnant women with a previously affected child [98]. Teratology information services support counselling in the exposed pregnancy: in the United Kingdom, UKTIS (https://uktis.org/), with monographs for professionals on the TOXBASE platform (https://www.toxbase.org, access with registration) and Bumps leaflets for patients (https://www.medicinesinpregnancy.org/); in Europe, the ENTIS network (https://www.entis-org.eu/); in the USA, MotherToBaby of OTIS (https://mothertobaby.org/; telephone 866-626-6847) [51, 75, 111]. The Canadian Motherisk programme, also listed by Oxford [51], could not be confirmed as an active service: in October 2026, the domain responded with no content.

8.5. Counselling of families and of minors

Counselling of girls should begin before puberty or when treatment is started after menarche, because the PPP applies to all girls and women of childbearing potential and the FDA calls for regular counselling, chiefly at the onset of puberty [18, 40, 42]. In the child with FVSD, the diagnosis has family and legal implications: patient associations support the path to diagnosis, social and educational rights and, in some countries, compensation proceedings [154, 155]. The ERN-ITHACA consensus provides information sheets for patients and parents, for the paediatric team, for the family doctor (with the adult checklist) and for teachers [12]. Young adults with FVSD and their parents describe needs that extend beyond paediatric age, in particular in employment, independent living and relationships, with limited adult services [88]. Oxford anticipated in 2017 that questions about in utero exposure and autism would become more frequent, with long-term neurodevelopmental information still limited for many drugs [51]. The French PNDS asks that the diagnosis be communicated at a dedicated consultation, by the doctor who raises it, preferably with a psychologist present; the explanation of the characteristics of the syndrome, of care and of follow-up should be taken up again later, in greater detail and without the child present; the contact details of a patient association are given to the parents; psychological support is offered to the parents and the siblings; and a letter to the referring doctors summarises what was discussed [22]. Therapeutic education begins at the moment of disclosure and is addressed to the parents and, later, to the adult patient: the diagnosis, the consequences for the genetic counselling of each family member, the medical consequences and the planning of care [22].

9. Special Considerations: Population Diversity and Epidemiology

9.1. What each figure measures

The epidemiology of FVSD has three layers that should not be confused: the frequency of exposure, the risk in exposed pregnancies and the prevalence of the recognised syndrome. About 0.5% of pregnant women take anticonvulsants for epilepsy, to which are added exposures in bipolar disorder and migraine [3]. The risk in exposed pregnancies comes from the pregnancy registers (about 10% MCM; section 3.2) and from the neurodevelopmental cohorts (section 3.9). The birth prevalence of the recognised syndrome comes from congenital anomaly registers, which capture cases with a diagnosed malformation and not the exclusively neurodevelopmental effects, so they underestimate the full spectrum [1, 6].

9.2. Europe

The EUROCAT study by Morris and colleagues counted 28 cases of valproate syndrome in 2.74 million births from 15 European registers between 2005 and 2014; prevalence fell from 0.22/10,000 in 2005–2006 to 0.03/10,000 in 2013–2014, and 17 of the cases came from a single register, that of Réunion Island, without which the downward trend was no longer significant [6]. Orphanet presents these data by area:

The sources consulted neither explain the concentration of cases in Réunion nor demonstrate a population susceptibility. In France, the ANSM estimated 41,200–75,300 exposed live births and 2150–4100 children with MCM between 1967 and 2016, and the modelled estimate of 16,600–30,400 children with neurodevelopmental disorders is known only through the specialist press (section 3.11) [59, 89]. The French ANSM and CNAM cohort, with 1,721,990 live births, quantified the neurodevelopmental risk (HR 3.7) and the dose effect [50], and the French action plan reduced births from exposed pregnancies by 66% [90]. In the European DARWIN EU network (six countries), the incidence and prevalence of valproate use in women aged 12–55 years have fallen since 2015 [139]; in England, the 2023 alert accelerated the fall in prescribing [117]; in Scotland, a national programme for the surveillance of medicines in pregnancy linked more than 900,000 pregnancies to maternal and child outcomes [74]; and in Ireland, a Dublin maternity hospital documented poor compliance with the recommended folate dose in pregnant women on valproate [124]. By way of comparison, the mean prevalence of neural tube defects in Europe is 0.94 per 1000 pregnancies (live births and terminations), against 1.3 in Australia, 2.0 in Peru and 4.5 in China [41]. In the Nordic registers, the population-based studies (Denmark, SCAN-AED) provide the most complete denominators for neurodevelopment [16, 35]. France and the United Kingdom were the countries that requested the European referrals, with active family associations in both countries [11, 18].

9.3. USA and other regions

In the USA, the NAAPR register, with confirmation of malformations by a dysmorphologist, gives 9.2% MCM with valproate between 1997 and 2023 [8]; the prevalence of spina bifida in the general population is 6–7/10,000 births according to the national birth defects prevention network of the Centers for Disease Control and Prevention (CDC), against 100–200/10,000 after valproate exposure [40]. In 17 academic centres in the USA, between 2021 and 2022, valproate was prescribed to 492 of 12,908 pregnant women with epilepsy (3.81%) and to 1419 pregnant women in all, a use that persists despite the warnings [140]; in the USA, open spina bifida is more prevalent in the pregnancies of Hispanic mothers and less so in those of Black mothers [41]. In Australia, the 30-month register showed a higher mean dose in pregnancies with defects [3, 43], and valproate initiation fell from 2.06 to 0.75 per 1000 women aged 15–49 years per year between 2013 and 2021, in a country without an equivalent PPP, in which 69% of the women who started valproate in 2021 had no prescribed contraception [141]. In South Korea, 0.24% of pregnancies from 2013–2021 had exposure to ASM in the 1st trimester [46], and in Taiwan 0.45% of births from 2004–2016 [82]; paternal exposure to valproate affected 0.05% of conceptions in Taiwan [142]. The Indian report by Goyal and colleagues notes that valproate remains widely used in resource-limited countries [156], and a centre in Kerala documented recurrence of seizures in more than half of the women who changed drug [127].

9.4. Portugal and Brazil

In Portugal, the official sources consulted document the transposition of the European measures by INFARMED (circulars of 2018 and 2024, communication of 2017, Pharmacovigilance Bulletin of 2024) and the list of marketed valproate medicines (sections 7.2 and 7.4) [19, 109, 115, 116], but present no epidemiological data on FVSD or on the number of exposed pregnant women; the abstract of the 2026 EURAP analysis gives no data by country [73]. Their absence from the sources consulted does not demonstrate the non-existence of published Portuguese data. In Brazil, the national source consulted has no valproate-specific data: it reports a prevalence of neural tube closure defects of 1–5 per 1000 live births depending on the region, more common in females and in the children of mothers aged 20–24 years, with anencephaly in about one third of cases, includes the use of valproic acid for epilepsy among the risk factors (with maternal diabetes, obesity, zinc deficiency and hyperthermia) and recommends folic acid 4 mg/day despite the fortification of flours [98]. The sources consulted do not identify measures of the Brazilian regulatory agency on valproate in pregnancy.

9.5. Special populations

In the woman with idiopathic generalised epilepsy, valproate is one of the most effective ASM, which creates the central conflict between seizure control and fetal risk [42, 120]. In women with significant intellectual or physical disability, the ILAE/EAN accepted in 2015 valproate as first line when pregnancy is very unlikely, a judgement that in the current framework must respect the PPP [18, 120]. The case in section 10 illustrates the combination of maternal intellectual disability, obesity, absence of preconception consultation and polytherapy [91]. Exposure through breast milk is low and breastfeeding is encouraged [12, 40]. In men, the European and British measures have applied since 2024, and in the United Kingdom the age limit of 55 years covers both sexes [37, 38].

9.6. Reference networks and patient associations

ERN-ITHACA (the European reference network for rare malformation syndromes, intellectual disability and other neurodevelopmental disorders) is the reference network for FVSD and the author of the 2019 consensus, with working groups for neurodevelopmental disorders and for spina bifida; no update of the consensus later than 2019 was found [12, 20]. ERN EpiCARE (rare and complex epilepsies) is relevant to the management of maternal epilepsy and in 2026 hosts the living meta-analysis on paternal exposure [21, 138]. In France, APESAC (Association d'Aide aux Parents d'Enfants souffrant du Syndrome de l'Anti-Convulsivant) supports families, with sections devoted to treated women and men and to compensation [155], and Épilepsie-France works with people with epilepsy [110]. In the United Kingdom and Ireland, OACS (Organisation for Anti-Convulsant Syndrome), a charity registered in England and Wales under no. 1116497 and founded in 1999, supports families affected by fetal anticonvulsant syndromes, with a helpline, parent groups, an advocacy service for social benefits and special educational needs and information for professionals; its site is www.oacscharity.org, which was not responding in October 2026 and was consulted in an Internet Archive copy of June 2026, and the former domain oacs.org.uk now belongs to an entity unrelated to the condition [154]. The MHRA also lists the Valproate Victims group and epilepsy and mental health associations [37]. Oxford listed the FACS Association (facsa.org.uk) [3], whose site is no longer available (HTTP 410 response in October 2026). For spina bifida and hydrocephalus, Oxford lists the British association Shine (https://shinecharity.org.uk/) [41]. The "Fetal Anti-Convulsant Syndrome Community" operates on the RareConnect platform [154]. The pregnancy registers (NAAED, EURAP, UK Medicines and Pregnancy Registry) and the teratology information services (UKTIS and the Bumps leaflets in the United Kingdom, ENTIS in Europe, MotherToBaby in the USA; section 8.4) are also resources for patients [7, 37, 40, 75, 111]. In Portugal, the INFARMED CIMI answers requests for information on medicines, and adverse reactions are reported on the RAM Portal [19, 115]. In France, the PNDS of 2017 was drawn up by the reference centre for developmental anomalies and malformation syndromes of the West (Rennes University Hospital, with the competence centre of Vannes), with a multidisciplinary working group that included the president of APESAC [22]. The care network rests on the six regional reference centres for developmental anomalies (West, in Rennes; East, in Dijon; South-East, in Lyon; South-West, Occitania and Réunion, in Bordeaux; Île-de-France, at the Robert-Debré Hospital, in Paris; and North-West, in Lille), federated in FeCLAD, on the reference centres for rare diseases of brain development and intellectual disability and on the AnDDI-Rares and DéfiScience filières [22]. Coordination falls to the clinical geneticist in the forms with multiple malformations, to the neuropaediatrician in cerebral malformation, epilepsy or severe neurodevelopmental disorder, and to the child psychiatrist in autism or where psychiatric or behavioural disorders predominate; in complex situations, recourse is had to a reference centre for language and learning disorders (CRTLA), to an autism resource centre (CRA) or to a reference or competence centre for developmental anomalies [22]. Local support uses the CAMSP, the medico-psychological and medico-psycho-educational centres, the special education and home care services (SESSAD), the school inclusion units (ULIS), the school support assistant, the medico-educational institutes and the application to the departmental centre for people with disabilities (MDPH) [22]. Besides APESAC, the protocol lists the Alliance Maladies Rares, ADAPEI (support for people with intellectual disability and their families) and the associations by organ or malformation listed in Orphanet; contact with an association should always be offered, the decision resting with the family [22].

10. Illustrative Clinical Case

Kikuchi and colleagues described in 2016 the prenatal ultrasound diagnosis of FVSD in Japan [91]. The mother was 27 years old, nulliparous, had had epilepsy since the age of 1 year (treatment stopped at 7 years and relapse at 17), mild intellectual disability and a body mass index of 30 kg/m², and had had no preconception consultation; she was referred at 13 weeks. She was taking valproate 1400 mg/day (level 80 µg/mL; reference range 40–100), phenobarbital 140 mg/day (26.1 µg/mL) and carbamazepine 1200 mg/day (10.5 µg/mL), and had about three seizures a week despite the polytherapy, so reducing or changing the regimen was considered difficult; informed of the risks, she continued the pregnancy with oral folic acid 5 mg/day [91]. The 20-week ultrasound detected no malformations, with imaging limited by obesity. At 25 weeks, a right wrist contracture suggestive of a radial ray defect was observed (Figure 1a and 1b, https://pmc.ncbi.nlm.nih.gov/articles/PMC5094134/figure/Fig1/). From 32 weeks there was growth restriction, monitored by weekly ultrasound; at 34 weeks, the estimated fetal weight was 1660 g (−2.5 standard deviations, SD), the biparietal diameter 74.9 mm (−2.8 SD), the right ulna measured 35 mm (−8.4 SD) against 53 mm on the left (−1.1 SD) and the right radius was not visible. At 37 weeks, three-dimensional ultrasound showed a saddle nose (Figure 1c) [91].

Delivery was spontaneous at 39 weeks and 2 days; a girl was born weighing 2056 g with Apgar scores of 8/8, with mild generalised hypotonia, respiratory difficulty, flexion and contracture of the right wrist and absence of the right radial ray on radiography (Figure 2, https://pmc.ncbi.nlm.nih.gov/articles/PMC5094134/figure/Fig2/, in which panel a shows the flexed wrist with the fingers flexed over the palm and panel b, the radiograph, the shortened right forearm with absence of the radius and a radially deviated hand, with a left forearm of complete appearance) [91]. She also had a ventricular septal defect, an atrial septal defect and a patent ductus arteriosus, not detected before delivery, and dysmorphism with a broad nasal root, saddle nose, poorly marked philtrum and low-set ears. The ductus arteriosus was ligated at 13 days and the right forearm immobilised in plaster at 89 days. At 12 months there was developmental and growth delay [91]. No genetic study was performed, so the case does not meet the essential criterion of a normal microarray and fragile X test of the ERN-ITHACA criteria, published three years later; it does, however, have documented exposure, compatible dysmorphism and supportive points from the cardiac defect and the radial defect [12, 91].

The case illustrates several points: polytherapy with valproate at a dose above 1000 mg/day; the difficulty of changing treatment in an already established pregnancy with frequent seizures; the limitation of ultrasound in obesity, with normal morphology at 20 weeks; the radial defect as a later ultrasound sign; and the occurrence of malformations despite high-dose folic acid, consistent with the absence of demonstrated protection [42, 91]. Continuation of the polytherapy was a decision of the report and does not correspond to a routine recommendation. The dose of 5 mg/day of folic acid coincides with that of the RCOG and UKTIS, but not with that of the current North American guideline, which recommends ≥0.4 mg/day and advises against high doses routinely (section 7.7) [42, 75, 96]. A second case, published in India by Goyal and colleagues, describes a boy of 11 months, the child of a consanguineous couple, whose mother took 1000 mg/day of valproate throughout pregnancy: severe global delay (no head control and no speech at 11 months), weight, length and head circumference below the 3rd centile, a high forehead with bifrontal prominence, a depressed and broad nasal root, anteverted nares, a thin upper lip, a posterior cleft palate, absence of the radius and thumb on the left, a rudimentary thumb on the right, overlapping toes, a left testis in the inguinal canal and a 1.6 mm ostium secundum atrial septal defect [156].

11. Prognosis and Long-Term Follow-Up

11.1. Prognosis

Prognosis depends on the severity of the malformations and, in the long term, on the neurodevelopmental profile. In Kozma's clinical series, which brings together chiefly severe cases published between 1978 and 2000, 12% died in childhood and 29% of survivors had developmental delay or intellectual disability [4]; these figures do not represent exposed children as a whole. In the prospective cohorts, the mean IQ at 6 years lies 8–11 points below that of children exposed to other ASM (97 in NEAD), with 20–30% of children below average and a risk of ASD of 2.7% to 15% according to the study [12, 14–16, 35]. The deficits do not disappear with age: children may "grow into" difficulties that become evident only with school demands, so a single assessment in early childhood is not enough, and verbal and working memory deficits persist into adulthood, with an impact on independence and employment [12]. Facial dysmorphism becomes attenuated with age [12]. There is no known adult-specific medical disease, and the life expectancy of survivors without severe malformations is not described in the sources consulted [12]. Many children need educational support, and behavioural problems are frequent [28]; according to the ILAE, the need for educational support affects 29–48% of exposed children and 74% of children with clinically diagnosed embryopathy [9]. The ConcePTION survey describes, besides the neurodevelopmental deficits, physical multimorbidity (a median of 6 of 12 systems affected with doses ≥1000 mg/day), pain and limitation of activities that persist into adulthood, which broadens the picture described by the 2019 consensus, although the cross-sectional design by questionnaire to carers limits inference [87].

When FVSD includes open spina bifida, the prognosis follows that of spina bifida. More than 90% of cases have hydrocephalus and more than 70% of meningomyeloceles a Chiari II malformation, which Smith describes as present in practically all [41, 48]. In a cohort of 117 consecutive unselected newborns operated on for open spina bifida (Hunt and Oakeshott, 2003), reviewed at 35 years, morbidity and mortality correlated with the sensory level; according to Oxford, 54 were alive, and among survivors with a sensory level below L3, 71% had a cerebrospinal fluid shunt, 88% an IQ of at least 80, 67% walked 50 m (with aids if necessary), 33% were continent, 58% lived independently and drove and 38% were in open employment; the book does not state the denominator of these percentages [41]. The Brazilian source consulted associates a worse prognosis with thoracic levels, kyphosis, hydrocephalus, early hydronephrosis and associated anomalies, and a better prognosis with lesions below L1 [98].

11.2. Surveillance table

The intervals follow Table 5 of the ERN-ITHACA consensus; the agreement and level of evidence are given in brackets [12].

Appendix 8 of the French PNDS proposes a schedule by age, summarised in Table 11.2a; genetic tests are performed once only and the remainder according to the clinical picture [22].

Table 11.2a. Medical and allied health follow-up by age according to the French PNDS (2017, appendix 8) [22]

The other sources consulted (ILAE 2019, RCOG 2016, NICE NG217, UKTIS, European SmPC) do not define a paediatric surveillance schedule: the ILAE confines itself to recommending that, at the maternal postpartum consultation at about 6 weeks, questions be asked about the infant's feeding, growth and development, and the SmPC directs that neonatal coagulation be investigated and hypoglycaemia and hypothyroidism monitored [9, 47]. Oxford, in an earlier formulation, recommended vision screening, audiometry, formal developmental assessment in the presence of delay and follow-up by a community paediatrician in the presence of significant medical, developmental or behavioural problems, without defined intervals [3]. The ERN-ITHACA consensus provided for updating every two years, but no version later than that of 2019 was found in the sources consulted [12].

12. Summary and Concluding Remarks

Fetal valproate spectrum disorder is the embryofetopathy with the highest malformation rate among the antiseizure medications and a preventable cause of malformation and of neurodevelopmental deficit. Maternal exposure in monotherapy is associated with about 10% major congenital malformations (10.3% in EURAP, 9.2% in NAAPR), with lumbosacral spina bifida, cardiac defects, cleft palate, metopic craniosynostosis, hypospadias and radial ray defects, in a dose-dependent effect with no safe threshold; and with developmental problems in 30–40% of children of preschool age, with an IQ 8–11 points lower, a verbal < non-verbal discrepancy and an excess of ASD and ADHD, present also without malformations or dysmorphism [7, 8, 11, 14, 35, 36]. Diagnosis is clinical, requires documented exposure, a normal microarray and fragile X test and exclusion of other teratogenic disorders, and uses the ERN-ITHACA criteria of 2019; the methylation episignature is a biomarker under investigation [12, 17]. There is no specific treatment; follow-up is multidisciplinary, with neonatal renal ultrasound and echocardiography, ophthalmological assessment in the first 6 months and annually thereafter, communication screening at 18 months, annual reviews until school entry and at 10 and 13–14 years, and a planned transition [12].

Prevention depends on prescribing. The EU contraindicates valproate in women and girls of childbearing potential outside the pregnancy prevention programme of 2018, the United Kingdom has required since 2024 two specialists to start treatment before 55 years of age in both sexes, and the USA maintains a boxed warning without a formal programme [18, 37, 40]. The AAN/AES/SMFM guideline of 2024 recommends avoiding valproate and preferring lamotrigine, levetiracetam or oxcarbazepine, but also caution in withdrawing an effective drug in an already established pregnancy, because convulsive seizures carry risks of their own [42]. The folic acid dose differs between sources (≥0.4 mg/day in the North American guideline and in the ILAE; 5 mg/day in the RCOG and UKTIS; 0.4–5 mg/day in the ERN-ITHACA consensus), and there is no proof that it prevents the effects of valproate [9, 12, 42, 75, 96]. For paternal exposure, the Nordic safety study that led to the European precautionary measures of 2024 (aHR 1.50; about 5% against 3%) has not been replicated by Danish, Norwegian-Taiwanese, Swedish-Norwegian and Taiwanese studies or by a living meta-analysis (HR 1.03–1.06), with indications of genetic confounding by indication, whereas the French EPI-PHARE report found a slight increase in neurodevelopmental disorders (HR 1.24) and a doubling of intellectual disability (HR 2.12); on 11 June 2026 the PRAC maintained the measures until the final report of the TANGO PASS, planned for 30 March 2028, and requested an update of the SmPC [53, 54, 113, 136–138, 142, 144, 145, 147]. In Portugal, INFARMED transposed the PPP in 2018 and the measures for men in 2024, with no national epidemiological data in the sources consulted; the Brazilian source consulted deals with neural tube closure defects and includes valproate among their risk factors, with no data specific to FVSD [19, 98, 109].

13. References

1. Orphanet. Fetal valproate spectrum disorder (ORPHA:1906). Paris: INSERM; updated 2026-07-02. Available at: https://www.orpha.net/en/disease/detail/1906
2. Orphanet. Fetal anticonvulsant syndrome (ORPHA:370068). Paris: INSERM; updated 2026-07-02. Available at: https://www.orpha.net/en/disease/detail/370068
3. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Fetal anticonvulsant syndrome (FACS). p. 732–735.
4. Kozma C. Valproic acid embryopathy: report of two siblings with further expansion of the phenotypic abnormalities and a review of the literature. Am J Med Genet. 2001 Jan 15;98(2):168-75. PMID: 11223853.
5. Online Mendelian Inheritance in Man. 609442 Valproate embryopathy, susceptibility to [Internet]. Baltimore (MD): Johns Hopkins University; created 27 Jun 2005, edited 7 Feb 2020 [accessed 2026 Oct 17]. Available at: https://omim.org/entry/609442
6. Morris JK, Garne E, Loane M, Addor MC, Barisic I, Bianchi F, et al. Prevalence of valproate syndrome in Europe from 2005 to 2014: A registry based multi-centre study. Eur J Med Genet. 2018 Sep;61(9):479-482. doi:10.1016/j.ejmg.2018.05.008. PMID: 29753923.
7. Tomson T, Battino D, Bonizzoni E, Craig J, Lindhout D, Perucca E, et al. Comparative risk of major congenital malformations with eight different antiepileptic drugs: a prospective cohort study of the EURAP registry. Lancet Neurol. 2018 Jun;17(6):530-538. doi:10.1016/S1474-4422(18)30107-8. PMID: 29680205.
8. Hernandez-Diaz S, Quinn M, Conant S, Lyons A, Paik H, Ward J, et al. Use of Antiseizure Medications Early in Pregnancy and the Risk of Major Malformations in the Newborn. Neurology. 2025 Aug 12;105(3):e213786. doi:10.1212/WNL.0000000000213786. PMID: 40669027.
9. Tomson T, Battino D, Bromley R, Kochen S, Meador K, Pennell P, et al. Management of epilepsy in pregnancy: a report from the International League Against Epilepsy Task Force on Women and Pregnancy. Epileptic Disord. 2019 Dec 1;21(6):497-517. doi:10.1684/epd.2019.1105. PMID: 31782407.
10. Meador K, Reynolds MW, Crean S, Fahrbach K, Probst C. Pregnancy outcomes in women with epilepsy: a systematic review and meta-analysis of published pregnancy registries and cohorts. Epilepsy Res. 2008 Sep;81(1):1-13. doi:10.1016/j.eplepsyres.2008.04.022. PMID: 18565732.
11. European Medicines Agency. Valproate and related substances – Article 31 referral: CMDh agrees to strengthen warnings on the use of valproate medicines in women and girls. EMA/709243/2014. Amsterdam: EMA; 2014 Nov 21 [accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/medicines/human/referrals/valproate-related-substances
12. Clayton-Smith J, Bromley R, Dean J, Journel H, Odent S, Wood A, et al. Diagnosis and management of individuals with Fetal Valproate Spectrum Disorder; a consensus statement from the European Reference Network for Congenital Malformations and Intellectual Disability. Orphanet J Rare Dis. 2019 Jul 19;14(1):180. doi:10.1186/s13023-019-1064-y. PMID: 31324220. PMC6642533.
13. Jentink J, Loane MA, Dolk H, Barisic I, Garne E, Morris JK, et al. Valproic acid monotherapy in pregnancy and major congenital malformations. N Engl J Med. 2010 Jun 10;362(23):2185-93. doi:10.1056/NEJMoa0907328. PMID: 20558369.
14. Meador KJ, Baker GA, Browning N, Cohen MJ, Bromley RL, Clayton-Smith J, et al. Fetal antiepileptic drug exposure and cognitive outcomes at age 6 years (NEAD study): a prospective observational study. Lancet Neurol. 2013 Mar;12(3):244-52. doi:10.1016/S1474-4422(12)70323-X. PMID: 23352199. PMC3684942.
15. U.S. Food and Drug Administration. FDA Drug Safety Communication: Valproate anti-seizure products contraindicated for migraine prevention in pregnant women due to decreased IQ scores in exposed children. Silver Spring (MD): FDA; 2013 May 6 [accessed 2026 Oct 17]. Available at: http://www.fda.gov/Drugs/DrugSafety/ucm350684.htm (archived copy: https://web.archive.org/web/2014/http://www.fda.gov/Drugs/DrugSafety/ucm350684.htm)
16. Bjørk MH, Zoega H, Leinonen MK, Cohen JM, Dreier JW, Furu K, et al. Association of Prenatal Exposure to Antiseizure Medication With Risk of Autism and Intellectual Disability. JAMA Neurol. 2022 Jul 1;79(7):672-681. doi:10.1001/jamaneurol.2022.1269. PMID: 35639399.
17. Haghshenas S, Putoux A, Reilly J, Levy MA, Relator R, Ghosh S, et al. Discovery of DNA methylation signature in the peripheral blood of individuals with history of antenatal exposure to valproic acid. Genet Med. 2024 Oct;26(10):101226. doi:10.1016/j.gim.2024.101226. PMID: 39097820.
18. European Medicines Agency. Valproate and related substances – Article 31 referral: new measures to avoid valproate exposure in pregnancy endorsed. EMA/375438/2018. Amsterdam: EMA; 2018 [EC decision 2018 May 31; accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/medicines/human/referrals/valproate-related-substances-0
19. INFARMED – Autoridade Nacional do Medicamento e Produtos de Saúde, I.P. Valproato e ácido valproico – novas medidas para evitar exposição na gravidez. Circular Informativa n.º 054/CD/550.20.001. Lisbon: INFARMED; 2018 Mar 29 [accessed 2026 Oct 17]. Available at: https://www.infarmed.pt/documents/15786/2409571/Valproato+e+%C3%81cido+Valproico+-+novas+medidas+para+evitar+exposi%C3%A7%C3%A3o+na+gravidez/8674040d-5633-46c5-90ab-0e341f8c8b61
20. ERN-ITHACA. European Reference Network for rare malformation syndromes, intellectual and other neurodevelopmental disorders [Internet]. [accessed 2026 Oct 17]. Available at: https://ern-ithaca.eu/
21. ERN EpiCARE. European Reference Network for rare and complex epilepsies [Internet]. [accessed 2026 Oct 17]. Available at: https://epi-care.eu/
22. Haute Autorité de Santé. Protocole National de Diagnostic et de Soins (PNDS): Embryo-Foetopathie au Valproate. Saint-Denis La Plaine: HAS; 2017 Mar 21 [reference centre text of May 2017; accessed 2026 Nov 7]. Available at: https://www.has-sante.fr/jcms/c_2751900
23. Jones KL, Jones MC, del Campo M. Fetal hydantoin syndrome (fetal Dilantin syndrome). In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 792–793.
24. Meadow SR. Anticonvulsant drugs and congenital abnormalities. Lancet. 1968 Dec 14;2(7581):1296. doi:10.1016/s0140-6736(68)91781-9. PMID: 4177489.
25. Fedrick J. Epilepsy and pregnancy: a report from the Oxford Record Linkage Study. Br Med J. 1973 May 26;2(5864):442-8. doi:10.1136/bmj.2.5864.442. PMID: 4712481.
26. U.S. Food and Drug Administration. Drugs@FDA: NDA 018081 (Depakene), NDA 018723 (Depakote), NDA 019680 (Depakote Sprinkle), NDA 021168 (Depakote ER) [openFDA drugsfda database; accessed 2026 Oct 17]. Available at: https://api.fda.gov/drug/drugsfda.json
27. Robert E, Guibaud P. Maternal valproic acid and congenital neural tube defects. Lancet. 1982 Oct 23;2(8304):937. doi:10.1016/s0140-6736(82)90908-4. PMID: 6126782.
28. Jones KL, Jones MC, del Campo M. Fetal valproate syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 794–795.
29. DiLiberti JH, Farndon PA, Dennis NR, Curry CJ. The fetal valproate syndrome. Am J Med Genet. 1984 Nov;19(3):473-81. doi:10.1002/ajmg.1320190308. PMID: 6439041.
30. Ardinger HH, Atkin JF, Blackston RD, Elsas LJ, Clarren SK, Livingstone S, et al. Verification of the fetal valproate syndrome phenotype. Am J Med Genet. 1988 Jan;29(1):171-85. doi:10.1002/ajmg.1320290123. PMID: 3125743.
31. Phiel CJ, Zhang F, Huang EY, Guenther MG, Lazar MA, Klein PS. Histone deacetylase is a direct target of valproic acid, a potent anticonvulsant, mood stabilizer, and teratogen. J Biol Chem. 2001 Sep 28;276(39):36734-41. doi:10.1074/jbc.M101287200. PMID: 11473107.
32. Göttlicher M, Minucci S, Zhu P, Krämer OH, Schimpf A, Giavara S, et al. Valproic acid defines a novel class of HDAC inhibitors inducing differentiation of transformed cells. EMBO J. 2001 Dec 17;20(24):6969-78. doi:10.1093/emboj/20.24.6969. PMID: 11742974. PMC125788.
33. Wyszynski DF, Nambisan M, Surve T, Alsdorf RM, Smith CR, Holmes LB, et al. Increased rate of major malformations in offspring exposed to valproate during pregnancy. Neurology. 2005 Mar 22;64(6):961-5. doi:10.1212/01.WNL.0000154516.43630.C5. PMID: 15781808.
34. Meador KJ, Baker GA, Browning N, Clayton-Smith J, Combs-Cantrell DT, Cohen M, et al. Cognitive function at 3 years of age after fetal exposure to antiepileptic drugs. N Engl J Med. 2009 Apr 16;360(16):1597-605. doi:10.1056/NEJMoa0803531. PMID: 19369666. PMC2737185.
35. Christensen J, Grønborg TK, Sørensen MJ, Schendel D, Parner ET, Pedersen LH, et al. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism. JAMA. 2013 Apr 24;309(16):1696-703. doi:10.1001/jama.2013.2270. PMID: 23613074. PMC4511955.
36. Christensen J, Pedersen L, Sun Y, Dreier JW, Brikell I, Dalsgaard S. Association of Prenatal Exposure to Valproate and Other Antiepileptic Drugs With Risk for Attention-Deficit/Hyperactivity Disorder in Offspring. JAMA Netw Open. 2019 Jan 4;2(1):e186606. doi:10.1001/jamanetworkopen.2018.6606. PMID: 30646190. PMC6324310.
37. Medicines and Healthcare products Regulatory Agency. Valproate use by women and girls [guidance]. London: MHRA; published 2025 Jun 10, updated 2025 Sep 23 [accessed 2026 Oct 17]. Available at: https://www.gov.uk/guidance/valproate-use-by-women-and-girls
38. European Medicines Agency. Potential risk of neurodevelopmental disorders in children born to men treated with valproate medicines: PRAC recommends precautionary measures. Amsterdam: EMA; 2024 Jan 12 [accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/news/potential-risk-neurodevelopmental-disorders-children-born-men-treated-valproate-medicines-prac-recommends-precautionary-measures
39. European Medicines Agency. Meeting highlights from the Pharmacovigilance Risk Assessment Committee (PRAC) 8-11 June 2026. Amsterdam: EMA; 2026 Jun 12 [accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/news/meeting-highlights-pharmacovigilance-risk-assessment-committee-prac-8-11-june-2026
40. AbbVie Inc. Depakote (divalproex sodium) delayed-release tablets, for oral use: prescribing information. NDA 018723. North Chicago (IL): AbbVie; SPL revision of 2026 Mar 31 [accessed 2026 Oct 17]. Available at: https://api.fda.gov/drug/label.json?search=set_id:08a65cf4-7749-4ceb-6895-8f4805e2b01f (DailyMed setid 08a65cf4-7749-4ceb-6895-8f4805e2b01f)
41. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Neural tube defects. p. 500–505.
42. Pack AM, Oskoui M, Williams Roberson S, Donley DK, French J, Gerard EE, et al. Teratogenesis, Perinatal, and Neurodevelopmental Outcomes After In Utero Exposure to Antiseizure Medication: Practice Guideline From the AAN, AES, and SMFM. Neurology. 2024 Jun;102(11):e209279. doi:10.1212/WNL.0000000000209279. PMID: 38748979. PMC11175651.
43. Vajda FJ, O'Brien TJ, Hitchcock A, Graham J, Lander C. The Australian registry of anti-epileptic drugs in pregnancy: experience after 30 months. J Clin Neurosci. 2003 Sep;10(5):543-9. doi:10.1016/s0967-5868(03)00158-9. PMID: 12948456.
44. Tomson T, Battino D, Bonizzoni E, Craig J, Lindhout D, Sabers A, et al. Dose-dependent risk of malformations with antiepileptic drugs: an analysis of data from the EURAP epilepsy and pregnancy registry. Lancet Neurol. 2011 Jul;10(7):609-17. doi:10.1016/S1474-4422(11)70107-7. PMID: 21652013.
45. EURAP Study Group. Seizure control and treatment in pregnancy: observations from the EURAP epilepsy pregnancy registry. Neurology. 2006 Feb 14;66(3):354-60. doi:10.1212/01.wnl.0000195888.51845.80. PMID: 16382034.
46. Kim HK, Lee H, Choi SA, Lee HJ, Lee D, Moon HJ, et al. Prenatal exposure to antiseizure medications and the risk of congenital anomalies: a nationwide population-based study in South Korea. Neurology. 2026 Jan 13;106(1):e214350. doi:10.1212/WNL.0000000000214350. PMID: 41370743.
47. European Medicines Agency. Valproate and related substances – Article 31 referral: Annex III – Amendments to relevant sections of the summary of product characteristics and package leaflets. Amsterdam: EMA; 2014 [accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/documents/referral/valproate-related-substances-article-31-referral-annex-iii_en.pdf
48. Jones KL, Jones MC, del Campo M. Meningomyelocele, anencephaly, iniencephaly sequences. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 874–877.
49. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Limb reduction defects. p. 212.
50. Miranda S, Zureik M, Dray-Spira R, Blotière PO, Coste J, Weill A. Risque de troubles neuro-développementaux précoces (avant l'âge de 6 ans) associé à l'exposition in utero à l'acide valproïque et aux autres traitements de l'épilepsie en France: étude de cohorte à partir des données du SNDS. Saint-Denis: ANSM, CNAM; 2018 Jun [accessed 2026 Oct 17]. Available at: https://archive.ansm.sante.fr/var/ansm_site/storage/original/application/5763eecfdee86fd4df5062e8838dcece.pdf (also https://ansm.sante.fr/uploads/2021/01/15/rapport-valproate-rapport-ansm-cnam-juin-2018.pdf)
51. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Drugs in pregnancy. p. 718–721.
52. Danielsson BR, Danielsson C, Nilsson MF. Embryonic cardiac arrhythmia and generation of reactive oxygen species: common teratogenic mechanism for IKr blocking drugs. Reprod Toxicol. 2007 Jul;24(1):42-56. doi:10.1016/j.reprotox.2007.04.005. PMID: 17570632.
53. Colas S, Longin J, Santos AC, Kaplan S, di Giuseppe R, Bigat D, et al. Paternal Valproate Use and Neurodevelopmental Disorder and Congenital Malformation Risk in Offspring. JAMA Netw Open. 2025 Nov 3;8(11):e2542581. doi:10.1001/jamanetworkopen.2025.42581. PMID: 41212559. PMC12603866.
54. Olstad EW, Nordeng HME, Bjørk MH, Selmer K, Gervin K. Paternal valproate use and impact of shared genetic susceptibility on child neurodevelopment. Sci Rep. 2025 Nov 7;15(1):39033. doi:10.1038/s41598-025-23377-1. PMID: 41203668. PMC12595016.
55. Dean JC, Hailey H, Moore SJ, Lloyd DJ, Turnpenny PD, Little J. Long term health and neurodevelopment in children exposed to antiepileptic drugs before birth. J Med Genet. 2002 Apr;39(4):251-9. doi:10.1136/jmg.39.4.251. PMID: 11950853. PMC1735079.
56. Morrow J, Russell A, Guthrie E, Parsons L, Robertson I, Waddell R, et al. Malformation risks of antiepileptic drugs in pregnancy: a prospective study from the UK Epilepsy and Pregnancy Register. J Neurol Neurosurg Psychiatry. 2006 Feb;77(2):193-8. doi:10.1136/jnnp.2005.074203. PMID: 16157661.
57. Seo M, Park S, Kim TE, Lee KE, Ko PW, Shin KH. Fetal outcomes of antiseizure medication use during pregnancy: a nationwide retrospective cohort study. Epilepsia. 2026 Apr;67(4):1639-1653. doi:10.1002/epi.70067. PMID: 41452258.
58. Yerby MS. Management issues for women with epilepsy: neural tube defects and folic acid supplementation. Neurology. 2003 Sep 1;61(6 Suppl 2):S23-6. doi:10.1212/wnl.61.6_suppl_2.s23. PMID: 14504306.
59. Valproic acid and pregnancy: assessment of malformations in France. Prescrire Int. 2017 Dec;26:294 [translation of Rev Prescrire. 2017;37(405):513]. Available at: https://english.prescrire.org/en/7A6DDE61333FBABDFF411DF1E0C31266/Download.aspx
60. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Craniosynostosis. p. 378.
61. Jones KL, Jones MC, del Campo M. Cleft lip sequence. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 336.
62. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Congenital heart disease. p. 112 (Table 2.5).
63. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Coarse facial features. p. 106.
64. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Seizures with developmental delay. p. 308.
65. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Radial ray defects and thumb hypoplasia. p. 297.
66. Jones KL, Jones MC, del Campo M. Baller-Gerold syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 598.
67. Santos de Oliveira R, Lajeunie E, Arnaud E, Renier D. Fetal exposure to sodium valproate associated with Baller-Gerold syndrome: case report and review of the literature. Childs Nerv Syst. 2006 Jan;22(1):90-4. doi:10.1007/s00381-004-1089-x. PMID: 15789214.
68. Glover SJ, Quinn AG, Barter P, Hart J, Moore SJ, Dean JC, et al. Ophthalmic findings in fetal anticonvulsant syndrome(s). Ophthalmology. 2002 May;109(5):942-7. doi:10.1016/s0161-6420(02)00959-4. PMID: 11986102.
69. Jones KL, Jones MC, del Campo M. Septo-optic dysplasia sequence. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 880.
70. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Optic nerve hypoplasia. p. 270.
71. Mairesse R, Lepetit M, Dubucs C, Cifuentes EA, Damase-Michel C, Lacroix I. In utero exposure to medications and congenital eye anomalies. BMJ Open Ophthalmol. 2026 Jan 5;11(1):e002049. doi:10.1136/bmjophth-2024-002049. PMID: 41494749.
72. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Neonatal encephalopathy and intractable seizures. p. 247.
73. Perucca P, Malpas CB, Berry-Noronha A, Walker SP, Kalincik T, Moche Kamga B, et al. Antiseizure medication use in pregnancy and the risk of poor fetal growth in the EURAP international registry: a prospective cohort study. Lancet Neurol. 2026 Sep;25(9):840-851. doi:10.1016/S1474-4422(26)00216-4. PMID: 42586099.
74. Moore E, Millar M, Merrick R, Mueller T, Stark V, Jarvis L, et al. Pregnancy, baby, and childhood outcomes from using anti-seizure medication during pregnancy. Commun Med (Lond). 2025 Dec 18;6(1):28. doi:10.1038/s43856-025-01285-9. PMID: 41413341.
75. UK Teratology Information Service (UKTIS). Sodium valproate [Best use of medicines in pregnancy (Bumps) leaflet]. Version 4.1. Newcastle: UKTIS; 2025 Nov [accessed 2026 Oct 17]. Available at: https://www.medicinesinpregnancy.org/leaflets-a-z/sodium-valproate/
76. Cohen MJ, Meador KJ, Browning N, Baker GA, Clayton-Smith J, Kalayjian LA, et al. Fetal antiepileptic drug exposure: motor, adaptive, and emotional/behavioral functioning at age 3 years. Epilepsy Behav. 2011 Oct;22(2):240-6. doi:10.1016/j.yebeh.2011.06.014. PMID: 21783425.
77. Cummings C, Stewart M, Stevenson M, Morrow J, Nelson J. Neurodevelopment of children exposed in utero to lamotrigine, sodium valproate and carbamazepine. Arch Dis Child. 2011 Jul;96(7):643-7. doi:10.1136/adc.2009.176990. PMID: 21415043.
78. Shallcross R, Bromley RL, Irwin B, Bonnett LJ, Morrow J, Baker GA, et al. Child development following in utero exposure: levetiracetam vs sodium valproate. Neurology. 2011 Jan 25;76(4):383-9. doi:10.1212/WNL.0b013e3182088297. PMID: 21263139.
79. Adab N, Kini U, Vinten J, Ayres J, Baker G, Clayton-Smith J, et al. The longer term outcome of children born to mothers with epilepsy. J Neurol Neurosurg Psychiatry. 2004 Nov;75(11):1575-83. doi:10.1136/jnnp.2003.029132. PMID: 15491979.
80. Viinikainen K, Eriksson K, Mönkkönen A, Aikiä M, Nieminen P, Heinonen S, et al. The effects of valproate exposure in utero on behavior and the need for educational support in school-aged children. Epilepsy Behav. 2006 Dec;9(4):636-40. doi:10.1016/j.yebeh.2006.08.007. PMID: 17049311.
81. Bromley RL, Mawer GE, Briggs M, Cheyne C, Clayton-Smith J, García-Fiñana M, et al. The prevalence of neurodevelopmental disorders in children prenatally exposed to antiepileptic drugs. J Neurol Neurosurg Psychiatry. 2013 Jun;84(6):637-43. doi:10.1136/jnnp-2012-304270. PMID: 23370617. PMC4115188.
82. Lu ML, Hung TH, Chen VC, Chen YL. Risks of neurodevelopmental disorders after prenatal exposure to antiepileptic drugs. J Child Psychol Psychiatry. 2026 Sep 14. doi:10.1111/jcpp.70238. PMID: 42736039. Online ahead of print.
83. Honybun E, Malpas CB, Graham JE, Hitchcock A, O'Brien TJ, Vajda FJE, et al. Neurocognitive outcomes of in utero exposure to antiseizure medication: an Australian cohort study. Neurology. 2026 Jul 14;107(1):e218122. doi:10.1212/WNL.0000000000218122. PMID: 42247656.
84. Bromley R, Weston J, Adab N, Greenhalgh J, Sanniti A, McKay AJ, et al. Treatment for epilepsy in pregnancy: neurodevelopmental outcomes in the child. Cochrane Database Syst Rev. 2014 Oct 30;2014(10):CD010236. doi:10.1002/14651858.CD010236.pub2. PMID: 25354543.
85. Karain A, Shakhshir LA, Pell JP, Mackay DF, Nelson SM, Singh S, et al. Educational outcomes associated with prenatal exposure to antiseizure medications: a systematic literature review and meta-analysis. Seizure. 2025 Nov;132:37-45. doi:10.1016/j.seizure.2025.08.016. PMID: 40902391.
86. Li Y, Loring DW, Cohen MJ, Matthews AG, Birnbaum AK, Pennell PB, et al. Development of cerebral lateralization in children exposed to fetal antiseizure medications: a prospective cohort study at age 6 years. Epilepsia. 2026 May;67(5):2412-2424. doi:10.1002/epi.70103. PMID: 41553327.
87. Wells PA, Richardson JL, Astill D, Bluett-Duncan M, Bromley RL, Clayton-Smith J, et al. Beyond congenital anomalies, the impact of sodium valproate exposure in utero on long-term health and well-being: a contribution from the ConcePTION project. Epilepsia. 2026 Sep;67(9):4523-4535. doi:10.1002/epi.70330. PMID: 42347559.
88. Khanom S, Astill D, Astill N, Cozens J, Mann B, Garratt J, et al. The lived experience of young adults with Fetal Valproate Spectrum Disorder, and the perspective of their parents: a qualitative study. Epilepsy Behav. 2024 Mar;152:109680. doi:10.1016/j.yebeh.2024.109680. PMID: 38335859.
89. Valproate : entre 16 600 et 30 400 enfants atteints de troubles neuro-développementaux. Le Quotidien du Pharmacien [Internet]. 2018 Jun 25 [accessed 2026 Oct 17]. Available at: https://www.lequotidiendupharmacien.fr/formation/specialites-medicales/valproate-entre-16-600-et-30-400-enfants-atteints-de-troubles-neuro-developpementaux
90. Agence nationale de sécurité du médicament et des produits de santé. Comité de pilotage «antiépileptiques et grossesse» – Étude sur le risque de troubles neuro-développementaux précoces associé à l'exposition in utero à l'acide valproïque et aux autres traitements de l'épilepsie en France – Communiqué. Saint-Denis: ANSM; 2018 [accessed 2026 Oct 17]. Available at: https://ansm.sante.fr/actualites/comite-de-pilotage-antiepileptiques-et-grossesse-etude-sur-le-risque-de-troubles-neuro-developpementaux-precoces-associe-a-lexposition-in-utero-a-lacide-valproique-et-aux-autres-traitements-de-lepilepsie-en-france-communique
91. Kikuchi N, Ohira S, Asaka R, Tanaka K, Takatsu A, Shiozawa T. Prenatal sonographic diagnosis of fetal valproate syndrome: a case report. J Med Case Rep. 2016 Nov 3;10(1):312. doi:10.1186/s13256-016-1094-1. PMID: 27809899. PMC5094134.
92. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Cleft lip and palate. p. 103.
93. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Hypospadias. p. 186.
94. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Plagiocephaly and skull shape abnormalities. p. 280.
95. Boyd PA, Wellesley DG, De Walle HE, Tenconi R, Garcia-Minaur S, Zandwijken GR, et al. Evaluation of the prenatal diagnosis of neural tube defects by fetal ultrasonographic examination in different centres across Europe. J Med Screen. 2000;7(4):169-74. doi:10.1136/jms.7.4.169. PMID: 11202581.
96. Royal College of Obstetricians and Gynaecologists. Epilepsy in pregnancy. Green-top Guideline No. 68. London: RCOG; 2016 Jun [accessed 2026 Oct 17]. Available at: https://www.rcog.org.uk/media/rzldnacf/gtg68_epilepsy.pdf
97. European Medicines Agency. Valproate and related substances – Article 31 referral: Annex III – Amendments to relevant sections of the product information. Amsterdam: EMA; 2018 [EC decision 2018 May 31; accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/documents/referral/valproate-article-31-referral-annex-iii_en.pdf (Portuguese version: https://www.ema.europa.eu/pt/documents/referral/valproate-article-31-referral-annex-iii_pt.pdf)
98. Banhos MEH, Barbosa MA, Ribeiro EM. Defeitos do fechamento do tubo neural (DFTN). In: Ribeiro EM, coord. Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 160–162.
99. Orphanet. Fetal hydantoin syndrome (ORPHA:1912). Paris: INSERM; updated 2026-07-02. Available at: https://www.orpha.net/en/disease/detail/1912
100. Orphanet. Fetal carbamazepine syndrome (ORPHA:370076). Paris: INSERM; updated 2026-07-02. Available at: https://www.orpha.net/en/disease/detail/370076
101. Orphanet. Fetal trimethadione syndrome (ORPHA:1913). Paris: INSERM; updated 2026-07-02. Available at: https://www.orpha.net/en/disease/detail/1913
102. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Approach to the consultation: dysmorphism and developmental delay. p. 4.
103. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Deletions and duplications (microdeletions). p. 660–665 (del 9p, p. 662).
104. Roberts AE. Noonan syndrome. In: Adam MP, Bick S, Mirzaa GM, Pagon RA, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Initial posting 2001 Nov 15; updated 2025 Dec 4. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1124/
105. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Noonan syndrome and RAS-MAPK syndromes. p. 513.
106. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Epilepsy in infants and children. p. 410.
107. Jones KL, Jones MC, del Campo M. Appendix: pattern of malformation differential diagnosis by anomalies. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 1001–1048.
108. ClinicalTrials.gov [Internet]. Bethesda (MD): National Library of Medicine; API v2 search of intervention studies on valproate and pregnancy/paternal exposure [accessed 2026 Oct 17]. Available at: https://clinicaltrials.gov/api/v2/studies
109. Severiano A, Martins AS. Valproato e ácido valpróico: medidas para evitar a exposição na gravidez e a exposição paterna antes da conceção. Boletim de Farmacovigilância (INFARMED). 2024;28(9–10) [accessed 2026 Oct 17]. Available at: https://www.infarmed.pt/documents/15786/10005301/Boletim+de+Farmacovigil%C3%A2ncia%2C+Volume+28%2C+n%C2%BA+9+e+10%2C+setembro+e+outubro+de+2024/73002432-5c2e-f394-97e3-425606f58401
110. Épilepsie-France. Risque confirmé de troubles neurodéveloppementaux (TND) chez les enfants dont les pères épileptiques ont été traités par valproate [communiqué de presse]. Paris: Épilepsie-France; 2025 Nov 9 [accessed 2026 Oct 17]. Available at: https://www.epilepsie-france.com/wp-content/uploads/2025/11/Communique_EPILEPSIE_FRANCE_Etude_EpiPhare_Valproate_Depakine.pdf
111. MotherToBaby | Organization of Teratology Information Specialists (OTIS). Valproic acid [fact sheet]. Brentwood (TN): OTIS; 2026 Jan 1 [accessed 2026 Oct 17]. Available at: https://mothertobaby.org/fact-sheets/valproic-acid-pregnancy/
112. European Medicines Agency. Direct healthcare professional communication (DHPC): Valproate-containing medicines: new measures regarding the potential risk of neurodevelopmental disorders in children of fathers treated with valproate in the 3 months prior to conception. Amsterdam: EMA; 2024 [agreed by the CMDh 2024 Jan 25; circulated 2024 Feb 19; accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/documents/dhpc/direct-healthcare-professional-communication-dhpc-valproate-containing-medicines-new-measures-regarding-potential-risk-neurodevelopmental-disorders-children-fathers-treated-valproate-3-months-prior_en.pdf
113. Pharmacovigilance Risk Assessment Committee (PRAC). PRAC recommendations on signals adopted at the 8-11 June 2026 PRAC meeting. EMA/PRAC/124080/2026. Amsterdam: EMA; 2026 Jul 6 [accessed 2026 Oct 17]. Section 1.3, Valproate and related substances – neurodevelopmental disorders with paternal exposure (EPITT 20191). Available at: https://www.ema.europa.eu/en/documents/prac-recommendation/prac-recommendations-signals-adopted-8-11-june-2026-prac-meeting_en.pdf
114. EPI-PHARE (GIS ANSM-Cnam). Exposition paternelle au valproate pendant la période de conception: nouvelle étude en faveur d'un risque accru de troubles neuro-développementaux chez l'enfant [communiqué de presse]. Saint-Denis: EPI-PHARE; 2025 Nov 6 [accessed 2026 Oct 17]. Available at: https://www.epi-phare.fr/app/uploads/2025/11/EPI-PHARE_ANSM_CNAM_CP_valproate-peres_20251106.pdf
115. INFARMED. Medicamentos contendo valproato: medidas para evitar a exposição paterna antes da conceção. Circular Informativa n.º 006/CD/550.20.001. Lisbon: INFARMED; 2024 Jan 19 [circulated by the Regional Health Directorate of Madeira, 2024 Jan 31; accessed 2026 Oct 17]. Available at: https://www.madeira.gov.pt//Portals/53/Documentos/GASA/circular_informativa_163_2024_medicamentos_contendo_valproato.pdf
116. INFARMED. Infarmed divulga informação sobre valproato/ácido valproico na gravidez [press release]. Lisbon: INFARMED; 2017 Jun 29 [accessed 2026 Oct 17]. Available at: https://www.infarmed.pt/documents/15786/1879176/Comunicado+de+Imprensa+-+Infarmed+divulga+informa%C3%A7%C3%A3o+sobre+valproato+ou+%C3%A1cido+valproico+na+gravidez/1836cead-5a6a-4bb8-8712-12be018f9996
117. Galloway J, Jones M, Gibson M, Song K, Bechman K, Mohanraj R, et al. Mandatory specialist approval requirements reduce valproate prescribing: population-based evaluation of a National Patient Safety Alert. BMJ Qual Saf. 2025 Nov 25. doi:10.1136/bmjqs-2025-019415. PMID: 41290398. Online ahead of print.
118. National Institute for Health and Care Excellence. Epilepsies in children, young people and adults. NICE guideline NG217. London: NICE; 2022 Apr 27 [updated 2026 Aug 5; accessed 2026 Oct 17]. Available at: https://www.nice.org.uk/guidance/ng217
119. Nabbout R, Santilli A, Tyvaert L, Schweitzer C, Maillard L, Kuchenbuch M. Trends in paediatric antiseizure-medication use and costs in France, 2014-2023: a nationwide population-based analysis. Lancet Reg Health Eur. 2026 Feb 3;63:101594. doi:10.1016/j.lanepe.2026.101594. PMID: 41675438.
120. Tomson T, Marson A, Boon P, Canevini MP, Covanis A, Gaily E, et al. Valproate in the treatment of epilepsy in girls and women of childbearing potential. Epilepsia. 2015 Jul;56(7):1006-19. doi:10.1111/epi.13021. PMID: 25851171.
121. Liu Z, Hong Q, Huang L, Sha L, Peng A, Chen L. Women with epilepsy during pregnancy: a systematic review of current guidelines. Epilepsy Behav. 2025 Oct;171:110658. doi:10.1016/j.yebeh.2025.110658. PMID: 40876195.
122. MRC Vitamin Study Research Group. Prevention of neural tube defects: results of the Medical Research Council Vitamin Study. Lancet. 1991 Jul 20;338(8760):131-7. PMID: 1677062.
123. Li Y, Bromley R, Meador KJ, Bjørk MH, Pennell PB, Voinescu PE; ECAM. Folate supplement dosage for pregnant people with epilepsy: survey of international experts recommendation and current practices. Epileptic Disord. 2025 Dec;27(6):1209-1216. doi:10.1002/epd2.70097. PMID: 40884533.
124. Turner C, McIntosh T, Gaffney D, Germaine M, Hogan J, O'Higgins A. A 10-year review of periconceptual folic acid supplementation in women with epilepsy taking antiseizure medications. J Matern Fetal Neonatal Med. 2025 Dec;38(1):2524094. doi:10.1080/14767058.2025.2524094. PMID: 40588438.
125. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Epilepsy in adults. p. 417.
126. Burbano MC, Couper RG, Espino PH, Burneo JG. Outcomes of valproic acid withdrawal in females before or during pregnancy: a systematic review and meta-analysis. Seizure. 2026 Feb;135:96-104. doi:10.1016/j.seizure.2026.01.019. PMID: 41621126.
127. Kalyanasundaram SS, Koshy KG, Menon RN, Radhakrishnan A. Impact of valproate discontinuation on seizure control in women of reproductive age. Epileptic Disord. 2026 May 30. doi:10.1002/epd2.70293. PMID: 42217171. Online ahead of print.
128. Vajda F, O'Brien TJ, Graham J, Hitchcock A, Perucca P, Lander C, et al. Balancing foetal disadvantage and seizure freedom in women capable of pregnancy - the Australian pregnancy register experience. Epilepsy Behav. 2025 Nov;172:110663. doi:10.1016/j.yebeh.2025.110663. PMID: 40882496.
129. Sandoz Inc. Ritalin (methylphenidate hydrochloride) tablets, for oral use: prescribing information. DailyMed setid d6fb2750-cdab-4749-ba0d-7534840a5892; version of 2025 Feb 5 [accessed 2026 Nov 7]. Available at: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=d6fb2750-cdab-4749-ba0d-7534840a5892
130. European Medicines Agency. Slenyto (melatonin) prolonged-release tablets: EPAR — product information (summary of product characteristics), sections 4.1 and 4.2. EMEA/H/C/004425; first authorisation 20 Sep 2018; extension to neurogenetic disorders (II/0025) 26 Aug 2024; ADHD indication (II/0028) 17 Mar 2025; product information updated 10 Jun 2026 [accessed 7 Nov 2026]. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/slenyto
131. European Medicines Agency. Circadin (melatonin) 2 mg prolonged-release tablets: summary of product characteristics. EMEA/H/C/000695. Amsterdam: EMA; initial authorisation 2007 Jun 29, renewal 2012 Apr 20 [accessed 2026 Nov 7]. Available at: https://www.ema.europa.eu/en/documents/product-information/circadin-epar-product-information_en.pdf
132. Janssen Pharmaceuticals, Inc. Risperdal (risperidone) tablets, oral solution and orally disintegrating tablets: prescribing information. DailyMed setid 7e117c7e-02fc-4343-92a1-230061dfc5e0; version of 2026 Oct 23 [accessed 2026 Nov 7]. Available at: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=7e117c7e-02fc-4343-92a1-230061dfc5e0
133. INFARMED. Risperdal (risperidona) 1 mg/ml solução oral e comprimidos revestidos por película: resumo das características do medicamento. Janssen Farmacêutica Portugal. Infomed, med_id 7607 [accessed 2026 Nov 7]. Available at: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=7607&tipo_doc=rcm
134. INFARMED. Infomed: search of authorised and marketed medicines with the active substance methylphenidate (Rubifen, Rubifen Retard, Ritalina LA, Concerta, Atexam) [accessed 2026 Nov 7]. Available at: https://extranet.infarmed.pt/INFOMED-fo/
135. INFARMED. Infomed: search of authorised and marketed medicines with the active substance melatonin (Circadin, Slenyto, Melatonina Alter, Melatonina Generis, Melatonina Mylan, Melatonina Teva, Melatonina Noxarem, Melatonina Pharma Nord) [accessed 2026 Nov 7]. Available at: https://extranet.infarmed.pt/INFOMED-fo/
136. Botton J, Rios P, Drouin J, Miranda S, Zureik M, Weill A, Dray-Spira R. Exposition paternelle au valproate au cours de la spermatogenèse et risque de troubles neurodéveloppementaux chez l'enfant: étude nationale à partir du registre EPI-MÈRES. Saint-Denis: EPI-PHARE (ANSM-Cnam); 2025 Nov 6. 61 p. [accessed 2026 Oct 17]. Available at: https://ansm.sante.fr/uploads/2025/11/06/20251106-actu-epi-phare-rapport-valproate-pere-vf3.pdf
137. Christensen J, Trabjerg BB, Dreier JW. Valproate Use During Spermatogenesis and Risk to Offspring. JAMA Netw Open. 2024 Jun 3;7(6):e2414709. doi:10.1001/jamanetworkopen.2024.14709. PMID: 38833248. PMC11151155.
138. Christensen J, Trabjerg BB, van Gelder MMHJ, Meng LC, Lai HY, Igland J, et al. Risk of neurodevelopmental disorders associated with paternal use of valproate during spermatogenesis: a living meta-analysis-version 1. J Neurol Neurosurg Psychiatry. 2026 Sep 14;97(10):890-898. doi:10.1136/jnnp-2026-338454. PMID: 42120066.
139. Bellas L, Català M, Burn E, Guo Y, Du M, Verhamme K, et al. Secular trends in the use of valproate-containing medicines in women of childbearing age in Europe: a multinational DARWIN EU network study. Pharmacoepidemiol Drug Saf. 2025 Oct;34(10):e70232. doi:10.1002/pds.70232. PMID: 41099219.
140. Voinescu PE, Seliger J, Gerard EE, Birnbaum AK, Johnson EL, Pennell PB, et al. Prescribing practices of antiseizure medications at US academic medical centers for pregnant people with and without epilepsy. Neurology. 2025 Oct 21;105(8):e214219. doi:10.1212/WNL.0000000000214219. PMID: 41026990.
141. Begum M, Giles L, Ilomaki J, Rumbold A, Davies M, Grzeskowiak LE. Longitudinal Australian trends in the initiation of valproate medication among women of reproductive age and concurrent use of prescription contraceptives. Womens Health (Lond). 2026;22:17455057251414916. doi:10.1177/17455057251414916. PMID: 41699930.
142. Meng LC, van Gelder MMHJ, Chuang HM, Lai HY, Chen LK, Hsiao FY, et al. Valproate Use by Fathers and Risk of Neurodevelopmental Disorders in Children. NEJM Evid. 2026 Mar;5(3):EVIDoa2500254. doi:10.1056/EVIDoa2500254. PMID: 41733407.
143. Meador KJ. Evidence against Neurodevelopmental Risks for Children Whose Fathers Use Valproate. NEJM Evid. 2026 Mar;5(3):EVIDe2500341. doi:10.1056/EVIDe2500341. PMID: 41733402.
144. Razaz N, Soderling J, Tomson T, Werenberg Dreier J, Christensen J, Bjørk MH, et al. Risk of neurodevelopmental disorders associated with paternal use of valproate during spermatogenesis. J Neurol Neurosurg Psychiatry. 2026 Jul 15;97(8):671-679. doi:10.1136/jnnp-2025-337441. PMID: 41781215. PMC13479227.
145. Feng YA, Lin MC, Tseng CH, Wu CS, Tsai MH, Wang SH. Paternal Valproate Exposure and Offspring Neurodevelopmental Outcomes. Neurology. 2026 Sep 22;107(6):e218375. doi:10.1212/WNL.0000000000218375. PMID: 42617146.
146. Voinescu PE. Paternal Valproate and Offspring Neurodevelopment: A Non-Nordic Perspective on an Unsettled Question. Neurology. 2026 Sep 22;107(6):e218516. doi:10.1212/WNL.0000000000218516. PMID: 42617145.
147. European Medicines Agency. HMA-EMA Catalogues of real-world data sources and studies: PaTernal exposure to vAlproate, further iNvestiGation on the risk of NeuroDevelopmental Disorders (NDD) and Major Congenital Malformation (MCM) in Offspring: a non-interventional post-authorisation safety study (TANGO). EUPAS1000000707. Updated 2025 Sep 26 [accessed 2026 Oct 17]. Available at: https://catalogues.ema.europa.eu/study/1000000707
148. Tennis P, Eldridge RR; International Lamotrigine Pregnancy Registry Scientific Advisory Committee. Preliminary results on pregnancy outcomes in women using lamotrigine. Epilepsia. 2002 Oct;43(10):1161-7. doi:10.1046/j.1528-1157.2002.45901.x. PMID: 12366730.
149. Mawhinney E, Craig J, Morrow J, Russell A, Smithson WH, Parsons L, et al. Levetiracetam in pregnancy: results from the UK and Ireland epilepsy and pregnancy registers. Neurology. 2013 Jan 22;80(4):400-5. doi:10.1212/WNL.0b013e31827f0874. PMID: 23303847.
150. European Medicines Agency. EMA review of data on paternal exposure to valproate [press release]. Amsterdam: EMA; 2023 Aug 16 [accessed 2026 Oct 17]. Available at: https://www.ema.europa.eu/en/news/ema-review-data-paternal-exposure-valproate
151. Christensen J, Trabjerg BB, Dreier JW. Risk of Neurodevelopmental Disorders and Paternal Use of Valproate During Spermatogenesis. JAMA Netw Open. 2025 May 1;8(5):e2512139. doi:10.1001/jamanetworkopen.2025.12139. PMID: 40402499. PMC12100447.
152. Garey JD, Damkier P, Scialli AR, Lusskin S, Braddock SR, Chouchana L, et al. Paternal Valproate Treatment and Risk of Childhood Neurodevelopmental Disorders: Precautionary Regulatory Measures Are Insufficiently Substantiated. Birth Defects Res. 2024 Aug;116(8):e2392. doi:10.1002/bdr2.2392. PMID: 39189597.
153. Damkier P, Shechtman S, Diav-Citrin O, Panchaud A, Weber-Schoendorfer C, Hodson K, et al. Drifting Too Far From Shore: Paternal Valproate Statement by the European Medicines Agency (EMA). Pharmacoepidemiol Drug Saf. 2024 Oct;33(10):e70016. doi:10.1002/pds.70016. PMID: 39420753.
154. Organisation for Anti-Convulsant Syndrome (OACS). Registered Charity No. 1116497 [Internet]. Home page, About us and Aims and objectives. [accessed 2026 Oct 17; Internet Archive copy of 2026 Jun 11]. Available at: https://www.oacscharity.org/ (archive: https://web.archive.org/web/20260611061526/https://www.oacscharity.org/)
155. APESAC. Association d'Aide aux Parents d'Enfants souffrant du Syndrome de l'Anti-Convulsivant [Internet]. [accessed 2026 Oct 17]. Available at: https://www.apesac.org/
156. Goyal M, Gupta A, Sharma M, Mathur P, Bansal N. Fetal Valproate Syndrome with Limb Defects: An Indian Case Report. Case Rep Pediatr. 2016;2016:3495910. doi:10.1155/2016/3495910. PMID: 28003925. PMC5143702.