Polycystic Ovarian Syndrome in Women with Epilepsy: A Review


Address correspondence and reprint requests to Dr. S. Duncan at Clinical Neurosciences Centre, Hope Hospital, Salford, Manchester, U.K. E-mail: susanduncan@compuserve.com


Summary: Polycystic ovarian syndrome (PCOS) remains a controversial issue in women with epilepsy. The syndrome is characterized by clinical signs of endocrine dysfunction, such as irregular menstruation, hirsutism, and infertility, but its pathogenesis and presentation are heterogeneous. There are few data on the relationship between epilepsy and PCOS. Studies by a Finnish group have raised the issue of an association between valproate (VPA) and PCOS in young women with epilepsy. These studies, however, were retrospective, laid emphasis on polycystic ovary morphology rather than on clinical endocrine dysfunction, and were undertaken in selected populations. Further studies, both in Italy and Germany, failed to replicate the findings of the Finnish group. Future research should ideally be prospective and include baseline data in untreated women. No compelling data lead to a specific contraindication of the use of VPA in young women, and the drug remains a first-line treatment option.

Menstrual disturbances are said to occur more frequently in women with epilepsy, especially those with complex partial seizures originating in the temporal lobe (1–3). Some of the data quoted to support this contention have been published only in abstract form (4), and there are no prospective studies in the literature in which women's menstrual patterns were studied before and after the inception of antiepileptic therapy. In addition, few authors attempted to address the frequency of menstrual irregularity in the female population at large, which has been reported as varying between 11% and 27%(5,6).

More recently there have been several publications purporting to show an association with polycystic ovarian syndrome (PCOS), epilepsy, and antiepileptic drugs (AEDs). PCOS is one of the most controversial issues in reproductive biology (7). Not only is there some debate over the best definition of PCOS, but the syndrome itself is associated with heterogeneous clinical and metabolic features and has a multifactorial, essentially unknown, pathogenesis. I examine the pathophysiology of PCOS and consider the possible implications of epilepsy and AEDs in its development.


It is important, at the outset, to differentiate PCOS from polycystic ovarian morphology (PCO). PCO is defined by ultrasonographic and anatomic criteria as the presence of multiple ovarian cysts (>10 on a single ultrasonographic plane) measuring 2–8 mm in diameter and usually located in the periphery, but sometimes disseminated, and increased ovarian stroma and/or size (8). This definition remains controversial, however, because the ovaries may not be enlarged, making the identification of multiple follicles difficult. Multiple cysts in ovaries of normal size may be found in hypothalamic amenorrhea, as seen in the context of weight loss (9). A “polycystic ovary–like” syndrome with cysts in the ovaries can be found in women with Cushing's syndrome, thyroid disease, hyperprolactinemia, and congenital adrenal hyperplasia (10).

In Europe, PCOS is generally defined as PCO in the presence of one or more specific clinical signs of endocrine dysfunction, such as menstrual irregularity, hirsutism, or infertility. In the United States, however, PCOS is described as a metabolic syndrome, and anatomic changes need not necessarily be present (11). A National Institutes of Health (NIH) consensus conference specified three criteria for diagnosing PCOS (11):

  • • Presence of ovulatory dysfunction, polymenorrhea, amenorrhea, or oligomenorrhea;
  • • Clinical evidence of hyperandrogenism or hyperandrogenemia; and
  • • Exclusion of other endocrinopathies (e.g., Cushing's syndrome, hypothyroidism, late-onset congenital adrenal hyperplasia).

This more restrictive definition of PCOS excludes isolated findings of PCO, hyperandrogenism, or multifollicular ovaries.

Clinical features

Menstrual irregularity is one of the defining characteristics of PCOS. This may first manifest itself at puberty, with a delayed menarche followed by the onset of irregular periods or as the breakdown of a previously regular cycle within a few years, often associated with weight gain (12). Hyperandrogenemia may show itself as alopecia, hirsutism, acne, male-pattern balding, and male distribution of body hair. Infertility is often cited as a feature of PCOS; one retrospective study failed to confirm this (13). However, other workers have reported lower fertility rates including those women with PCOS given clomiphene (12). Approximately 50% of women with PCOS are obese, and 20% of those obese women will have either impaired glucose tolerance or non–insulin-dependent diabetes mellitus by the age of 40 years (14).

Biochemical features

A multitude of endocrine abnormalities are associated with PCOS, but none is pathognomonic or indeed constant. One of the more consistent findings is an elevated luteinizing hormone (LH) level in urine and serum, increased LH/follicle-stimulating hormone (FSH) ratio, and elevated androgen levels. Both estrogens and FSH remain in the normal range. Sex hormone binding globulin (SHBG) is lower than in normal subjects, potentially increasing free testosterone levels (10). Women with PCOS tend to be hyperinsulinemic, irrespective of whether they are obese, compared with normal women (15).


Epidemiologic studies in randomly selected populations show the prevalence of PCO morphology to be 17–22%(16–20)(Table 1). Balen et al. (21) found the prevalence of PCO to be 33% in a randomly selected group of university undergraduates. Nearly 80% of women with PCO have one other clinical symptom (Table 1). In women with other clinical symptoms of PCOS, the prevalence of PCO is even higher: 32% in women with amenorrhea, 87% with oligomenorrhea, and 87% with idiopathic hirsutism (9).

Table 1.  Prevalence of polycystic ovaries and clinical symptoms of polycystic ovarian syndrome in the general population
Study populationPCO (%)nClinical symptoms (%)Reference
  1. PCO, polycystic ovaries; PCOS, polycystic ovarian syndrome; MD, menstrual disorder.

U.K. (n = 235)21PCO: 48 411418Cresswell et al. (16)
  Non-PCO: 186 27215
Greece (n = 1,078)17PCO + MD: 14748 45 Botsis et al. (17)
  PCO − MD: 3618 20 
  Non-PCO: 5010 10 
New Zealand (n = 255)21PCO: 39234623 Farquhar et al. (18)
  Non-PCO: 14419204 
U.K. (n = 353)22PCO: 43143014 Clayton et al. (19)
  Non-PCO: 1657272 
U.K. (n = 257)22PCO: 116 76  Polson et al. (20)
  Non-PCO: 33 <1  

Using U.K. and European definitions, epidemiologic evidence suggests that ∼15% of women meet criteria for PCOS (22). With the more stringent U.S. criteria, the prevalence is between 5% and 10%(22). Strict NIH criteria provide an even lower estimate, with a recent study reporting a prevalence of 4%(23).

Most authorities accept that the metabolic abnormalities in PCOS are a consequence of genetic and environmental factors, with no single mechanism accounting for all cases of PCOS. PCOS or PCO is present in a large proportion (31–87%) of first-degree relatives (24–28). Racial factors appear to play no role (23). As stated earlier, it is generally thought that PCOS begins at or around puberty, and PCO can be detected by transabdominal ultrasonography in prepubertal girls (29,30).

There are three hypotheses for the pathogenesis of PCOS:

  • • LH hypothesis;
  • • insulin hypothesis; and
  • • ovarian hypothesis.

LH hypothesis

In the LH hypothesis (12), an altered LH/FSH ratio causes bias toward androgen synthesis and thereby the expression of PCOS and associated physical symptoms. Most women with PCOS have an abnormally high LH pulse frequency and amplitude; this is thought to be due to increased pituitary sensitivity to gonadotropin-releasing hormone. This increased LH pulsatility then drives the ovarian thecal cells to produce more androgens. Normally these androgens then diffuse into the granulosa cells where, under FSH control, aromatase converts them to estrone and estradiol. Androgen excess can thus be stimulated by excess LH, and the situation compounded if there is a relative reduction in FSH activity.

Insulin hypothesis

The insulin hypothesis links the pathogenesis of PCOS with insulin resistance/hyperinsulinemia. Women with syndromes of severe insulin resistance have significant hirsutism and PCO. Conversely, women with PCOS tend to be hyperinsulinemic and insulin resistant, independent of obesity; ∼50% of women with PCOS have a defect in postreceptor signaling at the insulin receptor (31,32). Insulin also decreases the release of SHBG by the liver, and thus increases free androgen levels. In vitro studies have shown that insulin drives androgen production in thecal cells (33). Suppression of insulin levels in women with PCOS (for example, with diazoxide) leads to a reduction in serum androgen levels, lending support to the insulin hypothesis (34).

Ovarian hypothesis

The ovarian hypothesis is derived from in vitro studies showing that women with PCOS have increased 17-hydroxyprogesterone levels in thecal cells (35). 17-Hydroxyprogesterone is an androgen precursor, and thecal cells from women with PCOS have been demonstrated to produce increased amounts of androgens after stimulation, compared with those from their non-PCOS sisters. This increase in 17-hydroxyprogesterone response implies there is dysregulation of the P450 c17 enzyme function in the ovaries, which may occur in the thecal cells (36). This enzyme can be induced to enhance androgen production by serine phosphorylation, and if it becomes dysregulated, could explain the changes seen in PCOS. Furthermore, 50% of women with PCOS have a defect of phosphorylation at the insulin receptor (32). The other strand of evidence in favor of an ovarian cause is the presence of many small follicles in the ovaries with high androgen-to-estrogen ratios. The granulosa cells are viable and respond to stimulation, leading some authorities to suggest that the actions of FSH are blocked at ovarian level (12).

The interrelationships between these three hypotheses remain enigmatic, and at present, the consensus is that in the majority of women, it is impossible to identify the etiologic factors behind the syndrome.


It is frequently claimed that menstrual irregularity is more common in women with epilepsy. There are no prospective studies of menstrual function in women with epilepsy, whether treated or untreated. Moreover, most of the studies appear to be small, and individuals either have severe enough seizure conditions to require attendance at a specialized epilepsy clinic, where they are recruited into studies designed to examine menstrual and reproductive dysfunction by investigators looking for such individuals, or they have been treated surgically for their epilepsy (3). This is hardly a representative cross section of the epilepsy community.

Studies in the United States by Herzog et al. (1,37) suggest an association between epilepsy and PCOS. In their first study, five (25%) of 20 women with temporal lobe epilepsy were reported to have PCOS, characterized by oligomenorrhea and hirsutism (1). Four of the five women were not taking AEDs; Herzog et al. (1) concluded that there was a significant increase in PCOS in women with complex partial seizures. However, further examination of these findings questioned the definition used by Herzog of PCOS. Three of the five women had galactorrhea, and hyperprolactinemia is not a feature of PCOS. One of the women had increased prolactin, one had galactorrhea with borderline prolactin, and prolactin was not measured in the third. We have no way of knowing from this article whether these women had pituitary adenoma or other endocrine causes of increased prolactin.

In a second study, Herzog et al. (37) studied 50 women with temporal lobe epilepsy, including 20 untreated and 30 AED-treated patients. Ten had features of PCOS. However, two of the 10 women had galactorrhea, and two had no menstrual dysfunction. Thus, only six satisfied the criteria for PCOS. Overall, the prevalence of PCOS in these two studies was 10–12%, which does not differ greatly from the expected prevalence of 4–10% in the United States. The obvious weakness of these studies lies in their retrospective nature and the failure to distinguish between PCO, true PCOS, and other endocrine causes of polycystic-appearing ovaries. Nonetheless, for over 15 years, the assertions made in these articles have remained essentially unchallenged and have proved to be the basis of further work by European and Scandinavian groups who accept the findings of the Boston group uncritically.

Increased PCOS has also been associated with generalized epilepsy. A study by Bilo et al. (38) in 20 women with idiopathic generalized epilepsy reported a 15% prevalence of PCOS (38). All but three of the 20 women were receiving AEDs. In a study by Murialdo et al. (39) in 101 women with epilepsy (36 with idiopathic generalized epilepsy, 65 with partial epilepsy) treated with various AEDs, 83 patients underwent pelvic ultrasonography, and PCO was detected in 17% (21% with idiopathic generalized epilepsy, 14% with partial epilepsy). No patient fulfilled NIH criteria for PCOS, although 13 patients did meet U.K. criteria. Thus, there was no difference in the incidence of PCOS in these women compared with that in studies of women not taking AEDs. Interestingly, 40% of the women taking polytherapy had PCO. Recently, Murialdo et al. (40) compared 65 women treated with AEDs [21 valproate (VPA); 21 phenobarbital (PB); 23 carbamazepine (CBZ)] with 20 healthy untreated controls and found no differences in the incidence of PCO, hirsutism, or other PCOS markers. It is worth noting that Murialdo et al. (40) used NIH criteria for diagnosis of PCOS and did not consider isolated PCO or hyperandrogenism as markers for the condition.

There is no doubt that both tonic–clonic and complex partial seizures are associated with increases in prolactin (41,42) and increased LH and FSH levels (43). These changes are transient, and any long-term effects on the hypothalamus–pituitary–ovarian axis remains speculative. It has been suggested that subclinical discharges in mesial temporal structures cause changes in LH and FSH pulsatility, leading to cystic changes in the ovaries, a theory that would concur with the LH hypothesis of PCOS mentioned earlier. Herzog (2) reported that women with left-sided interictal discharges were more likely to have PCO than were those with right-sided discharges. There were no ictal studies in this article; neither was there any mention of imaging; thus the conclusions, although intriguing, must be treated with some caution.

The issue of the role of specific AEDs, in particular VPA, in the pathogenesis of PCOS has recently been raised by a series of retrospective studies by Isojärvi and colleagues (44–46). In the first study (44), 238 women with epilepsy were compared with 51 healthy untreated controls. Twenty-nine (12%) women with epilepsy were treated with VPA, 120 (50%) with CBZ, 12 (5%) with a combination of these two drugs, and 62 (26%) with other AEDs; 15 (6%) women with epilepsy remained untreated. Menstrual disturbances were experienced by 45% women receiving VPA, 19% receiving CBZ, and 25% receiving a combination of both. However, unusually, no patient with untreated epilepsy had abnormal menstruation, which in the light of other studies is unusual and raises the question of ascertainment bias. The investigators then performed vaginal ultrasonography on the 98 women with treated epilepsy and a history of menstrual problems, and found PCO in 43% of the VPA group, 22% of the CBZ group, and 50% of the combination group (Table 2). They compared these data with an incidence of 5% in regularly menstruating normal controls. It should be noted that up to 80% of women with irregular periods are expected to have PCO, and therefore there was a selection bias in this comparison. Nonetheless, the finding is an intriguing one.

Table 2.  Polycystic ovaries in women with treated epilepsy and in regularly menstruating normal controls (44)
TreatmentnPCO (%)
  1. PCO, polycystic ovaries.

  2. From ref. 44, with permission.

Sodium valproate  
 Total2310 (43)
 Menstrual disturbances106
 Regular cycles134
 Total4911 (22)
 Menstrual disturbances217
 Regular cycles284
Sodium valproate and carbamazepine  
 Total84 (50)
 Menstrual disturbances32
 Regular cycles52
Other medications  
 Total182 (11)
 Menstrual disturbances71
 Regular cycles111
Regularly menstruating normal controls  
 Total432 (5)

The major problem with this article is the consistent failure to differentiate between PCOS and PCO, and the erroneous belief that PCO equals PCOS despite the large body of literature in the area of reproductive endocrinology to the contrary. It would appear that six (25%) of 23 women receiving VPA appeared to have PCOS by the U.K. definition, compared with 14% of the CBZ group. By NIH criteria, three (13%) of the VPA group appear to have PCOS compared with none treated with CBZ. These incidences are lower than those suggested by Isojärvi using PCO data alone, and do not differ significantly from estimations of the prevalence of PCOS in the population at large.

In a second study, 22 women receiving VPA were compared with 43 women receiving CBZ (45). In the VPA group, 64% of women had PCO, hyperandrogenism, or both. Interestingly, 50% of women treated with VPA had marked weight gain (mean, 21.0 kg). But once again, because of the confusion over PCOS and PCO, it is difficult to know just how many women actually had the syndrome and how many had only ultrasonic changes in their ovaries. Because of this it is hard to draw any firm conclusions from the study.

Isojärvi et al. (46) then studied the effects of switching 16 VPA-treated patients who had PCO to lamotrigine (LTG) (46). During the following year, patients lost weight, with significant decreases in body mass index, androgen levels, and number of ovarian follicles visualized. Fasting insulin levels decreased, and lipid profiles improved. It is worth noting that four patients withdrew, one due to pregnancy, one because of rash, and two because of relapse of their juvenile myoclonic epilepsy. What is most interesting, however, is that weight loss was so marked and was associated with biochemical improvement and the restoration of regular menstrual cycles in the women. It is known that weight loss in women with PCOS will restore menstrual regularity and reduce insulin and lipid levels. Taking the insulin hypothesis into consideration, it could be argued that when women who are predisposed to develop PCOS by virtue of genetic or other factors are exposed to an agent that causes weight gain, they are tipped into developing the condition.

In two further studies, the Finnish group examined the effects of VPA on girls between the ages of 8 and 18 years, comparing 40 girls taking VPA with a group of 17 taking CBZ and 18 taking oxcarbazepine (OCBZ) (47). A control group of 49 healthy untreated girls was included. There was no adverse effect of any of the drugs on growth or sexual maturation, nor were there any statistical differences between the groups and controls in weight or metabolic markers, with the exception of elevated insulin growth factor 1 (IGF-1) in patients taking CBZ and OCBZ compared with controls. In a follow-up study, 41 girls aged 8–18 years taking VPA were compared with a group of 54 controls (48). Elevated testosterone and free androgen indexes were found in some of the VPA-treated group. Interestingly, there was no clinical evidence of hyperandrogenism in those girls with increased testosterone, and there was no difference in the incidence of menstrual irregularity between those girls with increased androgens compared with those who had normal levels. This begs the question of how significant these levels of androgens are.

Most recently Bauer et al. (49) carried out a study of 93 women with epilepsy. PCOS was defined using NIH criteria. The incidence of PCOS was compared across four groups of women: an untreated group, those taking VPA or CBZ, and those receiving polytherapy. The incidence was 10.5% in the untreated group, 10% in the CBZ group, and 11.1% in VPA group. This study, like that of Murialdo et al. (40), has the great virtue of using an internationally accepted definition of PCOS, and of making the distinction between PCO and PCOS. Like Isojärvi, Bauer recruited women with established epilepsy who were already taking AEDs, so no comment can be made about the endocrine status of these women before the inception of drug therapy. It is interesting that using a study design similar to that of Isojärvi, the German group failed to replicate the Finnish findings.


In the United Kingdom, physicians are being urged to practice “evidence-based medicine.” Thus, the quality of evidence becomes crucial. In an area as sensitive as young people's future reproductive function, there is no place for retrospective studies in selected populations. Studies should be prospective and, given the complexity of the subject, designed in close collaboration with reproductive endocrinologists. Our understanding of the whole issue would also be considerably enhanced if the organizers of international meetings invited experts in reproductive endocrinology to lecture on the topic. There appears at present to be a belief that this most controversial and complex condition can be explained satisfactorily by psychiatrists and neurologists with no training in the area addressing drug-sponsored satellite symposia. This policy has led to confusion in the minds of neurologists and has let drug companies set the agenda.

Conclusive evidence has yet to establish whether PCOS is more frequent in women with epilepsy than in the general population and whether there are relationships between specific AEDs and PCOS. Long-term prospective studies in women with epilepsy are needed to examine these issues. Importantly, such prospective studies should include baseline data.

If neurologists are concerned by the literature on PCOS and epilepsy, it would be reasonable for them to use VPA with caution in young women with a family history suggestive of PCOS. If young women have a marked weight gain or develop menstrual irregularity, they should be referred to an endocrinologist for assessment. To date, VPA remains a first-line option for the treatment of epilepsy in young women.