On the Association Between Valproate and Polycystic Ovary Syndrome: A Response and an Alternative View


Address correspondence and reprint requests to Dr. J. I. T. Isojärvi at Department of Neurology, University of Oulu, FIN-90220 Oulu, Finland. E-mail: jouko.isojarvi@oulu.fi

In their article in this issue of Epilepsia, Genton et al. (1) raise the important issue of an association of polycystic ovary syndrome (PCOS) with epilepsy and antiepileptic drugs (AEDs). They review the relevant literature extensively, but in our view, their presentation and interpretation of much of the existing data is biased and potentially misleading.

We fully agree with Genton et al. (1) that the issue of epilepsy-related reproductive endocrine disorders has been controversial (2–9). There are definitely data suggesting that epilepsy per se can be associated with development of reproductive endocrine disorders (2–4). However, there are also many studies suggesting that the role that AEDs play in the development of these disorders may be more important than the role of epilepsy itself (5–18). Many of the effects of AEDs on reproductive endocrine function have been confirmed in animal models (16,18).

Genton and his colleagues are highly critical of studies by our group (1). In our view, their criticism is unfounded and is based mainly on incorrect interpretations of the studies (6,12–14,17,19). Other related studies are also incompletely presented (8,9), and experimental animal data (16,18,20–22) are completely missing from their review. Our group of investigators have studied the endocrine effects of AEDs since the mid-1980s. Indeed, we have published the only long-term prospective studies on the reproductive and other endocrine effects of AEDs. Isojärvi et al. reported 1- and 5-year follow-up data in men and women who were given carbamazepine (CBZ) (5,7,11,23–26). Inexplicably, however, Genton et al. (1) did not include these prospective studies in their review of the endocrine effects of AEDs, even though they stress the importance of baseline measurements.

The major criticism that Genton et al. (1) raised against our studies is that they are retrospective and, therefore, biased with regard to patient selection. This is not the case. The studies were not retrospective, and, therefore, patient selection could not have been biased by data collected retrospectively (6,12,14). The methods and patient populations are carefully described in each of our articles. Genton and colleagues also criticize us for arguing that lamotrigine (LTG) is preferable to valproate (VPA) in young women with epilepsy, and even that use of VPA is contraindicated in this group. We never drew conclusions. In all of our articles, we have emphasized that while the findings raise concern, the benefits of VPA treatment must be weighed against possible risks (6,12,13,17). We believe that this should always be the practice when treating patients: the choice of any treatment should be based on a risk–benefit assessment. We have not advised physicians to replace VPA with LTG. Rather, we have merely stated that alternative treatments should be considered if VPA-related endocrine or metabolic problems are identified. In our article on obesity, hyperinsulinemia, and hyperandrogenism in women with epilepsy treated with VPA, we indicated that other effective medications, such as CBZ and oxcarbazepine (OCBZ), may be preferred in treating partial seizures in young women (12). In treating women with primary generalized epilepsies, LTG may be a reasonable alternative to VPA in some patients (13).

Genton et al. (1) extensively reviewed the literature related to PCOS and its pathogenesis. This is important. However, it is hardly relevant when trying to understand VPA-related endocrine problems. Many studies have shown that use of VPA is associated with endocrine changes that include polycystic ovaries (PCOs), hyperandrogenism, and menstrual disorders (6,8,9,12,18). In the studies by Isojärvi et al., these changes have not been defined as PCOS, because as a drug-related condition it may be different from the spontaneous form of PCOS.

In this response, we present important data related to the subject that were not included in the review by Genton et al. (1). Furthermore, we try to correct what we perceive as inaccuracies in their presentation which, we believe, results in a less biased interpretation of VPA-related reproductive endocrine problems based on the information currently available.


In their review, Genton et al. (1) discussed fertility rates in women with epilepsy and reviewed pertinent studies (27–31). They described the study by Webber et al. (28) as suggesting that fertility rates in women with epilepsy were close to, or only slightly below, average population values. However, in that survey, women with epilepsy had lower fertility than did women in general (28). Furthermore, the large study by Wallace et al. (32) in the U.K., omitted by Genton et al. (1), found significantly lower fertility rates in women with treated epilepsy than in the general population. Wallace et al. (32) concluded that their finding underlines the need for research to identify potentially preventable causes of low fertility rates in women with epilepsy.


We agree fully with Genton et al. (1) that epilepsy per se appears to be associated with an increased prevalence of reproductive endocrine disorders (2–4). However, the patients included in the studies by Herzog et al. (2,3) and Bilo et al. (4) cannot be regarded as representative of the general epilepsy population in the area where the studies were conducted. Furthermore, only 20 women participated in the study by Bilo et al. (3). In our analysis of all women that we had studied during the 1990s, PCOs or hyperandrogenism were associated with VPA treatment, but not with any particular type of epilepsy (15). Similarly, ovarian cysts develop after long-term VPA treatment in otherwise healthy, nonepileptic animals (16,18,20–22). As a whole, the results of our studies, the reports by Murialdo et al. (8,9), and recent experimental studies together suggest that AEDs play a more prominent role in the development of reproductive endocrine disorders in women with epilepsy than does epilepsy itself (5–18).


There are many reports since 1980 showing that AEDs have reproductive endocrine effects (5–19,23,25,33,34), and that these effects may be associated with changes in reproductive function (6–14). Most of the early studies focused on reproductive endocrine effects of AEDs in men with epilepsy; the first studies in women were published only at the beginning of the 1990s (5–9,12–15,17,19). CBZ induces the synthesis of sex hormone–binding globulin in women. This results in reduced bioactivity of the sex steroids. Reduced bioactivity of estradiol can lead to menstrual disorders whereas reduced bioactivity of testosterone may have a beneficial effect with regard to possible hyperandrogenism (5,7). However, VPA-related reproductive endocrine disorders seem to be far more frequent than CBZ-related menstrual disorders in women with epilepsy (6,8,9,14,17).

Margraf and Dreifuss (33) were the first to report amenorrhea as an adverse effect of VPA. The first report of a high incidence of PCO, hyperandrogenism, and menstrual disorders in women taking VPA was published in 1993 (6). Subsequently, an increased frequency of those kinds of disorders was reported in VPA-treated women from three centers in Finland, Norway, and the Netherlands (14); in girls aged 8–18 years taking VPA for epilepsy (17); and in two studies in Italian women (8,9). Animal studies have shown that VPA, but not LTG, induces changes in serum sex steroid levels and ovarian structure in female nonepileptic rats (16,18,22).

Genton et al. (1) selectively criticize some of the major studies in this field (6,12–14,17), and they review other studies inadequately or inaccurately (8,9). Important animal studies related to the topic (16,18,20–22) are not mentioned at all. In the following sections, we review the most important studies related to VPA and reproductive endocrine disorders in women with epilepsy, including some data that are missing from the review by Genton et al. (1).

The original report on a high frequency of PCO and hyperandrogenism in VPA-treated women

Genton et al. (1) found this study to be biased, mainly because patient material was selected retrospectively. This was not the case. The study was cross-sectional (6), as were most of the major studies in this field reviewed by Genton et al. (1–4,8–10,12,14,17,19,34–37). The only prospective studies on AEDs and reproductive endocrine function have been performed by Isojärvi et al., and most of these were not included in the review (5,7,11,13,23,25). Furthermore, our 1993 study (6) is the only one to date that included a large hospital-based patient population. In Finland, all patients from a region are seen in the major hospital serving the area. The population in the area of Oulu University Hospital is approximately 300,000, and virtually all epilepsy patients of working age in this region are seen in the Outpatient Department of Neurology at this hospital. All women between the ages of 18 and 45 with all types of epilepsy seen in the Outpatient Department of Neurology in Oulu University Hospital between April 1991 and March 1992 were interviewed for the study (n = 273). Thirty-five women were excluded because they were taking an oral contraceptive or other medication, had illnesses that interfered with pituitary–gonadal function, or had undergone hysterectomy. Accordingly, 238 women participated in the study (6).

Genton et al. (1) stated that we did not indicate the incidence of fertility problems. This is incorrect, as this information appears on page 1385 of the article. They also stated that PCO morphology was used as a surrogate for PCOS, but this, too, is a misinterpretation. We agree that PCO is only one of the criteria for PCOS. Consequently, we have not defined VPA-related reproductive endocrine problems as PCOS in our published articles (6,12–15,17). In our 1993 article, we stated, “Polycystic ovaries, hyperandrogenism, obesity, menstrual disturbances, and hirsutism are the clinical characteristics of the polycystic ovary syndrome, although the full-blown syndrome is relatively rare. These features also characterize the endocrine disorders associated with valproate therapy among women with epilepsy, especially those who began treatment in adolescence”(6).

In their review of the 1993 study, Genton et al. (1) stated that vaginal ultrasound examination was performed in all patients with menstrual disturbances and in “some others.” Thirteen of 16 women taking VPA who did not have menstrual disturbances had ultrasound examinations; altogether 23 of 29 women receiving VPA monotherapy were examined. Five of nine women taking CBZ and VPA who had regular menstrual cycles were examined. The figures were 28 of 97 for women taking CBZ, and 11 of 54 for women taking other medications. Altogether, 57 women with regular menstrual cycles and 47 women with menstrual disorders were examined (6).

Genton et al. (1) presented the frequency of PCO findings in this study inconsistently. The proportions for women with epilepsy are given as percentages of the whole patient group studied. However, in referring to control subjects, they stated that PCO was found in 5% of regularly menstruating control women, but they did not mention that the overall frequency of a PCO finding on ultrasound examination was 18% in control women (6).

The study on obesity, hyperinsulinemia, polycystic ovaries, and hyperandrogenism in VPA-treated women with epilepsy

When commenting on this study (12), Genton et al. (1) pointed to a bias due to its retrospective nature. This study, too, however, had a cross-sectional design. All available 65 women from the original study who were receiving either VPA (22 of the original 23 patients) or CBZ (43 of the original 49 patients) monotherapy participated. The patients were seen and examined at the Outpatient Department of Neurology, Oulu University Hospital (12).

The studies by Murialdo et al. related to reproductive endocrine effects of AEDs in women with epilepsy

Genton et al. (1) cited two articles by Murialdo et al. (8,9). However, the presentation did not include all the main results from these studies. In the first article, as mentioned by Genton and co-workers, Murialdo et al. found PCO in 21% of the patients with idiopathic generalized epilepsy and in 14% of the patients with localization-related epilepsy. Most of the patients were receiving polytherapy. Murialdo et al. divided patients receiving polytherapy into two groups according to whether they had taken VPA or not (8). Genton et al. (1) did not point out that Murialdo et al. found a 40% frequency of PCO among women taking polytherapy that included VPA but only a 13% frequency among women receiving polytherapy that did not include VPA. Furthermore, Genton et al. (1) omitted mentioning that Murialdo et al. found a 44.5% frequency of hyperandrogenic anovulation among women taking VPA (7). The total number of women with hyperandrogenic anovulation was 15 of 101 women. Among these 15 patients, 12 (80%) were taking VPA. Murialdo et al. stated, “our data agree with previous observations indicating a higher incidence of amenorrhea, PCO and hyperandrogenism during valproate treatment,” and made reference to our 1993 article (6).

With regard to the results of the second study by Murialdo et al., which included 21 women receiving VPA monotherapy, 21 women receiving phenobarbital (PB) monotherapy and 23 women receiving CBZ monotherapy, review by Genton et al. (1) does not indicate that the women taking VPA had elevated serum androgen levels, and that hyperandrogenic anovulation occurred in 63.5% of the women taking VPA (9).

The prospective study on endocrine and metabolic changes after replacing VPA with LTG

Sixteen women with a previously identified endocrine disorder (PCO or hyperandrogenism, or both) thought to be related to VPA treatment participated in the study (13). In their analysis of this report, Genton et al. (1) stated that the mean total number of ovarian follicles that were visualized in the epilepsy patients decreased from 20 receiving VPA therapy to 11 at 1 year after patients were switched to LTG. However, the figures cited did not refer to the number of ovarian follicles, but rather to the number of ovaries that fulfilled criteria for being polycystic. We quote from our article: “Only one of the women did not have polycystic ovaries on either side while taking VPA. Two women had polycystic ovaries on one side, and nine women had polycystic ovaries on both sides. After 1 year taking LTG, three women had polycystic ovaries on one side, and four women had such ovaries on both sides. Thus, the total number of polycystic ovaries decreased from 20 to 11 (p < 0.01, χ2 test).”

Genton et al. (1) wrote that the incidence of menstrual disorders in this study decreased from 44 to 13%. Actually, of the 12 women who completed the study, menstrual disorders were present in seven (58%) while they were taking VPA, and in two (17%) 1 year after VPA was replaced with LTG. Originally 16 patients had participated in the study, and 10 of these (63%) had menstrual disorders while taking VPA. However, one patient developed a rash from LTG, one patient became pregnant, and in two patients, VPA was reinstituted after LTG failed to provide equivalent seizure control. Therefore, the menstrual cycle patterns could not be analyzed in these four patients after switching medication.

Studies in young girls with epilepsy

All girls aged 8–18 years seen in the Outpatient Departments of Pediatrics or Neurology, Oulu University Hospital and who were taking VPA, CBZ, or OCBZ monotherapy for epilepsy and who did not have any exclusion criteria took part in these studies (17,19). Genton et al. (1) wrote that evidence of biologic hyperandrogenism was found in “some” girls taking VPA. Actually, elevated serum testosterone levels were found in 38% of the prepubertal, 36% of the pubertal, and 57% of the postpubertal girls (17). These frequencies would be considered high by people who treat young girls with epilepsy and who are concerned about the reproductive health of their patients. Genton et al. (1) claimed that in contrast to previous studies by Isojärvi et al., these two recent surveys apparently included all consenting patients. They inferred that the findings were not as spectacular as in previous studies, confirming the bias introduced by the mostly retrospective nature of previous reports by Isojärvi et al. in older women. As we have already stated, previous surveys by Isojärvi et al. on reproductive endocrine disorders in women taking VPA were cross-sectional, not retrospective (6,12–14). We found it surprising that Genton et al. (1) attempted to draw conclusions from the results of a later study on the methodology of earlier studies. Generally, the methodology of any study is usually assessed based on the data provided in the original article. We described carefully the methodology and patient population in each of our papers. As we stated in the articles (17,19), the older women had been taking VPA for ∼7–8 years, whereas the young girls had been taking VPA for only 2 years (5,16). Because both published data and clinical experience indicate that it may take many years for the endocrine abnormalities in question to develop, we suggested that the difference in findings between the older women and young girls might be due to the difference in the duration of medication exposure on the one hand, and to the differences between the developing and mature reproductive endocrine systems on the other (17). We were concerned by finding endocrine abnormalities in young girls, after only a few years of medication, and we could definitely not regard this as evidence against more severe problems in adult women (17).

The three-center study

Seventy-two women taking either CBZ or VPA alone took part in this study (14), which involved centers in three different European countries (Finland, Norway, Netherlands). Results were presented at a scientific meeting, and the abstract has been published. The full report has not yet appeared. Nonetheless, Genton et al. (1) claimed that the original group of VPA-treated women from Finland was included in this study. This is not the case. All Finnish subjects in the three-center study were new; they did not participate in the original study. This and all the other relevant information regarding the study was given in the oral presentation of the study, but it is difficult to give all details in the limited space of a short congress abstract.

Genton et al. (1) offer an unusual interpretation of the results of this study, stressing the important part played by hyperinsulinemia in the development of the VPA-related disorders. In fact, the three most important observations of this study are that the reproductive endocrine disorders in obese VPA-treated women are associated with hyperinsulinemia; that they are more severe in obese than in the lean women; and that they are also seen in the majority of lean VPA-treated women without hyperinsulinemia. This demonstrates that hyperinsulinemia is not necessary for the development of VPA-related reproductive endocrine disorders in women with epilepsy (14).

Animal studies related to the topic

Genton et al. (1) ignored recent important animal studies (16,18,20–22), the results of which have been presented at several international meetings (20–22). These studies demonstrated clearly long-term VPA treatment induces changes in ovarian morphology in rats by increasing the number of ovarian cysts, and altering sex steroid hormone levels. Interestingly, insulin levels are unchanged, indicating a unique mechanism of action of VPA not necessarily related to that of the PCOS in humans. Because changes are observed in nonepileptic animals, this argues against the hypothesis of epileptic activity as the primary cause of endocrine and morphologic changes in these animals. Further, serum concentrations measured 4–6 h after the last dose were well within the therapeutic range established in humans. On the contrary, LTG was not associated with ovarian morphologic changes in female rats (18). These findings support the view that VPA is associated with changes in ovarian structure and sex steroid serum concentrations in both human and animal subjects.


Analysis of scientific data must be based on sound and accurate presentation of the methods used, as well as the results. Unfortunately in the review by Genton et al. (1), many of the key studies related to VPA and endocrine effects are inaccurately or incompletely presented. Their conclusions are, therefore, misleading. Genton et al. (1) claimed that we advise physicians not to use VPA to treat young women and, rather, to use LTG instead. This is simply untrue. We have consistently and unambiguously expressed our belief that the findings and their impact on the health of young women with epilepsy warrant concern, and that if these kinds of problems emerge in women receiving VPA, alternative AEDs should be considered (6,12,13,17). In all cases, however, both risks and benefits of any medication change must be taken into account.

The original finding of a high frequency of PCO and hyperandrogenism in VPA-treated women has raised both interest and concern among epileptologists treating young girls and women with epilepsy. Therefore, after we completed our first study, we initiated other studies to (a) confirm whether these disorders are as common as the first survey indicated; (b) study the possible pathogenetic mechanisms leading to these endocrine changes; and (c) evaluate the reversibility of these abnormalities. Findings in our subsequent studies have been consistent. In the three-center study, the occurrence of reproductive endocrine problems were similarly high in VPA-treated women in three different countries (14). Hyperandrogenism was found in a large proportion of VPA-treated girls after only a few years of medication (17). Changes in the sex hormone profile also were seen in VPA-treated men (38). Most of the abnormalities were reversed when VPA was discontinued and another AED substituted (13). Other investigations have also shown a high prevalence of PCO, hyperandrogenism, and anovulation among VPA-treated women with epilepsy (8,9). These findings all indicate an association between VPA and reproductive endocrine changes in both women and men with epilepsy. These clinical data are supported by experimental observations that VPA induces changes in ovarian structure and serum sex steroid levels in female rats without epilepsy at concentrations approximating the therapeutic range in humans (16,18).

VPA is an efficacious broad-spectrum AED. However, when VPA-associated reproductive endocrine problems emerge, these should be carefully evaluated and an alternative drug considered, if appropriate. Monitoring body weight and menstrual cycles is important, as Genton et al. (1) suggested. However, we believe that baseline and subsequent information about ovarian structure and serum sex hormone concentrations is also important when evaluating possible endocrine effects of VPA treatment in women with epilepsy.

We advise professionals interested in the reproductive health of their young female patients with epilepsy to read the original articles and not rely solely on the view of Genton et al. We regard the safety and health of all our patients to be of the highest priority. Patients should be examined to identify possible endocrine (and other) adverse effects, and we should consider what actions might be indicated if abnormalities develop. The ethical principles of modern medicine and simply good clinical practice mandate that patients have the right to know possible untoward effects of treatment regimes on their health. Only by accurately prescribing and discussing available data in an honest and forthright manner can we make the most informed decisions to ensure the optimal care of our patients.