Valproate and the Polycystic Ovarian Syndrome: Final Thoughts

This commentary addresses four issues: (a) What is polycystic ovarian syndrome (PCOS)? (b) Is it particularly common among women with epilepsy? (c) Is PCOS attributable to epilepsy, valproate (VPA) use, both, or neither? And (d) Is there any practical significance to this issue? Polycystic ovarian syndrome (PCOS) is hyperandrogenic chronic anovulation, that is hyperandrogenism with menstrual, ultrasound, or endocrine evidence of anovulatory cycles 1. Hyperandrogenism is represented by elevated serum levels of testosterone, androstenedione, or dehydroepiandrosterone in its free or conjugated sulfated form. Hirsutism is also a characteristic feature, but its manifestation may vary considerably in relation to ethnic origin, even in the presence of hyperandrogenemia. Chronic anovulation is represented by (a) menstrual disorder (amenorrhea, oligomenorrhea, abnormal menstrual cycle intervals, or menometrorrhagia), and often by (b) ultrasound evidence of enlarged ovaries with multiple follicular cysts and increased stroma, and (c) abnormally low serum progesterone levels during the midluteal phase. Affected women, however, commonly revert spontaneously to having fertile ovulatory cycles. There is no known single specific cause. PCOS is a bona fide "syndrome" with multiple etiologies and contributory factors, including genetic (autosomal dominant), endocrine (ovary, adrenal), metabolic (obesity, insulin resistance), and neurologic (limbic, hypothalamic) causes. The broad definition of PCOS includes specific diagnostic entities that many today consider separately, such as the nonclassic forms of heredofamilial adrenal hyperplasia, ovarian virilizing tumors, various identifiable causes of hyperprolactinemia, and thyroid disorders. Our view is to include them under the rubric of the general syndrome and then to classify PCOS as to specific etiology or as idiopathic. Minimal criteria for designation should establish the occurrence of hyperandrogenic chronic anovulation and exclude hyperandrogenism or polycystic ovaries with no other concomitants. The prevalence of PCOS in the general female population has been a matter of debate, with estimates ranging from as low as 3–4% to as high as 18–19%, depending on the definitional parameters used and the population studied. Knockenhauer et al. 2 identified PCOS, consisting of hyperandrogenism (hyperandrogenemia and/or hirsutism) and oligomenorrhea, in 3–4% of women in a general community survey in the southeastern United States. Polson et al. 3 have reported polycystic ovaries (PCO) in 23% of a series of volunteers recruited among British hospital employees. Three fourths of these women (i.e., 18% overall) had menstrual disorders as well, and most of the women with PCO were reported to have either menstrual disorder, hirsutism, or both, thereby rating the designation of PCOS. Serum testosterone levels in the women with PCO, however, were similar to those of women with normal ovaries. Clayton et al. 4 identified a similar proportion of women as having PCO, specifically 22%, but only 30% of them exhibited menstrual disorders, and only 14% had hirsutism, again without a significant difference in serum testosterone levels. This series suggests that PCOS occurred in ∼6.6% of women. In our most recent control series of women of reproductive age, between 18 and 45 years 5, we have identified menstrual disorders (defined as amenorrhea, no menses for 6 months; oligomenorrhea, >32-day intervals; polymenorrhea, <26-day intervals; irregular cycles, >4 days' variability in interval; or menometrorrhagia, heavy menses and bleeding between periods) in 10% of 100 women in a primary care physician's community practice and 14% of 100 women who see a gynecologist in a teaching hospital setting. The combination of PCOS characteristics (i.e., hirsutism and menstrual disorder), were present in 4% and 6% of the two practices, respectively. The notion of an association between PCOS and epilepsy was first suggested in 1984 by Herzog et al. 6, who identified five women with menstrual disorders, hirsutism, and either hyperandrogenism or ovarian cysts among a series of 20 women with complex partial seizures of probable temporolimbic origin. The finding of 25% affected women when only 5% were expected, according to the generally accepted prevalence figures of the day, was considered a remarkable finding. In a subsequent investigation of 50 consecutive women outpatients with complex partial seizures of probable temporolimbic origin 7, 10 (20%) had menstrual disorders and hyperandrogenism, consistent with the diagnosis of PCOS. Bilo et al. 8 demonstrated PCOS features in three (15%) of 20 women with primary generalized epilepsy. Murialdo et al. 9 demonstrated combined hyperandrogenism and ovulatory dysfunction in 15 (15%) of 101 consecutive patients with epilepsy with similar frequencies of menstrual disorders and PCO in women with primary generalized and localization-related epilepsy. In our most recent investigation 5, eight (22.2%) of 36 consecutive women with unilateral temporolimbic epilepsy were found to have PCOS (combination of menstrual disorder, hirsutism, and hyperandrogenemia) as compared with 10 (5%) of 200 with menstrual disorders and hirsutism of the control women referred to earlier. This would suggest a statistically significant fourfold overrepresentation of PCOS (χ2 = 10.5, d.f. = 1; p = 0.001) for this select group. A possible association between PCOS and epilepsy has also been supported by evidence of a reverse association: the finding of EEG abnormalities, including paroxysmal epileptiform discharges, among 56.5% of anovulatory women, most of whom had PCOS 10. Epilepsy itself may play a role in pathogenesis. Reproductive endocrine disorders are unusually common among untreated women with epilepsy 7. Our own data have shown a 30% occurrence of PCOS among 20 untreated women with complex partial seizures of temporal lobe origin, comprising a group of patients with newly diagnosed or relatively mild refractory seizures 7, 11. The specific nature of reproductive endocrine disorders may be related to the laterality and focality of the temporolimbic discharges such that the occurrence of PCOS may be significantly more common with left than with right temporolimbic epileptiform discharges and possibly with right-sided nontemporal discharges. 12. There is now evidence to suggest that interictal as well as ictal temporolimbic discharges may exhibit a close temporal relationship with the development of abnormal patterns and levels of luteinizing hormone (LH) and prolactin secretion, independent of AED use 13, 14. These changes, moreover, may be associated with a significantly higher rate of occurrence of reproductive endocrine and menstrual disorders 5. Bilo et al. 15 found that untreated women with epilepsy have a higher luteinizing hormone pulse frequency than do normal controls. Drislane et al. 16 and Herzog et al. 5 have demonstrated higher gonadotropin-releasing hormone pulse frequencies with left- than with right-sided temporal foci. Kiely et al.(17) have found increased LH pulse amplitude in women with epilepsy. Increased LH pulse frequency and/or amplitude may promote LH secretion over that of follicle-stimulating hormone (FSH) 18, 19 and thereby promote gonadal steroid secretion without maturation of the follicle. This results in hyperandrogenism because the immature follicle is deficient in aromatase, which converts androgens to estrogens. Incomplete follicular maturation also results in anovulation and the collection of follicular cysts in the ovary. These features and the resulting menstrual disorder and hirsutism form the cardinal features of PCOS. The possibility that the apparent overrepresentation of PCOS in women with temporolimbic epilepsy may be related principally to medication use was considered unlikely in our 1986 investigation 7 because menstrual disorders occurred in similar proportions of medicated (53%) and unmedicated (60%) women, and PCOS was almost two and a half times more frequent among unmedicated (30%) than among medicated women (13%). None of the women in our early investigations 6, 7 took VPA. Medications included predominantly the enzyme-inducing AEDs: phenobarbital (PB), primidone (PRM), phenytoin (PHT), and carbamazepine (CBZ). In 1993, Isojarvi et al. 20 found a strikingly high occurrence of PCOs (>60%) in women with epilepsy who used VPA alone or in combination. The frequency was significantly greater than that among women with epilepsy who took CBZ alone (33%) or other drugs (14%), and was particularly high among women who started treatment before age 20 years. Among 23 women who took VPA and had ultrasound and endocrine investigations, four (17%) had hyperandrogenism (i.e., elevated serum testosterone levels), with three (13%) of these having the combination of hyperandrogenism and menstrual disorder suggestive of PCOS. Of note, none of the 67 women taking other AEDs or controls were reported to have hyperandrogenism. Moreover, none of the 15 untreated women with epilepsy were found to have menstrual disorders in comparison with eight (16%) of 51 normal controls. The finding of PCO in >60% of women who took VPA is impressive in comparison with controls in the study and the literature 3, 4. The finding of probable PCOS in 13% of women using VPA may be substantially greater than that in the general female population if the lower 3–6% prevalence rates are accepted but may not be impressive if the 18% rate is considered. The absence of the combination of hyperandrogenism and menstrual disorder in all other groups would suggest a lack of occurrence of PCOS in women with epilepsy, in the absence of VPA use, and apparently in normal controls. In their earlier investigation, Murialdo et al. 9 did not find an overrepresentation of PCO in 11 women with epilepsy who took VPA in comparison to women taking other AEDs. They did find, however, a statistically significantly greater occurrence of anovulatory cycles, as determined by luteal phase serum progesterone measurements. In their subsequent controlled study 21, they compared 65 women who had epilepsy, 21 taking VPA monotherapy, with 20 normal controls, and found that VPA use was associated with substantially and statistically significantly higher body mass index, androgen levels, and more ovulatory dysfunction (low luteal phase progesterone levels in 63.6%) than with CBZ or PB use or for normal controls. Isojarvi et al. 22 have suggested that the overrepresentation of PCOS among VPA-treated women with epilepsy may be attributable to VPA-induced weight gain and the resultant endocrine concomitants of insulin resistance (hyperinsulinemia, increased insulin-like growth factor and decreased insulin-like growth factor binding protein, and sex hormone binding globulin). These endocrine changes increase gonadal steroidogenesis and permit a greater proportion of the released serum testosterone to be bioactive. Increased bioactive androgen may act locally in the ovary to block ovulation or may accomplish this through aromatization to estrogen and negative feedback on FSH secretion. An interactional effect between AED use and epilepsy, nevertheless, must be considered 11. Specifically, there is the possibility that epilepsy may promote PCOS and that PCOS is treated by most enzyme-inducing AEDs but not by enzyme-inhibiting drugs such as VPA. Most commonly used AEDs, including barbiturates, PHT, and CBZ, induce cytochrome P-450 and accelerate hepatic biotransformation, whereas VPA does not. AEDs that induce hepatic enzymes reduce biologically active testosterone levels in the serum by increasing the binding and metabolism of testosterone. AEDs other than VPA, therefore, may treat hyperandrogenism and thus PCOS, whereas VPA therapy may not. This mechanism thereby could also contribute to a higher occurrence of PCOS in VPA-treated women with epilepsy. If this were the only mechanism, then VPA would not be the primary cause of PCOS, yet its selection as treatment may be an important factor. PCOS has neurologic as well as gynecologic and metabolic significance. Many steroids have both neuroactive and psychoactive properties. Estradiol, for example, blocks γ-aminobutyric acid (GABA) and promotes glutamate transmission 23. It has anxiogenic, proconvulsant, and convulsant effects. Progesterone and especially some of its reduced metabolites, such as allopregnanolone, exert potent GABAergic influences and have mood-stabilizing, sedative, hypnotic, and antiseizure properties 24. Seizure frequency in women with epilepsy correlates with serum estradiol-to-progesterone ratios 25. Anovulatory cycles are associated with more frequent seizures 25. PCOS is characterized by anovulatory cycles that expose the brain to continuous estrogen without cyclic progesterone effects. PCOS, therefore, may exacerbate seizures. There is also evidence to suggest that inadequate luteal phase cycles may play a role in the development of anxiety and mood disorders that frequently accompany PCOS 29, especially in the setting of anomalous brain substrates such as underlie epilepsy 26-28. Moreover, PCOS, like most reproductive endocrine disorders that are characterized by ovulatory dysfunction, may be associated with a higher frequency of migraine 29. Recognition of PCOS offers the opportunity to provide effective hormonal treatment that may have neurologic as well as reproductive benefits. Natural progesterone, for example, may reduce seizures 30-32 and exert beneficial sedative and mood-stabilizing effects 26-28 in addition to regulation of menstrual cycles. Gynecologically, one cannot dismiss the impact of disfiguring hirsutism, mittelschmerz, menstrual disorders, and infertility. Most important, however, the unopposed mitogenic effects of estrogen in PCOS have been linked to higher rates of endometrial carcinoma 33. Metabolically, PCOS is associated with insulin resistance and hyperlipidemia, which predispose to development of atherosclerotic cardiovascular disease and its complications 34, 35. In summary, our impression from the available evidence is that PCOS is a reproductive endocrine disorder with potentially significant neurologic and nonneurologic medical consequences. There is evidence from a number of investigations to suggest that PCOS occurs in 13–25% of women with epilepsy. The broad extent of this range may represent differences in the definition of PCOS and the effects of various epilepsy characteristics (e.g., epilepsy type, age at onset, seizure frequency and severity, laterality of EEG abnormalities, and AED use). The range of PCOS occurrence in women with epilepsy exceeds the 4—6% prevalence of PCOS found by a number of investigators in the general female population, but is comparable to the 18% proposed by Polson et al. 3. Both epilepsy and AEDs alter reproductive hormone levels and thereby may contribute to the development of reproductive endocrine changes. There is a differential effect of AEDs, however, in this regard. Two independent groups of investigators, Isojarvi et al. 20 and Murialdo et al. 21, have found that menstrual disorders, certain clinical, ultrasound, or endocrine manifestations of reproductive endocrine disorders, or PCOS appear to be more common with the use of VPA than with some other AEDs. The Isojarvi group has also demonstrated reversal of PCOS features in association with a change in AED use from VPA to lamotrigine (LTG) 36. There have been, moreover, animal investigations that also show ovarian and reproductive endocrine changes with VPA use 37 and suggest that this effect may differ from that of LTG 38. Although some limitations to these studies can be legitimately cited, the observations, in our opinion, cannot readily be entirely dismissed on these grounds. Moreover, plausible pathogenetic mechanisms for the induction of PCOS by VPA have been proposed 22, including mechanisms that may differentiate VPA effects from those of other AEDs 11, 36. Limitations of existing investigations include the absence of a uniform definition for PCOS. This reflects the ongoing uncertainty among reproductive endocrinologists who continue to debate which features (i.e., menstrual disorders, hirsutism, hyperandrogenemia, LH/FSH ratio, ultrasound demonstration of follicular cysts) are essential for the designation, and the categories of diagnoses that should be excluded from the designation. With regard to the latter, many argue that specific diagnostic entities, including some that represent the most common causes for the presentation of PCOS features, especially the nonclassic forms of heredofamilial adrenal hyperplasia, be excluded. The weakness of excluding these adrenal disorders is that the usual methods of excluding them such as measurement of serum levels of dehydroepiandrosterone sulfate or even, more recently, dehydroepiandrosterone sulfate and 17-hydroxyprogesterone, do not exclude forms of adrenal hyperplasia (e.g., 3-hydroxysteroid oxidoreductase deficiency) that are commonly detected by more extensive cortrosyn stimulation testing, which is frequently not done. It can even be argued that the generally accepted definition for a normal menstrual cycle interval (i.e., 21–35 days) may not be appropriate for women with epilepsy if it can be demonstrated that women who exceed a narrower range of cycle intervals (e.g., 26–32 days) have a significantly lower rate of anovulation or inadequate luteal phase cycles than do controls. This, in fact, is our impression and the basis for our continuing to use 26- to 32-day intervals as normal. Other limitations of some studies in the literature include the lack of large sample sizes to handle the relatively low frequencies of PCOS in women with epilepsy and the general population (i.e., ∼15–20% and 5%, respectively), the list of epilepsy- and medication-related variables noted earlier, and the selection of control groups. It is estimated that sample sizes of 88–160 subjects in each group would be optimal to demonstrate a significant difference between the two groups at the 0.05 level with a power of 0.80 39. These sample sizes, however, would have to be increased very substantially to take into consideration the number of epilepsy- and medication-related variables cited earlier. A lack of statistical power may explain frequently lacking corroborative evidence in investigations of reproductive endocrine changes in women with epilepsy. Another factor that increases the numbers of subjects required for investigations is the variable frequencies with which certain AEDs are used in various types of epilepsy (e.g., VPA may be used with proportionately greater frequency in primary generalized epilepsies, whereas PHT and CBZ may be used relatively more frequently with localization-related epilepsy). In our own investigation of reproductive endocrine secretion in women with unilateral temporolimbic epilepsy 5, the majority of women with localization-related epilepsy took CBZ and PHT. Only two of 36 women were receiving VPA monotherapy; one of the two had PCOS. Statistical comparisons among AEDs, however, lacked both significance and power. With regard to the selection of controls, women with major mood disorders who take VPA have been considered to help distinguish VPA versus epilepsy effects. The frequent association of mood disorders with EEG findings or functional imaging evidence of temporolimbic or frontal dysfunction, however, complicates the distinction 26-28, 40-43. Until large scale, multicenter, prospective, controlled investigations of reproductive function, reproductive endocrine function, and fertility are undertaken, some guidelines must be considered for the monitoring of these important functions in women with epilepsy regardless of which medication they take. The findings of weight gain, reproductive dysfunction, or a reproductive endocrine disorder should be recognized, evaluated, and treated appropriately. The contribution of AED use should be considered, and decisions regarding AED drug selection may now need to weigh the importance of a particular selection for seizure control against the possible reproductive endocrine consequences.