Introduction

The interactions between hormones, epilepsy, and antiepileptic drugs (AEDs) are complex. While there is ample evidence that hormones influence epilepsy, it is also apparent that epileptic activity influences hormones in both women and men. In addition, AEDs may disturb endocrine function. The clinical importance of these interactions is primarily related to the effects on reproductive hormones and is the focus of this article. Reproductive endocrine dysfunction is common among women and men with epilepsy [1]. Menstrual disorders, polycystic ovaries, and infertility have been described among women with epilepsy [1–3], while reduced potency and sperm abnormalities have been found in men [1, 4–5]. Sexual problems [6] and endocrine changes have been described frequently in both sexes [1–3]. There are also, however, interactions with other hormones, especially thyroid hormones (see [7]).

The close interplay between hormones, epilepsy, AEDs, psychosocial factors, comorbid conditions, and other non-epilepsy medication means that studying the effects of each factor in isolation can be problematic. However, animal models and experimental models using human tissue allow us to investigate the independent effects of the various factors.

Reproductive endocrine effects of epilepsy

The impact of the epilepsy itself on endocrine function has been recognised and discussed for many years. Clinical studies have shown that epilepsy itself affects the secretion of pituitary hormones, thereby affecting the secretion pattern, rhythmicity, and levels of the peripheral sex steroid hormones [1, 8]. Even laterality of epileptic activity may be important, as there are some indications that left-sided temporal foci increase the occurrence of polycystic ovaries in women, while right-sided foci increase the frequency of hypogonadotropic hypogonadism [9].

Anatomical and neurophysiological studies of tissue from animals and humans have shown that there is a close connection between the temporolimbic system and the hypothalamus, which controls the neuroendocrine system. The amygdala, in particular, has extensive direct reciprocal connections with regions of the hypothalamus that are involved in the regulation, production, and secretion of gonadotropin-releasing hormones (GnRH). In line with this, early studies demonstrated how amygdala-kindling in male cats led to hyposexuality [10], and limbic seizures elicited in female rats resulted in interruption of the estrous cycle and altered mating behavior [11]. Stimulation of the corticomedial amygdala can induce ovulation and uterine contractions in several species [12–13]. Bilateral amygdalectomy in adult male rats and cats has been found to cause marked degeneration of the testes [14], and bilateral ablations of the amygdala in adult female deer mice can induce anovulatory cycles and polycystic ovaries [15]. In addition, bilateral amygdalectomy in female monkeys induces amenorrhea and hypogonadal vaginal changes [16]. These studies demonstrate the close interactions between neuronal hyperactivity or hypoactivity in the temporolimbic brain areas and reproductive endocrine function.

A series of experiments conducted by Edwards and co-workers [17–18] showed the effects of seizures on reproductive functions in both female and male rats. In females, amygdala-kindled seizures halted ovarian cyclicity and also caused high serum estradiol concentrations, increased pituitary weight, theca cell hyperplasia, and polyfollicular ovaries consisting of many cystic follicles, as well as follicles in various stages of growth and atresia. Progesterone treatment, which is used to restore cyclicity, was effective in only 5 of the 28 animals that had stopped cycling, while all sham-kindled controls that had stopped cycling regained cyclicity. For intact male animals, amygdala-kindled seizures resulted in an increase in serum testosterone, estradiol, and prolactin, which was accompanied by a significant increase in testis, epididymis, and pituitary weight, as well as a significant decrease in prostate weight. Maximal electroshock seizures (MES) caused a short-term reduction in serum testosterone concentrations and in testis, epididymis, and prostate weight. These findings demonstrate that both focal limbic (amygdaloid) seizures and generalized (MES) seizures disturb the normal reproductive physiology in rats, independently of AEDs or other confounding factors.

The demonstration that different endocrine responses depend on the laterality of the epileptic activity provides a firm indication of the direct influence of epilepsy on endocrine function. Silveira and co-workers [19] reported asymmetric activation of hypothalamic regions after unilateral amygdala-stimulated seizures. By studying Fos-immunoreactivity in different hypothalamic areas, they identified three areas that are prominently involved in reproductive function, including the medial preoptic nucleus (MPO), the ventrolateral part of the ventromedial hypothalamus (VMHVL), and the ventral premammillary nucleus (PMV). These areas showed significantly greater and more asymmetric, ipsilaterally predominating induction of Fos following unilateral amygdala-stimulated seizures than other regions investigated, which are not involved in reproductive endocrine function (Fig. 1). Asymmetric activation of the hypothalamus could be the basis for the occurrence of such reproductive endocrine disorders in patients with left-sided or right-sided temporal lobe epilepsy, since there is a clear asymmetry in the reproductive functions of the hypothalamus, including asymmetric content of GnRH [20]. These animal studies provide a definitive demonstration of a direct effect of epilepsy and epileptic discharges on reproductive endocrine function.

Fig. 1
figure 1

Photomicrographs of hypothalamic nuclei from a right-amygdala-stimulated rat. a, c Similar numbers of Fos-ir neurons on the left and right-hand sides of the medial preoptic area (MPA) and paraventricular hypothalamic nucleus (PVH), areas not involved in reproductive endocrine function. b, d, e Laterally asymmetric, ipsilaterally predominating numbers of Fos-ir neurons in the medial preoptic nucleus (MPO), ventrolateral part of the ventromedial hypothalamic nucleus (VMH), and in the ventral premamillary nucleus (PMV), areas that are involved in reproductive function and reproductive endocrine secretion. (Copyright John Wiley & Sons Inc. [19])

Full size image

Effects of hormones on epilepsy

Both female and male sex steroid hormones influence brain excitability. Among the female sex steroid hormones, progesterone and its metabolites are anticonvulsant, while estrogens are mainly proconvulsant. Androgens are also mainly anticonvulsant, but their effects are more varied, probably due to the metabolism of androgens to, among others, estradiol.

Estrogens

Application of estrogen directly to the cortex has potent epileptogenic effects [21], while topical application of estrogen has even been used as a model of focal epilepsy (see [22]). Estradiol also reduces the electroshock threshold, increases paroxysmal spiking in epileptogenic foci in cats and rabbits, facilitates kindling, and potentiates seizures induced by different chemoconvulsants (see [23]).

Of considerable importance for the generally excitatory effect of estrogens is their ability to enable the rapid increase in responses of neurons to the excitatory effect of glutamate [24–26]. This potentiating effect is thought to primarily influence NMDA, but also non-NMDA types of glutamate receptors (see [26]). The excitatory responses to glutamate are increased by estradiol in a dose-dependent manner [24–25]. Typically, enhancement of glutamate excitation occurs within seconds after onset of local steroid application, indicative of an effect on membrane-bound glutamate receptors.

GABA-ergic mechanisms do not seem to be affected as an acute response [24, 27], but estrogens do affect the GABA-ergic system over time. Prolonged exposure (more than 24 hours) to estradiol suppresses GABA-ergic inhibition of hippocampal neurons that may be related to decreased GABA release at inhibitory synapses (see i.e. [28]). Also, it is assumed that estradiol decreases GABA synthesis by reducing the activity of glutamate decarboxylase (GAD) [29–31], although this was not observed in all studies [32].

Estradiol has also been found to alter brain morphology by increasing dendritic spine density via an NMDA receptor-dependent mechanism and altering the pattern of hippocampal synaptic connectivity [26, 33–35]. Estradiol selectively increased neuronal sensitivity to synaptic input mediated by the NMDA type of glutamate receptor, while responses mediated by the AMPA receptor were not affected. These continuous, plastic changes in morphology that are related to spine density and neuronal sensitivity to glutamate are probably of major importance with respect to the effect of estrogens on brain excitability.

Estrogens may be anticonvulsant in particular circumstances. As with other peripheral sex steroid hormones, estrogens also exert their effect at intracellular receptors, the estrogen receptors, ERα and ERβ. Of these, the effect of ERβ is the most important for the non-reproductive effects of the hormone, and some experiments have found that low doses of estrogen may actually reduce seizures by acting at ERβ (see [36]), while long-term estrogen exposure may decrease the susceptibility to kainate-induced seizures in some cases (see [28]).

Progesterone and its metabolites

Progesterone and its metabolites have been shown to increase the electroshock threshold [37, 81], protect against audiogenic and pentylenetetrazol (PTZ)-induced seizures, inhibit kindling, and also to protect against electroshock-induced seizures (see i. e. [23]).

The main mechanisms by which progesterone and its metabolites exert their effect on brain excitability are non-classical, being related to membrane receptors. Progesterone receptors are widely distributed in the brain (see [38]) and progesterone circulating in the blood gains rapid, and relatively unrestricted, access to all parts of the nervous system [39–40]. Neural tissue is able to convert progesterone to more potent antiepileptogenic progesterone metabolites within the brain (see [41]).

The main effects of progesterone and its metabolites are considered to be related to an enhanced postsynaptic GABA-ergic effect, as evidenced by several lines of investigation. Progesterone and its metabolites increase the inward chloride current induced by GABA [37, 42], increase the binding of muscimol [37, 43], stimulate the binding of flunitrazepam [44], and displace TBPS binding to the GABA receptor complex [44–45]. The effects on GABA-ergic inhibition of 3α-5α-THP, the most potent antiepileptogenic metabolite of progesterone, were investigated by Majewska et al [42], and dose-dependent increases in both peak amplitude and duration of the inward chloride current induced by GABA were demonstrated. In line with this, the effects of both progesterone and 3α-5α-THP on recurrent GABA-ergic inhibition have also been observed in hippocampal slices from rats [82].

The mechanisms by which the effect on GABA-ergic induced chloride influx is exerted and exactly where the steroid binds to the GABA-A receptor complex have been discussed. In the mid-1980 s the effect on chloride influx was shown to be largely related to an increase in the effective open time of the chloride channels [46]. However, although this may be the most important mechanism, later studies indicated that an increase in opening frequency may also be involved [47]. The binding site for progesterone and its metabolites within the GABA-A receptor complex is unique and differs from that of both barbiturates and benzodiazepines (BZ).

Progesterone and its metabolites may also affect excitatory mechanisms. A series of studies on Purkinje cells from a rat cerebellum [48–50], demonstrated that both progesterone itself and several of its metabolites, including 3α-5α-THP, decreased glutamate responsiveness after either systemic or topical application.

Seizure susceptibility may also be altered by changes in the subunit composition of the GABA-A receptor. The most striking finding following progesterone withdrawal, a model for catamenial epilepsy, is the marked up-regulation of the α-4 subunit of the GABA receptor [23, 51–54], which is insensitive to BZ. The increase in α-4 expression also leads to a decrease in inhibition gated by the GABA-A receptor (see [26]). Dynamic changes in the GABA receptor subunit composition in situations with progesterone withdrawal, which occurs premenstrually, also alter the seizure threshold and sensitivity to AED treatment on a cyclic basis. In this context, it is of relevance that some neurosteroids may be able to modulate all isoforms of GABA-A receptors, including those containing the α-4 subunit. This provides possibilities for specific treatment of women with catamenial epilepsy with new drugs, such as Ganaxalone – in essence a neurosteroid, which also acts at the α-4 subunit.

Although the effects on non-classical mechanisms are by far the most important for the role of progesterone and its metabolites as anticonvulsants, a possible effect on intracellular, classical progesterone receptors cannot be completely ruled out. A series of experiments, mainly on PTZ-induced seizures in ovariectomized rats, suggested a possible role for classical progesterone receptors as being relevant for the anticonvulsant effect [36].

Androgens

Although a general anti-seizure effect is most commonly observed, androgens have more varied effects [55]. Administration of androgens directly to the hippocampus of castrated rats reduces PTZ-induced seizures, while testosterone increases the electroconvulsive threshold in males at low dose, but in both sexes at higher doses (see [56]). The variable actions of testosterone may partly be due to its metabolism to 17β-estradiol, which is generally excitatory, but also to androstanediol and dihydrotestosterone, which exert potent antiepileptic effects (see [55]).

Regarding their role in epilepsy, androgens act primarily at non-classical, membrane receptors [57–58], as powerful and positive modulators of the GABA-A receptor. For instance, Reddy and Jian [58] showed how androstanediol produced a concentration-dependent enhancement of GABA-activated currents. Systemic doses of androstanediol (5–100 mg/kg) resulted in a dose-dependent suppression of seizures in a mouse hippocampal kindling model, which is a model of temporal lobe epilepsy, with high doses providing complete seizure protection.

Androgens also have effects on neuronal structure, increasing the number of spine synapses in the stratum radiatum of area CA1 in the rat hippocampus [59]. Androgens may also affect spine synapse density in the hippocampus in female rats and contribute to plastic changes over the course of the menstrual cycle [60].

Similar to the other sex steroids, androgens also act at classical, intracellular receptors, as demonstrated by the proconvulsant effect of flutamide, an antagonist of the intracellular androgen receptor. Further, it has been shown that testicular feminized mice, which are totally insensitive to androgens due to mutations in the intracellular androgen receptor, do not exhibit the anti-seizure effects from exposure to androgens that are observed in wildtype animals with intact intracellular androgen receptors (see [55]).

Endocrine effects of AEDs on sex steroid hormones

Several AEDs exert a direct effect on the production of sex steroid hormones. The first study to address this question in animals was published in 1990 and investigated the effects of valproate (VPA), carbamazepine (CBZ), and phenytoin on different steps of testosterone biosynthesis in isolated rat Leydig cells [61]. Using submaximally stimulating concentrations of human chorionic gonadotropin (hCG), leading to physiological testosterone secretion rates, half-maximal inhibition of testosterone formation occurred in the presence of 15 μM CBZ, 180 μM phenytoin, or 900 μM VPA. Only the values for CBZ were in the clinically therapeutic range.

Some, but not all, BZ may alter androgen production, perhaps due to the selective effects of different BZ on various types of BZ receptors. Peripheral-type BZ receptors have been characterized in various tissues like the ovary and testis. Clonazepam, which acts only on the central-type BZ receptors, did not affect androgen production, whereas diazepam, which binds to both central and peripheral BZ receptors, induced a significant increment of basal and hCG stimulated testosterone production [62]. However, other studies found that chronic treatment of male rats with diazepam is associated with lowered serum testosterone levels [63]. As there was no difference in serum the luteinizing hormone (LH) and the follicle-stimulating hormone (FSH), nor in the hypothalamic luteinizing hormone-releasing hormone (LHRH) content, it was suggested that diazepam could act directly on the testicular interstitial cells to reduce testosterone production [63]. This is supported by data that shows that the peripheral-type BZ receptor agonist Ro 5-4864 affects androgen production from suspensions of isolated rat interstitial cells, suggesting that BZ acting on peripheral BZ receptors have a direct effect on Leydig cells [62].

Long-term AED treatment may have profound effects on reproductive endocrine function. Due to the heated debate on the possible effects of VPA, the majority of studies focused on this drug. However, it should be borne in mind that the lack of reported effects on reproductive endocrine function from other drugs does not imply that they have been proven to be without such effects.

Long-term VPA treatment in female, non-epileptic rats has been shown to result in a marked increase in the testosterone/estrogen ratio, mainly by decreasing estrogen levels [64]. Regarding the gonadotropins, there was no increase in LH after VPA treatment; indeed, a trend towards reduced LH levels at high VPA doses was observed. No change was seen in FSH levels. Lamotrigine, on the other hand, did not affect any of the hormones studied. Taken together, the effects of long-term VPA treatment, with a pronounced reduction in estrogen concentrations, a notable increase in the testosterone/estrogen ratio, and minor effects, if any, on gonadotropins, suggest that VPA has a direct effect on the production of peripheral sex steroid hormones in the ovary.

In order to investigate the possibility of a direct effect of VPA on follicular steroidogenesis in more detail, the secretion of testosterone and estradiol from isolated porcine ovarian follicles has been studied [65]. In this study, co-cultures of theca and granulosa cells were used. Such co-cultures are thought to provide a better reflection of the in vivo situation than models using isolated cell cultures, as theca cells and granulosa cells work closely together in vivo as a functional unit. Using concentrations from 600 to 1500 μmol/l VPA, which includes clinically relevant concentrations, it was shown that VPA increased the secretion of testosterone, decreased estradiol secretion, and reduced the conversion of testosterone to estradiol. Two subsequent studies confirmed and extended these findings [66–67]. VPA was demonstrated to cause a significant increase in LH-stimulated testosterone secretion and a decrease in FSH-stimulated estradiol secretion (Fig. 2 and 3). VPA also decreased conversion of testosterone to estradiol in both basal and FSH stimulated cultures. This is of possible clinical importance as gonadotropins are always present in fertile women. These findings have been confirmed in studies using human ovarian follicular cells, which showed that VPA caused a significant and dose-dependent decrease in basal and FSH-stimulated estradiol secretion. Further, VPA reduced CYP19 aromatase activity in FSH-stimulated cells at higher concentrations [66].

Fig. 2
figure 2

The effect of levetiracetam (LEV) and valproate (VPA) on (a) basal and (b) LH-stimulated testosterone secretion. Significant differences compared with control. *p < 0.05; **p < 0.01. (Copyright John Wiley & Sons Inc. [67])

Full size image
Fig. 3
figure 3

The effect of levetiracetam (LEV) and valproate (VPA) on (a) basal and (b) FSH-stimulated estradiol secretion. Significant differences compared with control. *p < 0.05; **p < 0.01. (Copyright John Wiley & Sons Inc. [67])

Full size image

VPA may also alter enzyme activities and gene expression. In long-term cultures of human theca cells treated for 72 h with sodium VPA (30–3000 μmol/l), Nelson-deGrave and co-workers [68] observed an increase in basal and forskolin-stimulated dehydroepiandrosterone (DHEA), androstenedione, and 17alpha-hydroxyprogesterone production. The most pronounced effect of VPA on androgen biosynthesis occurred in the dose range of 300–3000 μmol/l, which includes therapeutic levels for the treatment of epilepsy and bipolar disorder. The study also showed that VPA can increase dehydroepiandrosterone (DHEA) and androstenedione and increase the expression of CYP17 and CYP11A genes, as seen in polycystic ovary syndrome. Thus, VPA may have a direct effect on steroidogenesis by affecting gene expression, converting normal theca cells to a polycystic ovary phenotype. In line with this, Gustavsen et al [69] used a model of human adrenal carcinoma cells that are capable of full steroidogenesis to show that VPA reduced estradiol levels and caused a general down-regulation of expression of genes encoding for enzymes early in steroidogenesis. Using the same cell line, gene analyses suggested that VPA affects NR0BI expression [70]. NR0BI inhibits promoters of other genes involved in steroidogenesis, and the altered expression of NR0BI might explain the observed down-regulation in hormone production. In the same study, expression of CYP19 was reduced following exposure to 900 μmol/l of VPA, which is a clinically relevant concentration. The CYP19 gene codes for aromatase catalyse androgens to estrogens, which is in line with the reduced conversion of testosterone to estradiol [67].

A direct effect of VPA is further supported by the findings of Hattori et al. [71], who demonstrated the presence of the enzyme microsomal epoxide hydrolase (mEH) in human ovaries. mEH is important in detoxification of various substances, and several studies have shown that VPA inhibits mEH activity [72]. Hattori et al. [71] showed that human granulosa cells expressed mEH, and that inhibition of mEH suppressed conversion of testosterone to estradiol. The action of VPA as an inhibitor of mEH may represent a route by which VPA reduces estrogen levels, thereby increasing the testosterone/estrogen ratio. This will lead to an androgen-dominant microenvironment in the ovary, and thereby possibly to polycystic changes, without an increase in LH levels [71].

However, a study in Rhesus monkeys [73] indicates a lack of endocrine effects after long-term treatment with VPA. Only 7 animals were studied, and this study also demonstrated a trend towards an increase in testosterone/estrogen ratio, lower estrogen levels, and an elevated LH: FSH ratio. In addition, body weight was significantly increased.

It is important that the potential direct effects of drugs other than VPA should also be investigated in more detail, especially “newer” AEDs. Levetiracetam (LEV) is of particular interest as it binds to the synaptic vesicle protein, SV2A [74]. SV2A is widely distributed in the nervous system and also in endocrine tissue. LEV may exert its effect both at the central and peripheral level as SV2A is expressed in the pituitary gland and the hypothalamus, but also in the ovary. In a primary investigation on the possible endocrine effect, we showed that LEV affected only basal, but not gonadotropin-stimulated, testosterone and estrogen secretion from porcine ovarian follicular cells [67]. This suggests that LEV could be an alternative drug for women of fertile age, as their gonadotropin status resembles the situation of gonadotropin stimulation. However, shortly after this first report was published, Svalheim et al [75] showed that long-term treatment of healthy female rats with LEV resulted in endocrine changes, with increased testosterone and reduced estradiol levels. In addition, there was a dose-dependent increase in ovary weight and an increase in the number of corpora lutea and secondary follicles, but no effect on number and dimension of ovarian cysts. These changes were observed at therapeutically relevant drug concentrations. It is also important to note that the changes differed from those associated with exposure to VPA, and thus represent a drug-specific pattern. Nevertheless, three more recent studies did not confirm any effect of LEV on hormone production and expression of genes related to steroidogenesis in both human adrenal cells (H295R cells) [69–70] and in human ovarian follicular cells [66]. In sum, at present LEV may be considered to have minor, if any, endocrine effects.

Topiramate is another drug that has been studied briefly with regard to endocrine effects, and was found to reduce fertility and ovarian weight in female rats that were exposed (100 mg/kg) for 12 weeks [76]. In male rats given the same dose for 8 weeks led to decreased spermatogenesis, sperm motility, and weight of reproductive organs [77]. However, although the serum concentrations were not measured and the clinical relevance of the observations is uncertain, these results underline the importance of studying all AEDs more closely with regard to possible endocrine effects.

Sex steroid hormones may also be influenced by AEDs via a centrally mediated effect by altering gonadotropin secretion. A study in male rats [64] demonstrated a dose-dependent increase in LH after long-term VPA treatment and also an increase in FSH at the highest dose. While this finding considered in isolation might imply a direct effect of the drug on the central nervous system, the increased gonadotropin levels may also be considered as compensatory mechanisms that are related to the marked peripherally induced effects observed in the animals. The lack of a centrally mediated effect is supported by the finding of unchanged FSH and prolactin levels in mice after 8 weeks of VPA treatment at doses that led to a reduction in pubertal maturation and alterations in ovarian and testicular function [78]. However, long-term VPA treatment has been shown to delay GnRH cell morphological maturation within the hypothalamus in young mice, although not in adults [79]. Long-term, low-dose, VPA or CBZ treatment in rats has also been shown to increase prolactin concentrations and reduce FSH and LH levels in male rats [80]. Furthermore, it should be remembered that GABA-ergic mechanisms are involved in the secretion of GnRH at the hypothalamic level, and several of our most commonly used AEDs exert their effects through actions on the GABA receptor. Thus, a centrally mediated effect of AEDs on sex hormone regulation cannot be ruled out and should be explored in further studies.

In conclusion, there is an intricate multidirectional interplay between epilepsy, sex steroid hormones, and AEDs. The complexity of interactions precludes investigation of the impact of each individual factor in clinical studies. Psychosocial aspects, comorbidity and use of drugs for other indications than epilepsy are further confounding factors. Animal models and experimental models using human tissue allow us to investigate the independent effects of the various factors. By comparing results from clinical and experimental studies several mechanistic questions regarding the complex interplay between epilepsy, hormones, and AEDs can be explored. Animal experiments should, therefore, be an integral tool in the study of reproductive endocrine disorders in epilepsy.