Changes in sex steroid levels in women with epilepsy on treatment: Relationship with antiepileptic therapies and seizure frequency


Address correspondence to Carlo Andrea Galimberti, M.D., Epilepsy Centre, IRCCS “C. Mondino Institute of Neurology” Foundation, Via Mondino, 2 I-27100 Pavia, Italy. E-mail:


Purpose: Reproductive dysfunction in epilepsy is attributed to the seizures themselves and also to antiepileptic drugs (AEDs), which affect steroid production, binding, and metabolism. In turn, neuroactive steroids may influence neuronal excitability. A previous study in this cohort of consecutive women with epilepsy showed that patients with more frequent seizures had higher cortisol and lower dehydroepiandrosterone sulfate levels than those with rare or absent seizures. The present study was aimed at evaluating, in these same women, the possible relationship between some clinical parameters, seizure frequency, AED therapies, and sex hormone levels.

Methods: Estradiol (E2), progesterone (Pg), sex hormone-binding globulin (SHBG), and free estrogen index (FEI) were measured during the luteal phase in 113 consecutive females, 16–47 years old, with different epilepsy syndromes on enzyme-inducing AED (EIAED) and/or non–enzyme-inducing AED (NEIAED) treatments, and in 30 age-matched healthy women. Hormonal data were correlated with clinical parameters (age, epilepsy syndrome, disease onset, and duration), seizure frequency assessed on the basis of a seizure frequency score (SFS), and AED therapies.

Results: E2, Pg, and FEI were lower, whereas SHBG levels were higher in the epilepsy patients than in the controls. However, sex steroid and SHBG levels were not different between groups of patients categorized according to SFS. Therapies with EIAEDs accounted for changes in E2 levels and FEI.

Conclusions: Despite globally decreased sex steroid levels in serum, actual hormone titers were not significantly correlated with SFS in consecutive epilepsy women; rather, these hormonal changes were explained by AED treatments, mainly when EIAED polytherapies were given.

Neuronal activity is modulated by “neuroactive steroids” synthesized in gonads and adrenal glands. These steroids easily cross the blood–brain barrier and may thus influence the neurotransmission and the function of receptors and membrane ion channels involved in epileptogenesis (Beyenburg et al., 2001). Estradiol (E2) and other estrogens enhance, whereas progesterone (Pg) and some of its active metabolites decrease neuronal excitability (Beyenburg et al., 2001; McEwen, 2002; Scharfman & MacLusky, 2006). Glucocorticoids also modify neuronal metabolism, excitability, and survival, but their effects on seizure susceptibility in clinics are less clear (Joels, 1997).

Conversely, changes in the peripheral steroid set-up in epilepsy may be caused by interictal discharges and seizures themselves, which affect brain areas involved in the regulation of the hypothalamic–pituitary function (Herzog et al., 2003). Finally, it is necessary to consider the effect of antiepileptic drugs (AEDs), which interfere with the synthesis, metabolism, and bioavailability of steroids by increasing their binding to sex hormone-binding globulin (SHBG) (Morrell et al., 2001; Isojarvi et al., 2005; Isojarvi, 2008).

In this cross-sectional observational study we assessed serum levels of some steroids known to influence neuronal activity and excitability in consecutive women with different epilepsy syndromes on various AED treatments. Data concerning cortisol and dehydroepiandrosterone sulfate (DHEAS) levels in this same group of patients have been published previously (Galimberti et al., 2005). In this paper, we focus on ovarian hormone findings.


One hundred thirteen female outpatients with epilepsy, age 16–47 years (mean ± SD: 28.1 ± 8.0 years), were enrolled in this study. Informed and written consent was obtained from all patients, in accordance with the principles of the Helsinki Declaration. The subjects’ clinical characteristics and the criteria of exclusion from the study have been detailed elsewhere (Galimberti et al., 2005).

According to the 1989 International League Against Epilepsy (ILAE) classification criteria, 40 patients had idiopathic generalized epilepsy (IGE) and 68 had cryptogenic or remote symptomatic partial epilepsy (PE); in five patients the type of epilepsy was undetermined. The patients’ age at onset of epilepsy ranged from 1–37 years (mean ± SD: 13.9 ± 7.4 years), and they reported a disease duration ranging from 6 months to 38 years (mean ± SD: 14.4 ± 8.8 years).

Sixty-nine patients were on AED monotherapy: 51 were taking enzyme-inducing AEDs (EIAEDs), namely, phenobarbital (PB), carbamazepine (CBZ), oxcarbamazepine (OXC), phenytoin (PHT), and primidone (PRM); 18 were on non–enzyme-inducing AEDs (NEIAEDs), namely valproic acid (VPA) and lamotrigine (LTG). Forty-four patients were on AED polytherapy (two or more EIAEDs in 23 cases, EIAED and NEIAED combinations in 19, and two NEIAEDs in the other 2).

All patients recorded seizure occurrence and menstrual cycles in a diary. Blood samples for the assays of E2, Pg, and SHBG were taken between 8:00 and 9:00 a.m. during the midluteal phase (20th–24th day of the ovarian cycle) in menstruating women or at a random time in the five amenorrheic patients (three IGE; two PE).

The patients’ seizure frequency scores (SFSs), over the previous 6 months, were reported using a four-point scale: (1) absent (N = 50 patients), (2) sporadic (one seizure per month or less) (N = 25), (3) frequent (four seizures per month or less) (N = 30), (4) very frequent (more than one seizure per week up to one or more seizure per day) (N = 8). Because of the small number of subjects with very frequent seizures, groups 3 and 4 were processed together.

Data in epilepsy patients were compared with those of 30 healthy, drug-free women, aged 19–47 years (mean ± SD: 28.3 ± 8.5), with normal menstrual cycle. Steroid and SHBG levels were assayed by currently available immunoradiometric assay (IRMA) methods. Free E2 index (FEI) was calculated on total E2 (pmol/L) divided by SHBG concentrations (nmol/L).


The SAS statistical package (SAS, Carey, NC, USA) was used for the statistical analysis. Student’s t test for unpaired data was used to assess statistical differences between two groups. The SAS general linear model (GLM) procedure was used throughout to analyze statistical differences in clinical parameters and hormonal data between healthy controls and multiple groups of patients formed according to their SFSs or AED treatments.

The GLM procedure and the REG procedure for multiple regression stepwise analysis were used to analyze correlations between clinical findings and hormonal data in healthy controls and in patients grouped according to their SFSs and AED treatments. A p-value < 0.05 (two-sided) was taken to indicate a statistically significant difference.


PE was diagnosed in 87% of the patients with more severe epilepsy (SFS 3–4), in 56% of those with SFS 2, and in 42% of those with SFS 1, whereas IGE affected 13% of the patients with SFS 3–4, and 44% and 48% of those with SFS 2 and SFS 1, respectively. Moreover, 50% of the patients with SFS 3–4, 40% of those with SFS 2, and 30% of those with SFS 1 were on AED polytherapy.

These women with epilepsy, analyzed as a whole group, showed significantly lower luteal E2 and Pg levels, whereas their SHBG levels were significantly higher than those recorded in the controls (Table 1). Consequently, the mean FEI was significantly lower in the epilepsy patients than in the healthy females (Table 1).

Table 1.   Levels of E2, Pg, and SHBG, plus FEI (mean values ± SD), measured during the midluteal phase of the ovarian cycle (or at a random time in the five amenorrheic patients) in 113 women with epilepsy and in 30 healthy controls
GroupE2 (pmol/L)Pg (nmol/L)SHBG (nmol/L)FEI
  1. Student’s t-test was used for statistical analysis.

  2. E2, estradiol; FEI, free estrogen index; Pg, progesterone; SHBG, sex hormone-binding globulin.

Epilepsy337.5 ± 182.320.0 ± 14.769.9 ± 41.08.0 ± 10.9
Healthy controls491.1 ± 134.127.8 ± 14.347.7 ± 9.910.7 ± 4.2
Student’s t-testp < 0.001p = 0.011p < 0.001p = 0.025

Lower mean E2 and Pg levels, and increased SHBG levels versus healthy controls were confirmed in all the subgroups of patients formed on the basis of their SFSs. These differences were statistically significant for the E2 and SHBG levels, whereas the difference in the Pg levels was at the limit of significance (Table 2). However, no statistically significant differences in sex steroid and SHBG levels emerged among the patients divided into subgroups on the basis of their SFSs. Sex steroid and FEI data did not significantly correlate with age, disease duration, SFS, or epilepsy syndrome (p > 0.05).

Table 2.   Levels of E2, Pg, and SHBG, plus FEI (mean values ± SD), measured during the midluteal phase of the ovarian cycle (or at a random time in the five amenorrheic patients) in 113 women with epilepsy divided according to disease severity scores and in 30 healthy controls
GroupE2 (pmol/L)Pg (nmol/L)SHBG (nmol/L)FEI
  1. Statistical analysis was performed using the SAS GLM procedure.

  2. E2, estradiol; FEI, free estrogen index; GLM, general linear model; Pg, progesterone; SFS, seizure frequency score; SHBG, sex hormone-binding globulin.

SFS 1 (N = 50)329.4 ± 166.419.3 ± 14.570.1 ± 44.17.5 ± 6.9
SFS 2 (N = 25)363.4 ± 200.321.9 ± 15.360.7 ± 41.09.1 ± 9.0
SFS 3–4 (N = 38)331.3 ± 193.219.6 ± 14.775.5 ± 36.58.0 ± 15.5
Healthy controls (N = 30)491.1 ± 134.127.8 ± 14.347.7 ± 9.910.7 ± 4.2

SHBG levels were significantly correlated with increasing age (F = 1.87; p = 0.0158) and disease duration (F = 1.63; p = 0.0351), but not with SFS or epilepsy syndrome. E2 and FEI, but not Pg and SHBG levels, showed significant differences among patients grouped according to their AED therapy schedules, the mean E2 and FEI values being highest in patients under treatment with a single NEIAED, and lowest in those on EIAED polytherapy (Table 3).

Table 3.   Levels of E2, Pg, and SHBG, plus FEI (mean values ± SD), measured during the midluteal phase of the ovarian cycle (or at a random time in the five amenorrheic patients) in 113 women with epilepsy divided according to the AED therapy with enzyme inducing AEDs (EIAEDs) and non–enzyme-inducing AEDs (NEIAEDs)
AED therapyE2 (pmol/L)Pg (nMol/L)SHBG (nMol/L)FEI
  1. Statistical analysis was performed using the SAS general linear model (GLM) procedure.

Monotherapy EIAED343.6 ± 66.222.1 ± 16.170.4 ± 8.17.3 ± 1.2
Monotherapy NEIAED440.6 ± 215.316.4 ± 14.755.5 ± 36.016.7 ± 22.0
Polytherapy EIAED276.8 ± 161.818.4 ± 16.073.1 ± 33.74.8 ± 4.0
Polytherapy EIAED + NEIAED308.1 ± 169.321.6 ± 16.174.6 ± 52.46.4 ± 5.9


In this study we evaluated, in a group of consecutive women with AED-treated epilepsy, the relationship between sex steroid levels and some clinical parameters, that is, epilepsy syndrome, disease severity (based on seizure frequency), and AED treatments. In a previous study we demonstrated, in this same cohort of subjects, increases in morning cortisol and in the cortisol/DHEAS ratio (C/Dr), but significantly decreased DHEAS levels in comparison with healthy controls. In addition, women with more frequent seizures showed a more marked increase of cortisol levels, higher C/Drs, and lower DHEAS levels than those with rare or absent seizures (Galimberti et al., 2005).

Although multivariate statistical analysis ruled out a role of epilepsy syndromes and AED treatments in inducing cortisol changes, disease severity (SFS) explained the cortisol and C/Dr increases. DHEAS variations were highly correlated with SFS, disease duration, and epilepsy syndrome, but EIAED therapies also contributed to reducing steroid levels (Galimberti et al., 2005).

We can now add data on sex steroid profiles in this same group of patients. Both luteal E2 and Pg levels were significantly diminished in these epilepsy patients compared with healthy controls. Changes in pulsatile secretion of gonadotropins, altered pituitary response to luteinizing hormone–releasing hormone, ovarian dysfunction, and a higher prevalence of polycystic ovaries (PCO), and PCO syndrome are all factors contributing to ovarian failure, and to more frequent menstrual disorders, reproductive dysfunction, and decreased fertility in both treated and untreated women with epilepsy compared with the general population (Murialdo et al., 1997; Bauer et al., 1998; Stoffel-Wagner et al., 1998; Morrell et al., 2001; Herzog, 2006; Lofgren et al., 2007).

PE, mainly when originating in the temporal lobe, is thought to be the form most frequently associated with reproductive dysfunction (Herzog et al., 2003; Herzog, 2006). In our groups of patients, PE accounted for 87% of the cases in the SFS 3–4 group; this group also presented a significantly earlier disease onset and a longer disease duration (Galimberti et al., 2005). Nevertheless, the few amenorrheic women we enrolled were distributed in the SFS 1 and 2 groups, and neither luteal E2 and FEI, nor Pg mean levels showed significant differences among SFS groups.

Furthermore, in our patients, sex hormones, unlike cortisol and DHEAS (Galimberti et al., 2005), were not significantly correlated with clinical parameters, such as age, disease duration, epilepsy syndrome, and SFS. On the other hand, the possibility that the physiologic surge in sex hormone levels during the luteal phase plays a role in eliciting seizures was emphasized by Bauer et al. (1998), who reported that women with ovulatory cycles more often presented with seizures during the days of menstruation than women with anovulatory cycles.

Most of the old AEDs given to our subjects (PB, CBZ, PHT) are known to affect levels of steroid hormones by enhancing their metabolism through activation of the cytochrome P450 oxidative system, thereby also inducing SHBG synthesis in the hepatocytes (Isojarvi et al., 2005; Isojarvi, 2008). Older patients and those with longer disease duration showed higher levels of SHBG. This can be attributed to the physiologic increase in binding protein levels with aging and to these subjects’ greater use of EIAEDs. These changes in serum SHBG levels, and the lowered total E2 levels, implied a global decrease in biologically active free E2, but we were unable to demonstrate significant differences in FEI in women with more frequent seizures.

Conversely, significant differences in E2 levels and in FEI were found between women on different AED regimens. Patients treated with EIAED polytherapies showed E2 and FEI values that were lower than those recorded both in patients treated with a single EIAED or NEIAED, and in those on combined (EIAED plus NEIAED) therapies.

We did not analyze the effects of single AEDs on sex hormones in our patients because of the large variety of pharmacologic treatments they were on and the small number of cases in some of the groups. However, most of the women we studied were treated with traditional EIAEDs, PB and CBZ primarily, whereas the NEIAEDs considered in this study, namely VPA and LTG, modified E2, SHBG, and FEI to a lesser degree when used in monotherapy.

Recently, Jacobsen et al. (2008) reported that some of the AEDs used in our patients (PB, PHT, VPA, OXC, LTG), but not CBZ and PRM, inhibit in vitro the aromatase complex (CYP19) activity that converts testosterone into E2. Additive enzyme inhibition has been observed in combination experiments with multiple AEDS, and was also found in a number of the subjects we studied. Therefore, aromatase inhibition might explain the decreased production of E2 we observed in vivo, mainly when combined AED therapies were employed.

In conclusion, consecutive women with AED-treated epilepsy showed decreased E2 and Pg levels. This finding is due, in a large part, to AED therapies, which interfere with sex steroid synthesis and metabolism, but a reduced conversion of androgens to estrogens might also be involved. In addition, impaired luteal function and anovulation could explain the lower luteal Pg surge observed in our patients.

Ovarian steroids are known to exert various biologic effects beneficial to women’s health, and ovarian function should be monitored closely as part of the comprehensive care of women with epilepsy (Herzog, 2006) to prevent adverse effects of ovarian failure, mainly in patients with more frequent seizures who need AED polytherapies.

However, although the neuroexcitatory effects of estrogens and anticonvulsant properties of progestins have been extensively documented (Beyenburg et al., 2001; Herzog, 2006), our findings suggest that actual levels of E2 and Pg in serum did not appear to be significantly correlated with seizure frequency in women with AED-treated epilepsy, and that the role of their changes on disease exacerbations remains to be better defined.

For this observational study, we randomly enrolled consecutive outpatients with different epilepsy syndromes and clinical conditions. Studies in more selected and homogeneous groups of patients are needed to better define the complex interactions between epilepsy, AEDs, and reproductive function.


This study was supported by grants from the University of Genoa and from the Ministry of Health to the “IRCCS C. Mondino Institute of Neurology” Foundation (RC 2006).

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: The authors declare no conflicts of interest relevant to this article.