Address correspondence and reprint requests to Dr. G. Murialdo at Department of Endocrinological and Metabolic Sciences, University of Genova, Viale Benedetto XV, 6, I-16132 Genova, Italy. E-mail: firstname.lastname@example.org
Summary: Purpose: Hormonal changes occur in epilepsy because of seizures themselves and of antiepileptic drug (AED) effects on steroid production, binding, and metabolism. Conversely, steroids may influence neuron activity and excitability by acting as neuroactive steroids. This cross-sectional observational study aimed to evaluating cortisol and dehydroepiandrosterone sulfate (DHEAS) levels in female epilepsy patients with different disease severity, as assessed by a seizure frequency score (SFS).
Methods: Morning serum levels of cortisol and DHEAS were assayed in 113 consecutive women, aged 16 to 47 years, with varied epilepsy syndromes, receiving mono- or polytherapy with enzyme-inducing and/or noninducing antiepileptic drugs (AEDs). Hormonal data were correlated with clinical parameters (age, body mass index, epilepsy syndrome, disease onset and duration, SFS, AED therapy, and AED serum levels) and compared with those of 30 age-matched healthy women.
Results: In epilepsy patients, cortisol levels and cortisol-to-DHEAS ratios (C/Dr) were significantly higher, whereas DHEAS levels were significantly lower than those in controls. Patients with more frequent seizures showed higher cortisol and C/Dr values and lower DHEAS levels than did those with rarer or absent seizures during the previous 6 months. SFS mainly explained the increase of cortisol levels and C/Dr in patients with more active disease. Changes in DHEAS levels correlated with SFS and epilepsy syndrome, as well as with AED treatments and ages.
Conclusions: Women with more frequent seizures had alterations of their adrenal steroids characterized by an increase of cortisol and a decrease of DHEAS levels. Such hormonal changes might be relevant in seizure control and in patient health.
Both seizures and antiepileptic drugs (AEDs) may affect hormone secretion through their influences on brain areas involved in the regulation of the hypothalamus–pituitary function or directly by acting on the pituitary. Besides, seizures may act as a “stressor” and thus activate the hypothalamus–pituitary–adrenal axis (HPAA) (1,2).
Neuron function and excitability are influenced by neurotransmitters and neurosteroids, the latter produced de novo from cholesterol by cerebral glial cells and neurons (3–5). However, the brain also may be regarded as a target for steroid hormones synthesized by the adrenal glands and gonads, which easily cross the blood–brain barrier and act as neuroactive steroids (6–8). These effects on neuron excitability are due to modulatory activity, either direct or mediated by steroid metabolites, on a variety of neurotransmitter receptors [e.g., γ-aminobutyric acid A (GABAA) and N-methyl-d-aspartate (NMDA)] and/or ion channels on the cell membranes, involved in stress response, epileptogenesis, and seizure occurrence (7–11).
Glucocorticoids appear to modify neuron metabolism, excitability, and survival (12–14). However, the relations between seizures and cortisol or other steroids produced by the adrenal gland, such as dehydroepiandrosterone (DHEA) and its sulfate (DHEAS), remain uncertain. This cross-sectional observational study was aimed at assessing the circulating levels of cortisol and DHEAS in a cohort of AED-treated consecutive women with various degrees of epilepsy severity and disease control, quantified by monitoring seizure frequency.
SUBJECTS AND METHODS
One hundred thirteen consecutive adult women with epilepsy, outpatients attending the Epilepsy Center of the Neurological Institute IRCCS “C. Mondino” of Pavia (Italy) between January and June 2002, were eligible for this cross-sectional study. Appointed patients were otherwise clinically healthy except for epilepsy and were selected according to the following criteria of exclusion: pregnancy or lactation ongoing or in the year before; therapies with corticosteroids, estroprogestins, or drugs interfering with hypothalamus–pituitary function, adrenal steroidogenesis, or steroid metabolism in the previous 6 months, except for AEDs; and alcohol abuse or drug addiction. Blood samples for hormonal assays were obtained on the occasion of a routine monitoring for serum AED concentrations. Informed and written consent was obtained from all subjects according to the principles of the Helsinki Declaration.
Patients were age 16 to 47 years, their body weight (BW) ranged between 44 and 103 kg, and the body mass index (BMI), between 18.1 and 39.1.
According to the 1989 International League Against Epilepsy (ILAE) Classification (15), idiopathic generalized epilepsy (IGE) was diagnosed in 40 cases, and cryptogenic or remote symptomatic partial epilepsy (PE) in 68 subjects. Five patients had epilepsy undetermined as partial or generalized (UNDE).
Seizure onset was between ages 1 and 37 years, and disease duration (the period elapsing from the first recognized seizure) lasted 6 months to 38 years. Clinical and demographic features of patients are summarized in Table 1.
Table 1. Biometric and clinical parameters, seizure frequency score, and body mass index in 113 female patients with idiopathic generalized epilepsy, partial epilepsy, and epilepsy undetermined as partial or generalized
Age at disease onset (yr)
Disease duration (yr)
BMI, body mass index; IGE, idiopathic generalized epilepsy; PE, cryptogenic or remote symptomatic partial epilepsy; UNDE, epilepsy undetermined as partial or generalized; scores (SFS).
IGE (n = 40)
25.2 ± 6.5
12.4 ± 5.1
12.8 ± 8.5
1.6 ± 0.9
23.1 ± 3.6
PE (n = 68)
29.6 ± 8.3
14.2 ± 8.2
15.4 ± 8.8
2.3 ± 1.1
23.1 ± 4.2
UNDE (n = 5)
34.4 ± 9.3
21.8 ± 7.3
12.6 ± 10.3
1.0 ± 0.0
25.9 ± 4.2
All patients (n = 113)
28.1 ± 8.0
13.9 ± 7.4
14.4 ± 8.8
1.9 ± 1.1
23.2 ± 4.0
Four possible regimens of AED treatments were defined on the basis of mono- or polytherapies with AEDs, and of their enzyme-inducing or noninducing properties. Thus 69 patients were receiving AED monotherapy, 56 were treated with enzyme-inducing AEDs [EIAEDs: phenobarbital (PB), carbamazepine (CBZ), oxcarbazepine (OXC), phenytoin (PHT), or primidone (PRM)] and 13 with non–enzyme-inducing AEDs [NEIAEDs: valproic acid (VPA) or lamotrigine (LTG)]. Forty-four patients were receiving AED polytherapy: 23 with more than one EIAED and 21 with combined EIAEDs and NEIAEDs.
Epilepsy severity and control over the 6-month period before the study were identified by a seizure frequency score (SFS) made up according to the following scale: 1, absent; 2, sporadic (one seizure monthly or less); 3, frequent (four seizures monthly or less); 4, very frequent (more than one seizure weekly up to one or more seizure daily). Characteristics of patients grouped according to SFS are shown in Table 2. Because of the small number of subjects in the SFS group 4, their data have been processed together with those of group 3.
Table 2. Biometric and clinical features in 113 women with epilepsy grouped according to their seizure frequency
No. of cases
Age at disease onset (yr)
Disease duration (yr)
Epilepsy syndrome IGE/PE/UNDE
AED regimen IM/NIM/IP/INP
BMI, body mass index; IGE, idiopathic generalized epilepsy; PE, cryptogenic or remote symptomatic partial epilepsy; UNDE, epilepsy undetermined as partial or generalized; SFS, seizure frequency score; AED, antiepileptic drug; IM, enzyme-inducing AED monotherapy; NIM, non–enzyme-inducing AED monotherapy; IP, enzyme-inducing AED polytherapy; INP, enzyme-inducing and non–enzyme-inducing AED polytherapy.
Statistical analysis by GLM frequency procedure (F values) or χ2 (for epilepsy syndrome and AED regimen distribution).
26.8 ± 0.7
22.9 ± 3.7
15.6 ± 7.1
11.2 ± 7.8
26.2 ± 0.7
23.6 ± 4.1
14.6 ± 2.1
11.7 ± 1.4
32.7 ± 9.2
23.8 ± 4.3
11.9 ± 1.4
20.8 ± 10.6
27.1 ± 4.2
22.3 ± 1.4
8.7 ± 0.7
18.4 ± 3.5
F or χ2 values
Data in epilepsy patients were compared with those of 30 healthy and drug-free women, age 19–47 year (mean ± SD, 28.3 ± 8.5 years), with normal menstrual cycles and BMIs ranging between 19.7 and 26.1 (mean ± SD, 23.0 ± 1.8).
Blood samples were drawn between 8.00 and 9.00 a.m., after a 12-h fasting period, and then rapidly centrifuged. Sera were maintained frozen at –20°C until processed. Cortisol and DHEAS levels were assayed by coated tube radioimmunoassay (RIA) methods by using reagents purchased from RADIM (Pomezia, Italy). The sensitivities and intraassay and interassay percentage coefficients of variation (CV%) for determined values were 2.8 nM (CV%, <3.8 and 7.1) for cortisol, and 140 nM (CV%, <2.9 and 5.9) for DHEAS assay.
Cortisol/DHEAS ratio (C/Dr) was calculated in each subject on steroid molar titers, and the obtained value was multiplied by 10.
PB, CBZ, PHT, and VPA serum levels were assayed by routine fluorescence polarization immunoassay system.
The SAS 8.2 statistical package (Carey, NC, U.S.A.) was used for statistical analysis. A univariate procedure was used for descriptive statistics of data and to evaluate their skewness and distribution.
Student's t test for unpaired data was adopted to assess statistical differences in clinical and hormonal data between groups.
The general linear model (GLM) procedure was used throughout to analyze statistical differences in clinical parameters (age, BMI, age at disease onset, disease duration, AED levels) and hormonal data between multiple groups of patients grouped according to their SFS, and χ2 to assess the distribution of epilepsy syndromes and AED treatments in SFS groups. GLM procedure and REG procedure for multiple regression step-wise analysis were used to analyze correlations between clinical findings, cortisol, DHEAS, and C/Dr values in patients grouped according to their SFS and AED treatments, and in healthy controls. GLM procedure also was used for a preliminary identification of variables determining changes in hormone levels.
Because of quantitative or categoric differences of variables, we used the CORR procedure for the calculation of Kendall's correlation coefficient to evaluate the association strength between clinical parameters and hormonal data that were found to be significantly different between the various SFS groups, as inferred by the preliminary GLM procedure.
Clinical features according to SFS
Patients with more frequent seizures (SFS groups 3–4) were older, their disease had begun earlier, and it showed a longer duration than in SFS 1 and 2 groups. No significant changes were found in BMI among patients with different SFS (Table 2).
SFS 1 group had a higher prevalence of cases with IGE (48%). Moreover, it included 42% of patients with PE and all patients with UNDE (10%). Fifty-six percent of patients with PE were in SFS group 2, but they prevailed in groups 3–4 (86.8% vs. 13.2% of cases with IGE; Table 2). Eleven patients with IGE were receiving EIAED and 12 NEIAED monotherapy, with one taking EIAED and 16 taking EIAED plus NEIAED polytherapy. Forty-one patients with PE were treated with a single EIAED; 22 with a single EIAED; and five with EIAED plus NEIAED polytherapy. All patients with UNDE took a single AED (four EIAED and one NEIAED).
Thus different AED regimens significantly correlated with epilepsy syndromes (F= 13.52; p ≤ 0.0001) and SFS (F= 2.95; p = 0.0236) (Table 2), but not with other clinical indices.
No significant difference was found between serum levels of the main AEDs assayed in patients grouped according to their SFS (PB, F= 1.37, p = 0.2620; CBZ, F= 1.51, p = 0.2313; VPA, F= 1.57; p = 0.2257). Only serum PHT levels appeared causally reduced in patients in SFS group 2 (F= 4.24; p = 0.0404).
Women with epilepsy showed significantly higher baseline cortisol levels, whereas those of DHEAS were decreased in comparison with controls. Consequently, C/Dr values were higher in women with epilepsy than in healthy women (Fig. 1).
Mean cortisol levels were higher in patients with more severe epilepsy than in seizure-free women and in healthy controls (F= 5.10; p = 0.002), but no significant difference was observed among SFS groups (Fig. 2). Conversely, DHEAS levels were significantly modified in relation to seizure occurrence, being further decreased in women with more frequent seizures than in those with sporadic or absent seizures and in controls (F= 48.47; p < 0.001). C/Dr values were high in patients in SFS groups 2 and 3–4 versus group 1 patients and in healthy subjects (F = 16.58; p < 0.001; Fig. 2).
Correlations between steroid data and clinical parameters are reported in Table 3. Disease duration correlated inversely with DHEAS and directly with C/Dr. An age-dependent decrease in DHEAS levels was found in healthy women (F= 19.04; p = 0.0002), but not in those with epilepsy. Cortisol levels and C/Dr were directly and those of DHEAS indirectly correlated with SFS.
Table 3. Correlations between cortisol and DHEAS levels, cortisol-to-DHEAS ratio values and age, body mass index, disease duration, seizure frequency scores, and epilepsy syndromes in 113 women with epilepsy
Values expressed as mean ± SD.
DHEAS, dehydroepiandrosterone sulfate; C/Dr, cortisol-to-DHEAS ratio; BMI, body mass index; SFS, seizure frequency score.
DHEAS levels were significantly different in relation to epilepsy syndromes, being lower in PE than in IGE or UNDE.
DHEAS levels were significantly higher in patients treated with NEIADs than in those receiving EIAD mono- or polytherapy. No significant changes in cortisol and C/Dr values were found between patients with various AED treatments (Table 4). No significant correlation was found between hormone and AED serum levels, the latter never exceeding the therapeutic range in any subject.
Table 4. Cortisol and DHEAS levels and C/Dr values in 113 women with epilepsy grouped according to mono- or polytherapy with enzyme-inducing and non–enzyme-inducing AEDs
Changes in DHEAS levels in epilepsy were determined by SFS as 44% (p < 0.0001), by age for 33% (p = 0.0003), by disease duration for 32% (p = 0.0007), AED therapy for 0.23% (p = 0.0128), whereas changes of C/Dr were determined by SFS as 27% (p = 0.0037), AED therapy for 22% (p = 0.0177), and age for 21% (p = 0.0222).
In this study, we evaluated consecutive women with epilepsy receiving AED treatment and attempted to correlate cortisol and DHEAS levels with some clinical and demographic parameters, disease severity as assessed by seizure frequency during the previous 6 months, and differing AED regimens.
Patients with more frequent seizures were slightly older than those with rarer or absent seizures, and their disease had begun earlier and showed longer duration.
PE affected 86.8% of patients with more active epilepsy, but only 56% and 42% of those in SFS groups 2 and 1, respectively. Fifty percent of patients in the SFS group 3–4, 40% in SFS group 2, and 30% in SFS group 1 were receiving polytherapy, whereas 92% of cases in SFS group 3–4, with seizures persisting weekly to daily were treated with EIAEDs alone or combined with other AEDs.
However, multivariate statistical analysis ruled out a significant or exclusive role of epilepsy syndromes and of serum AED levels in determining changes in cortisol levels. Only DHEAS levels correlated to epilepsy syndrome, AED therapy, and disease duration.
The most interesting finding in this study concerned the relation between epilepsy severity, increased cortisol, and decreased DHEAS levels. These changes were more relevant in patients with more frequent until daily seizures.
An increase in adrenocorticotropic hormone (ACTH) and in cortisol levels has been observed in the epileptic postictal period (1). Gallagher et al. (16,17) demonstrated that in humans, both spontaneous and temporal lobe seizures induced by depth electrode stimulation increased ACTH and cortisol levels and their secretion rates as well. Moreover, the elevation of serum cortisol levels in patients with status epilepticus has been proposed as a useful predictive indicator of prognosis and as playing a pathophysiologic role in this condition (18).
Moreover, CBZ has been found to increase 24-h urinary free cortisol, evening plasma cortisol, and cortisol-binding globulin capacity when administered to healthy volunteers, possibly by interfering with a negative-feedback mechanism in the pituitary (19). However, baseline cortisol levels in our patients were not related to AED treatments and single-drug serum levels, but rather to SFS.
In our series, decreased DHEAS levels correlated with disease activity and duration, epilepsy syndrome, and AED therapies, being affected to a greater extent in patients with more frequent seizures treated with EIAEDs, and less in those taking NEIAEDs.
Reduced DHEAS levels in patients with temporal lobe epilepsy on CBZ and other EIAED medication were previously attributed to the drug effects (20,21). In male epilepsy patients, CBZ reduced DHEAS levels to 62.7% and the ratio of DHEAS to DHEA to 63% of pretreatment values, but no changes in steroid levels were observed in subjects taking VPA (22). With respect to healthy women, DHEAS levels were reduced in women with epilepsy treated with PB or CBZ, but not in patients taking VPA therapy (23). No changes in DHEA levels occurred in epilepsy patients of both sexes taking NEIAEDs, such as VPA or LTG (24). Precocious sex-related changes in DHEAS levels have been reported after 3-month VPA therapy, being decreased in women and increased in men, whereas CBZ reduced steroid levels in both groups (25). These data emphasize the role of drugs in inducing changes of DHEAS levels in epilepsy. However, DHEAS levels remained substantially and significantly reduced after a 6-month period of AED discontinuation in men with epilepsy (26). This finding induces us to consider that factors other than AEDs may affect DHEAS levels in epilepsy.
The unbalancing in morning cortisol and DHEAS levels was emphasized by increased C/Dr. Similar changes are present in aging and characterize other neurologic conditions (i.e., primary depression and dementia) (27–30).
DHEAS is a neurosteroid acting also as a neuroactive steroid (6,31). Both DHEA and DHEAS permeate the blood–brain barrier and are detectable in CSF (6,32). Their levels decline with age (33,34), as we observed in our control subjects, but not in women with epilepsy, in whom the correlation between steroid levels and age was weak, possibly because of the interfering effects of AED treatment and/or of seizures themselves.
Opposite effects on neuron survival have been attributed to cortisol and DHEA or DHEAS. Cortisol affects cerebral glucose metabolism and enhances Ca2+ influx in hippocampal neurons, thus exerting a neurotoxic effect (12,13), whereas neuroprotective activity has been suggested for DHEA and DHEAS (6,35,36).
Moreover, some relations of corticosteroids with seizures have been described (8). Glucocorticoids affect neuron excitability (8,9,14) other than survival (12,13). Among brain areas, the hippocampus has high concentrations of receptors for gluco- and mineralocorticoids (12,13,36), but mRNA expression of these steroid receptors has been demonstrated also for other brain regions in patients with epilepsy (37). An exacerbation of seizure frequency induced by high corticoid concentrations and mediated by mineralocorticoid receptors has been demonstrated from experimental models of epilepsy (38). Temporal lobe epilepsy induced by cerebral injection of kainic acid in rats increased corticosterone levels, suggesting a seizure-mediated activation of the HPAA (39).
DHEAS is known to act as a noncompetitive allosteric antagonist of the GABAA receptor, to inhibit GABA-induced chloride transmembrane transport, and to antagonize NMDA toxic effects on neurons due to enhanced Ca2+ influx (6,8,35,40). All these factors are involved in neuron excitability and seizures. Intracerebroventricular injection of DHEAS in animals induced seizures in a dose-dependent way and antagonized the hypnotic-like effect of ethanol (41), so that the actual effects in vivo of DHEAS on neuronal excitability and seizure susceptibility remain uncertain. DHEAS levels may be found reduced in a number of clinical conditions other than epilepsy. Findings from a large population-based study in elderly people showed that the decrease in DHEAS levels correlated with diseases or clinical data that were different in male and female subjects. Thus the authors concluded that DHEAS might be considered an unspecific marker of disease rather than a definite etiologic factor (42).
In conclusion, adult women with epilepsy receiving long-term AED treatment showed increased cortisol and reduced DHEAS levels, thus confirming previous data in the literature. This finding is due, in part, to the therapy with EIAEDs, which decrease DHEAS levels possibly by their enzyme-inducing activity. However, our data showed that changes in adrenal steroid levels in women with epilepsy must be ascribed mainly to the frequency of seizures, which are known to activate HPAA function.
DHEAS is a weak androgen secreted by the adrenal cortex, but its biologic function has not yet been completely understood, and the clinical effects related to its decrease remain uncertain (43). DHEAS has been indicated to exert various favorable biologic and metabolic effects in women, in particular on bone and muscle metabolism, on protection against some sex-dependent tumors, and possibly on cognitive function (6,32,42). But the actual clinical role of these changes in glucocorticoid levels on neuron excitability and neurodegeneration, as well as in other health-threatening events in long-term epilepsy treatment deserves further investigations.
Acknowledgment: We are grateful to Prof. Maria Vittoria Gianelli, Faculty of Medicine, University of Genova, for the linguistic revision of the manuscript.