Address correspondence and reprint requests to Dr. Preeti Sahota, Department of Neurology Postgraduate Institute of Medical Education & Research, Chandigarh, India 160012. E-mail: email@example.com
Summary: Purpose: To assess the association of long-term sodium valproate therapy with reproductive endocrine disorders in Indian women with generalized epilepsy.
Methods: Clinical parameters, ovarian morphology, and serum reproductive hormone concentrations were evaluated in 30 clinically normal and eumenorrheic reproductive age women with generalized epilepsy who were newly initiated on valproate. Longitudinal evaluations were done in 25 of these women after 1 year, and in some of them after 2 and 3 years of therapy.
Results: Of the 25 women who completed 1 year follow-up, we observed clinically relevant weight gain in 40%, hirsutism in 20%, menstrual abnormalities in 24%, polycystic ovaries (PCO) in 16%, polycystic ovarian syndrome (PCOS) in 20%, and a significant increase in mean serum testosterone (p = 0.046). A significant positive correlation existed between weight gain and the development of menstrual abnormalities (r = 0.66, p < 0.0001), hirsutism (r = 0.53, p = 0.006) and PCO (r = 0.51, p = 0.012). No correlation existed between weight change and serum reproductive hormonal changes. Yearly follow-up for next 2 years in some of these women revealed persistence of menstrual abnormalities, hirsutism and PCO, a significant linear increase in mean body weight, body mass index, and serum testosterone concentrations, and an increase in serum LH levels from second year onwards.
Limitations: Limitations include small sample size and a high dropout rate on follow-up.
Conclusions: Long-term valproate therapy in Indian women with generalized epilepsy is associated with development of hirsutism, significant weight gain, stable or progressive alterations in reproductive hormonal function, and ultimately a higher occurrence of PCOS.
Reproductive endocrine disorders and sexual dysfunction occur more frequently among women with epilepsy than among normal subjects of similar age (Herzog et al., 1986; Webber et al., 1986; Bilo et al., 1988; Isojärvi et al., 1993; Bauer et al., 2000). These include polycystic ovarian syndrome (PCOS), isolated components of this syndrome such as polycystic ovaries (PCO) and hyperandrogenemia, hypothalamic amenorrhea, and functional hyperprolactinemia. Recent papers have evoked the role of antiepileptic drugs, more particularly sodium valproate (VPA), in the expression and possible increased frequency of these disorders, especially that of PCOS, in women with epilepsy.
PCOS is one of the most common endocrine disorders in women, with an estimated incidence between 4% and 6.7% (Herzog, 1996; Diamanti-Kandarakis et al., 1999). In addition to being the most common cause of anovulatory infertility, PCOS is associated with an increased risk for type 2 diabetes, impaired glucose utilization and cardiovascular disease (Meirow et al., 1996; Hopkinson et al., 1998; Lobo and Carmina, 2000; Dunaif and Thomas, 2001). Women with epilepsy who are treated with VPA have been reported to be at differential risk for developing PCO, and menstrual disorders and hyperandrogenism, two features of PCOS (Isojärvi et al. 1993, 1996; Mikkonen et al., 2004). Obesity, hyperinsulinemia, and low serum levels of insulin-like growth factor binding protein 1 have been implicated as the possible causative factors leading to PCO and VPA-related hyperandrogenism in these women (Isojärvi et al. 1996, 2001). Hyperandrogenism has also been reported in prepubertal and pubertal girls taking valproate for epilepsy (Vainionpăă et al., 1999). In their recent post hoc reanalysis in 148 women with epilepsy by antiepileptic drug use and epilepsy type, Lofgren et al. (2006) reported an increased risk of reproductive endocrine disorders in women with idiopathic generalized epilepsy and use of VPA was a predictor of the development of PCO and PCOS. Isojärvi et al. (1998) have reported reduction of VPA-related risks of obesity, hyperinsulinemia, hyperandrogenemia, and PCO one year after substituting lamotrigine for VPA. Whether this possible risk of reproductive endocrine abnormalities on long-term VPA use could outweigh its benefits in young women with epilepsy is still uncertain and needs further investigation.
There are no large longitudinal studies on the development of VPA-related reproductive endocrine disorders in women with epilepsy. The present study aimed at prospectively evaluating the effects of VPA therapy on development of reproductive endocrine abnormalities, in particular PCOS, in Indian women with generalized epilepsy.
The present study was conducted at the department of Neurology, Postgraduate Institute of Medical Education and Research, Chandigarh, India, with the approval of Local Ethics Committee. Informed consent was obtained from all the patients. The type of epilepsy was classified according to the recommendations of the International League Against Epilepsy (1989).
Thirty consecutive, clinically normal and eumenorrheic reproductive age group women with generalized epilepsy, who were newly initiated on VPA, were recruited for the study. Exclusion criteria included progressive brain disease, other chronic diseases, pregnancy or lactation and use of oral contraceptives or other hormonal preparations, and previous use of antiepileptic drugs for last one year. The initial evaluation was done in all the patients prior to starting VPA and all the patients had been seizure free for at least 4 weeks.
All the patients were evaluated at inclusion, and six monthly follow-up studies were planned for next two years. However, most of the patients were irregular in follow-up, for the reasons best known to them. All patients followed up at least once (range: 6 months to 2 years), while 11 patients had more than two follow-ups. Medical history, with special reference to menstrual history, and detailed clinical evaluation were conducted, and EEG and MRI of brain were performed on each patient. At inclusion and on every follow-up, the patients were clinically examined by one of the authors. The height was measured to the nearest 0.5 cm with the Harpenden wall-mounted stadiometer and weight to the nearest 0.5 kg on electronic scales. Body mass index (BMI) was calculated (weight in kilograms divided by the square of height in meters). Women with a BMI > 25 kg/m2 were considered obese (Kiddy et al., 1990). Hirsutism was evaluated and scored by modified Ferryman Gallwey system (Ferriman and Gallwey, 1961), and a score of more than eight was considered significant hirsutism. On follow-up, menstrual disorders were categorized as amenorrhea (no menstruation), oligomenorrhea (cycle length longer than 35 days) during the last 6 months, or irregular menstrual cycles (cycle length varying more than 4 days from cycle to cycle, between 22 and 35 days, at least once during the last 6 months) (Kiddy et al., 1990). Weight gain of 4 kg or more compared with weight at inclusion was considered clinically relevant.
On each occasion, transabdominal USG was performed during the early follicular phase using a 3.5-MHz transducer to look for ovarian diameter, number and mean size of ovarian follicles, and stromal echogenicity. The ovaries were considered polycystic if there were ≥10 cysts, 2–8 mm in diameter arranged either peripherally around a dense core of stroma or scattered throughout an increased amount of stroma (Swanson et al., 1981). Venous blood samples were drawn after an overnight fast in the early follicular phase (day 3–5) of the menstrual cycle for the analysis of serum testosterone, dihydrotestosterone (DHT), luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin, dehydroepiandrosterone sulfate (DHEAS), and thyroid stimulating hormone (TSH). Serum hormone binding globulin (SHBG) assays were unfortunately not available at our endocrinology laboratory during the study period and hence serum SHBG and free serum testosterone levels could not be assessed.
The samples were centrifuged immediately, serum separated, and frozen at −20°C until analyzed. Serum testosterone and DHT concentrations were assayed using radioimmunoassay (RIA) as per WHO protocol (1983); LH, FSH, and prolactin and TSH were assayed by RIA using kits obtained from BARC (Mumbai, India) where as DHEAS was evaluated by RIA using kit from RADIM (Iberica S.A., Barcelona, Spain). Intraassay and interassay coefficients of variation for all the hormones were less than 8%. The upper limits for testosterone and DHT in reproductive age group women in early follicular phase used were 1.2 ng/ml and 1.5 ng/ml, respectively. A serum testosterone level more than 1.2 ng/ml was considered as hyperandrogenemia. The reference range for LH and FSH was 5 to 15 mIU/ml, with lower detection limit of 1.56 mIU/ml and 1.25 mIU/ml, respectively. The normal range for prolactin and DHEAS were 5 to 25 ng/ml and 0.5 to 1.6 ng/ml, respectively.
PCOS, i.e., the coexistence of menstrual abnormalities with hyperandrogenism or hyperandrogenemia, was diagnosed using the criteria laid down at NIH conference (Zawadski and Dunaif, 1992).
SPSS 10.0 software was used for analysis of data, with two-tailed, significant level at p < 0.05. Paired t-test was used to perform two time point analysis where in the results of the clinical and hormonal assessments before VPA was started were compared with the results of assessment after a specified period of VPA therapy, i.e., at 6 months, 1 year and 2 years. Repeated-measure analysis of variance (RANOVA) with the Fisher's least significant difference test was used to perform analysis of the results of yearly assessments over 2-year period. The results of nominal data or not normally distributed data were analyzed using Wilcoxon signed-rank test. Spearman's correlation analyses were performed to correlate weight change with the occurrence of menstrual abnormalities, hirsutism, PCO, and hormonal changes on follow-up. Changes exceeding two SD from the participant's mean serum hormonal levels at inclusion were considered to describe an increase or decrease in the individual hormonal values on follow-up.
At inclusion, the mean age of 30 women with generalized epilepsy was 18.5 years (range: 15–35 years) and the mean duration of disease was 3.3 years (range: 1 month to 20 years). None of the women were obese, had menstrual disturbances, hirsutism, or evidence of hypothyroidism. All the women younger than 18 years of age had attained menarche at least 3 years prior and had attained Tanner stage 5 of puberty at the time of inclusion (Tanner, 1962). The mean serum testosterone and DHEAS levels at baseline (1.4 ± 0.6 ng/ml and 1.7 ± 0.7 ng/ml, respectively) were observed to be higher than the normal range. Both PCO and hyperandrogenemia together were detected in two women; one of them lost to follow-up and hence excluded.
Only five women had first follow-up at 6 months. All of them remained seizure free and received similar VPA doses. Clinically relevant weight gain (maximum of 6 kg) was observed in three of these women (60%) and the mean BMI significantly increased from 19.8 to 21.1 kg/m2 (p = 0.037). Irregular menstrual cycles developed in one (20%) and PCO was detected in two (40%) of them. A nonsignificant increase in the mean serum levels of hormones was observed (testosterone 1.4 ± 0.7 to 2.0 ± 0.7 ng/ml, DHEAS 1.5 ± 0.7 to 1.8 ± 1.4 ng/ml, LH 2.3 ± 1.5 to 6.9 ± 3.2 mIU/ml and LH/FSH ratio 0.7 ± 0.3 to 2.2 ± 1.4). The serum testosterone and LH levels increased in one (20%) and three women (60%), respectively. One woman (20%) developed PCOS, with 6 kg weight gain, irregular menstrual cycles, hirsutism (HS score = 10), PCO, and increase in serum testosterone, DHEAS, LH, prolactin levels, and LH/FSH ratio.
Twenty-five women (mean age: 18.3 years, range: 15–28 years) completed 1-year follow-up. The clinical characteristics of these women at inclusion are summarized in Table 1. Only one patient had two seizures in the previous 6 months and required an increase in her VPA dose; rest remained seizure free on therapy. Eighteen women (72%) gained weight (range, 1 to 11 kg), clinically relevant weight gain was seen in 10 (40%) of them, the mean BMI significantly increased from 20.0 to 21.0 kg/m2 (p = 0.002), and four women (16%) became obese (p = 0.046) (Table 2). Menstrual abnormalities were seen in six women (24%) (oligomenorrhea in two and irregular menstrual cycles in four), including the one who developed irregular menstrual cycles at 6-month follow up. PCO developed in four women (17%), including the one with PCO at inclusion. The mean serum testosterone levels significantly increased from 1.53 to 1.76 nmol/L (p = 0.046), while serum LH/FSH ratio tended to be low (p = 0.08) (Table 2). The mean serum levels of other hormones studied did not change. We observed an increase in the levels of serum testosterone in 3 of 25 (12%), FSH in 3/24 (12.5%), DHT in 1 of 23 (4.3%), LH in 2 of 25 (8%), LH/FSH ratio in 1 of 24 (4.2%), and prolactin in 2 of 22 (9%) women. One woman with oligomenorrhea showed evidence of hypothyroidism (S. TSH = 21.5 μIU/ml) with a normal reproductive hormonal profile, and was treated with thyroxine. The woman with PCO at inclusion now additionally developed oligomenorrhea and an increase in serum testosterone and FSH levels. PCOS was diagnosed in five (20%) women; all of them showed clinically relevant weight gain and evidence of clinical and/or biochemical hyperandrogenism. Among women who remained eumenorrheic, an increase in serum testosterone and prolactin levels was seen in two women each and in serum LH, FSH, DHT, DHEAS levels, and LH/FSH ratio in one woman each. A significant positive correlation was found between weight change and the development of menstrual irregularities (r = 0.66, p < 0.0001), hirsutism (r = 0.53, p = 0.006) and PCO (r = 0.51, p = 0.012) on follow-up. No correlation existed between weight change and serum reproductive hormonal changes.
Table 1. Clinical characteristics at inclusion of women with epilepsy who completed 1-year follow up (n = 25)
Mean ± SD
18.3 ± 3.7
Duration of illness (years)
3.3 ± 3.9
Dose of valproate (mg)
724 ± 150.8
49.2 ± 7.4
20 ± 2.6
Table 2. Clinical characteristics and serum hormone concentrations in women with epilepsy at inclusion, and on valproate therapy 1 year later (n = 25)
At 1 year
Values are mean ± SD
a Significant p value
NS, not significant; N, number; PCO, polycystic ovaries; DHT, dehydrotestosterone; LH, luteinizing hormone; FSH, follicle stimulating hormone; DHEAS, dehydroepiandrosterone sulfate.
Valproate dose (mg)
724 ± 150.8
736 ± 211.9
49.2 ± 7.4
51.4 ± 7.7
20.0 ± 2.6
Menstrual irregularities (%)
S. Testosterone (ng/ml)
1.5 ± 0.7
1.8 ± 0.7
S. DHT (ng/ml)
1.6 ± 0.7
1.5 ± 0.7
S. LH (mIU/ml)
3.0 ± 1.5
2.8 ± 1.9
S. FSH (mIU/ml)
3.3 ± 1.1
3.8 ± 2.0
S. LH/FSH ratio
1.1 ± 0.8
0.9 ± 0.5
S. Prolactin (ng/ml)
13.5 ± 9.3
13.8 ± 8.9
1.6 ± 0.6
1.8 ± 1.4
Eleven of these 25 women completed 2-year follow up, including the three with menstrual abnormalities at 1-year follow-up. Their seizures remained well controlled on therapy. Over 2 years, six women (45.5%) had clinically relevant weight gain (maximum up to 18 kg), and a linear increase in the mean body weight (p = 0.023) and BMI (p = 0.008) was observed. Irregular menstrual cycles and PCO were seen in three (27.3%) and four (36.4%) women, respectively. The mean serum testosterone levels significantly increased further during the second year of therapy (p = 0.015). The mean serum LH levels, which tended to be low during first year of VPA therapy (p = 0.085), showed a significant increase during the second year (p = 0.011) (Table 3). We observed an increase in the levels of serum LH in three (27.3%), DHT in two (18.2%), testosterone, LH/FSH ratio, and prolactin in one (9.1%) of women. Both the women with menstrual abnormalities at 1-year follow-up persisted to have irregular menstrual cycles and hirsutism; the woman with hypothyroidism became euthyroid now while the other one persisted to have PCO and showed a further increase in serum testosterone levels. A woman, who at 1-year follow-up was eumenorrheic but had hirsutism, PCO, and hyperandrogenimia, now developed irregular menstrual cycles and a further rise in serum testosterone levels. Hence, PCOS was diagnosed in two women (18.2%). Among eight women who remained eumenorrheic, PCO and an increase in serum testosterone, DHT and LH levels were seen in one woman, clinically relevant weight gain and an increase in serum FSH and prolactin levels in another woman, and weight gain with an increase in LH and FSH levels in two more women.
Table 3. Clinical characteristics and serum hormone concentrations in women with epilepsy at inclusion, and on valproate therapy 1 year and 2 years later (n = 11)
At 1 Year
At 2 Years
Values are mean ± SD; N, number.
aTest performed in 7 patients only.
bp=0.023, repeated measures ANOVA with Fisher's LSD, compared with at baseline.
cp=0.031, repeated measures ANOVA with Fisher's LSD, compared with at 1 year.
dp=0.047, repeated measures ANOVA with Fisher's LSD, compared with at baseline.
ep=0.008, repeated measures ANOVA with Fisher's LSD, compared with at baseline.
fp=0.015, repeated measures ANOVA with Fisher's LSD, compared with at baseline.
gp=0.011, repeated measures ANOVA with Fisher's LSD, compared with at 1 year.
Six of these 11 women further followed up once in next 9 months to a year (2.75 to 3 years on therapy). Compared with the hormonal profile at inclusion, the mean serum testosterone significantly increased at three years (1.1 ± 0.5 to 1.8 ± 0.7 ng/ml, p = 0.025), while serum LH levels and LH/FSH ratio tended to be high (2.0 ± 0.7 to 5.0 ± 3.1 mIU/ml, p = 0.07 and 0.6 ± 0.3 to 1.3 ± 0.9, p = 0.06, respectively). We observed an increase in the levels of serum testosterone and LH in two (33.3%), and DHT and LH/FSH ratio in one (20%) of the women. Clinically relevant weight gain, hirsutism, irregular menstrual cycles, and PCO persisted in the three women with menstrual abnormalities on previous follow-up. Lamotrigine (100 mg/day) was added to VPA therapy in one of them but her serum testosterone and DHT levels remained high; the other woman could not afford lamotrigine due to financial reasons and hence was continuing on VPA. The woman on thyroxine continued to be euthyroid but persisted to have menstrual abnormalities with normal reproductive hormonal profile. Among women who remained eumenorrheic, a new increase in serum LH and FSH levels was detected in one woman, and in testosterone and LH levels in another woman. The serum testosterone levels of these six women at inclusion and serially on VPA therapy are shown in Figure 1.
Long-term VPA therapy in Indian women with generalized epilepsy was associated with significant weight gain, development of hirsutism, menstrual abnormalities, and alterations in the reproductive hormonal function during the first year of medication and these changes seemed to be stable or progressive over next 2 years of therapy. The mean serum concentration of testosterone showed a significant and linear increase and that of gonadotropins tended to increase with time on VPA therapy.
We observed menstrual abnormalities in 24% and PCOS in 20% of women after 1 year of VPA therapy and these abnormalities persisted in women who were followed up for next 2 years. The frequency of PCOS was higher in our patients as compared to the reported prevalence of PCOS of 4–6.7% in general population (Herzog et al., 1996; Diamanti-Kandarakis et al., 1999).
We observed elevated mean serum testosterone and DHEAS levels and elevated individual testosterone levels above normal in some women at inclusion. On VPA therapy, we detected clinical and/or biochemical hyperandrogenism and PCO appearing prior to the appearance of menstrual abnormalities in some women. Also, the serum testosterone levels remained high or continued to rise in women who developed PCOS on follow-up. Rättyä et al. (2001) have reported an increase in serum testosterone and androstenedione levels as early as 1–3 months after initiation of VPA in almost 50% of their patients, in absence of weight gain or clinical signs of hormonal disorders. High serum testosterone concentrations may be a factor leading to the arrest of follicular maturation and eventually to the development of PCO in these women (Hsueh et al., 1984). Also, in some of the women who remained eumenorrheic, we observed an increase in gonadotropin and prolactin levels and LH/FSH ratio over time. These women correspondingly showed variable-to-significant increase in their serum testosterone levels and one of them developed PCO but remained eumenorrheic. This suggests an alteration in pituitary-hypothalamic ovarian axis and stimulation of ovarian androgen synthesis via increased LH secretion. DHEAS levels also increased in some women over time, suggesting possible direct effect of VPA on ovarian androgen synthesis.
A number of limitations precluded us from generalizing the results of this study. The main limitations are small sample size and large dropout rate during follow-up, thus constricting the power of the longitudinal analysis. This deficit reflects the complexity of collecting longitudinal data in Indian patients over a long follow-up period, major reasons being poor compliance with therapy after adequate seizure control, financial constraints affecting ability to buy drugs over long period of time, and girls stopping drugs and follow-up after marriage due to stigma attached to the disease. Also, women with clinical problems like weight gain and menstrual abnormalities are more likely to follow-up, producing a sample bias. Despite these limitations, these results provide important findings as the longitudinal evaluation of reproductive endocrine function in women with epilepsy on long-term VPA therapy, where subjects served as their own controls.
The potential pathophysiology underlying a VPA-mediated increase in PCOS occurrence is likely to be multifactorial with several factors acting simultaneously. VPA may influence the risk of PCOS by its CNS activity, increasing γ-aminobutyric acid synthesis and release and may block N-methyl-d-asparatate-type glutamate receptors (Loscher, 1999; Mosche, 2000). Another explanation for why VPA increases the risk of PCOS is related to the association between VPA use and weight gain, which increases insulin resistance (Isojärvi et al., 1996). However, obesity alone is unlikely to fully explain the association of VPA use with PCOS as some VPA-treated epileptic women are lean (Isojärvi et al., 1996; Bilo et al. 2001), and other medications that cause significant weight gain (e.g., clozapine) have not been associated with PCOS. Weight gain is possibly an additional factor contributing to the expression of PCOS in these women. Epilepsy may be required as a component factor in the association between VPA use and PCOS. The prevalence of PCOS is not well known in women receiving VPA for bipolar disorder. A small pilot study of PCOS in 22 women with bipolar disorder found no association of valproate with PCOS (Rasgon et al., 2000). In contrast, O'Donovan et al. (2002) reported high rates of PCOS (41%) and menstrual disturbances (47% vs. 13% not on VPA) in their study of women with bipolar disorder receiving VPA. McIntyre et al. (2003) also reported higher rates of menstrual abnormalities (50%) in women with bipolar disorder on VPA compared to those on lithium (15%). However, in a 2-year longitudinal evaluation of women with bipolar disorder on various mood stabilizing agents including VPA, Rasgon et al. (2005) did not observe any change in rates of oligomenorrhea and hyperandrogenism, although serum total testosterone levels increased over time in VPA-treated women. It is plausible that ictal and interictal discharges and altered GnRH pulsatility may make women with epilepsy particularly susceptible to the effects of VPA on hypothalamic-pituitary-ovarian axis (Morrell, 1999), and hence contribute significantly to the increased risk for PCOS.
In summary, the present study demonstrates a link between development of weight gain, altered reproductive hormonal profile, particularly that of serum testosterone, and PCOS in VPA-treated women with generalized epilepsy over time. Our findings raise concern about the long-term use of VPA as AED therapy in young women with epilepsy. Weight gain and related risks should be taken into consideration when choosing AED therapy for these women, as suggested by Isojärvi et al. (1998). Further randomized long-term longitudinal studies are needed to establish the consequences of these reproductive and hormonal changes.