Neurosteroid Withdrawal Model of Perimenstrual Catamenial Epilepsy


  • Doodipala S. Reddy,

    1. Neuronal Excitability Section, Epilepsy Research Branch, National Institute of Neurological Disorders and Stroke; and
    Search for more papers by this author
  • Hee-Yong Kim,

    1. Laboratory of Membrane Biochemistry and Biophysics, National Institute of Alcohol Abuse and Alcoholism, National Institutes of Health, Rockville, Maryland, U.S.A.
    Search for more papers by this author
  • Michael A. Rogawski

    1. Neuronal Excitability Section, Epilepsy Research Branch, National Institute of Neurological Disorders and Stroke; and
    Search for more papers by this author

Address correspondence and reprint requests to Dr. M. A. Rogawski at Epilepsy Research Branch, National Institute of Neurological Disorders and Stroke, National Institutes of Health, 10 Center Drive Room 5N-250 MSC 1408, Bethesda, MD 20892-1408, U.S.A. E-mail:


Summary:  Purpose: Perimenstrual catamenial epilepsy, the increase in seizure frequency that some women with epilepsy experience near the time of menstruation, may in part be related to withdrawal of the progesterone metabolite allopregnanolone, an endogenous anticonvulsant neurosteroid that is a potent positive allosteric γ-aminobutyric acidA (GABAA) receptor modulator. The objective of this study was to develop an animal model of perimenstrual catamenial epilepsy for use in evaluating drug-treatment strategies.

Methods: A state of prolonged high serum progesterone (pseudopregnancy) was induced in 26-day-old female rats by sequential injection of pregnant mares' serum gonadotropin and human chorionic gonadotropin. Neurosteroid withdrawal was induced by treatment with finasteride (100 mg/kg, i.p.), a 5α-reductase inhibitor that blocks the conversion of progesterone to allopregnanolone. Plasma progesterone and allopregnanolone levels were measured by gas chromatography/electron capture negative chemical ionization mass spectrometry. Seizure susceptibility was evaluated with the convulsant pentylenetetrazol (PTZ).

Results: Plasma allopregnanolone levels were markedly increased during pseudopregnancy (peak level, 55.1 vs. control diestrous level, 9.3 ng/mL) and were reduced by 86% 24 h after finasteride treatment (6.4 ng/mL). Progesterone levels were unaffected by finasteride. After finasteride-induced withdrawal, rats showed increased susceptibility to PTZ seizures. There was a significant increase in the number of animals exhibiting clonic seizures when challenged with subcutaneous PTZ (60 mg/kg) compared with control pseudopregnant animals not undergoing withdrawal and nonpseudopregnant diestrous females. The CD50 (50% convulsant dose) was 46 mg/kg, compared with 73 mg/kg in nonwithdrawn pseudopregnant animals and 60 mg/kg in diestrous controls. The threshold doses for induction of various seizure signs, measured by constant intravenous infusion of PTZ, were reduced by 30–35% in neurosteroid-withdrawing animals compared with control diestrous females. No change in threshold was observed in pseudopregnant rats treated from days 7 to 11 with finasteride, demonstrating that high levels of progesterone alone do not alter seizure reactivity.

Conclusions: Neurosteroid withdrawal in pseudopregnant rats results in enhanced seizure susceptibility, providing an animal model of perimenstrual catamenial epilepsy that can be used for the evaluation of new therapeutic approaches.

A hallmark of epilepsy is the unpredictable occurrence of seizures. However, in many women with epilepsy, seizures do not occur randomly but cluster in association with the menstrual cycle. Based on the review of a vast clinical experience, Newmark and Penry (1) defined catamenial epilepsy as epileptic seizures occurring in women of fertile age exclusively or significantly more often during a 7-day period of the menstrual cycle beginning 3 days before menstruation and ending 4 days after its onset. With this criterion as a rough guideline, catamenial epilepsy has been reported to occur in 10–72% of women with epilepsy (2–4). Recently, Herzog et al. (5) proposed an extension of the definition of catamenial epilepsy to include preovulatory and anovulatory forms. However, these authors observed that the conventional perimenstrual form of Newmark and Penry is the most common clinical type. In perimenstrual catamenial epilepsy, seizures decrease in the midluteal phase when serum progesterone levels are high and increase premenstrually when progesterone levels decrease and there is a decrease in the serum progesterone to estrogen ratio (6,7). The premenstrual exacerbation of seizures has been ascribed to the withdrawal of the antiseizure effects of progesterone (8).

Progesterone has long been known to have anticonvulsant activity in animal seizure models, and in clinical studies, progesterone has been found to reduce the frequency of interictal spikes and lessen the risk of seizures (9–12). Progesterone elicits anticonvulsant effects as a consequence of its metabolic conversion to the endogenous neurosteroid allopregnanolone (3α-hydroxy-5α-pregnan-20-one) (13,14), a powerful positive allosteric modulator of γ-aminobutyric acidA (GABAA) receptors (15–17). Perimenstrual catamenial epilepsy may therefore be, at least in part, attributed to withdrawal of allopregnanolone (6,18–20).

Despite the increased awareness and understanding of catamenial epilepsy, there are few specific treatment approaches. The dearth of attention to the development of therapies may be due to the lack of an appropriate animal model. Our objective in the present study was therefore to develop a convenient animal model of perimenstrual catamenial epilepsy for drug evaluation. A state of chronically elevated serum progesterone and allopregnanolone was induced in immature female rats with gonadotropins. Allopregnanolone was abruptly withdrawn by administration of finasteride, a selective blocker of 5α-reductase that in rats inhibits progesterone conversion to allopregnanolone in various tissues including brain (21,22). Neurosteroid withdrawal was associated with increased seizure susceptibility, mimicking the situation in catamenial epilepsy.



Female 25- or 26-day-old (70–80 g) and 40- to 50-day-old (150–200 g) Sprague–Dawley rats (Taconic) were housed in groups of four under a 12-h/12-h light/dark cycle in an environmentally controlled animal facility. The immature animals began the pseudopregnancy protocol within 48 h after arrival, but were not tested until they had reached an age (41 days) similar to that of the older adult animals that served as controls. All animals were allowed to acclimatize with free access to food and water for ≥24 h before use. All procedures were performed in strict compliance with the NIH Guide for the Care and Use of Laboratory Animals under a protocol approved by the NIH Animal Use Committee.

Induction of pseudopregnancy and neurosteroid withdrawal

The sequence of hormonal and drug treatments for the induction of pseudopregnancy and neurosteroid withdrawal is summarized in Fig. 1. The protocol for gonadotropin priming to induce pseudopregnancy is based on Kim and Greenwald (23). In brief, all subjects received pregnant mares' serum gonadotropin (20 IU/rat, s.c.) on postnatal day 27 at 10:00 h, followed 48 h later by human chorionic gonadotropin (10 IU/rat, s.c.). The day of human chorionic gonadotropin treatment (day 29) was considered day 0 of pseudopregnancy. On day 11 (11:00 h) of pseudopregnancy, the rats received an injection of finasteride in 50%β-cyclodextrin (100 mg/kg, i.p.) to block the enzymatic conversion of progesterone to allopregnanolone. Control pseudopregnant animals received an injection of the vehicle alone. All animals in these two groups were tested for seizure sensitivity on day 12, 24 h after finasteride or vehicle administration. A second control group consisted of adult cycling female rats in the diestrous phase of their cycles (associated with low plasma progesterone levels). The stage of the estrous cycle was evaluated by microscopic evaluation of the vaginal lavage as described (24). As an alternative to the use of finasteride, in a few animals, withdrawal was induced on day 11 of pseudopregnancy by ovariectomy under anesthesia induced by i.p. injection of a mixture of ketamine (100 mg/kg) and xylazine (50 mg/kg).

Figure 1.

Diagrammatic illustration of the pharmacologic treatments for induction of pseudopregnancy and neurosteroid withdrawal.

Estimation of plasma progesterone and neurosteroids

Blood was collected from the trunk into heparinized tubes under CO2 anesthesia. The ovaries were dissected from the adhering tissue and quickly weighed. Plasma was separated by spinning at 12,000 rpm for 10 min at 4°C. To the clear supernatant, 10 μL of 10% formic acid was added, and the samples were frozen immediately and stored at −20°C until assayed for steroids. Progesterone and neurosteroids (dihydroprogesterone and allopregnanolone) in plasma samples were quantified by gas chromatography/electron capture negative chemical ionization mass spectrometry as described previously (25). In brief, 50-μL plasma samples were spiked with 1 ng of deuterium-labeled standards and applied onto 100 mg C18 SPE columns (Varian Sample Preparation Products, Harbor City, CA, U.S.A.) that had been preconditioned with 4 mL methanol and 4 mL deionized water. After the loaded columns were washed with 4 mL deionized water, steroids were eluted with 2 mL methanol. Steroids were derivatized by reacting with 50 μL of 0.2% carboxymethoxylamine hemihydrochloride in pyridine at 60°C for 45 min, 100 μL of 1.25% pentafluorobenzyl bromide and 2.5% diisopropylethylamine in acetonitrile at 45°C for 20 min, and 100 μL of 50%bis(trimethylsilyl)trifluoroacetamide in acetonitrile at 45°C for 30 min. After each step, the reaction mixture was dried completely under a stream of N2. The fully derivatized steroids were dissolved in 5 μL of hexane and injected onto the thin-film capillary column (15 m × 0.25 mm, 0.05-μm film thickness; Quadrex Corporation, New Haven, CT, U.S.A.) of a Hewlett-Packard 5890/5989A gas chromatograph–mass spectrometer operated in the selected ion-monitoring mode. The oven temperature was raised from 150°C to 230°C at 30°C/min, 230°C to 250°C at 1°C/min, and 250°C to 320°C at 30°C/min. The injector and transfer line temperatures were maintained at 300°C and 310°C, respectively. The temperatures of the mass spectrometer source and quadrupole were set at 200°C and 100°C, respectively.

Pentylenetetrazol seizure test

Seizure sensitivity of the animals was evaluated using the subcutaneous (s.c.) PTZ test (14). One day after finasteride or vehicle treatment, each treatment group was divided into six subgroups, and the seizure reactivity of the animals was assessed by injecting s.c. one of six doses of PTZ within the range from 30 to 100 mg/kg. Rats were observed for a 30-min period after PTZ injection to determine the incidence and latency of clonic seizures. Rats exhibiting clonic seizures lasting >5 s were scored as positive for the seizure occurrence. If no clonic seizures appeared during this time, the animals were considered protected and were given a maximal latency score of 30 min.

Pentylenetetrazol seizure-threshold test

Seizure-threshold determinations were made by constant intravenous (i.v.) infusion of PTZ. The lateral tail vein of each rat was catheterized with a 25-gauge stainless steel butterfly needle (Becton-Dickinson Vacutainer Systems, Franklin Lakes, NJ, U.S.A.) attached to a 0.3-m section of polyethylene tubing that was secured to the tail by taping lightly. The tubing was attached to a 12-mL plastic syringe containing the PTZ solution (20 mg/mL in sterile saline) and mounted on a Harvard infusion pump. The tubing was of sufficient length to allow the rat to move freely in an open cage during the testing procedure. The infusion rate was set at 0.34 mL/min so that the total infusion time would not exceed 3–4 min to minimize discomfort. Three convulsion signs, which are typically observed in sequence, are used as end points in the PTZ threshold determination (26). These include (a) “myoclonic twitch” (sudden involuntary jerking of the whole body), (b) “forelimb clonus” (repeated clonic jerking of the head, neck, and forelimbs), and (c) “tonic hindlimb extension” (extreme rigidity of the body with forelimbs and hindlimbs extended caudally). The times between the start of infusion and onset to first myoclonic twitch, forelimb clonus, and tonic hindlimb extension were recorded in seconds and subsequently converted to threshold convulsant dose in milligrams of PTZ per kilogram of body weight as follows:

inline image

The infusion was terminated when the animals exhibited tonic hindlimb extension, and the animals were rapidly killed by CO2 anesthesia.

Drugs and hormones

Pregnant mares' serum gonadotropin and human chorionic gonadotropin were dissolved in sterile saline. Finasteride was made fresh in aqueous 50% hydroxypropyl-β-cyclodextrin (β-cyclodextrin; Research Biochemicals, Natick, MA, U.S.A.). Drug solutions were administered in a volume equaling 1% of the animal's body weight. All drugs and hormones were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.).

Data analysis

The significance of differences in the mean plasma concentrations was assessed by one-way analysis of variance (ANOVA), followed by Dunnett's t test. Group comparisons in the percentage of animals exhibiting seizures in the s.c. PTZ test were made with the Fisher's exact probability test. The significance of differences in the mean latency values in the s.c. PTZ test was determined using the Kruskal–Wallis one-way ANOVA by ranks, with post hoc comparisons using the Mann–Whitney U test. CD50 and CD97 values (the doses at which 50% and 97% of tested animals exhibited convulsions, respectively) with 95% confidence limits in the s.c. PTZ test were determined by log-probit analysis using the Litchfield and Wilcoxon procedure. The significance of differences between PTZ dose–response curves was assessed with the Litchfield and Wilcoxon χ2 test. The slope factor (S) was calculated from the CD16, CD50, and CD84 doses using the equation [(CD84/CD50+ CD50/CD16)/2]. In the construction of dose–response curves, at least six animals were tested at each dose. Comparisons of mean PTZ doses in the PTZ threshold test were made by one-way ANOVA, followed by Dunnett's t test. In all tests, the criterion for statistical significance was p < 0.05. Statistical analyses were carried out with PHARM/PCS Version 4.2 (Microcomputer Specialists, Philadelphia, PA, U.S.A.) except Fisher's exact test, which was with SYSTAT Version 7.0 (SPSS, Chicago, IL, U.S.A.).


Allopregnanolone levels

Figure 2 summarizes the results of measurements of plasma progesterone, dihydroprogesterone, and allopregnanolone in control diestrous female rats, in animals at various days of pseudopregnancy, and in pseudopregnant animals that had received finasteride on day 11. The mean plasma concentration of progesterone was markedly elevated during pseudopregnancy, reaching a maximum on day 11 that was 12-fold greater than the mean value in control animals. Allopregnanolone and dihydroprogesterone also were elevated, reaching peak values that were respectively 6- and 14-fold greater than the value in diestrous controls. Finasteride (100 mg/kg, i.p.) significantly reduced allopregnanolone (86%) and dihydroprogesterone (86%) to levels equivalent to that in diestrous controls, but did not significantly affect progesterone. The mean ovarian weight in animals on pseudopregnancy day 12 was 168 ± 6 vs. 77 ± 3 mg in diestrous controls (n = 20; p < 0.01). Plasma levels of all three steroids were significantly correlated with increased ovarian weight (r = 0.87; p < 0.05). The increase in ovarian weight during pseudopregnancy was not affected by finasteride (148 ± 5 vs.169 ± 6 mg in vehicle-treated pseudopregnant controls; n = 16; p > 0.05).

Figure 2.

Plasma concentrations of progesterone, dihydroprogesterone, and allopregnanolone during pseudopregnancy and after finasteride administration compared with levels during diestrous. Values are shown for diestrous controls, for days 10–13 of pseudopregnancy, and for 24 h after finasteride (100 mg/kg, i.p.) administration on day 11 of pseudopregnancy. Data are expressed as mean ± SEM of values from 20 rats, analyzed in triplicate. *p < 0.01 versus control; #p < 0.01 versus pseudopregnancy day 12 (Dunnett's t test).

Pentylenetetrazol seizure test

Susceptibility to PTZ-induced clonic seizures was examined in three groups of animals: (a) control diestrous females, (b) vehicle-treated pseudopregnant females (“pseudopregnant controls”), and (c) finasteride-treated pseudopregnant females (“withdrawn”). In all three groups, PTZ (30–100 mg/kg, s.c.) produced a dose-dependent increase in the fraction of animals exhibiting clonic seizures (Fig. 3). There was a corresponding dose-dependent decrease in the latency to seizure onset with increasing PTZ dose (Fig. 4). However, the fraction of animals exhibiting seizures in response to the 60-mg/kg dose of PTZ was greater in the withdrawn pseudopregnant animals (that had received finasteride 24 h before testing) than in the diestrous and pseudopregnant control groups (Fig. 3). In addition, the mean latencies to onset of PTZ-induced clonic seizures were significantly shorter in withdrawn animals than in pseudopregnant controls but not diestrous controls (Fig. 4). The increase in seizure susceptibility observed 24 h after withdrawal returned to diestrous control-like values when assessed 48 h after withdrawal (data not shown). A summary of the CD50 and CD97 values and the slope factors for the dose–response data in the s.c. PTZ test is given in Table 1. The CD50 and CD97 values in withdrawn animals were significantly less than in diestrous control or nonwithdrawn control pseudopregnant animals. The slope factors did not significantly differ between groups, indicating that the shifts in the dose–response curves are parallel.

Figure 3.

Pentylenetetrazol (PTZ) induction of clonic seizures in control, pseudopregnant, and neurosteroid withdrawn rats. PTZ (30–100 mg/kg, s.c.) was administered 24 h after neurosteroid withdrawal (by administration of finasteride) or vehicle treatment. Each point represents data from between six and 13 animals. *p < 0.05 versus control; #p < 0.05 versus pseudopregnancy (Fisher's exact probability test).

Figure 4.

Latency to the appearance of pentylenetetrazol (PTZ)-induced clonic seizures in control, pseudopregnant, and neurosteroid withdrawn rats. PTZ (30–100 mg/kg, s.c.) was administered 24 h after neurosteroid withdrawal or vehicle treatment. Values are expressed as mean ± SEM of data from six to 13 animals. *p < 0.05 versus control; #p < 0.05 versus pseudopregnant (Mann–Whitney U test).

Table 1.  CD50 and CD97 values of PTZ for induction of clonic seizures in control diestrous, vehicle-treated pseudopregnant, and finasteride-treated (neurosteroid-withdrawn) pseudopregnant animals
Slope factor
  • Numbers in parentheses are 95% confidence intervals. CD50 and CD97 values are in mg/kg.

  • a

     p < 0.05 versus control (Fisher's exact test, at 60 mg/kg PTZ).

  • b  p < 0.05 versus pseudopregnant (Litchfield and Wilcoxon χ2 test). Slope factors are not significantly different.


In five pseudopregnant animals, withdrawal was induced by surgical ovariectomy instead of by finasteride treatment. Four of these animals exhibited clonic seizures in response to a low challenge dose of PTZ (50 mg/kg, s.c.), indicating an increase in seizure susceptibility.

Pentylenetetrazol seizure thresholds

To provide a more sensitive assessment of changes in seizure susceptibility after neurosteroid withdrawal, we determined the minimal dose of PTZ required for induction of various convulsion signs by constant i.v. infusion of PTZ. The mean threshold values are presented in Fig. 5. The threshold for myoclonic twitch and forelimb clonus was significantly lower in withdrawn animals than in either nonwithdrawn pseudopregnant animals or diestrous controls. The threshold for onset to tonic hindlimb extension in withdrawn animals was significantly lower than that in nonwithdrawn pseudopregnant animals; the value was also lower than in diestrous controls, but the difference was not statistically significant. Overall, in withdrawn animals, there was a 30–35% reduction in the PTZ dose required to elicit the various seizure signs in comparison with diestrous controls. Although the threshold dose of PTZ for all three convulsion signs was higher in nonwithdrawn pseudopregnant animals than in diestrous controls, the difference did not attain statistical significance (p < 0.1).

Figure 5.

Threshold dose of intravenous pentylenetetrazol (PTZ) for induction of various convulsion signs in diestrous control rats, pseudopregnant rats, and pseudopregnant rats 24 h after neurosteroid withdrawal. Values are expressed as mean ± SEM of data from seven to nine animals. *p < 0.05 versus control; #p < 0.05 versus pseudopregnant (Dunnett's t test).

Long-term finasteride treatment

To assess whether the increase in seizure susceptibility occurring in response to allopregnanolone withdrawal is specifically related to the acute decrease in neurosteroid levels, animals were treated daily with finasteride (100 mg/kg, i.p.) from days 7 to 11 of pseudopregnancy to create a state of persistently low allopregnanolone levels in the face of elevated progesterone and estrogen. Seizure thresholds were then determined on day 12 as in the acute withdrawal protocol. In contrast to the situation with acute withdrawal, there were no significant changes in the thresholds for any of the three convulsion signs in the long-term finasteride treatment animals (Fig. 6). These results confirm that the increase in seizure susceptibility requires acute withdrawal from chronically elevated allopregnanolone and indicate that chronically elevated estrogen does not affect seizure susceptibility. The results also support the conclusion that the increase in susceptibility in the acute withdrawal experiments is not related to actions of finasteride, apart from its effects on neurosteroid levels.

Figure 6.

Threshold dose of intravenous pentylenetetrazol (PTZ) for induction of various convulsion signs in pseudopregnant rats receiving vehicle or finasteride (100 mg/kg) on days 7–11 of pseudopregnancy. Seizure threshold was assessed 24 h after the last dose of finasteride. Values are expressed as mean ± SEM of data from seven to nine animals. Differences between mean seizure threshold values in vehicle- and finasteride-treated groups were not significant (p > 0.05; Dunnett's t test).


The main observation in the present study is that finasteride-induced withdrawal from persistently high neurosteroid levels in pseudopregnant rats is associated with enhanced susceptibility to PTZ seizures. The decrease in neurosteroids induced by finasteride is comparable to that which occurs in women of reproductive age around the time of menstruation, providing a basis for the hypothesis that the enhanced seizure reactivity has a similar genesis as perimenstrual catamenial epilepsy and could represent a model of the disorder. As in the luteal phase of the menstrual cycle, a state of persistently elevated progesterone and progesterone-derived neurosteroids was produced in immature female rats with a sequential regimen of gonadotropins (27–29). The animals first received pregnant mares' serum gonadotropin (containing follicle-stimulating hormone and luteinizing hormone activities) to stimulate the development of multiple ovarian follicles. This was followed by human chorionic gonadotropin (which substitutes for luteinizing hormone that normally triggers ovulation at midcycle) to induce superovulation. In the superovulated state, the ovaries contain a large mass of corpora lutea that secrete progesterone in a pulsatile manner for ≤14 days (30). Although pregnant mares' serum gonadotropin alone can also induce pseudopregnancy, progesterone levels are higher, and the high levels are maintained longer with the sequential regimen. Pseudopregnancy also can be induced through various types of neural stimulation (31–33), but these methods are not as reliable and do not produce as high levels of progesterone (34,35).

Progesterone is converted to the GABAA-receptor–modulating neurosteroid allopregnanolone by two sequential reductions catalyzed by 5α-reductase (finasteride sensitive in the rat) and 5α-hydroxysteroid oxidoreductase, with 5α-dihydroprogesterone as intermediate (36). Thus we were able to verify that the augmented circulating progesterone during pseudopregnancy is associated with increases in the plasma concentrations of dihydroprogesterone and allopregnanolone. Administration of finasteride causes a decrease in plasma dihydroprogesterone and allopregnanolone to near control levels. Brain concentrations of allopregnanolone are expected to increase in parallel with that of plasma levels (37,38). Thus the increase in plasma allopregnanolone concentrations during pseudopregnancy mimics the high levels of neurosteroids during the luteal phase of the menstrual cycle (20,39), whereas the finasteride-induced withdrawal is similar to the marked decrease in neurosteroid concentrations that occurs before the start of menstruation (20). Indeed, after finasteride treatment, allopregnanolone levels were reduced by 86% in pseudopregnant animals. In a recent study by Concas et al. (40), finasteride produced similar decreases in the concentrations of allopregnanolone in the plasma of pregnant rats that lasted >19 h. In this study allopregnanolone levels were reduced to a greater extent in the brain than in plasma.

The pseudopregnancy paradigm has some advantages over other approaches to delivering progesterone or neurosteroids, such as intermittent injections or the use of implanted capsules that release their contents at a constant rate. In pseudopregnancy, secretion of progesterone by the luteinized ovaries occurs in a physiologically appropriate episodic fashion and leads to plasma progesterone levels that are within the physiologic range (30). Moreover, the 9-day elevation of progesterone and allopregnanolone levels (days 4–13) in our pseudopregnancy model (23) closely matches the 10-day increase in allopregnanolone levels during the luteal phase of the human menstrual cycle (20,39). In addition, the magnitude of the increase in serum progesterone is comparable to the 6- to 8-fold increase that occurs in women during the normal menstrual cycle (20,39). In contrast, the fluctuations in progesterone and allopregnanolone levels in true pregnancy may differ (40). In the development of the present model, we used finasteride rather than ovariectomy to induce withdrawal from neurosteroids because ovariectomy would be associated with a decrease in estrogens as well as neurosteroids. Ovariectomy would therefore not simulate the reduced progesterone (and allopregnanolone)-to-estrogen ratio that is believed to be critical to perimenstrual catamenial epilepsy (19). Nevertheless, ovariectomized pseudopregnant animals did exhibit an increase in seizure susceptibility, indicating that maintained estrogen is not required for enhanced seizure reactivity.

Our data show that withdrawal from elevated levels of neurosteroids in pseudopregnant rats increases seizure susceptibility. This increase was observed in the s.c. PTZ test with a 60-mg/kg dose of PTZ. There was no change in the latency to seizure onset in the withdrawn animals compared with diestrous controls, although there was a significant decrease in latency compared with pseudopregnant controls. Differences in seizure susceptibility are difficult to demonstrate in the all-or-none s.c. PTZ test unless the appropriate dose is chosen and a large number of animals are used. The i.v. PTZ threshold test, which examines various components of the response to PTZ, provides a graded measure of sensitivity of each component and avoids the possibility of selecting an incorrect dose. After neurosteroid withdrawal, there was a decrease of 30–35% in the threshold to elicit the convulsion end points, confirming the increase in seizure susceptibility seen in the s.c. PTZ test. We have previously shown that finasteride does not affect the convulsant effects of PTZ or alter the anticonvulsant activity of allopregnanolone (14), indicating that it does not act as a proconvulsant apart from inhibition of 5α-reductase activity. In the present study, we also observed an increase in seizure susceptibility after surgical ovariectomy, providing additional evidence that finasteride is not responsible for the enhanced seizure susceptibility. When finasteride was administered repeatedly to pseudopregnant rats on 5 successive days, there was no change in PTZ threshold, confirming that withdrawal from elevated neurosteroids and not elevated progesterone is critical to the enhanced seizure susceptibility. In addition, this experiment further confirms that finasteride is not proconvulsant. Overall, our results are in agreement with the recent reports of Smith et al. (41) and Moran and Smith (42), showing that progesterone withdrawal results in an increase in seizure-like behavioral activity to the convulsants picrotoxin, a GABAA-receptor antagonist, and methyl-β-carboline-3-carboxamide, a benzodiazepine-receptor inverse agonist.

The basis for the increased seizure susceptibility after neurosteroid withdrawal in pseudopregnant rats is not well understood. An alteration in the pharmacokinetic properties of PTZ produced by gonadotropin priming is unlikely to be a factor because pseudopregnant animals not undergoing withdrawal did not exhibit increases in seizure susceptibility. Indeed, these animals showed a trend toward reduced sensitivity in the s.c. PTZ seizure test and toward increased convulsant thresholds, likely attributable to the presence of greater levels of allopregnanolone (or other GABAA-receptor potentiating neuroactive steroids). In addition, however, changes in the number or properties of GABAA receptors could be a factor in the reduced convulsant sensitivity, inasmuch as pregnant rats exhibit an increase in the densities of brain [3H]GABA- and [3H]flunitrazepam-binding sites (43).

Smith et al. (41,44,45) have proposed that progesterone withdrawal is accompanied by alterations in the expression of GABAA-receptor subunits and a consequent change in GABAA-receptor properties that causes reduced inhibition and an overall increase in brain excitability. Specifically, these workers reported increased expression of the GABAA-receptor α4 subunit, which was associated with an acceleration in the decay of GABAA-receptor currents in CA1 hippocampal neurons. Conversely, in hypothalamic magnocellular oxytocin-secreting neurons, there is increased α1 GABAA-receptor subunit expression during pregnancy that is associated with enhanced excitability of these neurons when neurosteroid levels decrease at the time of parturition (46,47). Synaptic currents mediated by GABAA receptors containing the α1 subunit are highly allopregnanolone sensitive and, in the presence of the steroid, decay slowly, leading to strong synaptic inhibition (48). However, in the absence of steroids, synaptic currents mediated by these receptors decay rapidly (49) so that there is a relative decrease in inhibition in association with the decrease in circulating allopregnanolone at the time of parturition. In addition, at the end of pregnancy, there is a gradual reduction in α1-subunit expression, leading to a relative increase in the number of α2-containing, steroid-insensitive receptors, which also decay faster than alloprenanolone-potentiated α1-containing receptors (47,50). Both of these factors lead to disinhibition and increased excitability of oxytocin neurons. It is tempting to speculate that similar changes could occur in brain areas relevant to seizure susceptibility, and it is noteworthy that the hippocampus is one of the few areas in the adult brain (other than the hypothalamus) that express both α1 and α2 subunits (51). However, Concas et al. (40) failed to observe any change in the expression of the α1-4, β1-3, and γ2S GABAA-receptor subunits in rat cerebral cortex and hippocampus during pregnancy or after delivery, and Weiland and Orchnik (52) found that progesterone suppressed hippocampal α1-subunit expression in estrogen-primed animals. Therefore, the precise nature of any changes in GABAA receptors that occur under hormonal conditions similar to the present catamenial epilepsy model remains to be characterized.

Neurosteroids are not the only GABAA receptor–modulating agents whose withdrawal leads to brain hyperexcitability and to a precipitation of seizures. For example, it is well known that abrupt discontinuation of benzodiazepines (BZDs) (52) and ethanol (53) can predispose to seizures. However, there may be differences in the mechanisms accounting for brain hyperexcitability after withdrawal from prolonged exposure to endogenous neurosteroids and the hyperexcitability that occurs after withdrawal of BZDs. Thus tolerance does not develop to neuroactive steroids, unlike BZDs, during prolonged treatment (54,55). Moreover, we have found that the anticonvulsant activity of neurosteroids is enhanced after neurosteroid withdrawal, whereas BZDs may have reduced activity (56,57).

In conclusion, the present study shows that withdrawal from elevated levels of neurosteroids during pseudopregnancy results in increased susceptibility to PTZ-induced seizures. In view of the considerable evidence linking perimenstrual catamenial seizure exacerbations to the sharp decline in allopregnanolone levels at the time of menstruation, the pseudopregnancy–finasteride withdrawal model may be useful for the evaluation of therapeutic approaches to the treatment of catamenial epilepsy (56,57).