Predictors of pharmacoresistant epilepsy: Pharmacoresistant rats differ from pharmacoresponsive rats in behavioral and cognitive abnormalities associated with experimentally induced epilepsy

Authors

  • Alexandra M. Gastens,

    1. Department of Pharmacology, Toxicology, and Pharmacy, University of Veterinary Medicine Hannover, and Center for Systems Neuroscience, Hannover, Germany
    Search for more papers by this author
  • Claudia Brandt,

    1. Department of Pharmacology, Toxicology, and Pharmacy, University of Veterinary Medicine Hannover, and Center for Systems Neuroscience, Hannover, Germany
    Search for more papers by this author
  • Jens P. Bankstahl,

    1. Department of Pharmacology, Toxicology, and Pharmacy, University of Veterinary Medicine Hannover, and Center for Systems Neuroscience, Hannover, Germany
    Search for more papers by this author
  • Wolfgang Löscher

    1. Department of Pharmacology, Toxicology, and Pharmacy, University of Veterinary Medicine Hannover, and Center for Systems Neuroscience, Hannover, Germany
    Search for more papers by this author

Address correspondence to Dr. Wolfgang Löscher, Department of Pharmacology, Toxicology and Pharmacy, University of Veterinary Medicine, Bünteweg 17, D-30559 Hannover, Germany. E-mail: wolfgang.loescher@tiho-hannover.de

Summary

Purpose: Patients with intractable temporal lobe epilepsy (TLE) exhibit an increased risk of psychiatric comorbidity, including depression, anxiety, psychosis, and learning disorders. Furthermore, a history of psychiatric comorbidity has been suggested as a predictor of lack of response to therapy with antiepileptic drugs (AEDs) in patients with epilepsy. However, clinical studies on predictors of pharmacoresistant epilepsy are affected by several confounding variables, which may complicate conclusions. In the present study, we evaluated whether behavioral alterations in epileptic rats are different in AED nonresponders versus responders.

Methods: For this purpose, we used an animal model of TLE in which AED responders and nonresponders can be selected by prolonged treatment of epileptic rats with phenobarbital (PB). Behavioral and cognitive abnormalities were compared between responders and nonresponders as well as between epileptic rats and nonepileptic controls in a battery of tests.

Results: Fifteen epileptic rats with spontaneous recurrent seizures (SRS) either responding (11 rats) or not responding (4 rats) to PB were used for this study. The nonresponders differed markedly in behavioral and cognitive abnormalities from responders and nonepileptic controls in tests of anxiety (open field, elevated-plus maze test), behavioral hyperexcitability (approach-response, touch-response, pick-up tests), and learning and memory (Morris water maze).

Discussion: Our hypothesis that AED-resistant rats will show more severe behavioral and cognitive changes than AED-responsive rats was confirmed by the present experiments. The data substantiate that rodent models of TLE are useful to delineate predictors of pharmacoresistant epilepsy.

Despite more than 20 approved antiepileptic drugs (AEDs) and several nonpharmacological options, about 30–40% of patients with epilepsy are refractory to treatment ( Kwan & Brodie, 2000; Schmidt & Löscher, 2005). Poor prognostic factors include high frequency of seizures before onset of treatment, lack of response to the first AED, specific syndromes, a history of febrile seizures, brain lesions such as hippocampal sclerosis, and prior or current psychiatric comorbidity ( Regesta & Tanganelli, 1999; Kwan & Brodie, 2000; Kwan & Brodie, 2004; Hitiris et al., 2007). These observations may suggest that medically intractable epilepsy is a distinct condition that may be identified early and, thus, be targeted early for more aggressive therapeutic intervention, such as epilepsy surgery (Arroyo et al., 2002; Kwan & Brodie, 2002). Research in animal models of epilepsy also favors the notion that refractory epilepsy could be a distinct condition (Löscher, 2006). Thus, in both the kindling model and post-status epilepticus (post-SE) models of temporal lobe epilepsy (TLE), AED-responsive and AED-resistant rats are observed despite the similarity in all other variables, including rat strain, gender, drug elimination, type and severity of seizures, and onset of AED treatment (Löscher, 2006).

Psychiatric comorbidity is a major concern in the treatment of patients with intractable epilepsy (Hermann & Jones, 2006; Kanner, 2006; Hitiris et al., 2007). Furthermore, many patients with epilepsy suffer from impaired cognitive performance (Motamedi & Meador, 2003). In the search for predictors or surrogate markers of pharmacoresistant epilepsy, we evaluated whether AED-resistant rats differ behaviorally from AED-responsive rats or nonepileptic controls in a post-SE model of TLE. In this model, AED responders and nonresponders can be selected by prolonged treatment with phenobarbital (PB; Brandt et al., 2004). The majority (>80%) of PB-resistant rats are also resistant to phenytoin (Bethmann et al., 2007), thus meeting the operational definition of pharmacoresistance in animal models, that is, persistent seizure activity not responding to at least two AEDs at maximum tolerated doses (Stables et al., 2003). It is known that behavioral alterations occur in post-SE rat models of TLE (Heinrichs & Seyfried, 2006; Stafstrom, 2006), but our hypothesis was that the AED-resistant rats would show more severe behavioral and cognitive changes than AED-responsive rats.

Materials and Methods

Animals

As in our previous experiments in rats with spontaneous recurrent seizures (SRS) developing after SE induced by prolonged electrical stimulation of the basolateral amygdala (BLA) (Brandt et al., 2003a, 2004; Volk & Löscher, 2005; Brandt et al., 2006a; Volk et al., 2006; Bethmann et al., 2007), adult female Sprague–Dawley rats (Harlan-Winkelmann, Borchen, Germany) were used for this study. The rats were purchased at a body weight of 200–230 g. Following arrival, the rats were kept under controlled environmental conditions (24–25°C; 50–60% humidity; 12 h light/dark cycle; light on at 5:00 a.m.) with free access to standard laboratory chow (Altromin 1324 standard diet; Altromin, Lage, Germany) and tap water. Before being used in the experiments, the rats were allowed to adapt to the new conditions for at least 1 week. All animal experiments were carried out in accordance with the European Communities Council Directive of November 24, 1986 (86/609/EEC) and were formally approved by the animal subjects review board of our institution. All efforts were made to minimize the number of animals used and their suffering.

Electrode implantation and SE induction

For obtaining the rats for the behavioral experiments described in the present study, a total of 87 age-matched animals were used. The experiments were performed in two separate tranches (NIH-1 and NIH-2) with 45 rats each and subsequently analyzed together. In each tranche, 35 rats were implanted with BLA electrodes, while 9 (NIH-1) and 8 (NIH-2) rats were used as naive controls. Electrodes were stereotactically implanted into the right anterior BLA under anesthesia as described in detail recently (Brandt et al., 2003a) and served for electrical stimulation and recording of the electroencephalogram (EEG). The stereotaxic coordinates for the BLA electrode according to the atlas of Paxinos & Watson (1998) were 2.2 mm caudal, 4.7 mm lateral, and 8.7 mm ventral (all respective to bregma). These coordinates were based on extensive previous experiments in female Spague–Dawley rats, demonstrating correct placement of the stimulation and recording electrode in the BLA or at the border between BLA and ventral endopiriform nucleus (Brandt et al., 2003a). About 4 weeks after electrode implantation, 25 rats of each experiment (NIH-1, NIH-2) were electrically stimulated via the BLA electrode for induction of a self-sustained SE as described previously (Brandt et al., 2003a, 2004; Volk & Löscher, 2005). The following stimulus parameters were chosen: stimulus duration 25 min; stimulus consisting of 100 msec trains of 1 msec alternating positive and negative square wave pulses. The trains were given at a frequency of 2/s and the intratrain pulse frequency was 50/s. Peak pulse intensity was 700 μA. For this pulsed-train stimulation, an Accupulser A310C stimulator connected with a Stimulus Isolator A365 (World Precision Instruments, Berlin, Germany) were used. In all rats, the EEG was recorded via the BLA electrode during SE and up to 20 h after termination of SE by diazepam (see below). Only rats that developed a self-sustained SE with generalized convulsive seizures were used for further experiments to ensure that the type or severity of SE did not differ among rats. SE was interrupted after 4 h by diazepam (10 mg/kg i.p.). If necessary, the application of this dose of diazepam was repeated, which was only needed in five rats. Diazepam completely stopped motor and EEG seizure activity in all rats, so that SE duration was the same in subsequent PB responders and nonresponders. Starting 6 weeks later, all rats that had experienced a convulsive SE were monitored by EEG-video recordings for about 2 weeks until the first spontaneous seizures were detected as described recently (Brandt et al., 2003a, 2004). Only rats that exhibited SRS during this period were used for selection of responders and nonresponders by prolonged treatment with PB (see below). Twenty age-matched rats that were implanted with BLA electrodes but not stimulated were used as sham controls; one of these rats lost its electrode assembly after implantation, so that final sham group size was 19. Furthermore, 17 age-matched animals without BLA electrodes served as naïve controls. Both control groups were concurrently handled together with the epileptic rats throughout all experimental procedures, so that they had the same handling experience as the epileptic rats. Furthermore, in each tranche (NIH-1, NIH-2) the rats of the two control groups were tested concurrently with the epileptic rats in the behavioral battery described below.

Determination of seizure threshold and the acute anticonvulsant effect of phenobarbital

Starting 8 weeks after SE, the thresholds for induction of spikes (afterdischarges) in the EEG (afterdischarge threshold [ADT]) as well as the threshold for secondarily generalized seizures (generalized seizure threshold [GST]) were determined in part of the rats by electrical stimulation via the BLA electrode by a staircase procedure as described previously (Ebert et al., 2000). ADT was defined as the current producing afterdischarges with a duration of at least 3 s. Seizure severity occurring at ADT current was rated according to Racine (1972). GST was defined as the current producing stage 4 or 5 seizures. Some rats exhibited secondarily generalized (stages 4–5) seizures in addition to focal (stages 1–3) seizures at the ADT current, so that it was not necessary to determine the GST separately in such animals. The aim of this experiment was to examine whether epileptic rats, in which SRS could subsequently not be suppressed by PB, did also not respond to the acute anticonvulsant effect of PB on induced seizures. Seven epileptic rats that were subsequently selected into responders or nonresponders by prolonged treatment with PB were used for this experiment. First, ADT and GST were repeatedly determined once per day in the morning at intervals of 2–5 days to assure that seizure thresholds were stable. After three ADTs/GSTs had been determined, a cross-over design was used in which each rat received either saline or 30 mg/kg PB i.p. in the morning and ADT/GST were determined 1 h after injection, followed 2 days later by injections of either saline or PB, so that each rat served as its own control.

Selection of responders and nonresponders by prolonged treatment with phenobarbital

For suppression of SRS, PB was chosen because it is an efficacious AED in rat models of TLE with a sufficiently long half-life (17 h in female Sprague–Dawley rats) to allow maintenance of “therapeutic” drug levels during prolonged treatment (Löscher, 2007). As described in detail recently (Brandt et al., 2004), several preliminary experiments were performed to develop a dosing protocol with maximum tolerable doses, resulting in maintenance of plasma drug concentrations within or above the therapeutic range (10–40 μg/ml; Baulac, 2002) over 24 h/day, 7 days/week. Based on these preliminary experiments, a dosing protocol with an i.p. bolus dose of 25 mg/kg in the morning of the first treatment day, followed 10 h later by an administration of 15 mg/kg i.p., and then twice daily (between 7:00 and 8:00 a.m. and 5:00 and 6:00 p.m.) 15 mg/kg i.p. for the 13 subsequent days, was used in rats with SRS. For drug administration, the sodium salt of PB was dissolved in 0.9% saline and administered at a volume of 3 ml/kg. Before onset of drug treatment, baseline seizure frequency was determined over 2 weeks (predrug control period, which started about 12 weeks after the SE), then PB was administered over two weeks, followed by a postdrug control period of 2 weeks (Fig. 1). In this way, each animal served as its own control, accounting for differences between animals, for example, variability in baseline seizure frequency. During the control periods, 0.9% saline (3 ml/kg) was injected instead of PB. Aggressive rats were not anesthetized (because this could affect results) but one investigator fixed such rats while another investigator performed the i.p. injection. During drug treatment, rats were closely observed for adverse effects (ataxia, sedation, muscle relaxation). Blood was sampled by retroorbital puncture (after local anesthesia with tetracaine) 10 h after the first drug injection, 10 h after drug injection on day 7, and 10 h after the last drug injection for PB analysis in plasma by high-performance liquid chromatography (HPLC) with ultraviolet detection (Potschka et al., 2002). In all rats, seizures were continuously (24 h/day, 7 days/week) monitored by video/EEG recording over the 6 weeks of the experiment as described below. In NIH-1, 15 epileptic rats were used, while 16 epileptic rats were used in NIH-2.

Figure 1.

 Schematic illustration of the experimental protocol used to select drug-resistant and drug-responsive rats with spontaneous recurrent seizures for determination of behavioral alterations.

Monitoring and analysis of spontaneous recurrent seizures

For continuous EEG-monitoring, an 8-channel amplifier (CyberAmp 380, Axon Instruments, Inc. Foster City, CA, U.S.A.), eight 1-channel bioamplifiers (ADInstruments Ltd., Hastings, East Sussex, U.K.), and two analogue-digital converters (PowerLab/800s, ADInstruments Ltd.) were used in NIH-1 and 16 1-channel bioamplifiers in NIH-2. This system allowed simultaneous recording of EEGs from up to 16 rats, so that the 15 and 16 rats of NIH-1 and NIH-2 could be monitored together within each experimental period. The data were recorded and analyzed with the Chart4 for windows software (ADInstruments Ltd.). The sampling rate for the EEG-recording was 200 Hz. A high pass filter for 0.1 Hz and a low pass filter for 60 Hz were used.

Simultaneously to the EEG-recording, all rats used in this study were video-monitored continuously during the experimental periods, using four light-sensitive black-white cameras (CCD-Kamera-Modul, Conrad Electronic, Hannover, Germany) which allowed video recording of up to four rats per camera. The cameras were connected to a multiplexer (TVMP-400, Monacor, Bremen, Germany) which converted the signals from the four cameras to a video recorder (Time Lapse recorder, Sanyo TLS-9024P, Monacor). To allow video recording of seizures during the night, infrared light was used. Rats were housed in clear glass cages (one per cage) to allow optimal video observation.

For detection of spontaneous seizures, the EEG recordings were visually analyzed for characteristic ictal events. The typical ictal discharges occurring during spontaneous seizures in this model have been described in detail previously (Brandt et al., 2003a). To evaluate the severity of motor seizure activity during a paroxysmal alteration in the EEG, the corresponding video recording was viewed. In addition to seizures observed by video and EEG recordings, seizures observed during handling or other manipulations of the animals were noted. As described previously (Brandt et al., 2003a, 2004), most spontaneous seizures were generalized convulsive seizures, resembling stage 4 or 5 seizures of the Racine scale (Racine, 1972). Based on individual responses of rats to treatment, they were either considered responders or nonresponders. Responders were defined by complete seizure suppression during treatment or a seizure suppression of at least 75% compared to seizure frequency in the control periods.

Behavioral evaluation

After completion of the drug experiment, the rats were evaluated in different behavioral tests (Fig. 1). The experiments were performed in two separate tranches (NIH-1 and NIH-2) and subsequently analyzed together. In “NIH-1,” 9 naive controls, 11 sham controls, and 10 epileptic rats were compared, while “NIH-2” comprised 8 naive controls, 8 sham controls, and 5 epileptic rats (see Results). In order to minimize the possibility that alterations in behavior or cognition in epileptic rats were due to the PB treatment, we used a long wash-out period (6–7 months) between termination of PB treatment and start of the behavioral measurements (Fig. 1). All tests were performed within a time period of 5 weeks. Intertrial intervals ranged between 1 and 5 days. Before each test, the rats were transferred to a room next to the room in which the experiments were conducted. The two rooms were connected through a door, which was closed during the behavioral trials, so that the respective rat that was subjected to a behavioral test was not affected by the presence of an experimenter or the other rats. The experiments were performed between 9 a.m. and 1 p.m. For the elevated-plus maze, open field, and water maze, a video tracking system with the EthoVision software from Noldus (Wageningen, The Netherlands) was used. Behavioral testing was performed only if no motor seizure was observed for at least 1 h before the test. During behavioral testing, no spontaneous seizures occurred except in the Morris water maze (see below). During the behavioral tests and analysis of data from these tests, the investigators performing the tests were not aware whether epileptic rats were PB responders or nonresponders.

Open field

The open field is a very popular animal model of anxiety-like behavior (Prut & Belzung, 2003). The procedure consists of subjecting an animal to an unknown environment from which escape is prevented by surrounding walls. In such a situation, rodents spontaneously prefer the periphery of the apparatus to activity in the central parts of the open field. In the present study, the test was performed in a round open field with gray-painted wall and floor (ø 83 cm). The animals were placed individually in the center of the open field. Behavior was observed for 10 min. Before each trial, the field was cleaned thoroughly with 0.1% acetic acid solution. The open field was divided into three zones: center, internal ring, and outer ring. The center and internal ring were considered aversive places for the rats. For each rat, the total distance moved and the time spent in each zone were measured by a computerized tracking system (EthoVision, Noldus). Rearing and grooming were recorded manually by the experimenter.

Elevated-plus maze

The elevated-plus maze is a validated model to assess the level of anxiety in rodents (File, 1993). The apparatus was constructed with black plastic. It comprises two open arms (50 × 10 cm), two enclosed arms (50 × 10 × 30 cm), and a central platform (10 × 10 cm). The configuration has the shape of a plus sign, and the apparatus is elevated 80 cm above the floor level. Grip on the open arms is facilitated by inclusion of a small edge (0.5-cm high) around their perimeter, made of transparent Plexiglas. Before each trial, the maze was cleaned thoroughly with 0.1% acetic acid solution. At the beginning of the test, rats were placed on the central platform always facing the same closed arm. The test lasted 5 min and the behavior of rats in the test was analyzed using the EthoVision Software. Activity and anxiety-related behaviors were assessed. Standard measures comprised: the total distance moved (cm), the time spent in different sections of the maze (open and closed arms), the frequency of entries into open and closed arms, and the frequency of head dips and rearing.

Subgroups of the rats (9 naive controls, 11 sham rats, and 10 epileptic rats) were tested in the elevated-plus maze 4–5 days before induction of SE to examine whether rats that subsequently turned out to be PB-nonresponders differed in premorbid behavior from subsequent PB-responders or controls.

Test for behavioral hyperexcitability

Rice & DeLorenzo (1998) described four behavioral tests that potentially discriminate differences in behavioral excitability and sensory responsiveness between epileptic and control rats. These tests were taken from the functional observational battery described by Moser et al. (1988). The four tests, which are quick and easy to perform without prior training of the rats or special equipment, are: (1) Approach-response test: A pen held vertically is moved slowly toward the face of the animal. Responses were scored as 1, no reaction; 2, the rat sniffs at the object; 3, the rat moves away from the object; 4, the rat freezes; 5, the rat jumps away from the object; and 6, the rat jumps at or attacks the object. (2) Touch-response test: The animal is gently prodded in the rump with the blunt end of a pen. Responses were scored as 1, no reaction; 2, the rat turns toward the object; 3, the rat moves away from the object; 4, the rat freezes; 5, the rat jerks around toward the touch; 6, the rat turns away from the touch; and 7, the rat jumps with or without vocalizations. (3) Finger-snap test: A click noise with a clicker several centimeters above the head of the animal is performed. Responses were scored as 1, no reaction; 2, the rat jumps slightly or flinches or flicks the ear (normal reaction); and 3, the rat jumps dramatically. (4) Pick-up test: The animal is picked up by grasping around the body. Responses were scored as 1, very easy; 2, easy with vocalizations; 3, some difficulty, the rat rears and faces the hand; 4, the rat freezes (with or without vocalization); 5, difficult, the rat avoids the hand by moving away; and 6, very difficult, the rat behaves defensively, and may attack the hand.

There were four to five independent observers for each rat, and the means of their scores were calculated for each animal for each test. Between each observer there was a period of at least 60 min. The tests were accomplished in the home cage.

Forced swim test

A depressed state can be induced in rodents by forcing them to swim in a narrow cylinder from which they cannot escape (Porsolt et al., 1978). After a brief period of vigorous activity, the animals adopt a characteristic immobile posture (behavioral despair), which is readily identifiable. Rats remain immobile longer in the second of two swim tests, a phenomenon known as “learned despair.” In the present study, the animals were tested in a transparent plexiglas cylinder (ø 39 cm, height 60 cm) filled to a depth of 15 cm with water (at 25°C), so that the animals could touch the bottom. The glass cylinder was surrounded by dark walls. On the first experimental day, rats were gently placed in the water for a 15-min period of habituation. On removal from the water, they were placed in a standard Plexiglas box with the floor covered with paper towels under red light to dry. The next day, the rats were placed in the glass cylinder as on the previous day, but this time for only 5 min. At the end of the 5-min period, the rats were transferred to the red light warmed box and allowed to dry. On both days, the behavior of the animals in the glass cylinder was videotaped. During analysis of the videotapes, the duration of the following behaviors was recorded: swimming (all activities, including struggling, of the rats except from jumping and diving) and immobility. For comparison of these behaviors between day 1 and day 2 of the test, the first 5 min of the habituation trial on day 1 were compared with the 5-min trial of day 2. Furthermore, the duration of immobility on day 2 (learned despair) was compared between groups.

Morris water maze

The Morris water maze (Morris, 1984), in which rats learn to escape from water onto a hidden platform, is a widely used test of visuospatial memory and hippocampal integrity. In the present study, the water maze apparatus was made of a black circular tank (diameter 150 cm, 60 cm deep) filled with water to a height of 27 cm. The water temperature was 19 ± 1°C. The maze was surrounded by several visual cues. After a single habituation trial (day 0, 60 s, no platform), a submerged black escape platform (1.5 cm below the water level, 10 × 10 cm) was placed in the middle of the northeast quadrant on five consecutive days (acquisition). This quadrant was not preferred or avoided by the rats during the habituation trial. Each day, the animals were placed into the pool at four different starting points. On day 1, beginning from the south turning clockwise until four trials were accomplished, on day 2 beginning from the west, day 3 from the north, day 4 from the east, and day 5 from the south, respectively. Animals, which did not find the escape platform within 60 s, were placed there by the experimenter. All rats remained on the platform for at least 2 s. Between the four trials per day, the animals were allowed to rest for 60 s in a transport cage. After the four trials, they were placed in a standard Plexiglas box with the floor covered with paper towels under a red light to dry. For each trial, the escape latency (seconds) was measured by a computerized tracking system (EthoVision). For each rat, the data of the four trials per day were averaged except for day 5 of the acquisition period, for which only the first three trials were averaged. The fourth trial of day 5 was used for the spatial probe, for which the platform was removed and the crossings of the former platform position during a single trial over 60 s were recorded. Four of the 15 epileptic rats tested in the Morris water maze showed spontaneous seizures during a swimming trial, mostly on day 2 of the test. In these rats, the trial was interrupted and the rats were put back in their home cage for at least 1 h. Then the entire four trials of the respective day for these rats were repeated and only these repeated trials were used for data analysis.

Neuropathology

In 11 of the 15 epileptic rats used for final analysis in the present study (see Results), the hippocampus was histologically examined for damage, using thionin-stained sections as described in detail recently (Volk et al., 2006; Bethmann et al., 2008). In short, in CA1, CA3, and dentate hilus, loss of neurons was semiquantitatively assessed by a grading system: score 0, no obvious damage; score 1, lesions involving 20–50% of neurons; score 2, lesions involving >50% of neurons. Nonepileptic control rats were used for comparison. Three sections in the ipsilateral and contralateral hippocampal formation were analyzed for damage in each rat. These sections were located at 1–1.5 mm intervals between 2.1 and 5.8 mm posterior from bregma according to the atlas of Paxinos & Watson (1998). All evaluations were performed blind with respect to the treatment status of the rat. In four of the 15 epileptic rats (NIH 65, 67, 69, and 91), histological brain examination was not possible because of a technical problem in tissue preparation and storage. In addition to using thionin-stained sections for assessment of neuronal damage, the localization of the stimulation and recording electrode was verified. Electrode position corresponded to that described by us previously for this model (Brandt et al., 2003a), without any obvious difference between PB responders and nonresponders.

Statistics

Depending on whether data were normally distributed or not, either parametric or nonparametric tests were used for statistical evaluation. Seizure frequencies were not normally distributed, so that nonparametric statistics (Friedman test, followed by the Wilcoxon signed rank test for paired replicates) were used for within-group comparisons. Intergroup differences in seizure frequency were calculated by the Mann--Whitney U-test. For group comparisons of PB levels and behavioral alterations, analysis of variance (ANOVA) followed by the Bonferroni multiple comparison test was used. In case of not normally distributed data, ANOVA for nonparametric data (Kruskal–Wallis test), followed by post hoc testing for individual differences by Dunn's test were used. Data from Morris water maze were analyzed by two-way ANOVA with Bonferroni posttests to localize significant intergroup differences. Within-group comparisons in the Morris water maze and forced swim test were performed by the paired t-test. Frequencies were analyzed by Fisher's exact test. All tests were used two-sided; a p < 0.05 was considered significant.

Results

Selection of PB responders and nonresponders

Fifteen of the 31 epileptic rats that were used for selection into responders or nonresponders with PB in the present study could be included in the final analysis. The other rats had to be excluded because they did not show SRS during either pre- or postdrug control periods, could not be unequivocally attributed to the responder or nonresponder subgroups, or died unexpectedly before the behavioral analyses could be performed. Thus, all data described in the following are from 15 epileptic rats (plus controls) in which all experiments could be successfully performed.

With the dosing protocol used for selection of responders and nonresponders, PB-induced marked sedation and ataxia, indicating that maximum tolerated doses were used. Analysis of plasma drug concentrations showed that drug concentrations within or above the therapeutic range (10–40 μg/ml) known from patients with epilepsy were maintained throughout the period of treatment except for one rat (NIH 95), in which therefore the dose of PB was increased to 20 mg/kg twice daily, resulting in an increase of PB levels in plasma.

In the 15 epileptic rats, median seizure frequency in the 2 weeks before onset of treatment with PB was 1.8 seizures per day (Fig. 2A) with a large variation of individual data (range = 0.07–14.1 seizures/day). In all rats, seizures were evenly distributed over the 2 weeks of continuous predrug recording, without any indication of seizure clusters. Compared to predrug seizure frequencies, seizure frequency was not significantly changed by treatment with PB (Fig. 2A). In the postdrug control period, seizure frequency increased in several rats compared to that determined before onset of treatment (Fig. 2A), but the difference was not statistically significant. In contrast to the lack of significant difference between seizure frequencies recorded during the predrug and treatment periods, seizure frequencies recorded during treatment with PB differed significantly from postdrug control data (p = 0.0353).

Figure 2.

 Effect of phenobarbital (PB) on spontaneous recurrent seizures (SRS) in rats. SRS were recorded over a period of 2  weeks before onset of PB treatment (predrug control), followed by drug treatment for 2  weeks, and then a 2-week postdrug control period. All data are shown as means ± SEM. (A–C) Show the average number of seizures per day recorded over the three 2-week periods, while (D) illustrates the average plasma concentration of PB from the blood samples taken during treatment. In (A), average seizure data from the 15 epileptic rats used in the present study are given, while (B) shows respective data from the 11 responders and (C) data from the four nonresponders of this group (see text for definitions). Analysis of seizure frequency data by nonparametric repeated measures ANOVA (Friedman test) indicated significant differences for the responder group (p  = 0.0003), but not the nonresponder group (p > 0.9306). Post hoc analysis of the responder group indicated that PB significantly suppressed SRS compared to the pre- and postdrug periods (p < 0.01; indicated by asterisk), while such effect was not determined in nonresponders. The shaded area in (D) indicates the therapeutic plasma concentration range of PB. PB plasma levels of responders and nonresponders did not differ significantly.

However, by averaging data from all 15 rats as shown in Fig. 2A, striking differences in individual response to PB were masked. In eight of these rats, complete control of seizures was achieved and three other rats exhibited a >75% reduction of seizure frequency. These 11 rats were considered responders. Data from these rats are shown in Fig. 2B, demonstrating a significant anticonvulsant effect of PB in this subgroup when compared to either pre- or postdrug seizure frequency. In four other rats, PB did not exert an anticonvulsant effect, so that these rats were considered nonresponders. Data from these rats are shown in Fig. 2C. Analysis of seizure frequency in nonresponders by ANOVA did not indicate any significant difference between the predrug, drug, and postdrug data (p = 0.9306). Two of the nonresponders (NIH 18 and NIH 95) exhibited an extremely high seizure frequency (>9 seizures/day) during the control periods. The other two nonresponders did not differ from responders in seizure frequency during the control periods, so that the pre- and postdrug control values of all four nonresponders did not differ significantly from those of the 11 responders, arguing against any general difference in control seizure frequencies in the nonresponders versus responders. In the two rats with high seizure frequency, spontaneous seizures were observed throughout the period of treatment with PB, while in the two other nonresponders spontaneous seizures occurred in the second week of treatment. As in our previous studies (Brandt et al., 2004, 2006a; Bethmann et al., 2007), the severity or duration of the initial brain insult (the SE) did not differ between responders and nonresponders.

Plasma levels of PB did not differ significantly between responders and nonresponders at any of the three determinations (i.e., on days 1, 7, and 14) of the treatment period, so that plasma concentrations averaged from the three determinations in all 15 rats and in the responder and nonresponder subgroups are shown in Fig. 2D. No significant difference was found between these values. Furthermore, the severity of adverse effects of PB did not significantly differ between the responder and nonresponder subgroups (not illustrated).

Behavioral alterations in epileptic rats and differences between PB responders and nonresponders

In general, naive controls did not differ significantly from electrode-implanted (sham) rats in any test, so that alterations in epileptic rats were considered different from control when they differed significantly from either naive controls or sham controls or both. In the open field, locomotor activity, measured by the distance moved in the field, was similar in all rats, except that PB responders showed moderately more activity than controls and nonresponders (Fig. 3A). Both groups of epileptic rats, that is, PB responders and PB nonresponders, stayed significantly longer in the aversive internal ring of the open field than controls (Fig. 3C), which was associated with significantly reduced duration of stay in the less-aversive outer ring of the field (Fig. 3B). PB nonresponders showed this behavior significantly more intensive than responders. Furthermore, PB nonresponders tended to stay longer in the aversive center of the field (Fig. 3D), which, however, was not statistically significant. The normal locomotor activity seen in PB nonresponders (Fig. 3A) suggested that the markedly longer duration that these animals spent in the aversive parts of the field was not just secondary to increased ambulatory activity. No significant intergroup differences were seen in the frequency of rearing or grooming (not illustrated).

Figure 3.

 Behavior of epileptic rats and nonepileptic controls in the open field. Data are means ± SEM of 17 naive controls, 19 sham controls (implanted with BLA electrodes but not stimulated), and 15 epileptic rats, subdivided into 11 phenobarbital responders and 4 nonresponders, respectively. (A) Illustrates the total distance that the rats moved during the 10 min of the test. (B) Illustrates the time that rats spent in the external (outer) ring of the open field. (C, D) Illustrate the time that rats spent in the aversive internal ring and center of the open field, respectively. Analysis of data by ANOVA indicated significant differences between groups for the data shown in (A) (p  = 0.0128; F  = 4.005), (B) (p < 0.0001; F  = 27.49), (C) (p < 0.0001; F  = 32.8), but not (D) (p  = 0.1976; F  = 1.619). Significant differences of responders or nonresponders to controls (either naive or sham or both) are indicated by asterisk (p < 0.05), and significant differences between nonresponders and responders by circle (p < 0.05). Naive controls did not differ significantly from sham controls for any of the parameters shown.

In the elevated-plus maze test, no significant intergroup differences in locomotor activity were observed (Fig. 4A). Both groups of epileptic rats stayed significantly longer on the aversive open arms than controls (Fig. 4C) and also the frequency of entries into the open arms was significantly enhanced (Fig. 4D). Similar to the open field, this behavior was much more marked in the PB nonresponders compared to responders. Furthermore, both groups of epileptic rats exhibited significantly more exploratory head dips than controls, while the frequency of rearing was significantly decreased in PB responders, but not nonresponders (not illustrated).

Figure 4.

 Behavior of epileptic rats and nonepileptic controls in the elevated-plus maze. (A) Illustrates the total distance that the rats moved during the 5 min of the test. (B) Illustrates the time that rats spent in the closed arms of the maze. (C) Illustrates the time that rats spent in the aversive open arms of the maze and (D) the frequency of entries into the open arms. Analysis of data by ANOVA indicated significant differences between groups for the data shown in (A) (p  = 0.0367; F  = 3.078), (B) (p  = 0.0002; F  = 8.080), (C) (p < 0.0001; F  = 11.05), and (D) (p < 0.0001; F  = 14.85). Significant differences of responders or nonresponders to controls (either naive or sham or both) are indicated by asterisk (p < 0.05), significant differences between nonresponders and responders by circle (p < 0.05). Naive controls did not differ significantly from sham controls for any of the parameters shown. For further details see  Fig.  3 legend.

In subgroups of naive controls, sham controls, and epileptic rats, the elevated-plus maze test was performed twice, that is, before SE induction and after selection of PB responders and nonresponders. Thus, it was possible to evaluate whether subsequent PB nonresponders differed behaviorally from responders before induction of epilepsy. As shown in Fig. 5, no differences between groups were observed before SE induction (Fig. 5A–C), whereas after development of epilepsy PB nonresponders stayed significantly longer on the aversive open arms than PB responders (Fig. 5F).

Figure 5.

 Comparison of behavior in the elevated-plus maze before and after status epilepticus (SE). Data are means ± SEM of 9 naive controls, 11 sham rats, and 10 epileptic rats that were tested in the elevated-plus maze 4–5 days before induction of SE to examine whether rats that subsequently turned out to be PB-nonresponders differed in behavior from subsequent PB-responders or controls. No significant differences between groups were seen before the SE (A–C) while significant differences between nonresponders versus responders and controls were determined after SE (D–F). Thus, analysis of data by ANOVA indicated significant differences between groups for the data shown in (F) (p  = 0.0004; F  = 8.646). Significant differences of responders or nonresponders to controls (either naive or sham or both) are indicated by asterisk (p < 0.05), significant differences between nonresponders and responders by circle (p < 0.05). Naive controls did not differ significantly from sham controls for any of the parameters shown. For further details see  Fig.  3 legend.

For testing behavioral hyperexcitability of epileptic rats, the observational battery of behavioral tests described by Rice & DeLorenzo (1998) was used as described in the Methods section. In the approach-response test, only PB nonresponders differed significantly from control (Fig. 6A). Furthermore, PB nonresponders reacted significantly more intensely in this test than responders. Similarly, only PB nonresponders differed from controls in the touch-response test (Fig. 6B). In the finger-snap test, no differences between nonepileptic controls and the two epileptic groups were observed (Fig. 6C). In contrast, a significant difference to controls was observed for both epileptic groups in the pick-up test (Fig. 6D).

Figure 6.

 Behavior of epileptic rats and nonepileptic controls in an observational battery to test for behavioral hyperexcitability. Data are shown for the following tests: (A) approach-response test; (B) touch-response test; (C) finger-snap test; and (D) pick-up test. Analysis of data by ANOVA indicated significant differences between groups for the data shown in (A) (p = 0.0099; F = 4.227), (B) (p = 0.0144; F = 3.901), (C) (p = 0.0326; F = 3.177), and (D) (p = 0.0007; F = 6.743). Significant differences of responders or nonresponders to controls (either naive or sham or both) are indicated by asterisk (p < 0.05), significant differences between nonresponders and responders by circle (p < 0.05). Naive controls did not differ significantly from sham controls for any of the parameters shown. For further details see  Fig.  3 legend.

In the forced swim test, no significant intergroup differences in swimming were observed during the first 5 min of the habituation trial on day 1 (Fig. 7A). On day 2, the two control groups showed significantly less swimming activity in this test than on day 1, indicating development of learned despair (Fig. 7A). This, however, was not observed in the two epileptic groups. When the duration of immobility on day 2 was compared between groups, significantly less immobility was observed in the two epileptic groups compared to controls (Fig. 7B). In nonresponders, almost no immobility was observed, but the difference to responders did not reach statistical significance.

Figure 7.

 Behavior of epileptic rats and nonepileptic controls in the forced swim test. In (A), the first 5 min of the habituation trial on day 1 were compared with the 5 min trial of day 2. Significant differences between day 1 and day 2 in each group are indicated by asterisk (p < 0.05). In (B), the duration of immobility recorded on day 2 of the test is illustrated. Analysis of data by ANOVA indicated significant differences between groups for the data shown in (B) (p  = 0.0040; F  = 5.094). Post hoc testing by the Bonferroni test indicated that responders and nonresponders differed significantly from sham controls (p < 0.05; indicated by asterisk). Naive controls did not differ significantly from sham controls. For further details see  Fig.  3 legend.

In the Morris water maze test, nonepileptic controls rapidly improved to locate the hidden platform (escape latency) (Fig. 8). Average escape latencies on day 1 versus day 5 were 38.5 ± 2.7 s versus 16.2 ± 2.4 s for naive controls (p < 0.0001) and 44.8 ± 2.9 s versus 21.9 ± 1.9 s for sham controls (p < 0.0001), respectively. PB responders also showed some learning upon the repeated trials, but less rapidly than controls (Fig. 8). Average escape latencies on day 1 versus day 5 were 50.45 ± 4.5 s versus 38.8 ± 4.6 s (p = 0.0097). However, no learning was observed in PB nonresponders, in which average escape latencies were 57.1 ± 2.9 s on day 1 versus 51.6 ± 5.3 s on day 5 (p = 0.3888). When the four experimental groups were compared at each day of the acquisition period, ANOVA indicated significant intergroup differences over the whole period of the experiment (Fig. 8). Nonresponders significantly differed from controls already at the onset of the experiment, that is, on day 1, as well as on all subsequent days of the acquisition period. Responders differed from controls on days 3–5 of the experiment. The swimming velocities of the four groups of rats did not differ significantly (not illustrated).

Figure 8.

 Acquisition learning of platform location by epileptic rats and nonepileptic controls in the Morris water maze. The figure shows the mean time to reach the hidden platform (escape latency) during the 5 days of water maze testing. Each day, the rats were subjected to four 1-min trials in the maze, and the data of the four trials were averaged for each rat. Analysis of data by two-way ANOVA indicated significant differences for treatment (p < 0.0001) and time (p < 0.0001). For each day, significant differences of epileptic rats to controls are indicated by asterisk (p < 0.05). Within group comparisons indicated significant (p < 0.01) learning in naive controls, sham controls, and PB responders, but not nonresponders. Naive controls did not differ significantly from sham controls at any experimental day. For further details see  Fig.  3 legend.

In the spatial probe on day 5 of the trial, in which the platform was removed and the crossings of the former platform position during a single trial were recorded over 60 s, the four groups of rats showed the expected behavior. Epileptic rats significantly less often crossed the former location of the platform than nonepileptic rats, without significant differences between PB responders and nonresponders (not illustrated).

A general problem with several epileptic rats of both groups in the water maze was that these animals, in contrast to nonepileptic controls, appeared not interested to find the platform or stay on the platform, even if placed on the platform. It was not clear whether this unusual behavior in the water maze was due to a lack of motivation to leave the water or to a problem in understanding the context (or situation).

Effect of PB on seizure thresholds after SE

In part of the epileptic rats (7/15), the acute effect of PB on ADT and GST was determined before treating the animals for suppression of SRS. As shown in Fig. 9, PB did not significantly increase ADT, but caused a fourfold increase in GST (p = 0.0004). When the effect of PB was separately calculated for rats that were subsequently either responders or nonresponders in terms of suppression of SRS by PB, a significant, eigthfold increase of GST was determined in responders, whereas only a nonsignificant, twofold difference was observed in nonresponders. However, although this indicates that responders were also more sensitive to the acute effects of PB than nonresponders, there was no all-or-none difference between the two groups, but GST was markedly increased (by 1,500 and 200%) in two of the three nonresponders, while only the third nonresponder did not exhibit any increase in GST.

Figure 9.

 Acute anticonvulsant effect of phenobarbital (PB) in seven epileptic rats before treating the animals for suppression of SRS and selection of responders and nonresponders. The idea behind this experiment was to test whether failure of PB to suppress SRS can be predicted by failure of PB to increase electrically induced seizure thresholds. Two types of thresholds were determined; the afterdischarge threshold (ADT) and the threshold for secondarily generalized seizures (GST). PB was administered at a dose of 30 mg/kg 1 h before threshold determination; saline was injected for determination of control thresholds in the same rats with an interval of 2 days between saline and PB. The experiment was performed about 8  weeks after SE (about 4  weeks before treating the animals for suppression of SRS). Data are shown as means ± S.E.M. (A) In the whole group of rats (all), PB did not significantly increase ADT. Individual data indicated that ADT was increased by 45–83% in four rats, but no increase versus control was seen in three rats, so that the average group threshold determined after PB did not significantly differ from control. When data were separately illustrated for subsequent responders (n  = 4) and nonresponders (n  = 3), no significant increases in ADT were seen in any group. Individual data indicated that ADT was not increased at all in two responders and one nonresponder. (B) GST was significantly increased in the whole group (p  = 0.0004) as well as in the responder subgroup (p < 0.0001), whereas no significant effect was determined in the nonresponder group. Individual data indicated that GST was increased in all responders and two of three nonresponders. Control ADT or GST did not differ significantly between subgroups.

Differences in neuropathology between responders and nonresponders

As reported recently (Volk et al., 2006; Bethmann et al., 2008), PB nonresponders differed strikingly from responders in hippocampal damage. Visual inspection of thionin-stained sections did not indicate any difference between PB-responding epileptic rats and nonepileptic controls (sham or naive). The only exception was one rat (NIH 7) of the 7 PB-responders that could be analyzed in the present study (see Methods). This rat showed bilateral damage (scores 1–2) in CA1, CA3, and dentate hilus (Table 1). In contrast to most responders, all four PB nonresponders exhibited neuronal damage in the hippocampal formation, with neuron loss in CA1, CA3, and/or dentate hilus (Table 1). When the number of rats with damage in CA1, CA3, or hilus was statistically compared between the responder and nonresponder subgroups by Fisher's exact test, the difference in incidence of neurodegeneration in PB responders (1/7) and PB nonresponders (4/4) was significant (p = 0.0152).

Table 1.   Neurodegeneration in the hippocampal formation of rats with spontaneous recurrent seizures
RatsNeuronal loss in the hippocampal formation (scores)
CA1CA3aCA3c/CA4Dentate hilus
  1. Morphological alterations were semiquantitatively assessed by a grading system as described in the Methods section. Rats were grouped according to their response to phenobarbital into responders and nonresponders (see  Fig.  2). By comparison with nonepileptic controls (not included in the table), neurodegeneration was only seen in one of the seven responders, for which morphological data were available. In contrast, all four nonresponders exhibited different patterns of hippocampal cell damage as indicated in the table (p  = 0.0152 versus responders; Fisher's exact test). The most common damage in nonresponders was cell loss in the dentate hilus. In most rats, no differences between hemispheres were observed, so that data are not shown separately for the right and left hemisphere. If there was a difference between hemispheres, scores for both hemispheres are shown (e.g., “1/2”).

Responders
 NIH 40000
 NIH 7211/22
 NIH 110000
 NIH 210000
 NIH 230000
 NIH 250000
 NIH 310000
Nonresponders
 NIH 62212
 NIH 8221/22
 NIH 18100/12
 NIH 951011

Discussion

This is the first study showing that pharmacoresistant rats differ in several behavioral aspects from pharmacoresponsive rats in a rat model of TLE. Traditionally, TLE is considered to present a relatively specific risk factor for psychiatric comorbidities, notably for affective disorders, because of the major involvement of the limbic system both in seizure generation in TLE and in the regulation of affect and mood (Perini et al., 1996; Swinkels et al., 2006). Depression, anxiety, and psychosis are among the frequent comorbid psychiatric disorders identified in patients with TLE (Kanner, 2004, 2005; Jones et al., 2007). Recently, Hitiris et al. (2007) reported that psychiatric comorbidity predicts lack of response to AED therapy in patients with epilepsy. This finding was based on outcome data analyzed from 780 patients newly diagnosed with epilepsy and followed up over a 20-year period to investigate which clinical factors predicted pharmacoresistance (Hitiris et al., 2007). Psychiatric comorbidity was analyzed in 15% of AED responders versus 28% of nonresponders (p = 0.002).

In the present study in an animal model of epilepsy, PB resistant rats exhibited significantly more intense behavioral alterations than PB responsive rats in the open field, elevated-plus maze, and tests of behavioral hyperexcitability. In the Morris water maze, PB nonresponders were the only group in which no significant learning was determined over the period of the experiment. Thus, PB nonresponders differed from responders and nonepileptic controls in most  behavioral tests used in this study. Although behavioral abnormalities were also observed in PB responders in all tests, they were much less marked compared to PB resistant rats.

In the open field, a popular test for evaluating exploratory and anxiety-like behavior, both groups of epileptic rats stayed longer in the aversive areas of the field than controls, with a more than threefold difference between PB responders and nonresponders. A similar behavior was observed in the elevated-plus maze test of unconditioned anxiety-like behavior, in which PB responders and, more markedly, nonresponders stayed longer on the aversive open arms than controls. We previously reported that epileptic rats from this post-SE model of TLE did not exhibit any signs of increased anxiety-like behavior in the open field and elevated-plus maze, but instead displayed an increased time spent in the aversive locations of these tests (Brandt et  al., 2006b). Similar findings have recently been reported for the pilocarpine model of TLE (Detour et  al., 2005;  Dos Santos et  al., 2005). Because pilocarpine-treated rats exhibit lesions in the ventral hippocampus, entorhinal cortex, and amygdala, that is, networks which are involved in fear expression,  Detour et  al. (2005) suggested that the disruption of these networks causes a misevaluation of threatening situations, which could in turn reduce anxiety or, more likely, enhance impulsive unadapted behavior. In other words, the “anxiolytic” profiles of these epileptic rats may not be specifically linked to anxiety per se, but are maybe more reflective of impulsivity or the inability to normally perceive or process potentially threatening situations. In the post-SE TLE model used in the present study, we have previously described lesions in the hippocampus, entorhinal cortex, and amygdala (Brandt et  al., 2003a), so that the suggestion of  Detour et  al. (2005) could also be valid for the altered behavior of epileptic rats in the open field and elevated-plus maze observed in our previous (Brandt et al., 2006b) and present studies. This hypothesis is supported by our recent observation that treatment of rats with valproate after SE, which reduced neuronal damage, normalized the rats' behavior in the open-field and elevated-plus maze (Brandt et  al., 2006b). The present finding that the maladaptive behavior of epileptic rats in the open field and elevated-plus tests was much more intense in PB responders than responders is most likely related to our recent observation that neuronal damage is almost exclusively seen in PB nonresponders, while, at least in the hippocampus, no obvious damage occurs in responders (Volk et  al., 2006;  Bethmann et  al., 2008).

In the battery of tests for behavioral hyperexcitability proposed by Rice & DeLorenzo (1998) to discriminate differences in behavioral excitability and sensory responsiveness between epileptic and control rats, PB nonresponders differed from nonepileptic controls in three of four tests, while such difference was seen in only one of four tests for the responders, again demonstrating the marked behavioral differences between the two subgroups of epileptic animals. Furthermore, in more than 30 epileptic rats of the BLA model that we selected for PB response in recent years, aggressive behavior was exclusively observed in PB nonresponders, particularly in rats with a very high seizure frequency. In the pilocarpine model, epileptic rats scored significantly higher than controls in the touch-response and pick-up tests, but not in the approach-response and finger-snap tests (Rice & DeLorenzo, 1998). Inhibition of the N-methyl-d-aspartate (NMDA) subtype of glutamate receptors by MK-801 (dizocilpine) 20 min before pilocarpine—under conditions that did not alter the duration of SE—completely prevented these behavioral changes, indicating that NMDA receptor activation during SE may play an important role in causing behavioral morbidity in post-SE models of TLE (Rice & DeLorenzo, 1998). However, an alternative explanation is the neuroprotective effect of NMDA receptor antagonists such as MK-801 in such models (Fujikawa et al., 1994; Fujikawa, 1995; Brandt et al., 2003b), which was not taken into account in the study of Rice & DeLorenzo (1998). We (Brandt et al., 2006b) have recently shown that prevention of hippocampal damage after SE also completely counteracts the behavioral alterations in tests for behavioral hyperexcitability proposed by Rice & DeLorenzo (1998).

In the forced swim test, the most widely used model of depression-like behavior (learned despair) in rodents (Petit-Demouliere et al., 2005), both groups of epileptic rats did not develop increased immobility on day 2 of the test as seen in controls, which was particularly pronounced in PB nonresponders, which displayed no learned despair at all. The forced swimming test is a common preclinical procedure highly predictive of antidepressant action in patients (Cryan et al., 2005). Antidepressant pretreatment reverses the immobility in this test. However, we do not believe that the markedly reduced immobility of the epileptic rats is indicative of “antidepressant behavior” but rather regard the abnormal behavior of the epileptic animals as a result of a problem in understanding the context (or situation). In other words, whereas normal rodents rapidly learn to cope with the inescapable stress by developing immobility, the epileptic rats seem not to understand that they cannot escape and therefore continue to struggle. Similar observations were recently reported by us for epileptic mice of the pilocarpine model of TLE (Gröticke et al., 2007). In humans, problems in context processing are for instance observed in some types of psychoses such as schizophrenia and in patients with panic attacks (Dibartolo et al., 1997; Barch et al., 2003). Both types of psychopathology are frequently observed in epilepsy patients (Boro & Haut, 2003; Devinsky, 2003). In addition, cognitive impairment may be involved in the present observation in rats.

To directly assess cognitive impairment, we used the Morris water maze, a behavioral procedure designed to test spatial learning and memory (Morris, 1984). As previously reported for the BLA model used in the present study (Brandt et al., 2006b, 2007), epileptic rats exhibited impaired learning in this test. This was particularly pronounced in PB nonresponders, in which no significant improvement in escape latencies was observed over the duration of the experiment. Because learning in the water maze primarily depends on an intact hippocampus (Stafstrom, 2006), the complete failure of PB nonresponders in this test could be explained by the marked hippocampal degeneration found in such animals. This, however, would not explain the impaired learning determined in PB responders, in which no hippocampal damage is observed in the majority of animals, substantiating previous reports in such rats (Volk et al., 2006; Bethmann et al., 2008). How to explain this apparent paradox? One possible explanation is that the histological methods used to assess neurodegeneration in the hippocampal formation of PB responders may have missed subtle but functionally relevant morphological alterations. Another possible explanation refers to the observation described in the Results section that several epileptic rats of both the responder and nonresponder groups displayed abnormal behavior in the water maze in that they appeared not interested to find the platform, climb on the platform or stay on the platform. This could relate to a problem in understanding how to escape from the water, that is, the context of the test. Thus, similarly as suggested for the forced swim test, in addition to memory loss because of neurodegeneration, a problem in context understanding in a stressful situation could add to the abnormal behavior of epileptic rats in the Morris water maze test. This has recently also been suggested for the abnormal behavior of epileptic mice in this test (Gröticke et al., 2007).

There is increasing evidence that a history of psychiatric diseases, particularly depression along with anxiety, increases the risk of later epilepsy (Kanner, 2006). Furthermore, a history of depression has been reported in up to two-thirds of patients with medically intractable epilepsy (Lambert & Robertson, 1999). We therefore studied whether rats that later develop pharmacoresistant seizures after SE-induction differ behaviorally from pharmacoresponsive animals or nonepileptic controls before SE. In order to minimize behavioral habituation, only one test, the elevated-plus maze, was used to address this question. Although outbred rats may show marked interindividual variation in tests of baseline anxiety (Adamec et al., 2005), no significant intergroup differences were observed in the present experiments. Adamec et al. (2005) have recently reported that the level of anxiety in the elevated-plus maze in Wistar rats before BLA kindling determines the effect of kindling on anxiety. In other words, the premorbid affective state of the animals contributes to effect of kindling on affect, indicating that premorbid anxiety level interacts with BLA kindling to influence lasting behavioral outcome (Adamec et al., 2005). Highly anxious rats became less anxious, whereas rats with lower anxiety levels became more anxious after kindling of the right BLA (Adamec et al., 2005). Whether the premorbid affective state of the animals also contributes to pharmacoresistance in kindled or epileptic rats is not known, and the present experiments were only performed in a relatively small sample size. Nevertheless, the approach used in the study of Adamec et al. (2005) and in the present experiments is of obvious clinical significance and more studies on consequences of premorbid affective state on outcome after induction of epilepsy are needed.

Of course, it would have been desirable to include more PB nonresponders in the present study. In our model, about 30–40% of epileptic rats do not respond to PB, resulting in a higher yield of responders compared to nonresponders (Brandt et al., 2004; Bethmann et al., 2007). Furthermore, in the present experiments some nonresponders, particularly animals with extremely high seizure frequency, died in the period between the PB trial and onset of the behavioral experiments, further reducing the group size of nonresponders. Such unexpected deaths were not observed in responders. However, despite the small sample size of nonresponders, these rats significantly differed in most behavioral tests from responders and nonepileptic controls, indicating that these differences were quite robust. It is unlikely that the differences in behavior and cognition between PB responders and nonresponders determined long after a 2-week treatment period with PB are a result of this previous treatment, because both subgroups received the same PB treatment before the behavioral experiments. Furthermore, in order to minimize any effect of the PB treatment on behaviors determined after this treatment in responders and nonresponders, we used a long wash-out period (∼6 months) after PB treatment before starting with the behavioral experiments. Certainly, it would have been interesting to involve also additional groups, that is, a group of sham animals treated with PB for two weeks and a group of amygdala-stimulated (epileptic) animals treated with vehicle instead of PB and then tested 6 months later in the behavioral battery. However, the present study already comprised some 90 rats (see Methods), so that inclusion of additional control groups was not feasible. Furthermore, using the same TLE model, we have previously characterized behavioral alterations of vehicle-treated epileptic rats in the behavioral battery also used in the present study (Brandt et al., 2006b, 2007).

Overall, our data suggest that intense behavioral and cognitive alterations are a predictor of pharmacoresistance, which is in line with the recent clinical study of Hitiris et al. (2007), demonstrating that animal models can be used to evaluate clinical hypotheses. Another clinical factor that predicts pharmacoresistance is high seizure number or density before onset of treatment with AEDs (Sillanpaa, 1993; Kwan & Brodie, 2000; MacDonald et al., 2000; Leschziner et al., 2006; Hitiris et al., 2007). In the present study, the two rats with the by far highest seizure rate of the group of 15 epileptic rats were in the PB nonresponder group. However, because the two other nonresponders did not differ in seizure frequency from responders, the average seizure frequencies of responders and nonresponders did not statistically differ. Thus, in our model, pharmacoresistant epilepsy is not just a different, more severe type of epilepsy than pharmacoresponsive epilepsy. In part of the rats, high seizure frequency may contribute to resistance, but other factors are certainly also important.

One of these factors may be hippocampal damage, because neurodegeneration in CA1, CA3, or dentate hilus was determined in all PB nonresponders, but only one of the responders examined in this regard, confirming previous studies in such rats (Volk et al., 2006; Bethmann et al., 2008). Hippocampal damage is long known to be associated with AED resistance in patients with TLE (Schmidt & Löscher, 2005), substantiating that the TLE model used by us shows many parallels with human TLE.

Additional differences between PB responders and nonresponders reported by us previously include overexpression of the drug efflux transporter P-glycoprotein in nonresponders (Volk & Löscher, 2005), an increase in diazepam-insensitive GABAA receptor binding in the dentate gyrus granule cell layer of nonresponders (Volk et al., 2006), and marked differences in GABAA receptor subunit expression between responders and nonresponders (Bethmann et al., 2008). Because the duration and severity of the brain insult (the SE) that induced epilepsy in both groups are the same (Volk et al., 2006), the only explanation for these multiple differences between AED resistant and responsive subgroups of epileptic rats are premorbid genetic differences leading to distinct conditions after SE. This seems to confirm the previous clinically based suggestion that medically intractable epilepsy is a distinct condition that may be identified early (Arroyo et al., 2002; Kwan & Brodie, 2002).

In part of the epileptic rats characterized in the present study, we tested whether the acute effect of PB on electrically induced focal or secondarily generalized seizures can be used as a predictor of pharmacoresistance. If so, this could replace the time-consuming and labor-extensive selection of responders and nonresponders by prolonged PB treatment and video/EEG-recording and analysis of SRS. PB did not significantly increase ADT in epileptic rats, but markedly increased GST. Such marked GST increases were observed in all but one rat, which was subsequently identified a PB nonresponder. However, two other nonresponders exhibited large GST increases after PB, so that this acute testing of anticonvulsant activity on induced seizures was not a reliable predictor of resistance of spontaneous seizures.

In conclusion, our hypothesis that AED-resistant rats will show more severe behavioral and cognitive changes than AED-responsive rats, which was based on clinical data (Lambert & Robertson, 1999; Hermann & Jones, 2006; Hitiris et al., 2007), was confirmed by the present experiments. Our data demonstrate that post-SE models of TLE are useful to delineate predictors of pharmacoresistant epilepsy in the absence of several confounding variables that impede such studies in the clinical setting, including the underlying etiology, AED treatment, and developmental status of the patient.

Acknowledgments

We thank Prof. H. Emrich (Department of Clinical Psychiatry and Psychotherapy, Medical School Hannover, Germany) and Prof. J.P. Huston and his colleagues (Institute of Physiological Psychology, Center for Biological and Medical Research, University of Düsseldorf, Germany) for helpful discussions and advice during establishment of the behavioral models and interpretation of data obtained in these models. We are grateful for technical support by Nicole Ernst and Julia Förster. The study was supported by a grant (Lo 274/9) from the Deutsche Forschungsgemeinschaft, a grant (R21 NS049592) from the National Institutes of Health (NIH; Bethesda, MD, U.S.A.), and a Ph.D scholarship (to J.B.) from the Konrad-Adenauer-Stiftung (Sankt Augustin, Germany).

Conflict of interest: 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. None of the authors has any conflict of interest to disclose.

Ancillary