Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Sep 1.
Published in final edited form as: Epilepsia. 2016 Jul 28;57(9):1386–1397. doi: 10.1111/epi.13476

Acute Cognitive Impact of Antiseizure Drugs in Naïve Rodents and Corneal-Kindled Mice

Melissa L Barker-Haliski 1,2,3, Fabiola Vanegas 1, Matthew J Mau 1, Tristan K Underwood 1, H Steve White 1,2
PMCID: PMC5012933  NIHMSID: NIHMS798307  PMID: 27466022

Summary

Objective

Some antiseizure drugs (ASDs) are associated with cognitive liability in patients with epilepsy, thus ASDs without this risk would be preferred. Little comparative pharmacology exists with ASDs in preclinical models of cognition. Even less data exists on the acute effects in rodents with chronic seizures. Predicting risk for cognitive impact with preclinical models may supply valuable ASD differentiation data.

Methods

ASDs (phenytoin (PHT); carbamazepine (CBZ); valproic acid (VPA); lamotrigine (LTG); phenobarbital (PB); tiagabine (TGB); retigabine (RTG); topiramate (TPM); levetiracetam (LEV)) were administered equivalent to maximal electroshock median effective dose (ED50; mice, rats), or median dose necessary to elicit minimal motor impairment (TD50; rats). Cognition models with naïve adult rodents were Novel Object/Place Recognition (NOPR) with CF-1 mice, and Morris water maze (MWM) with Sprague Dawley rats. Selected ASDs were also administered to rats prior to testing in an open field. The effect of chronic seizures and ASD administration on cognitive performance in NOPR was also determined with corneal-kindled mice. Mice that did not achieve kindling criterion (partially-kindled) were included to examine the effect of electrical stimulation on cognitive performance. Sham-kindled and age-matched mice were also tested.

Results

No ASD (ED50) affected latency to locate the MWM platform; TD50 of PB, RTG, TPM, and VPA reduced this latency. In naïve mice, CBZ and VPA (ED50) reduced time with the novel object. Interestingly, no ASD (ED50) affected performance of fully kindled mice in NOPR, whereas CBZ and LEV improved cognitive performance of partially-kindled mice.

Significance

Standardized approaches to the preclinical evaluation of an ASD’s potential cognitive impact are needed to inform drug development. This study demonstrated acute, dose- and model-dependent effects of therapeutically-relevant doses of ASDs on cognitive performance of naïve mice and rats, and corneal-kindled mice. This study highlights the challenge of predicting clinical adverse effects with preclinical models.

Introduction

Epilepsy is not only characterized by chronic seizures, but also cognitive comorbidities that can be as detrimental to a patient’s quality of life as the seizures themselves1; 2. More harmful to a person’s normal quality of life is the fact that many antiseizure drugs (ASDs) also adversely affect cognition3, which may contribute to medication discontinuation4 and lead to an increased risk of epilepsy-related morbidity and mortality (e.g. sudden unexplained death in epilepsy5). Thus, identifying novel ASDs that demonstrate promising antiseizure profiles in the absence of adverse effects on cognition is of great translational value6.

The Morris water maze (MWM) is a widely-used rodent model of spatial learning and memory that is applicable to drug development for epilepsy7–9. Evidence indicates that rodent performance in the MWM is mediated by dorsal hippocampus10; 11, a region implicated in the pathophysiology of temporal lobe epilepsy (TLE), albeit other hippocampal regions (e.g. ventral) that contribute to human TLE are less well-studied. Chronic administration of phenytoin (PHT) or carbamazepine (CBZ; sufficient to achieve levels equivalent to those found in human plasma) adversely affects performance of naïve rats in the MWM12; valproic acid (VPA) and ethosuximide do not impair performance. However, the acute performance of rats in the MWM following administration of mechanistically-diverse ASDs at doses sufficient to block rodent seizures has yet to be evaluated. Additionally, it is unknown whether a rapid, single-day drug screening approach in the MWM can identify dose-dependent effects of prototype ASDs consistent with clinical experience.

The novel object/place recognition task (NOPR) is a non-aversive, innate model of working memory13 that challenges the ability of rodents to discern a novel object from a previously presented, familiar object14. Importantly, this rapid task is well-suited to drug screening efforts15 and predominately relies on a different brain region, namely perirhinal cortex16, than dorsal hippocampus utilized in MWM. Of note, perirhinal cortex is affected by temporal lobe epilepsy (TLE), with patients exhibiting reduced perirhinal cortex volumes17. Thus, identifying compounds that adversely affect perirhinal cortex-dependent processes may provide useful information for ASD differentiation and simultaneously interrogate an alternative region relevant to TLE and thus inform on the potential for adverse effects. There are presently no reports on the acute effects of ASDs on naïve rodent performance in NOPR, although CBZ enhances discriminative memory of rats with status epilepticus-induced chronic seizures18. Moreover, perirhinal cortex-kindled rats exhibit deficits in this task19, but whether other forms of kindling can affect NOPR performance is unknown. The corneal-kindled mouse is frequently utilized for the evaluation of investigational agents and demonstrates a pharmacological profile consistent with the hippocampal-kindled rat20. More importantly, the pharmacological profile of the corneal-kindled mouse is consistent with human partial epilepsy20; 21, thereby confirming the utility of this model for ASD discovery.

The early evaluation of investigational agents in several animal models of cognition may provide useful cumulative information to differentiate otherwise promising investigational ASDs, and inform on the potential for untoward effects in a clinical setting. The National Institute of Neurological Disorders and Stroke has indeed determined that standardized approaches to evaluating the potential for cognitive liability by novel ASDs are needed for preclinical drug differentiation22. More so, a chronically-seizing substrate may provide better translational data for compound testing23. No such comprehensive systematic study has yet been attempted to define a reasonable and feasible preclinical approach for the early identification of novel ASDs. To this end, a standardized assessment of approved ASDs is necessary to assess the predictive validity of the selected preclinical platform. Therefore, the present studies were designed to systematically characterize the response of rodents to a variety of clinically-available, mechanistically-diverse ASDs in several well-validated animal models of cognition. The results of this study, described herein, provide a standardized basis for the differentiation of investigational ASDs.

Methods

Animals and antiseizure drugs

Rodents included male CF-1 mice (13–15 g) and Sprague Dawley rats (100–125 g). Male Black Swiss (n=12/group; 22–28 days old) and C57Bl/6 mice (n=12/group; 22–28 days old) were used to confirm the effects of subcutaneous administration of scopolamine (1.5 mg/kg) vs CF-1 mice in the NOPR test. All animals had free access to food and water, except during behavioral testing. Rodents were acquired from Charles River Laboratories. Rodents were allowed at least a 1-hour acclimatization in the testing room before commencing any behavioral assay. All testing was conducted between 9 am and 5 pm during the animals’ light cycle; animals were housed on a 12:12h light:dark cycle. With the exception of corneal-kindled mice in the NOPR task, all rats and mice were used once. Animal care and use was approved by the University of Utah Institutional Care and Use Committee. For all behavioral assays described below, testing and data analysis were conducted by an experimenter blinded to treatment.

Prior to each test, the ASD was administered i.p. at a dose equivalent to the maximal electroshock (MES) median effective dose (ED50; mice and rats), or median dose necessary to elicit minimal motor impairment (median toxic dose, TD50; rats, Table 1)24. The ASDs evaluated included phenytoin (PHT); carbamazepine (CBZ); valproic acid (VPA); lamotrigine (LTG); phenobarbital (PB); tiagabine (TGB); retigabine (RTG); topiramate (TPM); levetiracetam (LEV; Table 1). The dose of LEV was based on the mouse 6 Hz test25, or 100 mg/kg (rat). Performance of rodents was tested at the time of peak effect (TPE) of each ASD. With the exception of LTG (AK Scientific, Inc.), all ASDs and scopolamine hydrochloride were acquired from Sigma Aldrich.

Table 1.

Antiseizure drugs and doses utilized in each test.

Antiseizure Drug Sprague Dawley Rat MES ED50 (mg/kg, i.p.) Sprague Dawley Rat TD50 (mg/kg, i.p.) CF-1 Mouse MES* ED50 (mg/kg, i.p.)
Valproic Acid (VPA) 147 316 214
Carbamazepine (CBZ) 4 33 17
Tiagabine (TGB) 7 8 5
Phenytoin (PHT) 1.5 15 7
Phenobarbital (PB) 2 41 14
Lamotrigine (LTG) 2.5 22 8.5
Levetiracetam (LEV)* 100* -- 33*
Retigabine/Ezogabine (RTG) 3 42 9.5
Topiramate (TPM) 3 32 22
Open in a new tab
*

The dose of LEV was based on the CF-1 mouse 6 Hz test at 32 mA stimulation intensity 25, or at 100 mg/kg for Sprague Dawley rats.

Corneal kindling of CF-1 mice

Mice were kindled electrically with 3 sec stimulation, 3 mA, 60 Hz, and corneal electrodes to a criterion of 5 consecutive Stage 5 seizures26 according to procedures previously described20. Fully-kindled mice were those that achieved 5 consecutive Stage 5 seizures20, whereas partially-kindled mice underwent corneal kindling but did not achieve full kindling criterion. Sham-kindled mice had 0.5% tetracaine and a stimulating electrode placed twice daily onto the cornea, but electrical stimulation was not delivered. Finally, age-matched controls were received in-house coincident with all kindling groups, but not subjected to kindling or sham-kindling. Testing of ASDs in the NOPR task commenced at least 5–7 days after the last kindling stimulation (or sham) session. The evaluation of selected prototype ASDs at the MES (or 6 Hz for LEV) ED50 in a cohort of fully corneal-kindled, partially-kindled, sham-kindled, and age-matched mice was completed after kindling acquisition using a Latin-square design such that each mouse served as its own control. Thus, corneal kindled mice were subjected to repeated ASD administration over the course of these studies, with a minimum of 3 days of washout between drug administration days.

Novel object place recognition with mice

The performance of untreated inbred C57Bl/6 mice (n=12) and outbred Black Swiss mice (n=10) was compared to untreated outbred CF-1 mice (n=10) in NOPR (described below). These strains were selected because they are more frequently used for behavioral testing and would thus demonstrate whether CF-1 mice are also suitable. Our group has extensively characterized the pharmacology of CF-1 mice in acute seizure models and corneal-kindled mice20; 25, which was the basis for using this strain. The effect of prototype ASDs in naïve, male CF-1 mice (n=12/group) was conducted in the NOPR test. Corneal-kindled mice (n=12/group) were also employed to assess the combined effect of chronic seizures and acute drug administration on cognitive performance in the NOPR test. Partially-kindled mice were included to examine what, if any, effect electrical stimulation might have on NOPR performance; results were compared to those obtained in the NOPR test for sham-kindled and age-matched mice.

NOPR was designed to assess hippocampus-independent16 working memory following acute administration of an ASD in naïve CF-1 mice, with the ultimate goal being to identify investigational agents that could adversely affect performance in this task. This hybrid episodic memory task was based on the novel object recognition task27; 28 and object displacement task27 to provide a platform to rapidly interrogate the potential for acute effects of ASD administration in a drug screening setting. In addition to designing these studies for moderate-throughput screening applications (i.e. short testing windows), the short (5-min) retention interval primarily engages extrahippocampal structures16, so that any effects of corneal kindling on hippocampal integrity or astrogliosis29 would not confound the performance of mice in the NOPR task. Mice were habituated to a clear, Plexiglas testing chamber (40L × 40W × 35H cm) 24-hours prior to the testing day. The following day, working memory of mice was tested at the ASD TPE by first exposing a mouse to a pair of identical “familiar” objects (for example, 2.2L × 2.2H × 2.2W cm red Lego) for 15 minutes at an approximate distance of 15 cm apart within the chamber (Acquisition phase). After a 5-minute retention interval wherein the mouse was isolated from the test environment, one “familiar” object was replaced with a “novel” object (for example, 5D × 3W cm yellow yo-yo toy) in a new location and the mouse’s behavior was recorded for an additional 5 min (Test phase). The time that the mouse spent with the novel object versus the time that was spent with the familiar object was determined based on the total time spent with both objects during the Test phase and results are represented as an “index of recognition” (RID): RID = (Time Novel − Time Familiar) / (Time Novel + Time Familiar). A cognitively-intact mouse will innately spend more time with a “novel” than with a “familiar” object14. Thus, a mouse that shows no preference for either object presented during the Test phase would be considered impaired by an acute ASD. The outcome measure was the ability of VEH- and ASD-treated mice to discern the familiar from the novel object during the 5 min Test phase. A cohort of VEH-treated mice (n=12) was tested with each ASD prototype to control for the potential for fluctuations in daily testing conditions. Additionally, total distance traveled in the chamber was recorded for mice treated with each ASD to control for any general effects on motor behavior. For simplicity, all RID scores and distance traveled are normalized to the VEH-treated group score for that ASD testing day and represented as individual graphs in Figure 4. All data were recorded by an automated video tracking system (Noldus) and reviewed by an experimenter blinded to treatment condition.

Figure 4.

Figure 4

Open in a new tab

Effect of Prototype ASDs on Novel Object Place Recognition task performance in naïve 4–6 week-old CF-1 male mice (n=12/group). All ASDs were tested with a VEH-treated control cohort run on the same day and results compared to the VEH-treated group by two-tailed t-test. Results are represented on the same figure for presentation purposes. A) Only VPA (48.9±14.3% of control) and CBZ (30.8±14.3% of control) significantly impaired working memory performance of CF-1 mice after a 5-minute delay. B) The effects of each ASD on total exploration suggest that VPA induced generalized suppression of motor behavior at the MES ED50, which may have contributed to reductions in novel object preference. However, CBZ treatment was not associated with significant reductions in general motor behavior in the open field. Surprisingly, PB was associated with hyperactivity in the open field, whereas RTG was associated with acute reductions in exploratory behaviors. Neither PB nor RTG treatment were associated with significant reductions in novel object preference of naïve CF-1 mice. * Indicates significantly different from control mice, p < 0.05; ** indicates significantly different from control mice, p<0.01.

Morris water maze with rats

Performance of rats (n=10/group) in the MWM was tested using a single-day, 5-trial drug screening protocol modified from our previous work with this model30. Briefly, rats were administered the ASD (ED50 or TD50; Table 1) and challenged at the TPE. The outcome measure was the rats’ ability to find the hidden platform. The maze was surrounded by a white curtain and spatial cues. Each trial lasted 120 seconds, with a minimum inter-trial interval of 5 minutes in a heated holding cage. The rat’s performance was recorded using automated video tracking (HVS Image).

Open field activity of rats

In an effort to further define the potential of a given ASD to produce behavioral deficits, the TD50 of selected ASDs was administered to rats (n=10/group) before they were evaluated in an automated OF, as described previously for mice31; 32. Briefly, a rat was placed at the ASD TPE into a Plexiglas chamber (40L × 40W × 30H cm) equipped with infrared sensors to detect movement for 10 minutes. During the 10-minute period, total distance travelled (cm), horizontal activity counts, and vertical rearing activity counts were measured and recorded by an automated system (OmniTech Electronics).

Statistical analysis

Performance in the MWM was evaluated by repeat measures ANOVA with Tukey’s post-hoc test. OF activity was evaluated by one-way ANOVA, with Dunnet’s post-hoc test. NOPR activity of naïve mice was compared by Student’s t-test, but represented as one graph for clarity in Figure 4. NOPR activity of kindled mice treated with each ASD was compared by two-factor ANOVA, with uncorrected Fisher’s post-hoc test to compare kindling groups. Data analysis was conducted with GraphPad Prism v.6.0, with p < 0.05 considered statistically significant.

Results

Performance of rats in MWM is acutely impaired by some ASDs

Acute administration of a number of the ASDs at their respective ED50 or TD50 resulted in notable variability in the spatial memory of rats in MWM (Figure 1). For all treatment conditions, rodents demonstrated significant reductions in the latency to locate the hidden platform (Figure 1A–1I; see legend for statistics). A significant overall treatment effect for VPA (F(2,26)=7.13, p=0.0034), PB (F(2,27)=17.5, p<0.0001), RTG (F(2,27)=10.1, p=0.0005), and TPM (F(2,27)=4.88, p=0.022), with post-hoc analysis was observed demonstrating significant differences between rats dosed at their TD50 and naïve rats at a minimum of one testing trial (see legend). Post-hoc analysis demonstrated that no ASD, at its respective ED50, was capable of significantly impairing performance in the MWM. In contrast, analysis of swim speed demonstrated dose-dependent effects of VPA, PB, and TPM on overall motor performance (Figure 1J). Average distance travelled for all trials was affected by the ED50 of TGB, and also the TD50 of VPA, PB, RTG, and TPM (Figure 1K).

Figure 1.

Figure 1

Open in a new tab

The performance of male Sprague Dawley rats treated acutely with either VEH (green line, naïve), an ASD at the MES ED50 (or 100 mg/kg for LEV; red line), or an ASD at the TD50 (blue line) was evaluated at the time of peak anticonvulsant efficacy (TPE) in the Morris water maze (MWM). Importantly, all treatment groups demonstrated significant time-dependent improvements in latency to locate the hidden platform (i.e. main trial effect, p < 0.05), but only a few ASDs induced significant acute treatment effects in this task. A) There was a significant effect of VPA treatment on performance in the MWM (F(2,26)=7.13, p=0.0034). Post-hoc tests revealed that rats treated with the VPA TD50 performed significantly worse than rats treated with VEH- (* trial 3 and 4) and the VPA ED50 (# trial 3), p<0.01 for all. B) Treatment with two doses of CBZ did not significantly affect performance of rats in the MWM task (F(2, 26) = 0.114, p>0.5). C) Treatment with two doses of TGB did not significantly affect latency to the platform of rats in the MWM task (F(2, 27)=2.86, p=0.075). D) Treatment with two doses of PHT did not significantly affect performance of rats in the MWM task (F(2, 27)=0.509, p>0.5). E) There was a significant effect of PB treatment on performance in the MWM (F(2,27)=17.5, p<0.0001). Post-hoc tests revealed that rats treated with the TD50 performed significantly worse than rats treated with VEH- (* trials 2–5) and the ED50 of PB (# trial 4 and 5), p<0.01 for all. F) Treatment with two doses of LTG did not significantly affect performance of rats in the MWM task (F(2, 27)=0.907, p>0.4). G) Treatment with 100 mg/kg LEV did not significantly affect performance of rats in the MWM task (F(1, 14)=0.012, p>0.9). H) There was a significant effect of RTG treatment (F(2,27)=10.11, p=0.0005). Post-hoc tests revealed that rats treated with the TD50 performed significantly worse than rats treated with VEH- (* trials 2–5) and the ED50 of RTG (# trial 3 and trial 5), p<0.05 for all. I) There was a significant effect of TPM treatment (F(2,27)=4.88, p=0.022). Post-hoc tests revealed that rats treated with the TD50 performed significantly worse than rats treated with VEH- (* trial 4) or the ED50 of TPM (# trial 2), p<0.05 for all. J) Swim speed was recorded for rats treated with the ED50 or TD50 of each ASD. There was a significant effect of ASD treatment (both ED50 and TD50) on swim speed (F(8,162)=7.30, p<0.0001). Only the TD50 of VPA, TGB, PB and TPM were associated with significant effects on average swim speed in the MWM, *p<0.5, **p<0.01, ****p<0.0001. K) There was a significant effect of ASD treatment on the average distance traveled during the five trials (F(8.162)=8.72, p<0.0001). Only TGB administered at the ED50 significantly increased the average distance traveled (p<0.05); an effect that was not statistically detected in the individual trial assessment (C). The TD50 of VPA, PB, RTG, and TPM significantly increased the average distance traveled of rats during the five training trials, *p<0.5, **p<0.01, *** p<0.001, ****p<0.0001.

Select ASDs suppress OF activity of male rats

The behavior of rats in the non-aversive OF was tested to determine whether TD50 impaired overall motor coordination, which may have contributed to acute deficits in MWM. Measures of vertical activity (Figure 2A), horizontal activity (Figure 2B), and total distance traveled (Figure 2C) were recorded for VPA, CBZ, PHT, PB, and TPM. Vertical rearing activity was significantly reduced by treatment (F(5,41)=5.32, p=0.0007; Figure 2A); in contrast, TPM did not reduce vertical activity. Horizontal activity was similarly reduced by some, but not all, ASDs (F(5,41)=5.52, p=0.0006); VPA (p<0.01), CBZ (p<0.05), and PB (p<0.001) all significantly reduced horizontal activity relative to VEH-treated rats (Figure 2B). Finally, the total distance traveled (F(5,41)=5.68, p=0.0005) was also significantly reduced by acute administration of the TD50 of the ASDs VPA (p<0.001), CBZ (p<0.05), PHT (p<0.05), and PB (p<0.001), but not TPM. Thus, VPA and PB were both associated with reductions in MWM performance and OF behavior, suggesting overall impairments in motor coordination at the dose tested. CBZ and PHT did not induce adverse effects on MWM performance, but they were both associated with significant negative effects on exploratory behavior in the OF. Conversely, TPM did impair MWM performance but was not associated with reductions in OF behavior.

Figure 2.

Figure 2

Open in a new tab

The activity of male Sprague Dawley rats (n = 8/treatment group) in the OF task was evaluated for the 10-minute period surrounding the TPE of VEH or ASDs at the TD50. Measures of A) vertical activity (e.g. rearing behavior), B) horizontal activity, and C) total distance traveled in the OF were measured for all treatment groups. There was a significant treatment effect for all parameters, with Dunnet’s post-hoc test indicating that administration of the TD50 of VPA, CBZ, and PB reduced motor behavior in all three measures. Administration of PHT at the TD50 was associated with significant reductions in A) vertical activity and C) total distance traveled, but rats treated with PHT were not significantly different from VEH-treated rats in the amount of horizontal activity. Administration of the TD50 of TPM did not significantly affect any OF measure. * Indicates significantly different from VEH, *p<0.5, **p<0.01, *** p<0.001, ****p<0.0001.

NOPR performance is not different between mouse strains

We first determined whether CF-1 mice were as capable of discerning the novel from the familiar object as two other strains that are more frequently utilized in cognition testing, by comparing the performance of CF-1 mice to age-matched C57Bl/6 and Black Swiss mice in NOPR (n=10–12/group; Figure 3A). Performance in this task was better than chance for all strains tested (C57Bl/6: t=4.13, p<0.003; Black Swiss: t=3.93, p<0.003; CF-1: t=9.01, p<0.0001; Figure 3A). The RID scores of naïve CF-1 mice were no different than that of C57Bl/6 or Black Swiss (F(2,28)=0.073, p>0.9). Object recognition of all three strains was sensitive to administration of the muscarinic antagonist, scopolamine (1.5 mg/kg, s.c.; Figure 3B–3E). Cholinergic signaling is essential to learning and memory processes33, thus administration of scopolamine is useful to validate the NOPR assay. The total time spent with the novel object was reduced for all groups following scopolamine administration (Figure 3B): C57Bl/6 (t=2.18, p=0.040), Black Swiss (t=2.35, p<0.03) and CF-1 mice (t=3.26, p<0.004). Furthermore, the RID was also significantly reduced following scopolamine administration: C57Bl/6 (t=2.17, p=0.041; Figure 3C); Black Swiss (t=2.40, p=0.025; Figure 3D); and CF-1 mice (t=3.11, p=0.005; Figure 3E). Because CF-1 mice were found to perform in NOPR to a similar degree as mice routinely utilized for cognition testing, and because they were also sensitive to pharmacological blockade of memory consolidation by scopolamine, all subsequent testing with ASDs was conducted in CF-1 mice.

Figure 3.

Figure 3

Open in a new tab

The novel object/place recognition (NOPR) task with a 5-minute delay between exchange of familiar and novel object was used to test the hippocampus-independent working memory of male naïve mice. A) There were no significant differences in the performance on this task between mouse strains that are commonly used for cognition testing and CF-1 mice. Naïve C57Bl/6, Black Swiss, and CF-1 mice all demonstrated a similar preference for the novel object. # All strains performed significantly better than chance, as measured by one sample t-test: C57Bl/6: t=4.13, p<0.003; Black Swiss: t=3.93, p<0.003; CF-1: t=9.01, p<0.0001. B) Subcutaneous administration of the muscarinic antagonist, scopolamine (1.5 mg/kg) significantly impaired the total time spent with the novel object for C57Bl/6 (t=2.18, p=0.040), Black Swiss (t=2.35, p<0.03), and CF-1 mice (t=3.26, p<0.004). Administration of scopolamine (1.5 mg/kg, s.c.) also significantly reduced the recognition index (RID) of C) C57Bl/6 mice 1 hour prior to testing, D) Black Swiss mice 2 hours prior to testing, or E) CF-1 mice 1 hour prior to testing. Thus, the NOPR performance of all of the mouse strains presently tested is sensitive to blockade of working memory by scopolamine administration. * Indicates significantly different from saline, *p<0.5, **p<0.01, *** p<0.001, ****p<0.0001.

Performance of naïve CF-1 mice in NOPR is impaired by acute administration of VPA and CBZ

The performance of CF-1 mice treated with each ASD was normalized to the RID of VEH-treated, age-matched mice tested on the same day. Precautions were taken to minimize variability between testing days; e.g. a comparison to a VEH-treated group for each ASD controlled for any potential differences in daily testing parameters. Only administration of VPA and CBZ were associated with significant reductions in RID of CF-1 mice (Figure 4). VPA-treated mice had an RID 48.9±14.3% (mean ± SEM) of that of VEH-treated mice (t=2.85, p=0.010) and CBZ-treated mice had an RID of 30.8±20.4% of that of VEH-treated mice (t=2.59, p=0.017). No other ASD at the ED50 significantly reduced RID relative to VEH-treated mice run on the same day.

NOPR performance of corneal-kindled CF-1 mice is not impaired by ASDs

The effect of selected ASDs on NOPR performance of corneal-kindled, partially-kindled, and sham-kindled mice was compared to effects in age-matched mice. Importantly, age-matched, VEH-treated mice (approximately 4–6 months old) did not exhibit significantly different performance relative to naïve, VEH-treated CF-1 mice used to validate the NOPR test (Figure 4 vs. Figure 5, p>0.4). Administration of VPA induced a significant main effect of treatment (F(1,131)=9.60, p=0.002; Figure 5A), with age-matched (p=0.0014) and sham-kindled (p=0.025) being significantly affected. CBZ induced a kindling × treatment interaction (F(3,132)=4.04, p=0.009; Figure 5B). Interestingly, CBZ impaired performance of age-matched (p=0.050) mice, but significantly improved performance of partially-kindled mice (p=0.007). Administration of PHT induced a significant main effect of treatment (F(1,131)=9.40, p=0.003; Figure 5C), with age-matched (p=0.033) and sham-kindled (p=0.030) being adversely affected. Administration of LTG did not significantly affect performance of any group (F(1,133)=1.38, p>0.2; Figure 5D). Treatment with LEV also induced a kindling × treatment interaction (F(3,132)=2.90, p=0.04; Figure 5E). LEV administration to partially-kindled mice also improved object recognition (p=0.028). Finally, acute administration of TPM significantly affected performance on NOPR (F(1,137)=7.15, p=0.008; Figure 5F), but only age-matched mice exhibited reduced performance after TPM (p=0.002). Thus, no acute treatment significantly affected cognitive performance of corneal-kindled mice. Interestingly, age-matched, partially-kindled, and sham-kindled mice were all differentially affected by the ASDs under evaluation.

Figure 5.

Figure 5

Open in a new tab

Acute administration of selected ASDs at their respective ED50 does not affect performance of corneal-kindled mice in the Novel Object Place Recognition task. Acute administration of VPA or PHT significantly impairs performance of partially-kindled mice. Acute administration of CBZ or LEV to partially-kindled mice actually improves performance in this task. Administration of VPA, CBZ, PHT, or TPM impairs performance of age-matched CF-1 mice. Significant differences from VEH-treated mice of the same kindling group are indicated as * p<0.05 or ** p<0.01.

Discussion

Herein we present a standardized evaluation of numerous mechanistically-novel, clinically-available ASDs on rodent models of cognition in both naïve mice and rats, as well as mice with chronic network hyperexcitablity. While the ASDs were evaluated at equivalent anticonvulsant doses (e.g. MES ED50), there were differences in the impact based on the species or model at-hand. The present results provide critical insight into the utility and challenge of cognitive effects testing for preclinical drug discovery. Clinical adverse effects often arise after multiple drug administrations, but we presently demonstrate that even acute administration of ASDs affected cognitive performance of rodents. Importantly, the behavioral tasks currently used activate different brain regions, all of which are adversely affected by TLE, to comprehensively interrogate the feasibility and predictive validity of the preclinical evaluation of cognitive impact of investigational ASDs.

Single-day screening in MWM and OF can detect acute, dose-dependent deficits induced by ASDs. VPA, PB, and TPM dose-dependently impaired performance of naïve rats in the MWM; an effect with each ASD that is consistent with clinical literature3. Our single-day protocol replicated earlier findings using a multiple-day MWM protocol demonstrating that LEV is without significant effect on MWM performance, whereas VPA is associated with dose-dependent effects in naïve rats34. PB exhibits mixed clinical effects, and although it is generally well-tolerated, patients have reported slowed movement times, impaired attention, and reduced processing speed3; consistent with the high-dose impairments presently observed in the MWM and OF. Only TPM was without effect on OF behavior, despite significant and dose-dependent slowing of MWM performance. These results suggest that effects of TPM, but not VPA or PB, on MWM performance may be more likely due to overt impairments in spatial memory. Such results for TPM align with clinical literature35. Absence of acute effect of CBZ and PHT on MWM performance is in disagreement with preclinical literature12, but may represent differences in drug administration (acute vs. sub-chronic) and methodology (single-day vs. multiple-day protocol). However, both CBZ and PHT significantly impaired OF activity, but did not elicit notable effects in MWM. Had CBZ or PHT been “unknown” investigational compounds under evaluation in this drug screening approach, the observed effect on OF activity would have signaled the need for further evaluation to assess their effects on motor performance. Altogether, this study highlights that acute adverse effects on rodent cognition can be detected in this rapid cognitive effects screening approach. Given that patients with epilepsy already report cognitive impairments that may be further compounded by ASDs, this approach could provide valuable early differentiation data.

While the screening of investigational ASDs in naïve rodents certainly provides information on the potential for cognitive effects, identifying such deficits in an epileptic substrate provides a more etiologically-relevant platform for drug differentiation23. It has long been known that kindled rodents are useful to uncover risk for adverse effect liability of an investigational drug36. For example, Löscher and Honack accurately predicted that NMDA-receptor antagonists would induce more severe adverse side effects in patients with epilepsy than in healthy controls using preclinical studies that demonstrated the heightened sensitivity of kindled rats to these compounds relative to naïve animals37. To this end, the evaluation of the effects of prototype ASDs on working memory with the corneal-kindled mouse, a model of chronic hyperexcitability frequently utilized in the drug discovery process20; 38, demonstrates that corneal-kindled mice are capable of performing in a cognitive task. Age-matched and kindled CF-1 mice (approximately 4–6 months old) can sufficiently differentiate a novel object as well as naïve CF-1 mice (approximately 4–6 weeks old; Figure 5 vs Figure 4, p>0.4). However, the observation that corneal-kindled CF-1 mice were not impaired on the NOPR contrasts with prior reports in the object-recognition task using hippocampal- 10; 11; 39 and perirhinal cortex-kindled rats19, which demonstrated deficits in spatial and object recognition memory, respectively. While corneal-kindled CF-1 mice do not exhibit hippocampal lesions29, which underlie the spatial learning deficits noted in the kindled rat11, it is unknown whether corneal kindling induces lesions in perirhinal cortex. Whether corneal-kindled mice also exhibit deficits in spatial memory-dependent tasks (e.g. MWM) and whether such performance is acutely or chronically sensitive to ASDs clearly requires further evaluation20; 38.

Naïve and corneal-kindled CF-1 mice are useful for screening the effects of investigational ASDs on cognitive comorbidities. CBZ and VPA acutely impaired the performance of aged CF-1 mice in NOPR, just as they did in naïve CF-1 mice. Interestingly, administration of either PHT or TPM reduced performance of age-matched mice; an effect that was not detected in young, naïve CF-1 mice (Figure 4). However, no compound produced any impairment in cognitive performance in mice subjected to kindling (e.g. kindled/partially-kindled). ASDs were administered in a Latin-square design so that disease progression and stimulation were controlled for across the study. However, neither dose-response nor chronic dosing paradigms were presently employed to comprehensively evaluate the effect of prototype ASDs in the NOPR task. NOPR with the corneal-kindled mouse may be a platform to evaluate the impact of investigational agents on working memory in a moderate-throughput model of epilepsy, but further studies with dose-response or chronic administration of ASDs may identify subtle adverse effects on working memory in a more clinically-relevant approach23.

Interestingly, CBZ and LEV actually enhanced performance of partially-kindled mice relative to VEH-treated, partially-kindled mice (Figure 5). This may suggest that partially-kindled mice have enhanced neuronal network connectivity in response to mild electrical stimulation, as is observed in electroconvulsive therapy (ECS) for major depression. Specifically, chronic ECS induces calcium channel upregulation and alters channel density40. Calcium channel dysfunction is heavily implicated in bipolar disorder and major depression in human patients. L-type calcium channels are essential to activity-dependent gene expression underlying synaptic plasticity processes41, and in fact, retrograde amnesia induced by ECS can be improved with pre-treatment with calcium-channel blockers (3). VEH-treated, partially-kindled mice were impaired in their performance relative to VEH-treated, naïve mice (p = 0.03) and VEH-treated, age-matched mice, suggesting that repeated electrical stimulation may induce some cognitive impairment that is not observed in fully-kindled mice. CBZ can inhibit calcium channels, contributing to the presently observed performance improvements in partially-kindled mice. LEV can itself reduce IP3-dependent calcium release42, which may be dysfunctional in partially-kindled mice. Calcium channels are coupled to adenosine receptors that are also upregulated in response to ECS40. Thus, it is possible that partially-kindled mice exhibit altered calcium channel or adenosine receptor expression, which may underlie the presently observed improvements in performance of partially-kindled mice acutely treated with CBZ or LEV.

The present results obtained using a multifaceted, standardized approach to screen the cognitive impact of ASDs in rodent models both aligned and contrasted with clinically observed cognitive effects3. Only LTG was without effect in any assay presented herein, which is consistent with its relatively benign profile in patients43. VPA demonstrated dose-dependent effects on performance across multiple models, but the other ASDs with known clinical effects on cognition3 were not identified in every task evaluated (e.g. PB or TPM). Moreover, we demonstrate dose-dependent effects of RTG on the performance of rats in MWM and the distance traveled of mice in NOPR task, however the risk for clinical cognitive adverse effects with RTG is presently inconclusive. The present behavioral battery thus failed to reproduce all clinically-observed cognitive adverse effects with the acutely administered prototype ASDs. Altogether, these results with prototype ASDs, some of which are known to affect human cognition, demonstrate the difficulty faced in predicting risk for clinical cognitive impact based on preclinical data alone. At best, a multidimensional approach in the early evaluation of investigational compounds will suggest the potential for beneficial or adverse effects, which may be clarified when a compound advances to chronic use in heterogeneous patient populations. At worst, significant financial investment will be made to prematurely suggest that an otherwise promising anticonvulsant agent exhibits the potential to induce untoward cognitive effect, which may have never actually materialized in the clinical arena. Indeed, the present results with a battery of approved ASDs demonstrate that it is most likely far easier to develop models and approaches that identify known adverse effects than it is to develop approaches that can accurately predict a priori risk. Future studies using a similar standardized approach with chronic administration of prototype ASDs in these and other models of learning and memory with naïve or chronically-seized rodents may further inform on the risk for cognitive liability in patients with epilepsy, thereby improving drug development efforts to identify transformative ASDs.

Key Points Box.

  • Antiseizure drugs are often associated with cognitive liability in patients with epilepsy.

  • Preclinical approaches to predict the risk for cognitive impact of therapeutically-relevant doses of prototype ASDs in animal models is not presently well defined; strategically filling this gap may inform future clinical development of investigational agents.

  • Morris water maze and novel object/place recognition models of cognition were used to evaluate acute liability of prototype ASDs.

  • Acute administration of high-dose VPA and CBZ consistently impaired performance of naïve mice and rats across multiple cognition models; no treatment affected NOPR performance of corneal-kindled mice.

  • Results in preclinical models both aligned and contrasted with known clinical outcomes, demonstrating the difficulty faced in developing preclinical models of, and approaches to, predicting cognitive adverse effects of ASDs.

Acknowledgments

The authors wish to thank Dr. Geoff G. Murphy, University of Michigan, for helpful preliminary discussions and input on this manuscript. This work was conducted at the University of Utah and supported by NIH HHSN 271201100029C contract to HSW.

Footnotes

Disclosures of Conflicts of Interest:

None of the authors has any conflicts of interest to disclose. 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.

References

  • 1.Brooks-Kayal AR, Bath KG, Berg AT, et al. Issues related to symptomatic and disease-modifying treatments affecting cognitive and neuropsychiatric comorbidities of epilepsy. Epilepsia. 2013;54(Suppl 4):44–60. doi: 10.1111/epi.12298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Bell B, Lin JJ, Seidenberg M, et al. The neurobiology of cognitive disorders in temporal lobe epilepsy. Nat Rev Neurol. 2011;7:154–164. doi: 10.1038/nrneurol.2011.3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Eddy CM, Rickards HE, Cavanna AE. The cognitive impact of antiepileptic drugs. Ther Adv Neurol Disord. 2011;4:385–407. doi: 10.1177/1756285611417920. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Witt JA, Elger CE, Helmstaedter C. Which drug-induced side effects would be tolerated in the prospect of seizure control? Epilepsy Behav. 2013;29:141–143. doi: 10.1016/j.yebeh.2013.07.013. [DOI] [PubMed] [Google Scholar]
  • 5.Devinsky O, Spruill T, Thurman D, et al. Recognizing and preventing epilepsy-related mortality: A call for action. Neurology. 2016;86:779–786. doi: 10.1212/WNL.0000000000002253. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Barker-Haliski ML, Friedman D, French JA, et al. Disease modification in epilepsy: from animal models to clinical applications. Drugs. 2015;75:749–767. doi: 10.1007/s40265-015-0395-9. [DOI] [PubMed] [Google Scholar]
  • 7.Morris R. Developments of a water-maze procedure for studying spatial learning in the rat. J Neurosci Methods. 1984;11:47–60. doi: 10.1016/0165-0270(84)90007-4. [DOI] [PubMed] [Google Scholar]
  • 8.Vorhees CV, Williams MT. Morris water maze: procedures for assessing spatial and related forms of learning and memory. Nat Protoc. 2006;1:848–858. doi: 10.1038/nprot.2006.116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Murphy GG. Spatial Learning and Memory-What’s TLE Got To Do With It? Epilepsy Curr. 2013;13:26–29. doi: 10.5698/1535-7511-13.1.26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Hannesson DK, Mohapel P, Corcoran ME. Dorsal hippocampal kindling selectively impairs spatial learning/short-term memory. Hippocampus. 2001;11:275–286. doi: 10.1002/hipo.1042. [DOI] [PubMed] [Google Scholar]
  • 11.Hannesson DK, Howland J, Pollock M, et al. Dorsal hippocampal kindling produces a selective and enduring disruption of hippocampally mediated behavior. J Neurosci. 2001;21:4443–4450. doi: 10.1523/JNEUROSCI.21-12-04443.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Churchill JD, Fang PC, Voss SE, et al. Some antiepileptic compounds impair learning by rats in a Morris water maze. Integr Physiol Behav Sci. 2003;38:91–103. doi: 10.1007/BF02688828. [DOI] [PubMed] [Google Scholar]
  • 13.Cohen SJ, Stackman RW., Jr Assessing rodent hippocampal involvement in the novel object recognition task. A review. Behav Brain Res. 2015;285:105–117. doi: 10.1016/j.bbr.2014.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Antunes M, Biala G. The novel object recognition memory: neurobiology, test procedure, and its modifications. Cogn Process. 2012;13:93–110. doi: 10.1007/s10339-011-0430-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Grayson B, Leger M, Piercy C, et al. Assessment of disease-related cognitive impairments using the novel object recognition (NOR) task in rodents. Behav Brain Res. 2015;285:176–193. doi: 10.1016/j.bbr.2014.10.025. [DOI] [PubMed] [Google Scholar]
  • 16.Hammond RS, Tull LE, Stackman RW. On the delay-dependent involvement of the hippocampus in object recognition memory. Neurobiol Learn Mem. 2004;82:26–34. doi: 10.1016/j.nlm.2004.03.005. [DOI] [PubMed] [Google Scholar]
  • 17.Bernasconi N, Bernasconi A, Caramanos Z, et al. Morphometric MRI analysis of the parahippocampal region in temporal lobe epilepsy. Ann N Y Acad Sci. 2000;911:495–500. doi: 10.1111/j.1749-6632.2000.tb06752.x. [DOI] [PubMed] [Google Scholar]
  • 18.Bernardi RB, Barros HM. Carbamazepine enhances discriminative memory in a rat model of epilepsy. Epilepsia. 2004;45:1443–1447. doi: 10.1111/j.0013-9580.2004.52403.x. [DOI] [PubMed] [Google Scholar]
  • 19.Hannesson DK, Howland JG, Pollock M, et al. Anterior perirhinal cortex kindling produces long-lasting effects on anxiety and object recognition memory. Eur J Neurosci. 2005;21:1081–1090. doi: 10.1111/j.1460-9568.2005.03938.x. [DOI] [PubMed] [Google Scholar]
  • 20.Rowley NM, White HS. Comparative anticonvulsant efficacy in the corneal kindled mouse model of partial epilepsy: Correlation with other seizure and epilepsy models. Epilepsy Res. 2010;92:163–169. doi: 10.1016/j.eplepsyres.2010.09.002. [DOI] [PubMed] [Google Scholar]
  • 21.Rogawski MA. Diverse mechanisms of antiepileptic drugs in the development pipeline. Epilepsy Res. 2006;69:273–294. doi: 10.1016/j.eplepsyres.2006.02.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Kelley MS, Jacobs MP, Lowenstein DH, et al. The NINDS epilepsy research benchmarks. Epilepsia. 2009;50:579–582. doi: 10.1111/j.1528-1167.2008.01813.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Klitgaard H, Matagne A, Lamberty Y. Use of epileptic animals for adverse effect testing. Epilepsy Res. 2002;50:55–65. doi: 10.1016/s0920-1211(02)00068-2. [DOI] [PubMed] [Google Scholar]
  • 24.Bialer M, Twyman RE, White HS. Correlation analysis between anticonvulsant ED50 values of antiepileptic drugs in mice and rats and their therapeutic doses and plasma levels. Epilepsy Behav. 2004;5:866–872. doi: 10.1016/j.yebeh.2004.08.021. [DOI] [PubMed] [Google Scholar]
  • 25.Barton ME, Klein BD, Wolf HH, et al. Pharmacological characterization of the 6 Hz psychomotor seizure model of partial epilepsy. Epilepsy Res. 2001;47:217–227. doi: 10.1016/s0920-1211(01)00302-3. [DOI] [PubMed] [Google Scholar]
  • 26.Racine RJ. Modification of seizure activity by electrical stimulation: II. Motor seizure. Electroenceph Clin Neurophysiol. 1972;32:281–294. doi: 10.1016/0013-4694(72)90177-0. [DOI] [PubMed] [Google Scholar]
  • 27.Barker GR, Bird F, Alexander V, et al. Recognition memory for objects, place, and temporal order: a disconnection analysis of the role of the medial prefrontal cortex and perirhinal cortex. J Neurosci. 2007;27:2948–2957. doi: 10.1523/JNEUROSCI.5289-06.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Dere E, Huston JP, De Souza Silva MA. Episodic-like memory in mice: simultaneous assessment of object, place and temporal order memory. Brain Res Brain Res Protoc. 2005;16:10–19. doi: 10.1016/j.brainresprot.2005.08.001. [DOI] [PubMed] [Google Scholar]
  • 29.Loewen JL, Barker-Haliski ML, Dahle EJ, et al. Neuronal Injury, Gliosis, and Glial Proliferation in Two Models of Temporal Lobe Epilepsy. J Neuropathol Exp Neurol. 2016 doi: 10.1093/jnen/nlw008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.White HS, Alex AB, Pollock A, et al. A new derivative of valproic acid amide possesses a broad-spectrum antiseizure profile and unique activity against status epilepticus and organophosphate neuronal damage. Epilepsia. 2012;53:134–146. doi: 10.1111/j.1528-1167.2011.03338.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Prut L, Belzung C. The open field as a paradigm to measure the effects of drugs on anxiety-like behaviors: a review. Eur J Pharmacol. 2003;463:3–33. doi: 10.1016/s0014-2999(03)01272-x. [DOI] [PubMed] [Google Scholar]
  • 32.Barker-Haliski ML, Dahle EJ, Heck TD, et al. Evaluating an Etiologically-Relevant Platform for Therapy Development for Temporal Lobe Epilepsy: Effects of Carbamazepine and Valproic Acid on Acute Seizures and Chronic Behavioral Comorbidities in the Theiler’s Murine Encephalomyelitis Virus Mouse Model. J Pharmacol Exp Ther. 2015 doi: 10.1124/jpet.114.222513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Blokland A. Acetylcholine: a neurotransmitter for learning and memory? Brain Res Brain Res Rev. 1995;21:285–300. doi: 10.1016/0165-0173(95)00016-x. [DOI] [PubMed] [Google Scholar]
  • 34.Lamberty Y, Margineanu DG, Klitgaard H. Absence of negative impact of levetiracetam on cognitive function and memory in normal and amygdala-kindled rats. Epilepsy Behav. 2000;1:333–342. doi: 10.1006/ebeh.2000.0098. [DOI] [PubMed] [Google Scholar]
  • 35.Loring DW, Williamson DJ, Meador KJ, et al. Topiramate dose effects on cognition: a randomized double-blind study. Neurology. 2011;76:131–137. doi: 10.1212/WNL.0b013e318206ca02. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Honack D, Loscher W. Kindling increases the sensitivity of rats to adverse effects of certain antiepileptic drugs. Epilepsia. 1995;36:763–771. doi: 10.1111/j.1528-1157.1995.tb01613.x. [DOI] [PubMed] [Google Scholar]
  • 37.Loscher W, Honack D. Anticonvulsant and behavioral effects of two novel competitive N-methyl-D-aspartic acid receptor antagonists, CGP 37849 and CGP 39551, in the kindling model of epilepsy. Comparison with MK-801 and carbamazepine. J Pharmacol Exp Ther. 1991;256:432–440. [PubMed] [Google Scholar]
  • 38.Matagne A, Klitgaard H. Validation of corneally kindled mice: a sensitive screening model for partial epilepsy in man. Epilepsy Res. 1998;31:59–71. doi: 10.1016/s0920-1211(98)00016-3. [DOI] [PubMed] [Google Scholar]
  • 39.Leung LS, Brzozowski D, Shen B. Partial hippocampal kindling affects retention but not acquisition and place but not cue tasks on the radial arm maze. Behav Neurosci. 1996;110:1017–1024. doi: 10.1037//0735-7044.110.5.1017. [DOI] [PubMed] [Google Scholar]
  • 40.Antkiewicz-Michaluk L, Michaluk J, Romanska I, et al. Effect of repetitive electroconvulsive treatment on sensitivity to pain and on [3H]nitrendipine binding sites in cortical and hippocampal membranes. Psychopharmacology (Berl) 1990;101:240–243. doi: 10.1007/BF02244133. [DOI] [PubMed] [Google Scholar]
  • 41.Deisseroth K, Heist EK, Tsien RW. Translocation of calmodulin to the nucleus supports CREB phosphorylation in hippocampal neurons. Nature. 1998;392:198–202. doi: 10.1038/32448. [DOI] [PubMed] [Google Scholar]
  • 42.Nagarkatti N, Deshpande LS, DeLorenzo RJ. Levetiracetam inhibits both ryanodine and IP3 receptor activated calcium induced calcium release in hippocampal neurons in culture. Neurosci Lett. 2008;436:289–293. doi: 10.1016/j.neulet.2008.02.076. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Meador KJ, Loring DW, Allen ME, et al. Comparative cognitive effects of carbamazepine and phenytoin in healthy adults. Neurology. 1991;41:1537–1540. doi: 10.1212/wnl.41.10.1537. [DOI] [PubMed] [Google Scholar]

RESOURCES