Standard dose Valproic acid does not cause additional cognitive impact in a rodent model of intractable epilepsy

Abstract

Children with epilepsy face significant cognitive and behavioral impairments. These impairments are due to a poorly characterized interaction between the underlying etiology, the effect of seizures and the effect of medication. The large variation in these factors make understanding the main drivers of cognitive impairment in humans extremely difficult. Therefore we investigated the cognitive effect of seizures and the antiepileptic drug valproic acid in a rodent model of cortical dysplasia. Rats were divided into seizure-receiving and non-receiving groups. Rats experienced frequent early life seizures using the flurothyl inhalation method: 50 seizures between postnatal day 5 and 15 and then one seizure a day following that. Rats were further divided into drug-treated and vehicle treated groups. Valproic acid treated animals were treated from 5 days preceding behavioral testing in the Morris water maze at a clinically relevant concentration. We show here that the main driver of cognitive impairments are the brain malformations, and that persistent seizures in animals with brain malformations and valproic acid caused no additional impact. These findings suggest that neither an appropriate dose of a standard antiepileptic drug or intractable seizures worsen cognition associated with a malformation of cortical development and that alternative treatment strategies to improve cognition are required.

Introduction

The epilepsies are amongst the most common serious neurological disorders in childhood and are associated with significant morbidity particularly in the domains of cognition and behavior [1]. The reasons for cognitive impairments in children with epilepsy are likely to include a complex interaction between the underlying etiology, the effect of seizures and the effect of antiepileptic drugs. Treatment strategies for children with epilepsy aim to create a balance between maximal seizure reduction, and minimal cognitive and behavioral impairment [2]. To achieve this goal it is important to understand the relative effects of each potential mechanism so that the most appropriate therapeutic strategies can be devised. However, in humans this is extremely difficult given the enormous variability in types and severity of both etiologies and seizures, and the variability in the number and types of antiepileptic drugs taken. Therefore, we took an animal model approach to addressing this clinically important issue using a standardized model of cortical dysplasia, the induction of a set number of tonic-clonic seizures and the use of a widely used antiepileptic drug: valproic acid (VPA). A clearer understanding of the drivers of cognitive impairment has the potential to lead to improvements in the care of children with epilepsy.

We chose to model malformations of cortical development (MCDs) as they are identified in approximately 25% of children with intractable epilepsy [3–4] and approximately 75% of children with MCDs are thought to have epilepsy [5]. MCDs are abnormalities of the cerebral cortex resulting from environmental or genetic disruption to neural plate formation. This causes disruption to neural circuitry and predisposes towards clinical symptoms, the most common of which are seizures and cognitive impairments [6]. We have previously shown in a model of MCD that seizures during development have little long term impact on cognitive outcomes suggesting that the most important driver of cognitive outcome is etiology [4]. However, that experiment did not address the additional cognitive impact of medically intractable seizures or of standard dose treatment with a single antiepileptic medication which is the recommended clinical approach.

Around 30% of children with epilepsy taking antiepileptic drugs have unsatisfactory seizure control [2,7] and even if children become seizure free, they can remain on AEDs for years. However, it remains uncertain whether antiepileptic drugs either in patients with continuing seizures or in those who become seizure free have significant cognitive impairments that can be attributed to the medication. Valproic acid (VPA) is broad-spectrum anti-epileptic drug which is a first-line treatment option for many childhood epilepsy syndromes [8]. Since VPA is so widely used, it is important to test this drug in an animal model of MCDs to make sure it doesn’t cause any differing cognitive effects compared to animals without brain abnormalities.

Here we show that the main driver of spatial cognition is the MCD and that there is no additional negative cognitive impact of VPA in either of the groups. However, the dose of VPA tested does not influence seizure susceptibility in the rats with MCD, even though it does in rats with normal brain structure. Continuing seizures have a negative impact on animals with no pre-existing brain abnormality, but no effect in those with MCD. Therefore, treatment with a single antiepileptic drug is unlikely to cause significant cognitive side-effects, cognitive outcomes are not heavily influenced by ongoing seizures and strategies that target etiology to improve outcomes should be developed.

Materials and Methods

Animals

All animal procedures were approved by the Institutional Animal Care and Use Committee at Dartmouth College, in accordance with National Institutes of Health guidelines, under United States Department of Agriculture and Association for the Assessment and Accreditation of Laboratory Animal Care International-approved conditions. Sprague-Dawley rats were housed with a 12 hour light/dark cycle and ad libitum access to food and water.

2 pregnant dam rats were injected with 20mg/kg intraperitoneally (i.p.) of Methylazoxymethanol acetate (MAM) at embryonic day 17 [4] to produce 2 MAM litters totaling 19 animals. 2 dams were injected with saline at E17 producing 2 control litters totaling 20 animals.

Valproic acid (VPA) injections commenced 5 days prior to, and until the end of the spatial memory testing in the Morris water maze in a random sample of half the animals in each group. Rats were injected with 250mg/kg (i.p.) of VPA at 12 hour intervals. This dose was based on previous experiments done in our laboratory [9]. Non-VPA injected control rats were injected with saline in the same manner.

Flurothyl-induced Seizures

Half of the rats in each of the control and MAM groups underwent Early-Life Seizures (ELS) using the bis-2,2,2-tri uoroethyl ether (flurothyl) inhalation method previously described in our laboratory [4,10,11,12]. They experienced 5 seizures a day 60 minutes apart from postnatal day 6–15, and then one seizure a day until they had completed the water maze. This was modeling intractable seizures in patients taking a single antiepileptic drug. Rats were sealed in an airtight plastic chamber, and liquid flurothyl was injected into the chamber onto a strip of absorbent paper, where it evaporated and filled the chamber. Rats were taken out of the chamber when they exhibited tonic extensions of the upper and lower limbs. Before P16, rats were in a chamber with a diameter of 13 cm and a height of 15 cm. After P16, rats were put into a 13 cm by 13 cm by 25 cm chamber to accommodate their increased size. Animals that did not receive seizures were removed from the Dam for the same length of time as those that had seizures.

Seizure Latency

Rats were held two at a time (one VPA-injected and one saline-injected) in the induction chamber (13 cm by 13 cm by 25 cm) and 0.05 ml of flurothyl was injected into the chamber per minute. Rats were taken out of the chamber when they started to exhibit tonic extensions of the upper and lower limbs. If one rat seized before another, it was quickly taken out of the chamber and the lid replaced. Rats were allowed to recover before being put back into their cages.

Water maze

Rats were tested in the water maze during adolescence (between p33 and p45). The starting age was when the animals reached 100g in weight to ensure that there was no size disadvantage during testing.

Testing took place in a circular tank with a diameter of 2 meters and a height of 50 cm. It was filled with water containing non-toxic white paint. A transparent platform was placed in the tank at a fixed location 1.5cm below the water’s surface. Surrounding the tank were black curtains with 2 fixed external cues visible from the water surface allowing the rat to orientate itself in space. 4 entry points were marked out to divide the tank into equal quadrants.

Rats were allowed to swim freely for 120 seconds without the platform being present to allow them to habituate to the environment. Before timed trials commenced, rats were held onto the platform for thirty seconds. Each rat completed 4 trials over 4 days, with 4 entries to the tank (randomized for each day) per trial. After finding the platform, rats were allowed a thirty second rest period before starting the next entry. If a rat failed to find the platform in 120 seconds, it was picked up and placed onto the platform to allow the 30 second rest period. The trials were recorded onto DVD using a camera directly overhead the tank. AnyMaze (version 4.63) software was used to track and analyze the tests, where several parameters could be measured- including escape latency, swim speed, rotations of the rat’s body, and path efficiency to find the platform.

Histology

Rats were deeply anaesthetized with isoflurane until they stopped breathing and perfused with 120ml of cold saline, and then 60ml of 4% paraformaldehyle (PFA). Rats were decapitated and brains were extracted from the skull and fixed in 4% PFA overnight in a brain vial; the PFA was then replaced with 30% sucrose for cryoprotection. The brains were considered ready for staining when they had sunk to the bottom of the vial.

Brains were sliced coronally at 50um sections in a cryostat (Leica CM3050, Leica Microsystems) across the dorsal and ventral hippocampi, ready for thionin staining. Brain slices were hydrated in decreasing concentrations of ethanol, submerged in 0.1% thionin solution (pH4) for 75 seconds, and then dehydrated in increasing concentrations of ethanol. Slices were then immersed in xylene for 10 minutes and coversliped.

Statistics

Evaluation of Morris water maze data was performed using a multivariable Cox regression time to event approach in STATA Intercooled (10.0; StataCorp), applied with frailty models to deal with repeated measures. The event was defined as the rat finding the platform and the time to event was defined as the latency to find the platform. The independent predictor variables tested included day of testing (days 1, 2, 3 or 4), MAM administration (yes or no), seizures (yes or no), swimming speed (continuous variable), age at start of testing and interaction terms. Latency to seizure measures were evaluated using a Student’s t test. Generalized estimating equation multivariable regression methods were used to evaluate path efficiency as these methods allow the assumptions of the most appropriate distribution and allow adjustment for repeated measures.

Results

20 Control animals and 19 MAM animals were included in this study. Table 1 shows the division of the rats into the 4 groups: Control, no seizures (C−), MAM, no seizures (M−), control with seizures (C+) and MAM with seizures (M+). VPA and saline injections took place twice a day (250mg/kg i.p.) from 5 days prior to water maze testing up until the end of water maze testing. Seizure groups received early life seizures (50 seizures between P5 and P15) and then a seizure a day after that until completion of testing in the water maze.

Table 1.

Groups and number of animals used

Group Name	Description of Group	VPA-injected rats	Saline-injected rats
C−	No seizures	5	5
C+	Control animals with ELS+ (50 seizures between P5 and P15 and then a seizure a day afterwards)	5	5
M−	MAM animals which did not have seizures	6	6
M+	MAM animals with ELS+ (50 seizures between P5 and P15 and then a seizure a day afterwards)	3	4

The probability of finding the platform in the Morris water maze was significantly predicted by the main effects; day of testing, whether the animal was from an MCD group and swimming speed. Compared to the first day of testing animals were likely to find the platform more quickly on Day 2 (Hazard Ratio [HR] 2.2±0.3; p<0.001), Day 3 (HR 3.2±0.44; p<0.001) and on Day (HR 4.2±0.58; p<0.001) with progressive improvement observed over days. There was also a strong relationship between probability of finding the platform and swimming speed (p<0.001). The administration of VPA did not influence performance in the water maze (p=0.75). In addition, the magnitude of any effect was similar across groups (Group*Treatment effect; p=0.71). There was a significant MAM*Seizure interaction (p=0.015). To evaluate this further we investigated the effect of seizures in separate analyses for MAM animals and for control animals (Figure 1). Animals with previously normal brains and seizures were less likely to find the platform than controls (HR 0.48±0.13; p=0.006). MAM animals with seizures did not show a deficit when compared to MAM animals with no seizures (HR 1.02±0.21; p=0.88). It is of interest to note that the animals that took the longest to reach 100g were the animals with previously normal brains and seizures and despite being the oldest at initial testing were impaired when compared to controls.

Figure 1.

Survival curves showing the effect of group. The Morris water maze was carried out during adolescence (postnatal days 33–37) once the rats reached a weight of 100g. The survival (proportion to reach the platform) is shown against time to reach the platform in seconds. Control animals with seizures were less likely to find the platform compared to the control animals without seizures. MAM animals with seizures did not significantly differ in their latency to find the platform compared to the MAM animals without seizures.

Path efficiency is an index comparing the actual path of the animal to a straight line (actual path length/direct path length). There was an improvement over days in all groups confirming that animals improve in the maze. On Day 1 path efficiency was 0.11±0.02. Efficiency was similar on Day 2 (0.14±0.02; p=0.016), but improved on Day 3 (0.21±0.02; p<0.001) and on Day 4 (0.23±0.01; p<0.001) (Figure 2). After adjustment for day of testing the path efficiency was 0.22±0.01 in animals with previously normal brains and 0.13±0.02 in animals exposed to MAM (p<0.001). Path efficiency was 0.2±0.01 in animals without seizures and 0.15 in animals with seizures (p=0.02). There was also a significant Group*Seizure interaction (p=0.01). The effect of the seizures was significant in animals not exposed to MAM (0.27±0.02 with no seizures compared to 0.17±0.01 in animals with seizures), but not in animals with malformations of cortical development (0.13±0.02 with seizures and 0.13±0.01 without seizures). There was no overall effect of VPA on path efficiency (p = 0.71). In addition the magnitude of any effect was similar across groups (Group*Treatment interaction; p=0.86) suggesting that there was no particular group that was susceptible to negative effects of VPA.

Figure 2.

Water maze path efficiency data. Path efficiency measures indicating how direct the animal route to the platform was: values closest to 1 imply a more direct route to the platform. Control animals were significantly different from C+ and M− animals. Non-significance between M+ animals is likely due to the small amount of animals in the group.

We next investigated whether VPA had a differing impact on seizure susceptibility depending on whether the animal had an MCD. One VPA-treated and one saline-treated rat was put in the induction chamber at one time. This meant that atmospheric conditions were the same for each of the treatment groups. Flurothyl was injected into the chamber onto a piece of filter paper where it vaporized. MAM animals treated with saline had a higher latency to seizure than control animals (p<0.001). In control rats VPA caused an increase in the latency to seize in control animals (Figure 3A), however, VPA had no effect on the latency to seize of MAM rats (Figure 3B).

Figure 3.

Average seizure latencies for C+ and M+ animals. Results are averages from measurements over a 6 day period (rats experienced one seizure per day). In measuring seizure latency, rats were sealed in a 13 cm by 13 cm by 25 cm chamber and 0.05ml of flurothyl per minute was injected onto absorbent paper (where it vaporised). Rats were taken out of the chamber once they had exhibited tonic extensions of their upper and lower body. (A) Control rats (receiving ELS and then a seizure a day). Significant differences were seen between VPA and saline injected control rats (P<0.001). (B) MAM rats (receiving ELS and then a seizure a day). No significant differences were seen in MAM animals in the latency to seize between VPA and saline injected rats.

At the end of the experiment histology was performed. We confirmed the characteristic hippocampal abnormalities previously described [4] (See Figure 4). On visual assessment there was no effect of VPA on hippocampal anatomy.

Figure 4.

Thionin staining of control and MAM hippocampi. (A) Normal hippocampus (B) MAM hippocampus showing dispersion of the pyramidal cell layer.

Discussion

The main findings in the current study are that brain malformations are a major driver of cognitive impairments, and that in the context of brain malformations persistent seizures have no additional impact. In the control animals ongoing seizures are associated with some impairment in spatial cognition although this is not as marked as the effect of the brain malformation. Importantly, VPA as a single medication at standard dose did not cause any cognitive impairment in either group.

These results suggest that recurrent seizures in the context of a severe etiology have little impact on cognition and aiming to prevent seizures at all cost may not be the most appropriate management strategy. It is often recommended that patients with epilepsy are treated with a single antiepileptic drug. As VPA had no impact on cognition this is a reasonable approach from a cognitive standpoint, although VPA may not be an effective agent in the context on MCD. It is likely that the cognitive data are generalizable to other AEDs (particularly newer ones which are known to have fewer cognitive side-effects).

Previous studies using the MAM model have suggested that current therapeutic strategies emphasizing antiepileptic drugs for children with MCDs are inadequate and that treatment should try to reflect the underlying neuropathology [4,13]. However these studies only investigated the long term impact of seizures occurring during a short period of development, and did not investigate the role of intractable seizures and of therapy on cognitive outcomes. Our current study therefore extends our previous studies in a clinically relevant way, and supports the view that recurrent daily epileptic seizures have little additional cognitive impact in the context of an MCD. There were only 2 MAM and 2 control litters in this study raising the issue of whether any of the findings are simply a function of litter. However, the results showing the effect of MAM and the effect of early life seizures in the context of a normal brain are entirely consistent with previous reports suggesting that any litter effect in the current study is minimal. Our results therefore suggest that the current views on how seizures impact cognition may not be entirely relevant in all clinical settings. Similar studies evaluating the impact of seizures on outcomes related to other (e.g. genetic and metabolic) etiologies are also likely to inform appropriate clinical practice.

A potential confound in our study relates to the age disparity in the animals at the time of testing. In controls it is clear that preweanling rats perform more poorly in the water maze when compared to adult rats. In the current study we are comparing performance across a much smaller age range with the youngest animal being P33 and the oldest P42 at the time that testing began. The question is whether a 10 day period in these animals which are all at least 12 days post-weaning could be markedly influencing our results. A previous report comparing P28 animals to adult animals shows that there is a difference between the groups but at no stage is the magnitude of the effect large [18]. In contrast, there is a dramatic change in ability between P17 and P28. As we stated in the manuscript the control animals that had early life seizures and were impaired when compared to those with no seizures were also older at the time of water maze testing. This suggests that our conclusion about the impact of seizures may be conservative but is not invalidated by the age at testing. We have previously tested MAM animals in the water maze at P25 and P45 to evaluate developmental effects [4]. In our previous study there was no difference in ability between these age groups. It is therefore unlikely that there would be differences in rats aged between P33 and P42. Given this we felt that it was important to have all rats reach a weight of at least 100g prior to testing to ensure that there was no size disadvantage for any of the rats and conclude that the potential developmental disparity is unlikely to be significantly impacting the conclusions of the study.

To evaluate the independent impact of AEDs on cognition we needed to isolate the effects of the seizures from the effects of the drug. Therefore, all VPA and saline-injected rats received the same number of seizures in their respective groups. This meant that rats being treated with VPA were forced to seize even when they were being treated with the AED. The rats were also treated with the VPA for a minority of the seizures they were receiving (approximately 12%), meaning any negative effects seen would be purely down to the actions of the drug. Our results are reassuring as VPA administration in this controlled experiment was safe from a cognition perspective. However, the measures we took to isolate the effects of the seizures from the effects of the drug did limit potential positive effects the drug would have on the rats- the damage from the seizures may already have been done before VPA was administered. We looked purely at whether a MAM brain which had experienced seizure activity exposed to VPA would have cognitive impairments compared to one which had not been exposed to VPA. Nevertheless, the finding that the seizures seemed to have little cognitive impact minimizes this concern.

To show the anticonvulsant effect that VPA had on our animals we measured their latency to seize (Figure 3). As expected, there was a significant difference in seizure latency (P<0.01) between VPA and saline-injected control animals. However, we saw no significant difference in seizure latency between VPA and saline-injected rats in MAM animals. This confirms results seen by Smyth et al. (2002) [13] which showed MAM rats to be pharmacoresistant to AEDs. This is a further similarity between MAM rats and children with intractable epilepsy.

It is thought that at least 14% of epilepsy patients are on polytherapy [16], and chronic AED polytherapy is very common in patients with intractable epilepsy. However, multiple AEDs have been associated with larger cognitive impairments than with one alone [17]. Even though polytherapy is largely not recommended, further studies could add a second AED to mimic treatment used in a small number of children with epilepsy.

In order to try and make a more clinically relevant early-life seizure model, we modified the standard flurothyl model in which animals are only exposed to seizures during early postnatal life. However, patients with MCDs are unlikely to stop having seizures spontaneously and therefore we judge our approach of giving daily seizures up to the time of testing in the water maze is more clinically relevant. We have shown early life seizures to affect MAM and control rats differently. In control rats recurrent seizures caused a mild cognitive impairment in the water maze. In MAM rats however, there was no significant difference in water maze performance between those receiving seizures and those that do not. This outcome is similar to results seen by Lucas et al. (2011) [4] using an ELS model with MAM animals, again questioning to what extent AED therapy is useful for minimizing cognitive impairments in patients with intractable epilepsy.

In conclusion this study has found VPA to cause no harmful cognitive effects on rats with a model of intractable epilepsy. VPA is widely used so this is a positive outcome, however no cognitive improvements were seen either. Thus, VPA is unlikely to be causing an improved outcome in cognition and behavior which are major morbidities that patients with intractable epilepsy face every day, even if seizures have been successfully treated (as modeled by the MAM animals with no seizures). It is therefore important that the mechanisms of cognitive impairment, beyond the impact of seizures, are investigated so that more successful treatments for these important morbidities can be established.

Highlights.

• We modelled intractable epilepsy through MAM treatment of prenatal rodents

• Brain malformations were the main driver of cognitive impairment in these rodents

• Ongoing seizures and/or VPA caused no additional cognitive impact in these rodents

• Other strategies to treat cognition in childhood intractable epilepsy are needed