Evaluation of subchronic administration of antiseizure drugs in spontaneously seizing rats

2.1. Generating a cohort of epileptic rats

Male Sprague Dawley rats (120‐135 g) were obtained from Charles River. Each treatment cohort began with 48‐64 rats (160‐205 g at time of treatment) induced with SE using the repeated low‐dose kainic acid paradigm (Tocris Bioscience, 7.5 mg/kg). ⁷ , ⁸ This results in the development of spontaneous recurrent seizures. Following SE, animals were housed separately in a temperature‐ and humidity‐controlled American Association for Laboratory Animal Science–approved vivarium with a 12‐hour light/dark cycle and free access to food and water. At approximately 10 weeks post‐SE, surviving animals were anesthetized with isoflurane (2%‐5%) and implanted with a TR50B electroencephalographic (EEG) wireless telemetry device (Kaha Telemetry Systems). A transmitter was placed in the intraperitoneal space and secured to the peritoneum with sutures. A trocar was used to separate the skin and peritoneum, and two wires were run underneath the skin to the cranium. The exposed wire tips were placed bilaterally in the epidural space. The wires were held in place using three fixation screws and dental acrylic. Animals were treated with Bicillin (MWI Animal Health) immediately following surgery and Rimadyl (MWI Animal Health, 15 mg/kg) for 2 days following surgery. Following surgery, a week‐long EEG recording period coupled to video recording was evaluated for every animal. For the purposes of evaluating compounds, animals with the highest seizure rates were selected for testing (minimum = 1 seizure per week). Therefore, from the initial surgical cohort, the 24 animals with the highest seizure rates and severity were selected to be used in the experiments. The remaining animals were retained to replace any animals lost to attrition.

2.2. Drug administration

A timeline of the experimental protocol is shown in Figure 1A. Drug treatment, as shown in Figure 1B, proceeded following the selection of the rats with the greatest seizure frequencies. For the first phase of a testing run, animals (n = 10‐12) were administered either drug or vehicle. After a 5‐day treatment period and a 2‐day washout, animals received the opposite treatment for 5 days. This protocol was repeated up to four times in the same cohort, with an optional 1‐week washout period included between tests for drugs with longer half‐lives (ie, 12 hours or more). As some rats do not complete the study due to a variety of factors (eg, telemeter failure, infection, moribund behavior, or death), animals that completed at least 3 days of treatment and observation were included in the final analysis. Animals lost at any point during the treatment paradigm were replaced at the next baseline period from the cohort of unused animals.

2.3. Data analysis

EEG and video data were continuously recorded (24 h/7 d/wk) using a custom software package ¹³ that synchronized the recording of EEG data from an MP150 (BIOPAC Systems) and video streams using a DVP‐7020BE capture card (Advantech). EEG data were first analyzed by a custom automated seizure detection system, and any seizurelike event was marked. A human reviewer then assessed the results of the seizure detection algorithm and confirmed or rejected any detected events as seizures. Additionally, the reviewer scored the behavior on a modified Racine scale. ¹⁴ Reviewers were blinded to treatment groups.

Reviewer results were compiled using custom software that records individual injections and video‐EEG seizure observations. ¹³ This database of seizure scores was analyzed in MATLAB (MathWorks) using custom software that provided seizure rates, seizure burdens, and seizure freedom. Seizure frequency was calculated as per‐day frequency. To calculate seizure burden, the sum of all Racine seizure scores was divided by the total number of testing days. The treatment period was considered completed at the time point of 12 hours following the final injection of either ASD or vehicle. Baseline seizure burden and frequency were defined as the period of exactly 7 days prior to the first injection. Seizure freedom was defined as zero seizures occurring between the first dose of drug and 12 hours following the final injection. Although these criteria do not take pharmacokinetics into account, it remains fixed for every drug as to not bias the results. However, all seizure results are provided to the participant who provided the drug to the ETSP program, so visual evaluation of when seizures occur allows for more detailed analysis to be performed on individual drugs.

Seizure burdens and seizure frequency were compared using a Kruskal‐Wallis test. Fisher exact test was used to compare the rates of seizure freedom between baseline and drug treatment as well as between drug treatment and vehicle treatments. Dose‐response curves for seizure freedom were analyzed using a probit analysis.

2.4. Drug preparation

All drugs, listed in Table 1, were prepared in 0.5% methylcellulose (Sigma), except for sodium valproate (VPA), which was prepared in saline (0.9% NaCl). CBZ, clobazam (CLB), clonazepam (CLZ), ethosuximide (ETX), ezogabine (EZG), phenobarbital (PHB), phenytoin (PHT), and tiagabine were obtained from TCI America). Lamotrigine (LTG) was obtained from AK Scientific. Lacosamide (LCM) was obtained from Axon Medchem. Everolimus (EVR) was obtained from Medchem Express. Perampanel (PER) was obtained directly from the ETSP. Drugs chosen for testing in this model were selected from the cohort of currently available clinical treatments for epilepsy. Doses were chosen based on efficacy and toxicity results from previous standard ¹⁵ , ¹⁶ tests in rats, including the maximal electroshock ED_50, the LTG‐kindled rat (LTG‐R) model, ¹⁶ and the minimal motor impairment toxicity test, as well as from literature references. A complete list of drugs, including their abbreviations, doses, and frequency of administration are included in Table 1, along with literature references that were used to determine an appropriate subchronic dose. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ In cases where the half‐life was shorter than the dosing indicates (ie, LCM, CLB, and CLZ), the secondary metabolites were considered as well. ²⁴ , ²⁵ , ²⁶

3.1. KA‐SE model of spontaneous seizures

To test against spontaneous seizures, we needed to create cohorts of rats with epilepsy. Four cohorts of rats were used in this study. Following KA‐SE, 41 of 48, 40 of 50, 57 of 64, and 40 of 56 animals survived; thus 82% of animals survived the initial treatment. As this model of epilepsy is a progressive disorder, ²⁷ we did not begin testing ASDs until 12 weeks post‐SE. This time point was selected to begin drug testing because previous work has demonstrated that most rats have achieved a fairly stable frequency of spontaneous seizures by this time point. ²⁷ We observed seizures with differing levels of severity, from nonconvulsive (Racine stage 0) to Racine stage 5. The electrographic activity of these seizures is included in Figure S1.

3.2. Evaluation of prototype ASDs

Visual examination of the results of the study allows for easy interpretation of the data. The seizure history of rats treated with four different compounds is shown in Figure 2: PHB (30 mg/kg once daily [qd]), TPM (300 mg/kg twice daily [bid]), gabapentin (GBP; 300 mg/kg bid), and ETX (150 mg/kg bid). These representative ASDs illustrate the range of effects noted in this study. Effects of drugs not shown here are included in Figures S2‐S12. At the doses tested, these compounds were either highly efficacious (eg, reduced seizure burden to at least 20% of baseline; PHB, TPM), moderately efficacious (20%‐80% of baseline seizure burden; GBP), or without efficacy (>80% of baseline seizure burden; ETX; see also Table 2).

Results from evaluation of each drug tested in this model are included in Table 2, sorted by mechanism of action. CBZ was the only sodium channel blocker that reduced seizure burden and increased seizure freedom, whereas LTG, LCM, and PHT had no effect on either seizure burden or seizure freedom. Of all the compounds tested in this study, only LTG resulted in adverse effects over the course of treatment, with animals exhibiting ataxia and lethargy starting on the 3rd day of treatment, requiring one animal's withdrawal from the study due to these effects. Compounds targeting γ‐aminobutyric acid type A receptors (eg, PHB, CLB, and CLZ) were all effective in this model and significantly reduced seizure burden and increased seizure freedom. The potassium channel opener EZG also significantly reduced seizure burden and increased seizure freedom. ASDs targeting calcium channels had mixed effects, with GBP (α2δ‐binding on voltage‐gated calcium channels) reducing seizure burden and increasing seizure freedom, whereas ETX (blockade of T‐type calcium channels) was ineffective. TPM, a compound with multiple mechanisms of action, reduced seizure burden and increased seizure freedom, whereas VPA (mixed mechanism of action) was without effect on seizure burden or seizure freedom. PER (α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid receptor antagonist) and levetiracetam (LEV; SV2A‐acting) moderately reduced seizure burden. Of note, none of the ASDs at the doses used in this study resulted in complete seizure freedom, and several ASDs had no significant effect on seizure burden (EVR, LCM, PHT, LTG, VPA, and ETX).

3.3. Dose‐response evaluation of TPM and PHB

To determine whether this screening approach was sensitive enough to determine the ED₅₀ of tested ASDs, dose‐response studies were performed using TPM and PHB. The initial screening dose of TPM was selected based on a maximum tolerated dose (300 mg/kg bid). ¹⁵ Although this dose was effective in reducing seizures, previously published data suggest it also corresponds to plasma levels well above the therapeutic range. ¹⁷ , ²⁸ Therefore, additional lower doses of TPM (10, 19, 37.5, 75, 150 mg/kg) were evaluated (see Table 3). In general, TPM reduced seizure burden by at least 50%, with a maximal reduction in seizure burden observed at doses of 75 and 100 mg/kg (see Table 3). In addition, TPM treatment increased seizure freedom in a dose‐dependent manner; 10‐19 mg/kg resulted in 33% seizure freedom, 37.5 mg/kg resulted in 67% seizure freedom, and 75‐300 mg/kg resulted in 70%‐88% seizure freedom (see Table 3). A probit analysis of seizure freedom calculated an ED₅₀ of 31.1 mg/kg, with a 95% confidence interval of 1.0‐69.8 mg/kg.

The initial screening dose of PHB was selected based on a maximum tolerated dose (30 mg/kg qd) in naive rats. ¹⁵ Although this dose was well tolerated, it was determined to be the maximum tolerable dose in this assay. Therefore, we also evaluated the effects on seizure burden and seizure freedom using 7.5 mg/kg (qd) and 15 mg/kg (qd). Treatment of rats at 15 mg/kg resulted in a significant reduction of seizure burden, whereas 7.5 mg/kg was without significant effect (Table 3). Furthermore, probit analysis of seizure freedom following PHB treatment calculated an ED₅₀ of 12.0 mg/kg, with a 95% confidence interval of 0.13‐21.7 mg/kg.

4.1. Considerations of the experimental approach

One issue inherent to testing against any model of chronic seizures is the variability of seizure frequency and severity. As demonstrated in Tables 2 and 3, the present studies confirm that seizure frequency following KA‐SE is highly variable. Although seizure variability is consistent with observations in human epilepsy, ⁴³ , ⁴⁴ evaluation of ASDs in this model is challenging and requires sufficiently powered treatment groups. Initial and post hoc analyses suggest that these studies are sufficiently powered to detect differences of at least 80% between drug and vehicle treatments with the sample sizes used. Therefore, differences in seizure burden of <80% (moderate efficacy) should be interpreted carefully. This approach to evaluating novel ASDs also suggests that compounds must produce substantial differences in seizure burden to be considered as potential therapies for pharmacoresistant epilepsy.

In this approach, animals with the highest seizure frequencies in each cohort were selected for inclusion in each study. This bias toward a higher seizure burden was intentional, as the goal of this model is to identify compounds effective against pharmacoresistant epilepsy. The higher frequency of seizures increases the chance for seizures to be captured during the subchronic window in which drugs are administered. In some cases, where animals were lost to attrition during the study, replacement animals were used to maintain a sufficient group size. The replacement animals were those not initially enrolled in the study and therefore had a lower relative seizure rate during the evaluation period. However, there is no specific distinction drawn between “high seizing” and “low seizing” animals. It is a continuous spectrum and varies from cohort to cohort.

A concern of using a 2‐day washout between treatment arms of the study is that it may be too short to avoid any carryover effects of drugs with a long half‐life. This is a legitimate concern, and future studies, depending on known pharmacokinetics of a compound, may allow for more time between treatment arms of an individual study. Furthermore, only four drugs in this study, PHB, LTG, EVR, and ETX, demonstrated a potentially longer elimination half‐life than the 2‐day washout period (ie, five half‐lives being longer than the 60 hours between a drug treatment and a vehicle treatment). Three of those treatments had no effect on seizure incidence or severity (LTG, EVR, and ETX) and therefore did not have an effect of carrying over into the vehicle arm of the study. For PHB, seizures were observed during and immediately following the crossover period (Figure 1A), suggesting that no carryover effect was present following the crossover to the vehicle treatment arm of the study. Furthermore, if the same cohort of animals was used for multiple drug studies, a minimum of 9 days occurred from the last day of drug treatment in the first study to the first day of the next drug treatment arm. Further, a new 7‐day baseline period is collected between drug studies. Thus, in the present study, there was likely very little carryover effect of a compound to either the next treatment arm of an individual study or the next study.

Each cohort of animals used in these studies received up to four different ASDs during the course of the experiments. Previous exposure to ASDs may impact subsequent responses to additional ASDs in a single cohort of animals. Such disease‐modifying changes have previously been demonstrated, and serve as the basis of the LTG‐resistant amygdala kindling model. ³¹ , ⁴⁵ Although ideally each compound would be tested in a new cohort, this would be cost‐prohibitive as part of a screening paradigm. One strategy to address this would be to perform dose‐response studies in naive cohorts of animals once a "hit" on an investigational compound is identified. However, given that patients with refractory epilepsy often receive a battery of different ASDs in the quest for seizure freedom, the approach we have undertaken to identify novel ASDs for refractory epilepsy is therefore translationally relevant.