Abstracts

Rationale: Growing doubts about the models used in anti-epileptic drug testing (White, Neurol. 2002; 59(suppl 5):S7-S14) have spurred development of etiologically realistic models of acquired epilepsies. These models are expected to provide superior prediction of clinical anti-epileptic effect while capturing both the chronic spontaneous recurrent seizures (CSRSs) that define epilepsy and the localization-related seizures that are most resistant to antiepileptic treatment. However, the substantial costs incurred to quantify infrequent chronic, spontaneous and often infrequent seizures (animal preparation, housing, data acquisition and analysis) call for efficient protocols for the use of such models in drug testing. We employed a bootstrapped nonparametric power analysis to evaluate the performance of several protocols and then used the best one to assess the effects of carbamazepine (CBZ) and valproate (VPA) on neocortical CSRSs after rostral parasaggital fluid percussion injury (rpFPI) in the rat. Methods: Seizure frequency data from epileptic rats (2.43±0.63 events/h; N=33) were used in bootstrapped analyses to estimate the statistical power of rpFPI-based test protocols to detect clinically interesting anti-epileptic effects. Frequencies were modeled on individual rats' observed frequency ranges so that both between- and within-subject variability were based on experimental data. Independent-groups and repeated-measures designs were analyzed, and effects of non-responders and measurement-to-measurement variability were assessed for n=5-32. The anti-epileptic effect of CBZ or VPA was tested on neocortical CSRSs at 4-5 weeks post-injury. Severe rpFPI (3.5 atm) was administered to 31-36 days male Sprague Dawley rats and 5 epidural electrodes were implanted 3 weeks later. Video/ECoG recordings (72hr/week) were obtained before and during continuous infusion of VPA (34±3 μg /ml plasma) and CBZ (7.8±1.2 μg/ml plasma) on post-injury weeks 4 and 5, respectively. Results: Power analyses showed that the best protocol consisted in a logarithmic transformation of seizure frequency data obtained from groups of 8 rats examined in a repeated measures design. Eight rats provide ≥0.8 power to detect a 70% decrease in seizure frequency in a group with 25% non-responders. With this protocol, we found the frequency of seizure was diminished by VPA, but not CBZ. The mean logarithm of the frequency of neocortical seizures in CBZ-treated rats was -0.09±0.31 prior to and -0.03±0.31 during treatment with CBZ. The mean logarithm of seizure frequency was decreased from -0.20±0.22 before, to -0.57±0.29 (p=0.036) during treatment with VPA. Conclusions: CSRSs induced by severe rpFPI, and detected by ECoG, provide good power (>0.8) to detect anti-epileptic effects with small groups of rats (n=8) even in the presence of 25% non-responders. FPI-induced neocortical CSRSs were refractory to treatment with CBZ, but were significantly decreased during VPA treatment. These data support the use of rpFPI epilepsy in anti-epileptic drug discovery. Supported by NIH NS053928 (RD).