Ranking of antiseizure medications in a panel of focal seizure models predicts their comparative efficacy in clinical add‐on trials in drug‐resistant focal epilepsy

Abstract Objective Most antiseizure medications (ASMs) have been discovered by testing in animal models, which are generally thought to predict antiseizure activity in patients. However, it is not known whether any of these models (or a combination of models) can predict whether a novel ASM exhibits higher clinical efficacy in focal drug‐resistant epilepsy (DRE) than benchmark ASMs such as carbamazepine or levetiracetam. This important question has plagued epilepsy drug discovery for over three decades and serves as the basis of the present analysis. Methods The present analysis includes 23 ASMs that the U.S. Food and Drug Administration (FDA) approved for the treatment of focal‐onset seizures. For assessing comparative preclinical activity, we used the median effective dose (ED 50 ) determined in six rodent models of induced generalized or focal‐onset seizures. For assessing comparative clinical efficacy, seizure‐freedom rates recorded during add‐on treatment in randomized controlled clinical trials in patients with focal DRE were used. The preclinical ED 50s were ranked from most potent to least potent by SUCRA ( surface under the cumulative ranking curve ). Correlation analysis of preclinical and clinical data was used to determine the predictive validity of the animal models. Results Except for the pilocarpine model, none of the seizure models predicted the comparative efficacy of add‐on treatment with ASMs in focal DRE. However, when we combined the ED 50 ‐based ranks from two to four models of focal‐onset seizures, highly significant positive correlations with seizure‐freedom rates in the clinical trials were obtained. Significance The present study shows that ranking of ASMs across a panel of mechanistically complementary but agnostic models of focal‐onset seizures yields greater translational accuracy and clinical predictivity for patients with focal DRE than rankings based on any single model. This novel finding can be used when defining a path forward that best identifies the most promising investigational compounds to advance to more expensive and time‐consuming clinical investigations.