Do preclinical seizure models predict success or failure in anti‐seizure drug development? A retrospective analysis of 89 compounds with divergent clinical outcomes for epilepsy

Anti-seizure medication (ASM) discovery remains a major challenge despite over 30 commercially available options. One persistent obstacle for any promising candidate compound to successfully move from preclinical study to clinical approval has been the ambiguity in preclinical model selection. Which models are most predictive of clinical response and failure? For decades, the maximal electroshock seizure (MES), subcutaneous pentylenetetrazole (s.c. PTZ) seizure, 6-Hz psychomotor seizure, and audiogenic seizure tests in rodents have served as foundational ASM screening models. These tests are generally regarded as suitable models of generalized tonic-clonic seizures (MES), non-convulsive (absence) seizures (PTZ), pharmacoresistant focal seizures (6-Hz test), and brain penetrance and generalized convulsive seizures (audiogenic seizures). Together, these assays have supported high-throughput ASM discovery for decades; however, they primarily measure acute anti-seizure response. Furthermore, they do not model the pathophysiology of epilepsy as a chronic human brain disorder. Critically, as shown here, these models do not accurately predict human treatment-resistant focal epilepsy success potential. The amygdala-kindled rat model (AKM) is a relevant human temporal lobe epilepsy model that is well suited to investigational ASM discovery and has been implemented for such use for decades. We hypothesized that AKM would be the most predictive of clinical success and failure among the widely implemented rodent screening models. Our present analysis of 89 ASMs and investigational new drugs (INDs) thus aimed to define which preclinical signals, if any, in the standard rodent seizure models would best predict clinical outcomes. The primary efficacy endpoint in AKM has traditionally been convulsive seizure suppression or electrographic outcomes (afterdischarge duration), with few groups reporting IND effects on focal seizures. In our retrospective analysis of 89 compounds, we find that the AKM focal seizure endpoint was the most predictive of clinical success. This structured retrospective comparative analysis of ASM and IND responses demonstrates the exceptional predictive validity of the AKM for early ASM discovery for treatment-resistant human focal epilepsy.