Abstracts

Rationale: Several previous studies of possible associations between specific genetic variants and variability in antiepileptic drug (AED) response have had conflicting results. This may be due in part to the lack of consistent methods to assess drug response, and the use of dichotomous measures, with no option for an indeterminate response. These problems with classification could have diluted the findings from previous studies. As part of the Epilepsy Phenome/Genome Project (EPGP), a study that will include 1500 sibling pairs with either idiopathic generalized epilepsy (IGE) or non-lesional localization-related epilepsy (LRE), we developed a structured decision approach for classifying AED response on an individual AED drug level, as well as for identifying drug resistant patients. Methods: Expert consensus was used to develop the key nodes for a decision tree approach to characterizing treatment success or failure for individual AED trials. The nodes were duration of therapy (> 3 months, ? 3 months), longest seizure free interval (< 12 months, ? 12 months), fraction of Defined Daily Dose [DDD] achieved (< 50%, ? 50%), number of seizures during titration or maintenance, and occurrence of provoked seizures. AED trials that do not meet the criteria for either success or failure are considered uninformative and will not contribute data to planned genome analyses. Based on response to individual AED trials, patients are characterized either as pharmacoresistant (no AED successes and 2 or more AED failures) or pharmacosensitive (at least 1 AED success and fewer than 2 AED failures). AED response is assessed by retrospective review of available medical records. Results: AED response data are available for 1358 individual AED trials in 419 sib-pair participants with IGE or LRE (average of 3 AED trials per participant). Of these AED trials, 373 (28%) were failures , 172 (13%) were successes and 508 (37%) were uninformative . For 18% of the AED trials, key data elements were missing and response could not be determined. Among all 419 participants, 176 (42%) could be classified as either pharmacoresistant or pharmacosensitive. The remaining 243 participants did not meet criteria for either pharmacoresistance or pharmacosensitivity. Among the 176 classifiable participants, 59 (34%) were pharmacoresistant and 117 (66%) were pharmacosensitive. Conclusions: Using a structured approach to the phenotypic characterization of AED response in EPGP sib-pair participants, we are able to use retrospective data to determine treatment success or failure in 40% of 1358 AED trials. We are also able to characterize patient-level response (pharmacoresistant or pharmacosensitive) in 42% of enrolled subjects. These data may contribute meaningful and unambiguous phenotype information for planned genome analyses.