Abstracts

Rationale: A number of studies investigating the association between specific genetic variants and variability in AED response have revealed conflicting results. In part, this may be due to the lack of a universal methodology to assess drug response. For example, prior classification schemes have not dealt with patients in whom response could not be determined. As part of the Epilepsy Phenome Genome Project (EPGP), a study of 3750 epilepsy patients, we developed a decision tree methodology for classifying AED response on an individual drug level, as well as identifying “drug resistant” patients. This scheme is currently in use. Methods: An expert consensus approach was used to develop the key nodes for a decision tree approach. The nodes were duration of therapy (> 3 months, ≤ 3months), 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. Results: Pharmacogenetic outcomes are either: AED success; AED failure; uninformative AED trial; or, refer for adjudication. This decision tree approach is used for each AED taken by every patient enrolled in EPGP. These pharmacogenetic phenotypes are to be used in future genetic association studies of AED response. Requirements to define a specific AED trial as a “success” include: > 3 months on that AED, ≥ 12 months seizure free on that AED without adding another AED, and a history of ≥ 3 seizures in a 6 month interval before attaining 50% of DDD. In contrast, requirements to define a specific AED trial as a “failure” include: > 3 months on that AED, < 12 months seizure free on that AED, achieving ≥ 50% of DDD and a history of > 3 seizures in a 3-12 month interval on maintenance dose without provoked seizures. AED trials that did not meet criteria for “success” or “failure” were categorized as “uninformative” and these AED trials will not contribute efficacy data to genetic association studies. Conclusions: This decision tree approach allows for unambiguous classification of each AED trial for every patient enrolled in EPGP. It is anticipated that most of the trials will be informative and provide sufficient well characterized response data to facilitate meaningful pharmacogenetic association studies.