Abstracts

Rationale: To determine clinical, environmental and genetic factors related to difficult-to-control epilepsy (DICE).Methods: Cross-sectional, epidemiological, case-control, multicenter study in patients ( 18 years) with focal epilepsy. Demographic, environmental and clinical data, plus DNA samples were collected. Univariate analysis, including a binary logistic regression model, was used to identify important/predictive characteristics of DICE. Goodness-of-fit was evaluated using Hosmer-Lemeshow statistics and accuracy via bootstrapped area under the Receiver Operating Characteristic curve (AUC) with 95%CI.Results: Among 564 patients (mean age 42.9 years; 52.3% women), 48.6% had controlled epilepsy. Using a subsequent multivariate stepwise procedure, five single nucleotide polymorphisms (SNP; CYP3A4, GSTM3, GSTP1, DRD3 and GSTT1) were selected to create an epilepsy risk allele score based on the sum of risk alleles. Lower frequency variant alleles of each polymorphism were associated with controlled epilepsy. On average, inheritance of the risk alleles increased the chance of controlled epilepsy by 32% (odds ratio per allele 1.32; 95%CI 1.13 1.55, p=4.0x10-4). Clinical variables that best predicted the probability of controlled epilepsy were focal seizures with secondary generalization (OR2.32; 95%CI 1.49 3.64; p=2.1x10-4) and age at diagnosis (OR1.04; 95%CI 1.03 1.06; p=3.1x10-08). In a constructed genetic score, the aforementioned genes had a predictive power of approximately 60% (AUC=0.599; cross-validation demonstrated AUC [bootstrap] of 0.56), while inclusion of clinical variables increased the predictive power to 75% (AUC=0.755).Conclusions: CYPP3A4, GSTM3, GSTP1, GSTT1 and DRD3 allele variations, plus focal seizures with secondary generalization and age at diagnosis were most predictive of controlled epilepsy.