Abstracts

Rationale: A number of drug metabolising enzymes (DMEs), including cytochrome P450s (CYPs), uridine diphosphate-glucuronosyltransferases (UGTs) and microsomal epoxide hydrolase (EPHX1), contribute to the inactivation of carbamazepine (CBZ). Single nucleotide polymorphisms (SNPs) in genes encoding these enzymes may alter their expression or activity and could, theoretically, affect serum concentrations of CBZ and individual dose requirement. We have assessed whether genetic variation across six major DMEs involved in CBZ metabolism is associated with maintenance doses of the drug when employed as monotherapy in people with newly-diagnosed epilepsy. Methods: A total of 161 individuals with newly diagnosed epilepsy were included in the analysis. Each had been seizure-free for at least 12 months on a stable dose of CBZ as monotherapy. A total of 91 tagging and known functional SNPs across six DMEs (CYP1A2, CYP2C8, CYP3A4, CYP3A5, EPHX1 and UGT2B7) were genotyped on a Sequenom MassARRAY iPLEX platform. Associations between maintenance dose of CBZ, demographic variables (age, gender, epilepsy type), and genotype of individual SNPs were identified using ANOVA and linear and stepwise regression analysis, as appropriate. Haplotype analysis was also performed, using the solid spine of linkage disequilibrium method to define blocks and PHASE software (version 2.1) to infer haplotypes. Correction for multiple comparisons was performed by false discovery rate (FDR). Results: Univariate analysis identified patient age and the genotype of seven individual SNPs as being independently associated with the maintenance dose of CBZ (all p<0.05; prior to correction for multiple testing). Three of these SNPs (rs4646450, CYP3A5, p=0.002; rs4149229, EPHX1, p=0.010; rs3924194, UGT2B7 p=0.014) remained significantly associated with dose after stepwise regression, with a significant overall model fit (adjusted r²=13.2%, p_residual=0.000015). Multivariate regression modelling of haplotype blocks, with age included as a covariate, identified four haplotypes that were associated with CBZ dose. Of these, only the six-SNP block 1 haplotype (ACGCCG) remained significantly associated with dose (p=0.00025, r²=9.7%) after stepwise regression. Conclusions: These findings suggest that variability in genes encoding CYP3A5, EPHX1 and UGTB7 may contribute to CBZ dose requirement in people with newly diagnosed epilepsy. Haplotype analysis better explains dose variability than single SNP analysis, as it is more representative of variation across entire genes and less likely to exclude any unidentified causal variants. However, even if these data were replicated in an independent cohort, it is unlikely, given the modest size of the effect reported here, that genetic variability in DMEs accounts for a sizeable proportion of inter-individual differences in CBZ dose requirement.