A functional polymorphism in the SCN1A gene is not associated with carbamazepine dosages in Austrian patients with epilepsy

To the Editors: The voltage gated sodium channel gene SCN1A can be regarded as a very attractive candidate gene for pharmacogenetic investigations in epilepsies. Work by the groups of Sisodiya and Goldstein has particularly implicated one functional polymorphism in this gene (IVS5N+5 G>A, rs3812718), which was suggested to influence the dosage requirements for the antiepileptic drugs carbamazepine and phenytoin in patients with epilepsy. This SNP in a 5′ splice donor site was shown to influence the alternative splicing of exon 5 which codes for one of the functionally important voltage sensors of the channel (Tate et al., 2005; Heinzen et al., 2007). Whereas the ancestral G-allele allows both the "neonatal" and "adult" versions of this exon to be alternatively incorporated into final mRNAs, the mutant A-allele almost abolishes the expression of mRNAs with the neonatal exon 5 (Heinzen et al., 2007). An impact of this SNP on the functioning of the channel appears possible given data from an analogous splicing process in insect sodium channels which dramatically influences pharmacological properties (Song et al., 2004). A large pharmacogenetic study on 425 British patients with epilepsy even suggested a clinical relevance as the A-allele was found to be linearly correlated with a need for higher carbamazepine and phenytoin dosages (Tate et al., 2005). A second study on 71 Chinese patients detected a similar association between phenytoin serum levels and genotypes (Tate et al., 2006). We attempted to further investigate the association of this SNP with the carbamazepine dosing in a cohort of 369 Austrian patients with focal epilepsy syndromes. Eighty-nine percent of our patients displayed a medically refractory disease course. We retrospectively assessed the carbamazepine maintenance doses which patients were prescribed during routine clinical visits. As an internal control, we showed that this dose significantly correlated with the severity of the epilepsy (i.e., the average seizure frequency). This confirmed that the prescribing pattern was sensitive to the clinical needs of the patients and that our retrospective study was sensitive enough to detect this trend in spite of a large confounding interindividual variability. The SCN1A IVS5N+5 genotype frequencies and the average carbamazepine dosages are shown in Table 1. We did not find a significant difference in the average carbamazepine doses between the genotype groups. As the highest doses were observed in the GG-homozygote group, we did not even find a trend in the same direction as reported by Tate et al. (2005). A power analysis (ANOVA, 3 groups) predicted a power of our study of 0.79 at an alpha of 0.05 to detect an effect size f equal to the study of Tate et al. (2005) of 0.16. Several reasons for the failure to replicate the initial results are conceivable. First, the true effect could be real but its size was overestimated in the original study. Although our study should have been adequately powered to detect an effect size of a similar magnitude, a smaller effect could well have been missed. Second, subtle differences in the phenotypes between the cohorts or in the prescription practice between the centers could be responsible. Third, population-specific effects could account for the observed discrepancy. The relevance of such population-specific gene effects in epilepsies has only recently been substantiated by the results of a large association study with multiple candidate genes (Cavalleri et al., 2007). To resolve this issue, it will probably be necessary to perform more association studies with even larger patient numbers and to await meta-analyses on this subject before prematurely accepting or rejecting this interesting SNP as a pharmacogenetically relevant factor in epilepsies.