Association of genetic loci

The replication addressed here is of a statistical nature—to be believable, reports of disease gene localizations by means of case-control association studies need to be confirmed (replicated) by independent studies. While this principle has generally worked rather well for genetic linkage analysis applied to Mendelian traits, a troubling number of “significant” associations have been reported that could not be confirmed in later studies. This issue of Neurology contains two articles, one confirming1 and one not confirming2 a previously published article showing a strong association between a single nucleotide polymorphism (SNP) genotype and multidrug resistance in epilepsy. Here is the situation. Last year, Siddiqui et al.3 reported a strong and significant association between the CC genotype at the C3435T polymorphism in the ABCB1 gene and multidrug resistance vs multidrug responsiveness in patients with epilepsy. Sample sizes were 200 resistant and 115 responsive patients, with an associated OR (approximate relative risk) of 2.04. However, with sample sizes close to twice those in the published report and an essentially identical ascertainment scheme, Tan et al.2 in this issue of Neurology are unable to confirm the reported association. On the other hand, Zimprich et al.,1 also in this issue of the journal, with a sample size less than two thirds of that in the original report, confirm the association, although their analysis was somewhat different from those in the other two articles: they worked with a haplotype at three closely spaced SNPs (one of them being the C3435T polymorphism) and defined three instead of just two categories of multidrug resistance. How do we interpret such seemingly contradictory findings? Various authors have addressed the reported replication of—or failure to replicate—statistical gene maps.4–8⇓⇓⇓⇓ Standards for association studies have been developed by many journals including Neurology . …