Linkage disequilibrium (LD) mapping is a useful method for detecting genes conferring susceptibility for common diseases such as rheumatoid arthritis (RA). Recently, LD mapping of the candidate region 5q31, which had been identified as a susceptibility region in other autoimmune diseases, was performed in a Japanese population (830 patients with RA and 658 controls) (1). As a result, significant association (P = 0.000034) between RA and the organic cationic transporter gene SLC22A4 was identified. The best evidence of association was observed with a single-nucleotide polymorphism (SNP) called slc2F1 (rs2073838), which is located in the intron of a Runt-related transcription factor 1 (RUNX1) binding site of SLC22A4. RUNX1 was suspected to regulate the expression of SLC22A4. Furthermore, an SNP called runx1 (rs2268277) in RUNX1 was also strongly associated with RA (P = 0.00035). The odds ratio (OR) for RA in individuals homozygous for slc2F1 versus other genotypes was estimated to be as high as 1.98 (95% confidence interval 1.43–2.75). Considering the magnitude of the effect of the HLA region, which is estimated to account for one-third of the genetic component of RA, this OR indicates that the SLC22A4 polymorphism plays a very important role in the pathogenesis of RA (2).
Although strong association was found in the original study of a Japanese population, Newman et al (3) and Barton et al (4) reported a lack of association in Caucasian populations. As a result, the association of SLC22A4 with RA remains controversial. Replicating findings in independent population samples is helpful in establishing genetic associations. Therefore, in order to validate the original findings, we undertook a large population-based study of these SNPs (slc2F1 and runx1).
The present study is part of an RA cohort project with an enrollment of nearly 4,000 patients, which was established by the Institute of Rheumatology, Tokyo Women's Medical University (5). Approval of this study was granted by the Genome Ethics Committee of Tokyo Women's Medical University. We genotyped 882 patients and 948 controls with the same ethnic backgrounds as those of subjects in the original study. In all patients diagnoses were made using standard methods, and all patients met the American College of Rheumatology (formerly, The American Rheumatism Association) classification criteria for a diagnosis of RA (6). All patients gave written informed consent after receiving a verbal explanation of the study. Control DNA samples were obtained from the Pharma SNP Consortium (http://www.jpma.or.jp/psc/index.html). All control subjects were also recruited in Tokyo and provided informed consent. Among the 882 patients with RA, 87% were rheumatoid factor (RF) positive, and most (81%) were female. Of the 948 controls, 40% were female. After genotyping, using a TaqMan fluorogenic 5′ nuclease assay according to the manufacturer's instructions (Applied Biosystems, Tokyo, Japan), we carried out a population-based association study using a chi-square test for statistical analysis.
Our data did not support association between the SLC22A4 or RUNX1 polymorphisms and RA, because the genotypic or allelic frequencies were not statistically significantly different between patients with RA and controls (Table 1). In addition, a gene–gene interaction between SNPs in SLC22A4 and RUNX1 was untenable in our study, using data from subjects who were genotyped for both SNPs (865 patients and 939 controls) (Table 2). Each marker was present in Hardy-Weinberg equilibrium for both patients and controls.
|SNP||Genotype||χ2||Genotype 11 + 12 vs. genotype 22|
|11||12||22||Total||11||12||22||Total||P||OR (95% CI)|
|Genotype, slc2F1||runx1||slc2F1/runx1, OR (95% CI)|
|Patients (n = 865)||Controls (n = 939)|
|11||138||180||60||156||197||72||referent||1.03 (0.76–1.40)||0.94 (0.52–1.42)|
|12||129||199||56||134||211||76||1.09 (0.78–1.52)||1.07 (0.79–1.44)||0.83 (0.55–1.26)|
|22||34||56||13||23||51||19||1.67 (0.94–2.97)||1.24 (0.80–1.93)||0.77 (0.37–1.62)|
Failure to confirm a previously identified association is not unusual in the search for genetic determinants of common diseases. The failure to confirm a previously determined association might arise due to a lack of statistical power. However, a lack of statistical power in the present study is unlikely, because the sample size was larger than that in the previous study. Based on the original study, it is expected that statistical power of 99.9% would be achieved with 882 patients and 948 controls (P = 0.05, with a 9% frequency of the homozygous risk allele in controls).
Similar negative results were also reported in Caucasian populations. Newman et al performed a case–control study of 918 patients with RA and 623 controls in a Canadian population and observed no association between RA and the SLC22A4 gene or its alleles, despite >99% power to detect the OR of 1.98 conferred by homozygosity for the risk allele (P = 0.88) (3). In a UK population, Barton et al also observed no association between RA and SLC22A4 in 909 patients and 594 controls; that study had enough power to investigate the association (4). In contrast to our study, allele frequencies in the Caucasian populations were quite different from those in the previous Japanese study. Ethnic differences might explain this dissimilar allele distribution and the failure to confirm the association. The SLC22A4 gene may possibly play a role in disease susceptibility only in the presence of ethnic-specific environmental or genetic factors. However, because subjects in the present study had the same ethnic background as those in the original study, population-specific differences in environment or LD between our samples and the original samples seem unlikely to have caused this discrepancy.
The lack of confirmation of the previous association by more powerful population studies, irrespective of ethnic background, casts doubt on the previously identified association between the SLC22A4 polymorphism and RA. However, the importance of the original findings warrants further study of the role of these genes in disease susceptibility.