A functional variant in the Fc receptor–like gene, FCRL3, has recently been reported to be associated with rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and autoimmune thyroid disease (AITD) in a collection of well-powered Japanese case–control studies (1). FCRL3 encodes a glycoprotein that is a member of the immunoglobulin receptor superfamily, and although its precise function remains unknown, it contains immunoreceptor tyrosine–based activation and inhibition motifs in its cytoplasmic domain, suggesting that it plays a role in immune cell regulation. The disease-associated functional polymorphism, −169T→C (rs7528684), has been shown to alter the binding affinity of NF-κB. Furthermore, the disease susceptibility allele, −169C, is positively correlated with higher levels of FCRL3 expression in B cells both in vitro and in vivo, as well as increased titers of rheumatoid factor (RF) and anti–citrullinated peptide (anti-CCP) antibodies in RA patients (1). These results suggest that a specific abnormality in the humoral arm of the immune system may play a role in susceptibility to RA, SLE, and AITD.
To determine whether the −169T→C variant in FCRL3 is associated with RA in white North Americans, we genotyped this single-nucleotide polymorphism (SNP) in 2 independent RA sample sets. The first RA sample set, obtained by Genomics Collaborative (Cambridge, MA), consisted of 475 white RF-positive RA patients and 475 individually matched North American controls (matched for sex, age [±5 years], and grandparental country of origin) (2). The second sample set, obtained by the North American Rheumatoid Arthritis Consortium (NARAC, based at North Shore University Hospital, Manhasset, NY) consisted of 573 unrelated RA patients (both RF-positive and RF-negative) from white multiplex families (3) collected from across North America, and 745 randomly selected healthy white control individuals from the New York Cancer Project (NYCP; URL: http://www.nycponline.org). All RA cases met the 1987 American College of Rheumatology (formerly, the American Rheumatism Association) diagnostic criteria (4). National and/or local institutional review boards approved all protocols and recruitment sites, and informed written consent was obtained from all subjects.
Genotyping of the −169T→C SNP in the first RA sample set was performed with conventional polymerase chain reaction (primers 5′-CACACAGTCAAGGTGTCAA-3′ and 5′-CGCTTGCTGATTTATCTCCTAA-3′) followed by a flow cytometry–based oligonucleotide ligation assay (major allele probe 5′-ATACAAATGTACAGATCAA-3′; minor allele probe 5′-ATACAAATGTACAGATCAG-3′, common probe 5′-phosphate-GGACTTCCCGTAATC-biotin-3′) (5). Genotypes were automatically called using custom software followed by hand curation. Genotyping in the second RA sample set was performed using the MassArray matrix-assisted laser desorption ionization–time-of-flight mass spectrometry–based SNP genotyping system, according to the protocols recommended by the manufacturer (Sequenom, San Diego, CA). Dideoxy-ACG terminator mix and the following primers were used: forward 5′ACGTTGGATGCAGATCTGGGTGAGATTACG3′, reverse 5′ACGTTGGATGTTATGAGGCTTCTGAACAGG3′, extension 5′-GAGATTACGGGAAGTCC3′. Genotypes were automatically called with SpectroCaller software (Sequenom). All nonconservative calls were checked and accepted, recalled, or rejected after manually evaluating the spectra.
Genotype frequencies for this SNP were in Hardy-Weinberg equilibrium in all case and control sample sets (P = 0.458 in sample set 1 controls and P = 0.925 in sample set 1 cases; P = 0.239 in sample set 2 controls and P = 0.672 in sample set 2 cases by an exact test). The distribution of genotypes in cases and controls is shown in Table 1. The −169C allele was more frequent in the white North American controls (frequency 0.448 in sample set 1, frequency 0.468 in sample set 2; Table 1) than in the Japanese controls (0.37 in 2,037 normal individuals ), which is consistent with the 0.45 frequency reported for 100 Americans of European descent (1). There was no significant evidence of association between the −169T→C SNP and RA in either sample set by allelic (P = 0.52 in sample set 1 and P = 0.55 in sample set 2, by chi-square test) or genotypic (for CC vs CT + TT; P = 0.38 in sample set 1 and P = 0.25 in sample set 2, by chi-square test) tests.
|Sample set 1||Sample set 2|
|Sample size, cases/controls*||467/473||565/743|
|Allelic χ2P value||0.52||0.55|
|Genotypic association, CC vs. CT + TT|
|95% confidence interval||0.63–1.20||0.65–1.12|
Stratification of the patients by carriage of the HLA–DRB1 shared epitope (SE) also revealed no evidence of association. The frequency of the −169C allele was not significantly different in SE-positive cases relative to controls (P = 0.52 in sample set 1 and P = 0.74 in sample set 2 by chi-square test) nor between SE-positive and SE-negative patients (P = 0.42 in sample set 1 and P = 0.49 in sample set 2 by chi-square test). We also examined whether there was a correlation between the −169T→C genotype and autoantibody titers in sample set 2 (a similar analysis could not be performed in sample set 1 because there was no available information on CCP status and RF titers). We found no evidence for association of the −169T→C genotype with either CCP or RF levels. Among patients with the TT genotype, median levels of anti-CCP (n = 119) and RF (n = 163) were 99 units/ml and 90 units/ml, respectively; among those with the TC genotype, median levels of anti-CCP (n = 218) and RF (n = 283) were 90 units/ml and 76 units/ml, respectively; and among those with the CC genotype, median levels of anti-CCP (n = 91) and RF (n = 114) were 94 units/ml and 98 units/ml, respectively. (P = 0.87 for anti-CCP and P = 0.58 for RF by Kruskal-Wallis one-way analysis of variance). Similarly, differences in anti-CCP and RF levels were not found when the analyses were limited to the anti-CCP–positive and the RF-positive patients (P = 0.68 for anti-CCP and P = 0.998 for RF). Anti-CCP–positive and RFG-positive patients had autoantibody levels greater than the upper limit of the reference range (20 units/ml for anti-CCP and 12 units/ml for RF).
To confirm these negative results, we genotyped an adjacent FCRL3 SNP (−110A→G; rs11264799), which is in strong linkage disequilibrium with −169T→C in Japanese subjects and also showed strong association with RA (1) in our second RA sample set (data not shown). No significant evidence of allelic association was observed (P = 0.49, by chi-square test).
Our data suggest that the previously identified FCRL3 −169C allele is not a disease susceptibility allele for RA in white North Americans. There are a number of possible explanations for our failure to replicate association with this allele. First, these findings may be the result of a Type II error; however, we believe this is highly unlikely, given that our 2 studies had 94.6% (sample set 1) and 98.8% (sample set 2) power to detect association based on the genotypic relative risks (GRR) calculated for this SNP from the Japanese data (GRR = 1.14 for the heterozygous genotype and 1.84 for the homozygous risk genotype ). The power of our study was calculated using the method of Purcell et al (6), assuming a 1% RA prevalence in white North Americans and a 0.46 allele frequency. The Type I error was set at 0.05.
Second, it is possible that this variant of FCRL3 is associated with disease severity, rather than susceptibility, and that the Japanese patients have a more severe form of RA than the patients in our North American sample sets. Again, we believe this is unlikely given that both our sample sets, in particular sample set 2 (unrelated patients from multiplex families selected for the presence of erosive disease), were selected for severe disease. In addition, stratification on the basis of markers of severity, RF positivity, and the number of copies of the HLA-DRB1 SE showed no evidence of association.
A third explanation is that this allele plays a role in susceptibility to RA only in the presence of certain environmental triggers, such as diet or a specific infectious agent, which are unique to Japan and not found in North America. Alternatively, the −169T→C SNP may not be the causal SNP, but could be in linkage disequilibrium with the true disease-causing SNP on a haplotype that is found in the Japanese population but not in white North Americans.
Finally, although we cannot absolutely exclude the possibility that our negative findings are the result of population heterogeneity, we believe this is unlikely, given that every attempt was made to individually match the controls to the patients based on age, sex, and grandparental country of origin, and that similar results were found in 2 independent sample sets. A family-based association approach would provide additional valuable confirmation; NARAC investigators are in the process of collecting up to 1,000 trio families for this purpose.
Failure to replicate association of specific SNPs between different ethnic groups is not surprising. Different risks of autoimmune diseases among various ethnic groups (7) have been reported, even when these groups reside in the same geographic region (8). This suggests that in addition to common genetic risk factors, such as the HLA– DRB1 SE, there may be distinct genetic factors involved in the RA disease initiation process in different ethnic groups. Examples of ethnic-specific genetic risk factors already exist. The R620W PTPN22 missense SNP has been consistently associated with susceptibility to RA in multiple studies of whites of European descent (for review, see ref. 9); however, the W620 susceptibility allele has not been detected in Asians (2, 10). Additionally, a functional haplotype of the PADI4 gene, which is a member of a family of peptidyl arginine deiminases and encodes an enzyme responsible for the posttranslational conversion of arginine to citrulline, is strongly associated with RA in the Japanese population (11) but not in British (12), French (13) or Spanish (14) whites.
In summary, the FCRL3 −169C allele does not appear to play a role in susceptibility to RA in our white North American sample sets. Future large-scale association studies across different ethnic groups will be useful to illustrate the mechanisms by which different genetic and environmental factors contribute to complex diseases in subjects with distinct ethnic backgrounds. These studies will also serve as powerful resources for understanding the history of various complex diseases.
Collection of the NARAC sample set has been funded by a National Arthritis Foundation grant and by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) and the National Institute of Allergy and Infectious Diseases, NIH. This research was supported in part by the Intramural Research Program of NIAMS, and with funds provided by the National Center for Research Resources to the General Clinical Research Center, Moffitt Hospital, University of California, San Francisco. The authors are grateful to the RA patients, the control subjects, and the collaborating clinicians for participation in this study. We would also like to thank the Celera Diagnostics High Throughput facility and Julie Le at the NIAMS Genetics and Genomics Branch for their invaluable help.