In rheumatoid arthritis (RA) and juvenile idiopatic arthritis (JIA), the synovium is infiltrated with T cells and macrophages, and chemokine receptors and their ligands are involved in regulation of the inflammation 12. Chemokines regulate lymphocyte trafficking by inducing cell motility and activating adhesion molecules. The chemokine receptor CCR5 is a G-protein-coupled seven-transmembrane-receptor, and it is expressed on monocytes, macrophages, memory T-cells and dendritic cells, where it serves as a receptor for RANTES, MIP-1-α, MIP-1-β and MCP 3. CCR5 has functionally been implicated in the pathogenesis of RA 2 and JIA 4. The expression of CCR5 is increased on synovial T cells in patients with JIA 4 or RA 5.

The gene encoding CCR5 is located on chomosome 3p21.3 6. A 32 base pair deletion (CCR5Δ32) in the CCR5 gene leads to a frame shift and a non-functional receptor. Homozygosity for CCR5Δ32 results in a complete lack of surface expression of CCR5 7. CCR5 happens to be a co-receptor for HIV-1 on CD4+ T-cells with the consequence that CCR5Δ32 homozygotes are protected against infections with HIV-1, while heterozygotes have a delay in the progression towards AIDS 8. The CCR5 deletion is common among Caucasians, especially in Nordic populations 9, but has not been found in native Africans, American Indians or East Indian ethnic groups.

The role of CCR5Δ32 has been investigated genetically in several immune-mediated diseases. In RA, a negative association of the deletion has been reported with divergent results 1011121314, while in childhood arthritis a weak negative association has been observed in a single study 15. Regarding clinical outcome, it has been reported that carriers of the CCR5Δ32 allele occurred at a significantly higher frequency among patients with non-severe compared to severe disease (N = 160) 14, whereas a later and larger study of 438 patients did not find any association between carriage of the CCR5Δ32 and milder disease severity in inflammatory polyarthritis 16. A recent meta-analysis of five reports addressing the association of CCR5Δ32 with susceptibility to develop RA in patients of European ancestry (1790 cases and 2717 controls), reported a significant negative association and estimated the deletion to provide disease resistance with an OR = 0.65 (95% CI [0.55–0.77])17.

The aim of our study was to investigate if there is an association between CCR5Δ32 and RA in the largest individual data set studied so far, as well as to address whether CCR5Δ32 is also associated with JIA, not yet extensively investigated.

853 RA patients, 524 JIA patients and 658 controls were included in the study. The RA patients, admitted to the Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway, fulfilled the American College of Rheumatology (ACR) criteria 18 and the JIA patients, admitted to the Department of Rheumatology, Rikshospitalet-Radiumhospitalet Medical Center, Oslo, Norway, were classified according to the ILAR (International League of Associations for Rheumatology) criteria 19. The control group was recruited through the Norwegian Bone Marrow Registry. All patients and controls were of Norwegian origin. The study was approved by the Norwegian Regional Ethics committee, and the participants had given their informed consent.

Whole-genome amplification was performed on genomic DNA using the Genomiphi^® kit (GE healthcare systems, Chalfont St. Giles, UK). Amplification of fragment surrounding the deletion was performed with the dye labelled forward primer 5'-ATCACTTGGGTGGTGGCTGTGTTTGCGTCTC and the reverse primer 5'-AGTAGCAGATGACCATGACAAGCAGCGGCAG under the following conditions; denaturation at 94°C for 15 seconds, annealing and extension at 70°C for 15 seconds, 30 cycles and final extension at 70°C for 5 min at a Perkin Elmer 9700 thermal cycler (Applied Biosystems, Foster City, CA, USA). Fragment length was determined on an ABI3730 DNA Analyzer (Applied Biosystems) with subsequent allele calling in GeneMapper v3.7 where a 193 bp PCR product corresponded to the wild type allele and a 161 bp product represented the deletion. Part of the control material was genotyped in our previous publication 20.

Statistical power was calculated prior to the study using the power calculator at 21, assuming the frequency of 10% for the CCR5Δ32 and an Odds Ratio of 0.56 (obtained from Zapico et al 14). The power (at α = 0.05) was calculated to be 84% for the RA and 75% for the JIA data set. Allele and genotype frequencies in patients and controls were compared by the Chi-Square test, or Fischer's exact test when appropriate, using the Public Domain Software for Epidemiology and Disease Surveillance EPI Info Version 5 (Center of Disease Control, Epidemiology Program Office, Atlanta, GA, USA).

The genotype and allele frequencies of CCR5Δ32 are summarised in Table 1. The genotyping success rate was >95% for both the patients and control populations, and all groups were in Hardy-Weinberg equilibrium. No significant negative association was observed with the Δ32 allele in either RA or JIA (Table 1), and only a negligible deviation in frequency was seen between the patients and controls. Furthermore, no decreased frequency for homozygosity of Δ32 was evident among the patients (0.9% in RA, 0.6% in JIA and 1.0% in controls). A revised meta-analysis, calculated by adding our results to those from the recently published meta-analysis of pooled RA cases and controls 17, reduced the initial published risk estimate from OR = 0.65; 95% CI [0.55–0.77], p < 0.0001, to OR = 0.83; 95% CI [0.73–0.95], p = 0.0068.

Our results, from the largest cohort of RA patients and controls investigated to date, do not support the existence of a negative association between CCR5Δ32 and disease (OR = 0.90; 95% CI [0.71–1.14]). Neither was the proposed disease resistance provided by homozygosity of the deletion sustained in our data set. The frequency of CCR5Δ32 in our control population is in the same range as that observed in other European populations and is highly similar to that reported in a Danish population 11. CCR5Δ32 has been found to have its highest prevalence among Nordic populations 9, which from a statistical power perspective should favour detection of a negative association in our Norwegian cohort, on the other hand deviating results between populations for disease associated alleles is not uncommon, e.g. for PADI4 22232425.

Altogether, two studies suggest an association between CCR5Δ32 and RA 1314, while three studies 101112, in addition to ours, do not reveal statistical significant evidence to support this. The number of patients included in the positive studies was 160 and 516 whereas in the negative studies 278, 163, 673 and 853 patients were analysed. A weak effect of CCR5Δ32 on the disease risk, combined with inadequate power, could explain the contradictory results. Alternatively, there could be true differences between the populations in the effect exercised by CCR5; i.e. due to genetic or allelic heterogeneity, variable influence of environmental factors, gene-gene interaction, or variable linkage disequilibrium patterns. All studies display a deviation in the direction of a lower frequency for the deletion among RA patients, but in our data set only a 0.9% reduction in frequency was seen in RA patients compared to controls. The results of a recent meta-analysis of pooled cases and controls were in favour of a negative association (OR = 0.65 (95% CI [0.55–0.77])). 17. We recalculated the power of our RA study post hoc based on the odds ratio for the CCR5Δ32 obtained in the recent meta-analysis 17, and given this risk effect our power was reduced to 63%. When our RA data set was included in a revised meta-analysis, the overall risk estimate was greatly reduced (OR = 0.83; 95% CI [0.73–0.95, p = 0.0068]). Furthermore, if the true risk estimate is in the range of that calculated after also including our data into the pooled meta-analysis, one would need almost 4000 cases and an equal amount of controls to replicate the finding with an 80% power. Notably, the CCR5Δ32 meta-analysis may also have been prone to type 1 error caused by e.g. publication bias or heterogeneity between the populations included in the combined analysis 26.

Our data does not reveal an association in JIA. After the initiation of our study, the possible association of CCR5 in JIA was addressed, and the authors reported a negative association between CCR5Δ32 and childhood arthritis, particularly in the early onset group and in those diagnosed with pauciarticular subtype 15. This finding was not sustained in our data set as no association was revealed when stratifying for either JIA subgroups, including the oligoarthirits patients (comparable to the pauciarticular group defined by the ACR criterias; OR = 0.80, 95% CI [0.54–1.17], p = 0.23), or early disease onset (<6 years; OR = 0.81, 95% CI [0.55–1.19], p = 0.27). Further investigations are warranted for JIA before any firm conclusions can be drawn. Nevertheless, our study does not suggest that CCR5Δ32 influences the risk of JIA, and if present the effect is likely to be small.

CCR5 has functionally been implicated in RA, and our study does not formally exclude CCR5 as a candidate gene for RA and JIA. Polymorphisms located elsewhere in the gene could potentially be predisposing, and if the associations observed for Δ32 is only secondary due to linkage disequilibrium, that could be an explanation for the discrepant results obtained by different studies. Interestingly, in the recently reported childhood arthritis study 15 other polymorphisms in the CCR5 gene were also investigated, and an additional association with the CCR5 -1835T allele was also observed. No strong linkage disequilibrium between CCR5Δ32 and -1835T was evident, but the haplotype lacking both these alleles displayed positive association, which could suggest that the true etiological variant of CCR5 in RA or JIA pathogenesis that is still unidentified.

In conclusion, whether the CCR5 gene is involved in RA or JIA susceptibility remains unclear, however, our study suggests that if a protective effect is exercised by CCR5Δ32, it is minor and less than predicted from the meta-analysis of previous publications17.

The author(s) declare that they have no competing interests.

EL and GBNN participated in the design of the study, performed the genotyping, statistical analysis and drafted the manuscript. EM participated in the design of the study and performed the genotyping of a part of the control material. BF, AMS and ØTF contributed with DNA samples and clinical information for the JIA patients. TKK contributed with DNA samples and clinical information for the RA patients. ET helped writing the manuscript. BAL participated in the design of the study, statistical analysis and drafted the manuscript. All authors have read and approved the final manuscript.