PATIENTS AND METHODS

Genomic DNA was obtained from samples of peripheral venous blood from 398 white patients from the United Kingdom with RA (Nuffield Orthopaedic Centre, Oxford, UK) and 289 ethnically matched healthy controls (Oxford Regional Transfusion Centre), and from 63 patients with RA and 93 ethnically matched controls from Uttar Pradesh, northern India. All patients with RA satisfied the 1987 revised American Rheumatism Association criteria.¹⁵

An experimental review of the published methods for typing the 559T/G polymorphism disclosed a high error rate in genotyping (over 10%), mainly due to the existence of the highly homologous FcγRIIIb gene which at position 559 has an invariant G (158V) allele. To ensure unequivocal results, typing was done using two different methods: amplification by polymerase chain reaction (PCR) followed by restriction digestion (PCR–RFLP)¹⁴ and allele specific PCR amplification.¹⁶

Statistical analysis

Allele and genotype frequencies were determined by direct counting. The level of significance for the phenotypic frequencies was determined from 2x2 contingency tables using the χ² statistic, with odds ratios (ORs) calculated from the cross product ratio. Two sided p values were set at the 5% significance level. Genotypic relative risk was determined by the method of Lathrop.¹⁷ In the comparison of the 158GG phenotype versus phenotypes with no or one 158G allele, the χ² statistic was used. This study had 90% statistical power to detect a genotypic relative risk (RR) of 1.6 and a significant allelic association with an OR=1.4 in the white group. The power to detect a significant allelic and genotypic association in the northern Indian group was significantly lower (70% power to detect a significant genotypic association with OR=2.5, and 80% power to detect a significant allelic association with OR=2.0).

RESULTS

The distributions of the FcγRIIIA 158F and 158V alleles between the patients with RA and the controls were similar in both the United Kingdom and Indian groups. The frequencies of the V allele in patients compared with controls were 35% versus 34% in the United Kingdom group and 28% versus 33% in the Indian group. In the Indian group, the frequency of the 158V allele was non-significantly reduced among the patients and the proportion of the 158VV homozygotes was also correspondingly non-significantly reduced (table 1). However, as there have been reports of a gene-dosage effect in this polymorphism, we analysed the RR for 158VV compared with that of 158FV or 158FF. A moderate protective effect from 158VV was found among the northern Indian patients with RA (RR=0.2; p<0.02, 95% confidence interval (95% CI) (0.04 to 0.9); table 2). The genotypic frequencies did not differ significantly in the United Kingdom group, although the proportion of 158VV homozygotes was slightly higher among the patients.

DISCUSSION

Testing polymorphisms in candidate genes across different ethnic groups is potentially a rigorous method for identifying relevant genetic influences. Studies on FcγRIIIA have now been undertaken in RA in white patients from the United Kingdom, and Indian, Spanish, and Japanese patients.^12–14 If similar associations had been found in these different racial groups there would be strong evidence of a causal relationship with the FcγRIIIA gene. However, results from this study do not support an association between the 158VV FcγRIIIA phenotype and RA.¹³ They are in agreement with the negative finding from a smaller study of Japanese patients with RA.¹² A comparison of the published frequencies of this polymorphism among healthy controls in various populations has disclosed significant discrepancies among the published studies, most of which have analysed samples of between 100 and 200 people.^7,8,14 The allelic and genotypic frequencies obtained in the United Kingdom control group in our study differed from the United Kingdom control frequencies published previously by Morgan et al on a smaller sample.¹³ Combining our data with those from Morgan et al (patients with RA 542, controls 544) showed no significant allelic or genotypic association. Thus it seems likely that the earlier positive association in the white United Kingdom group is misleading. The protective effect of the 158VV phenotype found in the Indians with RA in our study reached statistical significance but it is also likely to be spurious as it was not found in the other groups and the sample studied was small.

One possible explanation for the discrepancies in the results reported so far is that one of the other neighbouring FcγR genes (FcγRII or FcγRI) is the true disease susceptibility locus. Incomplete linkage disequilibrium between the 158FV polymorphism and the actual RA predisposing allele could account for different associations in different ethnic groups.

Another significant factor influencing the activity of Fcγ RIIIA is its density on the cell surface, as aggregation of Fcγ receptor triggers cell activation.¹⁸ A study of a mouse model of immune complex mediated arthritis (ICA) has shown an interdependence between the degree of joint inflammation and cartilage destruction and the levels of FcγRIII expression on synovial macrophages.⁵ If FcγRIIIA is a true RA susceptibility factor, then the regions regulating gene expression, rather than polymorphisms of the coding sequence, may harbour the genetic elements for its involvement in the cause of disease. Furthermore, the signalling function of FcγRIIIA is mediated by two closely related intracytoplasmic subunits: the ζ chain of the T cell receptor and the γ chain of the IgE receptor. Comparison of the abilities of these two subunits in mediating activation signals showed that the cross linking of FcγRIIIA associated with a γ chain was significantly more efficient in signal transduction and phagocytosis than signalling through a ζ subunit.^19,20 Genes for the γ and ζ subunits map to the same region of chromosome 1q as the Fcγ receptor cluster. It is conceivable that a potential polymorphism predisposing to RA in one of these two subunits may be reflected by the differential findings of genetic associations between RA and the FcγRIIIA. Further studies of the genes in linkage disequilibrium with this polymorphism in large samples of genetically diverse populations could clarify the involvement of the IgG receptors in the cause of RA.