Introduction

Polymorphisms in CTLA-4, the gene encoding cytotoxic T lymphocyte-associated antigen-4, have been widely studied in connection with genetic susceptibility to various autoimmune diseases [1], but studies have led to contradictory results in different populations.

Among CTLA-4 gene polymorphisms, a G to A transition at position 49 (+49A/G) of exon 1 leads to an alanine to threonine amino acid substitution at codon 17 in the leader peptide (A17T), and a C to T transition at position 60 (CT60) is located within the 3'-untranslated region [2]. The G allele of +49A/G has been associated with a predisposition to many autoimmune diseases (reviewed in [1]). Both polymorphisms are in linkage disequilibrium, which warrants haplotype analysis in studies of CTLA-4 polymorphisms. The CT60 G allele has been reported to increase susceptibility to several autoimmune diseases [2], and a functional approach provided evidence for lower mRNA levels associated with the CT60 G allele [2].

Downie-Doyle and colleagues have recently reported a significant association of the CTLA-4 +49A/G*A allele and of the CTLA-4 +49A/G*A allele carrier haplotypes with primary Sjögren syndrome (pSS), especially in patients with anti-SSA or anti-SSB antibodies, in a study including 111 Australian patients with pSS and 156 controls [3].

The aim of our study was to investigate in a large case-control study whether CTLA-4 CT60 and/or +49A/G SNPs were involved in genetic predisposition to pSS in French patients.

Materials and methods

Patients

A first cohort of 142 unrelated patients with pSS diagnosed in accordance with the European American consensus group criteria [4] (37% without autoantibodies, 30% with anti-SSA antibodies only, and 33% with both anti-SSA and anti-SSB antibodies) and 241 healthy blood donors were genotyped for CTLA-4 CT60 and +49A/G polymorphisms. A second independent cohort of 139 patients with pSS was further genotyped for CTLA-4 +49A/G polymorphisms in a replication study. The geographical origin and the clinico-biological characteristics of the patients in this second cohort were the same as those in the first. In this second cohort, 27% of patients were anti-SSA and anti-SSB negative, 35% had anti-SSA only and 38% had both anti-SSA and anti-SSB. All patients and controls were Caucasians and provided informed consent.

Genotyping

After the isolation of genomic DNA from peripheral blood mononuclear cells, CTLA-4 CT60 and +49A/G polymorphisms were genotyped by restriction fragment length polymorphism with the use of BbvI (+49A/G) and NlaIII (CT60).

Statistical analysis

Allelic and genotypic frequencies of CTLA-4 CT60 and +49A/G polymorphisms were compared between patients and controls by using a two-sided χ2 test. All genotyped SNPs were in Hardy–Weinberg equilibrium. CTLA-4 (+49A/G or CT60) haplotypes, constructed with the PHASE program, were also examined for association with pSS. P < 0.05 was considered significant.

Results

In the first cohort of patients with pSS, the A allele of the CTLA-4 +49A/G polymorphism was found on 73% of chromosomes in patients with pSS, in comparison with 66% in controls (p = 0.036, odds ratio (OR) 1.41, 95% confidence interval (CI) 1.02 to 1.95; Table 1). No significant difference in CTLA-4 +49A/G*A allele frequencies was observed among subgroups of patients according to their anti-SSB and/or anti-SSA status (Table 1). No difference in CTLA-4 CT60 allelic or genotypic distribution was observed between patients (n = 142) and controls (n = 241). CTLA-4 (+49A/G or CT60) haplotype distribution mirrored the CTLA-4 +49A/G*A allele excess among patients with pSS (A/A 48%, A/G 26%, G/G 26%, G/A 0.4%; in comparison with A/A 45%, A/G 21%, G/G 34% among controls), leading to an excess of +49A/G*A allele carrier haplotypes among patients (p = 0.03, OR 1.41, 95% CI 1.02 to 1.95).

Table 1 Allellic frequencies of CTLA-4 49A/G polymorphism among patient controls
Full size table

To avoid the possibility of a false positive association of CTLA-4 +49A/G*A with pSS as a result of the somewhat small sample size of our first cohort, and because the CTLA-4 +49A/G*A allele has been only marginally associated with autoimmune diseases compared with the CTLA-4 +49A/G*G allele [1], we performed a replication study on a second independent cohort of 139 patients with pSS. In this second cohort, the CTLA-4 +49A/G*A allele was found on 62% of chromosomes in patients with pSS, compared with 66% in controls (p = 0.30; OR 0.85, 95% CI 0.63 to 1.16; Table 1). Thus, the CTLA-4 +49A/G*A allele excess among patients with pSS from the first cohort was counterbalanced by its under-representation in the second cohort. When the results from the patients in both cohorts were pooled (n = 281), there was no difference in CTLA-4 +49A/G polymorphism allelic or genotypic distribution in comparison with controls (p = 0.53, OR 1.09, 95% CI 0.84 to 1.4; Table 1). The sex ratios among patients (0.97) and controls (0.06) were different. We therefore investigated CTLA-4 +49A/G polymorphism genotypic distribution among males and females in the control group and found that it was not statistically different (p = 0.1), thus excluding any possible gender effect.

Our results therefore demonstrate a lack of association between CTLA-4 CT60 or +49A/G polymorphisms and pSS among Caucasians.

Discussion

The results from our first cohort were very close to those from the study of Downie-Doyle and colleagues [3], with a significant association of pSS with the +49A/G*A allele and with the +49A/G*A allele carriers haplotypes. The association observed in the first cohort, of two haplotypes bearing the same allele (CTLA-4 +49A/G*A), was actually more probably due to the statistical weight of the CTLA-4 +49A/G*A allele than to a true functional effect of two different haplotypes, bearing either CTLA-4 CT60*C or CTLA-4 CT60*T alleles, each having opposite functional effects on CTLA-4 mRNA expression [2].

In fact, our results suggest a false positive association of CTLA-4 +49A/G*A allele with pSS in the first cohort of patients. When data were pooled (cohorts 1 and 2), no significant association was found with the CTLA-4 +49A/G polymorphism in our Caucasian population of patients with pSS. This was not the consequence of different origin or different clinico-biological characteristics of the patients from the two cohorts and could only be the result of a sampling bias. Indeed, the findings observed in our first cohort of patients, as those from Downie-Doyle and colleagues [3], were unexpected because there are only rare examples of association of the CTLA-4A/G*A allele with autoimmune diseases [5–7]. Consequently, we might have made premature conclusions if a replication study had not been performed.

Conclusion

Our study illustrates the necessity to include a large number of patients in genetic case-control studies. In fact, sampling bias may partly account for some contradictory results previously reported for CTLA-4 association studies in autoimmune diseases.