Rheumatoid arthritis (RA) pathogenesis is multifactorial, involving both genetic and environmental factors. Although the association of some HLA-DRB1 alleles with RA was reported nearly three decades ago, the underlying biological mechanism of this association remains unknown. The presence of the RAA sequence at positions 72–74 of the HLA-DR β chain molecule, for all HLA-DRB1 alleles known to be associated with RA, led to the shared epitope (SE) hypothesis.1 The modelisation of the SE component in RA has recently been successful.2 3

Rheumatoid factor (RF) production in RA is generally associated with a more severe phenotype of the disease. RF production has been associated with carriage of SE alleles. Furthermore, several publications showed that the presence of anti-cyclic citrullinated peptide (anti-CCP) antibodies together with SE allele carriage was associated with a very high relative risk for future development of RA.4 5 These reports finally led to the hypothesis that SE alleles contribute to the RA only by the development of anti-CCP antibodies and do not independently constitute a risk factor of RA.6^–8 Moreover, the combination of the PTPN22 1858T variant, another genetic factor RF-positive RA-associated and RA-linked, and anti-CCP antibodies was recently reported to give a much higher relative risk for developing RA than the combination of the PTPN22 1858T variant and HLA-SE.9

Numerous environmental risk factors have been studied in RA, but tobacco smoking is to date the only well-established environmental risk factor.10 A gene–environment interaction between tobacco smoking and SE alleles was reported to increase the risk of RF-positive RA.11 12 More recently, a model has been proposed in which smoking, in the context of SE alleles may trigger specific immune reactions to citrullinated proteins.13 However, smoking seems to be a risk factor for anti-CCP antibody production restricted to patients with RA who carry SE alleles.14 No association was reported between PTPN22 1858T variant and tobacco exposure in RA.

Here, we tested the hypothesis of an association between genes (HLA-DRB1 and PTPN22) and tobacco smoking, separately as well as combined, and serological markers of RA in a French population with familial and sporadic RA.

PATIENTS AND METHODS

Patients

We studied 274 patients with RA of French Caucasian origin as defined for each of the four grand-parents, 196 patients being index cases of trio families, ie, one patient with RA and both parents and 78 patients being index cases of affected sibling pair families. The diagnosis of RA fulfilled the 1987 American College of Rheumatology (formerly, the American Rheumatism Association) criteria.15 All individuals provided informed written consent and the study was approved by the Hôpital Bicêtre ethics committee (Kremlin-Bicêtre, Assistance Publique-Hôpitaux de Paris).

In this sample of patients with RA, 89% were females, 47% belonged to multicase families with RA, and at the time of serum collection, the mean age of RA onset was 34 (SD 11) years and the mean disease duration was 12 (SD 9) years. Erosions were present in 82% of the patients and rheumatoid nodules in 19%.

Genotyping

Blood samples were collected for DNA extraction and genotyping. HLA-DRB1 typing was performed with the polymerase chain reaction sequence specific primer method using the Dynal Classic SSP DR low resolution and the Dynal Classic high resolution SSP for subtyping of HLA-DRB1*01, *04, *11, *13 and *15 alleles (Dynal Biotech, Lake Success, NY, USA). SE alleles were HLA-DRB1*0101, *0102, *0401, *0404, *0405, *0408 and *1001. Alleles were then classified into three groups, S₂, S_3P and L, according to their 70–74 amino acid sequence, as previously reported.2 3 Genotyping of the PTPN22-1858 T/C variant was performed by polymerase chain reaction–restriction fragment length polymorphism, as previously reported.16 This T variant eliminated a restriction site for RsaI enzyme, and genotypes were secondly checked by a polymerase chain reaction–restriction fragment length polymorphism using the XcmI enzyme for which a restriction site was created when the T allele was present.

Autoantibody status

IgM RF was provided by ELISA method (QUANTA, Lite RF IgM, INOVA diagnostics, San Diego, CA, USA), and the anti-CCP status was provided by an anti-CCP antibody ELISA (Immunoscan RA, Euro-Diagnostic, Malmö, Sweden). Both ELISA tests were performed on the same serum sample according to the manufacturer’s instructions.

Tobacco exposure

Information about tobacco exposure was collected by using a questionnaire sent to each patient with RA.11 The following questions about smoking were asked: (1) Do you smoke? (2) If not, did you ever smoke? (3) In which year did you start smoking? (4) In which year did you stop smoking? (5) What sort of tobacco did you smoke: cigarettes, cigars, pipes? (6) Average number of cigarettes smoked per day: 1–5, 6–9, 10–19, ⩾20 cigarettes? (7) Duration in years of smoking exposure: <10, 10–19, ⩾20 years? Unanswered questionnaires were completed by telephone. Patients with RA who were smokers at the year of the serum collection were considered as current smokers, those who reported that they were smokers and stopped smoking before the year of serum collection were defined as ex-smokers. Current smokers and ex-smokers were defined as ever smokers, and patients with RA who reported they had never smoked were defined as never smokers. The cumulative dose of cigarettes smoked was then expressed as pack-year, one pack-year being the equivalent of 20 cigarettes smoked per day for 1 year.

Statistical analysis

Statistical analysis to search for association between exposure to genetic factors (SE alleles and/or PTPN22 1858T allele) and/or tobacco smoking with RF or anti-CCP positivity, relied on χ² or Fisher’s exact test when appropriate, odds ratio (OR) and 95% confidence interval (95% CI) with 2×2 or 2×3 tables. In order to search for biological interaction, we used the departure-from-additivity-based method. Attributable proportion (AP) due to interaction was determined by Ap = (R₁₁– R₁₀– R₀₁+R₀₀)/R₁₁; AP and 95% CI were calculated with the SYSTAT program.17 18 The correlation study between the cumulative dose of cigarette smoking expressed in pack-years and the anti-CCP antibody titres relied on a correlation test. Finally, the search for an association between exposure to genetic factors (SE alleles and/or PTPN22 1858T allele) and/or tobacco smoking with RF or anti-CCP positivity was performed with the same statistical tests as the global sample, in the subgroup of familial RA and in the subgroup of sporadic RA.

RESULTS

Rheumatoid arthritis sample characteristics

A total of 218 (79.6%) patients with RA carried at least one SE allele within 88 patients homozygous for SE alleles, whereas 89 (32.6%) patients with RA carried at least one T allele of PTPN22-1858 variant within nine patients homozygous for the PTPN22-1858 T allele. A total of 190 (69%) patients with RA were RF positive and 217 (79%) patients had anti-CCP antibodies, and 172 (63%) were both RF and anti-CCP antibody positive. Among the 243 (89%) patients with RA for whom the information about tobacco exposure was known, 122 (50%) were ever smokers, 43 (35%) being current smokers and 79 (65%) being ex-smokers. There were significantly more men in the subgroup of patients exposed to tobacco (18.0% vs. 1.6%, p = 2×10⁻⁵).

Association between autoantibody status and SE alleles

Rheumatoid factor and shared epitope alleles

The presence of at least one SE allele in the genotype was not associated with RF-positive RA (81.0% RF positive vs. 76.0% RF negative, OR = 1.3 (0.7 to 2.5)). The hierarchy of the genotype risk in the RF-positive subgroup was different from that previously reported in the literature, S₂/S₂ being the most at risk genotype followed by S₂/S_3P and S_3P/S_3P genotypes. The distribution of HLA-DRB1 alleles was similar in RF-positive and -negative patients with RA.

Anti-cyclic citrullinated peptide antibodies and shared epitope alleles

The presence of at least one SE allele in the genotype was significantly associated with anti-CCP antibody positive (82.5% in anti-CCP positive vs. 68.4% in anti-CCP negative, OR = 2.2 (1.2 to 4.2)), particularly for the subgroup of patients homozygous for SE (OR = 5.2 (2.0 to 13.6)) (table 1). The hierarchy of the genotype risk in the anti-CCP-positive subgroup was similar to that previously reported in the literature, although 95% CI largely overlapped (table 1).

The distribution of HLA-DRB1 alleles was similar in patients with RA who were anti-CCP positive and negative, except for HLA-DRB1*0401 allele (28.0% of HLA-DRB1*0401 alleles in anti-CCP positive vs. 16.4% in anti-CCP negative, OR = 1.9 (1.1 to 3.3)).

Association between autoantibody status, SE alleles and cigarette smoking

No effect of tobacco exposure on RF status was observed.

Tobacco exposure was significantly associated with anti-CCP antibody occurrence, but only in the presence of SE alleles (OR = 2.9 (1.2 to 7.4)) (table 2a). We found a negative, but not statistically significant interaction between those factors, AP = −0.83 (−1.75, +0.09). We identified an association between tobacco exposure and HLA-DRB1*0401 allele (*0401) for the presence of anti-CCP antibodies (table 2b), and a tendency toward a positive but not statistically significant interaction between those factors, AP = +0.24 (−0.21, +0.76).

Association between anti-cyclic citrullinated peptide antibody status and cigarette smoking

Tobacco exposure was not associated with RF positivity nor with anti-CCP positivity. However, the risk for anti-CCP positivity increased with the number of years of smoking (table 3) and was statistically significant at ⩾20 years of tobacco exposure (OR = 3.7 (1.1 to 12.8)).

The risk for anti-CCP antibodies also increased with the number of cigarettes smoked per day, rising from less than 1 for <10 cigarettes per day to 3.3 when ⩾20 cigarettes were smoked (table 4).

Finally, the risk for anti-CCP antibodies increased up to 4.2 for a cumulative dose of cigarettes smoked ⩾20 pack-years (table 5).

Indeed, we observed a correlation between the cumulative dose of smoked cigarettes expressed in pack-years, and anti-CCP antibody titres (r = 0.19, p = 0.04).

Association between autoantibodies, tobacco exposure, SE and PTPN22-1858 C/T genotype

We did not find any association between PTPN22-1858T allele and tobacco exposure neither for RF nor for anti-CCP antibodies positivity. We observed that the presence of both genetic factors, ie, SE and PTPN22-1858T alleles, was associated with a higher, but not statistically significant, risk to develop anti-CCP antibodies (OR = 2.9 (1.2 to 7.1)) than for each genetic factor separately (table 6a). A negative but not statistically significant interaction between those factors was observed, AP = −0.27 (−1.08, +0.54). Considering only the HLA-DRB1*0401 allele among the SE alleles, the risk for anti-CCP antibodies was similar in the presence of both genetic factors (HLA-DRB1*0401 and PTPN22-1858T alleles) or in the presence of HLA-DRB1*0401 allele alone, as 95% CI largely overlapped (table 6b), and a negative but not statistically significant interaction was observed, AP = −0.15 (−0.91, +0.60).

The search for association between genes, tobacco smoking and autoantibody positivity remained not significantly different in the subgroup of familial RA (78 affected sibling pairs families and 51 trio families with at least one first- or second-degree relative affected by RA). In the 145 trio families without any family history of RA, the presence of at least one SE allele in the genotype was significantly associated with anti-CCP positivity (OR = 2.7 (1.1 to 6.2)), particularly for the subgroup of patients homozygous for SE (OR = 14.5 (1.7 to 120.2)), and the presence of both genetic factors, ie, SE and PTPN22-1858T alleles, was associated with a higher, but not statistically significant, risk to develop anti-CCP antibodies (OR = 4.8 (1.4 to 16.2)) than for each genetic factor separately.

DISCUSSION

In this study, we aimed at evaluating the association between the two RA genetic factors (SE and PTPN22-1858T alleles), and/or tobacco exposure and autoantibody (RF and anti-CCP antibodies) positivity in a French population with familial and sporadic RA. We failed to identify any association between RF and SE, nor between RF and tobacco smoking. We observed that the presence of at least one SE allele was associated with anti-CCP antibody presence (82.5% vs. 68.4%, p = 0.02), particularly with HLA-DRB1*0401 allele (28.0% vs. 16.4%, p = 0.01). Tobacco exposure was significantly associated with anti-CCP antibodies, but only in the presence of SE. The hierarchy of HLA-DRB1 genotype risk was respected in the anti-CCP-positive subgroup, probably because of the high prevalence of the HLA-DRB1*0401 allele in this population. A tendency toward a positive but not statistically significant interaction was observed between tobacco, the presence of at least one HLA-DRB1*0401 allele and anti-CCP antibodies (AP = +0.24 (−0.21, +0.76)). The risk for anti-CCP positivity in RA index cases increased as the number of cigarettes smoked per day increased and as the number of years of smoking increased. The cumulative dose of cigarette smoking was correlated with anti-CCP antibody titres (r = 0.19, p = 0.04). The presence of both SE and PTPN22-1858T alleles was associated with a higher, but not significantly different, risk to develop anti-CCP antibodies than for each genetic factor separately. But this increased risk disappeared when considering only the HLA-DRB1*0401 allele within the SE. No association was found between PTPN22-1858T alleles and tobacco smoking for autoantibody positivity.

Here, we found an association between the anti-CCP positivity, tobacco exposure and SE alleles, particularly with the HLA-DRB1*0401 allele. Moreover the anti-CCP antibody titres were correlated with the intensity of the tobacco exposure, suggesting a strong effect of this environmental factor on autoantibody production, through a gene–environment association. This gene–environment association was not observed for PTPN22-1858 T alleles. However, the PTPN22 gene encodes a protein that is not involved in antigene recognition and this gene has a minor effect on RA susceptibility in comparison with SE; a larger sample size should be required to observe such a gene–environment association. Surprisingly, we failed to identify any association between RF and anti-CCP antibody presence and the PTPN22-1858 T allele. Indeed, Dieudé et al16 reported linkage to and association with this allele and the RF positivity in trio families. This difference may be explained by the fact that, in this study, we pooled these trio families and the affected sibling pairs families in an exposed–not exposed study. Furthermore RF status was determined by an ELISA test for IgM on a serum sample collected at the inclusion in the genetic study, whereas in the previous study, RF was considered as positive when at least one RF-positive result (determined by latex fixation, or Waaler–Rose assay or by laser nephelometry) was observed during the disease course.

Although the sample size was limited in this study, our findings of association between SE, anti-CCP antibodies and tobacco exposure were similar to those already reported in the literature.11 Recently, gene–gene and gene–environment interactions in RA were compared in three large case–control studies.19 This article reported an interaction between SE and PTPN22-1858T alleles for developing anti-CCP-positive RA, and the absence of an interaction between smoking and the PTPN22-1858T allele. The association between tobacco exposure, anti-CCP antibodies and HLA-DRB1*0401 allele is interesting as the citrullination of peptides such as vimentin selectively increased their binding to HLA-DR molecules containing the SE motif.20 Indeed, Hill et al21 reported that HLA-DRB1*0401 transgenic mice had a stronger immune response to citrullinated peptides than to native arginine-containing peptides.

In this study, we did not find any association between tobacco exposure, SE alleles and RF positivity although this association was previously reported in the literature. This observation could be due to the long RA duration in this sample and the possible disappearance of the RF during the evolution of the disease, and maybe to the fact that we chose to test only IgM RF and not IgA RF. Moreover the sample size is rather small to study the interaction between genetic factors and environmental factors, such as tobacco smoking. Replication studies with larger sample size of unselected patients with RA, should be required before clinical application.

Finally, it should be of great interest to go further by performing immunological studies to investigate the functional interaction between tobacco exposure, anti-CCP antibodies and HLA-DRB1*0401 allele, and to determine which component of the tobacco smoke should be responsible for such an autoimmune reaction.

In conclusion, our findings suggest an association between SE alleles and tobacco smoking for anti-CCP positivity and a tendency toward an interaction between the HLA-DRB1*0401 allele and smoking in the development of anti-CCP positivity in this French population with familial and sporadic RA.