Drs. Sibilia and Mariette contributed equally to this work.
In primary Sjögren's syndrome, HLA class II is associated exclusively with autoantibody production and spreading of the autoimmune response
Article first published online: 1 AUG 2003
Copyright © 2003 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 48, Issue 8, pages 2240–2245, August 2003
How to Cite
Gottenberg, J.-E., Busson, M., Loiseau, P., Cohen-Solal, J., Lepage, V., Charron, D., Sibilia, J. and Mariette, X. (2003), In primary Sjögren's syndrome, HLA class II is associated exclusively with autoantibody production and spreading of the autoimmune response. Arthritis & Rheumatism, 48: 2240–2245. doi: 10.1002/art.11103
- Issue published online: 1 AUG 2003
- Article first published online: 1 AUG 2003
- Manuscript Accepted: 14 APR 2003
- Manuscript Received: 2 DEC 2002
To reevaluate, in a large series of patients with Sjögren's syndrome (SS) recruited from 2 French centers, the question of whether HLA is associated with SS itself or with a pattern of secretion of autoantibodies.
One hundred forty-nine white patients fulfilling the American-European Consensus Group criteria for SS were divided into 3 subgroups, according to their anti-Ro/SSA and anti-La/SSB status, as follows: group 1 (n = 53), no antibody; group 2 (n = 46), anti-SSA only; group 3 (n = 50), both anti-SSA and anti-SSB. Patients were compared with 222 unrelated healthy subjects representative of the white population in France.
Comparisons between the 149 SS patients and 222 controls confirmed the association of SS with DRB1*03 (the frequency was 25% in patients versus 10% in controls) and DQB1*02 (32% versus 22%). The association between HLA and SS was restricted to patients with anti-SSA and/or anti-SSB; no association with HLA was observed in patients in group 1 (no antibody). The frequency of HLA–DRB1*15 was highest in group 2 (24%), compared with 11% in group 1 and 11% in controls, whereas the frequency of HLA–DRB1*03 was highest in group 3 (44%), compared with 12% in group 1, 19% in group 2, and 10% in controls. Group 2 and group 3 had more clinical and biologic markers of activity than did group 1 but were not clinically different. HLA alleles were not associated with clinical features of the disease, and were associated with only some biologic features: rheumatoid factor positivity, increased serum IgG, and thrombocytopenia were associated with HLA–DRB1*03, and neutropenia was associated with DQB1*01.
HLA class II markers confer genetic susceptibility to Sjögren's syndrome. The association between HLA and SS is restricted to patients with anti-SSA and/or anti-SSB antibodies; HLA is not associated with SS in patients without these autoantibodies. The absence of a difference in disease severity between groups 2 and 3, as well as the restricted association of HLA–DRB1*03 in group 3, strongly suggest that HLA alleles predispose to autoantibody secretion, without being associated with clinical outcome. HLA class II phenotype might support epitope spreading: HLA–DR15 favors anti-SSA synthesis, whereas HLA–DR3 is associated with both anti-SSA and anti-SSB production.
Primary Sjögren's syndrome (SS) is an autoimmune disorder characterized by lymphocytic infiltration of salivary and lachrymal glands leading to xerostomia and keratoconjunctivitis sicca, and systemic production of autoantibodies (1). The condition is characterized by numerous abnormalities of epithelial cells that can express class II major histocompatibility complex (MHC) and costimulation molecules (2, 3) and therefore can act as antigen-presenting cells to infiltrating lymphocytes, mainly of the CD4 type. The presented autoantigens are mainly SSA/Ro, SSB/La, α-fodrin and β-fodrin, or cholinergic muscarinic receptors. Help from CD4 cells leads to production of specific autoantibodies. Anti-SSA/Ro antibodies may be detected alone, whereas anti-SSB/La antibodies are always found in conjunction with anti-SSA/Ro, suggesting a spreading of the autoantibody response, the origin of which is still unknown.
The MHC genes are the best documented genetic risk factors for the development of autoimmune diseases and could be directly involved in SS because of their abnormal expression on the surface of epithelial cells, which could favor presentation of SSA and SSB epitopes to T cells, leading to specific help in autoantibody production (4). Most previous studies suggested an association between SS and the HLA–A1;B8;DR3/DQ2 haplotype (5–8). An association between this haplotype and autoantibodies to SSA or SSB was also shown in some studies (9–12). Until recently, however, the absence of consensual, objective diagnostic criteria for SS in patients without anti-SSA/SSB antibodies did not allow determination of whether this association was linked to the disease itself or to a pattern of secretion of autoantibodies. With the recently published consensual American-European criteria for SS (13), which require a focus score of ≥1 on a lip biopsy specimen obtained from a patient who is anti-SSA/SSB negative, it is now possible to be confident of the diagnosis of SS in patients without anti-SSA/SSB antibodies.
We took this opportunity to reinvestigate the question of whether HLA is associated with SS itself or with a pattern of secretion of autoantibodies, in a large cohort of 149 patients fulfilling the American-European Consensus Group criteria. The comparison between the whole cohort of SS patients and controls confirmed the known association of HLA with the disease. However, after stratification for the presence of anti-SSA/SSB, the HLA associations were restricted to autoantibody-positive patients.
PATIENTS AND METHODS
One hundred forty-nine white patients with primary SS according to the American-European Consensus Group criteria (which include a focus score of ≥1 or anti-SSA/SSB positivity) (13) were consecutively recruited from 2 French reference centers for SS (the rheumatology departments of Le Kremlin Bicêtre and Strasbourg). They were compared with 222 unrelated healthy subjects representative of the white population in France, who were not specifically screened for symptoms of SS; these subjects were previously included in HLA studies in France (14).
All patients were examined for clinical features of Sjögren's syndrome, including enlarged parotid glands, Raynaud's phenomenon, arthralgia, synovitis, bronchopulmonary involvement, purpura, myositis, neuropathy, and lymphoma. The erythrocyte sedimentation rate (ESR) was determined for each patient. Antinuclear antibodies were detected by indirect immunofluorescence using the HEp-2000 substrate, which consists of HEp-2 cells transfected with Ro 60 complementary DNA (15). Rheumatoid factor, complement components C3 and C4, and serum IgA, IgG, and IgM levels were measured by nephelometry. Anti-SSA/Ro and anti-SSB/La were detected using a commercial enzyme-linked immunosorbent assay (ELISA) (Varelisa Ro and La antibodies; Pharmacia & Upjohn, Freiburg, Germany), which used both baculovirus-expressed recombinant Ro 52 and Ro 60, coated in an unspecified ratio. All positive results were confirmed by either counterimmunoelectrophoresis using purified antigens obtained from rabbit and rat thymus powder (Pel-Freez, Rogers, AR) and from human spleen extract from our laboratory, or by double radial immunodiffusion.
Patients were divided into 3 subgroups according to their anti-SSA/SSB status, as follows: group 1 (n = 53), no antibody; group 2 (n = 46), anti-SSA only; group 3 (n = 50), both anti-SSA and anti-SSB.
HLA typing was performed using a serologic method for HLA–A and HLA–B, and HLA class II typing was performed using a molecular technique. Serologic typing for HLA–A and HLA–B was performed using the standard microlymphocytotoxicity method with monoclonal antibodies (One Lambda, Canoga Park, CA), which defined the 24 HLA–A and the 48 HLA–B antigens. For HLA class II typing, HLA–DRB1 and HLA–DQB1 medium resolution genotyping was performed using polymerase chain reaction single-strand oligonucleotide reverse dot-blot kits (InnoLipa DRB key and InnoLipa DQB kits, respectively; Innogenetics, Zwijndrecht, Belgium).
The distribution of HLA types was compared between patients and controls and between the different patient subgroups and controls. An association between clinical or biologic features of the disease and HLA was also investigated. The chi-square test was used to calculate the significance of the difference between groups. Analysis was performed using HLASTAT2000 software, which was developed by INSERM U396 (program available on request to firstname.lastname@example.org) and has already been used in other HLA studies (14). Allele frequencies were estimated using the maximum-likelihood method (16). Comparisons between the 3 groups for phenotype frequencies of each HLA allele were made, and each P value was corrected (Pcorr), multiplying by the number of tested alleles at each considered locus. The relative risk (RR) according to Haldane's method was also calculated for each allele (17).
Clinical and biologic characteristics.
The study group comprised 15 men and 134 women. The mean ± SD age at diagnosis was 56 ± 13 years, and the mean ± SD age at first symptom was 45 ± 7 years. Thirty-nine patients (26%) had enlarged parotid glands, 49 (33%) had Raynaud's phenomenon, 115 (77%) had arthralgias and/or synovitis, 5 (3%) had myositis, 18 (12%) had bronchopulmonary involvement, 13 (9%) had purpura, 16 (11%) had neuropathy, and 6 (4%) had lymphoma. Twenty-one patients (14%) were neutropenic, 8 (5.3%) were thrombocytopenic, and 22 (15%) had a monoclonal component (Table 1). The mean ± SD ESR was 30 ± 27 mm/hour. Sixty-eight patients (46%) had positive rheumatoid factor. Mean serum IgA, IgG, and IgM levels were 2.9 gm/liter, 14.7 gm/liter, and 1.5 gm/liter, respectively. Of the 149 patients, 96 (64%) had anti-SSA antibody, and 50 (33%) had both anti-SSB and anti-SSA antibodies (Table 1). No patient had anti-SSB antibodies in the absence of anti-SSA antibodies. Thus, the following 3 subgroups of approximately equal size were studied: patients in group 1 (n = 53) had no antibody, patients in group 2 (n = 46) had anti-SSA only, and patients in group 3 (n = 50) had both anti-SSA and anti-SSB. One hundred thirty-seven patients (92%) had a focus score of ≥1, including 100% of the patients in group 1.
|Characteristic||Group 1 (n = 53)||Group 2 (n = 46)||Group 3 (n = 50)||All patients (n = 149)|
|Enlarged parotid glands†||8||13||18||39 (26)|
|Raynaud's phenomenon||17||12||20||49 (33)|
|Peripheral nerve involvement||5||7||4||16 (11)|
|Lung involvement‡||3||2||13||18 (12)|
|Skin manifestations§||1||3||9||13 (9)|
|Monoclonal component||7||6||9||22 (15)|
|Mean ESR, mm/hour**||18.7||32.2||40.8||30.3|
Analysis of the whole group of SS patients.
There was no significant difference in autoantibody status or HLA type between patients referred to Le Kremlin Bicêtre and to Strasbourg; thus, the patients were pooled. Comparisons between the 149 patients with SS and the 222 controls confirmed the association of SS with both DRB1*03 and DQB1*02: the frequency of DRB1*03 was 25% in patients versus 10% in controls (RR 3.8, Pcorr = 10−7), and the frequency of DQB1*02 was 32% in patients versus 22% in controls (RR 2.2, Pcorr = 6 × 10−4) (Table 2). These alleles are in linkage disequilibrium on the A1;B8;DR3/DQ2 haplotype. No other HLA class I or class II marker was significantly associated with the disease.
|SS patients (n = 149)||Controls (n = 222)||Pcorr||RR|
|A1||55 (20)||49 (12)||0.02||2.2|
|B8||53 (19)||35 (8)||5 × 10−6||3.1|
|DRB1*03||72 (25)||44 (10)||10−7||3.8|
|DQB1*02||84 (32)||68 (22)||6 × 10−4||2.2|
Analysis of the subgroups.
Association of HLA with autoantibody secretion.
HLA distribution in patients in group 1 (SS patients without any antibody) was not significantly different from that in controls. The frequency of HLA–DRB1*15 was highest in group 2 (24%), compared with 11% in group 1 (RR 4.1, Pcorr = 0.019), and 11% in controls (RR 3.5, Pcorr = 0.0001) (Table 3). The frequency of HLA–DRB1*03 was highest in group 3 (44%), compared with 12% in group 1 (RR 15, Pcorr = 10−7), 19% in group 2 (RR 8.1, Pcorr = 0.0001), and 10% in controls (RR 20, Pcorr = 10−15). HLA–B8 and HLA–DQ2, which are in linkage disequilibrium with HLA–DRB1*03, were also significantly increased in group 3 compared with group 1, group 2, and controls.
|Allele||Group 1 (n = 53)||Group 2 (n = 46)||Group 3 (n = 50)||Controls (n = 222)|
|A1†||18 (17)||12 (13)||25 (28)||42 (12)|
|B8‡||11 (10)||10 (11)||32 (36)||35 (8)|
|DRB1*15§||9 (11)||21 (24)||14 (14)||44 (11)|
|DRB1*03¶||13 (12)||17 (19)||42 (44)||44 (10)|
|DQB1*02#||20 (21)||22 (26)||42 (49)||68 (22)|
Association of anti-SSA/SSB with clinical and biologic features of disease activity and severity.
Enlarged parotid glands, lung involvement, and skin manifestations were more frequent, and the mean ESR was higher, in groups 2 and 3 together compared with group 1 (Table 1). The differences in these parameters between groups 2 and 3 were not statistically significant, except that the frequency of lung involvement was higher in group 3.
Association of HLA alleles with clinical features of the disease.
No correlation between HLA and clinical features of SS was observed. Concerning the association between HLA markers and biologic results, HLA–DRB1*03 was not associated with a higher mean ESR (33 mm/hour in patients with HLA–DRB1*03 versus 28 mm/hour in patients without HLA–DRB1*03; Pcorr = 0.29). However, some biologic features associated with disease severity were associated with the DRB1*03/DQB1*01 haplotype. There was a slight association between DRB1*03 and thrombocytopenia (defined as <150 × 109 platelets per liter) and between DQB1*01 and neutropenia (defined as < 1 × 109 neutrophils per liter): 50% of patients with ORB1*03 were thrombocytopenic versus 24% of controls (Pcorr = 0.04), and 10% of patients with DQB1*01 were neutropenic versus 1% of controls (Pcorr = 0.016). HLA–DRB1*03 was also significantly associated with rheumatoid factor positivity (RR 3.01, Pcorr = 0.014) and a higher mean IgG level (16.2 gm/liter in patients versus 13.2 gm/liter in controls; Pcorr = 0.015).
This study included a racially homogeneous group of 149 patients with primary Sjögren's syndrome, who were carefully selected according to the American-European Consensus Group criteria. Findings in this series of 149 patients confirm the previously described association between SS and the HLA class II markers DR3 and DQ2 (5–8), which are in linkage disequilibrium with HLA–A1 and HLA–B8, in white SS patients. For example, Kang et al (7) also reported an increased frequency of the HLA–A1;B8;DR3/DQ2 haplotype in white patients. In addition, an association with DR2 (now called DRB1*15) has been reported in Scandinavian patients (18), and an association with DR5 has been reported in Greek patients (19). Last, the HLA–DRB1*0301/*1501 heterozygote genotype was observed to be increased in a population of white patients in France (20).
This study represents the largest ever performed in which an association between HLA and SS (as defined by strict consensus criteria) was investigated, and thus allows individualization of 3 subgroups of equal size (no antibody, anti-SSA only, both anti-SSA and anti-SSB) in a population in which every patient equally fulfilled the American-European Consensus Group criteria for Sjögren's syndrome. Thus, every patient in the group without anti-SSA or anti-SSB had a focus score of ≥1. Surprisingly, the association between HLA and SS was restricted to patients with anti-SSA and/or anti-SSB; no association with HLA was observed in the subgroup of patients in whom these autoantibodies were absent.
Patients with anti-SSA expressed the DRB1*15/DQB1*01 haplotype with a higher frequency, whereas patients with both anti-SSA and anti-SSB more frequently expressed the DRB1*03/DQB1*02 haplotype. The association between anti-SSB and this haplotype was entirely responsible for the association found between A1;B8;DR3/DQ2 and Sjögren's syndrome, because this haplotype was not associated with either of the other subgroups. Because of the very close linkage between DRB1*15 and DQB1*01, as well as between DRB1*03 and DQB1*02, it was impossible to determine whether the closest association involved the DR or the DQ locus. Independently from anti-SSB, HLA–DRB1*03 was also associated with 2 prominent features of primary SS: rheumatoid factor positivity and increased serum polyclonal IgG level.
Accordingly, Harley et al (9) suggested a correlation between HLA and the level of hypergammaglobulinemia and autoantibodies in SS, by showing that heterozygotes for DQ1 and DQ2 had the highest level of anti-SSA and anti-SSB antibodies. Most studies in the literature about HLA predisposition in primary SS also disclosed an association between DRB1*03/DQB1*02 and the presence of autoantibodies, but with anti-SSA as well as with anti-SSB (8–12). However, only 1 of these studies had a large enough number of patients to permit individualization of subgroups with anti-SSA only (10). In the study by Harley et al (as in ours), the latter subgroup was associated with the DR15/DQ1 haplotype (DR15 was formerly known as DR2), whereas DR3/DQ2 was associated with the presence of precipitating anti-SSB. Among patients with anti-Ro/SSA, the presence of anti-SSA antibody was confirmed by either double radial immunodiffusion or counterimmunoelectrophoresis. Positivity by these techniques, which detect precipitins, implies the mandatory presence of anti–Ro 60 antibody. Thus, we could be sure that all of the patients with anti-Ro/SSA had anti–Ro 60 antibodies, and that this specificity was associated with HLA–DR15. However, it was not possible to look for a possible association between HLA and anti–Ro 52 antibody, because the ELISA could not differentiate between anti–Ro 52 antibody alone and both anti–Ro 52 and anti–Ro 60 antibodies.
A recent study demonstrated that regions of SSB were able to generate a T cell proliferative response (21), provided these T cells were HLA–DR3/DQ2 restricted. Thus, together, these results and the literature data demonstrate that the association of HLA class II with SS susceptibility is related exclusively to the pattern of autoantibody diversification, rather than to the disease itself. Likewise, in animal models transfected with human HLA class II, natural spreading of immune response from Ro 60 to Ro 52 and La depended only on the nature of the transgenic human HLA class II (22), whereas the clinical course of the disease was not modified (23). In accordance with previous reports (7), this study did not show any association between HLA class II and clinical features of SS, but demonstrated an association between anti-Ro/La antibodies and extraglandular involvement and disease severity. Indeed, patients with anti-Ro alone or patients with anti-Ro and anti-La (groups 2 and 3) had a more severe disease than did patients in whom these autoantibodies were absent (group 1), but the group with anti-Ro alone and the group with anti-Ro and anti-La did not differ clinically from one another except in the frequency of lung involvement (Table 1), whereas HLA–DRB1*03 allele frequency was higher only in group 3. Thus, the absence of a difference in disease severity between groups 2 and 3, as well as the restricted association of HLA–DRB1*03 in group 3, strongly suggest that HLA alleles predispose to autoantibody secretion without being associated with clinical outcome.
Anti-Ro and anti-La were first described in sera from lupus patients. The role of HLA as a predisposition factor in systemic lupus erythematosus (SLE) has been widely studied. Reveille et al (24) showed that among black Americans with lupus, the frequency of HLA–DQA1*0401 was lower in patients with both anti-SSA and anti-SSB than in patients with anti-SSA only. In addition, the frequency of DQB1*0201 was higher in patients with anti-SSA only than in white lupus patients without anti-SSA. Scofield et al (25) found a cooperative association between T cell receptor β restriction enzyme polymorphisms and HLA–DQB1*0201 and one of DQA1*0101, *0102, *0103 alleles in SLE patients who had anti-Ro antibodies only. Results of these studies suggest that HLA essentially promotes autoantibody responses in SLE. Whether the HLA association with anti-Ro/La could be similar regardless of the disease (SLE or primary SS) deserves further evaluation.
In SS, as in lupus, anti-SSA may be found alone in serum, whereas anti-SSB antibodies are always detected in association with anti-SSA. This is interpreted as a spreading of autoimmune response from SSA to SSB epitopes. Rischmueller et al (10) speculated that anti-SSB antibody production requires efficient help to B cells, which could be provided only by DR3/DQ2-restricted T helper cells recognizing SSB determinants. DR15/DQ1-restricted T helper cells recognizing SSA determinants could stimulate anti-SSA antibody synthesis but could not contribute to anti-SSB antibody synthesis. This explanation may account for the differential HLA association between anti-SSA and anti-SSB but does not give any clue to the reason for the necessary presence of anti-SSA in production of anti-SSB.
We suggest the following 3 hypotheses concerning the role of HLA in autoantibody diversification: 1) HLA–DR15/DQ1 might present only SSA epitopes, whereas HLA–DR3/DQ2 could present both SSA and SSB determinants; 2) SSB processing could require anti-SSA synthesis first, then formation of an immune complex composed of anti-SSA, SSA, and SSB, which would be mandatory for SSB presentation to HLA–DR3–restricted cells; 3) DR15/DQ1 (but not DR3/DQ2) could be associated with the presence of an antiidiotypic response to anti-SSB, which could neutralize anti-SSB antibody (26).
This study, which included 149 patients in a homogeneous population of white patients with primary SS according to the American-European Consensus Group criteria, shows that HLA class II markers confer genetic susceptibility to Sjögren's syndrome but do not influence the clinical course of the disease. The association between HLA and SS is restricted to patients with anti-SSA and/or anti-SSB antibodies. HLA is not associated with SS in the subgroup of patients without any of these antibodies. Furthermore, this study suggests that HLA class II might support epitope spreading: HLA–DR15 favors anti-SSA synthesis, whereas HLA–DR3 favors both anti-SSA and anti-SSB production.
- 1Sjögren's syndrome. In: MaddisonPJ, IsenbergDA, WooP, GlassDN, editors. Oxford textbook of rheumatology. 2nd ed. Oxford: Oxford Medical Publications; 1997. p. 1301–17., , .