Particular HLA–DRB1 shared epitope genotypes are strongly associated with rheumatoid vasculitis

Authors


Abstract

Objective

To examine the relationship of the HLA–DRB1 shared epitope (SE) to rheumatoid vasculitis, using individual patient data (IPD) meta-analytic methods.

Methods

Published studies that enrolled adult patients with rheumatoid arthritis (RA) were identified by searches of Medline and Embase, and by manual searches of medical journals. All authors were contacted for IPD. Meta-analyses were performed to assess the association of SE presence, dose, and genotype with rheumatoid vasculitis.

Results

A total of 14 studies and 1,568 patients (129 with vasculitis) were included in the analysis. RA patients with vasculitis were significantly more likely to have rheumatoid nodules (odds ratio [OR] 2.5, 95% confidence interval [95% CI] 1.5–3.9], but there was no significant association with male sex, rheumatoid factor positivity, or erosive disease. No significant association was observed between the presence of the SE (i.e., 1 or 2 alleles versus 0 alleles) and rheumatoid vasculitis (summary OR 1.4, 95% CI 0.7–2.7). Analysis by SE genotype, however, demonstrated a striking relationship of vasculitis to 3 genotypes containing a double dose of the SE, specifically HLA–DRB1*0401/*0401 (OR 6.2, 95% CI 1.01–37.9), *0401/*0404 (OR 4.1, 95% CI 1.1–16.2), and *0101/*0401 (OR 4.0, 95% CI 1.4–11.6).

Conclusion

The HLA–DRB1 SE genotypes *0401/*0401, *0401/*0404, and *0101/*0401 may be of particular importance to rheumatoid vasculitis. It is hoped that additional investigation of these and other SE genotypes will lead to improved insight into the mechanisms influencing the clinical expression of RA.

Rheumatoid arthritis (RA) is a chronic systemic inflammatory disease affecting ∼1% of the population. Although the most characteristic feature of RA is erosive joint damage, systemic manifestations are common, and almost any organ system can be affected during the course of disease. One of the most serious outcomes is development of extraarticular manifestations, including rheumatoid vasculitis, which are associated with significant morbidity, including a 2- to 4-fold increase in RA mortality (1, 2).

The clinical phenotype manifested by individual patients with RA, such as the presence of extraarticular manifestations, demonstrates substantial variability that may reflect genetic differences. The most extensively studied genetic marker in RA is HLA–DRB1 and, more specifically, those alleles with a conserved amino acid sequence in the molecule's third hypervariable region, termed the shared epitope (SE) (3). Among individual studies, the most striking relationship of the SE to RA outcome was described in an analysis of the association with nodules and other extraarticular manifestations in a highly selected group of RA patients (4). Although numerous studies have since pursued clarification of this relationship, subsequent findings have been less remarkable, with many failing to show an association (5–7). It has been hypothesized that the strongest relationship of the SE may be to rheumatoid vasculitis (8, 9), but this hypothesis is based on a few individual studies and has not been systematically investigated. Moreover, speculation that specific SE genotypes, such as *0401 homozygotes, confer substantially greater risk for vasculitis has not been corroborated.

The paucity of information regarding the association of vasculitis with particular SE genotypes, in combination with our previous work suggesting that these genotypes did not increase the risk of rheumatoid nodules (10), motivated us to examine this relationship using individual patient data (IPD) meta-analysis. The receipt of IPD provided us with the unique opportunity to examine the role of specific SE genotypes in rheumatoid vasculitis and to investigate the influence of covariates such as disease duration and ethnicity.

MATERIALS AND METHODS

Inclusion criteria.

Studies eligible for inclusion in this meta-analysis were those in which HLA–DRB1 alleles in patients with RA were investigated, and in which information regarding vasculitis was provided. The HLA–DRB1 alleles defined as encoding the SE were as follows: HLA–DRB1*0101, *0102, and *0104; HLA–DRB1 *0401, *0404, *0405, *0408, *0409, *0410, *0413, *0416, *0419, and *0421; HLA–DRB1 *1001; and HLA–DRB1 *1402 and *1406 (11). The frequency of HLA–DRB1 alleles varies according to ethnic and racial background, with some alleles being extremely rare; therefore, articles were not required to identify all 16 SE alleles to be eligible for inclusion. Studies selected were those in which adult patients with RA were evaluated and in which vasculitis was assessed.

Search strategy.

Three overlapping search strategies were used to identify studies published between January 1987 and June 1999. We searched Medline with the truncated keywords of rheumatoid arthritis and 1 or more of the following: human leukocyte antigen, gene, prognosis, or severity. Embase was searched with the exploded terms rheumatoid arthritis and 1 or more of the following: histocompatibility, HLA system, epitope, etiology, or genetics. The database search terms were designed to be sensitive in order to identify all studies involving HLA–DRB1 genotyping in which there was any reference to RA severity. In addition, an updated PubMed search was performed to identify articles published between June 1999 and March 2003.

Hand searches of 6 journals (Annals of Rheumatic Diseases, Arthritis & Rheumatism, British Journal of Rheumatology, Journal of Rheumatology, Scandinavian Journal of Rheumatology, and Tissue Antigens) for articles (January 1987 through June 1999) were also performed. The start date for article identification was selected based on the initiation of molecular DRB1 typing in 1987 (12).

A single investigator (JDG) performed the literature searches, reviewed abstracts, and extracted data from studies. Training on database and hand searching was provided by formal instruction prior to the start of the study. Upon identification of articles appropriate for study inclusion, letters were mailed or electronically sent to the corresponding author to request missing information and/or IPD. If the same patients were reported in more than 1 article (i.e., duplicate data) and IPD were not available, the publication that presented the most complete recent information was selected for inclusion in the analysis. Because the IPD were provided without personal identifiers, this study was certified as exempt from Human Subjects Review by the University of California, San Francisco Committee on Human Research.

Statistical analysis.

The odds ratio (OR) for the association between the SE and vasculitis was the effect measure of interest; patients with no SE alleles served as the reference group for the computation of ORs. For the meta-analysis, summary ORs were computed using fixed-effects models (for data that did not demonstrate significant heterogeneity) or random-effects models (for data demonstrating significant heterogeneity) (13). The Breslow-Day test of homogeneity was used to assess heterogeneity (14), and a significance level of P = 0.1 was established to test consistency of the ORs across studies (15).

Additional analyses were performed to account for differences in disease duration between individuals. This approach required the use of multivariate logistic regression analyses to calculate ORs for each of the individual studies along with 95% confidence intervals (95% CIs), with rheumatoid vasculitis as the outcome variable, SE (i.e., presence, dose, or allelic genotype) as the explanatory variable, and disease duration as a covariate. IPD were required for this adjusted analysis because of the need for information regarding disease duration for each patient. The adjusted ORs after logistic regression were then pooled using fixed-effects or random-effects meta-analyses.

Publication bias was assessed graphically by funnel plot (16) and was statistically evaluated by the regression asymmetry test described by Egger et al (17), which quantitatively evaluates funnel plot asymmetry by a linear regression analysis incorporating information regarding the effect of genotype and standard error, an index related to study size. All analyses were performed using Stata statistical software, version 8.0 (18).

RESULTS

The searches identified a total of 10,367 articles. Upon review, however, only 33 articles mentioned HLA–DRB1 and extraarticular manifestations or vasculitis in the title or abstract and represented a primary data analysis. Further assessment revealed that in only 22 of these 33 articles were patients specifically evaluated for vasculitis. Four of these 22 articles were excluded: 2 because molecular typing was not performed, 1 that represented a duplicate publication, and 1 that was a review article. Thus, a total of 18 studies were appropriate for inclusion. Authors were contacted for additional data, and the authors of 12 of these 18 articles provided IPD. Two additional articles from the original 18 did not include IPD, but enough information was available within the published articles regarding the SE genotype and vasculitis to allow inclusion of these 2 studies in the primary analyses. Therefore, our final meta-analysis consisted of 14 studies (12 with IPD) and 1,568 patients, 129 of whom had vasculitis (7, 19–31).

Characteristics of the included studies are presented in Table 1. Eighty-one percent of the patients were Caucasian. Two studies included in the meta-analysis investigated Asians (i.e., Korean or Taiwanese) with RA, and a single study enrolled both Hispanic-American and African-American patients. IPD were available for 86% of included studies, corresponding to 89% of the total patient population. The mean duration of disease was 14 years, 79% of the patients were female, and 82% of patients were seropositive for rheumatoid factor (RF). A description of the criteria used for assessing rheumatoid vasculitis was present within the published article for 8 of the 14 studies. Contact with individual authors provided the definitions used in 3 additional studies. The definitions of vasculitis used in individual studies are shown in Table 2.

Table 1. Characteristics of studies included in meta-analysis of HLA–DRB1 shared epitope (SE) with rheumatoid vasculitis*
Race, author, year (ref.)Patient selection requirementsStudy designNo. of patientsEthnicityMean age, yearsMean disease duration, years% female% RF positive% SE positive
  • *

    RF = rheumatoid factor.

  • Available for at least 1 analysis.

  • Individual patient data not available.

Caucasian         
 Turesson et al, 2000 (7) Retrospective cohort74Swedish17775788
 Bas et al, 2000 (19) Cross-sectional238Swiss6516747377
 Toussirot et al, 1999 (20) Case–control119Southern French6110748674
 Criswell et al, 1998 (21)FemaleProspective cohort142American Caucasians63201007870
 Cortet et al, 1997 (22) Cross-sectional69Northern French5912807568
 Perdriger et al, 1997 (23)Erosive diseaseCase–control200Northern French5913768584
 Rowley et al, 1997 (24) Cross-sectional79European/Australian Caucasians18679278
 Voskuyl et al, 1997 (25) Case–control107Dutch729476
 Seitz et al, 1996 (26) Case–control83Swiss13758873
 Benazet et al, 1995 (27) Cross-sectional117Southern French767884
 Boki et al, 1993 (28) Cross-sectional45Greek11877889
Non-Caucasian         
 Kim et al, 1997 (29)SeropositiveCross-sectional102Korean910068
 Teller et al, 1996 (30) Cross-sectional59Hispanic American,     
    African American5411958153
 Yen et al, 1995 (31) Cross-sectional134Taiwanese10848951
Total  1,568      
Mean    61.413.779.182.474.1
Table 2. Individual study definitions of rheumatoid vasculitis
Race, author, year (ref.)Definition
Caucasian 
 Turesson et al, 2000 (7)Cutaneous vasculitis: purpura or other major cutaneous ulcer in the absence of significant atherosclerosis or varicose vein disease verified by biopsy or clinical judgment by dermatologist. Scleritis or retinal vasculitis as clinically judged by ophthalmologist. Vasculitis involving other organs as judged by clinical judgment by organ specialist and biopsy compatible with vasculitis.
 Bas et al, 2000 (19)Any clear manifestation of rheumatoid vasculitis during evolution of the disease.
 Toussirot et al, 1999 (20)Cutaneous ulceration without atherosclerotic arterial involvement, peripheral neuropathy, or mononeuritis.
 Criswell et al, 1998 (21)Diagnosis of vasculitis by treating rheumatologist, which includes cutaneous lesions, ocular disease, and major organ involvement.
 Cortet et al, 1997 (22)Systemic rheumatoid vasculitis defined as the presence of mononeuritis multiplex, peripheral gangrene, biopsy evidence of acute necrotizing arteritis plus systemic illness, and/or deep cutaneous ulcers or active extraarticular disease (e.g., pleurisy, pericarditis, scleritis) if associated with typical digital infarcts or biopsy evidence of vasculitis.
 Perdriger et al, 1997 (23)Major vasculitis defined as cutaneous ulcerations without atherosclerotic arterial involvement and peripheral neuropathy, and minor vasculitis as isolated nailfold vasculitis.
 Rowley et al, 1997 (24)Definition not available.
 Voskuyl et al, 1997 (25)All cases had histologic evidence of vasculitis. Petechiae or purpura were categorized as minor skin vasculitis. Major organ lesions indicative of vasculitis were considered to include peripheral neuropathy, deep skin ulcers, gangrene, necrotizing glomerulonephritis, ischemic bowel disease, and necrotizing scleritis.
 Seitz et al, 1996 (26)Rheumatoid vasculitis of the skin and peripheral nerves.
 Benazet et al, 1995 (27)Any diagnosis of vasculitis, including that affecting the skin or nervous system or nailfold vasculitis.
 Boki et al, 1993 (28)Cutaneous vasculitis.
Non-Caucasian 
 Kim et al, 1997 (29)Definition not available.
 Teller et al, 1996 (30)Definition not available.
 Yen et al, 1995 (31)Cutaneous vasculitis.

An investigation of the clinical characteristics of the patients (Table 3) revealed that those with vasculitis were significantly more likely to have rheumatoid nodules (OR 2.5, 95% CI 1.5–3.9). Moreover, the risk of vasculitis was significantly greater in those with longer disease duration (OR 1.3, 95% CI 1.1–1.5 [per 10-year increase in duration]). Although RA patients with vasculitis were more likely to be male and RF positive (OR 1.5 for both), neither of these variables achieved statistical significance in the analysis.

Table 3. Association between clinical characteristics and rheumatoid vasculitis*
CharacteristicNo. of studiesNo. of patients available for analysisSummary OR (95% CI)
  • *

    Statistically significant (P < 0.05) values are shown in boldface. OR = odds ratio; 95% CI = 95% confidence interval.

Male sex78371.5 (0.9–2.5)
Rheumatoid factor positive68361.5 (0.7–3.0)
Radiographic erosions32790.7 (0.1–3.0)
Rheumatoid nodules91,0822.5 (1.5–3.9)
Disease duration, per 10-year increase111,2541.3 (1.1–1.5)

Relationship of rheumatoid vasculitis to the presence and dose of the SE.

In analyses adjusted for disease duration, the presence of the SE (i.e., 1 or 2 versus 0 alleles) was not significantly associated with rheumatoid vasculitis (OR 1.4, 95% CI 0.7–2.7, P = 0.98 for heterogeneity). However, other analyses by SE dosage suggested the presence of a dose-dependent association (Table 4), because the OR estimate was greater for patients with 2 SE alleles than for those with a single allele (2.1 and 1.1, respectively [the reference group had 0 SE alleles]). More obvious evidence for a dose-dependent relationship of the SE to vasculitis was provided by the significant difference observed by direct comparison of those with 2 versus 1 SE allele (OR 1.8, 95% CI 1.1–3.1, P = 0.88 for heterogeneity).

Table 4. Association of the HLA–DRB1 SE genotype with vasculitis in 1,224 patients*
GenotypeNo. of studiesNo. of patients with genotypeNo. of patients available for analysisOR (95% CI)*P for heterogeneity
  • *

    Analysis adjusted for disease duration. The referent group for all comparisons was 0 shared epitope (SE) alleles. OR = odds ratio; 95% CI = 95% confidence interval; X = non–SE allele. Statistically significant (P < 0.05) values are shown in boldface.

  • Values <0.1 represent significant heterogeneity.

  • Grouped analysis of remaining SE genotypes excluding the specific ones listed.

2 SE alleles61873642.1 (0.98–4.7)0.90
 0401/04013101046.2 (1.0137.9)0.29
 0401/04044281464.1 (1.116.2)0.38
 0101/04014561834.0 (1.411.6)0.34
 SE/SE4401471.9 (0.6–6.2)0.60
1 SE allele74426831.1 (0.5–2.3)0.95
 0401/X51512991.6 (0.6–3.9)0.84
 0404/X3271183.7 (0.7–18.9)0.34
 0405/X2531521.2 (0.2–7.7)0.78
 0101/X41122381.2 (0.4–3.6)0.78

Relationship of rheumatoid vasculitis to the DRB1 SE genotype.

Because individual SE alleles and genotypes may differ in their association with specific outcomes (32), an analysis by individual genotypes was undertaken, adjusting for disease duration (Table 4). A total of 1,224 patients were available for these analyses, including 93 with vasculitis. Three genotypes demonstrated a statistically significant association with vasculitis, specifically HLA–DRB1*0401 homozygotes (OR 6.2, 95% CI 1.01–37.9), HLA–DRB1*0401/*0404 (OR 4.1, 95% CI 1.1–16.2), and HLA–DRB1*0101/*0401 (OR 4.0, 95% CI 1.4–11.6). However, the 95% CIs for these analyses were relatively wide and, moreover, the result for HLA–DRB1*0401/*0401 was only marginally significant. The association of vasculitis with other individual genotypes containing a double dose of the SE could not be assessed separately due to limited numbers within single studies. Analysis of these other double-dose genotypes as a single group demonstrated a much less striking and nonsignificant association with vasculitis (OR 1.9, 95% CI 0.6–6.2) in comparison with the 3 genotypes that could be independently assessed (Table 4). In order to evaluate a more homogeneous population, a separate analysis limited to Caucasians, the largest racial subgroup, comprising 707 patients with IPD (including 84 with vasculitis), was performed, and the results were not qualitatively different from those of the analyses that included all patients without regard to race (data not shown).

Assessment of publication bias.

A funnel plot of the association of SE positivity with rheumatoid vasculitis appeared relatively symmetric (data not shown). Moreover, statistical assessment by the regression asymmetry test described by Egger et al (17) did not reveal significant evidence for publication bias (P = 0.8).

DISCUSSION

Results from this meta-analysis suggest the particular importance of several SE genotypes to rheumatoid vasculitis, specifically HLA–DRB1*0401/*0401, HLA–DRB1*0401/*0404, and HLA–DRB1*0101/*0401. Moreover, the relationship to outcome was sizeable, with persons having inherited 1 of these genotypes demonstrating a ≥4-fold risk of vasculitis compared with those with no SE alleles. Although some researchers have reported individual importance of the HLA–DRB1*0401 homozygote genotype or specific SE compound heterozygotes (e.g., *0401/*0404) for extraarticular manifestations (33–35), our analyses could not demonstrate a substantially greater significance of one particular genotype over the other. However, these specific genotypes were fairly rare, as is the frequency of rheumatoid vasculitis, resulting in relatively imprecise estimates. Therefore, larger collaborative efforts are required to determine whether, in fact, essential differences exist between these particular SE allelic genotypes and the development of rheumatoid vasculitis.

In addition to genetic analyses, clinical comparisons revealed important differences between RA patients with and those without vasculitis. Specifically, patients with rheumatoid vasculitis were twice as likely to develop rheumatoid nodules, and those with disease of longer duration were more likely to have vasculitis; both of these findings are consistent with those from previous studies (36, 37). We did not demonstrate a significantly increased risk of vasculitis among men, subjects seropositive for RF, or those with radiographic erosions, in contrast to earlier work that demonstrated a significant association of all 3 characteristics with rheumatoid nodules among Caucasians (10). In addition, although significant associations of particular SE genotypes with vasculitis were observed in this study, our related meta-analysis failed to demonstrate a relationship of any SE genotype to nodules (10), suggesting that discrete pathogenetic processes may be important in the development of rheumatoid nodules and vasculitis.

It is of interest that all 3 genotypes demonstrated to be statistically significantly associated with vasculitis contained 2 copies of the SE, at least 1 of which was DRB1*0401. The association of DRB1*0401 with vasculitis is consistent with other observations that this same allele occurs with increased frequency in other inflammatory conditions. The most striking relationship has been demonstrated for the *0401 allele and Felty's syndrome (38), which is the clinical constellation of RA, splenomegaly, and neutropenia. Moreover, HLA–DR4 has been associated with both susceptibility to and the severity of giant cell arteritis (39–42), a systemic vasculitis of the cranial branches of aortic vessels. In this setting, particular importance has been suggested for DRB1*0401 and *0404 (43), which are the same alleles believed to be of greatest significance in the clinical expression of RA.

In addition to a suspected relationship of DRB1*0401 homozygosity to RA severity, several researchers have described the importance of the *0401/*0404 compound heterozygote genotype in terms of extraarticular manifestations (34, 44). In our analyses, 2 compound heterozygotes were significantly associated with rheumatoid vasculitis, specifically *0401/*0404 and *0101/*0401. Compound heterozygosity of HLA alleles has been of interest in autoimmunity, with perhaps the most compelling association being that of HLA–DR3/DR4 heterozygosity with type 1 diabetes mellitus (45). In contrast to diabetes, in RA the importance of DRB1 heterozygosity has been investigated to a lesser extent, and the results more controversial (33, 44, 46).

There are several limitations to our analysis that, in large part, relate to restrictions inherent in using existing data. These include the inability to examine other HLA–DRB1 SE allelic genotypes of potential interest (e.g., other compound heterozygotes or genotypes containing the *1001 allele) due to selective genotyping in the primary studies as well as the infrequency of several of these genotypes. However, even with these limitations, we did have the opportunity to investigate the role of SE genotypes identified to be of greatest interest to date. It should be noted that we are aware of only 5 studies that have described the association of specific HLA–DRB1 SE genotypes with the discrete outcome of rheumatoid vasculitis (as opposed to extraarticular manifestations generally) (7, 20, 23, 25, 26). However, in almost every case, these studies did not examine the role of single genotypes (e.g., *0401/*0404) but instead grouped multiple genotypes for analysis. This is not surprising, because it would be difficult for any individual study to have sufficient power to adequately examine these relationships. Therefore, findings from our IPD meta-analysis represent a substantial improvement in clarifying these associations compared with what is available within the current literature.

A problem with the existing literature pertinent to our analysis is the lack of specific information provided by individual studies regarding the criteria used in the assessment of rheumatoid vasculitis. The review of studies investigating rheumatoid vasculitis is complicated, not only because of the broad spectrum of vasculitis phenotypes included in a single study (e.g., cutaneous lesions, necrotizing scleritis, and peripheral neuropathy), but also because of the lack of uniformity of definitions across individual studies. Unfortunately, the rarity of these outcomes has prompted most researchers to group all vasculitis phenotypes together as a single entity in analyses reported in the literature.

In our study, we anticipate that this variability in definitions of vasculitis across studies attenuated our ability to observe significant relationships of SE genotype to outcome. This would be expected to occur for several reasons. First, the genetic association may be limited to specific vasculitis phenotypes. If this were the case, combined analysis of these varied processes would bias the estimate toward the null value, even if some of the phenotypes were associated with a particular SE genotype. However, even if all manifestations of rheumatoid vasculitis were equally associated with various SE genotypes, the variability in definitions across studies would still lead to a bias in the estimate toward the null. For example, studies limiting cases to those involving cutaneous lesions would misclassify those with other potentially relevant vasculitis outcomes, such as scleritis, as (nonvasculitis) controls. Moreover, because the frequency of vasculitis increases with disease duration, potential cases of vasculitis may have been misclassified as controls simply because sufficient time had not elapsed for development of the outcome.

In order to overcome some of these limitations, ideally one would separately evaluate the association of genotype with particular vasculitis phenotypes. However, even in the context of this meta-analysis, there were not enough individuals or information available to perform these analyses. Because of our collection of IPD, however, we could control for differences in disease duration in our analyses, thus minimizing the impact of this potential misclassification of our results. These limitations of the existing data strongly emphasize the need for more detailed reporting (and uniformity) of the definitions of various RA phenotypes. These changes would allow for better interpretation of individual studies and benefit future meta-analyses.

Last, linkage disequilibrium must be considered in interpreting our results. Thus, although our findings support an association of several SE genotypes with vasculitis in patients with RA, it is possible that another gene or genes located in close proximity to HLA–DRB1 are actually responsible for the relationship. Examination of allelic haplotypes across these and other HLA loci in RA is of increasing interest to researchers (47, 48).

In summary, this IPD meta-analysis demonstrates that 3 individual SE genotypes (i.e., HLA–DRB1*0401/*0401, HLA–DRB1*0401/*0404, and HLA–DRB1*0101/*0401) are significantly associated with vasculitis, one of the most severe outcomes of RA. These findings substantially contribute to our understanding of genetic risk factors for rheumatoid vasculitis through application of more sophisticated meta-analysis methods. Importantly, this meta-analysis was possible only because of the generosity of numerous independent investigators worldwide (see Acknowledgments). Further clarification of the genetic associations with individual phenotypes of RA, as well as with other complex diseases, will likely depend on similar collaborative approaches. Ultimately, it is hoped that this knowledge will translate into a better understanding of pathogenetic mechanisms and the development of novel and more effective therapies for these disease processes.

Acknowledgements

We thank Kirsten A. Pfeiffer, BA, Glenys Thomson, PhD, John J. Chen, PhD, Maria E. Suarez-Almazor, MD, PhD, Catherine Mallon, RN, John M. Colford, MD, MPH, PhD, and Mark Segal, PhD, for their insight and assistance with this project.

We are indebted to numerous colleagues who generously contributed data from their independent research for inclusion in this meta-analysis. These contributors include Graciela S. Alarcón, MD, PhD (Birmingham, AL), Sylvette Bas, PhD (Geneva, Switzerland), Alain Cantagrel, MD (Toulouse, France), Bernard Combe, MD, PhD (Montpellier, France), Bernard Cortet, MD, PhD (Lille, France), Anne Crilly, PhD (Glasgow, Scotland), Anne Davidson, MB, BS, FRACP (Bronx, NY), Kerstin Eberhardt, MD, PhD (Lund, Sweden), Jean-François Eliaou, MD, PhD (Montpellier, France), Markku Hakala, MD (Heinola, Finland), Ho-Youn Kim, MD (Seoul, Korea), Alex J. MacGregor, MD (London, UK), Loreto Massardo, MD (Santiago, Chile), Janet E. McDonagh, MD, MRCP (Birmingham, UK), Lachy McLean, MD, PhD (Loughborough, UK), Fiona M. McQueen, MBChB, MD, FRACP (Auckland, New Zealand), Olivier Meyer, MD (Paris, France), Timo Möttönen, MD (Paimio, Finland), Haralampos M. Moutsopoulos, MD, FACP, FRCP(Edin) (Athens, Greece), George Moxley, MD (Richmond, VA), Antonio Núñez-Roldan, MD, PhD (Seville, Spain), James R. O'Dell, MD (Omaha, NB), William E. R. Ollier, PhD (Manchester, UK), Aleth Perdriger, MD, PhD (Rennes, France), Janet E. Pope, MD, MPH, FRCPC (London, Ontario, Canada), Jean Roudier, MD, PhD (Marseilles, France), Merrill Rowley, PhD (Victoria, Australia), Güher Saruhan-Direskeneli, MD (Istanbul, Turkey), Michael F. Seldin, MD, PhD (Davis, California), Alan J. Silman, MD (Manchester, UK), Dharam P. Singal, PhD (Hamilton, Ontario, Canada), Fujio Takeuchi, MD (Tokyo, Japan), Wendy Thomson, PhD (Manchester, UK), Pierre Tiberghien, MD, PhD (Besançon, France), Yoshitaka Toda, MD (Osaka, Japan), Éric Toussirot, MD (Besançon, France), Carl G. Turesson, MD, PhD (Malmö, Sweden), Cor L. Verweij, PhD (Amsterdam, The Netherlands), Daniel Wendling, MD, PhD (Besançon, France), Jehsye-Hsien Yen, MD (Kaohsiung City, Taiwan), Steven A. Young-Min, BA, BMBCh, MRCP (Newcastle-upon-Tyne, UK).

Ancillary