ASSOCIATION OF HLA-DR WITH SUSCEPTIBILITY TO AND CLINICAL EXPRESSION OF RHEUMATOID ARTHRITIS: RE-EVALUATION BY MEANS OF GENOMIC TISSUE TYPING

Transcription

1 British Journal of Rheumatology 1998;37: ASSOCIATION OF HLA-DR WITH SUSCEPTIBILITY TO AND CLINICAL EXPRESSION OF RHEUMATOID ARTHRITIS: RE-EVALUATION BY MEANS OF GENOMIC TISSUE TYPING C. H. M. VAN JAARSVELD, H. G. OTTEN,* J. W. G. JACOBS, A. A. KRUIZE, H. L. M. BRUS and J. W. J. BIJLSMA Rheumatic Research Foundation Utrecht, Department of Rheumatology and Clinical Immunology and *Department of Immuno-Haematology, University Hospital Utrecht, PO Box 85500, 3508GA Utrecht and Department of Rheumatology, Hospital Hilversum, van Riebeeckweg 212, 1213XZ Hilversum, The Netherlands SUMMARY The clinical expression of rheumatoid arthritis ( RA) varies considerably among individual patients. Genetic variations in human leucocyte antigen ( HLA) may influence clinical expression. We re-examined the association of HLA-DR with susceptibility to and clinical expression of RA using genomic tissue typing, since most studies were based on ( less reliable) serological techniques. Seventy-eight patients with recent-onset RA, all participating in a clinical trial on therapeutic strategies, were HLA-DR typed by means of low-resolution genomic typing. Cumulative disease activity within the first 3 yr of disease was measured. Of the RA patients, 54% expressed DR4 (DR4+) vs 26% of healthy controls. Rheumatoid factor ( RF)-positive patients had a higher cumulative disease activity than RF-negative patients. Patients who were either DR1+ or DR4+ had a higher cumulative disease activity than those who expressed neither DR1 nor DR4. This association was less obvious after correction for RF status. The association of DR52+ (DR3, 5, 6) and a lower cumulative disease activity could also not be demonstrated after correction for RF status. Among RF-negative patients, DR51+ (or DR2+) was associated with a higher cumulative disease activity. Other HLA-DR types (including DR1 and DR4 separately) were not associated with the severity of RA. DR4 was associated with susceptibility to RA in our patients; HLA-DR low-resolution genomic tissue typing did not yield additional information to RF status for the clinical identification of individual patients with a poor prognosis. KEY WORDS: Rheumatoid arthritis, HLA-DR, Immunogenetics, Clinical expression. RHEUMATOID arthritis ( RA) is an autoimmune disease that between the presence of rheumatoid factor ( RF) in which lymphocytes and macrophages are assumed and a poor prognosis [8, 9]. to play a major role. Major histocompatibility complex The classical technique for HLA-DR typing is rough (MHC) molecules are essential for immune recogni- cellular leucocyte typing. A second technique is the tion. MHC is encoded in human leucocyte antigen serological method, which may yield 10 30% incorrect ( HLA) gene clusters on chromosome 6. Virtually all results when more than DR1 10 subtypes are studied autoimmune diseases studied show an association with [10]. Genomic techniques are more reliable, faster and some HLA specificity [ 1]. cheaper. In addition, the resolution improves: serolo- The third hypervariable region of the HLA-DR gical typing reliably identifies DR1 10, while the lowmolecule has been associated with susceptibility to RA resolution genomic technique is able to subdivide DR2, in several racial groups [2 5]. This shared epitope is 3, 5 and 6 each into two splits, which are called encoded by a number of alleles: DRB1 *0101, *0102, DR The high-resolution genomic technique can *0401, *0404, *0405, *0408, *1001, *1402 [1]. As a identify many subtypes; over 22 subtypes of DR4 have marker for RA in a general population, the shared been found [11]. In addition, DR51, 52 and 53 genes epitope has a specificity of 56%, a sensitivity of 95% can be identified which are located close to the DRB1 and a predictive value of 2% [1]. In spite of its indication of genetic susceptibility, this shared epitope genes on chromosome 6. Owing to the genomic organ- is not a useful diagnostic test for RA. ization, DR2 is associated with DR51; DR3, 5, 6 with Several studies report an association between the DR52; and DR4, 7, 9 with DR53 [1, 11]. presence of DR4 (DR4+) and the absence of DR2 In conclusion, most studies found an association of (DR2 ) and clinically severe RA ( Table I). However, DR4+, DR1+ or the so-called shared epitope with the findings are not consistent. All studies mentioned susceptibility to RA. Results on a possible association in Table I concern early RA patients, like our study of HLA-DR with the clinical expression of RA are population. In long-term RA patients, an association less consistent, which might partly be due to the between both DR4+ and DR1+ and severe RA is different typing techniques used. Differences in the described [6, 7]. The most consistent relationship is population studied may also be of importance. Since the association between RF and the clinical severity of RA is considered proven, we corrected for RF. The Correspondence to: C. H. M. van Jaarsveld, Department of aim of our study was to re-evaluate the association of Rheumatology and Clinical Immunology, University Hospital Utrecht, PO Box 85500, 3508 GA Utrecht, The Netherlands. HLA-DR with susceptibility to RA and its value for Submitted 10 April 1997; revised version accepted 3 September predicting individual disease activity (correcting for RF status), using low-resolution genomic tissue typing British Society for Rheumatology

2 412 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO. 4 TABLE I Review of the literature on the relationship between HLA-DR and severity of RA Follow-up HLA-DR DR4+/DR2 associated Discrepant Reference Population* (yr) typing with severe RA results [24] N = Ser DR2/Dw2, DR4/Dw4+ No association for men, or Dw4 homozygous{ patients over age 50 yr at more erosions onset, non-erosive disease [23] N = Ser DR4+{more severe RA women age (= higher joint/haq/steinbrocker <50 yr at onset score, more erosions/dmard use) [2] N = 87 3 Cel DR4 ~ unrelated to JD [9] N = Cel DR4+ or DR2 {more JD [25] N = 99 2 Ser DR4 homozygous{ DR4 ~ unrelated to disability more large JD and severe small JD [26] N = 58 2 Ser DR4 ~ unrelated to JD [27] N = Gen Dw4+ or Dw14+ (= DR4 variants){more JD *All study populations are patients with early RA. All studies are prospective. HLA-DR typing techniques: Cel, cellular leucocyte typing; Ser, serological typing; Gen, genomic tissue typing. JD, joint destruction. { indicates the presence of an association; ~ indicates the absence of an association. MATERIALS AND METHODS (PCR-SSP) [19]. In case only a single HLA-DRB1 Patients allele could be amplified, the alleles were assumed to Since 1990, all recent-onset RA patients (disease be present homozygously. The low-resolution genomic duration < 1 yr) seen at six rheumatological centres in typing was used which could identify DR1 DR18 and the Utrecht region of the Netherlands have been asked DR51, 52 and 53. No distinction between subtypes of to participate in a randomized trial to compare theraepitope could not be identified. In Caucasians, the DR1, 4, 10, 14 could be made, consequently the shared peutic strategies [12]. Of the first 100 patients, 78 blood samples for HLA-DR typing were collected by RA-related subtypes of DR1 and DR4 occur more rheumatologists for HLA typing within the period of frequently than those of DR10, DR14 and the other 1 yr. All patients but two were Caucasian and all subtypes of DR1 and DR4 [20, 21]; therefore, often fulfilled the 1987 ACR classification criteria for RA only DR1 and DR4 are studied. In our analyses, we [13]. combined DR1 and DR4 since relationships with both susceptibility and RA severity are described. Disease activity The RF test was positive if either the qualitative Five parameters of disease activity according to the latex fixation test at a dilution of 1:1 was positive or OMERACT recommendations were assessed [14]. the Rose Waaler test (titre 40 IU/ml ) was positive. Radiological damage was assessed after 3 yr of followup, RF status was determined at baseline. using the modified method of Sharp (range 0 448) [15]. The other parameters were measured every 3 Statistical analyses months during the first 2 yr and then every 6 months. Mann Whitney U non-parametric tests and odds Physical disability was assessed by means of a validated ratios (OR), with 95% confidence intervals (95% CI ), Dutch version of the Health Assessment Questionnaire were used to evaluate and estimate the differences (HAQ) (range 0 3) [16], pain score by a horizontal between patients with a specific HLA-DR type and visual analogue scale of 100 mm, joint score according clinical expression. Statistical analyses were carried out to Thompson (range 0 534) [17] and erythrocyte sedimentation using the SPSS/PC + statistical package [22]. rate ( ESR) by the Westergren method in mm/1st h. High values on all parameters indicate RESULTS more disease activity. A cumulative measure within Patient characteristics the first 3 yr was calculated for disability, pain, joint Of the 78 RA patients, 62% were female. At the score and ESR using the area under the curve method start of the study, the mean age was 54 yr (S.D. = 14) (AUC) [18]. For each individual patient, the AUC and 73% had a positive RF test. Mean disease activity provided a summary measure over time, i.e. cumulative scores at baseline were: a disability score of 1.3 (S.D. = disease activity, by adding up the serial scores. In the 0.7), a pain score of 44 mm (S.D. = 27), a joint score case of missing values, the time-weighted means of the of 151 (S.D. = 105), a radiological score of 4 (S.D. = previous and next measurements were taken. 5) and ESR of 46 mm/1st h (S.D. = 30). These baseline characteristics were comparable to those for the total HLA-DR typing population of the clinical trial (N = 439). All 78 RA HLA-DRB1 alleles were determined by the poly- patients were treated according to the study protocol merase chain reaction using sequence-specific primers [12]. At baseline, 32% were randomized to receive

3 VAN JAARSVELD ET AL.: HLA-DR AND RA SEVERITY IN DUTCH PATIENTS 413 treatment with methotrexate, 22% i.m. gold, 20% (P < 0.05). No differences were found for other DR hydroxychloroquine and 26% non-steroidal antiinflammatory types. drug (NSAID) only. In January 1996, In Table III, the associations between HLA-DR and when HLA-DR typing was carried out, 78% received clinical expression are shown for RF+ and RF a disease-modifying anti-rheumatic drug (DMARD) patients separately. Among RF+ patients, those with and 22% NSAID only. We focused on an association either DR1, DR4 or both had a higher cumulative between HLA-DR typing and the clinical expression joint score than those without DR1 and DR4 of RA, without correcting for treatment because medication (P < 0.05). No statistically significant differences were was randomized independent of HLA-DR found for the other parameters. Among RF patients, status and groups at baseline had comparable disease no association between DR1/DR4 and clinical expres- activity. sion was found; however, DR51+ was associated with a higher cumulative pain and joint score (P < 0.05). HLA-DR typing In contrast, radiological damage and cumulative ESR The frequency of HLA-DR types for our RA tended to be lower for DR51+ patients (not statistically patients was compared with that for 100 healthy controls significant). Other DR types were not associated with who had previously been typed in the same the clinical expression of RA in the first 3 yr, after laboratory. Eight RA patients ( 10%) and three controls correction for RF. We also performed multiple linear (3%) were homozygous for DR4. Forty-two RA and logistic regression analyses where appropriate patients (54%) and 26 controls (26%) had at least one (data not shown). These gave the same results: clinical DR4 allele. The percentage DR4 among RA patients expression is best predicted by RF status; HLA-DR was significantly higher (OR = 2.9, 95% CI = ), does not have an additional predictive value, nor do DR13 (a split of DR6) was lower (OR = 0.2, 95% other characteristics such as age and gender. CI = ). Likewise, DR53 (associated with DR4, 7, 9) was found more frequently among RA patients DISCUSSION (OR= 2.4, 95% CI = ), while DR52 (associ- The objective of this study was to re-evaluate the ated with DR3, 5, 6) was less common (OR = 0.4, role of HLA-DR in RA, using genomic tissue typing. 95% CI = ) compared to controls. Other For our Dutch patients with recent-onset RA, DR4 HLA-DR frequencies (including DR1) were comparable was associated with increased susceptibility to RA; the in RA and controls. A correction for the prepon- same applied for DR53 which is not surprising since derance of DR4 was made by excluding DR4+ people; DR53 is associated with DR4 (and DR7, 9). DR13 (a in this subgroup, the OR for DR1 was 1.1 (95% CI = split of DR6) and DR52 (associated with DR3, 5, 6) ). The OR for the association of having either were associated with a lower susceptibility to RA, DR1, DR4 or both with RA is 2.3 (95% CI = ). which also has been reported previously [23]. In contrast At baseline, clinical disease activity measures were not to reports in the literature, DR1 and DR2 could associated with HLA-DR typing or RF status. Gender and age at disease onset were also not related to RF status, DR4 or DR1. not be related to RA in our study. An association with the activity of RA was less obvious. Our results showed that patients with either DR1 or DR4, or both, had clinically more severe disease compared to patients without these DR types. Rheumatoid factor and clinical expression after 3 yr Table II shows that RF-positive ( RF+) patients had A positive RF test at baseline was closely associated suffered more radiological damage after 3 yr than with increased radiological damage after 3 yr RF-negative ( RF ) patients (P < 0.01). The other (P<0.01). The frequency of DR1+ or DR4+ did not clinical variables also indicated that RF+ patients, as differ between RF+ and RF patients. We studied the did females, exhibited higher cumulative disease activ- association between HLA-DR and clinical expression, ity, but these differences did not reach statistical significance. corrected for RF status, since RF status is, in several RF status was significantly related to DR52; studies, a confirmed predictor of severity. Results, among RF+ patients, DR52 was present more often presented in Table III, showed that among RF+ (P = 0.03). RF status and other DR types were not patients, the presence of either DR1 or DR4, or both, related, e.g. in RF+ and RF patients, respectively, was associated with a higher cumulative joint score. 58 and 43% were typed as DR4+ (P = 0.24). No associations were found with the other parameters. For RF patients, no association between DR1/DR4 HLA-DR typing and clinical expression after 3 yr and clinical expression was found. Therefore, we con- Cumulative disease activity after 3 yr was not associ- clude that DR1/DR4 is not clearly associated with the ated with DR4, nor with DR1 (Table II). After correction clinical expression of RA in the first 3 yr after disease for the preponderance of DR4, DR1+ was related onset, after correction for RF status. The association to more radiological damage (P = 0.03). Patients who between the presence of DR51 (DR2) and a higher were either DR1+ or DR4+, or both (n = 53), had pain and joint score in RF patients is remarkable, more radiological damage, a higher cumulative joint and in contrast with the literature [9, 24]. Even more score and a higher cumulative ESR than patients who radiological damage and cumulative ESR indicated the were both DR1 and DR4 (n = 25) (P < 0.05). opposite (not significant). DR52+ was associated with DR52+ patients had lower cumulative pain scores a higher cumulative pain score; this association was

4 414 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO. 4 TABLE II Differences in cumulative disease activity (3 yr) according to gender, rheumatoid factor (RF) test and HLA-DR typing Radiological Cumulative Cumulative Cumulative Cumulative N damage disability pain score joint score ESR Median Range Female Male RF+ at baseline RF at baseline 21 10** HLA-DR DR DR DR DR DR4 & DR DR4 & DR1 25 9* DR1+ or DR DR1 & DR4 25 8* * 295* DR DR DR DR * DR DR All variables are expressed in medians because of skewed distributions of the data. Analysis of DR1 excluding DR4+ patients. *P < 0.05; **P < TABLE III Differences in cumulative disease activity (3 yr) according to HLA-DR typing, after correction for rheumatoid factor test Radiological Cumulative Cumulative Cumulative Cumulative N damage disability pain score joint score ESR Rheumatoid factor positive (N = 57) DR1+ or DR DR1 & DR * 326 DR DR DR DR DR DR Rheumatoid factor negative (N = 21) DR1+ or DR DR1 & DR DR DR * 403** 268 DR DR DR DR All variables are expressed in medians because of skewed distributions of the data. RF status and DR1/DR4 expression are not significantly related: P value x2 test = 0.49; DR51 and DR53 expression are also not significantly related with RF status. RF status and DR52 expression are significantly related. Among RF-positive patients 35% are DR52+, while among RF-negative patients 62% are DR52+: P value x2 test = *P < 0.05; **P < not apparent after correction for RF status. DR52 and RF status were significantly correlated which might explain the association in the unstratified analyses. Other DR types showed no statistically significant association with clinical expression. Since results in the unstratified analyses differed from those in RF+ and RF patients, RF might be regarded as an effect modifier in epidemiological terms. This suggests that in RF+ patients the relationship of HLA-DR and RA differs from that in RF patients. However, in the unstratified analyses, DR1/DR4 was significantly related to three of the five parameters, while after stratification for RF status it was only significantly related to joint score. Multiple linear and logistic

5 VAN JAARSVELD ET AL.: HLA-DR AND RA SEVERITY IN DUTCH PATIENTS 415 regression analyses yielded the same conclusion: clin- REFERENCES ical expression is best predicted by RF status, and 1. Winchester R. The molecular basis of susceptibility to HLA-DR has no additional value as predictor. rheumatoid arthritis. Adv Immunol 1994;56: Although treatment was random and independent 2. Paimela L, Leirisalo Repo M, Helve T, Koskimies S. of HLA-DR, response to treatment may not have been The prognostic value of HLA DR4 and B27 antigens in independent of HLA-DR. In some studies HLA-DR early rheumatoid arthritis. Scand J Rheumatol is found to be associated with the occurrence of side- 1993;22: effects rather than with response to treatment. Taking 3. Wordsworth BP, Stedeford J, Rosenberg WM, Bell JI. Limited heterogeneity of the HLA class II contribution this into account, different treatments are not likely to to susceptibility to rheumatoid arthritis is suggested by have influenced our results. positive associations with HLA-DR4, DR1 and DRw10. Since an association of HLA-DR with the clinical Br J Rheumatol 1991;30: expression of RA was lacking in our study, the follow- 4. Jeng-Hsien Y, Jong-Rern C, Wen-Juun T, Jih-Jin T, ing aspects should be considered: different study popu- Hong-Wen L. HLA-DRB1 genotyping in patients with lations, different typing techniques and the power of rheumatoid arthritis in Taiwan. J Rheumatol the study (number of patients). We studied early-onset 1995;22: RA patients drawn from a population-based study, 5. Ollier WER, Stephens C, Awad J et al. Is rheumatoid which might give different results than a study of arthritis in Indians associated with HLA antigens sharing adrb epitope? Ann Rheum Dis 1991;50: patients with prolonged RA. We used reliable genomic 1 6. Stockman A, Emery P, Doyle T, Hopper J, Tait B, typing, suggesting that the weak associations between Muirden K. Relationship of progression of radiographic HLA-DR and RA found in other studies reflect the changes in hands and wrists, clinical features and true situation and are not due to the use of less reliable HLA-DR antigens in rheumatoid arthritis. J Rheumatol typing techniques, as we hypothesized in the 1991;18: Introduction. The numbers of patients for the analyses 7. Moreno I, Valenzuela A, Garcia A, Yelamos J, Sanchez by RF status were 57 RF+ and 21 RF patients, B, Hernanz W. Association of the shared epitope which might introduce a power problem. However, the with radiological severity of rheumatoid arthritis. aim of our study was to evaluate the possibility of J Rheumatol 1996;23: van der Heide A, Remme CA, Hofman DM, Jacobs using HLA-DR to predict the course of the disease for JWG, Bijlsma JWJ. Prediction of progression of radioindividual patients. If HLA-DR is clinically relevant, logic damage in newly diagnosed rheumatoid arthritis. such an effect should have become apparent during Arthritis Rheum 1995;38: our study, especially since we used a reliable HLA-DR 9. van der Heijde DM, van Riel PLCM, van Leeuwen MA, typing technique. Some associations between HLA-DR van t Hof MA, van Rijswijk MH, van de Putte LBA. and RA expression are found in our data; an effect of Prognostic factors for radiographic damage and physical HLA-DR at the subgroup level might therefore be disability in early rheumatoid arthritis. A prospective present. However, at a subgroup level, the effects of follow-up study of 147 patients. Br J Rheumatol HLA-DR are not likely to be large, since such effects 1992;31: Otten HG, Tilanus MGJ, Barnstijn M, van Heugten JG, should have been observed in our data; only for small Gast de GC. Serology versus PCR-SSP in typing for effects is the power of our study not sufficient. HLA-DR and HLA-DQ: a practical evaluation. Tissue We confirmed the association of DR4 with suscepti- Antigens 1995;45: bility to RA, but we were not able to demonstrate that 11. Bodmer JG, Marsh SGE, Albert ED et al. Nomenclature HLA-DR has an important influence on the clinical for factors of the HLA system, Vox Sang expression of RA for individuals. We suggest that 1995;69: HLA-DR typing by means of the low-resolution geneffectiveness of early treatment with second-line anti- 12. van der Heide A, Jacobs JWG, Bijlsma JWJ et al. The omic typing technique does not add to the value of rheumatic drugs. A randomized, controlled trial. Ann RF status for predicting individual disease activity in Intern Med 1996;124: the first 3 yr for early-onset patients. 13. Arnett FC, Edworthy SM, Blog DA et al. The American Rheumatism Association 1987 revised criteria for the ACKNOWLEDGEMENTS classification of rheumatoid arthritis. Arthritis Rheum 1988;31: The authors thank G. A. van Albada-Kuipers, A. 14. van Riel PLCM. Disease assessment and outcome in A. M. Blaauw, C. van Booma-Frankfort, E. J. ter rheumatoid arthritis. Rheumatol Eur 1994;23: Borg, H. C. M. Haanen, A. H. M. Heurkens, D. M. 15. Sharp JT, Young DY, Bluhm GB et al. How many joints Hofman, Y. Schenk and M. J. van der Veen, rheumato- in the hands and wrists should be included in a score of logists of the departments of rheumatology of radiologic abnormalities used to assess rheumatoid arth- University Hospital Utrecht, Sint Antonius Hospital ritis? Arthritis Rheum 1985;28: Nieuwegein, Diakonessen Hospital Utrecht, Eemland 16. Bijlsma JWJ, Oude Heuvel CHB, Zaalberg A. Hospital Amersfoort, Hospital Hilversum and Sint Development and validation of the Dutch questionnaire capacities of daily life (VDF) for patients with rheumat- Jansdal Hospital Harderwijk, joined in the Rheumatic oid arthritis. J Rehabil Sci 1990;3:71 4. Research Foundation Utrecht, The Netherlands, for 17. Thompson PW, Silman AJ, Kirwan JR, Currey HLF. their contributions to this paper. Supported by a grant Articular indices for joint inflammation with rheumatoid from the Dutch League against Rheumatism ( Het arthritis. Correlation with the acute-phase response. Nationaal Reumafonds). Arthritis Rheum 1987;30:

6 416 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO Matthews JNS, Altman DG, Campbell MJ, Royston P. Results of a followup study. Arthritis Rheum 1991;34: Analysis of serial measurements in medical research. Br Med J 1990;300: Young A, Jaraquemada D, Awad J et al. Association of 19. Olerup O, Zetterquist H. HLA-DR low resolution HLA-DR4/Dw4 and DR2/Dw2 with radiological PCR-SSP typing a correction and an update. Tissue changes in a prospective study of patients with rheumat- Antigens 1993;41:55 6. oid arthritis. Arthritis Rheum 1984;27: Nepom GT, Erlich H. MHC class-ii molecules and 25. Eberhardt K, Grubb R, Johnson U, Petterson H. HLA-DR antigens, Gm allotypes and antiallotypes in autoimmunity. Annu Rev Immunol 1991;9: early rheumatoid arthritis: their relation to disease pro- 21. Nepom GT, Byers P, Seyfried C et al. HLA genes gression. J Rheumatol 1993;20: associated with rheumatoid arthritis. Arthritis Rheum 26. Möttönen TT. Prediction of erosiveness and rate of 1989;32: development of new erosions in early rheumatoid arth- 22. Norusis MJ. SPSS/PC+TM base manual. Chicago: SPSS, ritis. Ann Rheum Dis 1988;47: Inc., Gough A, Faint J, Salmon M et al. Genetic typing of 23. Zeben van D, Hazes JM, Zwinderman AH et al. patients with inflammatory arthritis at presentation can Association of HLA-DR4 with a more progressive be used to predict outcome. Arthritis Rheum disease course in patients with rheumatoid arthritis. 1994;37: