Published ahead of print on December 4, 2003, doi:10.1164/rccm.200303-459OC
© 2004 American Thoracic Society
Human Leukocyte Antigen Class I Alleles and the Disease Course in Sarcoidosis PatientsDepartment of Medicine, Division of Respiratory Medicine, Karolinska Hospital; and Division of Clinical Immunology, Huddinge University Hospital, Huddinge and Karolinska Institutet, Stockholm, Sweden Correspondence and requests for reprints should be addressed to Johan Grunewald, M.D., Ph.D., Department of Medicine, Division of Respiratory Medicine Lung Research Laboratory L4:01, Karolinska Hospital, S-171 76 Stockholm, Sweden. E-mail: johan.grunewald{at}medks.ki.se
Several lines of evidence suggest a genetic predisposition to sarcoidosis, and strong associations have been shown with the major histocompatibility complex gene complex. In this study on Scandinavian sarcoidosis patients, we investigated any influence on the outcome of disease by human leukocyte antigen (HLA) class I alleles alone and in combination with selected class II alleles. HLA-B*07 independently increased the risk for persistent sarcoidosis (odds ratio [OR], 1.9; 95% confidence interval [CI], 1.0–3.7), as well as for resolving disease (OR, 2.7; CI, 1.1–6.2), suggesting an influence on factors common to both forms of sarcoidosis. The common allele combination A*03, B*07, DRB1*15 was most strongly associated with persistent disease (OR, 4.7; CI, 2.2–10.2) and was found in 25.3% of patients with persistent disease versus 7.1% of healthy control subjects. HLA-B*08 tended to increase separately the risk for resolving disease (OR, 2.4; CI, 0.7–8.0), as well as for persistent disease (OR, 2.2; CI, 0.8–6.1). Other HLA class I associations were mainly secondary to their linkages to DRB1*03 and DRB1*15, respectively. The influence of HLA class I alleles on sarcoidosis thus seems more pronounced than previously thought, and both HLA class I and class II should be relevant to evaluate in the clinical management of sarcoidosis patients.
Key Words: human leukocyte antigen • sarcoidosis • prognosis Sarcoidosis is a granulomatous disorder that is characterized by distinct immunopathologic features such as lung accumulated, activated, and oligoclonally derived Th1 cells (1). Such Th1 cells are considered to interact with one or a limited number of antigens, and they eventually may target autoantigens and initiate an autoimmune inflammatory reaction (2). One key event in the activation of T cells is the interaction between the T-cell receptor for antigen expressed on the T lymphocyte on the one hand and the HLA molecule with its antigenic peptide, expressed on antigen-presenting cells, on the other. The two major classes of HLA molecules have different capacities and functions, as HLA class I molecules primarily present self antigens (3), whereas the HLA class II molecules mainly present externally derived antigens (4). A genetic predisposition to sarcoidosis is suggested by different clinical appearances in distinct ethnic groups and by familial clustering (5, 6). The HLA genes seem to contribute substantially for this inheritable predisposition (7–10), although an influence by other genes may also be important (11–13). In line with an important role for HLA genes in the susceptibility to sarcoidosis, a large number of HLA class I and class II alleles have been reported to be overrepresented in sarcoidosis. Especially HLA-B8 and DR3 were early recognized to associate with sarcoidosis in general and/or with acute manifestations of the disease such as EN and ankle arthritis, that is, features linked to a good prognosis (14–19). All of these studies were performed on small numbers of sometimes heterogeneous patients and used serologic typing techniques, which may result in as much as 25% misinterpretations (20, 21). More recent studies instead applied polymerase chain reaction techniques, giving more correct classifications and also identifying subtypes of allelic variants (22). Our own previous studies revealed a strong association between good prognosis and HLA-DRB1*03 (serologically DR3) and between persistent disease and DRB1*15 (serologically DR2) in Scandinavian sarcoidosis patients (23, 24). A recent study corroborated these findings, as strong associations between good prognosis and DQB1*0201 (in strong linkage disequilibrium with DRB1*03) and between persistent disease and DQB1*0602 (strongly linked to DRB1*15) were presented (25).
However, genes other than HLA class II are also of importance for the disease. Recent studies have explored for example the relationship between HLA-DR3, tumor necrosis factor (TNF)- In agreement with our previous report on HLA and sarcoidosis (23), there was a very strong correlation between HLA-DRB1*03 and good prognosis and between DRB1*15 and a persistent disease. Focusing on these two HLA-DRB1 alleles, we revealed several HLA class I alleles to be able to modulate these associations. In addition, we found distinct combinations of HLA class I and HLA class II alleles to be strongly associated with resolution or persistence of the disease.
Study Subjects We investigated 166 consecutive Scandinavian sarcoidosis patients attending the outpatient clinic at the Karolinska Hospital, Stockholm, Sweden. There were 100 males and 66 females, with a median age of 37 (interquartile range of 32–45) years. One hundred ten patients were never smokers, 13 ex-smokers, and 32 current smokers. Every patient presented a clinical picture in accordance with sarcoidosis, often with symptoms such as fatigue, dyspnea on exertion, and dry coughing. The diagnosis was further established with a positive biopsy (n = 111) or, in case of a negative or absent biopsy, through typical disease manifestation as Löfgren's syndrome (n = 69), an elevated bronchoalveolar lavage (BAL) fluid CD4/CD8 ratio (more than 4.0; n = 38/70), or with an extensive clinical follow-up for more than 5 years (n = 5), with disease manifestations constantly being characteristic for sarcoidosis. When patients presented with bilateral hilar lymphadenopathy, fever, ankle arthralgia, and erythema nodosum, they were regarded as having Löfgren's syndrome. A few patients with fever, ankle arthritis, and bilateral hilar lymphadenopathy but without erythema nodosum were also classified as having Löfgren's syndrome, as they had the same clinical behavior and the same pattern of HLA alleles as the other patients with Löfgren's syndrome. Patient characterization further included chest radiographic staging (0, I, II, or III) at disease onset and evaluation of BAL CD4/CD8 ratio and lung function test (VC, FEV1, and diffusing capacity) at diagnosis (Table 1) .
In 156 of the 166 patients, clinical resolution defined as disappearance of symptoms (as dry cough, fever, fatigue, and arthralgia) and normalization of chest x-ray (without any signs of enlarged lymph nodes or parenchymal infiltrates) and of pulmonary function tests (more than 80% of predicted value for diffusing capacity, VC, and FEV1) was evaluated 2 years after diagnosis. Every patient was categorized as either having a persistent (with ongoing disease at 2 years; this group of patients include both those with a chronic stable as well as those with a chronic progressive disease) or a resolved (with no signs of sarcoidosis at 2 years) disease. All included subjects gave their informed consent for participation in the study, which was approved by the local ethics committee at the Karolinska Hospital.
BAL, BAL Fluid Cell Preparation, and Flow Cytometry
Preparation of DNA from Whole Blood Cells and HLA Typing Only HLA class I alleles found in more than 10% of the patients and control subjects are shown in the result section of the article, whereas the complete list of HLA class I alleles is shown in the online supplement.
Pulmonary Function Tests
Statistical Analyses To evaluate the effect of different alleles on persistent and resolving sarcoidosis when the effects of the other alleles is taken in account, a multiple logistic regression model was fitted for each patient group. The response is then whether a person is a case or a control, and the coefficient for a particular allele is the estimated odds ratio (OR) of being a case rather than a control if this allele is present compared with if it is not. In addition to the estimated OR, we calculated a 95% confidence interval and a p value for testing the hypothesis that the OR is equal to 1. Data are presented as adjusted (from the multiple regression model) and nonadjusted ORs for each patient group. In addition, ORs for certain HLA class I alleles were calculated in DRB1*0301- and DRB1*15-positive and -negative individuals. The statistical analyses were done by using the Stata statistical software (Stata 6.0 for Windows; Stata Corporation, College Station, TX).
Patient Characterization Most patients were nonsmokers (n = 110) and nontreated (n = 122). Chest radiographic stage I (n = 83) and stage II (n = 57) were most frequent (Table 1). Bronchoscopy with BAL and determination of the BAL CD4/CD8 ratio was done on 70 patients, with a median value of 5.0 (2.3–8.2). Pulmonary lung function tests showed that one-third of the patients had a diffusing capacity value of less than 80% of predicted (Table 1). Among 157 of the patients, the disease onset was possible to categorize as acute or insidious. Seventy-three (46%) patients had an acute onset, whereas 84 (54%) had an insidious onset. Sixty-nine (44%) patients fulfilled the criteria for classic Löfgren's syndrome. The disease course was evaluated in 156 of the patients 2 years after disease onset and found to be persistent in 87 (56%) and resolving in 69 (44%) patients.
HLA-DRB1*03, DRB1*15, DQB1*0201, and DQB1*0602 Alleles
HLA-A, HLA-B, and HLA-Cw Alleles HLA-A*01 was clearly overrepresented in the resolving disease group of patients (49.3%) compared with control subjects (25.7%, p < 0.001, corrected p = 0.01). In contrast, HLA-A*03 associated with the persistent disease group (36.8%) compared with control subjects (25.2%, p < 0.05, pc = ns) (Table 3) .
HLA-B*07 was significantly more common in the complete patient group (40.4%) versus control subjects (26.7%, p < 0.01, pc = ns) and significantly biased toward patients with persistent disease (49.4%, p < 0.001, pc = 0.01 vs. control subjects) (Table 3). Also HLA-B*08 was overrepresented in the entire patient group (36.1% vs. 19.5% in the control subjects, p < 0.001, pc = 0.02), reflecting a dramatic overrepresentation in the resolved patients (63.8%, p < 0.00001, pc < 0.00001 vs. control subjects) (Table 3). HLA-Cw*03 was present significantly less in the entire patient group (28.7%), as well as in the resolving disease group (25.5%) versus the control group (38.6%, p < 0.001, pc = 0.02, and p < 0.01, pc = ns, respectively). Also HLA-Cw*04 was underrepresented in the entire patient group (12.4%, p < 0.01, corrected p = ns) versus control subjects (20.0%).
HLA Class I and Class II Combinations According to the regression model, HLA-B*07 separately increased the risk for persistent disease (adjusted OR, 1.9 [1.0–3.7]; p = 0.05) (Table 4) . Moreover, HLA-B*07, which was not overrepresented in the resolving patient group, significantly increased the risk also for a resolving disease (adjusted OR, 2.7 [1.1–6.2]; p = 0.02) (Table 4). There was also a trend for HLA-B*08 to influence separately the risk for a resolving disease (adjusted OR, 2.4; 0.7–8.0) (Table 4) as well as for persistent disease (adjusted OR, 2.2; 0.8–6.1) (Table 4). Although HLA-A*03 had a significantly elevated nonadjusted OR for persistent disease, this risk was reduced when applying our regression model (adjusted OR). HLA-DRB1*15 was as expected found to increase significantly the risk for persistent disease (adjusted OR, 2.1; 1.1–3.7; p = 0.02; Table 4), whereas DRB1*03 was a risk factor for a resolving disease, with a high adjusted OR (11.8).
To explore any associations further between the HLA class I alleles and their closely linked respective HLA class II alleles, we calculated ORs for HLA-A*01 and HLA-B*08 in DRB1*03-positive and -negative individuals and for HLA-A*02, HLA-A*03, and HLA-B*07 in DRB1*15-positive and -negative individuals. There were significant OR values for persistent disease for HLA-A*03 and B*07 in DRB1*15-positive but not DRB1*15-negative individuals (Table 5) , although the difference between DRB1*15-positive and -negative individuals was not significant.
We also calculated any influence on persistent disease by the combined A*03, B*07, and DRB1*15 alleles (DRB1*15, B*07, A*03 is a common haplotype) and found that individuals with the allele combination A*03, B*07, and DRB1*15 had the most pronounced increased risk for a persistent disease, with an OR of 4.7 (2.2–10.2). HLA-A*03-, B*07-, and DRB1*15-positive patients made up 25.3% of all patients with persistent disease compared with 7.1% of healthy control subjects. Finally, among all DRB1*03-positive patients (n = 61), as many as 53 (87%) developed a resolving disease course. Thirty three of the DRB1*03-positive patients were HLA-A*01 positive, and 32 of these patients (97%) recovered within 2 years. All of these patients were HLA-B*08 positive, and 100% (32 of 32) were categorized as resolved. This group of patients constituted more than 20% of all patients in this study.
HLA Alleles and Clinical Manifestations There was an association between Löfgren's syndrome and HLA-DRB1*03, as 47 of the 69 patients (68%) with Löfgren's syndrome were HLA-DRB1*03 positive, compared with approximately 40% in the entire patient group (Table 2). Among DRB1*03-positive patients, 77% (47 of 61) had Löfgren's syndrome. Of 32 patients with the A*01, B*08, DRB1*03-deduced haplotype and pulmonary sarcoidosis, 29 (91%) had Löfgren's syndrome. Interestingly, the combination of Löfgren's syndrome and HLA-DRB1*03 gave an extremely good prognosis, as 98% of DRB1*03-positive patients with Löfgren's syndrome (n = 47) had a resolving disease compared with only 67% of DRB1*03-negative patients with Löfgren's syndrome (n = 21) (Figure 1) .
After excluding patients with Löfgren's syndrome, the distribution of the HLA class II alleles was assessed among the rest of the patients with either a resolving or a persistent disease. Almost identical to the entire patient group, HLA-DRB1*03 was found in 78% (7 of 9) patients with resolving disease, but only in 9.9% (7 of 71) of the patients with persistent disease. HLA-DRB1*15 was found in only 22% (2 of 9) of patients with resolving disease but in as many as 52% (37 of 71) of the patients with a persistent disease.
The aim of this study was to investigate whether distinct HLA class I alleles could influence the genetic susceptibility and the disease course in sarcoidosis and, if so, whether such an influence depended on HLA class II alleles. The strength of this study was the relatively large number of homogeneous patients that was closely characterized for disease manifestations and followed for at least 2 years, and the usage of the most recent polymerase chain reaction–based technique for HLA typing with a large number of allele specific primers. Because of strong associations with distinct HLA alleles in different subgroups of patients with sarcoidosis, that is, those with a persistent versus those with a resolving form of the disease, the patient selection will influence the results in studies of HLA-associated risks with sarcoidosis. The relative number of each subgroup therefore needs to be similar to enable comparisons between different studies. In this study, we had a high degree of patients with the acute form of the disease, as 44% had Löfgren's syndrome. We have here focused on studying associations between HLA class I and class II antigens, alone and in selected combinations, with the disease outcome. The ancestral MHC 8.1 haplotype (A1, B8, DR3, DQ2) associates with a number of autoimmune disorders such as insulin-dependent diabetes mellitus and interestingly seems to be linked to an exaggerated type of immune response (31). This particular haplotype was also previously suggested to associate in particular with acute sarcoidosis and a good prognosis (17, 18, 32, 33). This study, including a larger number of patients and more accurate method (20, 21), corroborates and strengthens these previous observations, as 100% of patients with the HLA-A*01, B*08, DRB1*03-deduced haplotype recovered within 2 years. Interestingly, this haplotype is linked to polymorphisms in the MHC III region, including allelic variants of TNF genes with a known influence on the TNF production (34, 35). An association between HLA-DR3 and the TNFA2 allele has been reported (35), and both alleles were found to associate with Löfgren's syndrome, that is, a good prognosis (11, 12, 36). However, yet other genes, possibly in linkage disequilibrium, also contribute to the risk for Löfgren's syndrome (26, 37). Our results show that the strong association of HLA-A*01 with resolving sarcoidosis seemed to be secondary to its link to DRB1*03, whereas HLA-B*08 had a tendency to separately increase the risk for resolving sarcoidosis. HLA-DRB1*03–positive individuals had a highly elevated risk for a resolving form of sarcoidosis (OR, 11.8). Interestingly, although only 19 patients were investigated, Hedfors and Lindström also calculated DR3-positive individuals to have a highly elevated risk for acute sarcoidosis (OR, 22.4) (17). On the other hand, when the disease outcome in DRB1*03-positive patients was investigated, an extremely favorable prognosis was found for DRB1*03-positive patients with HLA-A*01 and B*08 (100%). HLA-B*07 is in linkage disequilibrium with DRB1*15, and accordingly, the (nonadjusted) OR for persistent disease was reduced when applying our multiple regression analysis, that is, when taking this association into consideration in our statistical calculations. However, also, the adjusted OR was significant, and thus, B*07 is an independent risk factor for persistent sarcoidosis. Especially DRB1*15-positive individuals had a significantly increased risk for persistent disease if they were B*07 positive. The A*03-associated risk in DRB1*15-positive individuals was on the other hand most likely caused by its close association with B*07, as A*03 did not turn out as a separate risk factor according to our multiple logistic regression model. Furthermore, the common combination of A*03, B*07, and DRB1*15 was found to confer the highest risk for persistent sarcoidosis, and these patients made up a considerable part of all patients with persistent disease (25.3%). It is interesting that the same alleles were recently shown to increase significantly the risk for multiple sclerosis (27).
HLA-B*07 was in contrast to DRB1*15 found to increase independently the risk also for resolving sarcoidosis. A similar trend was observed for HLA-B*08, that is, B*08 independently and in sharp contrast to DRB1*03 tended to increase the risk for a persistent disease. Especially HLA-B*07, but to some extent also B*08, may thus predispose for features common to the different forms of sarcoidosis, that is, most likely early events in the "sarcoid" type of inflammation, including initial steps toward granuloma formation. In this regard, B*08 has been associated with an increased TNF- We confirmed our previously reported strong HLA class II associations with the development of a persistent disease (HLA-DRB1*15) or to resolution of the disease (DRB1*03) (23, 24). These results are in congruence with the reported associations between DQB1*0602 and persistent disease and between DQB1*0201 and good prognosis because of the strong linkage disequilibrium between DQB1*0602 and DRB1*15, respectively, between DQB1*0201 and DRB1*03 (23, 25). Interestingly, DRB1*03 was in this report not only found to be a risk factor for resolving disease, but also to protect against persistent sarcoidosis. Our previous findings of lung-accumulated TCRAV2S3+ CD4+ T cells in HLA-DRB1*03, DQB1*02-positive patients indicate the HLA-DR/DQ molecule itself to be important, with its antigen peptide-presenting capacities, although other immunomodulatory genes that are linked to the HLA regions may also be relevant for the disease associations (44). Our recent findings of TCRAV2S3+ BAL cell accumulated CD4+ T cells not only in DRB1*0301 but also in DRB3*0101-positive sarcoidosis patients would favor the HLA-DR genes to be important for the disease association because these specific DRB1 and DRB3 alleles show strong and in certain aspects unique similarities rendering them similar antigen-presenting capacities (45, 46). There was an overrepresentation of HLA-DRB1*03 in patients with Löfgren's syndrome, and in particular, among patients with the A*01, B*08, DRB1*03-deduced haplotype, Löfgren's syndrome was very common (91%). In general, patients with Löfgren's syndrome had a very good prognosis, similar to patients with HLA-DRB1*03. Interestingly, a synergistic and positive effect on the clinical outcome was noticed when these two groups were combined, that is, DRB1*03-positive patients with Löfgren's syndrome. In fact, DRB1*03-negative patients with Löfgren's syndrome did not at all have the same degree of recovery, supporting the idea that HLA typing gives additional information on the prognosis of the disease, also in patients with Löfgren's syndrome. In addition, our finding of almost identical distributions of HLA-DRB1*03 and DRB1*15 in "non-Löfgren's syndrome patients" as in the entire patient group, with similarly strong associations between DRB1*03 and resolving disease and between DRB1*15 and persistent disease, suggests that these HLA alleles are primarily important for the disease course. In conclusion, the main finding in this study is the significant contribution of HLA class I superimposed on the HLA class II–modulated disease risk. For the first time, distinct HLA class I alleles and combinations of HLA alleles are described to modify especially the HLA-DRB1*15–associated risk for a persistent course of the disease. HLA-A*03, B*07, DRB1*15–positive individuals were found to associate with persistent sarcoidosis most strongly. We also observed that HLA-A*01, B*08, DRB1*03–positive patients, making up as many as 20% of the investigated Scandinavian patients, had an extremely good prognosis. We conclude that HLA typing for both class I and class II is of clinical relevance to help in the management of Scandinavian sarcoidosis patients.
The authors thank Benita Dahlberg, Margita Dahl, Gunnel De Forest, and Beatrice Sjöberg for their assistance. Associate professor Jan-Olov Persson, Stockholm University, is acknowledged for his expertise and contributions with statistical calculations.
Supported by the Swedish Heart–Lung Foundation, the King Oscar II Jubilee Foundation, the Swedish Medical Research Council (K2002–74X-14182–01A), and the Karolinska Institute. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: J.G. has no declared conflict of interest; A.E. has no declared conflict of interest; O.O. has no declared conflict of interest. Received in original form March 31, 2003; accepted in final form December 2, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
alleles, and Löfgren's syndrome (
2 test. 
