Design: Case-control study.

Setting: Kuma Hospital, which specializes in thyroid diseases, in Kobe, Japan.

Subjects: 24 patients with Graves disease who had methimazole-induced agranulocytosis diagnosed by peripheral granulocyte counts of less than 0.5 x 10⁹/L, and 68 patients with Graves disease treated with methimazole, who were free from agranulocytosis. Controls were 525 healthy, unrelated Japanese student volunteers at Kyushu University in Japan.

Measurements: All HLA class II genes were analyzed for polymorphisms at the DNA level by using the polymerase chain reaction sequence-specific oligonucleotide probes method. The allele frequencies in the agranulocytotic Graves disease group were compared with those in the nonagranulocytotic Graves disease and control groups.

Results: A strong positive association was seen in DRB1*08032 between the agranulocytotic group and both the control and nonagranulocytotic Graves disease groups.

Conclusion: The HLA DRB1*08032 allele was strongly associated with susceptibility to methimazole-induced agranulocytosis, suggesting that cellular autoimmunity may be involved in its development.

The thioureylene antithyroid drugs methimazole and propylthiouracil have been widely used to treat Graves disease [1], but side effects are found in 1% to 5% of treated patients [2-4]. One important and serious side effect of this treatment is agranulocytosis, which occurs in 0.1% to 0.3% of treated patients [5-10]. Although the mechanisms responsible for the agranulocytosis are unclear, an immune phenomenon may be involved, because antigranulocyte antibodies [11-20] or lymphocyte sensitization to antithyroid drugs [17] are found in agranulocytotic patients. These autoantibodies may induce agranulocytosis by direct cytotoxicity or through blocking of membrane proteins that are important for the maturation of progenitor cells [19].

Antibodies are generally produced by plasma cells, which are maturated from B lymphocytes in the presence of various cytokines produced by activated CD4⁺ T lymphocytes and by direct T lymphocyte-B lymphocyte interactions [21]. CD4⁺ T lymphocytes are activated when they encounter antigenic peptide with major histocompatibility complex class II on antigen-presenting cells, such as B lymphocytes and macrophages. The recognition of major histocompatibility complex class II peptide complexes by T lymphocytes is therefore central to the development of immune responses and antibody production. The major histocompatibility complex class II molecules are highly polymorphic heterodimeric membrane glycoproteins composed of and ß chains. In humans, at least three major histocompatibility complex class II molecules—HLA-DR, HLA-DQ, and HLA-DP antigens—are expressed on the surface of antigen-presenting cells. In addition, these molecules are encoded by the genes in the HLA region: DRA, DQA, and DPA for chains and DRB, DQB, and DPB for ß chains of each HLA molecule. The function of major histocompatibility complex class II molecules is to bind short peptides derived mainly from extracellular proteins, in turn forming major histocompatibility complex class II peptide complexes that interact with appropriate T-cell receptors of CD4⁺ T lymphocytes [22]. Each major histocompatibility complex class II molecule binds different sets of proteolytic fragments of peptides, thus contributing to the diversity of immune responsiveness among individual persons [23-26]. Identification of susceptible HLA class II antigens to antithyroid drug-induced agranulocytosis is essential to understanding the development of the disorder [27, 28].

Methods

Top Methods Results Discussion Author & Article Info References

Participants

We studied 24 patients with Graves disease and methimazole-induced agranulocytosis, 68 patients with Graves disease and no agranulocytosis who were treated with methimazole, and 525 healthy unrelated student volunteers from Kyushu University. The agranulocytotic group consisted of 5 males and 19 females ranging in age from 16 to 68 years (mean, 38.4 years). Agranulocytosis was defined as peripheral granulocyte counts less than 0.5 x 10⁹/L. The nonagranulocytotic Graves disease group consisted of 10 males and 58 females with ages ranging from 11 to 58 years (mean, 32.4 years). Graves disease was diagnosed on the basis of clinical symptoms, thyroid function test results, positivity for thyroid-stimulating hormone-binding inhibitory immunoglobulin, and ¹²³I uptake. Blood samples were taken after informed consent was obtained.

DNA Typing of HLA Class II Genes

We extracted DNA from peripheral granulocytes using a previously described method [28]. Genomic DNA was subjected to 30 cycles of polymerase chain reaction in a thermal cycler (Perkin Elmer Cetus, Norwalk, Connecticut) to amplify the second exons of the DRB1, DQA1, DQB1, DPA1, and DPB1 genes using thermostable DNA polymerase (Ampli Taq, Perkin Elmer Cetus) [29]. The primers and sequence-specific oligonucleotide probes for the HLA-DNA typing were previously reported [29]. We used the nomenclature of HLA alleles recommended by the official nomenclature committee [30, 31].

Statistical Analysis

We compared frequencies of HLA-DRB1, DQA1, DQB1, DPA1, and DPB1 alleles in the agranulocytotic group with those in the control and nonagranulocytotic Graves disease groups. Strength of the statistical association between the disease and HLA allele was expressed by the odds ratio, and the statistical significance was examined by using the chi-square test with Yates correction. We calculated the corrected P value by multiplying the P value by the number of alleles tested for each class II gene (58 DRB1 alleles, 13 DQA1 alleles, 14 DQB1 alleles, 6 DPA1 alleles, and 46 DPB1 alleles).

Results

Top Methods Results Discussion Author & Article Info References

HLA Class II Alleles Associated with the Agranulocytotic Group

We observed positive associations between HLA class II alleles and agranulocytosis in DRB1*1501 (37.5% phenotype frequencies in the patients with agranulocytosis compared with 12.3% in controls; odds ratio, 4.25; P = 0.001), DRB1*08032 (54.2% compared with 17.9%; odds ratio, 5.42; P < 0.001), DQA1*0103 (66.7% compared with 40.4%; odds ratio, 2.95; P = 0.02), DQB1*0602 (33.3% compared with 11.8%; odds ratio, 3.73; P = 0.006), and DPB1*0501 (95.8% compared with 60.6%; odds ratio, 15.0; P = 0.001) (Table 1). The positive associations between HLA class II alleles and agranulocytosis in both DRB1*08032 and DPB1*0501 remained statistically significant, even when the P value was corrected by multiplying the number of tested alleles in each locus. We believe that the relatively increased frequencies of DQB1*0602 and DQA1*0103 were caused by linkage disequilibria between DRB1 and DQ alleles, because DRB1*1501-DRB5*0101-DQA1*0102-DQB1*0602 and DRB1*0803-DQA1*0103-DQB1*0601 haplotypes are frequently found in Japanese persons (11.6% and 17.9%, respectively) [32].

Identification of HLA Alleles Specific for Susceptibility to Agranulocytosis

Underlying Graves disease is a typical organ-specific autoimmune disease, and the association between the disease and HLAs has been reported in various ethnic groups [33-34]. We have reported that the susceptibility to Graves disease in Japanese persons is strongly associated with DPB1*0501 [35] and that resistance to the disease is associated with DQB1*0501 [36]. To distinguish genes associated with susceptibility to agranulocytosis from those associated with Graves disease, we compared frequencies of the HLA alleles associated with agranulocytosis in patients with Graves disease with those in nonagranulocytotic patients with Graves disease. Table 2 lists allele frequencies in the agranulocytotic and nonagranulocytotic Graves disease groups. The frequency of the DRB1*08032 allele was significantly increased in the agranulocytotic group (frequency of 54.2% in the agranulocytotic group compared with 22.1% in the nonagranulocytotic group; odds ratio, 4.18; P = 0.007). We found no significant difference in the frequencies of DRB1*1501 and DPB1*0501 between the two groups, although both alleles were more frequent in the agranulocytotic group than in the nonagranulocytotic group.

Discussion

Top Methods Results Discussion Author & Article Info References

The HLA class II oligotyping data that we report suggest that the susceptibility to methimazole-induced agranulocytosis in patients with Graves disease has a genetic factor. In Japanese persons, methimazole-induced agranulocytosis shows strong associations with the presence of DRB1*08032 and DPB1*0501. Among these alleles, the frequency of DRB1*08032 was increased significantly in agranulocytotic patients compared with patients without agranulocytosis, suggesting that DRB1*08032 specifically is associated with susceptibility to agranulocytosis. The DR8 (DR -DRB1*08032) antigen is probably directly involved in the development of methimazole-induced agranulocytosis in Japanese persons, although other genes in close linkage disequilibrium with HLA-DRB1*08032 may also be involved [37]. We previously reported an association between the susceptibility to Graves disease and DPB1*0501 in Japanese persons [35], and the strong association with agranulocytosis (in comparison with controls) observed in this study may reflect the association between the allele and susceptibility to Graves disease. However, we could not exclude the possibility that DPB1*0501 was also involved in susceptibility to agranulocytosis, because the odds ratio for this allele was 3.51 compared with nonagranulocytotic patients (Table 2). The misconception that the increased frequency of DPB1*0501 was not significant may have originated because this allele is common in patients with Graves disease (the frequency of DPB1*0501 in nonagranulocytotic patients with Graves disease was 86.8%). DRB1*1501 was also found to be associated with agranulocytosis, although the association was not significant by strict statistical analysis (corrected P value equals 0.08), probably because of the small number of patients tested in our study. Further study is needed to confirm this association.

Major histocompatibility complex class II molecules are essential for the recognition of antigenic peptides by CD4⁺ T lymphocytes, and this process is central to the development of immune responses [23]. Each major histocompatibility complex class II molecule binds a different set of proteolytic fragments of peptides and thus contributes to the differences in immune responsiveness [23-25]. Even single amino acid substitutions in major histocompatibility complex molecules substantially alter both preferences for peptides and T lymphocyte responses [38-40]. Thus, susceptible HLA alleles must be identified at the DNA level rather than the serologic level. This can be done efficiently by using the polymerase chain reaction sequence-specific oligonucleotide probes method [23, 24].

It has been suggested that antigranulocyte antibodies play a crucial role in the pathogenesis of agranulocytosis through direct cytotoxity or growth inhibition of progenitor cells [19]. However, the mechanisms by which these antibodies are produced remain unknown. Antibody production is generally controlled by CD4⁺ T lymphocytes, which recognize the proper major histocompatibility complex class II peptide complex. Our observations offer a new insight into the pathogenesis of methimazole-induced agranulocytosis. Wall and colleagues [17] have reported that in vitro T lymphocyte proliferation in response to antithyroid drugs shows an extensive cross-reactivity between methimazole and propylthiouracil in the presence of autologous serum. Antithyroid drugs themselves, however, can not be presented by the HLA molecules because of the structure of the molecules; HLA class II molecules bind and present short peptide fragments consisting of 10 to 20 amino acids. Our results may oppose the hypothesis that the drugs act as nonspecific stimulants (adjuvants) of the immune system [17]. The development of agranulocytosis is clearly related to a specific HLA-DRB1 allele, suggesting that pathogenic T lymphocytes may be activated by unknown pathogenic peptide fragments bound to a specific HLA-DRB1 complex. Thus, T lymphocytes that show cross-reactivity in response to antithyroid drugs may recognize the same peptide fragments, and cross-reactivity may be explained by antigenic modifications of the peptides mediated by common chemical functions among thioureylene drugs. Sources of the peptide fragments probably include granulocyte surface proteins to which autoantibodies might attach.

A possible mechanism for methimazole-related insulin autoimmune syndrome was recently proposed [41]. The insulin autoimmune syndrome is characterized by large amounts of total serum immunoreactive insulin, the presence of anti-insulin autoantibodies, and fasting hypoglycemia [42]. Although the mechanisms by which these anti-insulin antibodies are produced remain unknown, susceptibility to the syndrome is strongly associated with DRB1*0406 in Japanese persons. In addition, one quarter of the patients surveyed had been receiving medications (95% were sulphydryl compounds such as methimazole and gold sodium thiosulfate) for other diseases such as Graves disease and rheumatoid arthritis before the onset of the syndrome. The association between DRB1*0406 and the insulin autoimmune syndrome was explained by the preference in binding peptides to the HLA-DR -DRB1*0406 complex; the involvement of chemical functions of the sulphydryl compounds was also suggested [41]. By affecting the processing of antigen-presenting cells, a reducing compound such as methimazole may cleave the disulfide bond between flanking cysteine residues of insulin molecules and allow for the HLA-DR -DRB1*0406 complex to bind peptide fragments of the insulin chain [41, 43]. Similar mechanisms may be involved in the process of methimazole-induced agranulocytosis. In fact, besides the thioureylene antithyroid drugs, compounds such as levamisole that can act as free radical scavengers through their sulfur atoms have been reported to induce agranulocytosis [14, 44]. If this is the case, identification of pathogenic peptide fragments that 1) may be modified by these reducing reagents, 2) may be presented by susceptible HLA molecules, and 3) may activate T lymphocytes will elicit direct evidence for autoimmunity in the methimazole-induced agranulocytosis at the molecular level.

Ours is the first report of a genetic predisposing factor that may allow the prediction of methimazole-induced agranulocytosis. However, the clinical utility of the finding is currently limited for several reasons. First, HLA gene typing is impractical for the prediction, both in time and in cost. Second, even if a patient with Graves disease is positive for DRB1*08032, this does not imply that a clinician should withhold methimazole therapy, because incidence of agranulocytosis is sufficiently low in patients with these alleles. In addition, it should be noted that although this allele indicates susceptibility to agranulocytosis in Japanese persons, it might not be applicable to other ethnic groups. Diversity in the HLA genotypes among different ethnic groups is too large; various associations between Graves disease and HLAs have been reported among several ethnic groups [33, 34].

In conclusion, we report that an HLA-linked genetic factor is associated with susceptibility to methimazole-induced agranulocytosis. The susceptibility to this complication is strongly associated with a specific HLA-DRB1 allele, suggesting that T lymphocyte-mediated autoimmunity may be involved in the pathogenesis.

Drs. Sudo and Sasazuki: Department of Genetics, Medical Institute of Bioregulation, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812, Japan.

Dr. Kimura: Department of Tissue Physiology, Division of Adult Diseases, Medical Research Institute, Tokyo Medical and Dental University, 2-3-10 Kandasurugadai, Chiyoda-ku, Tokyo 101, Japan.

Dr. Kuma: Kuma Hospital, 8-2-35 Shimoyamatedouri, Chuou-ku, Kobe 650, Japan.

Dr. Nagataki: First Department of Internal Medicine, Nagasaki University School of Medicine, 1-12-4 Sakamoto, Nagasaki 852, Japan.