SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

In the K/BxN mouse model, autoantibodies against glucose-6-phosphate isomerase (GPI) cause arthritis. The relevance of this model for human disease remains a subject of controversy. We set out to determine whether GPI autoantibodies occur in patients with rheumatoid arthritis (RA) and, if so, at what stage of the RA.

Methods

Using an enzyme-linked immunosorbent assay, serum from 131 RA patients and 28 healthy controls was tested for autoantibodies against recombinant human GPI. Patients were grouped according to disease duration and presence of rheumatoid nodules, rheumatoid vasculitis, and Felty's syndrome, which are extraarticular complications of RA.

Results

Elevated levels of autoantibodies against GPI were present in 5% of patients with uncomplicated RA and 4% of controls. In RA complicated by extraarticular manifestations, anti-GPI antibodies were observed in 18% of patients with rheumatoid nodules, 45% of patients with rheumatoid vasculitis, and 92% of patients with Felty's syndrome.

Conclusion

In patients with RA, autoantibodies to GPI are associated with the occurrence of extraarticular complications.

Rheumatoid arthritis (RA) is a common chronic inflammatory disease of unknown cause that primarily affects the joints. One of the features of RA is the presence of antibodies against self antigens, both inside and outside the joint. Detection of autoantibodies such as rheumatoid factors (RFs) is commonly used as a diagnostic and prognostic tool (1).

To study the disease mechanisms underlying RA, several animal models of arthritis have been used. In 1996, Kouskoff et al (2) first described the KRN T cell receptor–transgenic mouse that spontaneously develops a severe, destructive arthritis (K/BxN mouse model). In subsequent studies, it was shown that in the K/BxN model, autoantibodies against glucose-6-phosphate isomerase (GPI) are present in high titers. More importantly, these anti-GPI antibodies were shown to cause disease on transfer into healthy mice (3).

Given the dominant role of GPI antibodies in this animal model, antibody titers against GPI were studied in humans. It has been reported that GPI is an autoantigen in RA, being present in 64% of patients but only a few healthy control subjects (4). However, the notion that GPI is an autoantigen in RA is surrounded by controversy, since a number of other groups of investigators could not reproduce these observations (5, 6).

Patients with RA are heterogeneous in their autoantibody response. While ∼70–80% of RA patients have RFs, other autoantibodies, such as anti-C1q antibodies (7) and antineutrophil cytoplasmic antibodies (8), are less common and are found in patients with more severe disease. Therefore, the apparently conflicting findings of anti-GPI antibodies in RA may be explained by their occurrence at a specific point in the disease process. By studying groups of RA patients at different stages of disease, we sought to determine whether autoantibodies to GPI occur in RA and, if so, at what point.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients.

Serum was obtained from 131 patients who were classified as having RA according to the 1987 American College of Rheumatology (formerly, the American Rheumatism Association) criteria for RA (9) (Table 1). Patients with early RA had been enrolled in a special early arthritis study cohort, had a diagnosis of RA for less than 4 weeks, and were not receiving disease-modifying antirheumatic drugs at the time serum was obtained (10). Patients with uncomplicated RA of 12 years' duration and RA patients with rheumatoid nodules were from a 12-year prospective study cohort of female RA patients that had been established to investigate disease activity and joint destruction over time (11). Patients with uncomplicated RA of 3 years' duration, RA with vasculitis, and RA with Felty's syndrome were from the inpatient and outpatient clinics at the Leiden University Medical Center. Those with rheumatoid vasculitis had symptoms that were suggestive of vasculitis (neuropathy, skin lesions), and the diagnosis was confirmed by tissue biopsy. Patients with Felty's syndrome had RA, splenomegaly, and spontaneous, sustained neutropenia of <2.0 × 109 cells/liter that could not be attributed to other diseases or to drug therapy.

Table 1. Clinical features of the RA study patients*
 Untreated, early RA (n = 20)Treated RARA with extraarticular complications
3 years (n = 21)12 years (n = 55)Nodules (n = 11)Vasculitis (n = 11)Felty's syndrome (n = 13)
  • *

    In 2 of the 11 rheumatoid arthritis (RA) patients with rheumatoid nodules, the rheumatoid factor (RF) level was above 200 IU/ml; 200 IU/ml was used to calculate the mean.

Age, mean (range) years61 (24–82)61 (42–78)49 (32–64)51 (35–62)67 (51–84)67 (55–81)
Disease duration, mean (range) months<126 (6–60)144 (96–228)149 (120–168)204 (72–384)242 (48–444)
Sex, % female71581001002745
RF, % positive76886910072100
RF, mean (range) IU/ml77 (7–200+)37 (11–93)82 (24–190)
Erosive disease, %1040821008092

The 28 healthy control subjects were women who had been referred to the Leiden University Medical Center outpatient clinic because of primary infertility. All were extensively screened by an internal medicine specialist and were declared to be healthy.

Enzyme-linked immunosorbent assay (ELISA) for GPI.

Recombinant human GPI was prepared as previously described (12). GPI was covalently coupled to 96-well assay plates (Polysorb; Nunc, Roskilde, Denmark) at 0.5 μg/well. Coupling was performed for 12 hours at 4°C and at pH 9.6. Nonspecific binding was evaluated by coupling bovine serum albumin (BSA) instead of GPI. Microtiter plates were blocked with 1% BSA in phosphate buffered saline (PBS). Patient and control sera were diluted 1:50, 1:100, 1:200, 1:400, 1:800, 1:1,600, 1:3,200 in PBS containing 1% BSA and 0.05% Tween 20 (weight/volume). After incubation for 2 hours at 37°C, plates were washed 4 times with PBS–0.05% Tween. Antibodies were detected using mouse anti-human IgG (clone HB 43; American Type Culture Collection, Manassas, VA) coupled to horseradish peroxidase diluted 1:1,000. Bound antibodies were visualized using ABTS as a substrate.

Plates were read at a wavelength of 415 nm using an automatic ELISA reader (BioTek Instruments, Winooski, VT). Optical density (OD) values outside the linear detection range of the ELISA reader were excluded from analysis. OD values at a serum dilution of 1:200 were used for the analysis, since values at this dilution were consistently on the linear portion of the OD curve. Samples negative at this dilution were also negative at other dilutions. Throughout the experiments, a standard of pooled serum from 30 donors with RA was used as a reference on every plate. After subtraction of the blank control, the measured OD415 from the reference was defined as 1 unit value. Plate-to-plate variation of the reference sample did not exceed 5%, and values were therefore not corrected. Values of experimental samples were expressed as the ratio of the reference value.

To ensure specificity of sera recognizing GPI, the following tests were done. All positive samples were tested in an ELISA where the GPI antigen had been replaced by mouse type II collagen, BSA, and an irrelevant random peptide synthesized using automated simultaneous multiple peptide synthesis. Specific binding of patient serum to recombinant GPI was confirmed by inhibition immunodot-blotting and Western blotting (data not shown). To control for false-positive results in the ELISA due to IgM and IgG RFs (anti-IgG autoantibodies), the mean level of IgM-RF and the mean level of GPI antibodies were compared between patient groups. In addition, the concentration of RF in GPI-positive patient serum was lowered by incubating the serum with Sepharose-aggregated IgG or with Sepharose-aggregated BSA (control) for 1 hour at room temperature and retesting the serum in the ELISA.

Statistical analysis.

The mean level of antibodies was calculated, and the nonparametric Mann-Whitney U test (significant at P < 0.05) was applied to test differences between groups. Samples were considered positive for anti-GPI antibodies when the antibody level was higher than the mean plus 2 SD in the controls (1.74 arbitrary units [AU]). To control for differences in demographic background between the patients and the controls, we also used as a cutoff value the mean plus 2 SD in the group of patients with RA of 3 years' duration (1.71 AU), since this group of patients was more similar to the groups with complicated RA in terms of sex, age, and RF positivity (Table 1). This reanalysis gave identical results.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Autoantibodies against GPI in patients with early RA.

To determine whether antibodies against GPI were present at a specific point in the disease process, we tested by ELISA serum from RA patients at different stages of disease. GPI antibodies in sera from patients with untreated early RA and patients with RA of 3 years' duration were determined. Patients with early RA, who had not been treated with second-line antirheumatic drugs at the time serum was obtained, had a diagnosis of RA for less than 4 weeks. The RA patients with disease duration of 3 years had all been treated with second-line antirheumatic drugs. In both groups, most patients were IgM-RF positive and a significant proportion had erosions on radiographs of the hands and feet (Table 1).

Figure 1 shows the anti-GPI antibody levels in patients with early RA, patients with RA of 3 years' duration, and healthy controls. The mean ± SD level of GPI antibody was 0.78 ± 0.46 AU in the group with RA of 3 years' duration, 0.52 ± 0.46 AU in the group with early RA, and 0.86 ± 0.44 AU in the control group.

thumbnail image

Figure 1. Antibodies to glucose-6-phosphate isomerase (GPI) in healthy controls and in patients with early untreated rheumatoid arthritis (RA) or treated RA of 3 years' duration, as determined by enzyme-linked immunosorbent assay (ELISA). Horizontal lines show the mean for each group; broken line shows the cutoff for positivity (mean plus 2 SD in healthy controls; see Patients and Methods). All positive samples were tested in an ELISA against type II collagen, bovine serum albumin, and an irrelevant random peptide, but none of them had levels above background (results not shown).

Download figure to PowerPoint

When the mean plus 2 SD level of GPI antibody in the healthy controls was taken as the cutoff point for positivity, no differences in anti-GPI positivity were observed. Elevated levels of GPI antibody were present in 5% of patients with early RA (1 of 20), 5% of patients with RA of 3 years' duration (1 of 21), and 4% of healthy controls (1 of 28).

Antibodies against GPI in patients with RA and extraarticular complications.

We tested serum samples from patients with longstanding RA with and without various extraarticular disease manifestations for the presence of anti-GPI antibodies (Table 1). The 4 groups tested were those with a 12-year disease duration without extraarticular complications and those with longstanding RA and either rheumatoid nodules, biopsy-proven vasculitis, or Felty's syndrome.

In patients with uncomplicated RA of 12 years' duration, the mean ± SD antibody concentration was 0.94 ± 0.46 AU (Figure 2). In patients with similar disease duration but with rheumatoid nodules, the concentration was 1.4 ± 0.43 AU. The GPI antibody concentration in those with vasculitis was 1.7 ± 0.73 AU. The mean antibody levels in the groups of patients with nodules and vasculitis were significantly higher than those in the controls (P < 0.05). Moreover, there was a trend toward an increased prevalence in these groups, since 2% of those with RA of 12 years' duration (1 of 55), 27% of those with rheumatoid nodules (3 of 11), and 36% of those with rheumatoid vasculitis (4 of 11) had elevated levels of GPI antibody. The mean anti-GPI antibody concentration was 3.3 ± 1.3 AU in RA patients with Felty's syndrome (P < 0.0001 compared with controls). In this group, 12 of the 13 patients (92%) had an antibody concentration above the cutoff value.

thumbnail image

Figure 2. Antibodies to glucose-6-phosphate isomerase (GPI) in patients with uncomplicated, longstanding rheumatoid arthritis (RA) of 12 years' duration and RA with extraarticular complications, including Felty's syndrome, as determined by enzyme-linked immunosorbent assay (ELISA). Horizontal lines show the mean for each group; broken line shows the cutoff for positivity (mean plus 2 SD in healthy controls; see Patients and Methods). All positive samples were tested in an ELISA against type II collagen, bovine serum albumin, and an irrelevant random peptide, but none of them had levels above background (results not shown).

Download figure to PowerPoint

To assess possible artifacts mediated by RFs in the ELISA, the mean concentration of IgM-RF was calculated for groups of patients with extraarticular complications (Table 1). In the group with rheumatoid nodules, which had the lowest level of GPI antibodies, the mean level of IgM-RF (77 IU/ml) was significantly higher than that in the group with rheumatoid vasculitis (37 IU/ml; P = 0.03) but was not significantly different from that in the group with Felty's syndrome (82 IU/ml; P = 0.8).

Serum samples from 9 patients positive for GPI antibodies (4 with Felty's syndrome, 2 with rheumatoid vasculitis, and 3 with rheumatoid nodules) were incubated with Sepharose-aggregated IgG and then retested. After incubation, the IgM-RF concentration had dropped an average of 33% (range 11–70%), but the anti-GPI antibody level had increased an average of 6% (range –8% to 22%).

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In this study, we investigated the occurrence of autoantibodies to GPI in patients with RA. Five percent of the patients with early, untreated RA and 4% of the healthy controls had increased levels of autoantibodies to GPI. Similar percentages were found in patients whose RA had been treated: 5% in those with a disease duration of 3 years and 2% in those with a disease duration of 12 years. In contrast, in patients with complicated RA, increased concentrations of anti-GPI antibodies were found in 18% of those with rheumatoid nodules, 45% of those with rheumatoid vasculitis, and 92% of those with Felty's syndrome.

These increased concentrations of antibodies were not correlated with either the presence or the levels of RF. In all 3 groups, the majority of patients had IgM-RF. However, among the patients who had extraarticular manifestations, those with rheumatoid nodules had the lowest levels of GPI antibodies, but the IgM-RF concentrations were higher than those in patients with rheumatoid vasculitis and were similar to those in patients with Felty's syndrome. Depletion of RF from patient sera with Sepharose-aggregated IgG resulted in slightly increased levels of anti-GPI antibodies; this was likely due to a concentrating effect of the procedure.

Rheumatoid nodules, which are skin granulomas precipitated by local trauma, are present in 20–35% of patients with longstanding, RF-positive RA. Vasculitis and Felty's syndrome are observed less frequently. Clinically significant vasculitis occurs in 5–10% of RA patients, while Felty's syndrome, which is characterized by neutropenia and an enlarged spleen, is found in less than 1% of RA patients (13, 14). These disease manifestations tend to develop late in the disease course and are associated with a severe disease course. The occurrence of these antibodies in RA patients with extraarticular manifestations suggests, at least in part, that there is a common pathogenesis. It is thought that the neutropenia of Felty's syndrome, rheumatoid vasculitis, and rheumatoid nodules are due to the deposition of excessive amounts of immune complexes (e.g., RF, IgG) on cells in the circulation and in the vascular endothelium. This may lead to margination of neutrophils along the vessel wall or to inflammation of the vessel itself (15). It has previously been shown that the GPI antigen is present on the surface of blood vessels in RA synovium (4). In the present study, granulocytes isolated from patients with Felty's syndrome and from healthy controls contained GPI antigen in the cytoplasm but not on the cell surface (data not shown).

From these data, it could be argued that in RA patients with extraarticular manifestations, anti-GPI antibodies contribute to disease by targeting the vessel wall, forming immune complexes locally. Another possibility is that GPI antibodies are the result of excessive neutrophil destruction, which leads to altered presentation of the GPI antigen to the immune system. To understand the possible role of GPI in disease, it must be taken into account that although GPI is best known as a cytosolic enzyme that is involved in glycolysis and gluconeogenesis, proteins with cytokine and growth factor activities are highly homologous to GPI (16). So, although the role of these antibodies in the pathogenesis of complicated RA is still to be elucidated, given the high prevalence of elevated titers of GPI autoantibody in complicated RA, the detection of these antibodies may be useful in the diagnosis and management of extraarticular complications, Felty's syndrome in particular.

The results of the present study might also explain the discrepancy between previous studies of the prevalence of anti-GPI antibodies in RA. The controversy might be explained by our finding that, when compared with healthy controls and patients with uncomplicated RA, patients with extraarticular manifestations of RA had elevated levels of GPI antibodies.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank Dr. E. Lentjes for providing the control serum, Jolanda van Bilsen and Ellen van der Voort for their technical expertise and for evaluating the experiments, and Leendert Trouw (Department of Nephrology, Leiden University Medical Center) for the kind gift of Sepharose-aggregated IgG and BSA.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES