Abstract
Objective. To detect autoantibodies to tumor necrosis factor (TNF) in patients with rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE), and to determine their clinical correlates.
Methods. Ninety-two patients with RA and 62 with SLE were studied. Sera were examined for autoantibodies to TNF by enzyme linked immunoassay. Levels of these autoantibodies were analyzed in respect to markers of inflammation such as erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and joint erosions, as well as other clinical, laboratory, and therapeutic aspects of RA and SLE.
Results. Anti-TNF levels were higher in those RA patients without erosions, but did not correlate with ESR or CRP.
Conclusion. These observations suggest that autoantibody anti-TNF may be part of the innate immune system and may contribute to decreased inflammation in patients with RA.
- TUMOR NECROSIS FACTOR
- ANTI-TUMOR NECROSIS FACTOR ANTIBODIES
- RHEUMATOID ARTHRITIS
- SYSTEMIC LUPUS ERYTHEMATOSUS
Tumor necrosis factor (TNF) is an important mediator of inflammation in rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE). Further, biological agents that block TNF have a significant effect in downregulating inflammation in patients with RA; similar observations have been made in a few patients with SLE. We are interested in autoantibodies in patients with rheumatic diseases, and their possible role in inflammation. Autoantibodies to self-proteins have been recognized in a number of human autoimmune diseases. These include autoantibodies to oxidized low density lipoprotein (LDL), ß2-glycoprotein I, prothrombin, cardiolipin, IgG, idiotypes, α-1-antitrypsin, fibrinogen, ceruloplasmin, serum amyloid A (SAA), and C-reactive protein (CRP)1–11. The role of these autoantibodies in clinical disease has been intensively investigated, especially in patients with RA and SLE12,13. We have been interested in the possible role of autoantibodies to acute-phase proteins. A number of investigators have recently described antibodies to SAA8, CRP9–11, and in one study antibodies to TNF in SLE14. Monoclonal antibodies to TNF have been very successful in the treatment of inflammation in patients with RA and in some patients with SLE. We therefore determined whether patients with RA and/or SLE had autoantibodies to TNF and whether increased serum levels of these autoantibodies were associated with less inflammation.
MATERIALS AND METHODS
Subjects
All sera sent to the Brigham and Women’s Hospital (BWH) Clinical Immunology Laboratory from the BWH Arthritis Center for CRP measurement over 1 year were selected for this analysis. The electronic medical records of over 300 patients were reviewed, and those samples with a history of either RA or SLE (by physician diagnosis) were selected for further study. Whether patients were being treated with nonsteroidal antiinflammatory drugs, antimalarials (e.g., hydroxychloroquine), corticosteroids, disease modifying antirheumatic drugs (in particular methotrexate), and TNF-blocking agents was recorded. Only patients not currently receiving anti-TNF-blocking agents were selected for further study. In addition, data on patient’s sex, age at disease onset, current age, duration of illness, erythrocyte sedimentation rate (ESR), rheumatoid factor (RF), anti-cyclic citrullinated peptides (CCP), and radiological investigation for erosions were collected.
Laboratory methods
CRP was performed as a high sensitivity-CRP by the nephelometric method, expressed as mg/dl15. ESR was performed by the Westergren standard method16.
IgM RF was measured by the nephelometric method and expressed as IU17. Antibodies to CCP were measured by the ELISA method and expressed as units18. Anti-CRP antibodies were performed as described and expressed as optical density (OD)11.
Anti-TNF antibody autoantibody detection method
An ELISA protocol was developed as follows: irradiated plates (Costar 3590, Cambridge, MA, USA) were coated with TNF recombinant human (E. coli-derived; R&D Systems, Minneapolis, MN, USA, catalog no. 210-TA) at a concentration of 100 ng/ml in phosphate buffered saline (PBS), ph 7.4, 50 μl/well. The plates were incubated overnight at 4°C and were then washed 1 time with PBS, blocked with 2% bovine serum albumin/PBS (block solution) 200 μl/well, and incubated 1 h at room temperature. After washing 1 time with PBS, 50 μl serum diluted 1:50 in block solution was added to a well containing antigen plus block and a well containing block only to control for reactivity to block. Wells containing a highly reactive serum, a negative serum, or buffer served as controls. The plates were incubated overnight at 4°C. They were then washed 3 times with PBS with the second wash remaining on the plate for a full 2 min. Conjugate (A-3150 goat anti-human IgG alkaline phosphatase; Sigma, St. Louis, MO, USA) was diluted 1:1000 in block solution; 50 μl was added to each well. The plates were incubated 1 h at room temperature and then washed 3 times as before. Fifty μl substrate (p-nitrophenyl phosphate, 20 mg in 10 ml glycine buffer; Sigma N 2765) was added to each well and incubated at room temperature for 1 h 20 min. This long incubation period was necessary to reach the peak of the reaction. The OD were read on a plate reader (Titertek Multiskan MCC/34D) at 405 nm. The difference between the OD in the well with antigen plus block and the OD in the well with block only was recorded as the result.
Sixty-three blood sera were tested by this method to determine the mean OD ± 2 standard deviation (SD) for healthy individuals (0.1535 ± 0.2285). Levels of anti-TNF above 2 SD above the mean, that is 0.3820 OD, were considered positive.
Specificity of the assay was performed as follows: 3 sera from patients with RA, and 1 serum from a patient with SLE known to have autoantibodies to TNF, CRP, or both, at a dilution of 1:50. Serial serum dilutions, of a known positive, were incubated on TNF or CRP antigen-coated plates overnight (plate 1). On Day 2, the supernatants from plate 1 were transferred to freshly TNF and CRP-coated plates (plate 2). The reaction on plate 1 was completed (as per above) to obtain a baseline. After overnight incubation, plate 2 was completed. The difference in the OD on plates 1 and 2 represent the amount of inhibition.
Statistical analysis
All data were entered into an Excel spreadsheet and analyzed by Excel statistics including t tests and correlations.
The Partners Healthcare System Institutional Review Board approved all aspects of this study.
RESULTS
The specificity of the autoanti-TNF antibody assay is given in Table 1 as OD. The OD values represent the reactions in antigen-coated wells minus the OD in uncoated wells. The results indicate that solid-phase bound antigen absorbs anti-TNF from serum without absorbing anti-CRP, or vice versa.
Specificity of the anti-TNF antibody assay.
Sera from the 92 subjects with RA and 62 subjects with SLE were studied. The demographics (mean ± 2 SD) and serum anti-TNF antibody levels in these 2 groups are given in Table 2.
Demographics and laboratory data on patients with RA and SLE.
Characteristics of those RA and SLE patients with elevated levels of anti-TNF anti-body. Each of the 6 patients with RA who had radiographs of their hands or feet had elevated levels of anti-TNF antibody and no erosions.
There was no significant difference in the mean anti-TNF autoantibody titer in those subjects with RA and those with SLE, nor between the mean levels between normals and patients with SLE. However, the mean level was greater in the patients with RA than normals (p = 0.01). Patients with SLE were younger; the mean CRP levels were significantly different between the patients with RA and SLE (p = 0.01); none of the other differences in Table 2 were significantly different. The OD of the normals and patients with RA and SLE is given in Figure 1. Two of the 63 (3%) normal individuals had OD greater than 0.382.
Optical densities of anti-TNF autoantibody levels in patients with RA, SLE, and normals.
The patients with RA were receiving different medications, often in combination: methotrexate 48, prednisone 27, hydroxychloroquine 20, leflunomide 9, and sulfasalazine 6. The patients with SLE were receiving hydroxychloroquine 36, prednisone 29, azathioprine 12, mycophenolate 8, cyclophosphamide 3, and leflunomide 2.
There was no significant difference in the mean anti-TNF autoantibody levels in those patients with RA taking or not taking methotrexate or other medications; there was no significant difference in the anti-TNF autoantibody levels in those patients with SLE taking or not taking plaquenil, prednisone, or other medication.
Patients with RA and without erosions had significantly higher levels of anti-TNF autoantibody (0.236 ± 0.152 vs 0.154 ± 0.063; p = 0.0007).
There was no significant correlation between anti-TNF autoantibodies and age of onset, CRP, or ESR, in patients with RA or SLE.
In patients with RA there was no significant association between anti-TNF autoantibody titer and RF titer, anti-CCP titer, or a history of ever smoking.
In patients with SLE there was no association between anti-TNF autoantibody titer and anti-DNA titer, anti-Sm titer, anti-RNP titer, anti-Ro titer, anti-La titer, white blood cell count, hemoglobin, serum albumin, platelet count, or treatment with hydroxychloroquine, prednisone or azathioprine.
DISCUSSION
In our study, we found elevated levels of autoantibodies to TNF in some patients with RA (8%) and in some patients with SLE (6%). Our observations regarding anti-TNF autoantibodies extend our observations regarding anti-CRP. In our initial studies we observed that many patients with RA had markedly elevated levels of serum antibodies to TNF. However, in chart review most of those patients with RA were receiving treatment with anti-TNF biologicals, and thus were excluded from further study. None of the patients with SLE were being treated with anti-TNF. However, some patients with RA not receiving anti-TNF agents had elevated levels of serum anti-TNF antibodies. Further, those RA patients with elevated levels of anti-TNF autoantibodies did not have erosions on radiographs of hands or feet. Those without erosions had significantly higher levels of serum anti-TNF autoantibodies. These observations suggest that these autoantibodies may play some role in decreasing the degree of inflammation in patients with RA, just as anti-TNF agents, given in much greater amounts, have a greater therapeutic effect, including preventing erosions.
Autoantibodies to TNF have been observed by others in patients with SLE, where they correlated inversely with disease activity as assessed by the SLE Disease Activity Index (SLEDAI)18. In our study, there was insufficient chart information to assess a SLEDAI score. There was no association in the patients with SLE between anti-TNF autoantibody levels and clinical and laboratory variables.
As with any other study that is dependent on a medical record review, one must rely on the completeness and accuracy of the physicians who recorded information, requested specific laboratory and radiological investigations, and selected different medications for treating their patients.
Whether these autoantibodies to TNF are the result of stimulation by TNF, and/or are part of the innate immune system’s response to inflammation, will require further study. Our observations add to the expanding repertoire of autoantibodies to human proteins, and their possible clinical significance.
Acknowledgments
We gratefully acknowledge the assistance of Holly Fossel for expertly entering and cleaning the data and for the data analysis; the staff of the Clinical Immunology Laboratory for performing the routine assays for CRP, RF, and anti-CCP; and Drs. Karen Costenbader and David Lee for reviewing the manuscript and making useful comments.
Footnotes
- Accepted for publication November 6, 2008.
