Tumour necrosis factor alpha and its soluble receptors in juvenile chronic arthritis

Texte intégral

(1)Rheumatology 2000;39:432–438. Tumour necrosis factor alpha and its soluble receptors in juvenile chronic arthritis M. Rooney, H. Varsani, K. Martin, P. R. Lombard1, J.-M. Dayer1 and P. Woo Centre for Paediatric and Adolescent Rheumatology, UCL, London, UK and 1Division of Immunology and Allergy, University Hospital, Geneva, Switzerland Abstract Objective. To identify possible imbalance of tumour necrosis factor alpha (TNFa) and its soluble receptors in the different subgroups of juvenile chronic arthritis (JCA). Methods. Serum and synovial fluid samples from 45 children were examined, 25 pauciarticular JCA, 13 polyarticular JCA and seven spondyloarthropathy. TNFa, sTNFRI and sTNFRII levels were measured by EASIA and enzyme-linked immunosorbent assay (ELISA). Analysis of the results was carried out using non-parametric tests: Kruskal–Wallis one-way analysis of variance was used to compare the three clinical subgroups; the Mann– Whitney U-test was used to compare group medians. Results. Thirty-three serum samples were assayed for TNFa. There was no significant difference between the three groups using the Kruskal–Wallis analysis of variance. Analysis of synovial fluid TNF levels showed significantly lower levels in the spondyloarthropathy group compared with the pauciarticular JCA (P = 0.01) and the polyarticular group (P = 0.002). Significantly higher levels of sTNFRI were observed in the synovial fluid of the polyarticular JCA group compared with the pauciarticular JCA group (P = 0.004) and similarly for sTNFRII (P = 0.03). Molar ratios were calculated for TNF vs sTNFRI. The sTNFRI/TNFa ratio was significantly higher in the spondyloarthropathy group compared with the pauci(P = 0.003) and the polyarticular JCA subgroups (P = 0.003). The combined soluble receptor levels expressed as molar ratio to TNF again showed a significantly higher ratio in the spondyloarthropathy group compared with the pauciarticular group (P = 0.01) and compared with the polyarticular group (P = 0.05). Conclusion. These results suggest that the increased joint destruction observed in polyarticular disease compared with the other two subtypes may be related to the lower sTNFR/TNFa ratios observed. K : Juvenile chronic arthritis, Synovial fluid, Cytokines, Tumour necrosis factor alpha, Soluble TNF receptors.. Juvenile chronic arthritis (JCA) refers to a group of idiopathic arthritides of childhood. Although arthritis is the salient feature of all these diseases, there are important clinical differences which suggest that the underlying pathology may be different. Pauciarticular JCA, involving four or fewer joints [1], usually runs a benign course. While clinically quite marked inflammation can be observed in individual joints, destruction of that joint is an uncommon finding long term [2, 3]. In contrast, polyarticular disease is frequently associated with joint damage and erosive changes [4, 5]. While serological indices of inflammation are more frequently raised in polyarticular disease, there is little to distinguish between the clinical findings of these two. subgroups with respect to individual joints. The juvenile spondyloarthropathies are members of a different clinical group. Although the laboratory markers of inflammation can be impressive, the natural history of juvenile spondyloarthropathies is towards new bone formation, fibrosis and ankylosis rather than bone resorption [6–8]. This would suggest that different mechanisms of inflammation and/or repair occur in the subgroups of JCA. The inflammatory cytokines have long been implicated in the pathogenesis of inflammation and in particular arthritis [9–12]. Tumour necrosis factor alpha ( TNFa) exhibits many biological actions both in vitro and in vivo pertinent to arthritis. It can induce an inflammatory response, collagenase production [13, 14], bone and cartilage resorption [15, 16 ] and cachexia [17]. TNFa is one of the primary stimulants of interleukin-1 (IL-1) in synovial cell cultures of adult rheumatoid arthritis patients [18]. This, along with its ability to induce HLA class I molecule expression on vascular endothelial cells and dermal fibroblasts [19], suggests. Submitted 16 July 1998; revised version accepted 5 April 1999. Correspondence to: P. Woo, Centre for Paediatric and Adolescent Rheumatology, UCL, Department of Molecular Pathology, The Windeyer Institute of Medical Sciences, 46 Cleveland Street, London, W1P 6DB, UK.. 432. © 2000 British Society for Rheumatology.

(2) TNFa in JCA. that TNFa plays a central role in the cytokine network in rheumatoid arthritis as well as cellular immunity. The cytokine network is essential to host defence and thus regulatory mechanisms are essential for homeostasis [20, 21]. Several mechanisms are currently known to down-regulate the pro-inflammatory cytokines including anti-inflammatory cytokines [22, 23] and cytokine antagonists, which are present in synovial fluids [24, 25]. TNFa binds to two high affinity cell surface receptors, TNFRI and TNFRII, which are present on virtually all cell types [26, 27]. The extracellular domains of these receptors exist in a soluble form and have been shown to bind circulating TNFa and thus inhibit its biological activity [28]. Among the factors which may influence cellular responses induced by TNFa are the number of cell-associated receptors and the concentration of soluble TNFRI and TNFRII receptors in the extracellular fluid. The relative proportion of TNFa to its soluble receptors in biological fluids has proved to be important in a number of clinical conditions [29–32]. In meningococcal septicaemia, a diminished ratio of soluble TNFRI and TNFRII to TNFa was observed in those patients with a fatal outcome compared with survivors [30]. We postulated, therefore, that levels of TNFa and its soluble receptors might be different in the synovial fluid of the various subgroups of JCA thus reflecting their contrasting clinical courses.. Patients and methods Patients and clinical characteristics Fifty children with JCA according to the EULAR criteria [1] entered the study. All required intra-articular steroid injection for synovitis of the knee. None had had a steroid injection to that knee in the preceding 6 months. Clinical details recorded included subtype of JCA, age, sex, disease duration and current medication. The degree of joint inflammation observed clinically was scored on a scale of 1–3 indicating mild, moderate or marked synovitis, by the same observer. Three children had systemic JCA and due to their small numbers were excluded from the study. Two children, one from the polyarticular group and one from the pauciarticular group had synovial fluid TNFRII and RII/TNFa ratios 10 times higher than the maximum ratios observed in either group and were excluded as these results were probably incorrect, and a re-test was not possible due to the small number of samples available. Thus, 45 children completed the study in which 47 knees were aspirated. There were 25 in the pauciarticular group, 13 in the polyarticular group and seven with a spondyloarthropathy. Values for erythrocyte sedimentation rate ( ESR) and C-reactive protein (CRP) were available on 12 pauciarticular, seven polyarticular and four patients with spondyloarthropathy and the clinical details and results are included in Table 1. Synovial fluid was aspirated using an aseptic tech-. 433. nique. Samples were immediately spun and stored at −70°C until assayed. Blood samples were available on 31 children. CRP was measured by nephelometry and ESR using the Westergren method. Serum samples were spun and rapidly frozen at −70°C until used. Ethical Committee approval was obtained for this study at Great Ormond Street Hospital National Health Service Trust and patient assent and parent consent given. Cytokine determination TNFa determination in synovial fluid and serum. Synovial fluid and serum samples were initially assayed for TNFa by EASIA (Medgenix, Fleurus, Belgium) with a limit of sensitivity of 20 pg/ml. In 60% of samples assayed, TNFa levels were below the limit of detection of the assay. These negative samples were subsequently reassayed using a high sensitivity enzyme-linked immunosorbent assay ( ELISA) (R & D Systems Europe, Abingdon, Oxon) with a limit of sensitivity of 2 pg/ml. Both assays were chosen as their antibodies detected both free and complexed TNFa (as confirmed by the manufacturers). Intra-assay variation was 11.7% with an interassay variation of less than 10% for the individual assays used. There was a tendency for the high sensitivity kit to give lower values to high TNF levels than observed using the low sensitivity kit. This was probably due to the fact that the TNFa antibodies used in these kits are directed against different TNFa epitopes, the different standards used for the calibration curves, or to a different sensitivity because of interference with other proteins. Since the values obtained from these two assays were not linearly related, the results obtained from both assays were analysed separately using nonparametric tests (vide infra). Determination of soluble TNF receptors RI and RII in synovial fluid and serum. Initial samples were assayed by a method previously reported [25]. Subsequent samples were assayed by ELISA (R & D Systems Europe). Both inter- and intra-assay variation was less than 10% for each individual system used. Using values from the initial in-house assay for sTNFRI as internal controls, we observed that values on the same samples from the R & D System Method were consistently twice as high as the original results, and thus R & D readings were halved in order to be able to combine results from samples measured only with the R & D kit and those measured by both methods. Statistical analysis Since specimens were available in limited amounts only it was not possible to test them by both assays. As synovial fluid values for the TNFa and TNFRI and TNFRII were obtained by different assays, nonparametric analysis by the Kruskal–Wallis one-way analysis of variance was used to compare the three clinical subgroups. Subsequently the Mann–Whitney U-test was used to compare each pair of group medians. For analysis of TNFa values using the low sensitivity kit,.

(3) M. Rooney et al.. 434. T 1. Clinical and laboratory details of patients studied. No. of patients No. of knees Mean age (range) (yr) Sex ( F/M ) Disease duration (range) (yr) Medication NSAID Sulphasalazine Methotrexate Prednisolone None Other Knee joint activity (No.) 1 = mild 2 = moderate 3 = marked NR CRP mg/dl mean (range) ESR mm/h (range). Pauciarticular JCA. Polyarticular JCA. Spondyloarthropathy. 25 26 9.1 (3.5–16.5) 19/4 3.7 (0.3–7.0). 13 14 12.9 (4–24) 12/1 7.3 (1.0–15). 7 7 13.4 (11.3–14.9) 1/6 1.3 (0.3–4.2). 15 0 0 0 8 2. 13 0 6 3 0 0. 7 3 0 1 0 0. 7 12 3 4 1.6 (0.5–8.2) 21 (1–68). 0 9 3 2 2.8 (0.5–2.8) 39 (13–91). 0 5 2 0 3.0 (0.9–6.5) 38 (20–60). NSAID, non-steroidal anti-inflammatory drugs; NR, not recorded.. samples below the limit of detection were given a value of 20 pg. For analysis of results obtained using the high sensitivity kit, values were censored at 33 pg, the highest value observed using this assay. The Bonferroni correction was used to adjust the P values obtained from the Mann–Whitney U-tests to take into account multiple comparisons. All serum samples were assayed by the same assay for TNFa, sTNFRI and sTNFRII and values were found to be normally distributed if log transformed using Shapiro and Francia’s W-test. However, the assumption of constant variances were not valid. Therefore, the results were expressed as medians and analysed by the Kruskal–Wallis one-way analysis of variance.. Results Thirty-three serum samples were assayed for TNFa and all samples were assayed by the high sensitivity kit from 17 pauciarticular JCA, 10 polyarticular JCA and six spondyloarthropathy patients (Fig. 1). Although the median value for the polyarticular group appears higher. F. 1. Serum TNFa levels within the three clinical subgroups. Horizontal bars indicate median values.. than the median of the other two groups, this was not significant. In 16 patients sTNFRI and II were measured. All samples were measured by the same assay system [27] as were the synovial fluid samples used for comparison. For all patients the mean serum sTNFRI was 1.25 (range 0.6–2.1 ng/ml ). The level of serum sTNFRII was twice that of sTNFRI, mean 3.5 (range 2.3–6.0 ng/ml ) P = 0.001. Insufficient serum samples were available for analysis of sTNFRs within the disease subgroups. Synovial fluid TNFa levels were compared in the three subgroups and the results obtained using both the low and high sensitivity assays are shown in Fig. 2a, b. Using the low sensitivity assay, TNFa levels were higher in the polyarticular group [median 64, 95% confidence interval (CI ) <20–136 pg/ml ] when compared with the pauciarticular group (median <20, 95% CI < 20–36 pg/ml ) but not significantly so, P = 0.12 ( Fig. 2a). The lowest levels of TNFa were observed in those children with spondyloarthropathy (median <20, 95% CI < 20–20 pg/ml ) which were significantly lower than the polyarticular group, P = 0.02, but not the pauciarticular group, P = 0.16. Using the high sensitivity assay, TNFa levels were again highest in the polyarticular group (median >33, 95% CI 18 to >33 pg/ml ) (Fig. 2b). Intermediate values were observed in the pauciarticular group (median 30.4, 95% CI 13.5 to >33 pg/ml ), which were not significantly different than the polyarticular group, P = 0.3. The lowest levels were observed in the spondyloarthropathy group (median 9, 95% CI 5.9–18.6 pg/ml ), which were significantly lower than both the pauci- and polyarticular groups, P = 0.01 and P = 0.002, respectively. Significantly higher levels of sTNFRI were observed in the synovial fluid of the polyarticular group than in the pauciarticular group (median 9.2, 95% CI 7.3–11.4 pg/ml vs median 6.3, 95% CI 5.6–7.5 pg/ml ) P = 0.004 (Fig. 3a). Levels were intermediate in the.

(4) TNFa in JCA. 435. (a). (a). (b). (b). F. 2. Synovial fluid TNFa levels within the three clinical subgroups. (a) TNFa levels using the low sensitivity assay; P = 0.02 polyarticular JCA vs spondyloarthropathy. (b) TNFa levels using the high sensitivity assay; P = 0.01 pauciarticular JCA vs spondyloarthropathy; P = 0.002 polyarticular JCA vs spondyloarthropathy. Horizontal bars indicate median values.. F. 3. (a) Synovial fluid levels of sTNFRI within the three clinical subgroups: P = 0.04 polyarticular JCA vs pauciarticular JCA. (b) Synovial fluid levels of sTNFRII within the three clinical subgroups: P = 0.03 polyarticular JCA vs pauciarticular JCA. Horizontal bars indicate median values.. spondyloarthropathy group but were not significantly different from the other two groups (P = 0.4 and P = 1.0, respectively). As with sTNFRI, significantly higher sTNFRII levels were observed in the polyarticular group when compared with the pauciarticular group (median 12.8, 95% CI 11.5–17.9 pg/ml vs median 9.7, 95% CI 7.4–12.1 pg/ml ) P = 0.03 (Fig. 3b). Again there was no significant difference between the subset with spondyloarthropathy (median 8.0, 95% CI 2.0–14.5 pg/ml ) and the other two groups (P = 0.11 and P = 1.0, respectively). Analysis of 16 paired serum and synovial fluid samples revealed sTNFRII to be almost eight times higher than serum sTNFRI: mean 9.1 (range 5.0–16.0 ng/ml ) vs mean 1.25 (range 0.6–2.1 ng/ml ) P = 0.001. sTNFRII levels were four times higher in synovial fluid than in serum: mean 13.7 (range 7.8–28.9 ng/ml ) vs mean 3.5 (range 2.3–6.0 ng/ml ) P = 0.001. Since the biological activity of TNFa appears to be related to the ratio of TNFa to its soluble receptors in vivo, these were calculated for the synovial fluid samples using the results of the high sensitivity assay. Molar ratios were calculated taking the molecular weight of TNFa to be 17 kDa. Since TNFa circulates in a trimolecular complex, apparent molecular weight values for. TNFa were taken to be 51. Values obtained using the high sensitivity assay were used. The molecular weights of sTNFRI and sTNFRII were taken to be 55 kDa and 75 kDa respectively, and molar concentrations of TNFa and its receptors calculated accordingly. The results are displayed in Fig. 4a–c. The TNFRI/TNFa ratio was much higher in the spondyloarthropathy group (median 815, 95% CI 581–1447) than in the pauci- and polyarticular groups (median 286, 95% CI 224–468; P = 0.003 and median 353, 95% CI 257–447; P = 0.003; respectively). There were no significant differences between the polyarticular and pauciarticular groups P = 1.0 ( Fig. 4a). A similar trend was observed in the TNFRII/ TNFa ratios, with the highest ratios being observed in the spondyloarthropathy group (median 980, 95% CI 367–2837), which was greater than that observed in the polyarticular group (median 793, 95% CI 554–980) and the pauciarticular group (median 535, 95% CI 392–688). However, the differences were not statistically significant, P = 0.9 and P = 0.3 (Fig. 4b). Finally, the soluble receptor levels were combined to represent total TNFa inhibitory function and expressed as a ratio to TNFa (Fig. 4c). As expected, the highest ratios were observed in the spondyloarthropathy group (median 1556, 95% CI 1052–4285), which was significantly higher than the pauciarticular group (median 835,.

(5) 436. M. Rooney et al.. (a). statistically significant, P = 0.11. Clinically mild, moderate and marked degrees of inflammation were equally divided between the three disease subgroups ( Fisher’s exact test P = 0.36).. Discussion. (b). (c). F. 4. (a) Synovial fluid molar ratios of sTNFRI/TNFa within the three clinical subgroups: P = 0.003 pauciarticular JCA vs spondyloarthropathy; P = 0.003 polyarticular JCA vs spondyloarthropathy. (b) Synovial fluid molar ratios of sTNFRII/TNFa within the three clinical subgroups. (c) Synovial fluid molar ratios of sTNFRI+TNFRII/TNFa within the three clinical subgroups: P = 0.01 pauciarticular JCA vs spondyloarthropathy; P = 0.05 polyarticular JCA vs spondyloarthropathy. Horizontal bars indicate median values.. 95% CI 720–1071) P = 0.01. Similarly there was a significant difference in the RI+RII/TNFa ratio between the spondyloarthropathy and polyarticular groups, the latter having a median of 1133 (95% CI 817–1418) P = 0.05. Synovial fluid levels of TNFa were compared according to the clinical severity of joint inflammation (data not shown). While there was a trend to higher levels in patients with moderate disease activity, this was not. A number of clinical findings differentiate JCA from adult rheumatoid arthritis. In all forms of JCA erosive change is much less commonly seen. On the contrary, new bone formation leading to ankylosis is a frequent occurrence particularly in the spondyloarthropathies [7]. In pauciarticular disease, even with recurrent episodes of inflammation within a given joint, neither erosion nor new bone formation is common, particularly in early disease [4]. In adult rheumatoid arthritis the inflammatory cytokines have been implicated in the observed erosive disease. TNF is widely expressed and detected in the tissues and fluid of the synovial joint, and has been shown to induce collagenase production [13] and to resorb both cartilage and bone both in vivo and in vitro [15, 16 ]. We and others have shown that the inflammatory cytokines are involved in the different subgroups of JCA [12, 33–36 ]. Lepore et al. observed elevated levels of IL-6 and TNFa in the synovial fluid of children with JCA [33]. Mangge et al. [35] noted a correlation between laboratory markers of inflammation and IL-6, TNFa and the soluble IL-2 receptor in JCA, which were markedly elevated in children with systemic disease. However, the soluble TNFRI receptor was the best indicator of disease activity [35]. Prieur et al. have shown that sTNFRII but not TNFa correlated with the fever spikes in children with systemic JCA [34]. In recent years there has been increasing interest in the role of the soluble TNF receptors in a variety of diseases [30–32]. Girardin et al. [30] noted that in children with severe meningococcal septicaemia, levels of sTNFRI and sTNFRII initially increased as TNF increased. However, at TNFa levels greater than 500 ng/ml no further increase in the soluble receptors was observed and this was associated with a high mortality. Thus, the investigators concluded that an imbalance between TNFa and its naturally occurring inhibitors was implicated in the increased morbidity and mortality. Van Zee et al. [31] observed that soluble TNF receptors increased with inflammation in experimental bacteraemia and circulate at sufficient levels in critically ill patients to block TNFa cytotoxicity in vitro. This would suggest that in experimental bacteraemia, soluble TNF receptors protect against the TNF-mediated effects observed in septic shock. The clinical situation is somewhat different in JCA when compared with acute septic episodes. First, serum TNFa levels are either low or undetectable and certainly never reach the levels observed in septic shock. However, in our study synovial fluid levels in the 100 pg/ml range were not unusual and unlike acute septic episodes may remain elevated for months or years while synovitis persists..

(6) TNFa in JCA. In contrast to our findings, Madson et al. [36 ] observed increases in a number of inflammatory cytokines but not TNFa in both the serum and synovial fluid in JCA. The explanation for this discrepancy may be a technical one, in that a number of anti-TNF antibodies used in assays do not detect complexed TNFa, its predominant form in biological fluids [37]. Cope et al. [24] and Roux-Lombard et al. [25] observed increased levels of both TNFa and its soluble receptors in the synovial fluid of patients with rheumatoid arthritis and that sTNFRs were four to five fold higher in synovial fluid than in serum levels. These results concur with our own where synovial fluid sTNFRs were on average six times greater than serum levels. Furthermore, as they observed, synovial fluid sTNFRII levels were on average 1.5 times higher than TNFRI. Interestingly, in our patients with JCA serum sTNFRII was twice as high as sTNFRI. This contrasts with the findings in sepsis where sTNFRII levels are at least three to four times higher than sTNFRI. These variations in the ratios of the sTNFRs in different disease states and biological fluids may be due to the different standards used in the individual assays employed. Alternatively, they may reflect the different mechanisms of shedding of the two receptors, in particular variations in responsiveness of TNFRI and TNFRII to TNFa stimulation [38], or the effect of proteolytic enzymes, particularly elastase or other metalloproteinases, known to be present in inflamed joints [39]. Furthermore, the cell populations within the joint are quite distinct from those observed in the peripheral blood. Thus, the observed sTNFRI and RII levels may reflect different receptor expression known to occur on different cell types [28–40]. Since the vast majority of synovial fluid cells are neutrophils it is surprising that TNFRII, the most abundant polymorphonuclear leukocytes TNFR, is not higher in JCA synovial fluid when compared with serum. However, the relative increase of TNFRI might be explained by endothelial activation, where TNFRI is in greatest abundance. In this study we observed the highest levels of synovial fluid TNFa in patients with polyarticular disease, with significantly lower levels observed in the pauciarticular group. However, the lowest levels were observed in children with spondyloarthropathy. These findings are not related to the degree of knee joint inflammation observed clinically, since similar proportions of moderate and marked knee joint inflammation were present in the polyarticular and spondyloarthropathy groups. For all children there was a tendency for the sTNFRs to rise with increasing TNFa levels (data not shown) and this concurs with the findings in other inflammatory conditions [30]. However, in the spondyloarthropathy group, despite having the lowest levels of TNFa, the levels of sTNFRI and RII were similar to the other two groups. Thus, significantly higher ratios of sTNFRs/ TNFa were observed in this group. High concentrations of sTNFRs have been shown to abrogate the biological activity of TNFa both in vitro and in vivo [38]. While 300–500 molar excess may be required for almost com-. 437. plete loss of TNFa bioactivity, as little as 30 molar excess of the soluble receptors can have some inhibitory activity [31]. In our study, sTNFRs were present in between 100 and 5000-fold molar excess, thus it is conceivable that there would be considerable differences in the bioactivity of TNFa in synovial fluid with low and high sTNFR levels. Since significantly lower ratios were observed in the polyarticular group (on average 180 molar excess) when compared with the spondyloarthropathy group (molar excess 770), it could be anticipated that the resorptive effects of TNFa on bone and cartilage within the joint in the former group would be expected to be more marked when compared with the latter. This would provide at least part of the explanation for the different clinical courses of these disease subgroups. It could be argued that diseasemodifying and anti-inflammatory agents such as methotrexate, sulphasalazine and steroids might in themselves alter cytokine and receptor concentrations. These drugs were taken by approximately 50% of both the spondyloarthropathy and polyarticular groups: 3/7 children with spondyloarthropathies were given sulphasalazine compound, and 6/13 children with polyarticular JCA were treated with methotrexate. In conclusion we have observed a significant difference in the levels of TNFa and the sTNFRs TNFa ratios in the different subgroups of JCA. Whilst recognizing that TNF is only one of the many cytokines, not to mention growth factors, involved, and this system of agonist/ antagonist is only one of many methods of regulation which might influence bone destruction and remodelling in JCA, these findings reflect different pathophysiologies at play within the different subgroups of JCA, in particular, juvenile spondyloarthropathy.. Acknowledgement Supported by a grant from the Medical Research Council, UK.. References 1. Wood PHN. Nomenclature and classification of arthritis in children. In: Munthe E, ed. The care of rheumatic children. Basel: EULAR, 1978:47. 2. Resnick P, Niwayara G. Juvenile chronic arthritis. In: Resnick P, ed. Bone and joint imaging. Philadelphia: WB Saunders, 1989:288–9. 3. Martel W, Holt JF, Cassidy JT. Roentgenologic manifestations of juvenile rheumatoid arthritis. Am J Roentgenol 1962;88:400–23. 4. Cassidy JT, Martel W. Juvenile rheumatoid arthritis: clinicoradiologic correlations. Arthritis Rheum 1977;20(Suppl. 2):207–11. 5. Lang BA, Schneider R, Reilly BJ et al. Radiologic features of systemic onset juvenile rheumatoid arthritis. J Rheumatol 1995;22:168–73. 6. Ball J. Enthesopathy of rheumatoid and ankylosing spondylitis. Ann Rheum Dis 1971;30:213–23. 7. Bywaters EGL. Pathology of the spondyloarthropathies. In: Colin A, ed. Spondyloarthropathies. Orlando, FL: Grove & Strather, 1984:43–68..

(7) 438. M. Rooney et al.. 8. Will R, Palmer R, Bhalla AK et al. Osteoporosis in early ankylosing spondylitis: a primary pathological event? Lancet 1989;ii:1483–5. 9. Dayer J-M, Arend WP. Cytokines and growth factors. In: Kelly WN, Harris ED, Ruddy S et al. eds. Textbook of rheumatology, 5th ed. Philadelphia: WB Saunders Co, 1997:267–86. 10. di Giovine FS, Nuki G, Duff GW. Tumour necrosis factor in synovial exudates. Ann Rheum Dis 1988;47:768–72. 11. Deleuran BW, Chu CQ, Field M et al. Localization of tumor necrosis factor receptors in the synovial tissue and cartilage–pannus junction in patients with rheumatoid arthritis. Implications for local actions of tumor necrosis factor alpha. Arthritis Rheum 1992;35:1170–8. 12. Rooney M, David J, Symmons J et al. Inflammatory cytokine responses in juvenile chronic arthritis. Br J Rheumatol 1995;34:454–60. 13. Dayer J-M, Beutler B, Cerami A. Cachectin/tumor necrosis factor stimulates collagenase and prostaglandin E2 production by human synovial cells and dermal fibroblasts. J Exp Med 1985;162:2163–8. 14. Sherry B, Cerami A. Cachectin/tumor necrosis factor exerts endocrine, paracrine, and autocrine control of inflammatory responses. J Cell Biol 1988;107:1269–77. 15. Saklatvala J. Tumour necrosis factor alpha stimulates resorption and inhibits synthesis of proteoglycan in cartilage. Nature 1986;322:547–9. 16. Bertolini DR, Nedwin GE, Bringman TS et al. Stimulation of bone resorption and inhibition of bone formation in vitro by human tumour necrosis factors. Nature 1986;319:516–8. 17. Torti FM, Dieckmann B, Beutler B et al. A macrophage factor inhibits adipocyte gene expression: an in vitro model of cachexia. Science 1985;229:867–9. 18. Brennan FM, Chantry D, Jackson A et al. Inhibitory effect of TNF alpha antibodies on synovial cell interleukin-1 production in rheumatoid arthritis. Lancet 1989;ii:244–7. 19. Collins T, Lapierre LA, Fiers W et al. Recombinant human tumor necrosis factor increases mRNA levels and surface expression of HLA-A,B antigens in vascular endothelial cells and dermal fibroblasts in vitro. Proc Natl Acad Sci USA 1986;83:446–50. 20. Vassalli P. The pathophysiology of tumor necrosis factors. Annu Rev Immunol 1992;10:411–52. 21. Dayer J-M. Regulation of IL-1/TNF, their natural inhibitors, and other cytokines in chronic inflammation. Immunologist 1997;5:192–201. 22. van Roon JA, van Roy JL, Gmelig-Meyling FH et al. Prevention and reversal of cartilage degradation in rheumatoid arthritis by interleukin-10 and interleukin-4. Arthritis Rheum 1996;39:829–35. 23. Lacraz S, Nicod LP, Chicheportiche R et al. IL-10 inhibits metalloproteinase and stimulates TIMP-1 production in human mononuclear phagocytes. J Clin Invest 1995;96:2304–10. 24. Cope AP, Aderka D, Doherty M et al. Increased levels of soluble tumor necrosis factor receptors in the sera and synovial fluid of patients with rheumatic diseases. Arthritis Rheum 1992;35:1160–9.. 25. Roux-Lombard P, Punzi L, Hasler F et al. Soluble tumor necrosis factor receptors in human inflammatory synovial fluids. Arthritis Rheum 1993;36:485–9. 26. Ikejima T, Okusawa S, van der Meer JW et al. Induction by toxic-shock-syndrome toxin-1 of a circulating tumor necrosis factor-like substance in rabbits and of immunoreactive tumor necrosis factor and interleukin-1 from human mononuclear cells. J Infect Dis 1988;158:1017–25. 27. Bazzoni F, Beutler B. The tumor necrosis factor ligand and receptor families. N Engl J Med 1996;334:1717–25. 28. Arend WP, Dayer J-M. Inhibition of the production and effects of interleukin-1 and tumor necrosis factor alpha in rheumatoid arthritis. Arthritis Rheum 1995;38:151–60. 29. Hale KK, Smith CG, Baker SL et al. Multifunctional regulation of the biological effects of TNF-alpha by the soluble type I and type II TNF receptors. Cytokine 1995;7:26–38. 30. Girardin E, Roux-Lombard P, Grau GE et al. Imbalance between tumour necrosis factor-alpha and soluble TNF receptor concentrations in severe meningococcaemia. The J5 Study Group. Immunology 1992;76:20–3. 31. van Zee KJ, Kohno T, Fischer E et al. Tumor necrosis factor soluble receptors circulate during experimental and clinical inflammation and can protect against excessive tumor necrosis factor alpha in vitro and in vivo. Proc Natl Acad Sci USA 1992;89:4845–9. 32. Feldmann M, Brennan FM, Maini RN. Role of cytokines in rheumatoid arthritis. Annu Rev Immunol 1996;14: 397–440. 33. Lepore L, Pennesi M, Saletta S et al. Study of IL-2, IL-6, TNF alpha, IFN gamma and beta in the serum and synovial fluid of patients with juvenile chronic arthritis. Clin Exp Rheumatol 1994;12:561–5. 34. Prieur A-M, Roux-Lombard P, Dayer J-M. Dynamics of fever and the cytokine network in systemic juvenile arthritis. Rev Rhum Engl Ed 1996;63:163–70. 35. Mangge H, Kenzian H, Gallistl S et al. Serum cytokines in juvenile rheumatoid arthritis. Correlation with conventional inflammation parameters and clinical subtypes. Arthritis Rheum 1995;38:211–20. 36. Madson KL, Moore TL, Lawrence JM et al. Cytokine levels in serum and synovial fluid of patients with juvenile rheumatoid arthritis. J Rheumatol 1994;21:2359–63. 37. Kreuzer KA, Rockstroh JK, Sauerbruch T et al. A comparative study of different enzyme immunosorbent assays for human tumor necrosis factor-alpha. J Immunol Methods 1996;195:49–54. 38. Spinas GA, Keller U, Brockhaus M. Release of soluble receptors for tumor necrosis factor (TNF ) in relation to circulating TNF during experimental endotoxinemia. J Clin Invest 1992;90:533–6. 39. Hohmann HP, Remy R, Brockhaus M et al. Two different cell types have different major receptors for human tumor necrosis factor (TNF alpha). J Biol Chem 1989;264: 14927–34. 40. Porteu F, Brockhaus M, Wallach D et al. Human neutrophil elastase releases a ligand-binding fragment from the 75-kDa tumor necrosis factor (TNF ) receptor. Comparison with the proteolytic activity responsible for shedding of TNF receptors from stimulated neutrophils. J Biol Chem 1991;266:18846–53..

(8)