J. Kaufmann, A. Mueller, A. Voigt, H. D. Carl 1,A.Gursche 2, J. Zacher 2,G.SteinandG.Hein

Transcription

1 Rheumatology 2003;42: doi: /rheumatology/keg102, available online at Hydroxypyridinium collagen crosslinks in serum, urine, synovial fluid and synovial tissue in patients with rheumatoid arthritis compared with osteoarthritis J. Kaufmann, A. Mueller, A. Voigt, H. D. Carl 1,A.Gursche 2, J. Zacher 2,G.SteinandG.Hein Objective. To investigate the relationship between inflammation markers and content of pyridinium crosslinks in hydrolysates of synovial tissue and to specify the significance of urinary excreted pyridinoline, released primarily from collagen I and II of bone and cartilage, and deoxypyridinoline released especially from collagen I of bone and dentin, dependent on disease activity in rheumatoid arthritis (RA). Methods. Synovial tissue and fluid from knee endoprosthesis surgery, as well as simultaneously obtained serum and urine, were collected from 12 patients with inactive RA or RA with low disease activity wira: C-reactive protein (CRP) <28 mgulx, 10 with active RA (: CRP 028 mgul)and21withoa.afterpreparationofthe synovial tissue, including hydrolysis, completely released synovial pyridinoline and deoxypyridinoline crosslinks as well as those from synovial fluid, serum and urine were investigated using a gradient ion-paired reversed-phase HPLC method. Crosslink levels in synovial tissue are expressed as molumol collagen, assuming 300 residues of hydroxyproline per collagen molecule, also measured by HPLC. Results. In the synovial tissue of patients we found significantly elevated total pyridinoline concentrations and pyridinolineudeoxypyridinoline (PyruDpyr) quotients compared with the ira and OA controls, indicating an elevated crosslinking density of mature synovial tissue collagen with increased activity of RA. Pyridinoline levels and the PyruDpyr ratio were correlated with those of urine and with acute-phase reactants in RA patients. Compared with serum crosslink levels, which were unrelated to disease activity, the urinary concentration of pyridinoline was increased by a factor of 2 and showed a simultaneous increase with increasing synovitis. Conclusion. Both crosslinking density and degradation of mature collagen from synovial tissue depend on the disease activity in RA. Urinary excretion of associated crosslinks, expressed as the PyruDpyr ratio, correlates with those in synovial tissue and may be confirmed as a marker of synovial tissue collagen degradation. We suggest that increased crosslinking of mature collagen in the synovial tissue of RA is related to an inflammation-dependent regulation of collagen synthesis in activated synovial fibroblasts, in which lysyl oxidase represents the final enzymatic step for crosslinking. KEY WORDS: Rheumatoid arthritis, Collagen degradation, Hydroxypyridinium crosslinks, Synovial tissue, Synovial fluid, Urinary crosslink excretion, HPLC method. Department of Internal Medicine IV, University of Jena, 1 Division of Orthopaedic Rheumatology, Department of Orthopaedic Surgery, University of Erlangen-Nuernberg and 2 Clinic for Orthopaedics, Klinikum Berlin-Buch, Germany. Submitted 27 July 2001; revised version accepted 9 August Correspondence to: J. Kaufmann, Clinic for Internal Medicine IV, Department of Rheumatology and Osteology, Friedrich-Schiller-University of Jena, Jena, Germany. joerg.kaufmann@med.uni-jena.de 314 ß 2003 British Society for Rheumatology

2 Collagen crosslinking density of RA synovium 315 Rheumatoid arthritis (RA) is characterized by an autoimmunological inflammation. Synovitis of affected joints generates a destructively growing tissue that causes degradative and erosive lesions of cartilage and juxtaarticular bones w1x. In contrast, osteoarthritis (OA) is characterized by degenerative loss of cartilage, bone remodelling and an intermittent synovitis without the destructive potency seen in RA w2, 3x. Different diagnostic methods such as conventional radiography, tomography procedures, sonography and arthroscopy have been developed to document the state of destruction in both RA and OA, but it remains difficult to describe the dynamics of degradation depending on the current inflammatory activity. It is commonly accepted that the occurrence of the collagen hydroxypyridinium crosslinks pyridinoline and deoxypyridinoline in urine is an indication of the breakdown of mature collagen w4x. Deoxypyridinoline is released mainly from bone and dentin and, because of its low turnover, in insignificant amounts from collagen of tendons and aorta, whereas pyridinoline is the major crosslink of cartilage and is also prevalent in the collagen of bone and other tissues w5x. There are some reports which suggest that increased excretion of pyridinium crosslinks correlates positively with disease activity in RA patients w6, 7x. But analysis of these crosslinks in joint tissues is rare and only performed on a small number of patients w8, 9x. Asdata from studies showing the level of synovial tissue collagen crosslinking in the presence of active RA are lacking, the origins of elevated concentrations of urinary crosslinks remains unclear. Other pathological processes associated with RA, such as osteoporosis, infection of internal organs, coexisting osteoarthritis or treatment with corticosteroids may contribute to increased crosslink excretion w10, 11x. In this study, we investigated pyridinoline and deoxypyridinoline simultaneously in synovial tissue, synovial fluid, serum and urine in patients with RA and OA, to examine the effect of inflammation on the crosslinking of synovial tissue mature collagen and to specify the significance of the urinary excreted pyridinium crosslinks as markers of collagen degradation in the presence of disease activity in RA. Patients and methods Subjects and samples Twenty-two patients suffering from RA, diagnosed according to the 1987 revised criteria of the American Rheumatism Association (ARA), and 21 patients with osteoarthritis were studied. According to their C-reactive protein (CRP), RA patients were divided into two groups: 12 patients with inactive RA or low disease activity of the RA (ira: CRP <28 mgul) and 10 patients with active RA (: CRP 028 mgul). The characteristics of patients are summarized in Table 1. Articular prechondral synovial tissue and synovial fluid was obtained from total knee replacement surgery and thereafter immediately frozen in liquid nitrogen and stored at 2808C. Serum and urine were collected in the morning before breakfast (before surgery and perioperative steroid boli) and were centrifuged at 3000 g for 5 min at 48C andstoredat2808c before being assayed. Collection procedures were in accordance with the principles of the 1975 Declaration of Helsinki, as revised in Analytical techniques Standard laboratory methods were used including nephelometry for CRP, Westergren method for erythrocyte sedimentation rate (ESR) and Jaffe rate technique in duplicate for urinary creatinine. The synovial tissue was fragmented, freeze-dried under vacuum, weighed and approximately 20 mg were restored with 3:1 trichloromethane:methanol for exhaustive extraction. One aliquot of the synovial tissue collagen sample as well as samples of synovial fluid, serum and urine were prepared for HPLC measurement of pyridinoline and deoxypyridinoline as described w7, 12x, including pretreatment of samples by hydrolysis (32% hydrochloric acid, 17 h, 1108C) and fractionation of hydrolysates by manual partition chromatography on CF1 cellulose to remove interfering fluorescent components w13x. Investigation of pyridinoline and deoxypyridinoline was performed using ion-paired reversed-phase HPLC and gradient separation under fluorescence detection w14x, modified according to Mueller et al. w7x. For quantification, the HPLC method was calibrated with a purified standard containing defined amounts of pyridinoline and deoxypyridinoline. The cycle time for one determination was 30 min. The limits of detection of the assay were 25 fmol for pyridinoline and 56 fmol for deoxypyridinoline. Another synovial tissue collagen sample aliquot was taken for HPLC measurement of hydroxyproline w15x. Assuming that 300 hydroxyproline residues represent one collagen molecule TABLE 1. Different diagnosis and activity groups in relation to epidemiological parameters and serological markers of inflammation ira Significance levels RA (n=22) CRP<28 mgul (n=12) CRP028 mgul (n=10) OA (n=21) vs ira ira RA all ESR (mm 1st h) 38.7" " " "17.3 P<0.05 P<0.05 n.s. P<0.05 CRP (mgul) 28.6" " " "9.2 P< P< n.s. P<0.01 Sex ratio (femaleumale) 20u2 10u2 9u1 16u5 Age (yr) (meanurange) 59.1u u u u49 82 n.s. P<0.05 n.s. P<0.05 Disease duration (yr) (meanurange) 10.3u u u u1 18 n.s. n.s. n.s. n.s. n.s., not significant.

3 316 J. Kaufmann et al. w8x, pyridinium crosslinks in synovial tissue were specified as molumol collagen. Pyridinoline and deoxypyridinoline levels in synovial fluid and serum were expressed as nmolul, and urinary crosslinks after creatinine correction as nmolumol creatinine. Statistical analysis Analysis of variance was performed using the Kruskal Wallis test. For comparison of non-parametric variables between the patient groups the Mann Whitney U-test was used. Correlation coefficients were analysed according to Spearman. P values (0.05 were considered to be significant. All calculations were performed using the statistical software package SPSS version Results Patient characteristics, including age, duration of disease, sex distribution and acute-phase reactant levels are shown in Table 1, together with comparison of means of these parameters within the diagnosis groups. The means for pyridinoline, deoxypyridinoline and calculated pyridinolineudeoxypyridinoline (PyruDpyr) ratio in synovial tissue, synovial fluid, serum and urine within the diagnosis groups, ira and OA are displayed in Table 2. Total pyridinoline concentrations in synovial tissue were found to be significantly elevated in patients with compared with those with ira (P<0.005) and OA (P<0.05). No difference in pyridinoline levels was seen between ira and OA patients and in deoxypyridinoline levels between all groups. Subsequently the PyruDpyr ratio was shown to be significantly increased in all RA patients (21.8"7.7) compared with OA (P<0.001) and in the group compared with ira (P<0.05) and OA (P<0.0005) (Fig. 1). The PyruDpyr ratio in synovial tissue of RA patients was correlated with the acutephase reactants ESR (r=0.571, P<0.05) and CRP (r=0.416, P<0.05) (Fig. 2) In synovial fluid and serum, collected on the same day, we measured a significantly smaller ratio of PyruDpyr compared with that of synovial tissue (P<0.05) (Fig. 1), but there was no significant difference in the PyruDpyr ratio in synovial fluid and serum between the diagnosis groups (except for ira ) and no correlation of PyruDpyr ratio in synovial fluid and serum with ESR and CRP in RA and OA. In synovial tissue of RA patients there was a strong correlation between pyridinoline and deoxypyridinoline levels (r=0.944, P<0.0001), but not in synovial fluid, serum and urine. The PyruDpyr ratio in urine was found to be elevated in patients compared with ira (P<0.005) and OA (P<0.005) (Fig. 1). In RA, but not in OA, the urinary PyruDpyr ratio was found to be correlated with that of synovial tissue (r=0.734, P<0.005) (Fig. 3). Compared with the serum PyruDpyr ratios, found in equal ranges in all diagnosis groups and independent of those of synovial tissue, the urinary ratios increased according to an averaged factor of 1.96 in RA and 2.14 in OA, indicating a renal concentration of pyridinoline. Discussion The levels of hydroxypyridinium collagen crosslinks in urine were found to be in agreement with the data of previous studies. Urinary excreted crosslinks in RA patients were repeatedly reported to be in ranges of (pyridinoline) and (deoxypyridinoline) nmolumol creatinine, and were related to disease activity w12, 16, 17x. In OA patients, urinary crosslinks were also similar to those reported previously w9, 10x. The quantitative detection of pyridinoline and deoxypyridinoline by HPLC in synovial fluid and serum has been reported. Mueller et al. w7x detected pyridinoline and deoxypyridinoline in synovial fluid and serum of RA patients and found similar amounts to those described in this study. TABLE 2. Levels of pyridinoline (Pyr) and deoxypyridinoline (Dpyr) in synovial tissue, synovial fluid, serum and urine ira Significance levels CRP<28 mgul (n=12) CRP028 mgul (n=10) OA (n=27) vs ira ira Synovial tissue Pyr (molumol collagen) 0.39" " "1.28 P<0.005 P<0.05 n.s. Dpyr (molumol collagen) 0.03" " "0.12 n.s. n.s. n.s. PyruDpyr 18.3" " "6.2 P<0.05 P< P<0.05 Synovial fluid Pyr (nmolul) 58.4" " "33.4 n.s. P<0.005 P<0.005 Dpyr (nmolul) 11.9" " "10.5 n.s. n.s. n.s. PyruDpyr 9.7" " "5.5 n.s. n.s. P<0.01 Serum Pyr (nmolul) 5.4" " "2.2 P<0.05 P<0.01 n.s. Dpyr (nmoluul) 7.1" " "12.1 n.s. n.s. n.s. PyruDpyr 3.4" " "3.2 n.s. n.s. n.s. Urine Pyr (nmolumol creatinine) 54.7" " "25.3 P<0.020 P<0.005 n.s. Dpyr (nmolumol creatinine) 19.8" " "11.1 n.s. n.s. n.s. PyruDpyr 3.5" " "5.8 P<0.005 P<0.005 n.s. Urinary crosslinks were corrected for creatinine. Calculation of PyruDpyr ratio and comparison of means within the diagnosis groups.

4 Collagen crosslinking density of RA synovium 317 FIG. 1. The PyruDpyr ratio estimated in synovial tissue, synovial fluid, serum and urine (for data and statistics see Table 2). j active RA (CRP028 mgul); j inactive RA (CRP<28 mgul); I osteoarthritis. James et al. w13x quantified pyridinoline but not deoxypyridinoline in synovial fluid and serum of OA patients, and both crosslinks in the sera of patients with Paget s disease using HPLC. But there are only two reports on the quantification of crosslinks in synovial tissue of RA patients, performed on small patient groups and without any relation to disease activity. Pearce et al. w8x described synovial tissue levels of approximately 1 mol pyridinolineumol collagen and 0.1 mol deoxypyridinolineu mol collagen with no differences between the four RA and nine OA patients. From our results we confirm pyridinoline and deoxypyridinoline levels in this range. Takahashi et al. w9x reported similar ranges of synovial tissue pyridinoline and deoxypyridinoline levels in 10 RA and 10 OA patients with no difference between diagnosis groups and found a PyruDpyr ratio of approximately 25 in synovial tissue, 50 in cartilage and 3 in bone. We found an increased amount of total pyridinoline crosslinks per mol of collagen in the synovial tissue of RA patients, significantly correlated with the activity of disease. Because of the complete hydrolysis of synovial tissue, the HPLC measurement estimates both the pyridinium compounds crosslinking the synovial tissue collagen and related collagen crosslinks which have migrated from the collagen of associated joint tissues such as cartilage and bone into the synovial tissue. Assuming that pyridinoline and deoxypyridinoline from cartilage and bone penetrate in insignificant amounts and similar rates into synovial tissue, the elevation of total pyridinoline and PyruDpyr ratio, respectively, in synovial tissue and urine indicates an increased crosslinking density of mature collagen in synovial tissue of. Therefore the increased urinary crosslink excretion found to be correlated with higher disease activity in RA may possibly be due to both the increased activity of fibroblast-derived collagenases and metalloproteinases w1x and an increased crosslinking activity that provides more substrate for degradation. The correlation of increased total PyruDpyr in synovial tissue and its urinary excretion indicates that the crosslink measurement in the synovial tissue of the knee joint seems to point to synovial metabolism of pyridinium crosslinks under conditions of inflammation. Polyarticular involvement and the extent to which other, partially inflammation-independent collagen-destructing processes (such as post-menopausal osteoporosis or steroid therapy) contribute to urinary pyridinium excretion do not influence the specifity and sensitivity and the significance of differences of urinary crosslink measurement between the diagnosis groups. Urinary pyridinoline excretion may be confirmed as a marker of inflammation-mediated synovial tissue collagen destruction in RA and OA patients w12, 18x. Moreover, for the first time we describe an increased pyridinoline crosslinking of synovial tissue mature collagen, correlating with activity parameters of RA. Results obtained demonstrate clearly that a higher grade of inflammation in RA provides not only an increased breakdown of synovial tissue collagen, but also an increased crosslinking in the inflamed tissue. Oxidative deamination of lysine and hydroxylysine is required for collagen crosslinking specifically catalysed by the copper-dependent lysyl oxidase. Intermediate aldehydes are formed and condense spontaneously to crosslinked collagen forming a mature extracellular matrix w19, 20x. An inflammation-dependent increased lysyl oxidase expression in fibroblasts has been described in different types of fibrosis w21x, in wound healing w22x and in the skin of scleroderma patients w23x. Conforming to these immunohistochemical studies, lysyl oxidase expression of cultured osteoblasts and fibroblasts could be regulated with inflammation mediators including transforming

5 318 J. Kaufmann et al. FIG. 2. Correlation between PyruDpyr ratio in synovial tissue and the acute-phase reactants ESR and CRP in RA patients (for ESR: r=0.571, P<0.05, equation y=1.73x23.92; for CRP: r=0.416, P<0.05, equation y=1.01x+5.72). growth factor-b1, interleukin (IL)-1b and prostaglandin E2 w24 26x. Furthermore, other cytokines w27x, oxidative stress w28x or direct T-cell contact w29x regulate the fibroblastic collagen synthesis in various tissues. However, there are no investigations on crosslinking density in collagen released from synovial fibroblasts with respect to these factors. Moreover, in collagen there is an underlying increase in crosslinking with age, which is further enhanced in active inflammation of RA. But these modifications seem to be associated with the formation of advanced glycation end products, such as the pentosidine crosslink w30, 31x, and are not relevant for the measurement of increased pyridinium crosslinks. Chen et al. w32x have described markedly increased pentosidine crosslinking with ageing in human yellow ligament hydrolysates, whereas the pyridinoline and deoxypyridinoline crosslinks were found to be unchanged with age. In FIG. 3. Correlation between PyruDpyr ratio in synovial tissue and urine in RA patients (r=0.734, P<0.005, equation y=0.27x21.0). accordance with that, we found no correlation of synovial tissue pyridinoline and deoxypyridinoline with the age of patients and duration of disease (data not shown). The consequences of the formation of highly crosslinked collagen in active RA are still unknown. Possibly the inflammed tissue shows an increased antigenic activity triggering further autoimmune processes in active RA. Recently Kim et al. w33x reported increased levels of anti-type II collagen IgG antibodies in RA compared with OA, correlating positively with acutephase reactants and the pro-inflammatory cytokines tumour necrosis factor-a and IL-6 in RA. Different results have been published concerning the excretion of deoxypyridinoline in RA and OA patients. Seibel et al. w18x found increased excretion of deoxypyridinoline in RA patients and concluded this to be a result of increased bone degradation. Similarly, Kollerup et al. w17x found a 40% increase of urinary deoxypyridinoline in RA patients and found a significant correlation with ESR, CRP and number of swollen joints. In contrast to this, unchanged deoxypyridinoline excretion has been reported in RA patients compared with normal controls in two other studies w6, 34x. Discrepancies have been discussed with respect to different stages of disease activity, menopausal status, age and sex between cases and controls w17x. We also did not find any significant differences in urinary deoxypyridinoline excretion between RA and OA patients and in comparison with normal controls, as measured with the same HPLC method and published from our group w12x. Furthermore, no difference in total deoxypyridinoline levels could be found in hydrolysed synovial tissue from RA and OA patients. This may be due to a very slow activity of the bone destruction process w17x anduor a slow penetration of deoxypyridinoline from bone into

6 Collagen crosslinking density of RA synovium 319 synovial tissue. Thus, increased PyruDpyr ratios most probably reflect elevated collagen II degradation of non-osseous joint tissues. In conclusion, observations from this study demonstrate urinary excretion of pyridinium crosslinks, expressed as the PyruDpyr ratio, to be a marker of collagen degradation of joint tissues in RA patients correlating with current disease activity. Furthermore, an increase of synovial tissue total collagen crosslinking that correlates with the ongoing level of disease activity implicates an influence of RA inflammation on the structure and properties of the extracellular matrix of joint tissues. References 1. Manicourt DH, Fujimoto N, Obata K, Thonar EJ. Levels of circulating collagenase, stromelysin-1, and tissue inhibitor of matrix metalloproteinases 1 in patients with rheumatoid arthritis. Relationship to serum levels of antigenic keratan sulphate and systemic parameters of inflammation. Arthritis Rheum 1995;38: Sambrook PN, Reeve J. Bone disease in rheumatoid arthritis. Clin Sci 1988;74: Bijlsma JWJ. Bone metabolism in patients with rheumatoid arthritis. Clin Rheumatol 1988;7: Gunja-Smith Z, Boucek RY. Collagen crosslink components in human urine. Biochem J 1981;197: Eyre DR, Koob JJ, van Ness KP. Quantitation of hydroxypyridinium crosslinks by HPLC. Anal Biochem 1984;137: Spector TD, James IT, Hall GM, Thompson PW, Perrett D, Hart DJ. Increased levels of urinary collagen crosslinks in females with rheumatoid arthritis. Clin Rheumatol 1993;12: Mueller A, Hein G, Franke S et al. Quantitative analysis of pyridinium crosslinks of collagen in the synovial fluid of patients with rheumatoid arthritis using high-performance liquid chromatography. Rheumatol Int 1996;16: Pearce D, Fisher E, Randall A, Will R, Kent GN, Garcia-Webb P. Urinary levels of pyridinoline and deoxypyridinoline may be influenced by turnover of non-osseous tissues in patients with joint disease. Br J Rheumatol 1995;34: Takahashi M, Kushida K, Hoshino H et al. Concentrations of pyridinoline and deoxypyridinoline in joint tissues from patients with osteoarthritis or rheumatoid arthritis. Ann Rheum Dis 1996;55: Delmas PD, Schlemmer A, Gineyts E, Riis B, Christiansen C. Urinary excretion of pyridinoline crosslinks correlates with bone turnover measured on iliac crest biopsy in patient with vertebral osteoporosis. J Bone Miner Res 1991;6: Luckert BP, Raisz LG. Glucocorticoid induced osteoporosis: Pathogenesis and management. Ann Int Med 1990; 112: Hein G, Franke S, Mueller A, Braeunig E, Eidner T, Stein G. The determination of pyridinium crosslinks in urine and serum as a possible marker of cartilage degradation in rheumatoid arthritis. Clin Rheumatol 1997; 16: James I, Crowley C, Perrett D. Assay of pyridinium crosslinks in serum using narrow-bore ion-paired reversed-phase high-performance liquid chromatography. J Chromatogr 1993;612: Colwell A, Russell RG, Eastell R. Factors affecting the assay of urinary 3-hydroxy pyridinium crosslinks of collagen as markers of bone resorption. Eur J Clin Invest 1993;23: Palmerini CA, Fini C, Floridi A, Morelli H, Vedovelli A. High-performance liquid chromatographic analysis of free hydroxyproline and proline in blood plasma and of free and peptide-bound hydroxyproline in urine. J Chromatogr 1985;339: Black D, Duncan A, Robins SP. Quantitative analysis of the pyridinium crosslinks of collagen in urine using ion-paired reversed-phase high-performance liquid chromatography. Anal Biochem 1988;169: Kollerup G, Hansen M, Horslev-Petersen K. Urinary hydroxypyridinium cross-links of collagen in rheumatoid arthritis. Relation to disease activity and effects of methylprednisolone. Br J Rheumatol 1994;33: Seibel M, Duncan A, Robins SP. Urinary hydroxypyridinium crosslinks provide indices of cartilage and bone involvement in arthritic diseases. J Rheumatol 1989; 16: Kagan HM. Characterization and regulation of lysyl oxidase. In: Mecham RP, ed. Biology and regulation of extracellular matrix: A series. Regulation of matrix accumulation, Vol. 1. Orlando: Academic Press, 1986: Kagan HM, Trackman PC. Properties and function of lysyl oxidase. Am J Respir Cell Mol Biol 1991;5: Kagan HM. Lysyl oxidase: mechanism, regulation and relationship to liver fibrosis. Pathol Res Pract 1994; 190: Chen CJ, Kang J, Shieh TY. Changes in lysyl oxidase activity and calcium content during the healing of tooth extraction wounds. Kaohsiung J Med Sci 1997;13: Chanoki M, Ishii M, Kobayashi H et al. Increased expression of lysyl oxidase in skin with scleroderma. Br J Dermatol 1995;133: Boak A, Roy R, Berk J et al. Regulation of lysyl oxidase expression in lung fibroblasts by TGF-b1 and prostaglandin E2. Am J Respir Cell Mol Biol 1994;11: Feres-Filho EJ, Menassa GB, Trackman PC. Regulation of lysyl oxidase by basic fibroblast growth factor in osteoblastic MC3T3-E1 cells. J Biol Chem 1996;271: Roy R, Polgar P, Wang YY, Goldstein RH, Taylor L, Kagan HM. Regulation of lysyl oxidase and cyclooxygenase expression in human lung fibroblasts: interactions among TGF-b, IL-1b, and prostaglandin E. J Cell Biochem 1996;62: Siwik DA, Chang DLF, Colucci WS. Interleukin-1 beta and tumor necrosis factor-alpha decrease collagen synthesis and increase matrix metalloproteinase activity in cardiac fibroblasts in vitro. Circulation Res 2000; 86: Siwik DA, Pagano PJ, Colucci WS. Oxidative stress regulates collagen synthesis and matrix metalloproteinase activity in cardiac fibroblasts. Am J Physiol Cell Physiol 2001;280:C53 C Rezzonico R, Burger D, Dayer JM. Direct contact between T lymphocytes and human dermal fibroblasts or synoviocytes down-regulates type I and III collagen production via cell-associated cytokines. J Biol Chem 1998;273: Miyata T, Ishiguro N, Yasuda Y et al. Increased pentosidine, an advanced glycation end product, in

7 320 J. Kaufmann et al. plasma and synovial fluid from patients with rheumatoid arthritis and its relation with inflammatory markers. Biochem Biophys Res Commun 1998;244: Chen JR, Takahashi M, Suzuki M, Kushida K, Miyamoto S, Inoue T. Comparison of the concentrations of pentosidine in the synovial fluid, serum and urine of patients with rheumatoid arthritis and osteoarthritis. Rheumatology 1999;38: Chen JR, Takahashi M, Kushida K et al. Direct detection of crosslinks of collagen and elastin in the hydrolysates of human yellow ligament using single-column high performance liquid chromatography. Anal Biochem 2000; 278: Kim WU, Yoo WH, Park W et al. IgG antibodies to type II collagen reflect inflammatory activity in patients with rheumatoid arthritis. J Rheumatol 2000;27: Astbury C, Bird HA, McLaren AM, Robins SP. Urinary excretion of pyridinium crosslinks of collagen correlated with joint damage in arthritis. Br J Rheumatol 1994; 33:11 5.