Twenty years ago, Verna Wright commented, “clinicians may all too easily spend years writing ‘doing well’ in the notes of patients who become progressively more crippled before their eyes.” Thankfully much of this has changed. Clinicians increasingly understand the advantages of early intervention, particularly in inflammatory joint diseases,1 ,2 and we now have better, more targeted treatments. During this time, however, our methods of objectively assessing and quantifying joint damage have remained largely unchanged. As a result, it is likely that early joint damage in patients goes undetected and untreated.

Although magnetic resonance imaging initially promised much, it has delivered little outside highly specialised centres, where software, coils, and scan sequences change with every passing season. The “gold standard” for assessing joint damage is still the plain radiograph. This mainly images only the bone and is insensitive to change, with reliable differences requiring at least 12 months to evolve. The scoring of radiographs is also time consuming and it does not lend itself to routine monitoring. Importantly, all imaging techniques also only provide a historical view of damage that has already occurred. Even after repeated measures, they are of limited use in informing the clinician of continuing or future damage. Existing serological measures, such as erythrocyte sedimentation rate or C reactive protein, also fall short of requirements.3 They are neither specific to joint disease nor of much use in non-inflammatory conditions. Known genetic and environmental factors that have been associated with various arthritic diseases might also be considered to be markers, but they are often not modifiable nor do they provide any direct information on the extent of joint disease. There is an urgent need for reliable, quantitative, and dynamic tests that will detect damage early and allow the response of treatments targeted at joint destruction to be measured.

Joints are complex organs where bone, cartilage, and synovial tissue are destroyed or altered in disease. Collagen types I, II, and III are present with associated proteoglycan molecules and other glycoproteins (reviewed in detail by Garnero et al and Goldring4 ,5). These components are organised into a highly structured matrix whose composition varies with anatomical site and age. Our knowledge of individual components that make up joint tissues is increasing. We are beginning to understand how matrix components interact with collagens and proteoglycans and how the matrix is dismantled by a variety of proteinases that are up regulated by different cytokines and growth factors in disease. If this release of matrix components could be reliably measured then it might be possible to detect early synovial expansion and the early destruction of cartilage and bone.

A variety of different proteins have been proposed as candidate markers and these include matrix components, products of matrix degradation, cytokines, proteinases (for example, matrix metalloproteinases (MMPs)), and enzyme inhibitors.4 ,5 Such a diverse range of molecules may eventually be needed to answer a variety of clinically relevant questions. For instance, the characteristics of a biological prognostic marker that predicts future damage will be subtly different from a marker that reflects continuing tissue destruction. A prognostic marker should correlate with modifiable, fundamental control points in tissue destruction, whereas a marker of continuing joint damage should reflect the rate at which tissue is lost. In the future, markers may also help us resolve the apparent heterogeneity of existing clinical conditions.

Early studies of biological markers looked at the gross amounts of either proteoglycan or collagen that were released and established that proteoglycan components were usually released before collagen fragments and that the byproducts of collagen assembly into fibrils could be distinguished from those that resulted from degradation.6 It became clear that the synthesis of new matrix was also increased in disease as the tissue attempted repair and so specific markers that could distinguish between repair and degradation were important. An early success was the use of specific cross links to follow degradation of collagen type I in bone, and an early application was to correlate bone loss in osteoporosis with the release of these cross links detected in urine.7Subsequently our knowledge has increased and assays have become more sophisticated.

However, problems need to be overcome and these are illustrated in fig1 and have been described in previous reviews.8 ,9 When a matrix component is released from cartilage, bone, or synovium the most reliable measure can be obtained from synovial fluid. However, this only gives information on a single joint and in a large proportion of patients joint aspiration may not be appropriate or feasible. Urine can be problematic and patients often have difficulty collecting 24 hour urine specimens. Serum is the most convenient body fluid. The passage of molecules from the joint to body fluid is complex and can involve the modification or metabolism of the marker. It is likely that differential processing by the liver or kidneys occurs before such markers reach a steady state in body fluids, and this metabolism may not occur reproducibly in all patients, particularly in the presence of systemic disease.8 In addition, there is a general dilution of components in the serum and urine, and some components are only present at very low concentrations that cannot be measured reliably. The normal extra-articular turnover of connective tissue matrix may also mean that any contribution from affected joints is small and may not significantly alter the overall level. The interpretation of marker levels may also be complex as decreased levels may equally reflect reduced matrix breakdown, decreased synthesis, or impaired marker release from tissues, and in late stage disease, joint tissues such as cartilage may be absent, resulting in misleadingly low levels of certain markers.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Biochemical markers of tissue destruction, inflammation, and repair. Abbreviations and examples of references referring to the use of these markers: S = synthesis marker; D = degradation marker. PINP = amino-terminal type I procollagen propeptide11; PICP = carboxy-terminal type I procollagen propeptide11; bone-specific ALP = bone-specific alkaline phosphatase12; osteocalcin13; DPYD = deoxypyridinoline14; PYD= pyridinoline15; Glc-Gal-PYD = glucosyl-galactosyl-pyridinoline; NTX= type I collagen N-terminal telopeptide16; CTX = type I collagen C-terminal telopeptide-217; ICTP = type I collagen C-terminal telopeptide-1; TRAP = plasma tartrate resistant acid phosphatase18; BSP = bone sialoprotein19; chondroitin sulphate epitopes 846, 3B3, and 7D420 ,21; PIICP = carboxy-terminal type II procollagen propeptide20; PIIANP = amino-terminal type IIA procollagen propeptide; YKL-4022; aggrecan core protein fragments23; keratan sulphate epitopes 5D424 AN9P125; collagen type II neoepitopes COL2–3/4m, COL2–1/4N126; CTX-II = type II collagen C-terminal telopeptide; COMP = cartilage oligomeric protein27; MMPs = matrix metalloproteinases28; TIMPs = tissue inhibitors of matrix metalloproteinases28; PIIINP = amino-terminal type III procollagen propeptide.13

Ideally, it would be helpful to be able to distinguish the precise components that are released by the separate tissues of the joint—namely, bone, cartilage, and the synovium. Table 1 lists the current markers that are thought to indicate either the synthesis of new matrix or the destruction of matrix from these three tissues. For example, the breakdown of collagen in bone can be followed by deoxypyridinoline cross links derived from type I collagen, whereas specific collagenase induced neoepitopes may be used to follow the cleavage of type II collagen from cartilage.

In this issue of the Annals Garneroet al use a panel of markers of bone, cartilage, and synovium in patients with knee osteoarthritis (see p619). They report that these patients have decreased bone turnover but increased cartilage and synovial metabolism. This interesting paper describes levels of urinary type II collagen telopeptide, serum procollagen fragments, and urinary Glc-Gal-PYD in association with cartilage loss. A urinary level of Glc-Gal-PYD, a modified form of the collagen cross link pyridinoline and a putative marker of synovial metabolism, was also shown to be the best predictor of pain and physical function. In osteoarthritis, synovial involvement is usually considered mild, if present at all. The finding that a synovial marker correlates best with symptoms may suggest that synovial involvement is significant.

These studies are interesting and, although limited, do suggest that it may be possible to follow specifically the activity of individual tissues within the joint in different diseases.

The potential for reliable and responsive markers is large. This and other recent studies have suggested that it is possible to separate inflammatory events from destructive events. For example, Cunnaneet al showed that following changes in serum levels of MMP-3 allowed inflammatory events to be studied in rheumatoid arthritis (RA), whereas changes in serum MMP-1 levels followed the destruction of cartilage and bone.10 Although the current study is in osteoarthritis it will be interesting to see if the new markers described in this study have a role in RA and whether levels in early joint disease allow the reliable prediction of those patients whose cartilage and bone will eventually be destroyed. Early warning of the initiation of matrix breakdown would prompt earlier treatment so preventing much of the destruction of cartilage and bone that leads to subsequent disability. Further studies, and particularly prospective studies, are still required to validate this and other markers as these are proposed/discovered.