Because of the limitations of current therapy, interest has arisen in agents that are not principally analgesic but that alter pathogenic mechanisms within the osteoarthritic joint—disease-modifying drugs [1]. Although claims have been made that various agents are "chondroprotective," no agent has yet been shown to modify osteoarthritis in a placebo-controlled clinical trial in humans. However, recent studies in animals show that pharmacologic modification is indeed feasible.

In considering guidelines for testing new drugs, a committee of the International League of Associations for Rheumatology [1] recommended that the designation "disease-modifying drug for osteoarthritis" be reserved for agents that prevent, retard progression of, or reverse morphologic changes; an effect only on the biochemistry or metabolism of cartilage matrix molecules or the concentration in body fluids of molecules derived from cartilage or bone in the joint ("osteoarthritis marker" molecules) was considered insufficient to qualify an agent as a disease-modifying drug.

In animal models, several agents meet the criterion defined above. They range from empirical compounds (for example, tissue extracts) to site-specific inhibitors designed to fit precisely into the catalytic site of collagenase; they include tribenoside, tamoxifen, diacerein, chloroquine, hyaluronic acid, glucocorticoids, and tranexamic acid. Excellent reviews have been published [2-4]. In this editorial, I summarize the evidence about the disease-modifying activity of NSAIDs, heparinoids, and tetracyclines.

Claims that some NSAIDs have a disease-modifying effect have been based almost exclusively on in vitro evidence that the drugs modify proteoglycan or collagen metabolism, cytokine-mediated matrix degeneration, secretion or activity of neutral matrix metalloproteinases, or the effects of toxic oxygen metabolites. Because fatal NSAID-induced gastrointestinal hemorrhage (related to inhibition of prostaglandin synthesis) often develops before arthritis becomes apparent in the species typically used in models of osteoarthritis (for example, mice, guinea pigs, dogs, rabbits), it has not been possible to extensively test the in vivo disease-modifying capacity of NSAIDs in animal models. No evidence indicates that NSAIDs have this effect in humans.

However, in two studies [5, 6], tiaprofenic acid (a proprionic acid derivative) has been shown to reduce the severity of osteoarthritis in dogs with cruciate ligament deficiency. In the same model, when gastrointestinal toxicity was prevented by administration of omeprazole [7], diclofenac did not significantly protect against joint damage, although tenidap (which inhibits synthesis of prostaglandins and leukotrienes and which may modulate synthesis of interleukin-1 [8]) had a striking disease-modifying effect [7]. Whether concomitant gastroprotective treatment will show that other NSAIDs also have this effect remains to be seen. Similarly, the disease-modifying activity of NSAIDs that selectively inhibit cyclo-oxygenase II should be assessed.

Several heparinoids have been shown to modify osteoarthritis in animal models of the disease [2-4]. Glycosaminoglycan polysulfate (Arteparon; an extract of bovine tracheal and bronchial cartilage) and glycosaminoglycan peptide complex (Rumalon; an extract of calf cartilage and bone marrow) contain chondroitin-4-sulfate, chondroitin-6-sulfate, and peptides. Both stimulate cartilage matrix synthesis and have activities as protease inhibitors. Sodium pentosan polysulfate (a polysaccharide sulfate ester prepared from beech hemicellulose) is a heparinoid (like glycosaminoglycan polysulfate and glycosaminoglycan peptide complex), but sodium pentosan has the advantage of lacking antigenic protein constituents. It is a potent inhibitor of matrix metalloproteinases and leukocyte elastase, and it may down-regulate metalloproteinase levels by interfering with the binding of transcription factor [9]. Each of these heparinoids requires intramuscular or intra-articular administration. It is therefore notable that calcium pentosan polysulfate, which is well absorbed after oral administration, reduced loss of cartilage proteoglycans in an animal model of inflammatory arthritis [10]; it has not yet been evaluated as a disease-modifying drug.

Neither of the above pentosans has been tested in humans. In patients with knee osteoarthritis who received a series of intramuscular injections of glycosaminoglycan polysulfate or glycosaminoglycan peptide complex every 6 months for 5 years, radiographic progression of osteoarthritis was reported to be slower in patients who received either agent than in controls, and improvement was noted in several functional measures [11]. However, failure of the study to include a placebo group or to control for NSAID use cast doubt on the importance of these findings. The lack of convincing evidence of a disease-modifying effect in humans, reports of bleeding attributed to the heparinoid structure of glycosaminoglycan polysulfate, reports of anaphylaxis related to the presence of antigenic protein components, and concern about possible transmission of the agent responsible for bovine spongiform encephalopathy have led to the removal of glycosaminoglycan polysulfate and glycosaminoglycan peptide complex from the market in the European countries in which they had been available.

Evidence that doxycycline inhibits the activity of collagenase and gelatinase in extracts of articular cartilage recently led to in vivo studies in a canine model of osteoarthritis that showed a marked disease-modifying effect with oral administration of the drug [12]. Reports that cartilage damage in guinea pig [13] and rabbit [14] models of osteoarthritis was also reduced by tetracycline administration bolster the observations in the canine model.

Previous studies suggested that the pathologic changes of osteoarthritis progress very slowly. This led to the assumption that many participants and years of treatment would be needed to show a disease-modifying effect in humans; this assumption deterred efforts by the pharmaceutical industry to develop disease-modifying drugs. How-ever, the mean rate of cartilage loss in osteoarthritis (based on the reduction of joint-space width in standard radiographs) is much more rapid than previously believed [1, 15]. Further, standardization of radioanatomical positioning of the knee can reduce the variability in measurements of joint space to only 1% to 2% [16], and computerized analysis of the digitized radiograph can reduce the variability inherent in manual measurements.

Clinical trials of a disease-modifying drug can be facilitated by focusing on a joint at high risk for osteoarthritis; for example, 50% of middle-age obese women with radiographic evidence of unilateral knee osteoarthritis will develop the disease in the contralateral knee within the next 2 years [17]. Several strategies can optimize retention of participants in a long-term clinical trial and can optimize compliance with the dosing regimen [18], as shown by the 82% retention rate in a 48-week trial of minocycline in patients with rheumatoid arthritis [19]. A "faintness-of-heart" test [20] (to exclude noncompliers before random assignment into treatment groups) and use of computerized medicine caps (to permit study personnel to direct their efforts toward enhancing compliance among those patients who can best benefit from such efforts) may further increase retention and compliance. With the use of these techniques, the logistics of a clinical trial of a disease-modifying drug need not be daunting; efficacy might be shown in a controlled clinical trial with only 2 to 2.5 years of treatment and with enrollment of fewer than 500 participants.

Clinical researchers in osteoarthritis, members of the pharmaceutical industry, and representatives of regulatory authorities (for example, the Food and Drug Administration) need to agree on the appropriate outcome measures and assessment tools for evaluation of disease-modifying drugs. Once a drug possessing this activity in humans has been identified, it will be possible to determine whether reduction in the rate of cartilage loss is accompanied by reductions in joint pain, the frequency of joint arthroplasty, and the rate of disability.