Glucosamine and Chondroitin for Treating Symptoms of Osteoarthritis: Title and subTitle BreakEvidence Is Widely Touted but Incomplete

Osteoarthritis (OA) is the most common form of arthritis and is a leading cause of physical disability, increased health care usage, and impaired quality of life.^{1 - 2} An estimated 12% of the US population aged 25 years and older (nearly 21 million persons in 1990) have clinical signs and symptoms of OA.³

Although nonsteroidal anti-inflammatory drugs (NSAIDs) are commonly used to treat OA and have been proved effective, their widespread use is associated with significant potential toxic effects, especially in the elderly population. Although the cyclooxygenase-2 inhibitors have a lower rate of gastrointestinal tract complications than conventional NSAIDS, there remains an urgent need for finding pharmacological therapies for OA that are both effective and relatively safe. In this regard, the nutraceuticals glucosamine sulfate and chondroitin sulfate have attracted substantial attention by the public and in the medical literature.^{4 - 5}

Glucosamine is a hexosamine sugar and a basic building block for the biosynthesis of the glycosaminoglycans and proteoglycans that are important constituents of the articular cartilage. In contrast, chondroitin is a glycosaminoglycan that is found in the proteoglycans of articular cartilage. Both are derived from animal products. The mechanism of action of glucosamine and chondroitin in the treatment of OA is not known. Both compounds have anti-inflammatory activities,^{6 - 7} and both compounds can favorably affect cartilage metabolism in vitro.^{8 - 9} Antiarthritic effects have also been demonstrated for both glucosamine and chondroitin in animal models.^{10 - 11}

Preliminary evidence has demonstrated that both glucosamine and chondroitin may favorably modify the radiological progression of OA. Reginster et al¹² reported a double-blind, placebo-controlled, randomized trial showing that glucosamine significantly reduced progression of knee OA over a 3-year period. Patients treated with placebo had an average joint space narrowing of approximately 0.1 mm per year, whereas no joint space narrowing occurred in the glucosamine group. Chondroitin may also stabilize radiological progression of OA of the hand and knee.^{13 - 14}

In this issue of THE JOURNAL, McAlindon and colleagues¹⁵ report a systematic review and meta-analysis of randomized controlled trials (RCTs) evaluating the efficacy of glucosamine and chondroitin in the symptomatic management of OA of the knee, hip, or both. This is a rigorously conducted and methodologically sound review with results that are likely to be valid and unbiased. The review satisfies the quality assessment criteria proposed by Oxman and colleagues.¹⁶ For example, the overview addressed a focused and explicitly stated clinical question. The criteria used to select articles for inclusion were appropriate and consistent with the objectives of the review. It is unlikely that important relevant studies were missed. The validity of the included RCTs were appraised. Assessments of the studies were reproducible with an acceptable degree of interobserver reliability. The statistical methods used in the meta-analysis were appropriate.

The main finding of McAlindon et al is that both glucosamine and chondroitin are likely to be effective therapies for the symptomatic management of OA. However, the actual degree of symptomatic benefit may be less than that predicted by the RCTs because of methodological flaws (inadequate allocation concealment and absence of intention-to-treat analyses) and probable publication bias. Randomized clinical trials with inadequate allocation concealment and absence of intention-to-treat analyses have been associated with exaggerated estimates of treatment benefit.^{17 - 20} Additional evidence of a probable exaggeration of treatment benefit stems from the finding of an inverse relationship between trial effect size vs trial size and quality.

Publication bias (ie, the greater likelihood that research with statistically significant results will be published compared with research with nonsignificant results)²¹ will tend to produce exaggerated treatment benefits in a meta-analysis if positive trials are preferentially being published. McAlindon and colleagues present qualitative and quantitative evidence for the presence of publication bias on the basis of asymmetrical funnel plots. (Funnel plots display effect size vs number of patients enrolled in the studies. If a greater effect size is found in studies with fewer patients enrolled, it creates an asymmetrical funnel plot and suggests that smaller trials with negative results are not published because of publication bias.) In addition, the fact that only 1 of the 15 RCTs was deemed completely independent in terms of manufacturer support raises additional support for the possibility of publication bias. Rochon and colleagues²² have shown that RCTs that are supported by the pharmaceutical industry may be associated with publication bias, a biased interpretation of results, or both.

McAlindon et al did not address 2 areas that are important considerations when applying the results of any systematic review to individual patient care. First, a detailed description of the patients enrolled in the RCTs was not provided by the authors, making it difficult to generalize results from the meta-analysis to individual patients. For example, important demographic details such as mean age, sex distribution, radiological stage of OA, duration of OA, and concomitant use of analgesics and NSAIDs were not summarized. Second, the toxic effect profiles of glucosamine and chondroitin compared with placebo were not summarized by the authors, making it difficult to evaluate whether the risks are worth the benefits.

Although the glucosamine and chondroitin RCTs have specific methodological flaws, collectively they are probably no worse methodologically compared with RCTs in other disciplines²³ and compared with NSAID trials in OA and in rheumatoid arthritis.^{24 - 26} For example, the mean quality score of the glucosamine and chondroitin RCTs of 35.5% (ranging from 12.3%-55.4%) is comparable with the mean quality score of 38.5% for trials published in journals and 33.6% for trials published in journal supplements.²⁷ These considerations notwithstanding, a number of important questions about the efficacy and toxic effects of glucosamine and chondroitin in OA remain. First, since most RCTs have been of a relatively short duration, the long-term efficacy and toxic effects of glucosamine and chondroitin have not been established. Second, since the relative purity and content of glucosamine and chondroitin in different preparations made by different manufacturers may vary, the relative efficacy (and toxic effects) of the various preparations may also vary. Third, additional large well-designed RCTs are needed to determine whether glucosamine and chondroitin can modify the radiological progression of OA. The National Institutes of Health has recently sponsored the first US study evaluating glucosamine and chondroitin in patients with OA of the knee. Subject recruitment is expected to begin later this year.²⁷ The 16-week, parallel group, double-blind RCT includes 4 treatment arms, in which patients will ingest orally 1 of the following combinations: placebo, 500 mg of glucosamine sulfate 3 times per day, 400 mg of chondroitin sulfate 3 times per day, and a combination of glucosamine and chondroitin.

As with many nutraceuticals that currently are widely touted as beneficial for common but difficult-to-treat disorders, the promotional enthusiasm often far surpasses the scientific evidence supporting clinical use. Until high-quality studies, such as the National Institutes of Health study, are completed, work such as that by McAlindon and colleagues is the best hope for providing physicians with information necessary to advise their patients about the risks and benefits of these therapies.