Glucosamine/Chondroitin Arthritis Intervention Trial (GAIT)

This study is a 24 week, placebo-controlled, parallel, double-blind, five-arm trial. The randomized treatment assignment will be stratified by participating study center and baseline pain score (mild, moderate to severe). In order to establish that the study is able to detect significant treatment effects for glucosamine, chondroitin sulfate and the combination of the two, celecoxib, already established to be effective, will also be compared to placebo. Patients will receive a rescue analgesic medication, acetaminophen, which they will be allowed to take for severe pain throughout the trial but not within 24 hours prior to a study visit. The visits include a screening visit (Visit 1.0), randomization visit (Visit 2.0), and follow-up visits at 4 weeks (Visit 3.0), 8 weeks (Visit 4.0), and 16 weeks (Visit 5.0) and a completion visit at 24 weeks (Visit 6.0) after randomization or at any time patients discontinue the study drug or withdraw from the study. All patients require an X-ray to document the presence of their disease and all patients must have clinical and X-ray evidence of OA. X-rays are read at the individual sites and then forwarded to a central radiology center for confirmatory readings. Patient evaluations consist of; 1) the Western Ontario and MacMaster (WOMAC) osteoarthritis index, 2) patient and investigator global evaluations of disease status and response to study medication, 3) evaluation of the index knee for swelling and tenderness, 4) SF-36, 5) the Health Assessment Questionnaire (HAQ), 6) Beck Depression Inventory (BDI), 7) clinical evaluation for adverse reactions and 8) reconciliation of study medications and rescue analgesia use. The percentage of treatment responders is the primary outcome measure. A patient will be classified as a treatment responder if there is a 20% reduction from baseline to the end of follow-up in the WOMAC pain subscale. Following the method by which the sample size target was derived, the analysis to address the primary hypotheses will involve 3 primary comparisons 1) glucosamine vs. placebo, 2) chondroitin vs. placebo, and 3) glucosamine + chondroitin vs. placebo. A side comparison will also be done between celecoxib and placebo. Each comparison will be done using Fisher's Exact Test, with a two-sided alpha of 0.0125. GAIT Ancillary Study Patients recruited at selected sites will be extended the opportunity to participate in a trial that administers blinded study agents for a total of 24 months. This study is a preliminary evaluation of whether glucosamine, chondroitin sulfate and/or the combination of glucosamine and chondroitin sulfate are more effective than placebo and whether the combination is more effective than glucosamine or chondroitin sulfate alone in altering radiographic progression in patients with osteoarthritis (OA) of the knee after two years of follow-up. Expected Total Enrollment: 791 (\*Indicates centers participating in the ancillary study, sites listed below) Study start: April 2000; Study completion: March 2006 Primary Analysis The primary analysis will be based on all patients with baseline and follow-up xrays, including those who withdraw from treatment. Inclusion of all patients with baseline and follow-up data will reduce the degree to which differential effectiveness biases the treatment comparisons. A second analysis will be based on all patients who remain on treatment. Following the method by which the sample size target was derived, the analysis to address the primary hypotheses will involve 3 primary comparisons 1) glucosamine vs. placebo, 2) chondroitin vs. placebo, and 3) glucosamine + chondroitin vs. placebo. A side comparison will also be done between celecoxib and placebo. Each comparison will be done using the t-test for 2 independent samples, with a two-sided alpha of 0.0125. The primary outcome measure is the change in joint space width between baseline and two years of follow-up as defined by the Buckland-Wright MTP protocol. To evaluate long-term efficacy of the treatments, the two-year treatment response rate (defined as a 20% reduction from baseline in the WOMAC pain score) will be calculated. Paralleling the main study protocol, Fisher's exact test will be used to compare each active treatment arm to placebo. Mixed-model analysis of variance using generalized estimating equations will be used to compare the % of treatment responders over time across treatment groups. Treatment of missing data will follow that discussed previously for the mixed-model analysis of change in JSW. Safety will be evaluated by comparing the percentage of people withdrawing from study medications due to adverse events during the two-year follow-up period using Fisher's exact test. Time to withdrawal due to an adverse event will be evaluated using Kaplan-Meier life table estimates and comparison of treatment groups will use the log-rank test.