Abstract| Volume 25, SUPPLEMENT 1, S408-S409, April 2017

No treatment effects of oral glucosamine for subgroups of knee and hip osteoarthritis patients; an individual patient data meta-analysis from the OA trial bank

      Purpose: To evaluate the effectiveness of oral glucosamine in clinical relevant subgroups of hip and knee osteoarthritis patients based on pain severity, BMI, sex, structural abnormalities and inflammation, using individual patient data from published trials.
      Methods: Pubmed, the Cochrane Central Register of Controlled Trials, Embase, Web of Science, Cinahl and Scopus were searched for published randomized controlled trials on the effectiveness of any oral glucosamine substance in patients with clinically or radiologically defined knee or hip OA. Additionally, and the ISRCTN registry were searched for ongoing studies. As a minimum, pain, age, sex and BMI needed to be assessed at baseline and adequate reporting of pain as an outcome measure needed to be available. All authors and institutions of all eligible studies were approached and asked to share the trial data. All anonymous trial data from trialists willing to share the data were transferred to a secured database of the OA Trial Bank, upon signing the license agreement. All available trials were assessed for their risk of bias, using the criteria recommended by the Cochrane. Missing data for covariates and outcome measures were imputed, using multiple imputation methods, within each original study. Subgroup factors were dichotomized, based on consensus of the OA Trial Bank Steering Committee. A multilevel regression analysis was performed to estimate the magnitude of the effect of glucosamine over the control intervention in the different subgroups with the individuals nested within each study. Pain at short-term (3 months) and long-term (24 months) served as primary outcomes measure, function as secondary. Outcome measures measured on different scales were standardized to a 0-100 scale to pool the data. All analyses were performed with and without stratifying for affected joint and type of glucosamine substance.
      Results: Out of 21 eligible studies, six studies (N = 1663) shared their trial data with the OA Trial Bank (see figure). Only one of these studies (N = 40) was industry driven. All six studies had a low risk of bias, of which five trials (all not industry driven) compared glucosamine to placebo. These five studies represented 50% of the total number of randomized subjects in all published trials for this comparison. No main effects of glucosamine were found on pain or function at short-term (mean difference 0.31 95%CI[-2.02 to 2.64] and 1.56 [-0.56 to 3.69], respectively) and at long-term follow-up (0.98 [-1.76 to 3.73] and 1.40 [-1.27 to 4.06], respectively). Also, no significant interactions with treatment were found for subgroups based on pain severity (WOMAC pain < 70 vs. ≥70), BMI ( < 27 kg/m2 vs. ≥27 kg/m2), sex (male vs. female) and structural abnormalities (KL-grade 0-2 vs. 3-4), see Table. Stratification for knee OA patients only and for type of glucosamine did not result in any differences in the outcomes. No data was available to adequately form subgroups based on degree of inflammation.
      Conclusions: The majority of industry-led glucosamine studies for osteoarthritis did not wish to share data, challenging optimal use of available data. There is currently no evidence for the use of glucosamine for the treatment of hip or knee osteoarthritis and an absence of support for clinically relevant subgroups of OA patients according to baseline pain severity, BMI, sex, and structural abnormalities.
      Tabled 1Estimated pooled differences between glucosamine and placebo on a 0-100 scale and p-values for treatment-subgroup interactions.
      All studies (N = 1394)Knee OA only (N = 1172)GH in knee OA (N = 671)GS in knee and hip OA (N = 723)GS in knee OA (N = 501)
      Pain at short-term*Estimated pooled differences and 95% confidence interval
      Glucosamine vs0.310.531.97−2.62−5.99
      placebo(−2.02 – 2.64)(−2.21 – 3.26)(−1.14 – 5.09)(−6.03 – 0.79)(−11.50 – 0.48)
      p-values for treatment-subgroup interactions
      Pain subgroup0.800.920.940.590.81
      BMI subgroup0.360.660.830.300.43
      Sex subgroup0.800.700.800.670.62
      KL subgroup-----
      Pain at long-term*Estimated pooled differences and 95% confidence interval
      Glucosamine vs placebo0.98 (−1.76 - 3.73)0.19 (−2.83 – 3.22)0.78 (−4.33 – 5.89)1.22 (−1.90 –4.33)−0.38 (−3.67 – 2.90)
      p-values for treatment-subgroup interactions
      Pain subgroup0.260.280.420.440.86
      BMI subgroup0.550.100.510.720.10
      Sex subgroup0.460.530.750.520.77
      KL subgroup-0.40---
      Function at short-term*Estimated pooled differences and 95% confidence interval
      Glucosamine vs1.561.572.130.74−1.38
      p-values for treatment-subgroup interactions
      Pain subgroup0.460.320.380.970.90
      BMI subgroup0.920.871.000.830.45
      Sex subgroup0.550.370.230.490.79
      KL subgroup-----
      Function at long-term*Estimated pooled differences and 95% confidence interval
      Glucosamine vs placebo1.40 (−1.27 – 4.06)0.63 (−2.31–3.58)0.85 (−4.43–6.13)2.02 (−0.82–4.86)0.62 (−2.29–3.52)
      p-values for treatment-subgroup interactions
      Pain subgroup0.490.380.550.940.91
      BMI subgroup0.820.420.650.560.68
      Sex subgroup0.720.610.801.000.94
      KL subgroup-0.77---
      *measured using WOMAC pain (0-100) and adjusted for age sex, BMI, WOMAC pain at baseline and study number, **measured using WOMAC function (0-100) and adjusted for age sex, BMI, WOMAC function at baseline and study number. Positive estimated pooled differences indicate a greater reduction in the outcome in the glucosamine group compared to the placebo group. GS = glucosamine sulphate; GH = glucosamine hydrochloride.