We conducted a double-blind RCT to compare the efficacy of the combination of pCGS plus diacerein with pCGS plus placebo in the treatment of knee OA. Our findings suggest that combined treatment does not reduce VAS pain score, WOMAC total score, or WOMAC subscores compared with pCGS monotherapy in patients with mild to moderate knee OA. Although the efficacy in both treatment groups did not differ, clinical signs and symptoms were improved in both treatment groups at 12–24 weeks, and this was particularly evident in the pCGS plus placebo group.
pCGS, which comprises essential components of the proteoglycans in normal cartilage, is found naturally in the knee joint of the human body. With its possible anabolic effect, it is used for inhibition of metalloproteinase activity, prostaglandin E
2 release, nitric oxide production, degradation of glycosaminoglycans, and stimulation of the synthesis of hyaluronic acid in the joint [
12,
27]. It has a slow onset of response, provides long-lasting pain relief and functional improvement, and delays progression of the joint space [
28,
29] in OA of the knee [
12,
30,
31]. Diacerein may be beneficial for OA in that it inhibits interleukin-1, controls lymphocytes, increases the number of bone marrow cells, and reduces degeneration of the bone joint [
13]. As a result, diacerein is also claimed to improve pain and function in OA of the knee [
30,
32‐
34]. In a recent animal study, researchers found chondroprotective effects of diacerein and pCGS, but a better range of motion of the knee joint was found in response to diacerein than to pCGS [
35]. With the different mechanisms of action of pCGS and diacerein, it could be expected that combined treatments should result in synergetic effects.
Our findings are similar to those of a previous study in which researchers compared combined pCGS plus chondroitin sulfate with pCGS or chondroitin sulfate alone in patients with OA of the knee whose Kellgren-Lawrence grade was 2–3 [
31]. That study was later combined in a network meta-analysis [
36], which showed similar results. Although the potential synergistic effects derived from a pharmacologic study of pCGS and diacerein [
27] looked promising, our findings indicate that combined treatments did not provide any benefit over monotherapy.
Strengths and limitations
To the best of our knowledge, this is the first double-blind RCT designed to assess the effects of combined pCGS plus diacerein versus pCGS monotherapy with 6 months of follow-up. The active treatments, both capsules and sachets, were identical in appearance and were administered to patients using coded drug packs; therefore, patients, investigators, and outcome assessors were truly blinded. We considered the most relevant outcomes, including subjective (i.e., VAS pain score, WOMAC total score, and WOMAC subscores) and objective (i.e., minimal tibiofemoral JSW) measures. In addition, all possible adverse effects were collected. Drug compliance was reasonably high, ranging from 88 % to 100 % and 79 % to 100 % in the combined treatment and monotherapy groups, respectively (P = 0.133). Cointervention with additional pain medications was also similar at 17.6 % versus 21.6 % in the combined treatment and monotherapy groups, respectively (P = 0.534). We applied an ITT analysis by considering all patients in the groups to which they were originally randomly allocated, thus minimizing bias.
Our study has some limitations. The dosage of diacerein that we used was 50 mg in the combined treatment group because the side effect of diacerein has been shown to have a correlation to the drug dosage in prevention of drug withdrawal, according to a previous study in which researchers compared diacerein in different dosages (50, 100, and 150 mg) with placebo. The highest safety profile is at a dose of 50 mg/day, and we did not up-titrate the dose. However, the positive effects on VAS pain symptoms are decreased in a dose-dependent manner as the dose is decreased [
34]. This could explain the lack of difference between the two treatment groups.
The sample size calculation was computed to assess primary outcomes between groups, but it may not be generalized to assessment of secondary outcomes; therefore, statistical insignificance might be due to the risk of type II errors. We considered mostly patients with knee OA with mild to moderate pain scores at baseline, which might have made it difficult to detect the benefits of combined treatment. In addition, our patients with knee OA were mainly diagnosed. The uncertain clinical diagnosis and classification may affect the outcomes of clinical studies [
37‐
41]. There is a widespread belief that there is a high discordance between clinical and radiographic knee OA. In an attempt to overcome this problem, we included the participants in our study on the basis of American College of Rheumatology criteria [
20] for diagnosis of knee OA with the radiographic criteria of Kellgren-Lawrence grade 2–3 [
42]. However, the Kellgren-Lawrence classification strongly depends on adequate patient positioning when taking x-rays. It is also not specific to cartilage loss. Magnetic resonance imaging (MRI) provides somewhat superior sensitivity to change compared with commonly used radiographs, and it has recently provided non-location-dependent measures of cartilage thickness loss and gain, which are potentially more sensitive in detecting symptomatic slow-acting drug for osteoarthritis effects than radiographic JSW. The cost of MRI is about 20 times greater than that of radiographs; therefore, we used radiographically measured JSW with strict positioning to obtain the highest-quality outcome measurements. Moreover, knee pain and function of patients with these inclusion criteria, which we assessed using subjective outcome VAS and WOMAC scores, may be imprecise owing to the nonspecific nature of knee pain (e.g., non-OA pathology such as tendinitis or muscle strain, referred pain, and nonphysical pain such as depression), and all these factors can coexist at the same time, composing multiple layers of causality of knee pain [
37].
The potential sources of bias that could substantially impact interpretation of the trial were noncompliance, cointervention, and contamination. For cointervention and contamination, there were no reports of use of nonprotocol medications or other cotreatments, which meant cointervention or contamination between the groups was unlikely. However, the data were analyzed by the ITT method. As for compliance, missing outcome data were imputed using a multivariate normal regression analysis with 20 replications. Complete data (i.e., age, sex, BMI, VAS and WOMAC subdomain scores at baseline) were used to simultaneously predict VAS and WOMAC subdomain scores after treatment.
We measured outcomes over 6 months owing to time restrictions, making the study a short-term assessment. However, according to previous studies of both drugs, there have been long-term effects lasting up to about 1 year (longest effect lasted 3 years). There were also sustained effects lasting longer than 3 months; therefore, when assessing stratified patients with knee OA in a subgroup with longer follow-up times, a sustained effect could be considered to deduce the benefits of combined treatment. Replication using larger samples with repeated measurements might show greater, more conclusive differences in all possible outcomes over time between the two groups. This information could be used to more properly address the treatment effects. Moreover, the results derived from this study were based on plain radiographs, which may not be sufficient to assess knee cartilage, the major component that responds to both drugs (based on the JSW of the knee joint). MRI may better facilitate the assessment of cartilage changes in the knee joint in the next study because MRI is much more sensitive than plain radiography. However, in this study, we assessed JSW at 3 and 6 months. These time periods may not be adequate for assessing changes in JSW and may be a reason for the insignificant results regarding radiographic changes in JSW. In postmarketing data, elevations in serum liver enzymes and acute hepatic injury have been recorded. In this study, although participants with preexisting liver problems were excluded, there were no measures to monitor liver toxicity; therefore, we were unable to detect any potential liver-related adverse drug events. Finally, there are various generic preparations of glucosamine that were approximately five times cheaper than the original ones. Bioequivalence trials between original and generic glucosamine samples should be conducted.