Study selection
In total, we included 2073 studies among these databases (343 studies in Medline, 1,331 studies in Embase, 754 studies in Cochrane CENTRAL, and 45 studies in ClinicalTrails) after removing duplications (Figure 1). By screening title and abstract, we excluded 2,064 articles with exclusive criteria, and only 9 eligible literatures were moving to the next filtration step. After reading the full text of these studies, we excluded 5 conference abstracts and 1 research in Russian, then only 3 articles were included in our systematic review.
Study characteristics
Three RCTs were included in our systematic review. All of them were published in last ten years. The study characteristics of these three RCTs are listed in Table 1. These three RCTs were all used double-blind study design. Two RCTs compared the efficacy of independent drugs in OA pain management, and one RCT compared combined use of medicines.
Participant characteristics
The characteristics of participant enrolled in the three RCTs were listed in Table 2. Sofat and colleagues(15) recruited 89 patients aged 40-75 years with hand OA diagnosed with American College of Rheumatology (ACR) criteria(16). The participants had a NRS score over 5, and were on usual care including acetaminophen and/or NSAIDs. 22 patients were treated with pregabalin, 21 patients were treated with duloxetine, and 22 patients received placebo. The three groups have no statistic difference in age, female ratio, NRS score before treatment, and AUSCAN pain score. Ohtori and colleagues(17) recruited 65 patients with knee OA complaining knee pain at least one month and having X-ray of affected knee. All patients were assessed for the severity of knee OA with Kellgren-Lawrence (KL) grading system by X-ray imaging. 28 patients were treated with pregabalin, 31 patients were treated with meloxicam, and 30 patients were treated with pregabalin and meloxicam. There was no statistically significant difference in age, female ratio, knee OA severity, VAS score before treatment, and WOMAC pain score between the three groups. Enteshari-Moghaddam and colleagues(18) recruited 150 patients aged 45-75 years with knee OA classified KL 3 and 4 level. 50 patients were treated with gabapentin, 50 patients were treated with duloxetine, and 50 patients were in acetaminophen group. The participants characteristics were also similar among the three groups, including age, female ratio, knee OA severity, VAS score before treatment, and WOMAC pain score.
Treatment details
The treatment details of included three RCTs were presented in Table 3. In Sofat et al. study, participants took one capsule of placebo or pregabalin or duloxetine at night in week one orally, then two capsules from week two to week ten, finally one capsule from week eleven to week twelve(15). In Ohtori et al. study, participants in meloxicam group took meloxicam orally 30 mins after breakfast, participants in pregabalin group took pregabalin orally before sleep, and combined medicine group took meloxicam 30 mins after breakfast and pregabalin before sleep for four weeks(17). In Enteshari-Moghaddam et al. study, participants took gabapentin or duloxetine or placebo for 3 months, and double the dose from week 3(18).
Efficacy and safety of GABA derivates in OA pain management
All included studies reported OA pain relief with GABA derivates treatment (Table 4). For hand OA, pregabalin reduced NRS score from 6.1 (95% CI: 5.4 to 6.7) to 3.4 (95% CI: 2.4 to 4.4) with a mean difference -2.7 (95% CI: -3.5 to -1.9). The difference of mean difference of pregabalin effect on NRS score was statistically significant compared with placebo (-0.9 (95% CI: -0.2 to 0.2), P = 0.023). Meanwhile, pregabalin reduced AUSCAN pain score from 317.0 (95% CI: 280.8 to 353.1) to 176.5 (95% CI: 123.9 to 229.1) with a mean difference -132.1 (95% CI: -181.1 to -82.9). The difference of mean difference of pregabalin effect on AUSCAN pain score was statistically significant compared with placebo (-46.61 (95% CI: -93.9 to 0.75), P = 0.008) (15). For knee OA, pregabalin reduced VAS score from 5.0±2.0 (mean ± standard error of the mean (SEM)) to 2.0±2.2 (mean ± SEM) in four weeks. Meanwhile, pregabalin reduced WOMAC pain score from 12.2±3.0 (mean ± SEM) to 6.6±3.0 (mean ± SEM) in four weeks(17). For knee OA, gabapentin reduced both VAS and WOMAC pain score in involved participants. Compared with acetaminophen (-31.20±12.58 (mean ± standard deviation (SD))), the mean difference of gabapentin (-63.36±8.87 (mean ± SD)) was statistically significantly larger (P < 0.001). Meanwhile, compared with acetaminophen (-50.30±10.78 (mean ± SD)), the mean difference of gabapentin (-73.94±12.79 (mean ± SD)) was statistically significantly larger (P < 0.001)(18).
Number and subtypes of recorded treatment-related AEs owing to side effects of GABA derivates were summarized in Table 5. Sofat et al. reported 55 recorded treatment-related AEs in pregabalin group, mainly including nervous system disorder and mental disturbance (15). Ohtori et al. reported no treatment-related AE in their study(17). Enteshari-Moghaddam et al. reported 9 recorded treatment-related AEs in 9 patients (18%) in gabapentin group (18).
Quality of evidence
There lacks available peer-reviewed literature and rigorous study design on this topic. In that case, the risk of confounding is moderate to severe for included studies. Bias for selection, performance, detection, attrition, and reporting of the included RCTs are lists in Table 6.