This is a multi-site randomized double-blind sham-controlled trial (N=392; 1:1). The trial protocol [in press] adhered to CONSORT guidelines (Supplementary material 1). It was conducted in the outpatient clinics in six hospitals (Shuguang Hospital, Renji Hospital, Shanghai East Hospital, Shanghai Changning Tianshan Traditional Chinese Medicine Hospital, Shanghai Tongren Hospital, and Shanghai Hudong hospital) in Shanghai, China. The institutional review board at each site approved the trial protocol. We established an international data and safety monitoring board (DSMB) to monitor data safety to ensure the quality of the trial and safety of patients in the trial.
A total of 603 patients were screened between January 2015 and November 2017 primarily through print advertisements on local newspapers and posters distributed in nearby communities (Figure 1). Participants were included if they were 50 years old or older, reported moderate or greater clinically significant knee pain on most days during the past month, had knee pain of at least 40/100 mm on a visual analogue scale (VAS), and had been diagnosed with idiopathic knee OA according to the American Rheumatism Association classification criteria.25 Kellgren-Lawrence grade ≥ 1 in the tibiofemoral joint on radiograph was also an inclusion requirement.5
Patients with other diseases affecting the knee, such as rheumatoid arthritis, fibromyalgia syndrome, chronic fatigue syndrome, and ankylosing spondylitis, were excluded. Other exclusion criteria were as follows: steroid medication or acupuncture/moxibustion treatment in the previous 3 months; intra-articular hyaluronate injection during the past 6 months; arthrocentesis or arthroscopy in the past 1 year; previous history of knee/hip replacement surgery and plan to have such surgery during the trial; use of other external treatments, such as topical medication; presence of serious medical conditions including cardiac diseases, pulmonary diseases, kidney diseases, liver diseases or malignant tumors, systemic infection or contagious diseases, and psychopathy; use of trial drug in the past 30 days; previous participation in other laser therapies; recruited in other clinical trial simultaneously; and unable to fill measurement questionnaires.
Randomization and Blinding
The 392 eligible participants were randomly assigned to receive either active LM or sham control. Randomization sequence with random blocks was generated using computer software. Allocation concealment was ensured with disguised letter codes of the LM devices (either active or sham devices) that were generated and sent to the site coordinators via a central randomization system. After receiving the device code from the site coordinator, the device operator used the LM device labeled with that code for patient treatment. The operators were unaware of the active or sham device as both produced the same red light. The whole procedure was supervised by the coordinators to ensure that the protocol was followed. Participants in the two groups were treated by trained operators. Communication among participants was discouraged and avoided as they were treated in separate rooms. Therefore, all involved personnel including participants, device operators, outcome assessors, research coordinators, and statistician were blinded to the treatment allocation.
The LM devices (SX10-C1) were manufactured by Shanghai Wonderful Opto-Electrics Tech. Co., Ltd. (Shanghai, China) and licensed by Shanghai Municipal Food and Drug Administration, China (20162210783). The wavelength of laser irradiation was 10.6 μm, and the output power was adjusted in the range of 160-180 mW. Energy density ranged from 61.2 to 68.8 J/cm2 for one treatment. After the patient laid supine on a treatment table, the laser irradiation tips of the two LM devices were aimed to the surface of the acupuncture points. The distance from the tips to the skin surface was 2 cm measured using a scale. Two acupuncture points were selected, namely, ST35 (Dubi; located in the depression on the lateral side of the patella and the patellar ligament) and Ashi point (tender point),26 at the affected knee. The selection of acupuncture points was based on the TCM theory used for Bi syndrome at the knee joints, and was successfully used in our previous studies.22,23 The treatments lasted 20 minutes and were performed 3 times a week for 4 weeks with a total of 12 sessions.
The sham treatment procedure was the same as the active treatment except no laser output irradiated from the device. However, in both active and sham devices, a red light-emitting diode with an output of 3 mW was used as visible indicator light on the skin to confirm accuracy of irradiation on the targeting acupoint. Participants were allowed to receive their usual care medications but were encouraged not to change to new drugs. In case of drug change, the name and dosage of the medication were documented.
The patients were assessed at baseline and at weeks 2, 4, 8, 12, and 24. All assessment instruments were in Chinese language version and previously validated.27, 28 The primary outcomes were the change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) 29 pain scores from baseline to 4 weeks. The index of WOMAC consists of 3 subscales (5 questions on pain, 2 questions on stiffness, and17 questions on functional status). The total score ranges from 0 (best) to 98 (worst). For bilaterally eligible knees, only the most symptomatic knee was evaluated.
Secondary outcomes included the change in WOMAC scores at weeks 2, 4, 8, 12, and 24; health-related quality of life (Short Form 36 Health Survey [SF-36]30); VAS; and use of medication and patients’ global assessment 31-33 (1=very good; 2=good; 3=fair; 4=poor). Adverse events, whether related to treatment or not, reported by the participants and practitioners were documented at each visit. We also communicated each participant weekly through telephone to follow up any adverse event or side effect. Possible side effects of LM include skin rash, redness, and blisters. To assess the masking effectiveness of the trial, the treatment providers and the participants were asked to guess their group assignment after the end of treatment at week 4.
Blood samples were collected at baseline and week 4 from the first one third of the participants (n=113, 56 from the LM group and 57 from the sham control group) to examine the changes in serum biochemical levels such as cartilage oligomeric matrix protein (COMP), interleukin (IL)-1β, IL-2, IL-6, IL-8, monocyte chemoattractant protein 1 (MCP-1), matrix metallopeptidase (MMP)-3 and MMP-13 (considered as important involved in the progress of OA). Blood samples (10 ml) were drawn at 10-11:30 am from each participant and then were stored in a refrigerator at -80 °C for later analysis.
Sample Size and Statistical Analysis
A minimum of 36% improvement in WOMAC score was considered to be clinically meaningful.34 Based on previous small-scale preliminary studies, 22-24, 35,36 a sample size of 324 participants (162 for each group) would be sufficient to detect the difference of 36% between the two groups to achieve a 2-sided 5% significance level with at least 80% power.37 Considering possible dropout (i.e., 17% dropout) during the trial, a total of 392 patients were thus required.
The analysis plan was determined and approved by the independent DSMB committee before the study was conducted. The primary analysis was an intention-to-treat analysis to compare the 4-week improvement in WOMAC pain score between the treated and the control in all randomized patients whose 4-week improvement was available. A chi-square test was used for categorical data and 2-sample t-test or Mann-Whitney U test was used for continuous data, to evaluate statistically significant differences in the distribution of different variables at baseline according to whether the data are normally distributed. Two-sample t-test or Mann-Whitney U test was performed for the primary (WOMAC pain) and secondary endpoints (WOMAC scores at other time points, SF-36, medication usage, and serum levels of different cytokines) at each time point. Chi-square test was performed for the categorical data (self-evaluation, credibility of the sham assessment, and safety assessment). For non-normally distributed variables, 95% bootstrap confidence instead of large sample normal based interval was calculated. All statistical analyses were conducted using SPSS (version 23.0; Chicago, USA). All reported P values were two-sided and used a significance level of 0.05.