Included will be patients with KOA who meet the definition of osteoarthritis as per the 1991 American College of Rheumatology (ACR) criteria . Patients will be recruited through advertisement in the orthopaedic clinic of the Shanghai Guanghua Integrated Chinese and Western Medicine Hospital, Shanghai, China. We followed the SPIRIT 2013 Statement  to guide the development and reporting of our trial protocol. This study has been approved by the ethics committee of Guanghua Hospital of Integrated Traditional Chinese Medicine and Western Medicine (Ethics Approval Number: 2020-K-109) and registered in the Chinese Clinical Trial Registry. All participants will be asked to provide written informed consent and will be informed that the trial will not involve the collection of biological specimens for storage.
The inclusion criteria for enrollment will be as follows: (1) 50–65 years of age and right-handness; (2) KOA-associated pain for a duration ≥ 6 months; (3) no other treatment pursued in the 1-month prior; (4) a score on the knee pain Numerical Rating Scale (0–10 NRS) ≥ 3 in the 1-week prior; (5) a Kellgren-Lawrence radiographic score I or II; (6) patient agreement to not pursue other therapies during the treatment period of the RCT; and (7) provision of written informed consent.
The exclusion criteria will be as follows: (1) skin damage around the knee joint; (2) prior history of knee surgery or severe knee joint trauma; (3) secondary KOA, as a complication of rheumatoid, gouty, or infectious arthritis, or other joint disease; (4) presence of diabetes, systemic infection, severe abnormality in liver and kidney function, and malignancy; (6) allergy to celecoxib, which will be used as the ‘usual care’ control; (7) pregnant or lactating women; (8) presence of a mental disorder or cognitive impairment; (9) presence of active gastrointestinal disorder, including ulcers, bleeding or recurrent ulcers, and bleeding; and (10) contraindications to magnetic resonance imaging, required for rsFC, including phobias, pacemakers, defibrillators, cardiac stents, and intrauterine devices.
Exit criteria and management
Exit criteria will be as follows: (1) requested by the participant; (2) severe postoperative complications (e.g., pulmonary embolism); and (3) intra-treatment side effects.
The focus of neuroimaging research is on determining neural mechanisms of a therapeutic effect. As such, sample size calculation for RCTs in which neuroimaging provides the primary outcome is different from that of classical RCTs. For neuroimaging studies, 12 to 15 participants per group is generally sufficient to provide statistically significant results [17, 18]. In our trial, we will include 20 patients per group, namely the Tuina and the celecoxib control group. Considering an attrition rate of 20% and the possibility of unquantifiable rsFC due to head movement, our final sample size will be set to 24 participants per group.
Recruitment strategies and enrollment
Participant registration will be conducted between June 2022 and June 2024. Written informed consent will be obtained from all participants and will include consent for the use their data in scientific publications. Figure 1 presents the trial flow, which includes participant recruitment, eligibility screening, randomization, intervention, and outcome assessments. Figure 2 presents an overview of the trial design, conduct, review, and analysis. A completed SPIRIT 2013 checklist (Word) is included (Additional File 1).
Randomisation and blinding
An independent research staff will generate a randomly-numbered sequence, using SPSS21.0 software (SPSS Inc., Chicago, IL, USA), for complete randomization of enrolled participants to the Tuina or celecoxib treatment groups,. The number sequence will be sealed by an independent assistant in an opaque envelope containing treatment information. Eligible and consenting patients with KOA will be randomly assigned to the Tuina group and celecoxib groups using a 1: 1 allocation ratio, with 24 patients per group.
Only the therapist in charge of treatment is authorized to open the envelope and enrol participants. All outcome assessors and data statisticians will be blinded to group allocation.
A panel of two therapists with an academic background in Tuina and complementary therapy treatments and hospital health care professionals agreed to develop a standard protocol for the Tuina treatment. The feasibility of the protocol was tested to determine whether it was realistic and workable and to identify practical issues and any adverse effects caused by the procedures. No changes were needed and no adverse events were identified. Participants in the pilot study were not included in the actual trial.
Clinical medicine professionals who have ≥ 3 years of working experience and expertise in Tuina were recruited. Training was provided to ensure that the agreed standard procedures are followed. Clinicians will be requested to strictly adhere to the treatment protocol, following the exact steps stated in the manual of standard procedure to minimize differences in the components of treatment provided. Participants in the Tuina group will receive a 20-min massage session, three times a week (ideally every other day), for 6 consecutive weeks. Participants in the celecoxib group will be treated using celecoxib capsules, using a daily dosage of 200 mg orally, for 6 consecutive weeks.
Tuina is based on the meridian system theory, which considers that there are many fixed channels of flowing Qi energy interlaced as a network throughout the body . There are mainly three kinds of Tuina techniques for the treatment of KOA: the first is soft tissue massage to stimulate acupoints in the meridian system; the second is joint mobilization; and the third is to guide patients to practice Qi Gong energy exercises . For this study we chose to stimulate acupoints in the meridian system around the knee and knee mobilization in alignment with the most common guiding ideology of Tuina to ‘emphasize both bones and tendons’ for the treatment of musculoskeletal diseases. Pressure will be applied to five acupoints which are common local points used to treat knee problems and reduce knee pain [20, 21]. Based on the World Health Organization standard acupuncture point location , the following acupoints were selected on the affected side: XUEHAI (SP10), LIANGQIU (ST37), YANGLINGQUAN (GB34), YINLINGQUAN (SP9) and XIYAN (EX-LE5). Knee joint mobilization will be performed by pressing on the inferior pole of patella and asking patients to slowly stand from sitting and return to sitting, with 3 repetitions completed.
Participants in the celecoxib group orally taken a celecoxib capsules (approval number J20140072, manufactured by Pfizer Pharmaceutical Co. Ltd.) at a daily dosage of 200 mg for 6 weeks.
MRI data acquisition
MRI will be performed at the MRI Center at Guanghua Hospital of Integrated Traditional Chinese Medicine and Western Medicine. Resting-state functional magnetic resonance imaging (fMRI) will be used to quantify rsFC at baseline, before treatment, and at the end of the 6 weeks of treatment. MRI data will be acquired using a 3.0-T magnetic resonance scanner (General Electric, Wauwatosa, WI, USA), with a 32-channel phase-array head coil, Participants will be asked to remain awake and keep motionless, with eyes closed, during the full scanning period.
Blood-oxygen-level-independent (BOLD) resting-state functional images will be acquired using the following parameters: TR = 2000 ms, TE = 30 ms, flip angle = 90°, 33 axial slices, and field of view (FOV) = 220 mm×220 mm. T1-weighted images will be collected using the following parameters: TR = 1900 ms, TE = 2.93 ms, flip angle = 9°, 160 axial slices, field of view (FOV) = 256 mm×256 mm. T2-weighted images will be collected with the following parameters: TR = 6300ms, TE = 86.0 ms, flip angle = 150°, 25 axial slices, field of view (FOV) = 240 mm×240 mm.
To ensure compliance with treatment, participants will be required to register for treatment.
The clinical outcomes will including pain sensation, pain emotion, and pain cognition.
All evaluations will be performed twice, at baseline and after the 6-week intervention. All evaluations will be evaluated by two licensed physicians who have received training in the use of the outcome measures selected. These are the pressure pain threshold, the Numerical Rating Scale, the Pain Catastrophizing Scale, the Hamilton Anxiety Scale score, and the Hamilton Depression Scale.
Demographic / medical variables
The following demographic and medical variables will be collected for analysis: sex, age, marital status, occupation, ethnicity, education level, blood pressure, temperature, respiration, pulse, height, weight, body mass index, the combination of disease and medication, the Kellgren-Lawrence classification, the KOA course, and history of major surgeries.
Adverse events of treatment
The safety of patients to participate will be evaluated before and after enrollment and allocation to group, and will be based on liver and renal function. The following events will be treated immediately by the researchers until resolved: syncope, ecchymosis, pain caused by massage techniques, and symptoms, such as nausea, abdominal pain, indigestion, or increased pain and swelling with the use of celecoxib. All adverse events that occur during the trial will be recorded, including the time of occurrence, symptoms of discomfort, specific signs, severity, specific treatment provided, time course of improvement, time to resolution, and date of termination of the treatment and trial.
Data management and monitoring
Clinical data will be carefully saved using printed and electronic case report forms (CRFs). To guarantee data quality, only outcome assessors will access the CRFs for data entry. CRFs will be verified for double-entry and accuracy. During the trial, the Guanghua Hospital of Integrated Traditional Chinese Medicine and Western Medicine will be responsible for making regular visits (once a week) to review the trial conduct. The ethics committee will monitor for protocol violations on a weekly basis and ensure that there are no conflict of interest with the sponsors or researchers. Only statisticians will access the final trial dataset, which will only contain coded data. The safety, progress, study integrity, and design aspects will be monitored at several meetings of the research team.
Before analysis, researchers will provide a statistical scheme to the statisticians. The scheme will include the required data and processing methods. The data will be processed and analyzed by statisticians in accordance with the agreed-upon scheme.
Behavioral data analysis
Measurement data: A Shapiro-Wilk test will first be used to verify the normality of distribution of continuous variables. For data with normal distribution, independent sample t-tests will be used to compare baseline characteristics among the two groups. For non-normally distributed data and categorical variables, a Mann-Whitney U test will be used to compare between-group differences.
Count and grade data: Count data will include the ratio of men and women (sex variable) and the knee joint constituent ratio of the affected side and will be compared between the two groups using a Fourfold Table Chi-Squared Test or row * list Fisher's exact probability Test. The body mass index (BMI) classification, Kellgren-Lawrence classification, and NRSR score are grade data, with the Mann-Whitney U test used for between-group comparison and the Wilcoxon symbolic test for intra-group comparison. Statistical significance will be set at a p-value < 0.05 (two-sided) for all tests.
Seed-to-voxel resting state functional connectivity analyses
Seed-to-voxel functional connectivity analyses will be calculated using the MATLAB R2013b platform with the CONN v17.C software functional connectivity toolbox (http://www.nitrc.org/projects/ conn). Pre-processing will be as follows: (1) removing data from the first 10 time points to reduce machine instability and the impact of the environment; (2) spatial calibration to estimate and correct head motion; (3) time alignment to unify collection time at all levels; (4) detection of time points with large head movements as covariates of subsequent regression models to remove their effects; (5) registration of the T1 structural image to the functional space and segmented to obtain gray matter, white matter, cerebrospinal fluid and related information matrix, and standardization of the functional image to Montreal Neurological Institute (MNI) space, according to the correlation matrix information; (6) Gaussian smoothing of the standardized functional image, using a 6 mm kernel; (7) removal of the smoothed functional image, which includes the influence of 12 head movement parameters, time point signal of large head movement, white matter, cerebrospinal fluid, and other covariables; and (8) a final filtering, using a 0.01Hz ~ 0.1Hz bandwidth.
Seed-based functional connectivity analysis
Given that the thalamus and periaqueductal gray (PAG) play important roles in the ascending and descending pain pathways respectively, they will be defined as a priori regions of interest (ROIs), using Automated Anatomical Labeling (AAL), for the functional connectivity analyses of resting-state fMRI data. A threshold of voxel-wise p < 0.005, uncorrected, and cluster-level p < 0.05 false discovery rate (FDR), corrected at cluster, will be applied in the data analysis.