The study was reviewed and approved by the Ethics Committee of the Hospital on 18th June 2022. All procedures were in accordance with the ethical standards of the institutional and/or national research committee. All patients were given informed consent in writing prior to inclusion in the study.

This study was the randomized controlled clinical trials. Total of 60 KOA patients were enrolled in the study. The subjects were divided into three groups based on random numbers assigned by computer. The SPSS Visual Binning function was used to helps us quickly group. Participants were randomly assigned to intra-articular prolotherapy group (IG), Peri-articular perineural injection group (PG) or combined treatment group (I + PG). The physicians of conducting injection and evaluation were blind to the subject groups. (Fig. 1).

The flowchart of this study

The inclusion criteria were as follow: (1) patient’s age was of 48–80 years old; (2) the diagnose of KOA; (3) the grade 2 and 3 of the Kellgern–Lawrence grading scale; (4) the pain, crepitation, and knee joint stiffness continuing for three months at least.

The exclusion criteria were as follow: (1) any infection involving the knee skin; (2) history of knee tumors or tuberculosis; (3) any intra- or peri-articular injection during the three last months; (4) prior total knee arthroplasty; (5) history of acute lumbosacral radiculopathy or peripheral neuropathy, bleeding disorders and pregnancy.

All injections were performed by using ultrasound guided (4–18 Hz linearprobe, Mindray, Crius MEBP). Injections were repeated at 2 and 4 weeks after the first injection. All analgesics were discontinued 2 days before and 2 weeks after injection. All injections were conducted by the same physician.

The patient was placed in a supine position with knee flexion of 20°–30°, and the injection site and surrounding skin were disinfected. A high-frequency ultrasound probe under musculoskeletal conditions was placed near the lower femur and above the patella to find the suprapatellar bursa (Fig. 2). The intra-articular prolotherapy injection group (IG) participant was injected (25-gauge needle) 8 mL of the 20% dextrose with assistance of ultrasound guidance.

The location of A suprapatellar bursa and 4 deep nerve branches including B superolateral genicular nerve (SLGN), C superomedial genicular nerve (SMGN), infrapatellar branch of saphenous nerve (IPBSN), and E saphenous nerve

After referring the former studies, the peri-articular perineural injection group (PG) participant was located 4 deep nerve branches including saphenous nerve, infrapatellar branch of saphenous nerve (IPBSN), superomedial genicular nerve (SMGN) and superolateral genicular nerve (SLGN) by using ultrasound guidance [19, 20]. Total of 2 mL of 5% glucose was injected (25-gauge needle) around each nerve under the real-time guidance of ultrasound images (Fig. 2).

Intra-articular prolotherapy injection combined with Peri-articular perineural injection.

The intra-articular prolotherapy injection combined with Peri-articular perineural injection (I + PG) participant was treated with the combination of IG and PG.

Baseline demographic criteria included age, BMI, sex and duration of current condition. Pain intensity was measured with Visual Analogue Scale (VAS) [21]. The Western Ontario and McMaster Universities arthritis index (WOMAC) was used to evaluate the patient’s knee function [22]. The pressure pain threshold (PPT) is the most commonly method to evaluate the degree of nerve sensitization [23]. The PPT decrease indicates peripheral sensitization, and the higher the value, the lighter the degree of sensitization. We selected 4 sites at the diseased knee joint 3 cm from the center of the patella by using a handheld pressure pain meter for PPT measurement. Each site was measured three times to obtain the average value at least 1 min apart.

As to the pain and disability measurement after the injection, all patients were scheduled for follow-up visits at 2, 4 and 8 weeks after treatment. The pain and disability measurement were conducted by different colleague which was blind to the groups. Also, patients were asked to mention adverse reaction.

Statistical analysis was performed using the SPSS for Windows version 25.0 software. Continuous variables were tested by K–S test. The t test was used for measurement data subject to normal distribution among groups, and Fisher exact probability method was used for counting data. The VAS, WOMAC index, and tenderness threshold were compared using repeated measurement analysis of variance. p < 0.05 represents a statistically significant difference.