Radiofrequency (RF) treatment is an invasive and promising procedure in the treatment of osteoarthritis (OA). A meta-analysis based on 12 RCT studies was to investigate whether invasive RF treatment is more effective in relieving knee pain and improving knee function. Relevant studies were searched on database of Pubmed, Embase, EBSCO, Cochrane library, Wanfang digital database, VIP database, and CNKI up to January 2018. A total of 841 participants from 12 publications were included. The weighted mean difference (WMD) and the corresponding 95% CIs were used to evaluate the difference in pain scores and OKS/WOMAC scores between RF treatment and control groups. The statistical analysis was performed by Stata 12.0. The pain scores (VAS) in the RF group were lower than those in the conservative treatment group after 1 week (WMD -1.77, 95% CI -2.93 to -0.61, P<0.01), 1 month (WMD -1.40, 95% CI -1.98 to -0.82, P<0.01), and 3 months (WMD -1.32, 95% CI -2.27 to -0.37, P<0.01) of treatment, while there was no significant improvement in knee function. In subgroup analyses by site of radiofrequency, RF mode showed some discrepancies in the WMD of VAS between the treatment and control groups. In addition, subgroup analysis and meta-regression showed that the efficacy of RF treatment for reducing pain is reversely related to female ratio, and we did not find any surgery-related adverse reactions. RF treatment significantly reduces the knee pain, but rarely improves the knee joint function. Radiofrequency ablation has better efficacy than pulsed radiofrequency ablation in reducing pain. Furthermore, subgroup analysis and meta-regression suggested that women are more sensitive to RF treatment than men.
Osteoarthritis (OA) is a common chronic noninflammatory joint disease characterized by articular cartilage damage, which involves the subchondral bone, synovium, and periarticular tissues. The probability of suffering from symptomatic knee OA in elderly patients aged >65 years is 20–30% [
Invasive radiofrequency treatment can fill the gap between conservative treatment and TKA. It is a promising procedure with advantages of minimal invasion, rapid recovery, and few adverse events, including radiofrequency ablation (RFA) on the genicular nerve, intra-articular pulsed radiofrequency ablation (PRF), and cooled radiofrequency ablation (CRF). It can be applied to patients with severe complications, who cannot undergo TKA, those with severe pain around the knee joint area after TKA, those with severe joint pain who refuse to undergo surgery, and those who have failed oral NSAIDs or knee intra-articular injection.
The exploration of RFA on the articular nerve was first proposed by Choi et al. [
The electronic databases, Pubmed, Embase, EBSCO, and Cochrane Central of Controlled Trials (Central), as well as Wanfang digital database, VIP database, and CNKI (Chinese database), were searched to identify eligible studies since inception to January 2018. An in-depth literature search was performed using the following keywords: “knee pain” or “chronic pain” or “radiofrequency” or “genicular nerve” or “nerve ablation” or “intra-articular” or “thermal radiofrequency” or “knee osteoarthritis”. Each search was limited to human clinical trials in English and Chinese languages. Initially, the studies were screened according to the title, abstract, and keywords to exclude the irrelevant studies. Then, the full text of the enrolled and potentially eligible studies was carefully screened to further identify the eligible studies. In addition, the reference lists of the retrieved studies were manually checked to find out relevant studies.
(1) RCTs on clinical efficacy of invasive radiofrequency treatment for OA were performed; (2) patients in the experimental group received any of the following three modes of radiofrequency: CRF, RFA, and PRF; (3) a control group was involved; (4) the visual analogue scale/numerical rating scale (VAS/NRS) or Oxford Knee Score/the Western Ontario and McMaster (OKS/WOMAC) at different time points post-treatment were provided, with their mean and standard deviations (SD) or the standard error (SE) available to calculating the weighted mean differences (WMD) and the corresponding 95% confidence intervals (CIs).
(1) Patients who had underwent knee arthroplasty and arthroscopic surgery will be excluded, though they might also suffer from serious knee pain; (2) when there were duplicate studies, only the newly published or largest sample size ones were included; (3) studies with unextractable or insufficient data were also excluded; (4) non-RCT studies were excluded; (5) unpublished studies such as conference abstracts were excluded because of their poor quality and insufficient data.
Two experienced researchers used a predesigned form to independently extract data. Any uncertainty or discrepancy was resolved through discussion. The following parameters were included: name of the first author, year of publication, country of publication, location of study population, study design, Kellgren–Lawrence classification, mode of RF, operating site of RF, sample size of the experimental and control groups, age, gender, VAS/NRS, and OKS/WOMAC score at different time points before and after treatment. Since VAS and NRS are comparable to some extent, we converted NRS to VAS score in the subsequent data extraction for uniform statistical analysis.
Risk of bias for each study was independently assessed by two researchers (HT and GXL) using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions [
Risk of bias summary: review authors’ judgements about each risk of bias item for each included study.
Risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies.
The WMDs and the corresponding 95% CIs were used to evaluate the difference in pain scores and OKS/WOMAC scores between the RF treatment and control groups at different time points (1 week, 1 month, and 3 months after treatment). Heterogeneity between studies was evaluated using the Cochran Q test and
According to our searching strategy, 187 studies were retrieved. Finally, 12 RCTs fulfilled the inclusion criteria, including 9 studies from Asia, 2 from Europe, and 1 from the USA (Figure
Basic information of the 12 observational studies included in the current meta-analysis.
First author | country | location | Sample size | Treatment | Control | Kellgren-Lawrence grade | Treatment VAS baseline | Control VAS baseline |
---|---|---|---|---|---|---|---|---|
(year) | M/F | M/F | ||||||
Choi WJ (2011)[ | Korea | Asia | 35 | 2/15 | 3/15 | 2-4 | 7.82 (1.38) | 7.72 (0.75) |
Shen WS (2017)[ | China | Asia | 54 | 7/20 | 9/18 | NA | 7.12 (1.08) | 7.14 (1.03) |
Yuan Y (2016)[ | China | Asia | 42 | 7/15 | 7/13 | NA | 5.9 (1.1) | 5.6(1.4) |
Gulec E (2017)[ | Turkey | Europe | 100 | 17/33 | 20/30 | 2-3 | 7.96 (1.78) | 5.5 (1.48) |
Rahimzadeh P(2014)[ | Iran | Asia | 70 | 11/13 | 9/11 | 1-3 | 7.08 (1.41) | 6.65(0.98) |
Sari S (2016)[ | Turkey | Europe | 73 | 7/30 | 9/25 | 2-4 | NA | NA |
Davis T (2018)[ | US | America | 151 | 26/50 | 26/49 | 2-4 | 7.2 (1.2) | 6.9 (1.4) |
Shen GC (2017)[ | China | Asia | 60 | 6/24 | 5/25 | 3-4 | 5.51(1.32) | 5.79(0.88) |
Yang GQ (2015)[ | China | Asia | 62 | 14/17 | 12/19 | 3-4 | 7.25 (1.33) | 7.21 (1.58) |
Yi YS (2012)[ | China | Asia | 36 | NA | NA | NA | 7.26 (1.34) | 7.19 (1.57) |
Hu Y (2016)[ | China | Asia | 92 | 17/28 | 20/27 | 2-3 | 6.53 (1.1) | 6.38 (1.03) |
Yang XL (2013)[ | China | Asia | 66 | 8/23 | 9/26 | 2-4 | 6.9 (0.3) | 6.7 (0.5) |
Intervention procedure parameters, results, adverse effects, and limitation of the 12 observational studies included in the current meta-analysis.
First author | Treatment | Treatment | Intervention | Scoring | Follow-up | Results | Adverse effects | Limitation | |
---|---|---|---|---|---|---|---|---|---|
(year) | target | mode | parameters | methods | Pain | Knee function | |||
Choi WJ (2011)[ | GN | RFA | 70°C, 90s | NRS, OKS, GPE | 1, 3mo | Significantly improved | Significantly improved | None reported | follow-up period was brief; small sample size; lack of data about postprocedure analgesic use. |
Shen WS (2017)[ | IA | RFA | 70°C, 120s | VAS, AKSS, SF-36 | 3mo | Significantly improved | Significantly improved | NA | The technical details of the procedure were unclear; Small sample size; adverse effects were not recorded. |
Yuan Y (2016)[ | IA | PRF | 42°C, 6min | VAS, WOMAC | 1wk,1,2,3, 6mo | Significantly improved | Significantly improved | Knee joint dropsy were recorded in both groups within 2 weeks after the treatment. | Small sample size; No recording about analgesic consumption. |
Gulec E (2017)[ | IA | PRF | 42°C, 10min | VAS, | 1wk,1 and 3mo | Significantly improved | Significantly improved | NA | Short follow-up time; adverse effects were not recorded; No recording about analgesic consumption. |
WOMAC | |||||||||
Rahimzadeh P(2014)[ | IA | PRF | 42°C, 15min | VAS, ROM | 2, 4, 12 wk | Moderate improved | Moderate improved | None reported | Short follow-up time; Small sample size |
Sari S (2016)[ | GN | RFA | 80°C, 90s | VAS, WOMAC | 1, 3mo | Significantly improved | Significantly improved | NA | No recording about analgesic consumption; Short follow-up time |
Davis T (2018)[ | GN | CRF | 60°C, 150s | VAS, OKS | 1, 3, 6 mo | Significantly improved | Significantly improved | None reported | - |
Shen GC (2017)[ | IA | RFA | 70 to 90°C, 3min | VAS, | 1, 6mo | Significantly improved | Significantly improved | NA | Small sample size; No recording about analgesic consumption; adverse effects were not recorded. |
Lysholm | |||||||||
Yang GQ (2015)[ | GN | RFA | 70°C, 120s | VAS, OKS | 1, 6, and 12 wk | Significantly improved | Significantly improved | None reported | Small sample size; Short Follow-up time; No recording about analgesic consumption |
Yi YS (2012)[ | GN | RFA | 70°C, 120s | VAS, PGA | 1, 6, and 12 wk | Significantly improved | NA | None reported | Follow-up period was brief; No recording about analgesic consumption. |
Hu Y (2016)[ | IA | PRF | 42°C, 15min | VAS, | 1 and 6mo | Significantly improved | Significantly improved | NA | Follow-up period was brief; No recording about analgesic consumption. |
WOMAC | |||||||||
Yang XL (2013)[ | GN | RFA | 80°C, 180s | VAS | 3d, 1 and 2mo | Significantly improved | NA | None reported | follow-up period was brief; small sample size; Lack of scoring method of knee function |
Flow diagram showing the process for study selection.
David et al. [
Three studies by Choi et al. [
In Choi et al. [
Choi et al. [
We identified no other potential sources of bias.
Because the between-study heterogeneity was statically significant (1 week,
Forest plots for the associations between radiofrequency treatment and VAS scores (a) at 1 week, (b) 1 month, and (c) 3 months.
Similarly, due to the limited number of included studies and a relatively large between-study heterogeneity for OKS/WOMAC scores, the random-effects model was used to merge the effect sizes except that the fixed-effects model was used for OKS score at 1 week after treatment. The majority of the results showed that RF treatment did not improve the knee function significantly (Figures
As illustrated in Table
Subgroup analyses of the WMD of VAS scores between RF group and control group at different time points (one week, one month, and three months).
Subgroups | No. of studies | WMD and 95% CI | Z-value | | Heterogeneity | |
---|---|---|---|---|---|---|
| | |||||
| ||||||
Location | ||||||
Asia | 5 | -2.02 (-3.15, -0.89) | 3.49 | <0.01 | 92.2 | <0.01 |
Europe | 1 | -0.65 (-1.22,-0.08) | 2.25 | 0.03 | – | – |
Blindness | ||||||
Yes | 2 | -0.64 (-1.14,-0.14) | 2.52 | <0.01 | 0.0 | 0.95 |
No | 4 | -2.34 (-3.47,–1.22) | 4.09 | 0.01 | 91.3 | <0.01 |
RF mode | ||||||
Pulsed | 2 | -0.80 (-1.26,-0.34) | 3.42 | <0.01 | 0.0 | 0.36 |
Radiofrequency ablation | 4 | -2.25 (-3.43,-1.08) | 3.77 | <0.01 | 90.9 | <0.01 |
Concomitant therapy | ||||||
Yes | 3 | -2.77 (-3.71,-1.84) | 5.83 | <0.01 | 83.9 | <0.01 |
No | 3 | -0.77 (-1.19,-0.35) | 3.59 | <0.01 | 0.0 | 0.63 |
Site of radiofrequency | ||||||
Genicular nerve | 3 | -1.80 (-2.81, -0.78) | 3.47 | <0.01 | 74.8 | 0.02 |
Intra-articular | 3 | -1.77 (-3.82, 0.27) | 1.70 | 0.09 | 97.5 | <0.01 |
Sex ratio (female | ||||||
< 2 | 2 | -1.48 (-3.12, 0.17) | 1.76 | 0.08 | 92.5 | <0.01 |
≥ 2 | 4 | -1.92 (-3.42, -0.43) | 2.52 | 0.01 | 94.0 | <0.01 |
| ||||||
Location | ||||||
America | 1 | -0.90(-1.62, -0.18) | 2.46 | 0.01 | - | |
Asia | 8 | -1.42 (-2.10, -0. 73) | 4.06 | <0.01 | 89.8 | <0.01 |
Europe | 1 | -1.86 (-2.87,-0. 85) | 3.60 | <0.01 | - | – |
Blindness | ||||||
Yes | 4 | -1.43 (-3. 27, 0.42) | 1.51 | 0.13 | 93.6 | <0.01 |
No | 6 | -1.50 (-1.91,–1.08) | 7.02 | <0.01 | 67.2 | <0.01 |
RF mode | ||||||
Cooled | 1 | -0.90 (-1.62, -0.18) | 2.46 | 0.01 | - | - |
Pulsed | 5 | -0.97 (-1.86,-0.08) | 2.13 | 0.03 | 84.9 | <0.01 |
Radiofrequency ablation | 4 | -2.08 (-2.90,-1.25) | 4.74 | <0.01 | 80.1 | <0.01 |
Concomitant therapy | ||||||
Yes | 3 | -1.72 (-2.24,-1.21) | 6.56 | <0.01 | 48.3 | 0.15 |
No | 7 | -1.34 (-2.20,-0.47) | 3.04 | <0.01 | 88.0 | <0.01 |
Site of radiofrequency | ||||||
Genicular nerve | 5 | -1.83 (-2.57, -1.09) | 4.82 | <0.01 | 83.0 | <0.01 |
Intra-articular | 5 | -0.97 (-1.86, -0.08) | 2.13 | 0.03 | 84.9 | <0.01 |
Sex ratio (female | ||||||
< 2 | 6 | -0.93 (-1.63, -0.23) | 2.62 | 0.01 | 80.9 | <0.01 |
≥ 2 | 4 | -2.13 (-2.95, -1.31) | 5.07 | <0.01 | 78.0 | <0.01 |
| ||||||
Location | ||||||
America | 1 | -2.40 (-3.07, -1.73) | 7.02 | <0.01 | - | |
Asia | 6 | -1.17 (-2.39, 0.05) | 1.88 | 0.06 | 92.2 | <0.01 |
Europe | 1 | -1.28 (-2.23,-0. 33) | 2.65 | <0.01 | - | – |
Blindness | ||||||
Yes | 4 | -0.69 (-2.82,-1.44) | 0.63 | 0.53 | 94.2 | <0.01 |
No | 5 | -1.88 (-2.39,-1.37) | 7.17 | <0.01 | 53.0 | 0.08 |
RF mode | ||||||
Cooled | 1 | -2.40 (-3.07, -1.73) | 7.02 | <0.01 | - | - |
Pulsed | 4 | -0.00 (-1.42,-1.42) | 0.00 | 0.99 | 88.7 | <0.01 |
Radiofrequency ablation | 4 | -2.23 (-2.80,-1.67) | 7.77 | <0.01 | 39.1 | 0.18 |
Concomitant therapy | ||||||
Yes | 3 | -2.01 (-2. 45,-1.57) | 8.99 | <0.01 | 0.0 | 0.99 |
No | 6 | -0.98 (-2.45, 0.50) | 1.30 | 0.19 | 93.3 | <0.01 |
Site of radiofrequency | ||||||
Genicular nerve | 4 | -2.36 (-2.92, -1.80) | 8.24 | <0.01 | 35.9 | 0.20 |
Intra-articular | 5 | -0.43 (-1.84, 0.98) | 0.59 | 0.55 | 92.5 | <0.01 |
Sex ratio (female | ||||||
< 2 | 5 | -0.76 (-2.30, 0.78) | 0.96 | 0.34 | 93.9 | <0.01 |
≥ 2 | 4 | -2.00 (-2.97, -1.04) | 4.07 | <0.01 | 76.3 | <0.01 |
WMD: weighted mean difference; CI: confidence interval.
Subgroup analysis showed that RFA of the genicular nerve could significantly relieve pain from 1 week to 3 months after treatment (1 week, WMD -1.80, 95% CI -2.81 to -0.78, P<0.01; 1 mon, WMD -1.83, 95% CI -2.57 to -1.09, P<0.01; 3 months, WMD -2.36, 95% CI -2.92 to -1.80, P<0.01), while the intra-articular PRF was slow-acting (1 week, WMD -1.77, 95% CI -3.82 to 0.27, P=0.09), which could relieve the pain only at 1 month (1 month, WMD -0.97, 95% CI -1.86 to -0.08, P=0.03), but showed an insignificant improvement at 3 months (3 months, WMD -0.43, 95% CI -1.84 to 0.98, P=0.55).
The RF mode was also one of the reasons for the improvement in pain. CRF (no data in 1 week) and RFA treatment showed a significant improvement in pain at all observation time points. However, PRF treatment did not show any statistically significant improvement in the pain till the
Due to the significant heterogeneity among the study groups and limited findings from subgroup analyses, meta-regression analyses were conducted to further verify the source of between-study heterogeneity. At 1 and 3 months after treatment, the sex ratio was inversely correlated to the WMD of VAS. The slope of the meta-regression curve at 1 month was -0.67, 95% CI -1.08 to-0.27, P<0.01, which might explain 71.56% heterogeneity across studies, while that at 3 months was -0.73, 95% CI -1.48 to 0.03, P=0.06, which might account for about 36.88% heterogeneity across studies (Figures
Meta-regression reveals inverse correlations between sex ratio and the weighted mean difference (WMD) of VAS scores at (a) 1 month and (b) 3 months.
Sensitivity analysis was used to evaluate the stability robust of the pooled WMDs. After the sequential exclusion of individual studies one by one, the WMDs were recalculated to identify the significant change in our results. Sensitivity analysis the elimination of any single study was unlikely to overturn our findings (Figure
Evaluation using Begg’s correlation tests (1 week:
In the present study, 12 RCTs involving a total of 841 participants were enrolled to evaluate the efficiency of RF treatment on long-term analgesia and improvement in the knee function in OA. Our results showed that the pain scores in the treatment group were lower than those in the control group at 1 week, 1 month, and 3 months after RF, while there was no significant improvement in knee function. In the subgroup analysis, CRF (no data in 1 week) or RFA in the genicular nerves significantly improved the pain within 1 week, and the analgesic effect could maintain up to 3 months. Nevertheless, intra-articular PRF was slow-acting with a short duration, its pain relief effects appeared until 1 month, but vanished at 3 months. Thus, it is suggested that RF thermocoagulation of genicular nerves can improve the refractory pain of the knee joint and the quality of life of patients. In addition, the pain relief effects of RF treatment were inversely associated with sex ratio. The higher the female ratio was, the more effective the pain relief effects were. We did not find any surgery-related adverse reactions. Furthermore, no publication bias was found in the current study, and the sensitivity analysis suggested that the pooled effect sizes did not alter significantly after the exclusion of individual studies sequentially, which confirmed the stability of the results.
The mechanism of intra-articular PRF or RFA of genicular nerves has not yet been clarified; however, all published studies suggest that RF treatment can significantly reduce pain around the knee [
In this meta-analysis, among the 7 studies on RCT of RFA in the genicular nerves, only 2 studies had performed prognostic nerve blocks with local anesthetic, in which patients with a remission rate of > 50% were enrolled in the RFA group, and 12–25.9% showed ineffective treatment or did not respond to the treatment [
Based on the published studies, the three RF methods can improve the knee function [
No surgical-related adverse reactions were reported during RF treatment group. Some studies stated severe pain during surgery, which might be attributed to the electrode tip of the needle that was close to the periosteum or tendon and could be reduced by replacing the needle site [
Subgroup analysis showed the sex ratio indicators of better pain improvement. The meta-regression analysis further proved that the higher the female proportion, the better the treatment effect, although the underlying mechanism is yet to be elucidated. Women are prone to suffering from chronic pain syndrome, and the frequency, areas of the body, duration, and severity of pain are significantly higher than those in men [
For all the RCTs included in our studies published in the past 9 years, their limitations are summarized as follows. (1) Some of the researches were conducted based on relatively small sample size (35 to 42 patients) [
In the present study, we conduct a pooled analysis of improvement in the pain and knee joint function with RF treatment at different time point using meta-analysis method. The absence of published bias in the enrolled studies and sensitivity analysis indicated the robustness of the results. Nevertheless, the present study has certain limitations. Firstly, the WMDs of all combined trials showed a statistically significant reduction in VAS scores at some time points, but this difference was below the widely accepted, minimally clinically significant decrease in VAS of at least 2/10. Second, the blind method was adopted in only 3 studies, indicating a potential high risk of bias. This might be because blindness was not applicable in some studies. Third, Few or no data was related to cartilage mechanism or change in morphology. In addition, only one RCT of CRF was included in our analysis. The comparability of this treatment with other types of RF treatment needs further investigation.
In summary, our meta-analysis showed that, in comparison with the conservative treatment, invasive RF treatment significantly improved the knee joint pain, but exerted a limited effect in improving the knee joint function. In addition, CRF or RFA on genicular nerves exhibited a more significant and long-lasting analgesic effect as compared to intra-articular PRF. Furthermore, women are more likely to benefit from RFA treatment on the genicular nerves around the knee joint than men. Both subgroup analysis and meta-regression suggested a positive correlation between knee pain remission and female in the gender ratio, although the mechanism is not yet clarified. Due to the limitations of included literatures, the authenticity of the results requires further verification by prospective multicenter randomized controlled trials.
The authors declare no conflicts of interest regarding the publication of this paper.
Financial support was provided by Professor Peng Yao’s Funding (no. 20170541032), from the Natural Science Foundation of Liaoning Province.
In order to evaluate knee function improvement among the different RF treatment groups at different observation points, we try to calculate the WMD of WOMAC and OKS. The majority of the results showed that RF treatment did not improve the knee function significantly (see Figures S1 and S2 in supplementary material). Because of the limited studies enrolled in our study, we did sensitivity analysis by excluding studies one by one. The removal of one study each turn showed no significant variation in WMD, which confirmed the stability of our results (see Figure S3 in supplementary material).