Long-term efficacy of genicular nerve radiofrequency ablation in patients with advanced knee osteoarthritis: A retrospective cohort study Authors: Salim Taner Gozukizil, Halil Ibrahim Altun Published: 2026 ABSTRACT OBJECTIVEConventional treatments for advanced knee osteoarthritis (OA) are often insufficient, and surgery is not always feasible. This retrospective cohort study aimed to evaluate changes in pain, function, and safety over 12 months following ultrasound-guided genicular nerve radiofrequency ablation (GNRFA).METHODSWe included 44 patients with chronic knee pain unresponsive to conservative treatments for at least 3 months. All patients underwent US-guided radiofrequency ablation of the superomedial, superolateral, and inferomedial genicular nerves. Pain numeric rating scale (NRS) and functional status Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores were assessed at baseline, and routinely at 1, 6, and 12 months post-procedure.RESULTSCompared to baseline, mean NRS and WOMAC scores decreased significantly at all follow-up points (p<0.001). Based on a ≥50% pain reduction threshold (clinical success), 81.8% of patients achieved success at 1 month, and 75.0% maintained it at 6 months. By 12 months, this rate declined to 11.4%. No statistically significant differences in outcomes were observed between radiological grades. A strong positive correlation between NRS and WOMAC scores was observed across all follow-up times. No serious adverse events were reported.CONCLUSIONUS-guided GNRFA was associated with improvements in pain and function, which were most pronounced within the first 6 months in patients with advanced knee OA. However, the marked decline in responder rates at 12 months suggests reduced durability beyond 6 months. These findings suggest that GNRFA may serve as a medium-term management option rather than a permanent solution, and that re-evaluation may be appropriate in patients with recurrent symptoms. Highlight key points Ultrasound-guided GNRFA was associated with reduced pain and improved function in advanced knee OA.Clinical success was observed in 81.8% of patients at 1 month and 75.0% at 6 months.The responder rate declined to 11.4% at 12 months, limiting long-term clinical relevance.GNRFA may serve as a medium-term management option, and re-evaluation may be appropriate for recurrent symptoms. Ultrasound-guided GNRFA was associated with reduced pain and improved function in advanced knee OA. Ultrasound-guided GNRFA was associated with reduced pain and improved function in advanced knee OA. Clinical success was observed in 81.8% of patients at 1 month and 75.0% at 6 months. Clinical success was observed in 81.8% of patients at 1 month and 75.0% at 6 months. The responder rate declined to 11.4% at 12 months, limiting long-term clinical relevance. The responder rate declined to 11.4% at 12 months, limiting long-term clinical relevance. GNRFA may serve as a medium-term management option, and re-evaluation may be appropriate for recurrent symptoms. GNRFA may serve as a medium-term management option, and re-evaluation may be appropriate for recurrent symptoms. Degenerative osteoarthritis (OA) is the most common cause of chronic knee pain [1]. The global prevalence of knee OA is reported to be 16% in individuals aged 15 and over, and 22.9% in those aged 40 and over [2]. Due to an aging population and rising obesity rates, the number of individuals affected by symptomatic knee OA is projected to increase significantly [3]. The pathogenesis of OA is multifactorial and involves biomechanical, inflammatory, and metabolic processes. These mechanisms contribute to cartilage damage, synovitis, and various subchondral bone abnormalities [4]. This condition, characterized by the progressive degradation of articular cartilage, primarily results in pain, activity limitations, and reduced mobility [5]. Beyond physical impairment, it also restricts social participation and negatively affects overall quality of life [6]. Management of OA in routine clinical practice generally follows a stepped-care approach. First-line conservative treatments include patient education, physical therapy, weight management, various pharmacological agents, and intra-articular injections with substances such as corticosteroids [7]. In patients with advanced or treatment-resistant disease, total knee arthroplasty (TKA) is often considered as a surgical option. Although randomized controlled trials have demonstrated the superiority of TKA over non-surgical treatment for end-stage knee OA, non-operative management remains an important option for certain patient populations. This group includes individuals with mild-to-moderate symptoms, those who are not surgical candidates due to significant comorbidities, and patients who decline surgical intervention. Conventional genicular nerve radiofrequency ablation (GNRFA) is one of the interventional procedures used for the symptomatic treatment of eligible patients. The goal of the GNRFA procedure is to partially denervate the anterior knee capsule by creating a thermal lesion, thereby reducing OA-associated pain. The superomedial, superolateral, and inferomedial genicular nerves are the traditional targets for radiofrequency (RF) therapy, based on the initial controlled trial by Choi et al. [8]. Although existing studies have found GNRFA to be effective for pain and function in knee OA, the majority of this research is limited to follow-up periods of 3 or 6 months [9]. Our study aims to demonstrate the effects of GNRFA over a longer follow-up period of 12 months. MATERIALS AND METHODS Study Design and Patients This retrospective observational cohort study was conducted between January 2022 and January 2023 at the pain management clinic of a training and research hospital. The study was designed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. Written informed consent for the procedure had been obtained from all patients at the time of treatment. Ethical approval for the retrospective review of patient records was obtained from the local Institutional Ethics Committee (No: 89, Date: April 21, 2025). Due to the retrospective nature of the study, ethics approval was obtained after completion of the clinical treatments and before data extraction and analysis. No interventions were performed for research purposes. The research was carried out in compliance with the ethical principles outlined in the Declaration of Helsinki, as revised in 2013. Patients aged 45 years and older were screened for eligibility. Inclusion criteria were: Knee pain for 3 months or longer due to advanced-stage (Kellgren-Lawrence Grade 3 and 4) knee OA; a numeric rating scale (NRS) score of 5 or higher despite conservative medical and physical therapy; and providing written informed consent. Exclusion criteria included any of the following: Knee injections within the previous 3 months, knee deformity or inflammatory arthritis of other causes, a history of knee arthroplasty, pacemaker or implantable cardioverter defibrillator, lumbar radicular pain, malignancy, psychosis or progressive neurological disease, allergy to study medications, bleeding disorders, or incomplete medical records. For each included patient, the clinical diagnosis and grade of knee OA were confirmed through physical examination and radiological imaging (magnetic resonance imaging or X-ray). Procedures All procedures were performed by physicians specializing in musculoskeletal interventions, using a Toshiba TUS-A300™ ultrasound system (USA) equipped with an 11L4 linear transducer. A Cosman™ RFG-4 RF generator was used for the RF procedures. The interventions were carried out using 60 mm long, 20G Unified™ EchoRF™ hybrid needles, which feature a 5 mm active tip and integrate an electrode, cannula, and injection tube. A diagnostic block was administered to all three target nerves simultaneously during a single session. Following the blocks, patients were asked to perform provocation maneuvers (e.g., walking, stair climbing) that typically elicited their baseline knee pain. A positive response was defined as a practical pain reduction of ≥50% compared to their pre-block activity-related pain, occurring within 30–90 min and lasting for at least 2 h. Patients with a positive response to the diagnostic block proceeded to the GNRFA procedure. Both the diagnostic nerve blocks and the GNRFA procedure were performed using the technique described by Merrin and Grimshaw [10]. The patient was placed in the supine position between the physician and the ultrasound equipment. The affected knee was supported by a triangular wedge to maintain approximately 30° of flexion. Using an in-plane technique, the transducer was positioned parallel to the bone shaft to identify the genicular arteries near the periosteum. The superolateral target was at the junction of the lateral epicondyle and the distal femoral shaft; the superomedial target was at the junction of the medial epicondyle and the distal femoral shaft; and the inferomedial target was at the junction of the medial epicondyle and the tibial shaft (Fig. 1). The location of the genicular arteries was confirmed with color Doppler. To facilitate the procedure and improve accuracy, the linear probe was rotated 90° to visualize the genicular arteries longitudinally. The superolateral, superomedial, and inferomedial genicular nerves, which accompany these arteries, were targeted. Transducer positions and sonoanatomy of genicular nerves. (A) Superolateral genicular nerve. (B) Superomedial genicular nerve. (C) Inferomedial genicular nerve. The RF needle was gently advanced to the target site. Prior to initiating GNRFA, sensory stimulation was performed at 50 Hz and below 0.6 V to confirm the proximity of the cannula to the sensory nerve. Successful sensory stimulation was confirmed when the patient reported sensory complaints (numbness, tingling, or pain) concordant with their typical knee pain. Subsequently, motor stimulation was performed at 2 Hz and 3 V to ensure the cannula was distant from motor nerve fibers, which was verified by the absence of muscle contractions in the lower extremity. Once the correct position was verified and appropriate responses were elicited, 1 mL of 1% lidocaine was injected, and thermal RF was applied to each nerve at 80°C for 90 s. This was followed by negative aspiration and the administration of one-third of a mixture containing 1 mL of dexamethasone (4 mg) and 2 mL of 1% lidocaine (1 mL of the mixture per lesion) to the lesion site to reduce the risk of post-procedural neuralgia/neuritis. All procedures were performed under sterile conditions, and local skin anesthesia was applied before needle insertion. Post-procedure, patients were transferred to a recovery room and monitored for 2 h for any potential complications. One week after the injection, all patients were provided with an exercise booklet to guide their home rehabilitation program. Patients were advised to use 500 mg of paracetamol (up to a maximum of four tablets per day) and/or Naproxen (750 mg) as needed for pain relief. They were instructed not to take any analgesics on the day of or the day before their follow-up assessments, as this could interfere with the clinical outcome measurements. Outcome Measures We evaluated all patients for both primary and secondary outcome measures at baseline (pre-injection) and at 1, 6, and 12 months post-injection. The primary outcome measure was the NRS score, while the secondary outcome measures included the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores. Any adverse events occurring during the procedure or follow-up visits were documented. Primary Outcome Measures Pain intensity was assessed using the 11-point NRS-11, where 0 represents “no pain” and 10 represents “the worst pain imaginable.” Secondary Outcome Measures The WOMAC is a standardized questionnaire widely used by healthcare professionals to assess the condition of patients with knee and hip OA, including pain, stiffness, and physical function of the joints. It was developed in 1982 by Western Ontario and McMaster Universities. Higher scores indicate worse pain, stiffness, and functional limitations. The WOMAC consists of five items for pain (score range: 0–20), two items for stiffness (score range: 0–8), and 17 items for functional limitation (score range: 0–68). The validity and reliability of the Turkish version have been previously established [11]. Statistical Analyses Statistical analyses were performed using Statistical Package for the Social Sciences software (version 27, IBM Corp., Chicago, IL, USA). The normality of data distribution was assessed using the Shapiro–Wilk test. As the data were not normally distributed, non-parametric tests were used for the analysis. Descriptive statistics were presented as mean, median, standard deviation, frequency, and percentage. The Wilcoxon signed-rank test was used for pre- and post-procedural comparisons, while the Mann–Whitney U test was used to analyze the difference between two groups. Spearman’s correlation coefficient was calculated to assess correlations. Effect sizes (r) for the Wilcoxon signed-rank tests were calculated as r = |Z|/√N, where N was defined as the total number of observations contributing to each paired comparison (N=88, i.e., 44 pairs). To evaluate clinical significance at the individual level, clinical success was defined a priori as a ≥50% reduction in the baseline NRS score at the follow-up time points. RESULTS A total of 63 patient records were screened for eligibility for the study. After excluding eight patients who met the exclusion criteria, an additional 11 patients were removed due to loss to follow-up. Forty-four patients (n=44) who met the inclusion criteria were included in the final analysis (Fig. 2). An analysis comparing the baseline characteristics (age, sex, and baseline NRS scores) of the 11 patients lost to follow-up with the 44 included patients revealed no significant differences (p>0.05). No significant baseline differences were identified between included and excluded patients. Flow diagram of the study. Of the included patients, 32 (72.7%) were female. Twenty-five patients (56.8%) had Grade 4 OA, and in 34 patients (77.3%), the symptoms were on the right side (Table 1). Table 1Demographic and clinical characteristics (n=44)%Sex Woman72.7 Man27.3KL stage 343.2 456.8Side Left22.7 Right77.3KL: Kellgren Lawrence. Demographic and clinical characteristics (n=44) KL: Kellgren Lawrence. Table 2 shows the NRS and WOMAC scores at baseline and follow-up. Both scores decreased significantly at 1, 6, and 12 months compared to baseline (p<0.001). Table 3 presents the pairwise comparisons, 95% confidence intervals, and effect sizes. The procedure demonstrated large effect sizes for most pairwise comparisons. Despite this, mean NRS scores began to increase gradually after the 1st month. Table 2Actual NRS and WOMAC scores at baseline and follow-up time pointsTime pointNRS, Mean±SDNRS, median (IQR)WOMAC, Mean±SDWOMAC, median (IQR)Baseline (T0)7.1±1.17 (6–8)65.2±5.964 (62–70)1 month (T1)2.8±1.13 (2–3)26.4±9.924 (20–36)6 months (T2)3.3±1.03 (3–4)28.6±8.828 (24–36)12 months (T3)5.1±1.35 (4–6)47.8±11.246 (36–56)Data are presented as Mean±Standard deviation and median with interquartile range. T0=Baseline; T1=1 month; T2=6 months; T3=12 months. IQR: Interquartile range; NRS: Numeric Rating Scale; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index; SD: Standard deviation. Actual NRS and WOMAC scores at baseline and follow-up time points Data are presented as Mean±Standard deviation and median with interquartile range. T0=Baseline; T1=1 month; T2=6 months; T3=12 months. IQR: Interquartile range; NRS: Numeric Rating Scale; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index; SD: Standard deviation. Table 3Pairwise comparisons of changes in NRS and WOMAC scoresComparisonMean difference±SD95% CI of differenceWilcoxon (Z)pEffect size (r)*NRS T0–T14.2±1.53.76 to 4.64-5.8<0.0010.61 (large) T0–T23.8±1.33.42 to 4.18-5.8<0.0010.61 (Large) T0–T31.8±1.41.39 to 2.21-5.2<0.0010.55 (large) T1–T2-0.4±0.7-0.61 to -0.19-3.40.0010.36 (medium) T1–T3-2.4±1.7-2.9 to -1.9-5.4<0.0010.57 (large) T2–T3-2.0±1.5-2.44 to -1.56-5.3<0.0010.56 (large)WOMAC T0–T138.9±12.335.27 to 42.53-5.7<0.0010.60 (large) T0–T236.6±1133.35 to 39.85-5.7<0.0010.60 (large) T0–T317.5±13.213.60 to 21.40-5.2<0.0010.55 (large) T1–T2-2.2±7.2-4.33 to -0.07-2.10.020.22 (small) T1–T3-21.3±16.7-26.23 to -16.37-5.3<0.0010.56 (large) T2–T3-19.1±13.9-23.21 to -14.99-5.5<0.0010.58 (large)95% CI: 95% Confidence interval. Effect size (r) calculated as |Z|/√N (N=88, i.e., 44 pairs). Interpretation: 0.1=Small; 0.3=Medium; 0.5=Large effect. T0=Baseline; T1=1 Month; T2=6 Months; T3=12 Months. NRS: Numeric Rating Scale; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index; SD: Standard deviation. Pairwise comparisons of changes in NRS and WOMAC scores 95% CI: 95% Confidence interval. Effect size (r) calculated as |Z|/√N (N=88, i.e., 44 pairs). Interpretation: 0.1=Small; 0.3=Medium; 0.5=Large effect. T0=Baseline; T1=1 Month; T2=6 Months; T3=12 Months. NRS: Numeric Rating Scale; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index; SD: Standard deviation. Regarding clinical success (≥50% pain reduction), 36 patients (81.8%) met this threshold at 1 month, and 33 (75.0%) met the same criterion at 6 months. By 12 months, only 5 patients (11.4%) still met this criterion. In patients with Grade 3 knee OA, baseline NRS scores were significantly lower than those in patients with Grade 4 OA (p<0.05). Post-procedure, the radiological grade did not create a significant difference in NRS scores (p>0.05). No significant difference was found between the radiological grade and WOMAC scores at any time point (p>0.05). No adverse events were observed in patients during the procedure or throughout the follow-up period. When examining the relationship between NRS and WOMAC scores, a strong positive correlation was observed between NRS and WOMAC scores at baseline (Spearman’s ρ=0.822), and this relationship persisted throughout the post-treatment period. At 1 month, ρ was 0.892, at 6 months ρ was 0.896, and at 12 months ρ was 0.973. DISCUSSION In this retrospective cohort, we evaluated the safety and efficacy of ultrasound-guided GNRFA in patients with advanced knee OA. Our findings suggest that GNRFA was associated with improvements in pain and function from baseline through the 12-month follow-up. After the initial decline recorded at 1 month, NRS scores showed a subsequent increase at 6 and 12 months. Pain scores increased over time but remained lower than baseline values. The statistically significant increase in pain after the 1st month suggests that the clinical benefit of GNRFA appears to diminish over time rather than representing a permanent solution. Although mean NRS and WOMAC scores remained statistically lower than baseline at 12 months, only 11.4% of patients maintained a clinically meaningful reduction in pain. This discrepancy highlights the distinction between statistical significance and patient-level clinical benefit, suggesting that the long-term clinical relevance of a single GNRFA procedure may be limited for most patients. Furthermore, these responder rates should be interpreted cautiously, as only patients with a positive diagnostic genicular nerve block proceeded to ablation, which may have enriched the cohort for potential responders. The observed loss of clinical benefit over time may be related to nerve regeneration, although this mechanism was not directly evaluated in the present study. RF energy causes Wallerian degeneration of the axons but largely preserves the nerve’s connective tissue sheaths, which theoretically allows the nerve to regenerate [12]. This biological plausibility may partly explain the observed decline in responder rates over time. These findings support the role of GNRFA as a medium-term management option rather than a permanent solution and suggest that re-evaluation may be appropriate in patients with recurrent symptoms around 6 months. In our study, the baseline NRS scores of patients with Grade 3 knee OA were significantly lower than those of patients with Grade 4. This is consistent with clinical observations and literature findings suggesting that pain intensity increases in the more advanced stages of OA [13]. However, no statistically significant differences were observed in post-procedural NRS scores between patients with Grade 3 and Grade 4 OA. Similar post-procedural outcomes across radiological grades suggest that the observed improvements were not restricted to patients with less advanced disease. These findings indicate that patients with advanced radiographic OA may also experience improvements in pain scores following GNRFA, although confirmation in controlled studies is required. Another key finding of our study was the high correlation between NRS and WOMAC scores. This relationship supports the notion that the intervention similarly affects both pain intensity and functional limitations, and that these parameters are reliable indicators of a patient’s quality of life. The particularly strong correlation of 97.3% at 12 months indicates that long-term pain control and functional improvement proceed in parallel. These findings indicate a close relationship between pain intensity and functional status, which is consistent with other studies in the literature [14]. While studies on GNRFA typically have follow-up periods of 3 or 6 months, the number of studies with a 12-month follow-up, like ours, is limited. While some studies report that the effects of GNRFA diminish after 3 months [15], others have reported sustained efficacy for 12 months or longer [16]. This discrepancy may partly reflect differences in the definitions and thresholds used to define treatment success across studies. Research into minimally invasive treatments for chronic knee OA has examined the efficacy of various methods beyond GNRFA, such as cryoablation, cooled RF, bipolar RF, and pulsed RF. Some studies have explored GNRFA with different durations and needles, sometimes targeting nerves other than the classic three. The results from these studies are conflicting. The lack of sufficient high-level evidence and the lack of comparative studies against traditional GNRFA indicate the need for further randomized controlled trials. A systematic review and meta-analysis of RCTs comparing cooled, monopolar, or bipolar RFA or cryoneurolysis with sham treatment in adults with symptomatic knee OA concluded that genicular neurolysis, particularly conventional monopolar RF ablation, provides clinically significant pain and functional improvement with a favorable safety profile [17]. In our protocol, a low dose of dexamethasone (4 mg) was administered following the creation of the RF lesion. The primary rationale for this was to minimize the risk of post-procedural neuritis and localized inflammation. While periarticular corticosteroids can provide immediate and short-term analgesia, their duration of action is generally limited to a few weeks. It is highly unlikely that the steroid injection significantly contributed to the sustained analgesic effects observed at the 6- and 12-month follow-ups. Nonetheless, its use represents a minor confounding factor when comparing our long-term outcomes with studies that utilize RF alone. Our findings also support the safety of GNRFA. Although serious adverse events such as septic arthritis [18], hematoma [19], and third-degree burns [20] have been reported, these are generally limited to case reports. The absence of any serious adverse events in our study, consistent with many others, supports the finding that GNRFA is a safe procedure. Our study has several important limitations. First, the retrospective observational design and the absence of a control or sham group make it impossible to definitively distinguish the specific effects of GNRFA from placebo effects, regression to the mean, natural symptom fluctuation, or the impact of concomitant treatments like rehabilitation exercises. Second, our sample size was relatively small. Although baseline characteristics did not differ significantly between groups, attrition bias cannot be completely excluded, given that nearly 20% of the initially eligible patients were lost to follow-up. Third, because only patients with a positive diagnostic block proceeded to ablation, the study population was inherently enriched for potential responders. Therefore, the observed responder rates may overestimate treatment outcomes compared with an unselected population of patients with advanced knee OA. In addition, the single-center nature of the study may limit the external validity of the findings. Finally, the routine administration of low-dose dexamethasone after creating the RF lesion represents a minor confounding factor; although its effects are short-lived, it should be considered when comparing our outcomes with studies using RF alone. Conclusion In this retrospective cohort, GNRFA was associated with meaningful improvements in pain and function, most prominently in the early post-procedural period, with reduced durability beyond 6 months. These improvements were observed across radiological grades. The decline in responder rates after 6 months suggests that repeat ablation may be necessary in selected patients. GNRFA appears to be a reasonable non-surgical option, particularly for patients at high surgical risk or unwilling to undergo TKA. Confirmation of these findings will require adequately powered randomized sham-controlled studies.