Long-Term Pain and Functional Outcomes After Cooled Genicular Radiofrequency Ablation With or Without Adjunctive Intra-Articular Botulinum Toxin Type A in Knee Osteoarthritis

Abstract

Background

Cooled genicular radiofrequency ablation (CRFA) is used to treat symptomatic knee osteoarthritis (OA), although treatment response may decline over time. This study evaluated 24-month outcomes of CRFA alone versus CRFA combined with intra-articular botulinum toxin type A (BoNT-A).

Material/Methods

In this retrospective observational comparative-effectiveness cohort study, 90 patients with moderate-to-severe knee OA underwent CRFA alone (n = 50) or CRFA plus intra-articular BoNT-A (n = 40). The combination group received BoNT-A at baseline and a planned repeat injection at ~ 12 months. Pain and function were assessed using the Numerical Rating Scale (NRS) and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). The primary durability endpoint was the between-group difference in longitudinal NRS trajectories over 24 months, assessed using adjusted mixed-effects modeling.

Results

Both groups demonstrated substantial early improvement; outcome trajectories diverged after ~ 12 months. Significant treatment-by-time interactions were observed for NRS (β = −0.84, 95% confidence interval [CI] −1.21 to −0.47; P = 0.0002) and WOMAC outcomes (β = −6.3, 95% CI −9.4 to −3.2; P = 0.0003). At 24 months, mean observed NRS scores were 4.2 ± 2.3 after CRFA alone and 1.8 ± 0.9 after CRFA plus BoNT-A; responder rates were 51.1% and 86.5%, respectively (P = 0.0004).

Conclusions

In this retrospective observational cohort, a longitudinal treatment pathway combining CRFA with intra-articular BoNT-A was associated with more sustained pain relief and functional improvement over 24 months than CRFA alone. Prospective randomized studies are needed to confirm these findings and clarify the independent contributions of adjunctive and repeat intra-articular BoNT-A.

Introduction

Knee osteoarthritis (OA) is a major cause of chronic musculoskeletal pain, disability, and reduced quality of life worldwide; its prevalence continues to rise with population aging and increasing obesity rates [1,2]. The condition is characterized by progressive cartilage degeneration, synovial inflammation, and structural joint remodeling, all of which contribute to persistent pain, impaired mobility, and functional limitations [2]. Although conservative management strategies—including pharmacologic therapy, physical rehabilitation, and intra-articular injections—may provide symptomatic improvement, many patients continue to experience clinically significant pain despite optimized nonsurgical treatment [3]. Current clinical guidelines emphasize a multimodal management approach for individuals with persistent symptoms, particularly those who are not immediate candidates for surgical intervention [4,5].

Total knee arthroplasty remains an effective treatment for end-stage knee OA, but many patients seek to delay surgery or are poor surgical candidates due to age, comorbidities, or personal preference. Consequently, there is an ongoing need for durable, minimally invasive interventions that provide sustained symptom relief and functional improvement while potentially postponing arthroplasty [4,5].

Among available interventional options, cooled radiofrequency ablation (CRFA) of the genicular nerves (also called cooled genicular radiofrequency ablation) has emerged as an effective treatment for chronic knee OA pain. Compared with conventional radiofrequency ablation, CRFA produces larger and more consistent thermal lesions, which may improve interruption of nociceptive transmission from the articular sensory branches innervating the knee joint [6]. Multiple clinical studies have demonstrated significant reductions in pain intensity and improvements in function after CRFA in patients with knee OA [7]. Randomized trials and prospective studies have also shown sustained analgesic benefits for up to 12 to 24 months in selected patients undergoing CRFA [8–10]. However, treatment response durability remains variable. Although many patients experience substantial relief during the first 6 to 12 months after treatment, pain recurrence has been observed during extended follow-up in a subset of patients [10–12]. This diminished benefit may reflect peripheral nerve regeneration along with persistent intra-articular inflammatory and sensory pain mechanisms that contribute to ongoing OA-related symptoms [13].

Botulinum toxin type A (BoNT-A) has emerged as a potential therapeutic option for chronic pain disorders based on its antinociceptive and neuromodulatory properties. In addition to its established neuromuscular blocking effects, BoNT-A inhibits the release of pain-related mediators (eg, substance P, glutamate, and calcitonin gene-related peptide), thus reducing peripheral sensitization and neurogenic inflammation [14–16]. Experimental and clinical studies suggest that these mechanisms can contribute to analgesic benefits in OA-related joint pain [17–21]. Randomized controlled trials and prospective studies have also identified improvements in pain and functional outcomes after intra-articular BoNT-A administration in patients with refractory knee OA symptoms [17–21].

Because CRFA and intra-articular BoNT-A act through distinct mechanisms, the combination of these interventions may offer a complementary approach to managing persistent OA pain. CRFA primarily interrupts peripheral pain signaling via thermal ablation of the genicular sensory branches [6–10], whereas intra-articular BoNT-A may modulate inflammatory and neurochemical pathways within the joint environment by inhibiting release of pain mediators [14–16]. Systematic reviews and meta-analyses have reported analgesic and functional benefits associated with radiofrequency ablation for knee OA [22–24]; studies evaluating intra-articular BoNT-A have also demonstrated symptomatic benefits in selected patients [17–21]. However, evidence concerning the combined use of these interventions remains limited, and comparative longitudinal data over extended follow-up are scarce.

Accordingly, the present study evaluated pain and functional outcomes after CRFA with or without adjunctive intra-articular BoNT-A in patients with moderate-to-severe knee OA over a 24-month follow-up period. The hypothesis was that the longitudinal treatment strategy incorporating adjunctive intra-articular BoNT-A would be associated with greater maintenance of analgesic and functional improvement relative to the CRFA-alone treatment strategy.

Material and Methods

Ethics Approval

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and approved by the Research Ethics Committee of the University of Tabuk, Tabuk, Saudi Arabia (Research Ethics Committee Reference No. 644; Approval No. UT-644-397-2025; approved on May 14, 2025). Given the retrospective observational design and use of anonymized clinical data obtained during routine care, the requirement for written informed consent was waived by the Research Ethics Committee. All data were de-identified before analysis to ensure confidentiality and compliance with applicable ethical and data protection standards.

Participants

Eligible patients were identified through a retrospective review of interventional pain clinic and procedural records at Prince Fahad bin Sultan Hospital. All consecutive patients with knee OA who underwent CRFA during the study period (January 2023 to May 2025) and met the predefined inclusion criteria were considered eligible. After application of the predefined inclusion and exclusion criteria, 90 patients with chronic symptomatic knee OA were included in the final analysis.

Patients were eligible if they were at least 50 years of age; had moderate-to-severe knee OA; had chronic knee pain persisting for more than 6 months; had radiographic evidence consistent with Kellgren-Lawrence (KL) grade II to IV OA; had experienced conservative management failure; underwent CRFA as part of routine clinical care; and had longitudinal clinical follow-up extending to at least 24 months within predefined follow-up assessment windows after the index procedure. Although KL grade II disease was eligible for inclusion, all patients who ultimately met the clinical and treatment-selection criteria and were included in the final cohort had KL grade III to IV OA. Conservative management failure—documented by the referring orthopedic team and confirmed during interventional pain clinic evaluation—was defined as persistent symptoms despite pharmacologic and nonpharmacologic therapies, including regular non-steroidal anti-inflammatory drug use; rescue analgesic therapy with tramadol 50 mg or codeine 30 mg administered at least twice weekly and/or gabapentinoid therapy exceeding 300 mg daily for at least 6 months; structured physical therapy (> 12 sessions during the preceding 6 months without clinically meaningful improvement, defined as < 2 points of improvement in the Numerical Rating Scale [NRS] pain score); and/or prior intra-articular interventions such as corticosteroid or hyaluronic acid injections.

Patients were excluded if they had undergone prior total knee arthroplasty; had an active joint infection; had inflammatory arthritis (eg, rheumatoid arthritis); had severe neurological disorders affecting lower-extremity sensation; had incomplete clinical records or insufficient longitudinal follow-up; or had undergone genicular nerve ablation within the preceding 12 months.

All included participants had documented longitudinal follow-up extending to 24 months after the index procedure, although intermittent missing outcome observations occurred at selected follow-up visits. Only 1 knee per patient was included in the analysis.

Treatment Allocation

Treatment allocation was not randomized. The decision to offer adjunctive intra-articular BoNT-A in addition to CRFA was based on individualized clinical judgment during routine practice. Factors influencing treatment selection included symptom chronicity, degree of functional limitation, diffuse nociceptive symptom burden, prior response to intra-articular therapies, anticipated long-term treatment requirements, patient preference, and physician assessment of the expected clinical benefit. No standardized institutional algorithm governed the use of adjunctive intra-articular BoNT-A. Patients in the combination-treatment group received an initial intra-articular BoNT-A injection at the time of CRFA and a planned repeat injection ~ 12 months later as part of the predefined longitudinal treatment pathway.

Interventions

All procedures were performed by fellowship-trained, board-certified interventional pain physicians experienced in fluoroscopy-guided genicular nerve interventions, using a standardized clinical treatment pathway as part of routine clinical care.

Diagnostic Genicular Nerve Block

Prior to CRFA, all patients underwent diagnostic genicular nerve blocks to confirm concordant pain relief and suitability for radiofrequency intervention. Under ultrasound guidance, 1 mL of 0.25% bupivacaine (Marcaine^®, AstraZeneca, Södertälje, Sweden) was injected around each target genicular nerve, including the superior medial, superior lateral, and inferior medial genicular nerves. Patients demonstrating greater than 50% reduction in pain intensity during the subsequent 24-hour period were considered appropriate candidates for CRFA.

Procedural Sedation

All CRFA procedures were performed under monitored procedural sedation using intravenous fentanyl combined with a titrated propofol infusion according to standardized institutional practice and patient tolerance.

CRFA Procedure

CRFA targeted the 3 principal articular sensory branches innervating the knee joint: the superior medial, superior lateral, and inferior medial genicular nerves. Under fluoroscopic guidance, cooled radiofrequency probes were advanced to the target locations using established anatomical landmarks adjacent to the femoral and tibial epicondyles. Standardized fluoroscopic views and procedural protocols were applied across all cases, including patient positioning, target nerve localization, probe placement, and confirmation of electrode positioning.

Before lesioning, sensory stimulation was performed at 50 Hz and motor stimulation at 2 Hz to verify accurate probe placement. Sensory stimulation reproduced concordant knee pain without motor activation, confirming accurate targeting of the articular sensory branches while minimizing the risk of unintended motor nerve injury.

Thermal lesioning was performed using cooled radiofrequency settings of 60 °C for 230 seconds at each target site via 17-gauge, 75-mm cooled electrodes equipped with 4-mm active tips (Coolief^®, Avanos Medical, Alpharetta, GA, USA). Each target nerve underwent 3 ablation cycles to ensure adequate lesion coverage.

Intra-Articular BoNT-A Injection

For patients treated within the predefined combination-treatment pathway, intra-articular BoNT-A was administered immediately after the CRFA procedure. A dose of 100 units of onabotulinumtoxinA (Botox^®, Allergan Inc., Irvine, CA, USA), diluted in 2 mL of normal saline, was injected into the knee joint under ultrasound guidance (Sonosite^®, Bothell, WA, USA) using a standardized lateral intra-articular approach with a 22-gauge needle to ensure accurate needle placement.

The 100-unit dose was selected based on previously published clinical studies evaluating intra-articular BoNT-A for symptomatic knee OA, which have shown favorable efficacy and safety outcomes using similar dosing regimens [17–21,25,26]. As part of the predefined longitudinal treatment strategy, a planned repeat intra-articular injection of 100 units of BoNT-A was administered ~ 12 months after the index procedure in an outpatient procedure room. This repeat injection reflected routine longitudinal clinical management intended to maintain symptom control during extended follow-up and was not introduced as a protocol-mandated research intervention.

Procedure Standardization

To minimize operator-dependent variability, all procedures were performed using standardized techniques, including fluoroscopic localization of target nerves, ultrasound-guided diagnostic blocks and intra-articular injections, stimulation testing, lesioning parameters, sedation protocols, and procedural workflow. No additional routinely planned interventional knee procedures were scheduled during the predefined follow-up period other than the planned BoNT-A reinjection at ~ 12 months in the combination-treatment group. Only the index knee selected for longitudinal outcome analysis underwent the standardized intervention pathway included in this study. No immediate procedure-related complications were observed during the intervention sessions.

Follow-Up Protocol

Patients were routinely followed in the outpatient pain clinic at 1, 3, 6, 12, 18, and 24 months after the index procedure. Follow-up visits were considered valid if completed within predefined assessment windows of ± 2 weeks for early follow-up visits (1–6 months) and ± 4 weeks for later follow-up visits (12–24 months).

At each scheduled visit, clinical outcomes were assessed using standardized evaluation forms and documented by trained clinical personnel as part of routine longitudinal follow-up care. Patients continued to receive routine individualized clinical care throughout the follow-up period. Postprocedural analgesic management, rehabilitation participation, and supportive noninterventional treatments were not standardized as part of the study; they were managed according to clinician judgment and patient-specific clinical needs. Because the study was a retrospective analysis, longitudinal information regarding medication adjustments, rehabilitation adherence, and supportive therapies was not consistently available for all participants.

Pain intensity and functional outcomes were evaluated at all scheduled follow-up visits. Intermittent missing observations at selected time points were addressed via linear mixed-effects models, which permit inclusion of participants with partially incomplete repeated measurements under the missing-at-random assumption. Radiographic follow-up was not systematically performed because serial imaging was not routinely indicated after CRFA or intra-articular BoNT-A administration in the absence of new clinical concerns. Accordingly, follow-up assessments focused on predefined clinical outcomes, including pain intensity and functional status, rather than structural imaging endpoints.

Safety outcomes were retrospectively assessed via review of outpatient follow-up records, procedural documentation, emergency department visits, and hospital admission records throughout the 24-month follow-up period. Because safety ascertainment relied on routine clinical documentation rather than prospective standardized adverse-event surveillance, minor transient events (eg, postprocedural soreness, localized bruising, transient pain flare, or self-limited knee discomfort) may not have been consistently captured.

Outcome Measures

Clinical outcomes were assessed using validated measures of pain intensity and functional impairment in knee OA. The same outcome instruments and assessment procedures were consistently applied across all follow-up visits.

Pain Intensity

Pain severity was evaluated using the NRS, a validated 11-point scale ranging from 0 (no pain) to 10 (worst imaginable pain). Higher scores indicated greater pain intensity.

Functional Outcome

Functional status was assessed via the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), a widely used and validated instrument that evaluates pain, stiffness, and physical function in patients with knee OA. The WOMAC total score was analyzed as a continuous composite measure ranging from 0 to 96; higher scores indicated greater pain, stiffness, and functional impairment. Outcome assessments were recorded at baseline and at 1, 3, 6, 12, 18, and 24 months after the index procedure.

Primary Durability Outcome

The predefined primary durability endpoint was the between-group difference in longitudinal NRS pain trajectories over the 24-month follow-up period, assessed using the treatment-by-time interaction in the adjusted mixed-effects model. To facilitate clinical interpretation of long-term outcome differences, model-derived adjusted between-group contrasts at the 24-month assessment were also estimated.

Secondary Outcomes

Secondary outcomes included longitudinal WOMAC functional trajectories, adjusted model-based between-group differences in WOMAC outcomes during follow-up, and responder analyses based on clinically meaningful reductions in NRS pain scores.

Responder Analysis

Responder analysis was performed to determine the proportion of patients achieving at least 50% reduction in NRS pain scores relative to baseline, a threshold widely considered to represent clinically meaningful improvement in chronic pain studies. Responder rates were calculated using observed-case denominators at each follow-up assessment.

Statistical Analysis

All statistical analyses were performed using IBM SPSS Statistics for Windows, version 29.0 (IBM Corp., Armonk, NY, USA). Continuous variables are presented as mean ± standard deviation, whereas categorical variables are reported as frequencies and percentages. The normality of continuous variables was assessed using the Shapiro-Wilk test before parametric comparisons. Baseline demographic and clinical characteristics between treatment groups were compared using independent-samples t-tests for continuous variables and chi-square tests for categorical variables.

Because this study utilized a retrospective observational cohort design with consecutive inclusion of eligible patients, no prospective a priori sample size calculation was performed. The final sample comprised all consecutively identified patients who met the predefined inclusion and exclusion criteria during the study period and thus reflected the available real-world clinical population, rather than a predetermined recruitment target.

The primary durability endpoint was the between-group difference in longitudinal NRS pain trajectories over the 24-month follow-up period, assessed using the treatment-by-time interaction in the adjusted mixed-effects model. Secondary outcomes included longitudinal WOMAC functional trajectories, responder rates defined as at least 50% reduction in NRS pain scores relative to baseline, and adjusted model-predicted outcome trajectories. Additionally, a model-derived adjusted between-group contrast at the 24-month assessment was estimated to facilitate clinical interpretation of long-term outcome differences.

Longitudinal changes in NRS and WOMAC scores were analyzed using linear mixed-effects models to adjust for repeated within-participant measurements across scheduled follow-up assessments. Time was modeled as an ordinal follow-up variable representing the sequence of scheduled assessments (baseline and 1, 3, 6, 12, 18, and 24 months). Treatment group was modeled as a binary categorical variable representing the predefined longitudinal treatment strategies (CRFA alone versus CRFA combined with intra-articular BoNT-A administered at baseline with planned reinjection at ~ 12 months).

Adjusted mixed-effects models included time, treatment group, and treatment-by-time interaction terms as fixed effects; patient identifier was incorporated as a random intercept to adjust for within-participant correlations across repeated observations. Repeated longitudinal measurements were modeled using a first-order autoregressive [AR(1)] covariance structure with restricted maximum likelihood estimation. The AR(1) covariance structure was selected to adjust for the expected temporal correlation between repeated observations obtained at sequential follow-up assessments. To reduce measured confounding associated with the nonrandomized observational design, adjusted analyses incorporated predefined baseline covariates, including age, sex, body mass index, KL grade, and the corresponding baseline outcome measure. Baseline NRS score was included in NRS models, whereas baseline WOMAC score was included in WOMAC models. KL grade was entered as a binary categorical variable comparing grade IV OA with grade III OA. Responder rates were compared between treatment groups using chi-square tests based on observed-case denominators at each scheduled follow-up assessment.

Model results are reported as regression coefficients (β) with corresponding standard errors (SEs), 95% confidence intervals (CIs), and exact 2-sided P-values. A P-value < 0.05 was considered statistically significant. No formal adjustment for multiple comparisons was performed because the analyses were predefined and exploratory in nature; secondary outcome findings should be interpreted cautiously.

Intermittent missing observations occurred at selected follow-up visits and were handled using mixed-effects models under the missing-at-random assumption. This approach allowed inclusion of all available repeated observations without requiring complete-case restriction.

Model adequacy was assessed by inspection of residual plots, evaluation of residual normality and homoscedasticity, assessment of covariance structure fit, and convergence diagnostics. Inspection of residuals and standardized residuals demonstrated no major deviations from approximate normality or homoscedasticity across follow-up assessments; all primary mixed-effects models successfully converged.

Bias Control and Missing Data Handling

To reduce potential bias inherent to retrospective observational cohort studies, all consecutively identified eligible patients who met the predefined inclusion criteria during the study period were included. Baseline demographic and clinical characteristics were compared between treatment groups to assess measurable imbalances and potential confounding by indication associated with clinician-guided treatment selection during routine practice. Because treatment allocation was not randomized, adjusted mixed-effects models incorporating predefined baseline covariates were used to reduce measured confounding. However, information regarding concomitant analgesic therapy, rehabilitation adherence, psychosocial factors, supportive noninterventional treatments, and physician-specific treatment preferences was not uniformly documented in the retrospective dataset and thus could not be reliably standardized or incorporated into the adjusted analyses.

Intermittent missing observations at selected follow-up assessments were addressed using mixed-effects models under the missing-at-random assumption, thereby permitting inclusion of participants who had partially incomplete repeated measurements without excluding all available longitudinal data.

Results

Study Population

In total, 112 patients underwent evaluation for CRFA during the study period. After application of the predefined inclusion and exclusion criteria, 90 patients with persistent moderate-to-severe symptomatic knee OA were included in the final analysis. Reasons for exclusion were incomplete clinical records (n = 12), insufficient longitudinal follow-up of less than 24 months (n = 6), and prior total knee arthroplasty (n = 4). Participant selection, treatment-group allocation, and longitudinal follow-up are illustrated in Figure 1.

Figure 1.

Study flow diagram showing patient identification, eligibility assessment, application of the inclusion and exclusion criteria, treatment-strategy classification, follow-up, and final analysis. Abbreviations: BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation.

Of the included participants, 50 underwent CRFA alone, whereas 40 received CRFA combined with intra-articular BoNT-A administered at baseline with a scheduled repeat injection at ~ 12 months as part of the predefined longitudinal treatment strategy. All participants had persistent symptomatic knee OA despite prior conservative management, including pharmacologic therapy, physical therapy, and intra-articular therapies.

Baseline Characteristics

Baseline demographic and clinical characteristics were comparable between treatment groups, with no statistically significant differences across the measured variables (Table 1). Mean ages were 64.2 ± 8.5 years in the CRFA group and 63.7 ± 9.1 years in the CRFA plus BoNT-A group. Female participant proportions were 31/50 (62%) in the CRFA group and 24/40 (60%) in the CRFA plus BoNT-A group; mean body mass indices were 31.5 ± 4.2 kg/m² and 32.1 ± 4.6 kg/m², respectively. KL grade III OA was present in 29/50 (58%) patients in the CRFA group and 22/40 (55%) patients in the CRFA plus BoNT-A group; grade IV OA was present in 21/50 (42%) and 18/40 (45%) patients, respectively.

Table 1.

Baseline patient characteristics.

Variable	CRFA (n = 50)	CRFA + BoNT-A (n = 40)	P
Age, years	64.2 ± 8.5	63.7 ± 9.1	0.74
Female sex, n (%)	31 (62%)	24 (60%)	0.84
BMI, kg/m2	31.5 ± 4.2	32.1 ± 4.6	0.56
KL grade III, n (%)	29 (58%)	22 (55%)	0.78
KL grade IV, n (%)	21 (42%)	18 (45%)	0.78
Baseline NRS	7.9 ± 0.4	8.1 ± 0.5	0.32
Baseline WOMAC	78.2 ± 6.3	77.4 ± 5.8	0.61

Abbreviations: BMI, body mass index; BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation; KL, Kellgren-Lawrence; NRS, Numerical Rating Scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index. Legend: Values are presented as mean ± standard deviation or n (%).

Mean baseline NRS pain scores were 7.9 ± 0.4 in the CRFA group and 8.1 ± 0.5 in the CRFA plus BoNT-A group. The limited variability in baseline pain scores likely reflects the inclusion criteria requiring chronic symptomatic moderate-to-severe knee OA with persistent symptoms despite conservative treatment. Mean baseline WOMAC scores were 78.2 ± 6.3 and 77.4 ± 5.8, respectively. Independent-samples t-tests and chi-square analyses confirmed no statistically significant differences across the measured demographic or clinical variables (all P > 0.05), indicating good baseline comparability between treatment groups.

Safety Outcomes

No major procedure-related adverse events requiring hospitalization or additional intervention were identified in either treatment group during the 24-month follow-up period. Observed major adverse event rates were 0/50 (0%) in the CRFA group and 0/40 (0%) in the CRFA plus BoNT-A group.

Specifically, no cases of joint infection, clinically significant bleeding, persistent neurological deficit, procedure-related hospitalization, or other serious adverse events were identified. No immediate complications related to diagnostic genicular nerve blocks, CRFA lesioning, procedural sedation, or intra-articular BoNT-A administration were documented. No sedation-related adverse events requiring escalation of care or hospitalization were observed.

Longitudinal Pain Outcomes

Observed longitudinal pain outcomes are summarized in Table 2, and adjusted model-predicted trajectories derived from the mixed-effects analyses are illustrated in Figure 2. Both treatment groups demonstrated substantial reductions in pain intensity during early follow-up; the greatest improvement was observed within the first 3 months after treatment. Mean observed NRS scores decreased from 7.9 ± 0.4 at baseline to 5.6 ± 0.3 at 1 month and 1.9 ± 0.8 at 3 months in the CRFA group. Similarly, NRS scores in the CRFA plus BoNT-A group decreased from 8.1 ± 0.5 at baseline to 4.0 ± 0.3 at 1 month and 1.5 ± 0.5 at 3 months (Table 2).

Table 2.

Longitudinal NRS and WOMAC outcomes.

Time point	CRFA NRS	CRFA + BoNT-A NRS	CRFA WOMAC	CRFA + BoNT-A WOMAC
Baseline	7.9 ± 0.4	8.1 ± 0.5	78.2 ± 6.3	77.4 ± 5.8
1 month	5.6 ± 0.3	4.0 ± 0.3	55 ± 6	38 ± 5
3 months	1.9 ± 0.8	1.5 ± 0.5	19 ± 8	16 ± 5
6 months	2.2 ± 0.9	1.6 ± 0.6	22 ± 9	17 ± 6
12 months	3.0 ± 1.5	1.9 ± 0.7	30 ± 14	20 ± 8
18 months	3.7 ± 2.0	1.6 ± 0.8	36 ± 18	15 ± 9
24 months	4.2 ± 2.3	1.8 ± 0.9	41 ± 20	18 ± 10

Abbreviations: BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation; NRS, Numerical Rating Scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index. Legend: Values represent observed mean ± standard deviation at each assessment time point.

Figure 2.

Observed and adjusted model-predicted Numerical Rating Scale (NRS) trajectories over 24 months according to treatment strategy. Solid lines represent mixed-effects model-predicted trajectories, and shaded areas represent 95% confidence intervals. Abbreviations: BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation; M, months.

Longitudinal pain trajectories were analyzed using adjusted random-intercept mixed-effects models that adjusted for repeated within-participant observations across follow-up assessments. Adjusted linear mixed-effects modeling demonstrated a significant overall longitudinal time effect for NRS scores (β = −2.84, SE = 0.31, 95% CI −3.45 to −2.23, P = 0.0001), indicating progressive improvement in pain intensity across follow-up assessments. A significant treatment-by-time interaction was also observed for longitudinal NRS outcomes (β = −0.84, SE = 0.19, 95% CI −1.21 to −0.47, P = 0.0002), suggesting differential pain trajectories between treatment strategies over time.

The predefined primary durability endpoint was the between-group difference in longitudinal NRS pain trajectories over the 24-month follow-up period, assessed using the treatment-by-time interaction in the adjusted mixed-effects model. Throughout the 24-month follow-up period, patients managed with the predefined CRFA plus BoNT-A longitudinal treatment strategy demonstrated consistently lower observed pain scores relative to those of patients managed with CRFA alone. Exploratory assessment of longitudinal trajectories showed similar early improvements in both treatment groups during the first 6 months; progressive divergence became apparent after ~ 12 months. Given that the scheduled treatment strategy in the combination group included planned BoNT-A reinjection at ~ 12 months, the post-12-month divergence in outcome trajectories should be interpreted in the context of the overall longitudinal treatment strategy (not as an isolated effect of baseline treatment exposure).

At the predefined 24-month assessment, observed NRS outcome data were available for 45/50 patients in the CRFA group and 37/40 patients in the CRFA plus BoNT-A group (Table 3). Mean observed NRS scores at 24 months were 4.2 ± 2.3 in the CRFA group and 1.8 ± 0.9 in the CRFA plus BoNT-A group (Table 2). To facilitate clinical interpretation of long-term outcome differences, a model-derived adjusted between-group contrast was also estimated at the 24-month assessment. At 24 months, adjusted mixed-effects modeling estimated an NRS score ~ 2.4 points lower in the CRFA plus BoNT-A group than in the CRFA-alone group, exceeding commonly reported minimal clinically important difference thresholds for chronic pain outcomes. These findings should be interpreted as observational longitudinal associations under routine clinical care, rather than as causal treatment effects.

Table 3.

Longitudinal outcome availability.

Time point	CRFA NRS	CRFA + BoNT-A NRS	CRFA WOMAC	CRFA + BoNT-A WOMAC
Baseline	50/50 (100%)	40/40 (100%)	50/50 (100%)	40/40 (100%)
1 month	49/50 (98%)	40/40 (100%)	49/50 (98%)	40/40 (100%)
3 months	48/50 (96%)	39/40 (97.5%)	48/50 (96%)	39/40 (97.5%)
6 months	47/50 (94%)	39/40 (97.5%)	47/50 (94%)	39/40 (97.5%)
12 months	46/50 (92%)	38/40 (95%)	46/50 (92%)	38/40 (95%)
18 months	45/50 (90%)	38/40 (95%)	45/50 (90%)	38/40 (95%)
24 months	45/50 (90%)	37/40 (92.5%)	45/50 (90%)	37/40 (92.5%)

Abbreviations: BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation; NRS, Numerical Rating Scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index. Legend: Values represent observed outcome availability at each assessment time point and are presented as n/N (%). Missing observations were handled using mixed-effects modeling under the missing-at-random assumption.

Functional Outcomes

Observed longitudinal WOMAC outcomes are summarized in Table 2, whereas adjusted model-predicted functional trajectories derived from the mixed-effects analyses are illustrated in Figure 3. Both treatment groups demonstrated substantial improvement in functional status during early follow-up; the greatest reductions in WOMAC scores were observed within the first 3 to 6 months after treatment. Mean observed WOMAC scores in the CRFA group decreased from 78.2 ± 6.3 at baseline to 55.0 ± 6.0 at 1 month and 19.0 ± 8.0 at 3 months. Similarly, WOMAC scores in the CRFA plus BoNT-A group decreased from 77.4 ± 5.8 at baseline to 38.0 ± 5.0 at 1 month and 16.0 ± 5.0 at 3 months (Table 2).

Figure 3.

Observed and adjusted model-predicted Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) trajectories over 24 months according to treatment strategy. Solid lines represent mixed-effects model-predicted trajectories, and shaded areas represent 95% confidence intervals. Abbreviations: BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation; M, months.

Adjusted linear mixed-effects modeling demonstrated a significant overall longitudinal time effect for WOMAC outcomes (β = −18.6, SE = 2.4, 95% CI −23.3 to −13.9, P = 0.0001), indicating progressive improvement in functional outcomes across follow-up assessments. A significant treatment-by-time interaction was also observed (β = −6.3, SE = 1.6, 95% CI −9.4 to −3.2, P = 0.0003), suggesting differential longitudinal functional trajectories between treatment strategies during extended follow-up (Table 4).

Table 4.

Adjusted linear mixed-effects model estimates for longitudinal NRS and WOMAC outcomes.

Outcome	Fixed effect	β (SE)	95% CI	P
NRS model	Time	−2.84 (0.31)	−3.45 to −2.23	0.0001
	Treatment group (CRFA + BoNT-A vs CRFA alone)	−0.42 (0.28)	−0.97 to 0.13	0.13
	Treatment group × time	−0.84 (0.19)	−1.21 to −0.47	0.0002
	Age	0.02 (0.01)	−0.01 to 0.05	0.18
	Female sex	0.16 (0.22)	−0.27 to 0.59	0.46
	BMI	0.04 (0.03)	−0.02 to 0.10	0.17
	KL grade	0.29 (0.18)	−0.06 to 0.64	0.10
	Baseline NRS	0.48 (0.14)	0.21 to 0.75	0.001
WOMAC model	Time	−18.6 (2.4)	−23.3 to −13.9	0.0001
	Treatment group (CRFA + BoNT-A vs CRFA alone)	−4.1 (2.9)	−9.8 to 1.6	0.15
	Treatment group × time	−6.3 (1.6)	−9.4 to −3.2	0.0003
	Age	0.18 (0.09)	0.00 to 0.36	0.051
	Female sex	1.7 (2.3)	−2.8 to 6.2	0.45
	BMI	0.42 (0.28)	−0.13 to 0.97	0.13
	KL grade	2.8 (1.9)	−0.9 to 6.5	0.14
	Baseline WOMAC	0.56 (0.11)	0.34 to 0.78	0.0001

Abbreviations: β, regression coefficient; BMI, body mass index; BoNT-A, botulinum toxin type A; CI, confidence interval; CRFA, cooled genicular radiofrequency ablation; KL, Kellgren-Lawrence; NRS, Numerical Rating Scale; SE, standard error; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index. Legend: Models were adjusted for age, sex, BMI, KL grade, and corresponding baseline outcome measure (baseline NRS for NRS model and baseline WOMAC for WOMAC model). Mixed-effects models were estimated via restricted maximum likelihood with a first-order autoregressive [AR(1)] covariance structure and patient-level random intercepts. Negative treatment group × time interaction coefficients indicate greater sustained longitudinal improvement associated with the CRFA + BoNT-A group. Time was modeled as an ordinal follow-up variable representing the sequence of scheduled assessments across the 24-month follow-up period; reported time and treatment group × time coefficients reflect overall longitudinal trends, rather than visit-specific contrasts.

Exploratory assessment of longitudinal trajectories demonstrated similar early functional improvement in both treatment groups during the first 6 months, followed by progressive divergence after ~ 12 months. As noted concerning pain outcomes, the divergence observed after 12 months should be interpreted in the context of the predefined longitudinal treatment strategy.

At the predefined 24-month assessment, observed WOMAC outcome data were available for 45/50 patients in the CRFA group and 37/40 patients in the CRFA plus BoNT-A group (Table 3). Mean observed WOMAC scores at 24 months were 41.0 ± 20.0 in the CRFA group and 18.0 ± 10.0 in the CRFA plus BoNT-A group (Table 2). At 24 months, adjusted mixed-effects modeling estimated a WOMAC score ~ 23 points lower in the CRFA plus BoNT-A group than in the CRFA-alone group (95% CI −31.4 to −14.8, P = 0.0001). The magnitude of the observed between-group WOMAC difference exceeded previously reported minimal clinically important difference thresholds for functional outcomes in knee OA. Both the observed WOMAC scores and the adjusted model-predicted trajectories indicated more sustained functional improvement in the combination-treatment group during long-term follow-up (Figure 3).

Adjusted Mixed-Effects Model Results

Adjusted linear mixed-effects analyses demonstrated significant longitudinal treatment-associated differences in pain intensity and functional outcomes after controlling for predefined baseline covariates, including age, sex, body mass index, KL grade, baseline NRS score, and baseline WOMAC score.

For longitudinal NRS outcomes, the adjusted treatment-by-time interaction remained statistically significant (β = −0.84, SE = 0.19, 95% CI −1.21 to −0.47, P = 0.0002), suggesting differential longitudinal pain trajectories between treatment strategies during extended follow-up. Similarly, the adjusted WOMAC analyses demonstrated a significant treatment-by-time interaction (β = −6.3, SE = 1.6, 95% CI −9.4 to −3.2, P = 0.0003), indicating more sustained functional improvement in patients undergoing CRFA combined with intra-articular BoNT-A.

Responder Analysis

Responder outcomes are summarized in Table 5 and were calculated using observed-case denominators at each scheduled follow-up assessment, consistent with the longitudinal outcome availability presented in Table 3. Responder analysis evaluated the proportion of patients achieving at least 50% reduction in NRS pain scores relative to baseline, a threshold widely considered to represent clinically meaningful improvement in chronic pain studies. To improve interpretability, absolute between-group differences in responder rates (CRFA plus BoNT-A minus CRFA) with 95% CIs were also calculated at each follow-up assessment.

Table 5.

Patients achieving at least 50% reduction in NRS scores.

Time point	CRFA, n/N (%)	CRFA + BoNT-A, n/N (%)	Absolute between-group difference,% (95% CI)*	P
1 month	36/49 (73.5%)	34/40 (85.0%)	11.5 (−5.1 to 28.1)	0.19
3 months	44/48 (91.7%)	37/39 (94.9%)	3.2 (−7.2 to 13.6)	0.56
6 months	41/47 (87.2%)	36/39 (92.3%)	5.1 (−7.6 to 17.8)	0.44
12 months	32/46 (69.6%)	34/38 (89.5%)	19.9 (3.4 to 36.4)	0.027
18 months	26/45 (57.8%)	33/38 (86.8%)	29.1 (11.1 to 47.1)	0.004
24 months	23/45 (51.1%)	32/37 (86.5%)	35.4 (17.1 to 53.7)	0.0004

Abbreviations: BoNT-A, botulinum toxin type A; CI, confidence interval; CRFA, cooled genicular radiofrequency ablation; NRS, Numerical Rating Scale. Legend: Responders were defined as patients achieving at least 50% reduction in NRS pain scores relative to baseline. Values are presented as n/N (%).

^*Absolute between-group difference was calculated as CRFA + BoNT-A responder rate minus CRFA responder rate

Responder rates in both treatment groups were high during early follow-up, particularly within the first 3 to 6 months after treatment. At 3 months, responder rates were 44/48 (91.7%) in the CRFA group and 37/39 (94.9%) in the CRFA plus BoNT-A group, corresponding to an absolute between-group difference of 3.2% (95% CI −7.2% to 13.6%; P = 0.56). At 6 months, responder rates were 41/47 (87.2%) and 36/39 (92.3%), respectively, corresponding to an absolute between-group difference of 5.1% (95% CI −7.6% to 17.8%; P = 0.44).

During extended follow-up, responder rates progressively declined in the CRFA-alone group, whereas patients undergoing CRFA combined with intra-articular BoNT-A maintained substantially higher responder rates over time. At 12 months, responder rates were 32/46 (69.6%) in the CRFA group and 34/38 (89.5%) in the CRFA plus BoNT-A group, corresponding to an absolute between-group difference of 19.9% (95% CI 3.4% to 36.4%; P = 0.027). At 18 months, responder rates were 26/45 (57.8%) and 33/38 (86.8%), respectively, corresponding to an absolute between-group difference of 29.1% (95% CI 11.1% to 47.1%; P = 0.004). At the predefined 24-month follow-up assessment, responder rates were 23/45 (51.1%) in the CRFA group and 32/37 (86.5%) in the CRFA plus BoNT-A group, corresponding to an absolute between-group difference of 35.4% (95% CI 17.1% to 53.7%; P = 0.0004), indicating that a greater proportion of patients maintained clinically meaningful pain reduction during extended follow-up with the combination-treatment strategy.

As noted for the longitudinal pain and functional outcomes, the higher responder rates observed during extended follow-up should be interpreted in the context of the predefined longitudinal treatment strategy.

Overall Treatment-Associated Differences

Across the 24-month follow-up period, patients managed with the predefined CRFA plus BoNT-A longitudinal treatment strategy demonstrated consistently lower observed pain and WOMAC scores relative to those of patients managed with CRFA alone. Separation between the longitudinal outcome trajectories became increasingly apparent after ~ 12 months and remained evident through the final follow-up assessment (Figures 2, 3).

Because treatment selection reflected individualized clinical management within a retrospective observational framework and the combination-treatment strategy incorporated planned repeat intra-articular BoNT-A administration at ~ 12 months, these findings should be interpreted as observational longitudinal associations rather than definitive evidence of independent treatment-specific causal effects.

Model-Predicted Outcome Trajectories

Adjusted model-predicted trajectories derived from the fully adjusted mixed-effects analyses demonstrated clear divergence between treatment groups during long-term follow-up. Predicted NRS trajectories showed progressive separation beginning at ~ 12 months; the CRFA plus BoNT-A group maintained consistently lower predicted pain scores through 24 months (Figure 2). Similarly, predicted WOMAC trajectories demonstrated more sustained functional improvement in the combination-treatment group, with persistent separation between treatment groups during extended follow-up (Figure 3). Corresponding 95% CIs supported statistically significant separation between treatment groups during extended follow-up.

Individual Trajectory Visualization

Figure 4 presents individual patient-level NRS trajectories across the 7 scheduled follow-up assessments. Each thin line represents an individual patient trajectory, whereas thicker lines represent group-level smoothed mean trajectories. The plots demonstrated substantial early improvement in both treatment groups, followed by divergent long-term patterns. Patients treated with CRFA alone exhibited gradual pain recurrence after ~ 12 months; patients undergoing CRFA combined with intra-articular BoNT-A generally maintained lower pain scores during extended follow-up. These patient-level trajectory plots visually supported the findings derived from adjusted mixed-effects analyses.

Figure 4.

Individual patient-level Numerical Rating Scale (NRS) trajectories over 24 months according to treatment strategy. Thin lines represent individual patient trajectories, and thicker lines represent group-level mean trajectories. Abbreviations: BoNT-A, botulinum toxin type A; CRFA, cooled genicular radiofrequency ablation; M, months.

Discussion

Principal Findings

In this retrospective comparative-effectiveness observational cohort study of 90 patients with moderate-to-severe knee OA, both CRFA alone and CRFA combined with intra-articular BoNT-A were associated with substantial early reductions in pain intensity and functional disability. However, important longitudinal differences in treatment response maintenance emerged during extended follow-up.

The predefined primary durability endpoint was the between-group difference in longitudinal NRS pain trajectories over the 24-month follow-up period, assessed using the treatment-by-time interaction in the adjusted mixed-effects model. Adjusted mixed-effects analyses demonstrated significant overall longitudinal improvement over time for both NRS outcomes (β = −2.84, SE = 0.31, 95% CI −3.45 to −2.23, P = 0.0001) and WOMAC outcomes (β = −18.6, SE = 2.4, 95% CI −23.3 to −13.9, P = 0.0001), indicating clinically meaningful improvement after genicular nerve CRFA under routine clinical practice conditions. Significant treatment-by-time interaction terms were observed for both NRS outcomes (β = −0.84, SE = 0.19, 95% CI −1.21 to −0.47, P = 0.0002) and WOMAC outcomes (β = −6.3, SE = 1.6, 95% CI −9.4 to −3.2, P = 0.0003), suggesting differential longitudinal outcome trajectories between treatment strategies over time. Complete adjusted model outputs are summarized in Table 4.

Both treatment groups demonstrated substantial improvement during the first 3 to 6 months after treatment. Throughout the 24-month follow-up period, patients managed with the predefined CRFA plus BoNT-A longitudinal treatment strategy demonstrated consistently lower observed pain and WOMAC scores than patients treated with CRFA alone; separation between trajectories become increasingly apparent after ~ 12 months and remained evident through the final follow-up assessment. At 24 months, mean observed NRS scores were 4.2 ± 2.3 in the CRFA group and 1.8 ± 0.9 in the CRFA plus BoNT-A group, whereas mean observed WOMAC scores were 41.0 ± 20.0 and 18.0 ± 10.0, respectively (Table 2). To facilitate clinical interpretation of long-term outcome differences, model-derived adjusted between-group contrasts were also estimated at the 24-month assessment. These estimations demonstrated approximate between-group differences of −2.4 points for NRS scores and −23 points for WOMAC scores (95% CI −31.4 to −14.8, P = 0.0001), favoring the combination-treatment strategy. Such differences exceeded previously reported minimal clinically important difference thresholds for chronic pain and functional outcomes in knee OA [27,28], supporting the clinical relevance of the observed long-term separation.

Responder analyses reinforced these findings. Although responder rates were high in both groups during early follow-up, the proportion of patients achieving clinically meaningful pain reduction progressively declined in the CRFA-alone group during extended follow-up, whereas patients managed with the CRFA plus BoNT-A treatment strategy maintained substantially higher responder rates over time. At the predefined 24-month assessment, responder rates were 23/45 (51.1%) in the CRFA group and 32/37 (86.5%) in the CRFA plus BoNT-A group (P = 0.0004). The consistency of findings across the observed longitudinal outcomes, responder analyses, adjusted mixed-effects models, and trajectory visualizations strengthens confidence in the overall pattern of treatment-associated differences observed during follow-up.

The present findings should be interpreted within the context of the predefined longitudinal multimodal treatment pathway. Patients in the combination group received both baseline intra-articular BoNT-A and a planned repeat BoNT-A injection at ~ 12 months; accordingly, the post-12-month divergence between groups should not be interpreted as the isolated long-term effect of the baseline adjunctive injection alone. Additionally, because treatment selection reflected individualized clinician-guided decision-making during routine interventional pain practice, residual confounding by indication could not be fully excluded despite the adjusted longitudinal analyses. Patients managed with adjunctive intra-articular BoNT-A may have differed in unmeasured prognostic characteristics, symptom patterns, psychosocial burden, rehabilitation adherence, physician-specific treatment philosophy, or anticipated treatment needs that were not fully captured in the retrospective dataset. Concomitant therapies received during follow-up may also have influenced outcomes because medication regimens, rehabilitation participation, psychosocial support, and supportive noninterventional treatments were not uniformly documented or standardized throughout the study period.

Mechanistic Interpretation

The observed differences between treatment strategies are biologically plausible and may reflect the complementary mechanisms of CRFA and intra-articular BoNT-A. The greater reduction in pain observed at the 1-month assessment in the CRFA plus BoNT-A group may reasonably reflect the early intra-articular antinociceptive and anti-inflammatory effects of BoNT-A in addition to CRFA-mediated denervation. Experimental and clinical studies have demonstrated that BoNT-A inhibits the release of pain-related neuropeptides and inflammatory mediators involved in peripheral sensitization and neurogenic inflammation [14–16,29,30]; previous clinical studies evaluating intra-articular BoNT-A in knee OA have documented early improvements in pain and functional outcomes after injection therapy [17–21,25,26,31–33]. In contrast, CRFA produces analgesia via thermal ablation of the articular sensory branches of the genicular nerves, thus interrupting nociceptive transmission from the osteoarthritic knee joint [6–10,34,35]. Multiple randomized and prospective studies have demonstrated that CRFA significantly improves pain and function during early follow-up [7–12,22–24].

The durability of denervation-based interventions may gradually diminish due to peripheral nerve regeneration or restoration of nociceptive signaling [10–13,35–37]. The progressive increases in pain intensity and functional impairment observed in the CRFA-alone group during later follow-up are consistent with previously reported long-term outcome patterns after radiofrequency ablation procedures. In contrast, BoNT-A primarily acts by modulating neurochemical pain pathways, rather than through direct neural destruction. Experimental evidence suggests that BoNT-A inhibits the release of nociceptive mediators (eg, substance P, glutamate, and calcitonin gene-related peptide), thereby reducing peripheral sensitization and neurogenic inflammation [14–16,29]. Because OA pain arises from a combination of structural degeneration, inflammatory activation, and nociceptive sensitization, a multimodal intervention targeting both neural transmission and intra-articular neurochemical signaling may plausibly provide more sustained symptom control.

Nevertheless, mechanistic interpretation should remain cautious. The present study was not designed to establish biological synergy; the proposed explanations should therefore be regarded as hypothesis-generating, rather than confirmatory. Because the combination-treatment strategy incorporated both baseline and planned repeat intra-articular BoNT-A administration at ~ 12 months, the relative contributions of each treatment component to the observed long-term outcome patterns cannot be determined. Accordingly, the findings should be interpreted as observational comparative-effectiveness associations observed during routine clinical practice, not definitive evidence of independent treatment-specific causal effects.

Functional Outcomes

Functional outcomes measured using the WOMAC closely paralleled the longitudinal pain trajectories. Both treatment groups demonstrated substantial early reductions in WOMAC scores, reflecting improved joint function and reduced disability during early follow-up, consistent with previously published studies of radiofrequency ablation for knee OA [8–12,34,38]. Adjusted mixed-effects analyses demonstrated a significant treatment-by-time interaction for WOMAC outcomes (β = −6.3, SE = 1.6, 95% CI −9.4 to −3.2, P = 0.0003), indicating more sustained functional improvement in the CRFA plus BoNT-A group during extended follow-up. By 24 months, WOMAC scores in the CRFA-alone group had progressively increased toward baseline values; patients undergoing the combination-treatment strategy maintained substantially lower functional disability scores.

Because the WOMAC is a validated instrument for assessing outcomes in knee OA [27] and previously reported minimal clinically important difference thresholds generally range from ~ 9 to ~ 12 points [28], the magnitude of functional separation observed in this study suggests clinically meaningful preservation of function associated with the combination-treatment approach. Moreover, WOMAC improvement likely reflects multiple interacting factors beyond analgesia alone, including rehabilitation participation, physical activity, body weight, comorbidities, and patient motivation. Because rehabilitation exposure was not standardized or systematically documented in this retrospective study, its contribution to functional outcomes cannot be determined.

Comparison With Previous Literature

The present findings are broadly consistent with the growing body of literature supporting the effectiveness of CRFA for symptomatic knee OA. Randomized trials and prospective cohort studies have demonstrated that CRFA can provide significant improvements in pain and functional outcomes for up to 12 months after treatment [8–12,34,38]. However, several longer-term investigations have noted variable maintenance of treatment response beyond 12 to 18 months, with gradual recurrence of symptoms in a subset of patients [10–13,39]. The trajectory pattern observed in the present CRFA-alone cohort is consistent with these previously reported long-term observations.

Evidence supporting intra-articular BoNT-A in knee OA has also expanded in recent years. Randomized controlled trials, prospective studies, and systematic reviews have demonstrated that intra-articular BoNT-A may reduce pain and improve functional outcomes in patients with refractory knee OA and other chronic joint pain conditions [17–21,25,26,30–33,40,41]. Nevertheless, most previous investigations have evaluated BoNT-A as a standalone intra-articular therapy; radiofrequency ablation studies have primarily focused on denervation-based interventions. Consequently, evidence regarding combined peripheral nerve ablation and intra-articular neuromodulatory treatment approaches remains limited.

Most previous CRFA investigations have evaluated outcomes after a single interventional treatment episode, whereas the present study evaluated a predefined longitudinal multimodal treatment strategy incorporating planned repeat intra-articular BoNT-A administration as part of routine clinical practice. Accordingly, this study extends the existing literature by providing comparative-effectiveness observations regarding long-term symptom trajectories within a real-world management framework and contributes additional longitudinal evidence concerning a multimodal interventional strategy that comprised peripheral nerve denervation plus intra-articular neurochemical modulation over an extended 24-month follow-up period.

Clinical Implications

From a practical perspective, longitudinal multimodal minimally invasive treatment pathways may offer clinical utility for patients with refractory knee OA who are poor surgical candidates or who wish to delay arthroplasty. More sustained symptom control may reduce symptom recurrence and the need for additional interventions; however, these outcomes were not directly evaluated in the present study. Repeated minimally invasive interventions may also introduce additional procedural and healthcare resource considerations. Formal prospective cost-effectiveness and healthcare utilization studies incorporating procedural costs, medication burden, repeat intervention frequency, and delayed arthroplasty outcomes are therefore warranted.

CRFA is increasingly recognized as an effective minimally invasive treatment for chronic knee OA pain [22–24,35–37]. However, analgesia durability remains an important clinical challenge in some patients. The present findings suggest that adjunctive intra-articular BoNT-A, administered as part of a predefined multimodal treatment strategy, is associated with more sustained analgesic and functional improvement after CRFA. The minimally invasive nature of the combination-treatment approach may be particularly relevant for older adults with medical comorbidities who are poor surgical candidates or who wish to delay arthroplasty while preserving mobility and symptom control.

No major procedure-related complications were identified in either treatment group during the available 24-month follow-up period. However, because safety ascertainment relied on a retrospective review of routine clinical documentation rather than prospective standardized adverse-event surveillance, minor transient adverse events may not have been consistently captured.

Strengths

This study has several notable strengths. First, it evaluated outcomes over a 24-month follow-up period, exceeding the duration reported in many previous CRFA studies [10–13,39]. Extended follow-up is particularly important when evaluating denervation-based interventions because symptom recurrence may emerge beyond the first posttreatment year. Inclusion of all consecutively identified eligible patients also reduced the likelihood of selection bias and enhanced the real-world applicability of the findings.

Second, longitudinal outcome availability remained high throughout follow-up—exceeding 90% at most scheduled assessments—supporting the robustness of the longitudinal analyses. Third, the study utilized adjusted linear mixed-effects modeling, which appropriately adjusted for repeated within-participant observations and permitted inclusion of partially incomplete longitudinal datasets under the missing-at-random assumption. The models incorporated clinically relevant baseline covariates, including age, sex, body mass index, KL grade, and baseline outcome measures, thus reducing the potential influence of measured confounding. Fourth, multiple complementary analytical approaches were used, including observed longitudinal outcome analyses, responder analyses, adjusted model-predicted trajectories, and individual patient-level trajectory visualizations. The consistency of findings across these approaches strengthened confidence in the internal coherence of the results. Finally, the study provides real-world comparative-effectiveness observations regarding a multimodal interventional approach that has received limited evaluation in existing literature.

Limitations

Several limitations should be considered when interpreting these findings. First, the retrospective observational design introduces the possibility of selection bias and residual unmeasured confounding. Because treatment selection reflected individualized clinical management and clinician-guided decision-making rather than randomization, confounding by indication cannot be fully excluded despite covariate-adjusted longitudinal analyses. Although the adjusted mixed-effects models incorporated predefined baseline covariates (eg, age, sex, body mass index, KL grade, and baseline symptom severity measures), unmeasured differences in clinical presentation, disease chronicity, psychosocial factors, rehabilitation adherence, physician-specific treatment preferences, and anticipated treatment needs may have influenced treatment selection and longitudinal outcomes.

Second, the study evaluated comparative treatment strategies implemented under routine clinical care, rather than within a controlled experimental framework. Accordingly, the findings should be interpreted as observational comparative-effectiveness associations (not definitive evidence of independent treatment-specific causal effects). Third, the study was conducted at a single interventional pain center, which may limit generalizability of the findings to other institutions, procedural settings, and patient populations.

Fourth, the combination-treatment strategy incorporated planned repeat intra-articular BoNT-A administration at ~ 12 months, whereas patients managed with CRFA alone did not undergo a routinely scheduled repeat intervention. Consequently, post-12-month differences in pain and functional trajectories reflect the overall multimodal treatment strategy, rather than the isolated effects of the baseline intervention; the relative contributions of the baseline and repeat BoNT-A administrations cannot be determined from the present study. Fifth, the absence of a BoNT-A-only comparator limits assessment of the independent contribution of adjunctive intra-articular BoNT-A and precludes formal evaluation of additive versus synergistic effects.

Sixth, mixed-effects modeling permitted inclusion of participants with partially incomplete repeated measurements under the missing-at-random assumption, but the potential influence of missing data on the longitudinal outcome estimates cannot be entirely excluded. Seventh, formal propensity score matching or inverse probability weighting was not performed due to the modest retrospective sample size and the limited number of uniformly available baseline covariates for robust propensity model specification. Additionally, the absence of a prospective sample size calculation may have hindered the ability to detect smaller between-strategy differences or uncommon clinical outcomes.

Eighth, medication use, rehabilitation participation, and other supportive noninterventional treatments were not prospectively standardized and could not be comprehensively tracked throughout the 24-month follow-up period. Information regarding rehabilitation adherence, changes in concomitant management, comorbidity burden, symptom duration beyond the eligibility threshold, and complete medication history was not consistently available in the retrospective records and thus could not be incorporated into the adjusted analyses. Consequently, their potential effects on treatment selection and longitudinal pain and functional outcomes cannot be fully excluded.

These collective limitations should be considered when interpreting the findings. Future prospective randomized studies with standardized follow-up protocols, comprehensive assessment of concomitant therapies, and appropriate comparator groups are warranted to further clarify the long-term comparative effectiveness of combined CRFA and intra-articular BoNT-A treatment strategies.

Conclusions

In this retrospective comparative observational study, a longitudinal treatment pathway consisting of CRFA combined with intra-articular BoNT-A, including planned repeat BoNT-A administration at ~ 12 months, was associated with more sustained pain relief and functional improvement over 24 months relative to CRFA alone. Given the observational design and potential for residual confounding, prospective randomized studies are needed to confirm these findings, clarify the independent contributions of adjunctive and repeat intra-articular BoNT-A, and define its role within multimodal treatment strategies for knee OA.

Data Availability

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.