Article Text
PDF
Abstract
Background and objectives Refractory chronic knee pain from osteoarthritis (OA) is commonly treated with total knee arthroplasty (TKA). TKA can be associated with severe postoperative pain and persistent postsurgical knee pain. Poorly controlled postoperative pain can negatively effect functional outcomes following TKA, and effective opioid-sparing analgesia is key to the ideal recovery. Genicular nerve radiofrequency ablation (GN-RFA) has been shown in several trials to be clinically effective in patients with severe refractory knee pain from OA. We aimed to assess if preoperative GN-RFA would improve postoperative pain outcomes following TKA.
Methods This was a sham-control prospective clinical trial in which blinded participants were randomized to image-guided GN-RFA or a simulated sham procedure 2-6 weeks prior to elective TKA. Outcomes were assessed at 48 hours and 1, 3 and 6 months following TKA.
Results Seventy participants enrolled in this study. As compared with sham controls, GN-RFA had no treatment effect on postoperative opioid consumption, pain or functional measures at any time point.
Conclusions Cooled RFA of the superior lateral, superior medial and inferomedial genicular nerves, when performed 2–6 weeks prior to elective TKA as part of a multimodal postoperative pain management regime, had no measurable effect on postoperative opioid use, analgesia use or function in the 48 hours following surgery. In addition, we found no longer term effect on outcome measures 1, 3 and 6 months after TKA.
Trial registration number NCT02746874.
- radiofrequency ablation
- knee osteoarthritis
- knee pain
- genicular nerve ablation
Statistics from Altmetric.com
Introduction
Chronic knee pain from osteoarthritis (OA) is commonly treated with total knee arthroplasty (TKA) when conservative therapies fail to provide pain relief.1 Although TKA is successful in reducing knee pain and stiffness in most cases, it can be associated with severe acute postoperative pain2 and a 7%–35% incidence of persistent postsurgical knee pain.3–6 High-intensity acute postoperative pain, in addition to catastrophization and anxiety, is an independent risk factor for persistent postoperative pain.5 7 8 Poorly controlled postoperative pain inhibits early knee mobilization and hinders physical therapy that are critical to prevent delays or limitations in the ultimate functional outcome following TKA. Effective analgesia is known to enhance functional recovery, improve joint mobility and decrease morbidity following TKA, and therefore, effective multimodal, opioid-sparing analgesia regimens are recommended in these patients.6 9–12
Genicular nerve radiofrequency ablation (GN-RFA) has been shown to be effective in reducing knee pain from OA in several randomized trials.13–15 There are also reports of substantial pain relief when GN-RFA is performed for the treatment of chronic knee pain following an otherwise uncomplicated TKA.16–18 GN-RFA disrupts articular sensory afferent pathways from the knee joint to the central nervous system, potentially decreasing postoperative nociceptive signaling in those who undergo TKA. This could, in turn, decrease overall postoperative pain levels or decrease dependence on other analgesic methods, including pan management with opioids. To date, preoperative GN-RFA and its effect on postoperative pain outcomes following TKA has not been investigated.
In this study, we aimed to assess the effects of preoperative genicular nerve cooled RFA on postoperative analgesia consumption and functional outcomes when combined with standardized multimodal pain management in patients undergoing elective TKA. We tested the hypothesis that opioid consumption (oral morphine equivalents (MEs)) for the first 48 hours after surgery would be less in patients who had undergone GN-RFA before TKA than in patients who had undergone a sham radiofrequency (RF) ablation.
Methods
This study was a prospective, randomized, sham-controlled trial with 6-month postoperative follow-up. It was registered on ClinicalTrials.gov. The manuscript adheres to the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Inclusion criteria were: (1) age between 30 years and 80 years; (2) planned elective primary unilateral TKA for knee pain from OA; and (3) willingness and consent to undergo the fluoroscopy-guided study intervention. Exclusion criteria were: (1) medical conditions that precluded the study intervention (eg, severe cardiac or pulmonary compromise; acute illness or infection; coagulopathy or bleeding disorder; allergic reactions; contraindications to local anesthetic; and pregnancy); (2) prior GN-RFA; (3) prior TKA; (4) TKA performed outside of the period of 2–6 weeks following the study intervention; (5) inability to write, speak or read in English; and (6) and lack of consent.
Participants were made aware of the study by their surgeon (DM, MB or KH) or were contacted by research personnel regarding study participation once surgery was scheduled. Participants underwent an informed consent process including provision of written consent before any study procedure was performed.
On the day of the study intervention, a computer-generated 1:1 block randomization scheme was used to assign participants to receive either GN-RFA or a sham procedure (simulated GN-RFA using identical supplies and devices). Randomization was performed by the interventionalist (DW) by opening an envelope to reveal the participant number and group assignment printed on an index card.
Participants were positioned supine on a fluoroscopy table with a bolster under the knee to provide 30° of knee flexion. The knee was prepped with chlorhexidine and draped in a sterile fashion. A grounding pad was placed on the contralateral thigh and connected to an RF generator (Halyard Health, Pain Management Radiofrequency Generator, Alpharetta, Georgia, USA), positioned out of the participant’s view.
Regardless of group assignment, participants were provided conscious sedation (intravenous midazolam in aliquots 0.5–1 mg as needed, maximum 6 mg; intravenous fentanyl 25–50 µg aliquots as needed, maximum 200 µg). Using fluoroscopic image guidance, a 17-gage 75–100 mm introducer cannula (Coolief Kit, Halyard Health) was placed at three unique anatomic sites after 2–3 mL of 1% lidocaine was injected in the skin and subcutaneous tissues superficial to the superior lateral, the superior medial and the inferior medial genicular nerves.
The superior lateral genicular nerve site is located at the confluence of the lateral femoral shaft and condyle in the anteroposterior (A-P) plane and at the midpoint of the femur in the lateral plane. The superior medial genicular nerve site is located at the confluence of the medial femoral shaft and the condyle at the midpoint of the femur in the lateral plane. The inferior medial genicular nerve site is located at the confluence of the medial tibial shaft and the tibial flare in the A-P plane and the midpoint of the tibia in the lateral plane.19
An internally cooled, disposable 18-gage RF electrode was then placed through the introducer cannula and stimulation at 2 Hz up to 2.0 volts confirmed safe distance from motor fibers at all three treatment sites. Sensory testing was not performed given the depth of conscious sedation provided. Motor testing was followed by the injection of 2% lidocaine 2 mL at each site. In the treatment group, each site was lesioned for 150 s with estimated tissue temperatures 77–80°C.20 In the sham group, simulated lesioning was performed at each level for 150 s, providing identical auditory and visual cues as the treatment group but without any RF current. Cannulas/electrodes were then removed, and the cannula sites were bandaged. Participants were transferred to a recovery area and monitored for at least 30 min.
All study interventions were performed by a single interventionalist (DW) with 18 years of clinical experience who had performed GN-RFA on more than 200 patients prior to this study. Surgeons, anesthetists and providers involved in patient care during and following the TKA, as well as research personnel tasked with participant follow-up were blinded to the group assignment.
Within 2–6 weeks following the study intervention, participants underwent elective TKA performed by one of three surgeons at a single urban tertiary medical center. Unless contraindicated, 1 hour preoperatively, patients were given a single dose of extended release oxycodone 10 mg, acetaminophen 650 mg and celecoxib 400 mg, per surgeons’ institutional practice, and received spinal anesthesia consisting of 2.5 mL of 0.5% bupivacaine or general anesthesia with a balanced inhalational technique. Quadriceps-sparing TKA via a limited arthrotomy using customized instrumentation was performed in all cases. Before closure of the arthrotomy, patients received a 100 mL local infiltration that included 30 mL 0.5% bupivacaine with 1:200 000 epinephrine, morphine 5–10 mg (depending on age and medical frailty) and ketorolac 15 mg as part of a standard multimodal analgesia regimen used at this institution.10 All non-diabetic patients received dexamethasone 4 mg during skin closure.
Once discharged from the postanesthesia care unit, patients were given acetaminophen 1 g every 8 hours, tramadol 50 mg every 8 hours and ketorolac 15 mg intravenous every 12 hours for 48 hours. Patients could refuse their medications. Furthermore, they could request hydromorphone 2 mg by mouth every 2 hours as needed for rescue analgesia.
At hospital discharge, participants were prescribed extended release oxycodone 10 mg twice daily, hydrocodone/acetaminophen 5–10 mg/325 mg every 4 hours as needed; they were counseled to use as needed analgesics for Numerical Rating Scale >4. Oxycodone was tapered within 4 weeks after surgery.
A standardized physical therapy protocol was initiated on the day of surgery or on postoperative day 1 if patients were transferred from the postanesthesia care unit after 16:00. Baseline measures were collected on the day of the study intervention and at the bedside 48 hours after TKA. At 1, 3 and 6 months following surgery, participants were contacted by telephone to collect outcome measures.
The primary outcome of this study was opioid consumption (oral MEs) for the first 48 hours after surgery. Secondary outcomes included the Medication Quantification Scale III (MQSIII)21 score at 48 hours, the number of stairs climbed and distance ambulated on postoperative day 2 prior to discharge, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores, Hospital Anxiety and Depression Inventory scores (HADS), 12-Item Short Form Health Survey (SF-12) scores, McGill Pain Questionnaire scores and ME and MQSIII at 1, 3 and 6 months after surgery.
Statistical power
The power analysis for this study was guided by prior publications on preoperative interventions for TKA postoperative analgesia including assumptions regarding perioperative opioid requirements in controls undergoing TKA.10–12 With a sample size of 56 (28 per treatment group), a difference in analgesic use of 15 MEs per 24-hour cycle, assuming the controls use 47 ME per 24-hour cycle with an SD of 19 ME with power of 0.81 at alpha=0.05, could be detected. The sample size calculation was computed using a two-sided Mann-Whitney test (Power Analysis and Sample Size software (PASS)). To correct for any incorrect assumptions and anticipated loss to follow-up, we enrolled a total of 70 subjects for this study.
Statistical analysis
Variables that were characterized by nominal data (eg, reduction in MQSIII scores by at least 3.4) are summarized as the number of patients and the percentage of all the patients in the group that they represent. These variables were compared between the randomized groups using the risk ratio and its 95% CI (StatsDirect, Cambridge, UK). Variables that were characterized by ordinal data and non-normally distributed continuous data (eg, ME at 48 hours) are summarized as median and IQR. These variables were compared between the randomized groups using the Mann-Whitney U test and within the randomized groups using Wilcoxon’s signed ranks test (StatsDirect). Median differences and their 95% CIs were calculated for the planned analyses, but 99% CIs of the median differences were determined for the post hoc analyses because of the large number of unplanned comparisons that were made. Variables that were characterized by normally distributed continuous data (eg, MQSIII scores at 48 hours) are summarized as mean and SD. These variables were compared between the randomized groups using the unpaired Student’s t-test (StatsDirect) or the two-way repeated measures analysis of variance (one factor repetition) with pairwise multiple comparison procedures made using the Holm-Sidak method when indicated by the results of the analysis of variance (SigmaPlot for Windows, Systat Software, San Jose, California, USA). Mean differences and their 95% CIs were determined for the planned analyses, but 99% CIs of the mean differences were determined for the post hoc analyses because of the large number of unplanned comparisons that were made. The criterion for rejection of the null hypothesis was a two-tailed p<0.05 for all planned comparisons and a two-tailed p<0.01 for all post hoc comparisons.
Results
A CONSORT diagram22 (figure 1) demonstrates the participant flow through the study. Of 208 patients screened, 70 met inclusion criteria and provided consent to participate. One participant withdrew consent on the day of the study intervention. Two participants postponed TKA surgeries for several months outside the period of 2–6 weeks following the study intervention due to medical contraindications. All three withdrawals had been allocated to the sham group. Enrollment took place between December 2016 and June 2017. There were no adverse events related to the study intervention.
CONSORT flow diagram. Genicular radiofrequency ablation prior to TKA. CONSORT, Consolidated Standards of Reporting Trials; TKA, total knee arthroplasty.
The characteristics of the patients in the two groups were similar (table 1).
Patient characteristics
The majority of participants were white, female and obese (BMI >30). Six of the control participants used an opioid preoperatively (range 10–60 ME) as did four participants in the treatment group (range 10–70 ME). There was no significant difference in the primary outcome, ME use at 48 hours, between the control group (144 ME, IQR 112–314) and the treatment group (192 ME, IQR 105–274) (table 2).
Oral morphine equivalents at baseline and 48 hours postoperatively
There were also no group differences in the incidence of postoperative analgesia side effects (sedation, nausea, pruritus and delirium) in the first 48 hours after TKA (29.6% in control group, 24.2% in the treatment group, p=0.086) or in MQSIII scores at 48 hours (table 3).
Medication Quantification Scale version III (MQSIII) scores at baseline and 48 hours postoperatively
Physical functioning on postoperative day 2 as measured by distance ambulated and number of stairs climbed did not differ between groups (table 4).
Physical function characteristics on postoperative day 2
Similarly, there were no group differences in any outcome measure at 1, 3 or 6 months after TKA. There was a very low incidence of opioid use in both treatment groups at 6 months, and the percent with a significant reduction in MQSIII score did not differ between treatment groups (table 5).
Medication Quantification Scale version III (MQSIII) scores and morphine equivalents at 6 months
Six-month WOMAC scores were similar between groups (table 6).
Western Ontario and McMaster Universities Arthritis Index (WOMAC) at 6 months
To confirm that anxiety and depression, which are known to have a negative impact on postoperative outcomes5 did not confound our results, outcomes were reanalyzed including only those participants with a baseline HADS score <11. This post hoc analysis also found no significant group differences in ME or MQSIII, or WOMAC scores (online supplementary table 7,8).
Discussion
The most important finding of the current study is that cooled RFA of the superior lateral, superior medial and inferomedial genicular nerves, when performed 2–6 weeks prior to elective TKA as part of a multimodal postoperative pain management regime, had no measurable effect on postoperative opioid use, analgesia use or function in the 48 hours following surgery. In addition, we found no longer term effect on outcome measures 1, 3 and 6 months after TKA.
Given the profound treatment effects of GN-RFA in prior controlled trials of patients with severe knee pain from OA,13–15 this lack of treatment effect is unexpected. In the most recent published trial of GN-RFA, Davis et al studied 151 patients with chronic knee pain from OA in a prospective randomized trial in which subjects underwent either cooled GN-RFA or an intra-articular knee joint injection with steroid. The cooled GN-RFA group had more favorable outcomes at the 6-month follow-up including superior pain scores, improved Oxford Knee Scores and superior Global Perceived Effect but had no difference in opioid use. Also at 6 months, 74.1% of those in the cooled GN-RFA group achieved >50% reduction in pain scores, whereas only 16.2% of those in the intra-articular steroid injection group achieved this outcome. Forty per cent of those in the cooled GN-RFA group reported ‘satisfactory knee function’ as compared with 3% in the intra-articular steroid group.14 Choi et al studied conventional GN-RFA in elderly patients with chronic knee pain. Those who underwent conventional GN-RFA reported at least a 50% decrease in pain scores at 1 and 3 months following the procedure, whereas the sham control group had no demonstrable change in pain scores at these time points. Oxford Knee Scores and patient satisfaction scores were significantly improved in the GN-RFA group but unchanged in the sham control group.13
Our findings contrast with the long-term clinical success seen in these trials and are likely due to multiple factors. First, the non-responder rate for cooled GN-RFA is at least 26%, even when prognostic genicular nerve blocks are performed.14 15 23 This is likely related to the complexity of knee innervation and the significant patient-to-patient neuroanatomic variability noted in recent literature.24 25 Our protocol did not require a prognostic block prior to the GN-RFA that may have increased the non-responder rate or created misclassification bias. However, prognostic nerve blockade may not effect or predict GN-RFA treatment outcomes.15
Second, newer cadaveric dissection studies of the genicular nerves24 25 call into question the anatomic targets described by Franco et al, 19 which are the original basis for cooled GN-RFA and the foundation of its technical execution in this study.
Furthermore, the recommended anatomic targets for GN-RFA13 14 19 result in only partial denervation of the knee joint, which may be inadequate for meaningful blockade of nociception after TKA.13 19 In fact, many patients who undergo TKA report posterior knee pain as the location of their initial perceived pain after surgery.12 The genicular nerves that supply the posterior knee are in close proximity to the tibial nerve and the arterial vasculature in the popliteal fossa,26 a region that is purposely avoided to prevent iatrogenic thermal injury to motor nerve fibers and vascular structures.27 Similarly, the anterolateral knee is supplied by the recurrent common peroneal nerve, which is in close proximity to the common peroneal nerve and is also avoided. Therefore, the lack of treatment effects we identified in this study could be related to our inability to safely lesion the genicular nerves most responsible for pain following TKA.
Lastly, consensus agreement on the timing and duration of pain relief after GN-RFA is lacking. We selected the time period of 2–6 weeks prior to TKA to perform the study intervention based on our extensive clinical experience performing cooled GN-RFA on hundreds of patients, wherein most patients report an analgesic effect within 2 weeks of cooled GN-RFA. Furthermore, this time period of 2–6 weeks is consistent with the onset of analgesia of GN-RFA in other published clinical trials,13 14 17 but it is not clear if this duration was adequate to assess a clear treatment effect on the primary outcome of this study at 48 hours after TKA.
No previous study has investigated the effect of GN-RFA on postoperative outcomes in TKA. Although this was a well powered, blinded, randomized sham control trial, our findings must be interpreted within the context of the study's limitations.
In order to maintain successful enrollment and cooperation from our orthopedic surgeons and study population, study participants were managed with a multimodal pain management protocol, standard to those undergoing TKA at our institution.6 10 This protocol includes articular injections of local anesthetic and adjuvant analgesics that are known to provide 6–12 hours of joint analgesia.28 In theory, this practice may have confounded ME and MQSIII scores in the first 48 hours after TKA. The ME in the first 48 hours after surgery were extracted from the medical record and are deemed to be accurate. However, follow-up medication use after hospital discharge was self-reported and could not be verified.
We also acknowledge that although patients could exercise the option to refuse analgesics or to request additional opioid ‘rescue’ analgesia every 2 hours in the immediate postoperative period, we do not know if patients were influenced by providers, family members or others to request less (or more) opioids. In addition, the nearly complete opioid tapering in all patients by 6 months after TKA may have been influenced by the practice habits of the orthopedic surgeons prescribing these medications after surgery, not patient preference. Thus, participants may have lacked the agency to increase or continue their opioid use after hospital discharge or later in their postoperative recovery if pain was severe but opioids were discontinued by the provider. Indeed, it is possible that during the 2016–2018 conduct of this study, increased patient and provider awareness of the opioid abuse epidemic in the USA, as well as fears of opioid addiction and increased pressures on physicians to be better stewards of opioid prescriptions, together may have influenced postoperative opioid use29 and ultimately influenced the primary outcome selected for this study.
Nonetheless, even if ME levels and MQSIII scores were confounded by the aforementioned factors, the lack of treatment effect on knee pain, knee stiffness and overall knee function (WOMAC subscores and total scores) does not appear to be confounded. Although we did not observe a treatment effect with GN-RFA in this study population, we also did not see a significant incidence of chronic pain after TKA in our study sample. Future studies should investigate the effects of GN-RFA on patients who do experience significant chronic knee pain following TKA, as the incidence of persistent postsurgical knee pain after this surgery can exceed 30%.
Supplemental material
Supplemental material
Acknowledgments
We are grateful for the contributions of Ms. Trista Reynolds and Ms. Suzanne Banuvar who were research coordinators for this study appreciate Dr. Kevin Hardt and Dr. Matthew Beal involvement in participant recruitment for this study.
References
Footnotes
Presented at Interim data from this work were presented at the American Society of Regional Anesthesia and Pain Medicine in New York City, 19–22 April 2018 and the World Institute of Pain Meeting in Dublin, Ireland, 9–12 May 2018.
Contributors DW was instrumental in the design, plan and conduct of this study; he lead the team who acquired the data, interpreted results and wrote the manuscript. DW received investigator-initiated research support (funding and disposable equipment) from Halyard Health to conduct this study. ZM assisted in the study design and plan, interpretation of results and manuscript production and editing. DM assisted in the study design, plan and study conduct, in addition to the interpretation of results and manuscript production and editing. MA assisted in the study design, statistical plan and analysis, the interpretation of results and manuscript editing.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests DW has received investigator initiated research funding and honoraria from Halyard Health; he receives funding from the National Institute on Aging (R01AG049924-01) for an unrelated research study in women. He receives royalties from UptoDate for a chapter on an unrelated topic. The other authors claim no other conflicts of interest.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.