Article Text

Original research
Genicular artery embolisation versus sham embolisation for symptomatic osteoarthritis of the knee: a randomised controlled trial
  1. Tijmen A van Zadelhoff1,
  2. P Koen Bos2,
  3. Adriaan Moelker1,
  4. Sita M A Bierma-Zeinstra3,
  5. Rianne A van der Heijden1,4,
  6. Edwin H G Oei1
  1. 1 Department of Radiology and Nuclear Medicine, Erasmus Medical Center, Rotterdam, Zuid-Holland, Netherlands
  2. 2 Department of Orthopaedics and Sports Medicine, Erasmus Medical Center, Rotterdam, Zuid-Holland, Netherlands
  3. 3 Department of General Practice, Erasmus Medical Center, Rotterdam, Zuid-Holland, Netherlands
  4. 4 Department of Radiology, University of Wisconsin-Madison, Madison, Wisconsin, USA
  1. Correspondence to Dr Tijmen A van Zadelhoff; t.vanzadelhoff{at}erasmusmc.nl

Abstract

Objective To determine the efficacy of genicular artery embolisation (GAE) compared with sham GAE for pain reduction in patients with symptomatic mild-to-moderate knee osteoarthritis (KOA).

Design Double-blind randomised sham-controlled clinical trial conducted from June 2019 to December 2021. The follow-up period was 4 months.

Setting Single-centre study conducted at a university medical centre in Rotterdam, Netherlands.

Participants 58 adults with symptomatic mild-to-moderate KOA not improving with conservative treatment.

Interventions Participants were randomised to receive either GAE treatment or a sham GAE treatment.

Main outcome measures The primary outcome was reduction of pain measured with the Knee Injury and Osteoarthritis Outcome Score pain subscale (0–100, with 0 representing the worst pain outcome and 100 the best) after 4 months. Outcomes were assessed at baseline and 1 and 4 months.

Results From June 2019 to December 2021, 58 patients were included. 29 patients were randomised to the GAE group and 29 to the sham group. All participants completed the study. The mean pain reduction after 4 months was 21.4 (95% CI 13.9 to 28.8) for the GAE group and 18.4 points (95% CI 11.6 to 25.1) for the sham group. The between-group difference for the mean pain reduction was 3.0 (95% CI −7.1 to 13.0) with an estimated Cohen’s d effect size of d = 0.15 (95% CI −0.37 to 0.66). Group allocation was not a significant contributor to pain reduction (p = 0.31). No serious adverse events (AEs) occurred. 23 mild AEs occurred in the GAE group and 5 in the sham group.

Conclusion We did not establish a clinical effect of GAE in patients with mild-to-moderate KOA as GAE produced a similar effect on pain reduction as a sham GAE procedure.

Trial registration number NCT03884049.

  • interventional radiology
  • knee
  • randomized controlled trial

Data availability statement

Data are available upon reasonable request. Data collected for this study will be shared upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • The addition of a control group receiving a sham treatment is an improvement to the evidence from previously performed cohort studies.

  • There were no dropouts during the study period, decreasing the risk of bias.

  • The success of adequate blinding was formerly tested.

  • Although justifiable considering the invasiveness of the treatment, this was a study with a relatively small sample size.

  • All procedures were performed by one interventional radiologist at one healthcare institution, increasing the consistency of the data but decreasing generalisability.

Introduction

Knee osteoarthritis (KOA) is a leading cause of morbidity worldwide with an estimated global prevalence of 16%.1 2 Therapies for early-stage OA aim to alleviate symptoms and prevent or postpone surgical treatment (eg, total knee arthroplasty) at end-stage KOA.3 However, there is a group of patients resistant to conservative treatment (eg, physiotherapy and non-steroidal anti-inflammatory drugs (NSAIDs)) but who are also ineligible for or unwilling to undergo surgical treatment. In the USA, this group is estimated to consist of 3.6 million patients, and this prevalence is projected to increase to 5 million by 2025.4 Patients in this so-called treatment gap experience severe KOA-related symptoms for a prolonged time, resulting in high morbidity and decreased quality of life (QoL). There is also a substantial economic burden associated with KOA through both direct and indirect costs.5 Therefore, new treatment strategies are needed for this patient group to bridge this treatment gap.

Genicular artery embolisation (GAE) is a novel minimally invasive procedure aimed to alleviate pain in patients with symptomatic KOA. The target of this treatment is to reduce perigenicular angiogenesis, which has been stated to be a contributor to KOA pain.6 The aim of GAE is to reduce blood flow to the hyperaemic synovium through embolisation, which hypothetically reduces synovitis and the pain it produces. Several uncontrolled cohort studies reported clinically important improvement in pain scores lasting for as long as 48 months.7–10 A meta-analysis of 11 uncontrolled cohort studies reported a pain reduction on the visual analogue scale (VAS) of 64% and 80% 6 and 24 months after GAE, respectively, suggesting a clinically meaningful and durable effect of GAE.11 Although these results show the potential of GAE and have resulted in an increased attention by treating physicians and intervention radiologists, a stronger scientific basis is needed before this treatment can be offered to the millions of KOA patients worldwide.

The scientific limitation of the aforementioned studies is the lack of a control group. Therefore, these studies did not account for a potential placebo effect, which is known to be effective in reducing pain complaints of KOA patients.12 Only two randomised controlled trials to date reported on the efficacy of GAE and showed conflicting results. One was relatively small (n = 21) and used a crossover design with a short crossover time point at 1 month. A greater improvement of 50.1 points on the VAS compared with the sham group was demonstrated.13 The other showed a non-significant difference of 2.6 points on the Knee Injury and Osteoarthritis Outcome Score (KOOS) pain score between treatment and control groups.14 The aim of this study was to further investigate whether GAE provides clinically relevant pain reduction in KOA patients resistant to conservative therapy compared with sham GAE.

Methods

The 2010 CONSORT checklist on reporting parallel-group randomised controlled trials was used.15 This study was a randomised sham-controlled trial conducted at Erasmus Medical Centre (MC), University Medical Centre Rotterdam, Netherlands. The trial protocol, available at http://www.clinicaltrials.gov/ under identifier NCT03884049, was approved by the medical ethical committee of the Erasmus University MC (MEC identifier: MEC-2018-081). A data safety monitoring board was appointed to monitor patient safety. Following the advice of the MEC, all adverse events (AEs) and the group allocations were reported to the data safety monitoring board (DSMB) after the first 10 inclusions. The DSMB deemed it safe to continue the trial after analysing the AEs.

Participants

Patients were recruited at the orthopaedic surgery outpatient clinics of Erasmus MC and four regional hospitals (IJsselland Hospital, Maasstad Hospital, Ikazia Hospital and Spijkenisse MC). All participants provided written informed consent.

The inclusion criteria were as follows: age ≥ 18 years, knee pain for ≥6 months, knee pain (NRS ≥4 to ≤8) on at least half of the days in the preceding month, insufficient response to conservative treatment (defined as physiotherapy aimed at strengthening the lower extremities, NSAIDs, intra-articular steroid or hyaluronic acid injections and use of walking aids) for ≥6 months determined by an orthopaedic surgeon and radiographic KOA (Kellgren and Lawrence (KL) grade 1–3)16 determined by a musculoskeletal radiologist. Patients with KL grade 4 KOA were excluded since these patients would be more suitable for total knee arthroplasty. If patients were bilaterally eligible, the most symptomatic side was chosen for treatment.

The exclusion criteria were as follows: contra-indications for MRI (eg, metallic foreign bodies, implants or devices incompatible with 3T MRI and claustrophobia), contra-indications for angiography (coagulation deficiency and allergy to iodinated contrast media), previous surgical treatment for KOA (high tibial osteotomy) except for knee arthroscopy, musculoskeletal comorbidity (eg, rheumatoid arthritis or gout) potentially blurring the effect of the treatment, renal insufficiency (glomerular filtration rate < 30 mL/min/1.73 m²), known allergy to contrast agents, known allergies to barium sulphate, 3-aminopropyltrialkoxysilane or polyphosphazene, pregnant or lactating women, intermittent claudication of the affected limb and intra-articular injections in the ipsilateral knee <6 months ago, in the waiting list for joint replacement surgery, amitriptyline usage implying a neuropathic pain component, insufficient command of the Dutch or English language and legally incompetent adults.

Randomisation and blinding

Immediately after inclusion, patients were randomly assigned to either GAE treatment or sham treatment, with a 1:1 ratio, in random blocks of 4 and 6, using an online web tool (www.aleaclinical.eu). An email with the randomisation result was sent to the interventional radiologist performing the procedure. He was the only study team member aware of treatment allocation for the entire study period. Blinding of patients for the allocation of treatment group was assessed using James’ blinding index, which ranges from 0 (total lack of blinding) to 1 (complete blinding).17 If the 95% CI lower bound was above 0.5, successful blinding was assumed. Three hours after the procedure, patients were asked in which treatment group they believed they were allocated, with three options to answer: ‘treatment group’, ‘placebo group’ or ‘I do not know’.

Procedures

Patients in the intervention group received GAE treatment. Following local anaesthesia with 10 mL 2% lidocaine, a small incision in the groin was made after which a 4 Fr catheter was inserted into the common femoral artery with an anterograde approach. A multi-purpose catheter with haemostatic valve and a 1.8 Fr microcatheter were advanced. After obtaining an overview series of the common femoral artery above the adductor canal, selective angiography was performed to determine where a hyperaemic blush was present indicating angiogenesis (figure 1). Eight genicular arteries were systematically depicted: the lateral superior genicular artery, medial superior genicular artery, lateral inferior genicular artery, medial inferior genicular artery, descending genicular artery, superior patellar artery, median genicular artery and superior patellar recurrent artery (figure 2). When identified, the blush was embolised using Embozene microspheres (Varian, Palo Alto, California, USA) of 75 or 100 µm, a typical size when using microspheres for GAE.8 10 13 The end-point of embolisation was near-stasis of flow in the hyperaemic blush while preserving the normal genicular vasculature. Before embolisation, ice packs were applied around the knee to induce vasoconstriction of small cutaneous arteries and reduce non-target embolisation. All interventions were performed by one interventional radiologist (AM) with 15 years of experience in the field of interventional radiology and embolisation procedures. This interventional radiologist was not involved in any data collection.

Figure 1

(A) Pre-embolisation angiography of the medial superior genicular artery of a right knee. Notice the blush appearance at the medial femoral condyle. (B) In the post-embolisation situation, the blush disappeared but the medial superior genicular artery is still open. This was considered a technically successful result.

Figure 2

A schematic depiction of all genicular arteries evaluated for a hyperaemic blush and how many times a specific artery was selected for embolisation (absolute number, proportion). (A) Superior patellar artery, (B) descending genicular artery, (C) lateral superior genicular artery, (D) medial superior genicular artery, (E) median genicular artery, (F) medial inferior genicular artery, (G) lateral inferior genicular artery and (H) anterior tibial recurrent artery.

Patients in the sham group received sham GAE treatment. Local anaesthesia was administered, and a small incision in the groin was made after which the interventional radiologist and his team pretended to insert the catheter and perform the procedure.

The utmost effort was made to ensure that patients in both groups were unaware of group allocation. The patients’ line of sight was cut off from the interventional field by a surgical drape. All patients were given a noise-cancelling headphone with music to block any ambient sounds. For patients in the sham group, the same procedure as in the intervention group was mimicked as much as possible. For example, during the sham procedure, the C-arm and table were moved according to a real angiography, while shutters blocked the X-rays. This way, patients in the sham group were given the impression of undergoing an angiography but were not exposed to unnecessary radiation. The sham procedure took 1–2 hours, equivalent to the actual procedure. After the procedure, the puncture site was compressed for 10 min, and a pressure bandage was applied. All patients stayed in bed laying supine for at least 3 hours to prevent a rebleed. Thereafter, patients were mobilised and discharged.

Outcomes

Follow-up measurements were collected during in-person visits at Erasmus MC after 4 and 16 weeks and were conducted by the same assessor (TAvZ), blinded for treatment allocation. If patients were unable to physically attend a follow-up visit, questionnaires were sent by mail. KOA symptoms were measured using the KOOS questionnaire.18 The KOOS consists of five subscales: symptoms, pain, daily living, sports and recreational activities and QoL. The subscales are summarised in a score ranging from 0 (no complaints) to 100 (worst complaints). The KOOS is considered a valid and responsive measurement for pain in KOA patients.18 Pain was also evaluated using a VAS pain score ranging from 0 (no pain) to 100 (worst pain). Patients were asked to rate their average knee pain and worst knee pain on a VAS. To distinguish between intermittent and constant OA pain, the Intermittent and Constant Osteoarthritis Pain (ICOAP) questionnaire was used.19 The painDETECT questionnaire assessed the presence of a neuropathic pain component.20 The EQ-5D-5L questionnaire was used to determine QoL.21

During follow-up visits, patients were actively questioned about any AE that could have occurred in the preceding period. Furthermore, participants were asked to report any hospitalisation or sudden change in health status to a study team member. AEs were documented using the common terminology criteria for AE. The AEs were assigned to a system organ class and graded for severity ranging from 1 (mild) to 5 (death related to AE).22 Attribution of AEs was determined by the assessor conducting follow-up measurements in consultation with an orthopaedic surgeon (PKB).

Patients were asked to not start any new therapies during the trial. Patients were allowed to stop any therapy they were receiving at the start of the trial. Concurrent therapies were documented during all follow-up visits.

Statistical analysis

The primary analysis was performed according to the intention-to-treat principles. The primary outcome was the KOOS pain subscale score (0–100) change after 4 months. We assumed an SD of 16 points for the KOOS pain score estimated from multiple KOA studies at our institution. A sample size of 48 was calculated to detect an increase of 8 points in the control group compared with 21 points in the intervention group (power 0.80, two-sided significance level 0.05) based on the effect observed in previous studies.10 23 24 The projected sample size was increased to 58 to account for approximately 20% loss to follow-up. Longitudinal data were analysed with the generalised estimating equations (GEE) method using an exchangeable working correlation structure to test between-group differences. Group allocation and follow-up time point were added to the model, and the interaction effect between these variables determined whether group allocation was of influence to the dependent variable (KOOS pain score change from baseline) at a certain time point. The distribution of body mass index (BMI), sex and radiographic OA severity and the degree of neuropathic pain according to painDETECT were checked, and if there was a >10% difference between both groups at baseline, the influence of this variable on the estimate of the GEE model was tested for every variable separately, without any other confounding variables in the model. If this changed the estimate by >10%, the variable was included in the model in order to adjust for this variable. Secondary outcomes were change of the other KOOS dimensions (symptoms, daily living, sport and recreation and QoL), VAS pain score change, ICOAP change, painDETECT change and the EQ-5D-5L. These were tested similarly as the primary outcome. A p < 0.05 was considered statistically significant for the primary outcome. For the secondary outcomes, we applied a Bonferroni correction to control for multiple comparisons. The corrected significance level for all secondary outcomes was set at α = 0.005 (0.05/10). Additionally, Cohen’s d effect sizes for between-group differences were calculated.

Between-group differences in concurrent medication and physiotherapy usage were tested using a χ2 test.

The data were analysed and interpreted by all study team members using a blinded interpretation approach without the knowledge of which study arm received GAE and which study arm received the sham treatment.25 The analysis and interpretation were documented and signed by all study team members. Subsequently, these analysis and interpretation were verified by an independent researcher (M.G.M. Hunink, MD, PhD) with a background in methodology and biostatistics (online supplemental material independent review). Only after this independent researcher agreed with the analysis and interpretation, the blinding was broken.

Supplemental material

Patient and public involvement

No patients or members of the public were directly involved in setting the research question, developing the research design, collecting the research data or interpreting the results.

Results

From June 2019 to December 2021, 58 patients were included. Details on the recruitment process can be found in the participant flow diagram (figure 3). The primary reason for not participating was unwillingness to undergo a sham treatment. 29 patients were randomised to the GAE group and 29 to the sham group. There were no dropouts or missing data. James’ blinding index was 0.70 (95% CI 0.61 to 1.00), indicating successful blinding of the participants.

Figure 3

Participant flow diagram. GAE, genicular artery embolisation; KL, Kellgren and Lawrence.

Baseline characteristics were similar between the groups (table 1). Of all participants, 60.3% was female, and the mean age was 58.1 years (SD = 8.3). The mean BMI was 30.0 kg/m2 (SD = 4.8), and the average duration of their complaints was 8.0 years (SD = 7.8). Most patients had KL grade 3. The average painDETECT score was 9.8, implying that a neuropathic pain component was unlikely among participants.

Table 1

Baseline characteristics

The mean procedure time was 91.9 min (SD = 18.5) in the sham group compared with 108.2 (SD = 21.7) in the treatment group. In the treatment group, the average amount of contrast administered and dose area product of ionising radiation were 148.1 mL (SD = 197.4) and 6695.9 mGy/cm2 (SD = 6830.4), respectively. The mean number of arteries selected per patient for embolisation was 3.7 (SD = 1.6), and the medial superior genicular artery was most frequently selected for embolisation (figure 2). No blush was identified in three patients from the GAE group, and thus no embolisation was performed in these patients.

Sex and KL grade showed a difference >10% between both groups; however, when added to the model, the primary outcome did not change >10%. The mean pain reduction in the KOOS pain subscale after 4 months was 21.4 (95% CI 13.9 to 28.8) for the GAE group and 18.4 (95% CI 11.6 to 25.1) for the sham group. The difference between both groups was 3.0 (95% CI −7.1 to 13.0). The effect size was small (0.15). There was no significant interaction effect for the term group allocation follow-up time (p = 0.31), and thus, group allocation was not a significant contributor to KOOS pain reduction. No secondary outcome demonstrated a significantly different improvement between both groups (table 2).

Table 2

Primary and secondary outcomes. A p value of 0.05 for the primary outcome and a p value of 0.005 for the secondary outcomes were considered significant

28 AEs occurred within the 4-month follow-up period. Of these, 5 occurred in the sham group and 23 in the GAE group. The most common AE was transient skin erythema, which occurred in 11 patients, who all received GAE treatment. The erythema was self-limiting after 1–4 weeks. Two patients receiving GAE treatment developed paraesthesia of the lateral ankle at the treated side, which did not resolve during this study period of 4 months. Since both cases had the onset of complaints within 3 days after the procedure, both AEs were considered to be related to GAE treatment. All AEs that occurred in the sham group were unrelated to the sham treatment. No serious AEs occurred.

Concurrent medication and physiotherapy usage can be found in online supplemental table 1. At no time point was there a significant difference in patients using medication or receiving physiotherapy.

Supplemental material

Discussion

This randomised controlled trial (RCT) compared the effect of GAE versus sham GAE treatment for pain reduction in patients with symptomatic mild-to-moderate KOA resistant to conservative therapy. In both groups, a clinically relevant pain reduction was observed, but the difference was not statistically significant. Therefore, we were unable to demonstrate the hypothesised large effect size of GAE 4 months after treatment. No secondary outcome demonstrated a significantly different improvement between both groups. Patients were adequately blinded throughout the study.

The largest strength of this study is the use of a control group receiving sham GAE treatment. Previously, multiple uncontrolled studies reported substantial pain relief after GAE ranging from 24- to 57-point improvement on the VAS.7 8 10 13 24 The GAE group in this study achieved a pain reduction of 22.9 points on the VAS. This is comparable to the low end of the aforementioned cohort studies. However, a similar pain reduction was achieved in our placebo group, suggesting that the effect is explained by a placebo effect rather than GAE.

A potential weakness of this study is the patient selection. The proposed effect of GAE hypothetically is achieved through the reduction of synovitis and pain stimuli by embolisation of angiogenesis and subsequent perivascular sensory nerve hypoxygenation.10 However, we did not select patients based on the presence or severity of synovitis but rather based on symptoms and KL grade. If the mechanism of pain alleviation in GAE relies on the severity of synovitis, the inclusion of patients exhibiting minimal or no synovitis diminishes the likelihood of detecting a discernible effect of GAE in our study. This might have influenced our primary outcome and is a possible shortcoming of this study. However, previous research has shown that synovitis severity at baseline measured by MRI is not associated with better outcomes of GAE.26

Our results are in line with the findings of a recent RCT including only KL grade 2 patients reporting no significant difference between GAE and sham GAE (19.5 vs 23.6 points) in KOOS pain improvement after 6 months.27 It should be noted that the intervention procedure changed during the study from embolisation of a single vessel, to multiple vessels and eventually to all vessels during the study. A subgroup analysis between the control group and patients for which all genicular arteries were embolised showed a non-significant difference of 11.0 point in the KOOS pain subscale after 12 months. A substantial difference is that in our study, we used permanent particles in the GAE group compared with a suspension of imipenem/cilastatin and contrast agent in the study by Landers et al. One could hypothesise that imipenem/cilastatin produces an additional local anti-inflammatory effect that could have influenced the pain-reducing effect. Furthermore, a formal test to assess the adequacy of blinding was lacking, potentially introducing bias given the known difficulties of blinding in interventional studies. Finally, our results are more generalisable due to the broader inclusion criteria of KL grades 1–3 in comparison to only KL grade 2.

Our trial was powered to demonstrate a 13-point difference in the KOOS pain subscale between both groups. Pain reduction in the GAE group (21.4 points) was as expected. Nevertheless, the pain reduction in the sham group (18.4 points) exceeded expectations, primarily attributed to a placebo effect more than twice as pronounced as initially anticipated. Consequently, we were unable to demonstrate the hypothesised large effect size of GAE. However, both groups achieved the minimal important difference (MID). The specific MID for GAE in the KOOS pain subscale is unknown, but the general MID for KOA patients is estimated at 12 points.28 The placebo effect tends to diminish to below the MID after 6 months in KOA patients treated with intra-articular saline injection.29 It is possible that after a longer period of time, the placebo effect fades but the effect of GAE remains. Future studies need to determine whether there is a significant effect of GAE compared with sham GAE after a longer follow-up period.

In conclusion, in our RCT comparing GAE versus sham GAE, both GAE and sham GAE groups exceeded the MID, yet no significant difference in pain reduction was observed due to a substantial placebo effect. Future trials should consider this when determining sample size and study duration in order to advance our understanding of GAE effectiveness.

Data availability statement

Data are available upon reasonable request. Data collected for this study will be shared upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

The trial protocol, available at http://www.clinicaltrials.gov/ under identifier NCT03884049, was approved by the medical ethical committee of the Erasmus University Medical Centre (MEC identifier: MEC-2018-081). All patients provided written informed consent before participation.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Deceased AM since deceased

  • Contributors All study authors were involved in the trial design. AM performed the interventions. TAvZ performed the follow-up visits and the data collection. TAvZ, AM, SMAB-Z, PKB, RAvdH and EHGO were involved in writing the manuscript, statistical analysis and had full access to all the data in the study and had final responsibility for the decision to submit for publication. TAvZ acted as guarantor.

  • Funding Funding was provided by Stichting Coolsingel (grant number: 562), COOK Medical (grant number: EUFRB-22-04-2016-IR-NL) and Erasmus MC MRACE (grant number: 2017-17202). The funders had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.