The short-to-midterm outcomes of geniculate artery embolization for mild-to-moderate osteoarthritis of the knee: a systematic review

Abstract

Background

Geniculate Artery Embolization (GAE) is a novel, minimally-invasive treatment for mild to moderate Osteoarthritis (OA) associated with musculoskeletal pain.

Purpose

To investigate the outcomes of GAE in patients with mild to moderate OA of the knee using a Quantitative-analysis of the available literature.

Methods

The PubMed, EMBASE and Cochrane databases were searched for GAE-related studies. Qualitative and quantitative analyses were performed following PRISMA-guidelines. Quantitative-analysis was performed using windows based ‘MedCalc-Statistical-Software version 19.6.1 (2020). Statistical analysis was performed in Stata–IC–13.1(Stata corp-USA). Quantitative-analysis was done using the random-effects model, and the Standardized-Mean-Differences (SMD) were calculated.

Results

After a full-text review, 13 studies with 399 knees (345-patients) were included in the qualitative synthesis, 10 were included in the quantitative synthesis. The total WOMAC score improved by a [SMD (95% CI)] of 3.46 points (1.27, 5.65), 3.50 (1.28, 5.71), 3.77 (0.58, 6.96), 5.46 (1.59, 9.33), 2.96 (−0.93, 6.85) compared to baseline at 1, 3, 6, 12, 24 months respectively. The VAS score improved by 2.06 (1.35, 2.76), 2.13 (1.39, 2.87), 2.36 (1.85, 2.90), 2.09 (0.91, 3.28) compared to baseline at 1, 3, 6, 12 months respectively. The Pain WOMAC score improved by SMD 2.97 (0.51, 5.43), 3.77 (0.58, 6.96), 2.27 (0.31, 4.22), 2.27 (0.31, 4.22) compared to baseline at 1, 3, 12, 24 months, respectively.

Conclusion

There was a statistically significant change from baseline in all outcome measures after GAE. GAE is a safe and effective method for pain control in mild to moderate OA-associated knee pain in the short and medium term.

Highlights

• •This systematic review shows that GAE is safe and effective in managing pain associated with mild to moderate OA.
• •GAE causes a significant improvement in the Total WOMAC scores indicating increased joint mobility and function.
• •GAE is a suitable alternative for OA patients who are unable or unwilling to undergo surgery.

1. Introduction

Osteoarthritis (OA) is a common and potentially debilitating disease affecting 10 and 18% of men and women above 60, respectively.¹ OA is the degenerative destruction of joints resulting from excessive wear and tear. A chronic, low-grade inflammatory response dominated by innate immune mechanisms causes progressive damage, resulting in pain, disability, and Quality of Life (QoL) issues.¹ Currently, Clinical evaluation of OA is done by the Western Ontario, and McMaster Universities Arthritis Index (WOMAC) scores.² Other evaluation measures include the Knee injury and Osteoarthritis Outcome Scale (KOOS).³ Pain is also evaluated by the Visual Analog Scale (VAS).⁴

The current guidelines recommend symptomatic management of the condition in the early stages, with pain relief (such as NSAIDs), weight loss, and physical therapy for function preservation being the first-line treatment.⁵ The standard of care (and ultimate last resort) for severe, chronic pain in end-stage OA is Total Knee Arthroplasty (TKA).⁶ Recently there has been an increase in alternative, minimally invasive therapies, such as Geniculate Artery Embolization (GAE) which have shown promising results in individual studies by reducing pain and increasing functionality.⁷

A significant contribution to the understanding of the pathophysiology was made by Okuno et al., in 2014 by discovering the presence and role of hypertrophied synovium and joint hyper-vascularization in nociception.⁷ GAE involves selectively embolizing 1 or more geniculate arteries supplying the knee joint, which feeds the abnormal synovium to reduce inflammation and pain7. At the senior author's (S.B.) institution, GAE is performed with ultrasound-guided antegrade catheterization of the common femoral artery, followed by introducing a 4F vascular sheath. Next, a 4-French catheter is advanced to the distal superficial femoral artery, and a DSA is performed to visualize the geniculate arterial anatomy and identify the artery supplying the hypervascular area in the knee joint. A microcatheter (2.1 or 1.7 Fr or 2.7F) is then advanced coaxially into the geniculate artery for super-selective embolization with particulate embolic agents (Fig. 2 shows the angiographic anatomy while Fig. 3 shows an angiographic image before and after embolization of the artery supplying neovessels). The embolic agent is prepared by mixing 2 mL of Embosphere with 19 mL of contrast and 19 mL of saline. Embolic materials such as Embospheres (Merit Medical, USA), Imipenem/Cilastatin (IP/CS) (Primaxin, Merck & CO., Whitehouse Station, New Jersey), Embozene (Varian Inc., USA) and PVA (Boston Scientific, USA) are used. Currently, particles of sizes varying from 75 μm to 300 μm made of different embolic materials are used for GAE in OA.7, 8, 9 Imipenem/Cilastatin mixture turns to 10–140 μm size particulate material when mixed with iodinated contrast¹⁰

Fig. 2.

DSA image showing the geniculate artery anatomy. Legend: DGA: Descending Geniculate Artery.

SLGA: Superior Lateral Geniculate Artery

ILGA: Inferior Lateral Geniculate Artery.

Fig. 3.

A: Super-selective catheterization of Superior Lateral Geniculate Artery with black arrows showing Hyperaemic blush in neovessel formation. 3B: Post-embolization image with black arrows showing the same area is now without Hyperaemic blush.

Since the report in 2015 by Okuno et al.⁷ that transcatheter arterial embolization could improve pain and knee function in patients with mild to moderate OA resistant to conservative treatment, the role of GAE in managing OA has been increasingly emphasized. GAE has been granted breakthrough designation by the FDA.¹¹ While research is ongoing, there is a paucity of literature evaluating the pooled effect of GAE on OA, which is essential for clinicians and agencies to recommend the optimal treatment for their patients.

The objective of this study is to review and answer the following question with a Quantitative-analysis of the available literature: In patients with osteoarthritis, how does GAE affect pain and quality of life within the short to medium term?

1.1. PICOs

Participants	Intervention	Comparison	Outcomes	Description of Outcome
Patients with mild to severe osteoarthritis in one or both knees	Geniculate Artery Embolization	Baseline	Visual Analogue Scale (VAS)	A self-reporting visual scale representing pain from a scale of 1–10 cm or 1–100 mm
			Western Ontario and McMaster Universities Osteoarthritis Index. (WOMAC)	A questionnaire designed to evaluate the impact of OA on joint function. The Scale is scored from 0 to 96.
			WOMAC - Pain	A subscale of the WOMAC questionnaire for assessing pain. The Scale is scored from 0 to 20.

Search strategy: A search was performed in PubMed, EMABSE and the Cochrane Library database, in accordance with PRISMA guidelines for systematic review and Quantitative-analysis using a combination of terms such as “geniculate artery”, “embolization”, “knee”, “osteoarthritis”, and “pain”, along with Boolean operators and received a total of 524 results. No language restrictions were applied. The complete strategy is available in the supplemental information. These were exported to Microsoft Excel for further review. The search was performed in June 2022 and July 2022.

Study selection: The selection process is shown in Flowchart-1. Excel was used to identify and remove duplicates, which left 405 results. The inclusion criteria were; Original studies in which GAE was performed, and the clinical outcomes were quantified via established outcome measurement tools such as VAS, WOMAC, and KOOS.

Flowchart 1.

PRISMA diagram showing the Study Selection process.

Two authors independently reviewed the titles and abstracts to match them to the inclusion criteria. Irrelevant studies, reviews, case reports, animal experiments, and editorials were excluded from the quantitative analysis. Case reports were included in the qualitative synthesis. This left a total of 25 studies which were then reviewed in full text. Studies that referred to the same data set and population were eliminated, as were the irrelevant studies or those using different study end goals. Disagreement was addressed through consensus.

A total of 13 studies were finally selected for qualitative analysis, and 10 were selected for quantitative analysis. The studies with their individual characteristics have been presented in Table 1. Both authors reviewed each study using the NIH quality assessment tool¹² to evaluate the quality of the studies, the result of which has been presented in Supplementary TABLE 2.

Table 1.

Summary of studies selected for qualitative review.

Table1A: Characteristics of included studies. The data collected shows clear preferences of imaging modality, types of studies conducted and the geographical distribution of studeis
Author(s)	Country of origin	Year	Study type	Imaging Modality	Patients	Total Knees	Side affected (R:L)	Mean age	BMI(kg/m^2)
										KL grade
Bagla et al.	USA	2020	P/MC	MRI	20	20	11; 9	59.4	35
Okuno et al.7	Japan	2015	P/SC	MRI	14	14	5; 9	65.2	26.3	3 (n = 9); 2 (n = 9); 1 (n = 2)
Okuno et al.21	Japan	2017	P/SC	MRI	72	95	49; 46	64.4	25.1	0 or 1 (n = 8), 2 (n = 6)
Lee et al.19	South korea	2019	R/SC	MRI	41	59	29; 30	66.2 ± 6.7	24.6 ± 3.6	1 or 2 (n = 62), 3 (n = 33)
						12	6; 6	68.1 ± 6.5	25.0 ± 4.1	1 to 3
Piechowiak et al.16	USA	2017	P/SC	MRI	5	5	NR	50-88 (age range)	0	4
Padia et al.8	USA	2021	P/SC	MRI	40	40	16; 24	69	29.3	nr
Padia et al.36	USA	2021	P/SC	DSA/CT	40	40	15; 25	69	28	2 to 3
Bagla et al.9	USA	2022	P/MC	MRI	7	7	5; 9	63.9 ± 8.37	30.8 ± 8.14	nr
					14	14	6; 1	62.9 ± 7.13	33.4 ± 10.5	2 to 3
Little et al.17	UK	2021	P/SC	MRI	38	38	nr	60(median)	30 (median)	1 to 3
Jalaeian et al.18	USA	2021	P/SC	MRI	21	33	17; 16	70.45		1 to 3
Landers et al.15	Australia	2021	P/SC	MRI	11	18	nr	nr	nr	1 to 4
Kumar et al.37	USA	2020	P/SC	MRI	21	21	nr	48–71 (age range)	nr	1 to 3
Lauko et al.	USA	2020	Case report	CT	1	1	left	64	nr	2 (median)
Legend: P - Prospective, SC - Single center, MC - Multicenter, NR - Not Reported, E - Embozene, I/C - Imipenem/Cilastatin, NA - Not Applicable, SIR - Society of interventional Radiology.

Table1B: Characteristics of included studies. Most studies have evaluated outcomes in the WOMAC scale. Embozene and IP/CS are the most commonly used embolic agents
	Embolic agent used	Agent dose and volume	Clincal success rate at endpoint	Technical success rate	Outcome scale used	complications		Follow up time (months)	Sample size
						overall/severe	Based on SIR
Bagla et al.	E	75/100 μm E spherical particles with 9 mL contrast material added to the 6 mL particles (sol)	80% (95% CI, 56%–94%) in WOMAC and 85% (95% CI, 62%–97%) in VAS.	100%	WOMAC and VAS	16/0	A	6 months	20
Okuno et al.	E (n = 3) and I/C (n = 11)	75 mm E spherical particles (mean volume; 0.068/2 mL particle volume) plus 10–70 mm I/C (mean volume; 2.5 mL/5 mL suspension)	NR	100%	WOMAC and VAS	1/0	B	12 ± 5	14
Okuno et al.	E (n = 7) and I/C (n = 65)	75 μm E + 0.5 g I/CS suspension in 5–10 mL iodinated contrast agent	85.2% (95% CI, 72%–92%) for KL 1–2, 69.8% (95% CI, 49%–84%) for KL 3	100%	WOMAC and VAS	16/0	B	24	95
Lee et al.	IC	0.5 g I/C in 7 mL iodinated contrast medium	NR	100%	VAS	10/0	B	12	59
	IC		NR	100%				12	12
Piechowiak et al.	E	75 μm Embospheres	NR	100%	WOMAC and VAS	0	NA		5
Padia et al.	E	100 μm E particles	68%	100%	WOMAC and VAS	7/0	A	12	40
Padia et al.	E	100 μm E particles	NR	100%	WOMAC and VAS	4/0	B	12	40
Bagla et al.	E	100–300 μm particles. 2 mL of particles (sol.) with 18 mL of contrast material	79%	100%	WOMAC and VAS	6/0	A	12	21
	E	100–300 μm absorbable particles. 2 mL of particles (sol.) with 18 mL of contrast material	0%	100%		5/0	B		7
Little et al.	Embospheres	100–300 μm Embosphere particles (Merit Medical, USA) diluted in 20 ml (300 mg/ml) iodinated contrast (Iomeron, Bracco, Italy)	69%, 56%, 53%, 59%, and 69% of patients at 3 months for symptoms and stiffness, pain, daily living, sports and recreation, and quality of life, respectively.	84%	WOMAC and VAS	5/0	B	12	32
Jalaeian et al.	E and IP/CS	100-300 ES and IP/CS	61.5% versus 53.8% of in ES and IP/CS groups	100%	WOMAC	8/0	A	24	33
Landers et al.	E	100-to 300-μm ES	NR	100%	KOOS outcome scale	1/0	B	12	18
Kumar	PVA	NA	NR	100%	WOMAC and VAS	0/0	A	1
Lauko et al.	E	NR	100%*	100%*	WOMAC	1/0	A	6	1
Legend: P - Prospective, SC - Single center, MC - Multicenter, NR - Not Reported, E - Embozene, I/C - Imipenem/Cilastatin, NA - Not Applicable, SIR - Society of interventional Radiology.

Data extraction and outcome measure: Both authors extracted the data independently to ensure accuracy in data collection. The type of study, study design, sample size, type of embolic agent used, side of embolization, complication type and rate, success rate, Whole-Organ Magnetic Resonance Imaging Score (WORMS) score, WOMAC score at baseline, 1 month ±2 weeks, 3 months, 6 months, 12 months, and 24 months, VAS scores at the same time points as the WOMAC score were the outcome measures noted. The baseline demographics included the sex distribution, mean BMI, mean age, OA grade, imaging modality used, the side affected, medication use, previous treatments, and Range of OA grade. Patient-level data were available for Bagla et al., 2022 via publication and Jalaeian et al., 2021 due to the association of SB in the same study.

The entire process was overseen by SB, an Interventional radiologist with 16 years of experience.

Statistical analysis: Quantitative-analysis was performed using windows based ‘MedCalc Statistical Software version 19.6.1 (2020)’. Data computations and imputations were done in Stata-IC 13.1 (Stata corp., USA). Quantitative-analysis was done using the random-effects model for comparisons between the study and control for the change in PD and CAL from baseline to 6 months (continuous measure). The principal summary measure used was the Standardized Mean Differences (SMD). SMD was calculated using the Hedges g statistic, the difference between the two means divided by the pooled standard deviation, with a correction for small sample bias. Wherever possible, Quantitative-analysis was repeated after excluding studies showing publication bias (Funnel plot). The heterogeneity of the data is estimated using Cohran's Q and I² statistics. P < 0.050 was considered significant.

2. Results

2.1. Qualitative analysis of the studies

Overarching themes of the studies included in this Systematic Review were that they were primarily single-arm, small (no study had >100 knees), included early osteoarthritis (KL grade 1–3), and were predominantly from the United States and East Asia. The publication dates range from 2015 to 2022. The individual study details are summarized in Table 1. A total of 13 studies with 399 knees in 345 patients were included in the qualitative synthesis.

All studies recorded the Kellegren-Lawrence (KL) grade¹³ of the patients’ OA, the ranges of which have been presented in Table 1. All studies included patients with KL grades 1–3, and 3 studies also included patients with KL grade 4. Plain radiographs depicting the contrast between KL grades are presented in Fig. 4. The patient evaluation was done by MRI in all studies, except for Padia et al. and Lauko et al., who used CT for evaluation. All studies included patients who had pain refractory to conservative therapy (Physiotherapy, NSAIDs, and Acetaminophen). 7 Studies recorded and followed up with the patient about requiring pain medication, physiotherapy, and other conservative measures. These have been recorded and presented in Table 3.

Fig. 4.

A series of Plain Radiographs showing KL grades. From Left to right - KL grade 1: No apparent radiological features visible; KL grade 2: Joint Osteophytes seen; KL grade 3: Decreased Joint Space; KL grade 4: Definite bone deformity with severe sclerosis.

Table 3.

Summary of post-GAE treatments in all studies.

Author and year	Physiotherapy		NSAID + acetaminophen		Opiate		PRP		HA		IACS
	Baseline	Final follow-up	Baseline	Final follow-up	Baseline	Final follow-up	Baseline	Final follow-up	Baseline	Final follow-up	Baseline	Final follow-up
Bagla et al., 2019	NR	NR	17	8	6	1	NR	NR	NR	NR	NR	NR
Okuno et al., 2014	10	NR	10	1	NR	NR	NR	NR	6	0	NR	NR
Okuno et al., 2017	NR	NR	39	4	22	2	NR	NR	43	2	NR	NR
Lee et al., 2019 (KL 1–3)	39	2	46	9	NR	NR	NR	NR	39	2	NR	NR
Lee et al., 2019 (KL 4)	10	1	10	3	NR	NR	NR	NR	10	1	NR	NR
Little et al., 2021	21	NR	25	8	5	3	1	0	NR	NR	26	0
H. Jalaeian et al., 2020	22	5	19	6	NR	NR	NR	NR	18	4	9	6
Landers et al., 2020	6	0	8	5	0	0	NR	NR	NR	NR	1	0

The trend shows a decrease in the use of adjuvant therapies post GAE.

Legend: NR – Not reported; PRP: Platelet Rich Plasma; HA: Hyaluronic acid; IACS: Intra-Articular Cortico-Steroid injection.

Two studies (Okuno et al. and Little et al.) recorded the WORMS score of the participants; this is the oldest and most widely used score for evaluating OA. It is a semi-quantitative, multi-feature score using conventional Magnetic Resonance techniques to evaluate joint changes.¹⁴ These have been presented in Supplementary Table 1.

Almost all studies measured the clinical outcomes via the VAS or WOMAC questionnaire. Landers et al. reported the outcomes exclusively as KOOS scores.¹⁵ This study was not included in the quantitative analysis but was part of the qualitative synthesis.

The embolic agents used were Embozene (Varian Inc., USA) (9 studies), Embospheres (Merit Medical, USA) (1 study), and Imipenem/Cilastatin (Whitehouse, New Jersey, USA) (2 studies), while IP/CS and Embozene both were used in 2 studies. PVA was utilized as the embolic agent in one study. The range of Embosphere particles used was 100–300 μm. The definition of technical success was varied, but embolization of at least one geniculate artery was considered in most studies as technical success. Other definitions included devascularisation of the entire organ or pruning of the hypervascular region (Okuno et al., 2015 and Piechowiak et al., respectively).^7,16 Technical success rates were 100% in all studies except Little et al., who reported a technical success rate of 84%. This lower rate was because of a conservative approach to the risk of non-target embolization and lack of a hyperemic target in some patients.¹⁷

Okuno et al., Jalaeian et al., and Lee et al. used a 50% reduction in WOMAC or VAS scores compared to baseline as the benchmark for clinical success.^7,18,19 Bagla et al. used a 20% reduction in WOMAC or a 16% reduction in the VAS score as the benchmark for clinical success.²⁰ Landers et al. used a 10-point decrease in the KOOS score as the minimal clinically important difference¹⁵ (Table 1). The remaining studies did not report a definition of clinical success. Clinical success rates were variably reported and were more varied than technical success rates, with clinical success rates ranging from 61.5% to over 80% at the study endpoint (Table 1). This deviation indicates that technical success is not correlated with long-term clinical outcomes, suggesting that other factors, such as disease process, patient PT, and lifestyle change, may play a role in the long-term clinical outcome.^21,22 Some studies were small, and the technical success rates of 100% may not be universally applicable to larger cohorts. Abundant collateral circulation between the geniculate arteries may also account for the clinical failure and recurrent symptoms.^8,21

All studies followed up for at least 3 months, but the follow-up times were variable. The longest follow-up period was 24 months (Okuno et al.), while the average follow-up times were 6–12 months (Table 1). Since GAE is a novel treatment approach, enough time has not passed since it was operationalized in most institutions to have longer-term follow-ups.

Post-GAE use of alternative therapies such as Platelet Rich Plasma (PRP) injections, Intra Articular Corticosteroid (IACS), and Physical Therapy was recorded by 7 out of 13 studies. PRP and IACS injections involve the instillation of Platelet Rich Plasma or Corticosteroids into the joint.²³ These therapies have improved pain and stiffness in the short and medium term versus placebo.²³ Using such therapies as adjunctive treatment can have a beneficial or confounding effect on clinical outcomes. Cessation of use compared to baseline may indicate a successful clinical outcome.

Almost all studies reported minor side effects, and Piechowiak et al. reported no side effects (Table 1). All adverse effects were graded as SIR complication grade A or B. Table 1 displays the total adverse effect rates and grading. A total of 80 patients out of 399 had complication(s), yielding a complication rate of 20%. The complications included fever, access site hematomas, transient cutaneous changes, and transient numbness. No major adverse effects were reported, such as new-onset pain, paraesthesia, or knee instability.

2.2. Quantitative-analysis of GAE efficacy

The metrics used to quantify the outcome of pain and functionality were the total WOMAC score, Pain WOMAC score, and the VAS scale (100 mm). The Quantitative-analysis showed statistically significant decreases in all three metrics at the primary temporal endpoints. Pooled effects of the studies were calculated at 1 month, 3 months, 6 months, and 12 months. Only two studies reported a follow-up score at 24 months. Other study follow-up times, such as 1 day, 4 months, 9 months were not included in the pooled effects analysis due to the uniqueness of such time points.

The total WOMAC score (Fig. 6) showed a Standardized Mean Difference (SMD) of 3.46 points (1.27, 5.65 [95%CI]) (p = 0.00) from the baseline at 1 month. The follow-up at 12 months showed an SMD of 5.46 points (1.59, 9.33 [95%CI]) (p = 0.01) from baseline, with the SMDs at the intervening time points shown in Fig. 4, Fig. 5, Fig. 6, Fig. 7. The final pooled analysis for 24 months showed a statistically insignificant change in Total WOMAC, although this is likely the result of only 2 studies being imputed at this point and the random-effects model applied to them. It can be summarized from the results that the patient functionality improved significantly.

Fig. 6.

Clockwise from top left: Change in total WOMAC score compared to baseline, at 1, 3, 6, and 12 months post GAE. The change in the Total WOMAC score from baseline is statistically significant for all time points evaluated. Other timepoints can be found in the supplemental material.

Fig. 5.

Clockwise from top left: Changes in VAS pain score at 1, 3, 6, and 12 months post-procedure. The change in VAS Pain score from baseline is statistically significant for all time points evaluated. Other timepoints can be found in the supplemental material.

Fig. 7.

Pain WOMAC: Top to bottom: Pain WOMAC change from baseline at 1, 3 and 12 months post GAE respectively. Other timepoints can be found in the supplemental material.

VAS pain scores (Fig. 5) were also reported and analyzed at similar outcome times as the total WOMAC scores. It was also the most consistently used and reported outcome measure across studies to assess pain. Analysis at 1 week showed an effect size of 2.10 (95% CI, 0.13–4.07), indicating the decrease in pain starts early (supplementary data). The effect sizes at 1 month, 3 months, 6 months, and 12 months were 2.06 (95% CI, 1.35–2.76) (p = 0.00), 2.13 (95% CI, 1.39–2.87) (p = 0.00), 2.38 (95%CI, 1.85–2.90) (p = 0.00), 2.09 (95%CI, 0.91–3.28) (p = 0.00) respectively, compared to baseline. This result showed a progressively increasing and sustained reduction in pain after GAE over time. An SMD of 0 indicates that a treatment is as effective as a placebo/no treatment. A positive SMD indicates that the treatment is more effective than a placebo/no treatment, while a negative SMD indicates that the treatment is less effective than a placebo. SMD was used due to its greater generalizability over the Mean difference²⁴ since the variant of the WOMAC scale used was unclear in various studies. An SMD of 0.20, 0.50 and 0.80 indicate small, moderate and large effect sizes, respectively.²⁵

The WOMAC pain subscale scores (Fig. 7) were also reported and analyzed at similar temporal outcome points as the total WOMAC scores and VAS pain scores. Since only a few studies reported the outcomes on the WOMAC pain subscales, the pooled effect sizes are only available for 1, 3, and 24 months. The remaining time points are single studies that prevent a pooled effect size analysis. The pooled effect sizes at 3 and 24 months, respectively, are 2.91 (95%CI, 1.03–4.80) (p = 0.00) and 2.27 (95% CI, 0.31–4.22) (p = 0.02), indicating a significant and sustainable improvement in the WOMAC pain score after GAE. The heterogeneity in the results is likely due to the varying study designs, outcome evaluation endpoints, variable embolic agents, definitions of technical success, and sample sizes. Given that these studies were evaluated with the random-effects model, the pooled effect size might be more conservative than the actual pooled effect size.

3. Discussion

GAE is a novel treatment for the treatment of OA-associated pain. This Quantitative-analysis showed that GAE is an effective and safe method for pain and function improvement in patients with knee osteoarthritis in the short and medium term.

The first breakthrough and the foundation of GAE as a treatment for OA was established by Okuno et al.⁷ when the presence of abnormal neovessels inside and surrounding the synovium was reported. Subsequently, Okuno et al. performed a more elaborate study with 95 knees showing the efficacy of GAE for OA.²¹ Subsequently, a 2017 study by the same group, also using IP/CS, reported durable improvements in OA-associated knee pain and function. This study established GAE as a serious contender for the interventional standard of care for early-grade OA.²¹

Multiple studies were then conducted that investigated the efficacy of GAE in treating early OA pain. Padia et al. contributed significantly to work on GAE in the United States by using 100 μm Embozenes and reported that the total WOMAC and VAS pain scores decreased by 61% and 67% at 12 months from a median baseline of 52 (out of 96) and 8 (out of 10), respectively (Table 2A, Table 2B, Table 2C).⁸ Little et al. performed GAE with Embospheres and found significant improvement in the VAS (60 from baseline to 30 at 6 months and 45 at 12 months, all out of 100) and KOOS score compared to the baseline (Table 2A, Table 2B, Table 2C). Bagla et al. performed multiple prospective GAE studies (Table 2A, Table 2B, Table 2C).^9,20 In early 2022, they published a randomized clinical trial with a crossover design, with promising results for GAE as a treatment.⁹ This trial showed that GAE was superior to placebo by reporting a significant improvement in the mean total WOMAC score at 1 month, compared to the sham group.⁹ This study was the first interventional design with a control group for GAE(9). This Quantitative-analysis found that GAE improved pain and physical function in the short term, with a sustained effect in the medium term.

Table 2A.

Total WOMAC scores of each study.

Authors	Year	sample size	Total WOMAC score
			pre-procedure		1 month		3 months		6 months		12 months		24 months
			Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	mean	SD
Bagla et al.	2020	20	61	12	24	17	31	21	31	26	NR	NR	NR	NR
Okuno et al.	2015	14	47.3	5.8	11.6	5.4	6.3	6	NR	NR	NR	NR	NR	NR
Okuno et al.	2017	95	43	8.3	24	14	14.8	11	11.2	10	8.2	8.5	6.2	6.4
Piechowiak et al.	2017	5	38.6	9.4	19.8	9.4	21.2	9.4	NR	NR	NR	NR	NR	NR
Padia et al.	2021	40	52.4	2.8	27.8	3.3	23.32	3.5	24.3	3.5	22.7	3.4	NR	NR
Padia et al.	2021	40	52 (median)	NR	NR	NR	NR	NR	NR	NR	21 (median)		NR	NR
Bagla et al.	2022	21	65.2	14.8	38.75	25.9	27.5	24.2	28.2	25.5	18.6	16.1	NR	NR
Jalaeian et al.	2021	33	46.6	15.2	NR	NR	26.8	19.3	NR	NR	NR	NR	29.2	19.6
Kumar	2020	NR	NR	NR	−18 compared to baseline	NR	NR	NR	NR	NR	NR	NR	NR	NR
Lauko et al.	2020	1	44	NR	5	NR	4	NR	3	NR	NR	NR	NR	NR

Legend: SD – Standard Deviation. NR = Not Reported, the study in question did not record the outcome at that particular time. Lee et al., Landers et al., and Little et al. did not use the Total WOMAC score as an outcome measure, hence these studies have not been included in this table.

Table 2B.

Pain WOMAC scores of each study.

Authors	Year	Pain WOMAC (0–20)
		pre-procedure		1 month		3 months		6 months		12 months		24 months
		Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	mean	SD
Bagla et al.	2020	14		5	NR	7	NR	5	NR	NR	NR	NR	NR
Okuno et al.	2015	12.2	1.9	3.3	2.1	1.7	2.2	NR	NR	NR	NR	NR	NR
Okuno et al.	2017	12.1	2.3	6.2	4	4.4	3.5	3.7	1.8	3	3.1	2.6	3.4
Padia et al.	2021	11	3.5	−6.0 (−17 to 3)	NR	−7 (−15 to 6)	NR	−7 (−14 to 6)	NR	−7.5 (−14 to 6)	NR	NR	NR
Jalaeian et al.	2021	12.0	3.0	NR	NR	6.0	4.6	NR	NR	NR	NR	7.0	4.6

Legend: SD – Standard Deviation. NR = Not Reported, the study in question did not record the outcome at that particular time.

The studies not included in this table did not report the Pain WOMAC at any time point.

Table 2C.

Pain VAS scores of each study.

Authors	Year	VAS pain
		pre-procedure		1 week		1 month		3 months		6 months		12 months		24 months
		mean	SD	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD
Bagla et al.	2020	76	14	NR	NR	22	19	34	26	31	28	NR	NR	NR	NR
Okuno et al.	2015	70	5	29	17	21	16	13	15	NR	NR	NR	NR	NR	NR
Okuno et al.	2017	72	16	NR	NR	38	23	29	22	19	21	13	21	14	17
Lee et al.	2019	5.5	2.2	3.1	1.9	2.9	1.7	2.2	1.7	1.9	1.5	1.8	2.1	NR	NR
		6.3	2.2	4.1	2.1	4.4	2.1	5.4	2	5.9	2.1	5.3	1.1	NR	NR
Piechowiak et al.	2017	6.7 ± 1.6	1.6	NR	NR	4	NR	3.8	1.6	NR	NR	NR	NR	NR	NR
Padia et al.	2021	NR	NR	NR	NR	−4.5 (−9 to 2)	NR	NR	NR	−5.0 (−8.5 to −1)		−5.0 (−9 to 0)	NR	NR	NR
Padia et al.	2021	80 (median)	NR	NR	NR	NR	NR	NR	NR	NR	NR	30(median)	NR	NR	NR
Bagla et al.	2022	80.3	11.3	NR	NR	33.8	28.5	28.8	29.5	19.9	27.4	20.3	25.3	NR	NR
Little et al.	2021	60	20	NR	NR	32	25	30	24	NR	NR	45	30	NR	NR
Kumar	2020	7.6/10	NR	NR	NR	3 out of 10	NR	NR	NR	NR	NR	NR	NR	NR	NR

The trend shows an improvement in Pain VAS from baseline to each of the recorded time-points. NR = Not Reported, the study in question did not record the outcome at that particular time.

Legend: SD – Standard Deviation. Jalaeian, Landers and Lauko did not report outcomes in VAS scores, hence they have not been included in this table.

While most studies excluded severe OA patients, Lee et al. included KL grade 4 (severe OA) patients. They reported that the improvement in the severe OA cohort was transient, and the total WOMAC score climbed back to the pre-procedural level at 3 months. The KL grade 1–3 (mild to moderate OA) cohort in the same study showed a sustained improvement in the total WOMAC score at 3 months.¹⁹ This distinction may reflect a disease process too advanced for GAE to be as effective as it is for earlier OA grades.¹⁹ This review did not find any evidence regarding KL grade 4 treatment other than Lee et al., making comparisons difficult.

Okuno et al. utilized Imipenem/Cilastatin (Primaxin, Merck & CO., Whitehouse Station, New Jersey), an antibiotic that acts as an embolic agent in their 2015 and 2017 study (In the latter study, Embozene was used for patients who were allergic to the antibiotic). IP/CS, on mixing with radiocontrast in standardized mixtures, forms embolic particles of 70–150μm,¹⁰ which causes cessation of flow for up to 48 h.¹⁰ While IP/CS takes several weeks to attain its full clinical effect,¹⁰ it is also associated with a decreased non-target embolization rate.²⁶ In addition, while IP/CS has proven effective, it is transiently embolic due to its solubility and is contraindicated in patients with hypersensitivity or allergy to antibiotics or valproic acid.¹⁰ The lack of availability of this antibiotic in the United States has led to using alternative agents such as Embosphere (Merit Medical, USA), PVA (Boston Scientific, USA), and Embozene (Varian Inc., USA) (Table 1). Follow-up with the patients showed no clinically significant difference between patients treated with IP/CS vs. Embozene, a finding confirmed by later studies.¹⁸

This systematic review evaluated the safety of GAE by assessing and grading the reported adverse events. Most adverse effects were mild and scored A (No Therapy needed and no consequences of Adverse Event) or B (Nominal therapy needed up to admission for observation, no consequences of Adverse event) according to the SIR criteria for adverse events.²⁷ Adverse effects such as transient cutaneous ischemia, redness, and hematomas are minor and self-resolving. These commonly reported events occurred in, around, and over the knee joint. Most of these minor complications are due to either the inherent nature of embolization, producing ischemic changes in the surrounding tissue, or the result of non-target embolization.^8,20 Non-target embolization can happen due to extensive collateral circulation or reflux of particles away from the catheter tip.⁷

East Asian studies had a mean BMI between 25 and 27, while the western studies almost uniformly had mean BMIs above 29 (Table 2A, Table 2B, Table 2C). This may reflect the higher obesity rates in the west compared to the east; Obesity is a risk factor for OA, and this trend may be explored in further detail in future studies.

Most studies used MRI for pre-operative OA evaluation and GAE suitability (Table 1). While The role of CT scans in Osteoarthritis imaging is well established, given the ability of CT scans to visualize calcified cartilage, and trabecular subchondral bone, while offering improved visualization of subchondral bone cysts and osteophytes compared with MRI.²⁸ CT scans are faster, cheaper, and more widely available than MRIs.²⁹ MRI can provide detailed visualizations of the soft tissue changes in OA, and the results can be quantified by the WORMS score.^14,29 MRI is limited to situations where the diagnosis is in doubt, the OA needs to be classified by subtype, or pre-procedural planning for GAE is necessary. Because of the ability of MRI to visualize soft tissue, most studies used MRI to evaluate suitability for GAE. The value of pre-operative MRI for assessing post-GAE outcomes and technical feasibility has been documented.^30,31 Effusion synovitis, bone marrow lesions, full-thickness cartilage defects, osteophytes, and cartilage surface area scores on pre-operative MRI at baseline are associated with poor GAE outcomes in the medium term.^30,31

Studies have reported the higher cost of GAE compared to NSAIDs while showing that GAE is more cost favorable than COX-2 inhibitors.³² Costs related to GAE may decrease in the future due to wider adoptions, innovations, decreased costs of disposables, and other factors determining the final cost to the patient.

Two studies in this analysis utilized the KOOS (Knee injury and Osteoarthritis Score) score, a more comprehensive scale compared to WOMAC,⁴ to measure their outcomes (Landers et al. and Little et al.).^15,17 Most studies evaluated utilized the WOMAC scores for measuring outcomes. This distribution has resulted in an inability to combine both scores in a pooled analysis. Investigators' uniform use of either metric in the future may yield a more robust and comparable dataset. Another trend was the lack of reporting on the type of WOMAC scale since several versions are available.³³ This lack of reporting made the analysis more challenging and necessitated using WMD over Mean Difference as a summary measure.

Based on this analysis, the authors concluded that GAE might be offered to patients suffering from mild to moderate OA who cannot undergo TKA or have failed conservative therapy. It remains an open question whether GAE should be offered to eligible and willing patients for TKA. This study contributes to the literature by providing a pooled analysis of the studies published so far, evaluating the safety and efficacy of GAE. This review may help clinicians make informed choices regarding the needs of individual patients, especially those who are poor surgical candidates or have refractory OA. This review may also be useful to policymakers and administrators who can make informed decisions while allocating resources for future research. Future research may also benefit from the limitations of this review and the studies included in it. More optimal clinical study designs can be formulated, such as large, multi-centric RCTs.

This study has several limitations. Overall, only 399 knees were evaluated. The WOMAC score, especially the physical function component, has knee interdependence. The use of bilateral GAE may distort the actual effect size of individual GAE procedures. Only 1 RCT was available at the time of writing, and the rest are single-arm studies. Only 1 study included KL grade 4, and conclusions cannot be drawn about the efficacy of GAE in severe OA. A Random effects model may lead to a more conservative pooled effects estimate than the actual effect size. Randomized clinical trials which compare treatments (and placebo) head-on are needed to evaluate the efficacy of GAE vs. other treatments, such as intraarticular injections of PRP or steroids³⁴ or TKA. A number of trials are currently underway and will provide further evidence regarding the efficacy of GAE for OA related Pain.³⁵

4. Conclusion

This Systematic Review and Quantitative-analysis found that GAE is a safe and effective procedure for early and low-grade OA in at least the short and medium term. Different embolic agents are available and can be used without significant adverse effects. Randomized controlled trials are only beginning to come out as of this writing, and more robust evidence in the form of RCTs is needed to cement the role of GAE in mild to moderate OA.

Sources of support

This study was not funded by any source.

Financial disclosure and conflict of interest

Ansh Bhatia: I affirm that I have no financial affiliation (including research funding) or involvement with any commercial organization that has a direct financial interest in any matter included in this manuscript, except as disclosed in an attachment and cited in the manuscript. Any other conflict of interest (ie, personal associations or involvement as a director, officer, or expert witness) is also disclosed in an attachment.

Shivank Bhatia has consulted for embolx and has received Research funding from Merit medical inc which has been addressed here. I affirm that I have no financial affiliation (including research funding) or involvement with any commercial organization that has a direct financial interest in any matter included in this manuscript, except as disclosed in an attachment and cited in the manuscript. Any other conflict of interest (ie, personal associations or involvement as a director, officer, or expert witness) is also disclosed in an attachment.

Statement of Institutional Review Board or Ethics Committee approval of the study protocol: Not applicable.

Funding/sponsorship

“This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors”.

Informed consent (patient/guardian)

Not applicable.

Institutional ethical committee approval

Not applicable.

Authors contribution

Ansh Bhatia: Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Software; Validation; Roles/Writing – original draft; Writing – review & editing.

Shivank Bhatia: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing.

Findings

GAE is a safe and efficacious, minimally invasive technique for the treatment of Osteoarthritis associated knee pain in the short and medium term.

Implications

GAE should be considered a viable alternative for treating OA-associated knee pain and should be offered to those patients who are not surgical candidates for total knee replacements or have mild to moderate OA.

Caution

Almost all studies included in this Systematic review are single-arm studies, hence the need for large-scale, multi-centric, Randomized Controlled Trials to validate the findings of previous studies and account for the placebo effect.

Declaration of competing interest

Shivank Bhatia reports a relationship with Embolx Inc that includes: consulting or advisory. Shivank Bhatia reports a relationship with Merit Medical Systems Inc that includes: consulting or advisory, funding grants, and travel reimbursement.

Appendix A. Supplementary data

The following are the Supplementary data to this article.