A network meta-analysis on the efficacy and safety of thermal and nonthermal endovenous ablation treatments

Abstract

Objective

We assessed the mid-term efficacy and safety of thermal and nonthermal endovenous ablation for the treatment of lower limb superficial venous insufficiency.

Methods

We performed a systematic review in accordance with the PRISMA (preferred reporting items for systematic reviews and meta-analyses) statement and a Bayesian network meta-analysis. The primary end points were great saphenous vein (GSV) closure and venous clinical severity score (VCSS) improvement. A meta-regression using GSV diameter as a covariate was undertaken for the two primary end points.

Results

We included 14 studies and 4177 patients, with a mean follow-up of 25.7 months. Radiofrequency ablation (RFA; odds ratio [OR], 3.99; 95% confidence interval [CI], 1.82-10.53), cyanoacrylate ablation (CAC; OR, 3.09; 95% CI, 1.35-8.37), and endovenous laser ablation (EVLA; OR, 2.72; 95% CI, 1.23-7.38) displayed increased odds for GSV closure compared with mechanochemical ablation (MOCA). MOCA inferiority compared with RFA (mean difference [MD], 0.96; 95% CI, 0.71-1.20), EVLA (MD, 0.94; 95% CI, 0.61-1.24), and CAC (MD, 0.89; 95% CI, 0.65-1.15) was also depicted regarding VCSS improvement. EVLA resulted in an increased risk of postoperative paresthesia compared with MOCA (risk ratio [RR], 9.61; 95% CI, 2.32-62.29), CAC (RR, 7.90; 95% CI, 2.44-38.16), and RFA (RR, 6.96; 95% CI, 2.31-28.04). Although the overall analysis identified nonstatistically significant differences for Aberdeen varicose vein questionnaire score improvement, thrombophlebitis, ecchymosis, and pain, further investigation revealed an increase pain profile for EVLA at 1470 nm compared with RFA (MD, 3.22; 95% CI, 0.93-5.47) and CAC (MD, 3.04; 95% CI, 1.05-4.97). A sensitivity analysis displayed a persistent underperformance of MOCA compared with RFA (OR, 4.33; 95% CI, 1.15-55.54) for GSV closure and both RFA (MD, 0.99; 95% CI, 0.22-1.77) and CAC (MD, 0.84; 95% CI, 0.08-1.65) regarding VCCS improvement. Although no regression model reached statistical significance, the GSV closure regression model revealed a trend for considerably decreased efficacy for both CAC and MOCA with larger GSV diameters compared with RFA and EVLA.

Conclusions

Although our analysis has produced skepticism regarding the efficacy of MOCA in the mid-term period for VCSS improvement and GSV closure rates, CAC showed equivalent results compared with both RFA and EVLA. Additionally, CAC displayed a decreased risk of postprocedural paresthesia and pigmentation and induration compared with EVLA. Also, both RFA and CAC had an improved pain profile compared with EVLA 1470 nm. The potential underperformance of nonthermal, nontumescent ablation modalities in ablating large GSVs necessitates further research.

Chronic venous insufficiency is a common medical condition with significant socioeconomic implications. Interventions for the treatment of chronic venous insufficiency have consistently displayed quality of life and cost-effectiveness benefits compared with conservative therapy. Thus, the European Society for Vascular Surgery in their recently published guidelines have strongly recommended the performance of interventions for patients presenting with symptomatic CEAP (clinical, etiology, anatomy, pathophysiology) class >2 superficial vein insufficiency (class I; level of recommendation, B).

Because of their proven efficacy and safety, thermal ablation techniques have largely replaced the traditional high ligation and stripping method as the first-line treatment (class I; level of recommendation, A). The two most widely used approaches are radiofrequency ablation (RFA) and endovenous laser ablation (EVLA). Other thermal ablation modalities include endovenous steam ablation (EVSA) and endovenous microwave ablation (EVMA). Despite the success of thermal-based methods, factors such as the need for tumescence anesthesia and complications such as bruising, nerve injury, and endothermal heat-induced thrombosis (EHIT) have propelled the development of novel, nonthermal, nontumescent (NTNT) ablation techniques.

The two novel NTNT venous ablation alternatives include cyanoacrylate glue ablation (CAC) and mechanochemical ablation (MOCA). CAC is a catheter-based technique in which the catheter is inserted into the vein, releasing an adhesive substance (n-butyl cyanoacrylate), which then polymerizes and solidifies, resulting in inflammation and vein ablation. Despite its advantages and that several reported studies have displayed its noninferiority compared with thermal methods, CAC has its own unique complications.^1,2

MOCA is a bimodal technique. First, a rotating wire or hook causes endothelial and medial abrasion injury to the venous wall.³ Next, sclerosant foam (usually polidocanol or sodium tetradecyl sulfate) is injected.³ Although a few studies have shown the safety and efficacy of MOCA, its performance compared with the other available modalities has not yet been investigated.

Both CAC and MOCA are reserved as second-line treatment methods, with a recommendation status of class IIa (level of recommendation, A) and class IIb (level of recommendation, A), respectively.

Methods

Eligibility criteria

We included studies reporting on ablation of the great saphenous vein (GSV) for the treatment of lower limb symptomatic venous insufficiency classified as CEAP class ≥2 and the use of any of the following: RFA, EVLA, CAC, MOCA, EVSA, EVMA, or foam sclerotherapy (FS). Randomized control trials (RCTs) and comparative cases series providing extractable data on any of the two primary end points were included. When multiple reports from the same institution were identified, the latest report was included.

We excluded studies reporting on GSV ablation via stripping and ligation, studies targeting the treatment of incompetent perforates, studies with a patient population of <20 patients per treatment arm, studies with follow-up of <12 months, and studies including patients with previous GSV interventions. Additionally, we excluded studies reporting on combined procedures on the same limb and studies exclusively depicting the treatment of the small saphenous vein and/or accessory saphenous vein. We also excluded studies that had failed to provide extractable data and/or a definition of GSV closure. Moreover, we excluded studies that had not reported the CEAP classification and studies of patients classified as having CEAP class <2. Finally, studies depicting comparisons failing to form a “closed loop” network and consequently not allowing for the evaluation of the consistency assumption (open loop networks are inconsistent by definition) were also excluded.⁴

Information sources and search strategy

A systematic review in accordance with the PRISMA (preferred reporting items for systematic reviews and meta-analyses) and the PRISMA for network meta-analyses (NMAs) extension statements was undertaken (Supplementary Fig 1, online only).^5,6 Following a prefabricated protocol, two of us (V.B. and A.B.) conducted a systematic search of Medline, Scopus, and Google Scholar for articles published by August 2022. An integrated vocabulary with keywords was used during the investigation. The keywords and terms implemented during the search process included the following: “thermal endovenous ablation,” “non thermal endovenous ablation,” “thermal ablation” AND “vein,” “thermal ablation” AND “venous,” “non thermal ablation” AND “vein,” “non thermal ablation” AND “venous,” “EVLA” OR “endovenous laser ablation,” “radiofrequency ablation” OR “RFA” AND “venous,” “radiofrequency endovenous ablation” OR “RFA” AND “vein,” “mechanochemical endovenous ablation,” “MOCA” AND “vein,” “MOCA” AND “venous,” “cyanoacrylate ablation,” “glue ablation,” “n-butyl-2-cyanoacrylate” OR “NBCA” AND “venous,” “n-butyl-2-cyanoacrylate” OR “NBCA” AND “vein,” “microwave ablation” OR “EVMA” OR “MWA” AND “vein,” “microwave ablation” OR “EVMA” OR “MWA” AND “venous,” “steam ablation” OR “EVSA,” “foam sclerotherapy,” “ultrasound-guided foam sclerotherapy” OR “UGFS.” The protocol is available on reasonable request from the corresponding author.

Supplementary Fig 1 (online only).

PRISMA (preferred reporting items for systematic reviews and meta-analyses) 2020 flow chart for new systematic reviews, which included searches of databases and registers only. CEAP, Clinical, etiology, anatomy, pathophysiology; EVLA, endovenous laser ablation; FS, foam sclerotherapy; GSV, great saphenous vein.

Selection and data collection process

The two responsible investigators evaluated the publications returned from the search. Articles with titles and abstracts that seemed to fulfill the inclusion criteria were collected and processed using identical terms. Where disagreement emerged, an agreement was reached through discussion with a third author (A.K.).

Data items

The primary end points were GSV closure and venous clinical severity score (VCSS) improvement by the end of follow-up. The secondary end points included Aberdeen varicose vein questionnaire (AVVQ) score improvement by the end of follow-up, pain experienced during the intervention measured using the visual analog scale (VAS), paresthesia, deep vein thrombosis (DVT), superficial thrombophlebitis, pigmentation and induration, and ecchymosis.

Definitions

GSV occlusion was defined as closure (incompressible vein with no flow) for the entire segment of the treated vein, with no evidence of patency for venous segments >5 cm long during duplex ultrasound examinations. The revised version of the VCSS was implemented by all included studies.⁷ Neurologic disturbances included sensory disturbances and saphenous neuralgia. The VAS used a score range from 0 (no pain) to 10 (most severe pain). Studies in which other forms of anesthesia in addition to local tumescent anesthesia were used (eg, spinal anesthesia) were excluded from the analysis of pain outcomes.

Effect measures and synthesis methods

We initially extracted raw data from the included studies either in the form of the mean ± standard deviation or as the number of events per the total number of interventions, depending on the outcome of interest. Because of the already standardized format of the VCSS, AVVQ, and VAS (score range, 0-10), we proceeded with the calculation of the mean differences (MDs) and not the standardized MDs. After data extraction, we performed a Bayesian NMA, applying the Markov chain Monte Carlo simulation with vague priors for posterior distribution estimation. The generalized linear model with four chains, burn-ins of 50,000 pursued by 100,000 iterations with 10,000 adaptations was applied. We evaluated network geometry operating network plots. Model selection between fixed and random effects models was based on leverage plot examination and appraisal of deviance information criterion values. We evaluated model convergence using trace and Gelman-Rubin-Brooks plot examination. Model inconsistency was assessed using leverage and posterior deviance plots, deviance information criterion estimates, and the node-splitting technique, for which P values ≤ .05 represented statistically significant variations between direct and indirect evidence. The results of the NMA were illustrated using heat plots, for which, according to the end point under investigation, either MDs, odds ratios (ORs), or risk ratios (RRs) and their corresponding 95% confidence intervals (CIs) are presented. Intervention ranking was processed using the surface under the cumulative ranking (SUCRA) score and visualized using SUCRA plots. We performed a network meta-regression using the GSV diameter as a covariate and exchangeable treatment–interaction model priors. Regression results are presented through forest plots.^8,9 We additionally performed a sensitivity analysis for all end points, excluding all non-RCTs. The risk of bias was assessed using the ROBINS-I (risk of bias in non-randomized studies of interventions) and the RoB 2 (revised Cochrane risk-of-bias tool for randomized trials).^10,11

Results

Study characteristics

Fourteen studies, including seven RCTs, two prospective case series, and five retrospective case series, with 4177 patients and a mean follow-up of 25.7 months, were included in the present analysis. Five studies had compared RFA to CAC, two RFA to MOCA, one RFA to EVLA, one EVLA to CAC, one MOCA to CAC, and one MOCA to EVLA. In addition, two studies had performed three-arm comparisons of the performance of EVLA, RFA, and CAC, and one of EVLA, RFA, and MOCA.12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25 Our research failed to identify eligible studies depicting EVSA and EVMA interventions or studies comparing CAC, MOCA, or RFA to FS (only EVLA to FS comparisons met the inclusion criteria). Thus, EVSA, EVMA, and FS were not included in the analysis (Table).

Table.

Study characteristics

Investigator	Study type	Treatment	Product	Patients, no.	Follow-up, months	GSV diameter, mm	CEAP class
							C2	C3	C4	C5	C6
Morrison et al,22 2020	RCT				60		125	68	29	0	0
		RFA	ClosureFast	114		5.91
		CAC	VenaSeal	108		6.44
Eroglu et al,16 2018	RCT				24
		CAC	VariClose	168		7.6	4	93	71	0	0
		RFA	ClosureFast	149		7.8	2	86	57	3	1
		EVLA	1470 nm	139		8	4	77	58	0	0
Vähäaho et al,24 2021	RCT				36
		MOCA	ClariVein	59		6.7	32	14	13	0	0
		EVLA	1470 nm	34		6.6	20	6	7	0	0
		RFA	ClosureFast	32		6.5	18	7	7	0	0
Holewijn et al,19 2019	RCT				24
		MOCA	ClariVein	102		6.2	7	63	33	2	0
		RFA	ClosureFast	103		6.8	5	69	27	3	0
Belramman 2022	RCT				12
		MOCA	ClariVein	80		5	25	19	27	3	6
		CAC	VenaSeal	83		5.87	23	22	27	1	5
Mohamed et al,21 2021	RCT				12
		MOCA	ClariVein	71		8.6	21	23	26	3	1
		EVLA	1470 nm	72		9	15	29	25	6	0
Hamann et al,17 2019	RCT				12
		EVLA	980 nm	148		7.85	51	76	14	3	3
		RFA	ClosureFast	150		7.55	49	82	17	2	0
Bozkurt et al,14 2016	PCS				12
		EVLA	1470 nm	156		7.1	119	33	4	0	0
		CAC	VariClose	154		7.2	104	38	12	0	0
Daylan 2019	RCS				60
		RFA	ClosureFast	634		>5.5	C2-C6
		CAC	VenaBlock	246		>5.5	C2-C6
Kubat et al,20 2019	RCS				12
		EVLA	980-1470 nm	257		13.3	155	68	22	12	0
		RFA	ClosureFast	249		12.9	142	69	32	6	0
		CAC	VariClose	79		13.1	47	23	6	3	0
Ovalı et al,23 2019	RCS				12
		RFA	ClosureRFG-ClosureFast	128		7.2	115			13
		CAC	VenaBlock	116		7	102			14
Ay et al,12 2021	PCS				12
		RFA	ClosureFast	70		7.6	57			13
		CAC	VariClose	85		7.9	67			18
El Kilic et al,19 2022	RCS				60
		EVLA	1470 nm	77		7.1	51	20	6	0	0
		CAC	VariClose	73		7.2	51	19	3	0	0
		RFA	ClosureFast	82		6.6	59	19	4	0	0
Bademci et al,25 2019	RCS				12
		CAC	VariClose	75		7	49	19	7	0	0
		RFA	ClosureFast	84		7.25	54	20	10	0	0

CAC, Cyanoacrylate ablation; CEAP, clinical, etiology, anatomy, pathophysiology; EVLA, endovenous laser ablation; GSV, great saphenous vein; MOCA, mechanochemical ablation; PCS, prospective case series; RCS, retrospective case series; RCT, randomized controlled trial; RFA, radiofrequency ablation; RFG, radiofrequency generator.

Risk of bias

The overall risk of bias for the seven included RCTs was moderate, with only one study displaying a high risk of bias.²⁴ The risk of bias for the nonrandomized studies (NRSs) was low to moderate, with only one study depicting a serious risk of bias (Supplementary Fig 2, online only).^1,5

Supplementary Fig 2 (online only).

Risk of bias determined using RoB 2 (version 2 of the Cochrane risk of bias tool for randomized trials; a) and ROBINS-I (risk of bias in non-randomized studies; b).

Network meta-analysis

GSV closure

The GSV closure network included all 14 studies and 4177 patients.12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25 Of the 14 studies, 7 had defined GSV closure as an incompressible vein with no flow for the entire length of the treated vein, and 7 had defined successful closure as no evidence of patency for venous segments >5 cm long (Fig 1, a). We identified nonstatistically significant differences between RFA, EVLA, and CAC. RFA (OR, 3.99; 95% CI, 1.82-10.53), CAC (OR, 3.09; 95% CI, 1.35- 8.37), and EVLA (OR, 2.72; 95% CI, 1.23-7.38) displayed superior outcomes compared with MOCA regarding the GSV closure rates by the end of follow-up (Fig 1, b). The SUCRA probability ranking graded RFA (SUCRA score, 81.32) as the most effective intervention, followed by CAC (SUCRA score, 12.02; Fig 2, a). During regression analysis, although our regression model failed to reach statistical significance (β = 0.43; 95% CI, −1.42 to 1.98), plot exploration revealed a tendency for both CAC and MOCA to have dramatically decreased efficacy for larger GSV diameters compared with RFA and EVLA (Fig 2, b).

Fig 1.

Great saphenous vein (GSV) closure by end of follow-up: diagram showing network (a) and heat plot showing odds ratios (ORs) and 95% confidence intervals (CIs; b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

Fig 2.

Great saphenous vein (GSV) closure by the end of follow-up: surface under the cumulative ranking (SUCRA) curve (a) and regression plot (b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

The sensitivity analysis included seven RCTs with 1236 patients. Although RFA (OR, 4.33; 95% CI, 1.15-55.54) sustained its statistically significant superior outcomes compared with MOCA, EVLA (OR, 3.98; 95% CI, 0.96-48.05) and CAC (OR, 2.56; 95% CI, 0.29-20.51) failed to display similar consistency.^13,16, 17, 18^,21,22,24

VCSS improvement

The VCSS network included 11 studies and 3453 patients (Fig 3, a).12, 13, 14, 15^,17, 18, 19, 20^,22,23,25 MOCA displayed statistically significant inferior results concerning VCSS improvement from baseline to the end of follow-up, resulting in the loss of approximately one VCSS improvement point compared with the three remaining interventions (RFA: MD, 0.96; 95% CI, 0.71-1.20; EVLA: MD, 0.94; 95% CI, 0.61-1.24; and CAC: MD, 0.89; 95% CI, 0.65-1.15; Fig 3, b). According to the SUCRA values, RFA (SUCRA score, 51.08) had the highest probability of being the most efficacious modality regarding VCSS reduction, succeeded by EVLA (SUCRA score, 38.00; Fig 4, a). Although our regression analysis failed to predict the outcomes (β = 0.20; 95% CI, −1.26 to 1.77), RFA and CAC displayed a trend toward outperforming EVLA for a GSV diameter >8 mm and >9 mm, respectively (Fig 4, b).

Fig 3.

Venous clinical severity score (VCSS): diagram showing network (a) and heat plot showing mean differences (MDs; b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

Fig 4.

Venous clinical severity score (VCSS): surface under the cumulative ranking (SUCRA) curve (a) and regression plot (b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

The sensitivity analysis concerning VCSS reduction included three studies comparing RFA, CAC, and MOCA, with 590 patients.^13,18,22 The analysis displayed a persistent pattern of MOCA inferiority compared with both RFA (MD, 0.99; 95% CI, 0.22-1.77) and CAC (MD, 0.84; 95% CI, 0.08-1.65).^13,18,22

Pain

The pain network consisted of six studies and 1367 patients (Supplementary Fig 3, a, online only).^13,14,17, 18, 19^,25 Although EVLA increased intraprocedural pain compared with CAC (MD, 2.19; 95% CI, −0.40 to 4.79), RFA (MD, 1.78; 95% CI, −0.78 to 4.39), and MOCA (MD, 1.52; 95% CI, −2.20 to 5.27), the differences failed to reach statistical significance (Supplementary Fig 3, b, online only). SUCRA ranked CAC (SUCRA score, 49.26) and RFA (SUCRA score, 24.89) as the two interventions most likely to cause the least amount of pain. Further investigation on the effects of EVLA wavelengths on pain outcomes resulted in a network of five studies with 1069 patients investigating EVLA 1470 nm (EVLA 980 nm failed to form a closed loop), RFA, CAC, and MOCA. EVLA 1470 nm displayed a statistically significant increased risk of pain compared with both RFA (MD, 3.22; 95% CI, 0.93-5.47) and CAC (MD, 3.04; 95% CI, 1.05-4.97). A sensitivity analysis for pain was not feasible because of the limited number of available studies consequently failing to form a closed loop network.

Supplementary Fig 3 (online only).

Pain experienced during intervention: diagram showing network (a) and heat plot showing mean differences (MDs; b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

Paresthesia

Ten studies with 2380 patients had reported on postintervention paresthesia (Supplementary Fig 4, a, online only).^12,14,18, 19, 20, 21, 22, 23, 24, 25 EVLA significantly increased the risk of postintervention paresthesia compared with MOCA (RR, 9.61; 95% CI, 2.32-62.29), CAC (RR, 7.90; 95% CI, 2.44-38.16), and RFA (RR, 6.96; 95% CI, 2.31-28.04; Supplementary Fig 4, b, online only). SUCRA ranked MOCA (SUCRA score, 52.72) and CAC (SUCRA score, 31.58) as the two interventions less likely to result in paresthesia.

Supplementary Fig 4 (online only).

Postprocedural paresthesia: diagram showing network (a) and heat plot showing risk ratios (RRs; b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

The sensitivity analysis network included three studies with 461 patients that had compared ELVA, RFA, and MOCA. Although EVLA continued to portray an increased risk of paresthesia compared with MOCA (RR, 8.50; 95% CI, 1.03-141.44), nonstatistically significant differences were identified between EVLA and RFA and between RFA and MOCA.^18,21,24

AVVQ score improvement

Eight studies with 2419 patients reported the AVVQ scores.13, 14, 15^,17,21, 22, 23^,25 Although EVLA displayed an increased AVVQ score reduction compared with CAC (MD, 1.49; 95% CI, −0.46 to 3.41), RFA (MD, 1.14; 95% CI, −0.87 to 3.16), and MOCA (MD, 0.92; 95% CI, −1.41 to 3.17), the outcome differences failed to reach statistical significance. SUCRA ranked EVLA (SUCRA score, 72.07) as the intervention more likely to result in the largest AVVQ score improvement, followed by MOCA (SUCRA score, 17.57) and RFA (SUCRA score, 8.19).

The AVVQ sensitivity analysis network included four studies of 826 patients that had compared EVLA, MOCA, CAC, and RFA. EVLA displayed a statistically significant increase in the AVVQ reduction score compared with CAC (MD, 3.07; 95% CI, 0.76-5.67), with nonstatistically significant differences identified among the remaining interventions.^13,17,21,22

Deep vein thrombosis

Twelve studies and 3138 patients were included in the DVT network.12, 13, 14^,16, 17, 18, 19, 20, 21, 22, 23, 24 The crude rate was 0.09% (3 of 3148), rendering further analysis irrelevant.

Pigmentation and induration

The pigmentation and induration network included nine studies and 2158 patients (Supplementary Fig 5, a, online only).^12,14,18, 19, 20, 21^,23, 24, 25 EVLA increased the risk of pigmentation and induration compared with CAC (RR, 2.85; 95% CI, 1.21-7.06) and RFA (RR, 2.16; 95% CI, 1.03-4.56; Supplementary Fig 5, b, online only). SUCRA ranked CAC (SUCRA score, 76.9) and RFA (SUCRA score, 21.3) as the two interventions less likely to result in pigmentation and induration.

Supplementary Fig 5 (online only).

Pigmentation and induration: diagram showing network (a) and heat plot showing risk ratios (RRs; b). CAC, Cyanoacrylate ablation; EVLA, endovenous laser ablation; MOCA, mechanochemical ablation; RFA, radiofrequency ablation.

The pigmentation and induration sensitivity analysis network included three studies with 461 patients. The three studies had compared RFA, EVLA, and MOCA, with no statistically significant differences between the interventions.^18,21,24

Ecchymosis and superficial thrombophlebitis

The ecchymosis and superficial thrombophlebitis networks included 7 studies with 2149 patients^{12,14,16,20,22,23,25} and 10 studies with 2289 patients,12, 13, 14^,16,18,19,21, 22, 23^,25 respectively. Although failing to reach statistical significance, CAC reduced the risk of ecchymosis compared with both RFA (RR, 0.51; 95% CI, 0.19-1.38) and EVLA (RR, 0.43; 95% CI, 0.11-1.49). Similarly, RFA displayed a nonstatistically significant risk reduction regarding thrombophlebitis compared with CAC (RR, 0.79; 95% CI, 0.34-1.58), MOCA (RR, 0.77; 95% CI, 0.24-3.24), and EVLA (RR, 0.61; 95% CI, 0.20-1.58).

The nonsignificant outcomes persisted in the ecchymosis sensitivity network, which compared RFA, CAC, and MOCA and included two studies and 678 patients,^16,22 and in the thrombophlebitis network, which included five RCTs with 1189 patients and compared EVLA, MOCA, RFA, and CAC.^{13,16,18,21,22}

Discussion

RFA, CAC, and EVLA displayed considerably higher odds regarding GSV closure compared with MOCA. This trend of MOCA inferiority was also displayed in the VCSS. EVLA, RFA, and CAC enhanced VCSS reduction by almost one point compared with MOCA. EVLA increased the risk of paresthesia compared with RFA, CAC, and MOCA and increased the risk of pigmentation and induration compared with RFA and CAC. EVLA 1470 nm increased the pain experienced during the intervention by almost three points compared with CAC and RFA. The reported DVT crude rate was 0.09% (3 of 3148), with one recorded incident of EHIT stage V. Nonstatistically significant differences were observed between the interventions regarding AVVQ, ecchymosis, and superficial thrombophlebitis.

In recent years, several NMAs have attempted to summarize the outcomes regarding the plethora of available interventions for the treatment of GSV insufficiency. The NMA reported by Kolluri et al²⁶ displayed statistically significant superior closure rates for CAC (VenaSeal; Medtronic, Dublin, Ireland) compared with EVLA, MOCA, and RFA at 6 months of follow-up. However, they failed to provide intercomparison outcomes between the remaining interventions.²⁶ Siribumrungwong et al,²⁷ in their network, exhibited statistically significant inferior short-term (maximum follow-up of 1 year) closure rates for MOCA compared with CAC and ultrasound-guided foam sclerotherapy (UGFS) compared with high ligation, open surgery, RFA, and EVLA, an outcome that did not persist for the mid-term period. Additionally, the CAC vs RFA comparison displayed CAC superiority for the short-term period.²⁷ Gasior et al²⁸ exhibited CAC superiority over MOCA and RFA regarding the short-term results (6 weeks), with no statistically significant differences among thermal and nonthermal interventions at 1 year and 2 to 5 years of follow-up.²⁸ Although the two studies with patient follow-up exceeding the short-term period agreed on the nonstatistically significant differences regarding the closure rates in the mid- and long-term periods between thermal and nonthermal techniques and all three studies had their own limitations (eg, Kolluri et al²⁶ only included CAC comparisons, Gasior et al²⁸ included MOCA comparisons for a maximum of 6 weeks of follow-up), they share two major methodologic weaknesses. First, all three networks were open-loop networks and, by definition, inconsistent. Although the elaboration of the transitivity and consistency assumptions were not the subject of our review, the violation of these assumptions threatens the validity of their results.⁴ Additionally, all three networks incorporated diverse closure success definitions (ranging from nonrecanalization to recanalization ≤10 cm in length), potentially resulting in the introduction of a misclassification bias. In our closure network, in an effort to minimize bias, we included studies that had either defined GSV closure as an incompressible vein with no flow or as a vein with evidence of recanalization of ≤5 cm.

Through a consistent network, we displayed the inferior results displayed by MOCA compared with EVLA, RFA, and CAC, with RFA ranking as the intervention most likely to result in the highest occlusion rates. Our results regarding the nonstatistically significant differences among EVLA and RFA are in-line with the reported literature. Regarding MOCA inferiority, similar results have been reported by several studies assessing the short-term efficacy but lacking mid- and long-term data. In our review, which summarized the mid-term data (mean follow-up of 25.7 months), MOCA portrayed results resembling those of UGFS compared with RFA and EVLA.^29,30 Because of the scarcity of studies directly comparing FS with RFA, MOCA, or CAC (only studies comparing EVLA to UGFS met the inclusion criteria), the formation of a closed loop network was not achieved, halting any further assessment of the relationship between MOCA and FS. Nevertheless, given the similarity between MOCA and FS (FS works by injuring the venous wall and MOCA causes an initial mechanical injury, followed by chemical injury), these results are hardly surprising. Whether the addition of mechanical injury could potentially enhance the occlusion rates requires further investigation through studies comparing plain FS to MOCA.

Although the reported data regarding the efficacy of endovenous thermal ablation in relation to the GSV diameter are contradictory, several studies have demonstrated the efficacy of EVLA 1470 nm and RFA in treating large GSVs.31, 32, 33, 34 Regarding EVLA in our network, six of the seven included studies described interventions using the EVLA 1470 nm device. Owing to its inherent water affinity and the consequent noncontact (to the vein wall) ablation ability, EVLA 1470 nm showed superior ablation results compared with EVLA 980 nm, resulting largely in the abandonment of the latter.^20,35 Information regarding the efficacy of CAC and MOCA in ablating larger GSVs is rather limited. However, the potential inferiority of both CAC and MOCA depicted by our model could be explained by the need for protocol adjustments during ablation of large veins when using these techniques, such as placing the patient in a Trendelenburg position in the case of MOCA or the use of extra glue in the case of CAC.³⁶ Additionally, inherent characteristics presented by these systems, such as the nonadjustable one-size rotating wire tip in the case of MOCA (6.5-mm diameter), present further obstacles in ablating veins of various diameters.³⁷

Regarding improvement in the VCSS, and given both the clinical and the objective nature of the VCSS, the trend toward MOCA inferiority displayed by the VCSS model was anticipated, considering the inferior ablation rates. In contrast, the AVVQ results did not reflect the closure rates, suggesting nonstatistically significant differences between the interventions. Correlation between the AVVQ and VCSS has been controversial, with previous studies demonstrating either a strong correlation or no correlation altogether.38, 39, 40 Although the AVVQ and VCSS correlation debate is not the subject of our review, one should consider the contrasting nature of these tools. The VCSS is an objective, physician-endorsed scoring system, and the AVVQ is a subjective, patient-reported questionnaire.

Although EVLA demonstrated an increased pain profile, the differences failed to reach statistical significance. Pain is a well-known disadvantage of EVLA compared with RFA.⁴¹ This was particularly true with the older generation of EVLA devices for which wavelengths <1000 nm constituted the mainstay of treatment. Although various studies have reported the improved pain profile displayed by EVLA 1470 nm compared with EVLA 980 nm, the paucity of available data included in our review did not allow for a head-to-head comparison of the two.^42,43 Nonetheless we managed to separately evaluate the pain profile for EVLA 1470 nm, with the analysis displaying increased intraprocedural pain for EVLA 1470 compared with both RFA and CAC, an outcome in line with the literature. Regarding MOCA and pain, although previous reports have displayed its improved pain profile compared with RFA, in our network, MOCA did not portray statistically significant differences compared with the three remaining interventions.^44,45

Concerning paresthesia, MOCA and CAC displayed a decreased risk of sensory complications compared with EVLA. Given the absence of thermal injury, these outcomes are reasonable and in line with previously reported results.⁴⁶ Additionally, the increased risk exhibited by EVLA compared with RFA adds to the controversy regarding the ultimate thermal technique to avoid sensory disturbances.

Regarding the cosmetic results, nonstatistically significant differences were observed for the ecchymosis and superficial thrombophlebitis networks. CAC and RFA displayed a statistically significant decreased risk compared with EVLA for the composite end point of pigmentation and induration, in line with the literature. The DVT crude rate in our review was 0.09%. One incident occurred during RFA treatment, and another incident (which had resolved within 2 weeks after low-molecular-weight heparin treatment) occurred in a gastrocnemius vein during MOCA ablation of small saphenous vein. The third and last incident, involved thrombus extension to the common femoral vein, classified as EHIT stage V, which had occurred after EVLA ablation.^16,19,21

Although the combination of RCTs and NRSs in an NMA could offer complementary information, investigators are often reluctant to combine both study designs for fear of introducing bias. Nonetheless, several methods to account for NRS bias exist, including naive pooling, using NRSs as prior information, and hierarchical modeling. In favor of creating closed loop networks, we proceeded with naive pooling. We additionally controlled for bias via the risk of bias assessment tool (ROBINS-I), sensitivity analysis, and data compatibility assessment (node splitting). In our review, the risk of bias for NRSs was low to moderate. Also, in the sensitivity analysis (for which only RCTs were included), the pattern of the results persisted, although some outcomes failed to reach statistical significance (presumably because of loss of evidence volume) Finally, through node splitting, we have displayed agreement between the direct and indirect evidence for both RCTs and NRSs.⁴⁷

The analysis of mid-term data (minimum of 12 months of follow-up), through which we have demonstrated the consistent inferiority of MOCA concerning closure and VCSS improvement, constitutes the highlight of our review. Additionally, and although further research is warranted, we found a tendency for NTNT techniques to underperform compared with thermal ablation for closure of larger GSVs.

Study limitations

The two major limitations of our review were the combination of RCTs and comparative NRSs and our inability to further investigate the entire spectrum of available thermal and nonthermal modalities owing to the scarcity of available data.

Conclusions

Albeit our analysis has produced skepticism regarding the efficacy of MOCA in the mid-term period as evaluated by VCSS improvement and GSV closure rates, CAC showed equivalent results compared with both RFA and EVLA. Additionally, CAC displayed a decreased risk of postprocedural paresthesia and pigmentation and induration compared with EVLA, and both RFA and CAC had an improved pain profile compared with EVLA 1470 nm. The potential underperformance of NTNT modalities in ablating large GSVs requires further research.

Author contributions

Conception and design: VB, AB, AK, AC, VR, AG, KK

Analysis and interpretation: VB, AB

Data collection: VB, AB, AC

Writing the article: VB, AB, AK, AC

Critical revision of the article: VB, AB, AK, AC, VR, AG, KK

Final approval of the article: VB, AB, AK, AC, VR, AG, KK

Statistical analysis: VB, AB

Obtained funding: Not applicable

Overall responsibility: KK