Update Onyx embolization for plexiform arteriovenous malformation: Ante-grade drifting technique

Abstract

Objective

The conventional arteriovenous malformation (AVM) Onyx embolization technique is the extrusion Onyx injection technique, with blood-flow control after a certain distance casting through the head end of the microcatheter. This method has elevated periprocedural AVM bleeding complications. In this study, the authors reported safety and efficacy of an updated ante-grade drifting Onyx injection for plexiform AVM embolization.

Methods

Between January 2016 and December 2018, 101 consecutive patients with plexiform AVMs were treated with ante-grade drifting Onyx injection. The patients’ clinical status was classified using the modified Rankin Scale (mRS). To measure associations, logistic univariate or multivariate regression analyses were used.

Results

Complete AVM obliteration was achieved in 51.2% (52/101) of patients. Two (2/101, 2.0%) arterial perforations occurred without causing neurological deficits. In univariate and multivariate logistic regression analyses, younger patient age (odds ratio (OR) = 1.06, 95% confidence interval (CI) 1.01–1.12, p = 0.014), haemorrhagic presentation at admission (OR = 7.14, 95% CI 1.52–33.33, p = 0.013) and low Spetzler–Martin grade (OR = 10.00, 95% CI 3.45–25.00, p < 0.001) were significantly correlated with complete obliteration. Pretreatment mRS was correlated with perforation complication (OR = 3.44, 95% CI 1.05–11.29, p = 0.041) in univariate logistic regression analysis but not in multivariate logistic regression analysis (OR = 2.956, 95%CI 0.745, 11.731, p = 0.123). Patients’ clinical status was significantly improved after endovascular AVM embolization.

Conclusions

With ante-grade drifting Onyx injection, it was possible to prevent serious bleeding complications and elevated complete embolization rate in plexiform AVMs. Younger patient age, haemorrhagic presentation at admission and low AVM Spetzler–Martin grade were significantly correlated with complete obliteration. Although there is not enough statistical power to show that the pretreatment mRS and the arterial perforation complication have a significant correlation, but its OR value is large, and there may be more data in the future to obtain further conclusion.

Introduction

Arteriovenous malformation (AVM) anatomy is essentially composed of one or more feeding arteries, a nidus and one or more draining veins. The nidus composes the bulk of the AVM and is a cluster of venous loops that are entangled. It has also been observed that the AVM nidus is composed of haemodynamically independent compartments.¹ Exact knowledge about the nidus of an AVM and the connected vessels is very important in AVM embolization.² Haemodynamic AVM compartments can have one or more separate feeding arteries and one or more draining veins, and multiple compartments can share the same venous drainage. AVM compartments are in haemodynamic balance, as no transfer of contrast material was observed from one compartment to another in digital subtraction angiography.² Therefore, attempting to occlude another compartment through one compartment with continued extrusion Onyx injection has resulted in a disastrous outcome because of the premature obstruction of draining veins.^3–5 To overcome this potential problems, we have promoted the ante-grade drifting Onyx injection technique, which is flow-free and pressure-control Onyx embolization in the AVM nidus based on the anatomy and biology of AVM.^6,7 This study reports on the safety and efficacy of AVMs treated by ante-grade drifting Onyx injection. The approval of the protocol of the study was approved by the ethical committee of our hospital.

Methods

From January 2016 to December 2018, 101 consecutive patients with plexiform AVMs were treated with ante-grade drifting Onyx injection (Table 1). Single-hole pial arteriovenous fistulas and high-flow arteriovenous fistulas were excluded. There were 59 (58.4%) male patients and 42 (41.6%) female patients, with ages ranging from 4 to 64 years (M = 26 years). The patients’ clinical status was classified according to the modified Rankin Scale (mRS), with a score of 0 in 36 (35.6%) patients, 1 in 47 (46.5%), 2 in 11 (10.9%), 3 in five (5.0%) and 4 in two (2.0%). Thirty-five (34.7%) AVMs (10 seizures and 25 incidental) were unruptured and 66 (65.3%) were ruptured. AVMs were located in the frontal lobe in 18 (17.8%) patients, in the occipital lobe in 16 (15.8%), in the parietal lobe in 16 (15.8%), in the temporal lobe in 21 (20.8%), in the thalamus in eight (7.9%), in the cerebellum in eight (7.9%), in the corpus callosum in six (5.9%), in the basal ganglion in three (3.0%), in the brain stem in three (3.0%) and in the ventricle in two (2.0%). Thirty-six (35.6%) AVMs were with deep-venous drainage. Fifty-one (50.5%) AVMs were in an eloquent location. Forty-two (41.6%) AVMs were small ( < 3 cm), 58 (57.4%) were mid-sized (3–6 cm) and one (1.0%) was large (>6 cm). Seventeen (16.8%) AVMs were Spetzler–Martin grade I, 42 (41.6%) were grade II, 35 (34.7%) were grade III, six (5.9%) were grade IV and one (1.0%) was grade V. Coexisting aneurysms were found in 17 (16.8%) AVMs, and high-flow fistulas were found in 14 (13.9%) AVMs. Pretreatment and follow-up clinical status were classified according to the mRS.

Table 1.

Clinical base data of 101 plexiform arteriovenous malformation patients.

Characteristics	Patients, n (%)
Age (years)	Mean 26, range 4–64
Male	59 (58.4%)
Sequential	51 (50.5%)
Non-sequential	50 (49.5%)
Small, <3 cm	42 (41.6%)
Mid-sized, 3–6 cm	58 (57.4%)
Large, >6 cm	1 (1%)
Deep-venous drainage	36 (35.6%)
S–M grade I–II	59 (58.4%)
S–M grade III	35 (34.7%)
S–M grade IV–V	7 (6.9%)

S–M grade: Spetzler–Martin grade.

Table 2.

Spetzler–Martin grades and embolization results of 101 patients.

S–M grade	Patients, n	Complete occlusion	Perforation complication
I	17	17 (100%)	1 (5.9%)
II	42	24 (57%)	0
III	35	11 (31%)	1 (2.9%)
IV	6	0	0
V	1	0	0

Ante-grade drifting Onyx injection technique

After informed consent was obtained, embolization was performed under general anaesthesia. A 5 F sheath was placed into the femoral artery, and a 5 F guiding catheter (Envoy; Codman & Shurtleff, Inc., Raynham, MA) was inserted through the sheath. The guiding catheter was continuously flushed with heparinized saline (3000 IU/500 mL). Onyx18 (Medtronic-ev3, Minneapolis, MN) was used as the embolic agent, which was delivered through a Marathon microcatheter (Medtronic-ev3) and navigated to the feeding arteries by a CHIKAI-10 microguidewire (Asahi Intec Co. Ltd, Seto, Japan) through the guiding catheter in the internal carotid or vertebral arteries. The microcatheter tip was advanced distally to select the proper feeding or contributing arteries of the AVM. A coexisting aneurysm and high-flow fistula were embolized as a priority. In the ante-grade drifting Onyx injection technique, Onyx is injected directly without proximal plug formation. The injection speed is proportional to the size of the nidus. Onyx penetrates into the nidus and fills the compartment of the nidus backwards to the microcatheter tip (Figure 1). Once it reaches a safe point, the injection is stopped, and the microcatheter is retrieved. This method is capable of embolizing nidal venules and shunting arterioles without overflow to the feeding arteries proximal to the shunting arterioles. Progressive Onyx deposition can be identified in the compartment as its injection advances, and the extent of deposition can be estimated accurately. The embolization rate of the AVMs can be judged by the last run of the procedure, except for cured AVMs, which are then treated with surgery or radiosurgery. The definition of safety was a haemorrhagic complication rate of  < 5%, and the definition of efficacy was a cure rate >30%.

Figure 1.

Unsubtracted angiogram of the right vertebral artery, lateral view, showing a thalamus arteriovenous malformation (AVM) and intra-nidal penetrating Onyx cast, which was injected using ante-grade drifting technique via the right posterior lateral choroidal artery without any reflux (arrow).

Statistical analysis

Statistical analyses were performed using R v3.6.1 (The R Foundation for Statistical Computing, Vienna, Austria; www.R-project.org). The pretreatment and follow-up clinical status was test using the Wilcoxon signed-rank test. The associations of patient sex, age, presentation (ruptured or unruptured), poor pretreatment clinical status (mRS 3–4), Spetzler–Martin grade (low: I to III; high: IV and V) with complete occlusion and perforation complication were tested using univariate logistic regression analysis. Factors of p < 0.1 were tested using multivariate logistic regression analysis. A p-value of  < 0.05 was considered statistically significant.

Results

One hundred and eighty feeders were catheterized and embolized with ante-grade drifting Onyx injection during 116 treatment sessions in 101 patients. Complete AVM obliteration was achieved in 52/101 (51.2%) patients (Figures 2 and 3). Complete obliteration was obtained in all Spetzler–Martin grade I AVMs (17 cases), 57% (24/42 cases) in grade II and 31% (11/35 cases) in grade III. The curative embolization in grades I and II was 69%, and there was no curative embolization in grades IV and V (Table 2).

Figure 2.

A 25-year-old man with a ruptured AVM in the right temporal lobe. (a) Right vertebral artery injection, frontal view, showing a grade I AVM fed by temporal branches of right posterior cerebral artery (arrowhead). (b) Super-selective injection after one branch catheterization (arrowhead). (c) Super-selective injection of the second branch catheterization (arrowhead). (d) Unsubtracted image of working angle showing the intra-nidal Onyx cast (arrowheads). (e) Left vertebral artery injection after two branches ante-grade drifting Onyx injection, frontal view, showing complete occlusion of the nidus.

Figure 3.

A 33-year-old man presented with an intracranial haematoma. (a) Left internal carotid artery injection, lateral view, showing a grade II AVM fed by branches of the left anterior and middle cerebral arteries and had a venous aneurysm on the draining vein, which was rupture site (arrowhead). (b) Super-selective injection after the anterior cerebral artery feeding branch catheterization (arrowhead). (c) Super-selective injection after the middle cerebral artery feeding branch catheterization (arrowhead). (d) Left internal carotid artery injection after embolization, lateral view, showing complete embolization of the AVM (arrowheads).

Two per cent (2/101) arterial perforation occurred in two patients because the feeders were very small and tortuous. There was no new neurological deficit in these two perforations because they were embolized as soon as possible, since they were found intraoperatively.

Mean follow-up was 1.7 years (range 0.2–3 years). Clinical outcomes were mRS 0–1 in 88 (87%) patients, mRS 2 in 10 (9.9%) patients and mRS 3 in three (3.0%) patients. There was no morbidity or mortality caused by embolization treatment during follow-up.

In univariate logistic regression analyses (Table 3), younger patient age (odds ratio (OR) = 1.05, 95% confidence interval (CI) 1.01–1.10, p = 0.007), haemorrhagic presentation at admission (OR = 4.35, 95% CI 1.76–11.11, p = 0.001), pretreatment mRS (OR = 1.85, 95% CI 1.13–3.06, p = 0.015) and low AVM Spetzler–Martin grade (OR = 5.56, 95% CI 2.70–11.11, p < 0.001) were significantly correlated with complete obliteration, while only poor pretreatment mRS correlated with perforation complication (OR = 3.44, 95% CI 1.05–11.29, p = 0.041). Sex, age and haemorrhagic presentation at admission were not significantly associated with perforation complication in univariate analysis. In multivariate logistic regression analyses (Table 4), younger patient age (OR = 1.07, 95% CI 1.01-1.13, p = 0.013), haemorrhagic presentation at admission (OR = 7.44, 95%CI 1.52, 36.47, p = 0.013) and low AVM Spetzler–Martin grade (OR = 10.03, 95%CI 3.48, 28.90, p < 0.001) were significantly correlated with complete obliteration, and poor pretreatment mRS was not correlated with perforation complication (OR = 2.956, 95%CI 0.745, 11.731, p = 0.123 ).

Table 3.

Results of univariate logistic regression analysis undertaken to quantify associations between risk factors and perforation and complete occlusion of plexiform arteriovenous malformations after treatment of 101 patients with ante-grade drifting Onyx injection.

Risk factors	Probability of perforation complication		Probability of complete occlusion
	OR (95% CI)	p (Wald’s test)	OR (95% CI)	p (Wald’s test)
Sex	0 (0–∞)	0.995	1.11 (0.5–2.44)	0.801
Age	0.87 (0.71–1.06)	0.166	1.05 (1.01–1.10)	0.007
Haemorrhagic presentation at admission	0 (0–∞)	0.995	4.35 (1.76–11.11)	0.001
Pretreatment mRS	3.44 (1.05–11.29)	0.041	1.85 (1.13–3.06)	0.015
S–M grade	0.62 (0.11–3.53)	0.588	5.56 (2.70–11.11)	<0.001

mRS: modified Rankin Scale; OR: odds ratio; CI: confidence interval.

Table 4.

Results of multivariate logistic regression analysis undertaken to quantify associations between risk factors and perforation and complete occlusion of plexiform arteriovenous malformations after treatment of 101 patients with ante-grade drifting Onyx injection.

Risk factors	Probability of perforation complication		Probability of complete occlusion
	OR (95% CI)	p (Wald’s test)	OR (95% CI)	p (Wald’s test)
Sex	NP	NP	NP	NP
Age	NP	NP	1.07 (1.01, 1.13)	0.013
Haemorrhagic presentation	NP	NP	7.44 (1.52, 36.47)	0.013
Pretreatment mRS	2.956 (0.745, 11.731)	0.123	1.02 (0.45, 2.31)	0.968
S–M grade	NP	NP	10.03 (3.48, 28.90)	<0.001

NP: not performed because of p>0.1 in univariate analyses.

The mean age (22.6±11.1 years) of patients with complete AVM obliteration was younger than that of patients with incomplete AVM obliteration (29.4±12.2 years). In the Wilcoxon signed-rank test, the follow-up mRS score (median =  0) was significantly lower than the pretreatment mRS score (median = 1; Z = –6.189, p < 0.001).

Discussion

In this study, we changed the strategy for embolizing plexiform AVM from the conventional method of extrusion Onyx embolization to ante-grade drifting Onyx embolization. The initial complete embolization rate was up to 50% of AVM patients. The incidence of acute bleeding complication rate was 2.0% caused by arterial perforation without morbidity or mortality, which is lower than that of 6.6% in a large meta-analysis.⁸ Ruptures due to premature venous occlusion and nidal damage caused by extrusion Onyx injection diminished with the ante-grade drifting technique, whereas the rate of arterial perforation remained steady. Our results are comparable to 5% intraoperative AVM rupture and 50% complete resection of AVMs in a cohort of 591 AVMs treated surgically.⁹ In a recent study of 224 AVM patients,¹⁰ extrusion Onyx embolization was used for low-grade (Spetzler–Martin grade I-II) AVMs. Complete AVM extrusion was achieved in 92% patients, of which 62.1% extrusions were obtained by a single endovascular session. However, they led to a mortality rate of 0.4% and a morbidity rate of 5%. In fact, the extrusion Onyx injection technique is more useful for dural arteriovenous fistula (DAVF) embolization, in which Onyx can be pushed into the initial venous compartment, allowing the embolic material to occlude the arteriovenous shunts, thereby closing them.^11,12 Extrusion Onyx injection elevated the DAVF and AVM cure rates significantly than the cure rates obtained by n-butyl-cyanoacrylate (NBCA) injection, but it also caused higher haemorrhagic complications in AVM embolization.

This dichotomy is based on the different vascular architecture between DAVF and AVM. There are five major parts to AVM: shunting arterioles (the nidus), made up of entangled anomalous venules; shunts, directly connected to draining veins; draining veins, which transfer arterial blood from the nidus to cortical veins or sinuses; and communicating venules, which connect the core vessels and surrounding cerebral veins.² Therefore, draining veins are not part of the nidus, and these pial veins must be kept intact in AVM embolization as long as the AVM nidus is not completely obliterated. Shunting arterioles have no cerebral branches for some distance and connect to the nidal venules. There are a large amount of aneurysms and aneurysmatic bulging on the nidal venules.^13,14 A few collecting veins arise from larger loops of the nidus and then unite to form a draining vein. Draining veins are noted to be passively dilated by the pressure transmitted from the nidus (i.e. venules). These nidal venules are crowded in three dimensions, and even with high-resolution angiography, it is impossible to take a tangential view of each venule loop to which shunting arterioles are connected. During angiograms, arterioles are angiographically demonstrable. The outlet point of the draining vein is the site where the density of contrast material decreased suddenly on angiograms. The more distal the catheterization, the simpler the architecture, down to single-channel communications between arteriole and venule.

Treatment based on ante-grade drifting Onyx embolization of AVMs has the advantages of preventing premature occlusion of the draining vein, decreasing Onyx volume and procedure time and avoiding nidus damage and catheter burst.⁷ The ante-grade drifting Onyx embolization eliminates adverse effects on AVM haemodynamics of extrusion Onyx embolization and transvenous Onyx embolization.¹⁵ Size-reducing extrusion Onyx embolization had been indicated as preparation for radiosurgery,^16,17 in which an aggressive over-volume Onyx injection to reduce the AVM size can explain why pre-radiosurgery embolization was associated with an increased rate of haemorrhage and more complications than mono-embolization and pre-surgical embolization.⁸ Pre-radiosurgery flow-reductive AVM embolization has recently been reported as a valid indication.¹⁸

In conventional extrusion Onyx embolization studies, there are no patient characteristics or AVM structures associated with AVM cure rate and complication rate.⁷ However, in our study, we found that younger patient age, haemorrhagic presentation at admission and low AVM Spetzler–Martin grade were significantly correlated with complete obliteration in univariate and multiple variate analyses. Although there is not enough statistical power to show that the pretreatment mRS and the arterial perforation complication have a significant correlation, but its OR value is large, and there may be more data in the future to obtain further conclusion. AVM in younger patients may be simple in vascular structure and small in size because AVM is a haemodynamic lesion.¹⁹ Haemorrhagic AVM and low Spetzler–Martin grade were also associated with small AVM size in previous studies.^13,19 Poor pretreatment of clinical status is associated with parenchymal haematoma stretching the AVM feeding vessels, making catheterization difficult and risking arterial perforation.²⁰

Haemorrhagic presentation, deep-venous drainage or associated aneurysms have an approximately twofold greater likelihood of a future AVM haemorrhage.²¹ It is well established in the literature that features such as intra-nidal aneurysms, high pressure in feeding vessels and obstruction of venous outflow increase the risk of haemorrhage.^22,23 According to Miyasaka et al.,²⁴ a small AVM represents a higher resistant shunt and a higher mean feeding artery pressure compared to medium-sized and large AVMs. This would imply that the likelihood of multi-compartmental large AVM bleeding should be dictated by the smallest compartments, which might represent a more vulnerable part of the AVM.²⁵ This explains why in partial elimination of an AVM by surgery, embolization or radiotherapy is claimed to increase or decrease the risk of bleeding. Partial elimination, irrespective of treatment modality, can leave small compartments with a tendency to bleed, or the treatment might have eliminated the small compartments and thus reduced the risk of bleeding.²⁶ In our early experiences, a small residual AVM bled and caused death after radiosurgery. Therefore, we prefer complete elimination of small AVMs.²⁷

Conclusions

With ante-grade drifting Onyx injection, it was possible to prevent serious bleeding complications and increase the complete embolization rate in plexiform AVMs. Younger patient age, haemorrhagic presentation and low AVM Spetzler–Martin grade were significantly correlated with complete obliteration. Although there is not enough statistical power to show that the pretreatment mRS and the arterial perforation complication have a significant correlation, but its OR value is large, and there may be more data in the future to obtain further conclusion.

ORCID iD

Xianli Lv https://orcid.org/0000-0001-8270-8464