Onyx extravasation during embolization of a brain arteriovenous malformation

Abstract

During Onyx embolization to treat brain arteriovenous malformation (AVM), carefully observing the penetration of Onyx to the nidus is important in order to avoid complications such as hemorrhage, ischemia, and difficulty with microcatheter removal. We encountered a case of Onyx extravasation during embolization of a cerebellar AVM confirmed by surgical resection and pathological analysis. The patient was a 44-year-old man with Spetzler-Martin grade I cerebellar AVM who underwent Onyx embolization prior to resection of the brain AVM. While injecting Onyx into the nidus using the “plug-and-push” technique, Onyx extravasation was observed. Onyx injection was paused and subsequently restarted, thereby allowing continuation of embolization. An oblate Onyx cast that was entirely covered in cerebellar tissue was removed during total resection of the AVM, performed the same day. The surgically removed oblate Onyx cast did not contain brain tissue or vessel wall, and immunohistochemical staining against glial fibrillary acidic protein (GFAP) showed Onyx penetration into GFAP-positive cerebellar tissue. Onyx extravasation was confirmed based on intraoperative findings during resection as well as pathological findings. The patient has been followed for four years postoperatively, and adverse events caused by Onyx extravasation have not been observed. Unexpected cast of Onyx, remote from the vascular architecture of the AVM, may represent an intra-parenchymal extravasation.

Introduction

In 1990, Taki et al.¹ first described the use of a non-adhesive liquid embolic agent comprising ethylene vinyl alcohol (EVOH) copolymer and radiopaque metrizamide dissolved in organic solvent dimethyl sulfoxide (DMSO) for the embolization of brain arteriovenous malformation (AVM). This liquid embolic agent was approved as Onyx (eV3 Covidien, Irvine, CA, USA) in 2005 by the Food and Drug Administration in the United States, and tantalum powder has been used as a radiopaque substance.² Onyx embolization is currently in wide use as an effective method in multimodal treatment for brain AVM,³ and a complete cure has been attained with Onyx embolization alone in some cases.^4,5 When Onyx comes into contact with blood, the DMSO rapidly diffuses into the blood and the EVOH precipitates, leading to the hardening of Onyx.^1,2 The characteristics of Onyx mean that the substance takes a longer time to harden than N-butyl-cyanoacrylate (NBCA), forming a soft, sponge-like substance. Satisfactory penetration of Onyx into the nidus can be feasibly achieved through intermittent injection over a long period of time.² To avoid complications such as hemorrhage, ischemia, and difficulty with microcatheter removal, careful observation of Onyx penetration into the nidus is recommended during Onyx embolization for brain AVM.^6–9 We report a case of Onyx extravasation during embolization of a cerebellar AVM confirmed by surgical resection and pathological analysis.

Case report

History and examination

A 44-year-old man with no significant past medical history presented with sudden onset of headache with Glasgow Coma Scale score 13 and no focal neurological deficits. Head computed tomography (CT) at admission showed Fisher grade 3 subarachnoid hemorrhage that was distributed unevenly in the left posterior cranial fossa along with intraventricular hemorrhage (Figure 1). The subarachnoid hemorrhage was classified as severity grade III on the Hunt & Kosnik grade and grade II on the World Federation of Neurosurgical Societies scale. Cerebral angiogram showed a flow-related cerebral aneurysm in the proximal segment of the left posterior inferior cerebellar artery (PICA) and a left cerebellar AVM measuring 24 × 20 × 18 mm in size with arterial supply from the hypertrophied left PICA and left anterior inferior cerebellar artery with the venous drainage into the inferior vermian vein and inferior cerebellar hemispheric vein (Figure 2). Based on the Spetzler-Martin grade system (1 point for size, 0 for venous drainage, 0 for eloquent area), it was classified as a grade I. The subarachnoid hemorrhage was thought to be related to the ruptured PICA aneurysm that was treated with coil embolization. The patient was discharged home on day 20 without neurological abnormalities. Curative treatment for the cerebellar AVM was planned, and embolization and craniotomy resection were scheduled for day 38.

Figure 1.

Initial computed tomography (CT) shows subarachnoid hemorrhage located mainly in the left posterior fossa with intraventricular hemorrhage.

Figure 2.

Initial left vertebral ((a), (b), (d), (e)) and three-dimensional rotational ((c), (f)) angiography shows an aneurysm originating from the proximal segment of the left posterior inferior cerebellar artery and a left cerebellar arteriovenous malformation in the early ((a), (c), (d), (f)) and late ((b), (e)) phases. This arteriovenous malformation is supplied by the dilated left posterior inferior cerebellar artery (small arrow) and left anterior inferior cerebellar artery (large arrow), and drains into the inferior vermian vein (small arrowhead) and inferior hemispheric vein (large arrowhead). (a)–(c): anteroposterior view; (d)–(f): lateral view.

Endovascular surgery

The surgical plan involved Onyx injection using the “plug-and-push” technique from the dilated primary supplying artery PICA. Under general anesthesia, a 7-Fr guiding catheter (FUBUKI; Asahi Intecc, Aichi, Japan) was advanced from the left femoral artery and placed at the extracranial V2 segment of the left vertebral artery. To increase the stability of the microcatheter and support during microcatheter removal, a 4.2-Fr intermediate catheter (FUBUKI, Asahi Intecc) was placed near the area where the PICA branches from the intracranial vertebral artery, and the tip of a DMSO-compatible microcatheter (Marathon, eV3 Covidien) was advanced just before reaching the left PICA nidus (Figure 3(a), (e)). Under blank roadmap imaging guidance, Onyx 34 was injected from the microcatheter, and a plug was created. The injected Onyx penetrated from the nidal surface to the deep area and drained into the inferior vermian vein. At this point, the injection was temporarily paused. When Onyx injection was restarted, penetration was seen from the nidal surface to the deep area, and an oblate Onyx cast appeared at the caudal side of the deep nidus and gradually expanded in size after approximately five minutes of injection at the same speed (Figure 3(b), (f)). At this point, 28 minutes had passed since the start of Onyx injection. Since aneurysms and varicose veins were not present at the site where the oblate Onyx cast appeared, Onyx extravasation from the nidus was determined to have occurred and Onyx injection was consequently paused temporarily. After confirming the absence of contrast agent extravasation on angiography, we decided to continue the embolization. When Onyx injection was resumed after pausing for 40 seconds, the direction of Onyx penetration into the nidus changed to penetrate primarily into the lateral area near the nidal surface. The left anterior inferior cerebellar artery was embolized in a retrograde manner, and embolization was subsequently concluded because Onyx had ultimately leaked to the inferior cerebellar hemispheric vein (Figure 3(c), (g)). The duration of Onyx injection was 41 minutes from start to finish, and 21 minutes if the pausing time was subtracted. The total volume of injected Onyx was 1.17 ml, and mean Onyx injection speed was 0.056 ml/min. Post-embolization angiography showed occlusion of the majority of the nidus along with patency of draining veins, albeit with delayed visualization (Figure 3(d), (h)). After confirming the absence of intracranial hemorrhage on head CT after Onyx embolization, the patient was transferred to the operating room.

Figure 3.

Preoperative left vertebral angiography ((a), (e)) shows a left cerebellar arteriovenous malformation. Native image ((b), (f)) just after Onyx extravasation shows an oblate cast (arrow) at a deep location of the Onyx cast. Native image ((c), (g)) after the Onyx embolization shows final cast of Onyx. Final left vertebral angiography ((d), (h)) after the Onyx embolization shows remarkable flow reduction in the nidus with patency of the draining veins with delayed visualization. (a)–(d): working angle view; (e)–(h): lateral view.

Operation

Left suboccipital craniotomy was performed with the patient in a prone position, the foramen magnum was exposed, and laminectomy of the first cervical vertebra was performed. An incision was made into the dura mater, the supplying artery was clamped in a conventional manner, and the entire circumference between the nidus and cerebellar parenchyma was dissected. An oblate Onyx cast was encountered when dissecting the caudal side of the deep nidus (Figure 4). Vessel wall was absent around this Onyx cast, which was covered in cerebellar tissue and adhered to the caudal side of the nidus. The border with the brain tissue was distinct across the entire circumference, and dissection and removal were easily performed. Intraoperative findings also indicated the presence of Onyx extravasation. Brain angiography was performed immediately postoperatively, and total resection of brain AVM was confirmed.

Figure 4.

Intraoperative microscopic view shows an oblate Onyx cast covered with cerebellar tissue and adhered to the caudal side of the nidus (arrow).

Pathological findings

The surgically removed oblate Onyx cast was 7.5 mm × 5 mm in size, and did not contain brain tissue or vessel wall (Figure 5(a)). Hematoxylin and eosin staining did not show vessel wall in or around the cast (Figure 5(b)), and immunohistochemical staining for glial fibrillary acidic protein (GFAP) showed Onyx penetration into GFAP-positive cerebellar tissue (Figure 5(c)). Pathological findings also indicated the presence of Onyx extravasation. Onyx extravasation was not evident in the surgically resected nidal specimen.

Figure 5.

(a) Macroscopic findings of the oblate Onyx cast section. The Onyx cast measures 7.5 mm × 5 mm. (b) Histological finding with hematoxylin and eosin staining of the Onyx cast. No vessel wall is observed inside or around the Onyx cast. (c) Immunohistochemical staining of the Onyx cast for glial fibrillary acidic protein (GFAP). The Onyx has infiltrated GFAP-positive cerebellar tissue. Arrows in (b) and (c) point toward the inside of the Onyx cast. Original magnification 100 × (b); 200 × (c).

Postoperative course

Postoperatively, no neurological abnormalities were seen and the patient made satisfactory progress. The patient has been followed for four years postoperatively with head magnetic resonance imaging and brain angiography, but no findings of brain AVM recurrence or adverse events attributable to Onyx extravasation have been identified.

Discussion

To the best of our knowledge, no reports have described cases where Onyx extravasation occurred during Onyx embolization to treat brain AVM. In a pathological study by Natarajan et al., none of the resected specimens after Onyx embolization for brain AVM have described even microscopic Onyx extravasation,¹⁰ indicating that Onyx extravasation in brain AVM is extremely rare. On the other hand, Onyx extravasation has been reported in a few studies with diseases other than brain AVM.^11–15 For instance, Onyx embolization used to stop gastrointestinal arterial bleeding has led to Onyx extravasation from the site of hemorrhage.^13,14 However, extravasation of contrast agent from the site of hemorrhage was evident on pre-embolization angiography in these cases, so Onyx extravasation from sites of hemorrhage appears natural. In addition, Onyx extravasation has been reported to occur out of the aneurysm sac and stray into the inferior vena cava when Onyx embolization was performed to treat endoleaks that persisted and increased within the aneurysm sac after stent-grafting for abdominal aortic aneurysm.¹⁵ Onyx extravasation has also been reported to occur into the subarachnoid space, leading to subarachnoid hemorrhage when Onyx embolization was performed to treat a giant internal carotid aneurysm.¹¹ Rangel-Castilla et al. performed preoperative Onyx embolization in 100 patients with hypervascular head, neck, and spinal tumors, and found that Onyx extravasation had occurred around the tumor in three patients.¹² Those reports suggested that Onyx may perforate the vessel wall and extravasation may occur when injected into a weakened aneurysm wall or tumor vasculature.

The hardening process for injected Onyx has been described as akin to volcanic lava-like flow, signifying that solidification occurs from the outside to the inside. This is thought to result from gradual dissipation of DMSO from the outer surface of the cast into the blood. Although the Onyx cast forms an outer skin immediately after injection, the center of the cast remains in a semi-liquid state. Onyx injected at this point extrudes from the outer skin, and the process of Onyx penetration starts over.² Viscosity of Onyx is dependent on the viscosity chosen, from Onyx 18 to Onyx 34. The pressure of injection is indeed correlated to the viscosity but even more to the volume being injected as there is a progressive increasing during injection that is made visible by an increasing vessel dilation seen during injection. Thus, the ability for Onyx 34 to diffuse is lower than Onyx 18 and its ability to induce extravasation is higher. In the present case, Onyx 34 was injected through all of the procedure using the “plug-and-push” technique and was injected with pressure into the nidus. Pressure applied during Onyx injection is therefore likely to have broken through the outer skin to perforate the weakened nidal vessel wall, leading to Onyx extravasation. Extravasation occurred at the deep nidus in the cerebellar tissue, and Onyx expanded in a spherical shape because of spread from the site of extravasation under uniform pressure. If the extravasation had occurred from a nidus in contact with the brain surface, then the Onyx would likely leak and diffuse into the subarachnoid space. In the present case, we initially thought that the Onyx had been injected into the aneurysm or varicose vein when the extravasation occurred, and thus took some time until making the decision to pause injection. Intraoperative extravasation of contrast agent was absent and postoperative head CT did not detect intracranial hemorrhage, indicating that the site of vessel wall perforation had become completely blocked by Onyx. Moreover, Onyx penetrated in a different direction in the nidus when injection was resumed, indicating that embolization can be continued because of Onyx hardening at the extravasation site even if extravasation has occurred.

The injection speed of Onyx is currently based on experimental results from studies in swine models of AVM.^16–18 DMSO comprises ≥90% of Onyx used in AVM embolization, and vasospasm and vascular toxicity can develop when DMSO is injected into arteries rapidly or in a large volume. However, the vascular toxicity of DMSO can be ignored under the recommended injection speed¹⁸ and a slow injection speed ≤0.1 ml/min is therefore recommended for Onyx.^2,3 Moreover, from the perspective that Onyx has a relatively high viscosity, the recommended injection speed should be followed in order to prevent Onyx extravasation since a fast injection speed would require Onyx injection under excessive pressure. In the present case, the injection speed of Onyx was unlikely to be the cause of extravasation, since the mean injection speed did not exceed 0.1 ml/min. Moreover, while Onyx extravasation is likely to occur from the site of hemorrhage in cases of AVM after bleeding, the present case involved an unruptured AVM. Although Natarajan et al. found angionecrosis in 43% and 59% of resected specimens through pathological examination of the brain AVM after Onyx embolization,^3,10 angionecrosis due to DMSO commonly occurs during the subacute stage,^10,16 indicating that DMSO-induced angionecrosis as the cause of Onyx extravasation is unlikely in the present case.

In the present case, we could not definitively conclude why Onyx extravasation occurred from the nidus. Nonetheless, our findings indicated that Onyx may perforate the nidal vessel wall and extravasation may occur under injection into a weakened nidus. In brain AVM, particularly in an eloquent area, symptomatic brain tissue injury may occur because of Onyx extravasation, so Onyx injection should be suspended as soon as possible after identification of extravasation. To accomplish this, thorough confirmation of the vascular structures of the nidus, feeding artery, draining vein, aneurysms, and varicose veins is needed preoperatively and Onyx must be injected slowly and carefully for penetration into the nidus.

Conclusion

We have reported a case of Onyx extravasation during embolization of a cerebellar AVM confirmed by surgical resection and pathological analysis. Since Onyx extravasation can occur during brain AVM embolization, thorough preoperative confirmation of the vascular structure of AVM may be desirable, along with slow, careful injection of Onyx into the nidus.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.