Article Text

Download PDFPDF

Case series
Extending the indications for transvenous approach embolization for superficial brain arteriovenous malformations
  1. Dinark Conceição Viana1,
  2. Luis Henrique de Castro-Afonso1,
  3. Guilherme Seizem Nakiri1,
  4. Lucas Moretti Monsignore1,
  5. Felipe Padovani Trivelato1,
  6. Benedicto Oscar Colli2,
  7. Daniel Giansante Abud1
  1. 1 Division of Interventional Neuroradiology, Department of Internal Medicine, University of São Paulo, Ribeirao Preto, São Paulo, Brazil
  2. 2 Division of Neurosurgery, Department of Surgery, University of São Paulo, Ribeirao Preto, São Paulo, Brazil
  1. Correspondence to Dr Luis Henrique de Castro-Afonso, Division of Interventional Neuroradiology, Medical School of Ribeirão Preto, University of São Paulo, Avenida Bandeirantes, 3900, Monte Alegre, Ribeirão Preto, São Paulo, Brazil, 14048-090; castroafonsolh{at}


Introduction Transarterial embolization is the standard endovascular treatment strategy for intracranial arteriovenous malformations (AVMs). The transvenous approach has been indicated for the embolization of deep AVMs meeting a set of strict eligibility criteria. The present study aims to assess the safety and efficacy of the transvenous approach for superficial AVMs.

Methods A retrospective series of 12 patients presenting with cortical AVMs were treated by endovascular embolization using a transvenous approach with a curative intent.

Results Nine patients (75%) had ruptured AVMs at admission. The mean nidus size was 1.9 cm, six patients (50%) had a nidus in eloquent areas and the median Spetzler–Martin grade was 2. The rate of immediate angiographic occlusion of the AVMs was 91.6% (11/12). One patient in whom immediate angiographic occlusion was not achieved showed spontaneous occlusion at the 6-month follow-up. No procedural or clinical complications were observed. The mean and median modified Rankin scale (mRS) scores at discharge were 1.7 and 2 (range 0–3, SD=0.96), and the mean and median mRS scores at 6 months were 1.6 and 2 (0–3, 1.16). Nine patients (75%) were independent (mRS ≤2) at discharge and 11 patients (91.6%) were independent (mRS ≤2) at the 6-month follow-up.

Conclusions The curative transvenous embolization of superficial intracranial AVMs is feasible and appears safe and effective when strict anatomical selection is respected. This technique extends the current indications for transvenous embolization of intracranial AVMs and may improve cure rates while reducing embolization-related complications.

  • Intracranial arteriovenous malformations
  • Onyx
  • Transvenous embolization.

Statistics from

Request Permissions

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


Recent multicenter studies have demonstrated that invasive treatment of unruptured intracranial arteriovenous malformations (AVMs) results in poor clinical outcomes when compared with conservative management.1 2 On the other hand, the medical arm of the ARUBA trial resulted in high rates of stroke or death (10.1%) and of neurological disability or death (14%), especially when considering the relatively short follow-up of 33 months.1 The hemorrhage rate in the medical arm of the ARUBA trial was consistent with the results of previous studies that showed an annual hemorrhage risk of 2–4%.1 3 Therefore, improvements in the invasive treatment of AVMs are needed and should be urgently pursued.

The development of new materials and embolization techniques has been associated with improvements in cure rates and reduced complication rates during AVM embolization.4–9 The transvenous embolization of AVM is a retrograde technique aiming to occlude AVMs by injecting Onyx through the venous route into the entire AVM. The current indications for embolization of AVMs using the transvenous approach are essentially anatomical: deep AVMs, unfavourable arterial access for arterial embolization, small nidus, or single draining vein. This approach is also indicated when no other treatment option can be safely performed.10–17 The aim of the present study was to investigate the feasibility, safety, and efficacy of curative transvenous embolization for superficial intracranial AVMs.


Patients and clinical decision-making

A retrospective database of 12 consecutive patients presenting with intracranial AVMs who underwent transvenous embolization with Onyx-18 (Medtronic, Irvine, California, USA) between January 2011 and November 2016 was analyzed. The study protocol was approved by the institutional review board, which waived the need for written informed consent from the participants. Baseline clinical and angiographic data were collected from all patients. All patients were examined by an independent neurologist who assessed neurological outcomes in the hospital using the Glasgow Coma Scale (GCS). Neurological outcomes at discharge and at the 6-month follow-up were assessed using the modified Rankin scale (mRS). Four-vessel digital subtracted angiography (DSA) was obtained before embolization, immediately after embolization, and at the 6-month follow-up.

In accordance with our institution’s multidisciplinary strategy for treatment of AVMs, patients admitted with acute hemorrhage associated with intracranial AVMs were first evaluated by the neurosurgical team and an emergent hematoma evacuation and/or external ventricular drainage were performed if necessary. After 2 weeks of hemorrhage, patients were evaluated for endovascular embolization using Onyx-18 with an intention to cure. If complete occlusion of the AVM could not be achieved by the endovascular approach alone, then the patients were referred for microsurgical resection or radiosurgery. The decision to perform an invasive treatment of intracranial AVMs was always defined by a multidisciplinary team. Once an embolization was indicated, we performed the transvenous approach based on the available criteria in the literature.

The current criteria for transvenous embolization as described by Professor Mounayer’s group include the following:11–13 inability to catheterize the arterial feeder(s); lesions not amenable to surgery or radiosurgery, favourable venous angioarchitecture of the AVM;11 high Spetzler–Martin grades; lesions in deep or eloquent locations; absence of arterial access in cases of residual nidus after previous embolizations, radiosurgery, or neurosurgery; narrow arterial feeders, very tortuous course, or en passage arterial feeders; single draining vein; relatively small nidus or nidus remnants; nidus size < 3 cm and drainage by a single venous collector without venous aneurysm.13

In addition to the criteria listed above, the present study included patients presenting with cortical or superficial AVMs.

Embolization procedure

All procedures were performed under general anesthesia. Catheterization was performed using a coaxial system through a femoral approach with a 6 F sheath and full selective DSA was performed before each treatment in all patients. No heparin was infused in any patient. The right or left jugular vein was punctured, depending on the location of the AVM, and a 5 F sheath was inserted. Jugular access was chosen rather than femoral access in order to avoid permanent lodging of microcatheters in the heart chamber in case the microcatheter became trapped in the Onyx cast and could not be safely withdrawn.

A 6 F guiding catheter (Softip; Boston Scientific) was positioned in the cervical artery. A microcatheter (Marathon or Apollo; Medtronic) was advanced through the 6 F guiding catheter as close as possible to the nidus. This microcatheter was used for angiographic control injections during embolization, for combined double injections through the arterial and the venous access routes. A 5 F vertebral or a 6 F guiding catheter (Neuron; Penumbra) was advanced into the intracranial venous sinus. A detachable-tip microcatheter (Apollo; Medtronic) was then navigated through the draining vein of the AVM and placed into the nidus.

The transvenous embolization was initiated by injecting Onyx-18 into the nidus through the venous access route; once an Onyx reflux was obtained, penetration of the Onyx all the way through to the arterial branches was achieved. After filling the nidus with Onyx and anatomically obliterating the AVM, the microcatheter was withdrawn. Ideally, the microcatheters were removed completely. However, in cases of entrapment, the microcatheters were cut with a blade at the level of the jugular sheath. All patients underwent brain CT scanning after treatment in the angiography suite. All patients were hospitalized and monitored for 4 days post-intervention, and IV dexamethasone was administered during this period. Anatomic AVM cure was defined as angiographic occlusion of the nidus in the absence of early venous drainage.

Statistical analysis

Categorical variables are presented as numbers and percentages and compared between groups using the χ2 or Fisher’s exact tests, as appropriate. Continuous variables are presented as mean (range, ±SD) and the Mann–Whitney test or Student’s t-test was used, as appropriate. One independent blinded statistician received all the data collected for statistical analysis. We considered p values <0.05 as significant. The IBM SPSS Statistics software version 20.0 (Chicago, Illinois, USA) was used for statistical analysis.


All patients presenting with cortical intracranial AVMs who underwent transvenous embolization at our institution were included in the study (n=12). Five of the patients were men (41.7%) and seven were women (58.3%). The mean and median ages were 33.4 and 28 years (range 11–68, SD 19.6), respectively. Among nine patients (75%) who had a ruptured AVM, eight had parenchymal hemorrhages (66.7%), one had a subarachnoid hemorrhage (8.4%), and one had a ventricular hemorrhage (8.4%). No patient required surgical evacuation of a hematoma during hospitalization.

Ten patients (83%) had supratentorial AVMs and two patients (16.6%) had infratentorial AVMs. Six patients (50%) presented with AVMs located in an eloquent area, 11 patients (91.7%) had AVMs located in the right hemisphere (including the cerebellum), and one AVM (8.3%) was in the left hemisphere. The mean nidus size was 1.9 cm (0.4–4, 1.2). Five patients (41.7%) had nidal aneurysms. The mean and median Spetzler–Martin grades were 2 (1–4, 0.95) and 2, respectively.

Regarding venous drainage of the AVMs, the mean and median number of draining veins were 1.2 (1–3, 0.62) and 1, respectively. Nine patients (83.4%) presented with exclusively superficial venous drainage, two (26.8%) had deep venous drainage, and one (8.4%) had mixed venous drainage. Ten patients (83.3%) had a single draining vein, one patient (8.3%) had two draining veins (patient 12), and one patient (8.3%) had three draining veins (patient 6). Among all 12 patients, no patient presented with dilated draining veins. Regarding the arterial supply, nine patients (75.0%) had AVMs supplied by anterior circulation feeders whereas three (25.0%) had AVMs supplied by posterior circulation feeders. Only two patients (16.6%) had AVMs with an arterial supply from perforator branches.

The study achieved 91.6% (11/12) immediate angiographic occlusion of the AVMs without any technical or clinical complications. Patient 4, in whom a complete immediate occlusion of the AVM was not achieved, showed complete AVM occlusion on DSA after 6 months. Two patients (5 and 12) did not undergo DSA at the 6-month follow-up. Both patients were contacted by telephone when mRS scores were obtained, and angiographic assessment was planned for the 12-month follow-up. The rate of complete angiographic occlusion was 10/10 (100%) at 6 months. The mean number of embolization sessions was 2 (1–5, 1.12) per patient. The mean volume of Onyx-18 injected per patient was 3.2 (0.4–20, 5.5) mL. Before curative transvenous embolization, eight (66.6%) patients had undergone previous arterial embolization sessions. During the final and curative embolization session, eight (66.7%) patients were treated only using the transvenous approach whereas, in four patients (33.3%), both transarterial and transvenous approaches were used (patients 5, 9, 11 and 12) in the final session. Only patient 11 underwent a final embolization using both arterial and venous access simultaneously while, for patients 5, 9 and 12, arterial access was performed to allow procedural angiograms and also as a security catheter.

The Glasgow Coma Scale scores of all patients were 15 before embolization and remained 15 at discharge. The mean and median mRS scores were 1.7 (0–3, 0.96) and 2 at discharge and 1.6 (0–3, 1.16) and 2 at 6 months. Nine patients (75%) were independent at the time of discharge and 11 patients (91.6%) were independent at the 6-month follow-up. Six patients (50%) showed improvement of at least 1 point in the mRS score at 6 months whereas scores for the remaining six patients were unchanged. Tables 1 and 2 summarize the individual baseline characteristics and outcomes of the patients. Figures 1, 2 and 3 show endovascular procedures of patients 2, 8 and 11, respectively. 

Table 1

Patients’ clinical characteristics, classification and outcome

Table 2

AVM angioarchitecture and characteristics of the procedure

Figure 1

Patient 2: (A) Digital subtraction angiography (DSA) of the left internal carotid artery, lateral view, showing a small left parietal arteriovenous malformation (AVM) (white arrowhead). (B) X-ray of skull showing a microcatheter placed in the left pericallosal artery (black arrow) close to the nidus (black arrowhead) and a single draining superficial vein (white arrow). (C) Selective angiogram performed through a microcatheter in the draining vein close to the nidus (white arrow). (D) X-ray of skull showing the two microcatheters placed in the left pericallosal artery (black arrow) and in the draining vein (white arrow). (E) Final cast of Onyx after transvenous embolization. (F) Final DSA of the left internal carotid artery showing complete angiographic occlusion of the AVM.

Figure 2

Patient 8: (A) Digital subtraction angiogram (DSA) of the right internal carotid artery, oblique view, showing a right frontal arteriovenous malformation (AVM) (black arrow). (B) Angiogram performed through a microcatheter placed in a distal frontal arterial branch of the right middle cerebral artery close to the nidus (white arrow). This frontal arterial branch bifurcated in two branches (black arrowheads) and fed the nidus by en passage branches (not directly visible), and a single draining superficial vein is shown by the white arrow head. (C) X-ray showing a microcatheter placed in the frontal arterial branch (black arrow) and another microcatheter placed in the draining vein of the AVM (white arrow). (D) X-ray showing the initial reflux of Onyx-18 after injection through the draining vein of the AVM. (E, F) After an initial venous reflux, a progressive penetration of Onyx-18 into the nidus is shown. (G) The final cast of Onyx after complete transvenous embolization. (H) Final angiogram performed through the microcatheter in the frontal arterial branch showing complete occlusion of the nidus while maintaining a normal flow through the arterial branches. (I) Final DSA of the right internal carotid artery, oblique view, showing complete angiographic occlusion of the AVM.

Figure 3

Patient 11: Second embolization session. (A, D) Digital subtraction angiogram (DSA) of the right vertebral artery, frontal view (A) and lateral view (D), showing a right cerebellar arteriovenous malformation (AVM) (white arrowhead). (B, E) DSA performed through the microcatheter in the dural arterial branch of the right occipital artery, frontal view (B) and lateral view (E), showing a residual nidus (white arrowhead) after the first embolization session. (C) X-ray of the skull, frontal view, showing a detachable-tip microcatheter placed in the draining vein of the AVM close to the nidus (white arrow). (F) X-ray of the skull, lateral view, showing three microcatheters near the residual nidus. A detachable-tip microcatheter in the draining vein close to the nidus (white arrow), a detachable-tip microcatheter in the dural arterial branch of the right occipital artery (black arrow), and a microcatheter (black arrowhead) placed in the dural branch of the right occipital artery between the distal and detachable points of the microcatheter, ready for embolization using both the modified pressure cooker technique and transvenous embolization. (G) X-ray of the skull, lateral view, showing the final cast of Onyx after arterial embolization by the arterial access (white arrowheads) and by the venous access (white arrow). (H, I) Final DSA of the right vetebral artery, lateral view (H) and frontal view (I), showing complete occlusion of the cerebellar AVM.


The goal of curative embolization of intracranial AVMs is to completely occlude the AVM’s circulation, which is usually achieved by complete transarterial filling of the nidus with Onyx followed by occlusion of the draining veins as the final step of the procedure.14–17 Intracranial hemorrhage is the leading cause of neurological disability associated with endovascular treatment of AVMs.18 Premature occlusion of draining veins before complete occlusion of the nidus has been reported as one of the main causes of hemorrhagic complications.18 In light of such complications, the concept of AVM embolization through the transvenous route may appear paradoxical for general intracranial AVMs. However, previous reports have demonstrated that, for a selected group of AVMs with the specific eligibility criteria described in the Methods section, transvenous embolization seems to be safe, effective, and the only possible treatment strategy to achieve a cure.1–17

The success of the transvenous approach in arteriovenous fistulas did not translate into a similar management paradigm for intracranial AVMs. The biggest challenge in embolization of AVMs has been adequate nidal penetration without premature compromise of the venous drainage. Houdart et al,19 in their angiographic classification of intracranial arteriovenous fistulas, highlighted that arteriolovenulous fistulas comprise the majority of intracranial AVMs, with plexiform niduses and upstream shunts that are challenging to reach via the transvenous route. However, advances in liquid embolic materials and the advent of non-adhesive agents such as Onyx have allowed for longer durations of injection and better injection control compared with previous adhesive liquid materials. This characteristic of Onyx allowed the development and evolution of transvenous embolization techniques.10–17

Few case series reporting the results of the transvenous approach for endovascular treatment of AVMs are available in the literature.11–17 These series comprise a total of approximately 50 patients presenting with mostly deep intracranial AVMs treated by the transvenous approach. The enrolment criteria, angiographic results, and clinical outcomes are very similar across these studies. In the present study we expanded the indications of the transvenous approach to a group of patients presenting with cortical or superficial AVMs, and we found similar results to previous studies on the transvenous approach for AVMs.11–17 Moreover, when we extracted the data of individual patients presenting with cortical AVMs in previous studies, we found a total of seven patients whose AVMs were managed by the transvenous approach; all these patients were cured by transvenous embolization and no complications were reported.11–17

Although the transvenous approach was the embolization strategy investigated in this study, three patients (25%) were treated by the transvenous approach alone in a single curative session whereas four patients (33.3%) underwent combined transarterial and transvenous embolization (table 2). The combined arterial approach was used to decrease nidus inflow and optimize retrograde penetration of Onyx-18 by decreasing intravenous resistance.10–17 In such situations, the injection of Onyx by the arterial route was performed before or at the same time as injection through the venous side. Whenever possible, we preferred to catheterize an arterial afferent before performing the transvenous approach because a microcatheter placed on the arterial side as close as possible to the nidus allows for superselective angiograms of the AVM. Better visualization of the AVM improves the chance of successful complete Onyx filling of the entire nidus. Arterial access also provides a security catheter if an advertent hemorrhage occurs during transvenous embolization.

In addition, some other techniques have been described to facilitate and improve Onyx penetration and improve the safety of embolization of intracranial AVMs.20–22 Some authors have described the use of a temporary balloon occlusion of an arterial trunk aiming to reduce the arterial pressure in the AVM, which may improve Onyx penetration while reducing the risk of a rupture of the AVM.20–22 Although these techniques could be useful to our patients, we did not perform them in the present study because all AVMs in the present series had a small and slow flow nidus. We tend to indicate embolization under arterial temporary balloon occlusion mainly for high flow shunts. Another recent technique described by Zhang et al, the transvenous pressure cooker technique, may facilitate embolization of some AVMs using the transvenous approach.22

In our study we were able to catheterize both the draining vein and the main arterial feeder in nine (75.0%) patients whereas, in three (25.0%) patients, arterial catheterization could not be achieved because of the very narrow and tortuous arterial pedicles. Despite usually tortuous and recurrent characteristics of cortical veins, we were able to catheterize all cortical draining veins. However, for patients 10, 11, and 12 we used a balloon-assisted technique to allow catheterization of the draining veins.5

In terms of clinical outcomes after treatment, although 75% of patients presented with intracranial hemorrhage at admission, 91.6% had mRS scores of 0–2 at 6 months (table 1). Although surgical resection of AVMs located in eloquent areas is associated with poor clinical outcomes, in the present series all the AVMs were completely occluded and no new neurological deficits nor any ischemic lesions were observed.

With regard to the withdrawal of the transvenous microcatheter after embolization, most authors advocate cutting off the microcatheter in the jugular vein in order to avoid rupture of the vein by traction.11–17 In our experience, we used detachable microcatheters that we were able to safely remove from 10 patients (83%). In two patients (16.6%) the microcatheters were cut in the jugular vein and no complications were observed during follow-up at 6 months.

A key criticism of the present study is the use of an investigative endovascular technique for a group of patients for whom classic open surgical resection has a relatively safe profile when the AVM is not located in an eloquent region of the parenchyma. Among the 12 patients in this series, six (50%) had AVMs in eloquent cortical regions and we observed no neurological deficit after curative transvenous embolization. One of the postulated advantages of the transvenous approach is that this technique allows occlusion of the entire nidus while avoiding significant arterial reflux and reduces the risk of embolization of normal arterial territories, which could cause ischemic complications. For the remaining six patients (50%) whose AVMs were not located in eloquent regions, we opted for the transvenous approach because it is a minimally invasive endovascular procedure and in the spirit of continuously pursuing the development of safe and effective treatments for neurovascular conditions while avoiding craniotomy and reducing hospital stays.

The limitations of the present study include the small sample size, the absence of a control group, and enrolment from a single centre. These limitations impair the statistical power of our results and increase the risk of random error and selection bias. We particularly recommend caution regarding generalizations of the present study results to general clinical practice because at this point indications for the invasive treatment of intracranial AVMs are under debate and because treatment policies and doctors’ experiences are highly diverse and variable worldwide. This study represents an opposing viewpoint to the generalized concepts of the ARUBA trial and is consistent with the emerging concept of precision medicine which advocates for condition-specific treatment strategies. The present study may generate hypotheses for future larger studies.


Curative transvenous embolization of superficial intracranial AVMs is feasible and appears to be safe and effective when strict anatomical selection is respected. Expanding the current indications for this technique in the treatment of intracranial AVMs may improve cure rates and reduce embolization-related complications.



  • Contributors DCV contributed to the concept of the study and drafted and approved the manuscript. LHdCA contributed to the study concept and design, data analysis, revision and final approval of the manuscript. GSN, LMM, FPT and OBC contribution to data acquisition, figures and tables, revision and final approval of the manuscript. DGA contributed to the concept of the study, acquired data, critically revised the manuscript and approved the final work.

  • Competing interests None declared.

  • Ethics approval Our institutional ethics committee and Plataforma Brasil, number of CAAE: 61913414.5.0000.5440.

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

  • Data sharing statement Unpublished or unprocessed data, protocols, or images are available upon request from the corresponding author.

  • Correction notice Since this paper was first published online an author’s name has been updated. Oscar Benedicto Colli has been corrected to read Benedicto Occar Colli.