Article Text

Download PDFPDF

Original research
Endovascular treatment as the main approach for Spetzler–Martin grade III brain arteriovenous malformations
  1. Humain Baharvahdat1,2,
  2. Raphaël Blanc1,
  3. Robert Fahed1,3,
  4. Ashkan Pooyan2,
  5. Ashkan Mowla4,
  6. Simon Escalard1,
  7. François Delvoye1,
  8. Jean Philippe Desilles1,
  9. Hocine Redjem1,
  10. Gabriele Ciccio1,
  11. Stanislas Smajda1,
  12. Mylène Hamdani1,
  13. Mikael Mazighi1,
  14. Michel Piotin1
  1. 1 Department of Interventional Neuroradiology, Fondation de Rothschild, Paris, France
  2. 2 Neurosurgery, Mashhad University of Medical Sciences, Mashhad, Iran (the Islamic Republic of)
  3. 3 Medicine – Neurology, Ottawa Hospital, Ottawa, Ontario, Canada
  4. 4 Neurosurgery, University of Southern California Keck School of Medicine, Los Angeles, California, USA


Background Because Spetzler–Martin (SM) grade III brain arteriovenous malformations (bAVMs) constitute a heterogeneous group of lesions with various combination of sizes, eloquence, and venous drainage patterns, their management is usually challenging. The aim of this study is to evaluate the clinical/imaging outcomes and the procedural safety of endovascular approach as the main treatment for the cure of SM grade III bAVMs.

Methods In this retrospective study, prospectively collected data of SM grade III bAVMs treated by endovascular techniques between 2010 and 2018 at our hospital were reviewed. Patients older than 16 years with angiographic follow-up of at least 6 months after endovascular treatment were entered in the study. The patients had a mean follow-up of 12 months. The data were assessed for clinical outcome (modified Rankin Scale), permanent neurological deficit, post-operative complications, and optimal imaging outcome, defined by complete exclusion of AVM. The independent predictive variables of poor outcome or hemorrhagic complication were assessed using binary logistic regression.

Results Sixty-five patients with 65 AVMs were included in the study. Mean age of the patients was 40.0±14.4. Most common presentation was hemorrhage (61.5%). The patients underwent one to eight endovascular procedures (median=2). Mean nidus diameter was 30.2±13.0. A complete obliteration of AVM was achieved in 57 patients (87.7%). Post-procedure significant hemorrhagic and ischemic complications were seen in 13 (20%) and five (7.7%) patients respectively, leading to five (7.7%) transient and four (6.2%) permanent neurological deficits. Eight patients (12.3%) experienced worsening of mRS after embolization. Ten patients (15.4%) had poor outcome (mRS 3–5) at follow-up and two (3%) died.

Conclusions Endovascular treatment can achieve a high rate of complete exclusion of grade III AVM but may be associated (as in other treatment modalities) with significant important complications.

Clinical trial registration number NCT02879071.

  • arteriovenous malformation
  • liquid embolic material
  • hemorrhage
  • intervention

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.


Spetzler–Martin (SM) grading was developed to assess the risk of microsurgical resection of brain arteriovenous malformation (bAVMs).1 It is also now widely applied to plan the management strategy of bAVMs. Despite the ARUBA trial results, treatment is sometimes being offered to low-grade bAVMs (SM grade I and II) because of the relatively low rate of associated morbidity/mortality, while high-grade bAVMs (SM IV/V) are often managed conservatively. The highly variable SM Grade III bAVM lies within these two categories. It includes four possible combinations of size, localization, and venous drainage defined by the Lawton sub-classification2 of the SM grading: type IIIA or S1V1E1 (small size, deep venous drainage, and eloquent location), type IIIB or S2V1E0 (medium sized, deep venous drainage, and noneloquent location), type IIIC or S2V0E1 (medium sized, superficial venous drainage, and eloquent location), and type IIID or S3V0E0 (large sized, superficial venous drainage, and noneloquent location). Because of its heterogeneity, the management of SM grade III bAVMs is frequently a debate. Several studies addressed the feasibility and outcome of surgery and radiosurgery as a main treatment of SM grade III bAVM,2–5 but none specifically analyzed the outcome of embolization alone. In this study we assessed the efficacy and outcome of endovascular treatment (EVT) as the main treatment for cure of SM grade III bAVMs.


Basic characteristics of patients and bAVMs

From our prospectively collecting database we reviewed all bAVMs treated by the endovascular approach in our department between January 2010 and December 2018. Non-pregnant patients, ≥16 years old with SM grade III AVMs were included in the present study. Baseline clinical characteristics of patients including age, sex, clinical presentation, neurological status on admission according to modified Rankin scale (mRS), were recorded. Cross-sectional imaging (cerebral MRI and CT scan) and angiograms were reviewed to determine bAVM location, size (mm), and angioarchitecture. BAVMs were classified as cortical/subcortical, deep or infratentorial. Based on size (S), involvement of eloquent area (E), and venous drainage pattern(V), SM grade III bAVMs were further classified to four subtypes: IIIA (S1E1V1), IIIB (S2E0V1), IIIC (S2E1V0), and IIID (S3E0V0).6 The anatomical grading was established by two senior operators independent from the endovascular treatments. The study protocol was approved by the local ethics committee, the board waived the need for signed consent for patients included in the study.

Endovascular procedures

Patients were allocated to treatment after multidisciplinary discussions involving interventional neuroradiologists and at least one vascular neurosurgeon. In general, treatment was offered to all patients with a history of ruptured AVM and to the patients with unruptured AVMs who had high natural risk of AVM rupture, such as single venous drainage and intranidal aneurysms. During the period of investigation, EVT was the first-choice of treatment for SM grade III bAVMs with the intention to cure in one or multiple sessions.

EVTs were performed under general anesthesia. The following liquid embolic agents were used for the embolization of one or several pedicles in each session: Onyx (Covidien/Medtronic, Irvine, California), n butyl-cyanoacrylate glue (Histoacryl; Braun, Melsungen, Germany), and Glubran2 (GEM, Viareggio, Italy). Sometimes microcoils were used to shut down intra-AVM fistulae. Most commonly a trans-arterial approach was used. Only two patients were treated by a transvenous approach. Occasionally, other approaches including double catheter techniques were used. A balloon-micro-catheter was not used for embolization in any of the patients. In each session, one or multiple pedicles were chosen for embolization. The number of embolized pedicles in each session was decided by the operator. During each pedicle embolization, the procedure was stopped when complete occlusion (no residual early venous filling including intentional occlusion of the draining vein/veins) was achieved, when there was 2 cm reflux of Onyx in a non-detachable microcatheter or Onyx reflux to the level of the proximal marker of the detachable part of the catheter. A head CT scan was performed with the C-Arm Flat panel detector in the operating room after each procedure or during operation if any perforation occurred. After the procedure, all patients were admitted to the intensive care unit with control of systolic blood pressure for 24–48 hours. Later, patients were transferred to the neurosurgical ward for 2–3 days of observation and were routinely discharged within 4–5 days of procedure if there was no complication. Cerebral MRI was performed before and after each procedure for all patients. Any new ischemia or hemorrhage in post-procedure cerebral MRI was recorded. In the case of multiple sessions of treatment, the interval between the two procedures was between 1 to 4 months. The procedural data including: number of sessions for each patient, type of technical approaches, complications during or after operation, or any post-procedure surgery were prospectively collected. The complications were classified as hemorrhagic or ischemic and were considered significant if they required surgical treatment such as an external ventricular drainage, a hematoma evacuation, a decompressive craniectomy, or if they caused an altered or decreased level of consciousness, a transient neurological deficit, a permanent neurological deficit, or a mRS grade worsening.

Clinical and angiographic outcome and follow-up

The clinical assessment of the patient was performed by a physician independent of the operators performing the EVT at the time of discharge and at 3–6 months or more of follow-up. Any new neurological deficit was collected and considered as transient if it resolved within 1 month after EVT or as permanent if it remained for more than 1 month. The clinical outcomes were recorded according to mRS and further classified as poor for mRS 3 to 6 and as good for mRS 0 to 2. Worsening of mRS was defined as any change from preoperative mRS score to higher score after the procedure and was recorded. bAVM occlusion was considered complete if the nidus was completely embolized with no residual arteriovenous shunt and no early venous filling, assessed by cerebral angiography 6 months after the last procedure.

Statistical analysis

Continuous data are presented as means±SDs and categorical data as count and percentage. Statistical comparisons were performed by student t-test for normally distributed data, the Mann–Whitney U test for data with skewed distribution, and the chi-square and Fisher’s exact tests for the categorized data. To assess the risk factors of poor outcome (mRS >2) and post-procedural hemorrhagic complication, univariate analysis was performed using baseline characteristics of patients, AVM angioarchitecture, and endovascular procedures variables. The risk factors that achieved a P-value less than 0.1 were entered in multivariate analysis to define independent predictive variables of poor outcome or hemorrhagic complication using binary logistic regression. A P-value less than 0.05 was statistically significant. The data were analyzed by using the Statistical Package for Social Sciences (Version 16.0; IBM, Armonk, New York).


Basic characteristics of patients and bAVMs

Sixty-five patients with 65 SM grade III bAVMs were included in the study (baseline characteristics in table 1). The mean age of patients was 40.5 ± 14. They presented with hemorrhage in 40 patients (61.5%), seizure in nine (13.8%), headache in seven (10.8%), isolated neurological deficits in four (6.2%), and were incidentally found in five (7.7%). Twelve patients had been partially treated by embolization before being referred to our hospital. The mean nidus diameter was 30.2 mm ± 13 (SD). The most common subtypes were IIIA (43.1%) and IIIC (35.4%). Fifty-one bAVMs (78.5%) were located in the eloquent area and the most common involved eloquent area was the sensorimotor cortex (33.8%). Except for a deep venous drainage that was more common in the ruptured bAVMs group (P=0.035), there was no significant difference between the two groups for other parameters.

Table 1

Basic characteristic of patients and bAVMs*, comparing ruptured and unruptured bAVM groups

Follow-up imaging outcome

Twenty-nine patients (43.9%) had only one session of EVT, 20 (30.3%) patients had two sessions, 11 (16.7%) patients had three sessions, and six (9%) patients had more than three sessions of embolization. Mean follow-up of patients was 12 months (min-max=6–72 months). Fifty-seven (86.4%) bAVMs were completely obliterated by EVT at angiographic follow-up. In two patients an immediate post-embolization surgery was performed and follow-up angiography confirmed the absence of residual shunt. Endovascular management was interrupted in four patients because two of them died because of hemorrhagic complications after EVT and two experienced severe complications with clinical state disqualifying them for further treatment. Twenty-four (42.1%) bAVMs were completely excluded after one session of EVT and 17 (29.8%) after two sessions. The number of EVT sessions was significantly higher in the unruptured group than the ruptured group. Table 2 shows clinical and imaging outcome based on SM grade III bAVMs subtypes.

Table 2

Clinical and imaging outcome after EVT, stratified by SM grade III bAVMs subtypes

Procedural complications and clinical outcomes

Significant hemorrhagic complications occurred in 13 patients (20%), including one subarachnoid hemorrhage (SAH), five intraventricular hemorrhage (IVH) with or without SAH, and seven intraparenchymal hemorrhage (IPH). Eleven patients required surgery, including six external ventricular drainage and five craniotomy and IPH removal. Five patients (7.7%) experienced ischemic complications, all of them treated medically. There was no significant difference for the occurrence of hemorrhagic complications between the ruptured and unruptured groups (table 3). Ischemic complications were significantly more common in the unruptured group compared with the ruptured group (0.041). In univariate analysis, age >40, diffuse nidus, presence of nidal aneurysms, single venous drainage, and presence of draining vein ectasia were predictors of post-operative hemorrhage (P=0.013, P<0.001, P=0.10, P=0.024, P=0.034, P=0.041, respectively). In multivariate analysis, age >40 was the independent predictor of post-operative hemorrhage (OR=16.222, 95% CI: 1.688 to 155.924, P=0.016).

Table 3

Procedural complications and clinical outcome, comparing ruptured and unruptured groups

Permanent neurological deficits were detected in four patients (6.2%). A motor deficit was the most common deficit. Post-EVT, good mRS was achieved in 55 patients (84.6%). Ten (15.4%) patients had a poor mRS at last visit. Two (3.1%) patients died during hospitalization due to hemorrhagic complications following embolization. In total, eight (12.3%) patients experienced mRS worsening. The rate of permanent deficit and poor outcome was not different between the ruptured and unruptured groups (table 3). Both in univariate and multivariate analysis, pre-op clinical condition (poor pre-op mRS) and hemorrhagic complications were significant independent predictors of poor outcome (in univariate analysis: OR=11.357, 95% CI: 1.608 to 80.235, P=0.023 and OR=19.056, 95% CI: 3.862 to 94.026, P<0.001 respectively and in multivariate: OR=59.744, 95% CI: 3.550 to 1005.38, P=0.005 and OR=42.455, 95% CI: 4.015 to 448.58, P=0.002 consecutively).


Multiple studies have assessed the outcome of surgical resection and radiosurgery for grade III bAVMs but no study so far has assessed the outcome of EVT as the main treatment modality for the cure of grade III bAVMs. Among the small number of grade III bAVMs in the endovascular reported series, the rate of complete obliteration ranges from 0% to 67% by trans-arterial approach.7–10 This obliteration rate increases when using the trans-venous approach, the double catheter approach, and detachable microcatheters.11–13 A trans-venous approach is applied mostly for deep-located bAVM or bAVMs without safe trans-arterial access.14 15

Surgical resection is considered to achieve complete exclusion of grade III bAVMs in over 90% of patients (table 4). de Oliveira et al achieved 98% of complete resection in grade III bAVMs.16 In Morgan et al's study, the long-term complete obliteration of SM grade III bAVMs was 97%: 20% of their patients had preoperative embolization.17 After embolization of all except 1, Lawton et al achieved complete resection in 97% of cases by surgery.2 Luzzi et al report complete resection rate of 92.5% by post-embolization surgical resection.5

Table 4

Comparison of the efficacy and complication rates of the three treatment modalities in prior studies of SM grade III bAVMs

The successful obliteration rate of radiosurgery for grade III bAVMs is less than surgery even after 5 and 10 years of follow-up (table 4). Kano et al report 72% and 77% rate of complete obliteration of grade III bAVMs after 5 to 10 years of follow-up, respectively, with radiosurgery.6 In an international multicenter study, Ding et al show obliteration rates of 63% and 78% at 5 and 10 years of follow-up after stereotactic radiosurgery.18

Pandey et al describe multimodality management of grade III bAVMs including post-embolization surgery and post-embolization radiosurgery with a 88.5% rate of complete obliteration.19 Abecassis et al report 61.9% of complete obliteration of grade III bAVMs after multimodality management, including 92.5% after surgery with or without embolization, and 32.2% after radiosurgery with or without embolization.3 Absence of a previous hemorrhage,6 18 absence of previous embolization,6 and higher margin dose6 are predictors for grade III bAVMs complete obliteration after radiosurgery in prior studies. A brief analysis of these series results is summarized in table 4.

With growing data supporting EVT as the main treatment of SM grade I/II bAVMs,20–22 we reviewed our experience on EVT of SM grade III bAVMs as the first-line treatment in this study. Our results show that embolization as a stand-alone technique can reach complete obliteration of grade III bAVMs in 86% of patients. The presence of an ectatic draining vein and a single venous drainage can decrease the rate of complete obliteration of grade III bAVMs. Since ectatic draining veins require larger amounts of embolic agents for occlusion, it is expected that with injection through trans-arterial approach, the embolic agent could obstruct the nidal outflows before sufficient volumes arrives to the ectatic vein. Consequently, the draining vein could remain open despite closure of most of the nidus and recruit small arteries. This also explains why the trans-venous approach might be a better choice to exclude the draining vein and achieve complete obliteration in bAVMs with single draining vein.

The morbidity and mortality rate of surgical resection with or without embolization have been reported as 4% to 30% and 1% to 4%, respectively.1–3 17 19 23 The latency-period hemorrhagic rate was reported as about 1.2% per year and up to 9% to 10% within 10 years after radiosurgery of grade III bAVMs.6 18 Symptomatic permanent adverse effects of radiosurgery were reported between 4% to 6% after radiosurgery of grade III bAVMs.6 18 In our study, hemorrhagic and ischemic complications occurred in 20% and 8% of bAVM3 patients leading to 6.2% permanent neurological deficit, 12% mRS worsening, and 3% death. Patients with ruptured bAVM present with imaging &/or clinically relevant findings of brain damage that could obscure post-embolization diffusion weighted imaging or clinical evaluation of post-procedural ischemic complications in comparison with patients with unruptured bAVM. Despite the high rate of post-op hemorrhagic complications, a prompt and immediate surgery, including external ventricular drainage and ICH evacuation, could prevent severe neurological deficit and poor outcome in many of our patients.

Several authors divided the heterogenous group of grade III bAVMs into subtypes for better prediction of the radiological/clinical outcome as well as the complication rate with different management strategies. de Oliveira et al defined subtype IIIA as large AVM IIIA and subtype IIIB, including small/medium AVMs with deep venous drainage or/and eloquent area involvement.16 Lawton et al classified grade III bAVMs into four subtypes: IIIA (small size with deep vein drainage and eloquent area involvement, S1E1V1); IIIB (medium size bAVMs with deep venous drainage, S2E0V1); IIIC (medium size bAVMs with eloquent area involvement, S2E1V0); and IIID (large bAVMs).2 The Lawton classification of IIIA to IIID2 is valuable in predicting the clinical outcome of surgery and multimodality management of this group of AVMs and finally to select patients for these treatments. In Lawton et al's studies, surgical risk of new deficit or death is 3% for IIIA, 7% for IIIB, and 15% for IIIC subtypes.2 Davidson and Morgan report 9% surgical morbidity in IIIA, 15% in IIIB and IIIC, and 17% in IIID subtypes.24 Subtyping of bAVM3 is not relevant in the prediction of radiation induced complications: While Andrade-Souza et al25 reported 25% radiation-induced complications in IIIA subtype and 27% in IIIB and IIIC subtypes, Kano et al present 3.5% permanent complications in subtype IIIA, 0% in subtype IIIB, and 2% in IIIC.4 In the present series, permanent neurological deficit developed in 6.9% of IIIA, 0% of IIIB, 8.7% of IIIC, and 0.0% of IIID subtypes and poor outcome was 21% in subtype IIIA, 8% in subtype IIIB, 13% in IIIC, and 0% in IIID (table 2). In our study, the rate of bAVM obliteration has a trend to decline from subtype IIIA to IIIC, however it did not reach statistical significance. Based on the result of the present study, Lawton’s subtype classification cannot be used for prediction of functional outcomes in EVT of grade III bAVMs.

In our study, pre-operative functional status (mRS >2) and significant hemorrhagic complications are independent predictors of poor outcome after EVT of bAVM patients. Hemorrhagic events during or after AVM embolization are the worst complication of EVT and are associated with a high rate of severe morbidity and mortality.26 Age over 40 is an independent predicting factor of hemorrhagic complications in multivariate analysis and age >40, single venous drainage, draining vein ectasia, diffuse nidus, and intranidal aneurysms are predicting factors of hemorrhagic complications in univariate analysis. It may be suggested that grade III bAVMs with compact nidus, multiple venous drainage, and no draining vein ectasia in a patient 40 years old or younger could be a better candidate for safe and effective EVT. EVT of grade III bAVMs might be less safe in patients older than 40 years, especially when there is a diffuse nidus, draining vein ectasia, or single draining vein. In our series, the trans-venous approach was used for two patients and it is not enough to assess the effect of the trans-venous approach on this group of patients with diffuse AVM nidus draining into a single vein. Several studies show the safety of the trans-venous approach in bAVMs with a single draining vein and diffuse nidus,14 15 so this approach is a possible option with lower hemorrhagic complication and better clinical outcome for some high-risk grade III bAVMs.


Our study presents several limitations. First, there are inherent selection biases due to the observational design of this work and the monocentric nature of the study could also affect the results. During the period of investigation, although the decision for treatment was multidisciplinary, the combination of endovascular techniques followed by a surgical resection was not considered. This type of combined management is the subject of further prospective studies in our department.

Further multicenter studies, with different treatment modalities such as surgery, EVT, and radiosurgery, may better clarify the different aspects in the management of grade III bAVMs.


EVT has the potential for a high cure rate of grade III bAVMs especially for those with compact nidus, multiple venous drainage, and no ectatic draining veins in young patients. Like other treatment modalities, EVT is associated with an important complication rate. Further prospective studies, including all different treatment modalities, could further delineate the best treatment approach for different subtypes of SM grade III bAVM.



  • Contributors HB: design, analysis, and preparing the draft. RB: design and supervision of the study and revising draft with final approval. RF: revising critically the draft. AP: acquisition of data and interpretation of data. AM: critically revising the draft. SE: acquisition of data and analysis. FD: acquisition of data. JPD: acquisition of data and interpretation of data. HR: acquisition and analysis of data. GC: acquisition of data. SS: supervision of correct data collection. MH: analysis of data. MM: acquisition of data. MP: design and supervision of the study and final approval of version published.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial, or not-for-profit sectors.

  • Disclaimer The study protocol was approved by the hospital local ethics committee. There is no conflicting interest for this study. All authors have their own specific responsibility and contribution in one of the following parts of this study: designing/conducting the study, collecting data, analyzing the data, preparing the manuscript, critically revising the draft, approving the final draft, and supervision of the study.

  • Competing interests None declared.

  • Patient consent for publication Not required.

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