Adjuvant Embolization with N-butyl Cyanoacrylate in the Treatment of Cerebral Arteriovenous Malformations: Outcomes, Complications, and Predictors of Neurologic Deficits

Abstract

Background and Purpose

To assess the frequency, severity, and predictors of neurologic deficits following adjuvant embolization for cerebral arteriovenous malformations (AVMs).

Methods

From 1997-2006, 202 of 275 AVM patients received embolization prior to microsurgery (n=176) or radiosurgery (n=26). Patients were examined before and after endovascular embolization, and at clinical follow-up (mean 43.4±34.6 months). Outcome was classified according to the modified Rankin Scale (mRS). New neurological deficits after embolization were defined as minimal (no change in overall mRS), moderate (mRS≤2), or significant (mRS>2).

Results

202 patients were treated in 377 embolization procedures. There were a total of 29 new clinical deficits after embolization (8% of procedures; 14% of patients), of which 19 were moderate or significant. Post-embolization deficits resolved in a significant number of patients over time (p<0.0001). Five patients suffered persistent neurological deficits due to embolization (1.3% of procedures; 2.5% of patients). In multivariate analysis, the following variables significantly predicted new neurological deficit following embolization: complex AVM with treatment plan specifying more than one embolization procedure (OR=2.7; 95% CI, 1.4-8.6), diameter <3cm (OR=3.2; 95%, CI 1.2-9.1), diameter >6cm (OR=6.2; 95% CI, 1.0-57.0), deep venous drainage (OR=2.7; 95% CI, 1.1-6.9) or eloquent location (OR=2.4; 95% CI, 1.0-5.7). These variables were weighted and used to compute an AVM Embolization Prognostic Risk Score for each patient. A score of 0 predicted no new deficits, a score of 1 predicted a new deficit rate of 6%, a score of 2 predicted a new deficit rate of 15%, a score of 3 predicted a new deficit rate of 21%, and a score of 4 predicted a new deficit rate of 50% (p<0.0001).

Conclusions

Small and large size, eloquent location, deep venous drainage, and complex vascular anatomy requiring multiple embolization procedures are risk factors for the development of immediate post-embolization neurological deficits. Nevertheless, a significant number of patients with treatment-related neurological deficits improve over time. The low incidence of permanent neurological deficits underscores the utility of this technique in carefully selected patients.

Introduction

The goal of treatment in cerebral arteriovenous malformations (AVMs) is elimination of intracerebral hemorrhage risk, alleviation of clinical symptoms, and preservation or improvement of neurological function.¹ Microsurgery, radiosurgery, and endovascular embolization have all been used successfully in various combinations. Treatment planning requires selection of a modality or a combination of modalities with the greatest success rate according to patient characteristics and AVM morphology.^2-6 Embolization-related morbidity and mortality vary greatly in reports. ^6-35 Risks depend on patient selection, treatment modalities, and outcomes measures.^{6, 8-13, 27, 28, 31, 35, 36} Risk is also related to the goals of endovascular embolization therapy. In the past, embolization was commonly used as “primary therapy”.¹⁰ However, more recently studies demonstrated that AVMs treated only with embolization have low obliteration rates.^{7, 11, 16-19, 24, 31, 34, 35} Therefore, embolization is usually not recommended as single modality therapy except for palliation of non-surgical or non-radiosurgical AVMs.^{3, 19}

Beginning in 1997, the treatment paradigm at our institution changed significantly with the introduction of gamma-knife radiosurgery, regular application of intra or post-operative angiography, and application of the Spetzler-Martin grading system biased against treating high grade (Spetzler-Martin 4 and 5) AVMs. Moreover, except in uncommon circumstances requiring palliation only, embolization has been generally used only as a pre-operative adjuvant prior to microsurgical resection or radiotherapy. The goals of this study were: (1) to analyze the frequency, severity, and types of neurological deficits following pre-operative embolization of cerebral arteriovenous malformations; (2) to determine how these deficits evolve over time; (3) to assess the predictors of new neurological deficits after embolization; and (4) to use multivariate analysis to identify predictors of endovascular treatment outcomes.

Clinical Materials and Methods

Between 1997 and 2006, a total of 275 AVM patients were treated at Columbia University Medical Center. Two hundred and two of these patients (74%) underwent catheter cerebral arteiography and endovascular embolization as a part of multimodality therapy. Following embolization treatment 176 patients (87%) were underwent microsurgical resection and 26 (13%) received gamma-knife radiosurgery.

Patient Selection

A team of cerebrovascular microsurgeons, endovascular neurosurgeons, and radiosurgeons evaluates each brain AVM to determine the best treatment plan. The goal of combined multimodality intervention was complete elimination of the AVM along with preservation of normal neurological function or alleviation of neurological deficits. Treatment planning was based on selecting a modality or a combination of modalities with the greatest success rate according to patient characteristics and AVM morphology.

Outcome Measures

We retrospectively analyzed the charts of 275 patients from a historical AVM database. All patients were examined immediately before and after each embolization procedure. Long-term outcomes were recorded through in-person follow-up (88 patients, 41%) or structured telephone interviews (110 patients, 51%). Seventeen patients (8%) were lost to long-term follow-up after microsurgery. In these patients outcome was assessed on hospital discharge after completion of treatment (mean time from embolization to discharge after surgery and follow-up assessment in 17 patients lost to long-term follow-up, 2.0 ± 2.9 months). All patients were alive after embolization and surgery. Mean follow-up in all patients was 43.4 ± 34.6 months.

Neurological outcomes were stratified according to the modified Rankin scale.³⁷ New neurological deficits after embolization were defined as minimal if there was no change in modified Rankin Scale (mRS), moderate (mRS≤2), or significant (mRS>2). If a patient with significant pre-existing disability (mRS >2) had a new deficit due to treatment that did not change his/her overall function, the deficit was still recorded as a new significant deficit as a result of treatment. If a deficit was made worse due to microsurgery or radiosurgery (i.e. occurring in the same distribution), assessment on follow-up was made as to the overall change in mRS.

Embolization Technique

Endovascular embolization was performed using transfemoral approach with patients under monitored anesthetic care general. Biplane, high-resolution digital subtraction angiography was used for treatment planning in each case. Microcatheter navigation of the cerebral vasculature was performed using subtraction roadmap imaging. Super-selective angiography was performed prior to embolization, and provocative anesthetic testing was performed as needed using amobarbitol and lidocaine. N-butyl cyanoacrylate (Trufill^® NBCA, Cordis Neurovascular, Miami Lakes, FL) was used to occlude the vascular nidus once the microcatheters were advanced into position.

The purpose of pre-operative embolization was staged reduction of blood flow in an AVM by stepwise occlusion of the AVM nidus over several weeks, elimination of deep-feeding arteries such as lenticulostriates that could require extension of the resection plane beyond the margin of the nidus, occlusion of any other feeding arteries not readily accessible during surgery such as skull base dural collaterals, and coil occlusion of feeding artery aneurysms distant from the operative exposure of the AVM itself. Occlusion was performed according to the goals of treatment: 1) nidal occlusion to reduce AVM size and for flow reduction through the fistula; 2) provocative anesthetic testing and occlusion of deep feeding arteries or arteries not readily accessible to surgery; 3) endosaccular occlusion for feeding artery aneurysms without impairment of the parent artery lumen. Standard microsurgical techniques were carried out as previously described.³⁸ All patients received intra-operative or post-operative catheter angiography to confirm complete obliteration.

Statistical Analysis

Univariate analysis was used to test the ability of patient and AVM characteristics to predict new post-embolization neurological deficits. Factors predictive in univariate analysis (p<0.15)³⁹ were entered into a stepwise backwards multivariate logistic regression analysis to ascertain the effects of patient and AVM characteristics on immediate post-embolization neurological complications while controlling for patient gender and age. These variables were weighted and used to compute an AVM Embolization Prognostic Risk Score for each patient. The Mantel-Haenszel test for linear association was used to assess the significance between ascending AVM Embolization Prognostic Risk Score and incidence of deficits.

Results

After 1997, more narrow selection criteria for surgery were adopted; intra-operative or immediate post-operative angiography was routinely performed; the institution's gamma knife center was opened, and endovascular treatment was regularly used for pre-operative treatment. Two hundred-two patients received a total of 377 embolizations. One hundred-ten (55%) had one embolization treatment, 47 (23%) two treatments, 28 (14%) three treatments, and 17 (8%) greater than 3 treatments. Baseline characteristics of patients undergoing endovascular embolization are shown in Table 1.

Table 1. Baseline Characteristics of 202 Patients Treated for Cerebral AVMs^*.

Age	35±13 (4-75)
Female gender	118 (58%)
Race
White	131 (65%)
Hispanic	31 (15%)
Black	9 (5%)
Asian	9 (5%)
Other	22 (10%)
Medical History
Hypertension	11 (6%)
Diabetes mellitus	5 (3%)
Hypercholersterolemia	6 (3%)
Cardiac Disease	3 (2%)
Smoking	30 (15%)
Initial Presentation
Hemorrhage	79 (39%)
Seizure	60 (30%)
Headache	101 (50%)
Neurological deficit	80 (40%)
Syncope	11 (6%)
Asymptomatic	7 (4%)
AVM characteristics
Number of major arterial feeders	1.5±0.7
Concurrent Aneurysm	86 (43%)
Spetzler-Martin criteria
Deep Venous Drainage	89 (44%)
Eloquent location	60 (30%)
Small diameter (<3cm)	118 (58%)
Medium diameter (3-6cm)	80 (40%)
Large (>6cm)	4 (2%)
Spetzler-Martin Score
1	49 (24%)
2	78 (39%)
3	66 (33%)
4	9 (4%)
5	0
RBGS‡
<0.5	7 (4%)
0.5-1.0	68 (34%)
>1.0 <1.5	65 (32%)
>1.5	62 (30%)
Treatment
Embolizations	377
Surgery after endovascular treatment	176 (87%)
Radiosurgery after endovascular treatment	26 (13%)

^*Mean follow-up 43.4±34.6 months

^‡Radiosurgery Based Grading Scale (RBGS) = (0.1)(volume) + (0.02)(patient age) + (0.3)(location)

Prior to the first embolization, 91% of patients had no significant disability mRS (0-2) and after the last embolization 86% of patients had no significant disability (Table 2a). On long-term follow-up at 43.4 ± 34.6 months on average, 94% of patients had no significant disability. There were 29 new clinical deficits after embolization (8% of procedures; 14% of patients, Table 2b). Ten were minimal (2.7% of procedures), 9 were moderate (2.4% of procedures) and 10 were significant (2.7% of procedures). On long-term follow-up, post-embolization deficits remained in no patients, there, 4 had persistent moderate deficits (1.1% of procedures), and one had a significant deficit (0.3% of procedures). Post-embolization deficits resolved in a significant number of patients on long-term follow-up (mean 43.4±34.6 months, p=0.0001).

Table 2a.

Overall 0utcome over time

verall mRS	Prior To 1st Embolization	After Last Embolization	Long-Term Follow-Up
0-2	184 (91%)	174 (86%)	173 (94%)
3	15 (7%)	20 (10%)	10 (5%)
4	3 (2%)	8 (4%)	2 (1%)
5	0	0	0
6	0	0	0
Table 2b. New neurological deficits due to embolization over time
Immediate Deficits	Any Deficits	Moderate or Significant deficit	Significant Deficit
	29 (14%; 8%)	19 (9%; 5%)	10 (5%; 3%)
Long-Term Deficits	Any Deficits	Moderate or Significant deficit	Significant Deficit
	5 (2.5%; 1.3%)	4 (2%; 1%)	1 (0.5%; 0.3%)
Table 2c. Complications in 377 Embolizations
Complications	Number (%)
Peri or Post-procedurial hemorrhage	10 (2.6%)
Infarction	6 (1.6%)
Dissection	2 (0.5%)
Femoral/Retroperitoneal Hematoma	2 (0.5%)
Allergic reaction	1 (0.3%)
Retrograde Arterial Thrombosis	1 (0.3%)

Presented as n (% per patients; % per embolization treatment) Outcome was classified according to the modified Rankin Scale (mRS). New neurological deficits after embolization were defined as minimal (no change in overall mRS), moderate (mRS≤2), or significant (mRS>2).

Technical complications of the embolization procedures are shown in Table 2c. Intracranial hemorrhage (10 patients) and infarction (6 patients) were the most common peri or post-procedural complications. Two patients (2.5%) with medium diameter AVMs had hemorrhage versus 7 patients (6%) with small diameter AVMs (univariate logistic regression: OR=5.3; p=0.13) or 1 patient (25%) with a large diameter AVM (OR=13; p=0.018, Figure 1). There were no peri or post-procedural infarctions in medium sized AVMs, but there was a trend towards increased risk of infarction in large AVMs versus small AVMs (univariate logistic regression: Figure 1).

Figure 1.

Relationship between Spetzler-Martin Grading Scale size and peri-procedural or post-procedural hemorrhage, infarction, or new deficit following embolization

Factors predictive of new deficit due to embolization in univariate analysis (p<0.15) were staged embolization using more than one procedure (OR=2.6; 95% CI, 1.1-5.9), deep venous drainage (OR=2.3; 95% CI, 1.0-5.5), and eloquent location (OR=1.8; 95% CI, 0.8-4.1, Table 3). Additionally, patients with medium diameter AVMs (3-6cm) were less likely to experience a neurological deficit due to embolization (OR=0.3; 95% CI 0.1-1.0), while patients with large diameter AVMs (>6cm) were more likely (OR=7.4; 95% CI 1.0-57.0). As might be expected, patients with peri- or post-embolization complications were more likely to have post-embolization deficits: hemorrhage (OR=32.5; 95% CI 6.5-163.7) and 5 of 5 patients with infarction.

Table 3. Factors Predictive of New Neurological Deficit in Univariate Analysis.

Characteristic	N (%)	OR	95% CI	Significance
Age	35±13 (4-75)	1.02	0.99-1.05	0.180
Female gender	118 (58)	0.98	0.38-2.54	0.961
Race
White	131 (65)	1.00
Hispanic	31 (15)	0.97	0.11-8.27	0.975
African American	9 (5)	0.97	0.11-8.27	0.975
Asian	9 (5)	0.26	0.32-2.02	0.198
Other	22 (10)	0.36	0.05-2.94	0.346
Medical History
Hypertension	11 (6)	0.96	0.11-7.94	0.971
Smoking	30 (15)	1.61	0.59-5.23	0.428
Initial Presentation
Hemorrhage	79 (39)	1.03	0.38-2.75	0.947
Seizure	60 (30)	1.74	0.66-4.56	0.262
Headache	101 (50)	0.84	0.35-2.04	0.704
Neurological deficit	80 (40)	1.79	0.69-4.63	0.227
Syncope	11 (6)	0.96	0.11-7.94	0.971
Asymptomatic	7 (4)	1.02	0.22-4.74	0.984
No neurological deficit at baseline	184 (91%)	1.53	0.70-3.39	0.295
AVM Characteristics
Number of arterial feeders	1.5±0.7	0.85	0.41-1.74	0.658
Concurrent Aneurysm	86 (43)	0.98	0.37-2.55	0.965
Spetzler-Martin criteria
Small diameter (<3cm)	118 (58)	1.00
Medium diameter (3-6cm)	80 (40)	0.29	0.08-1.04	0.058
Large (>6cm)	4 (2)	7.43	0.97-57.00	0.054
Deep Venous Drainage	89 (44)	2.31	0.97-5.51	0.058
Eloquent location	60 (30)	1.83	0.82-4.13	0.141
Spetzler-Martin Score
1	49 (24)	1.00
2	78 (39)	1.85	0.55-6.16	0.318
3 or 4	75 (37)	1.58	0.38-6.63	0.533
RBGS location*†	202 (100%)	1.25	0.52-3.02	0.625
RBGS	202 (100%)	1.10	0.77-1.56	0.608
>1 Embolization procedure planned	92 (46%)	2.60	1.1-5.9	0.023
Peri/post-procedural hemorrhage	10 (2.7%)	32.50	6.5-163.7	<0.001
Peri/post-procedural infarction⌃	5 (1.32)	5 of 5

^*Radiosurgery Based Grading Scale (RBGS) = (0.1)(volume) + (0.02)(patient age) + (0.3)(location)

^†RBGS location: frontal, temporal=0; parietal, occipital, intraventricular, corpus collosum, cerebellar=1; basal ganglia, thalmic, brainstem=2. When an AVM involves multiple sites, fractional values are used according to the number of sites (0.5 for two sites, 0.33 for three sites)

^⌃Infarction predicts any new neurological deficit in 5 out of 5 patients

In multivariate analysis controlling for patient gender and sex, the following pre-embolization variables were predictive of new neurological deficit: staged embolization using more than one procedure (OR=2.7; 95% CI, 1.4-8.6), deep venous drainage (OR=2.7; 95% CI, 1.1-6.9), eloquent location (OR=2.4; 95% CI 1.0-5.7), and small diameter (<3cm, OR=3.2; 95% CI, 1.2-9.1, Table 4). Although there were only 4 patients with large diameter AVMs, size greater than 6 cm also significantly predicted new deficit after embolization. (OR=6.2; 95% CI 1.0-57.0). The relationship between AVM size and neurological deficit following embolization is demonstrated in Figure 1.

Table 4. Multivariate Logistic Regression Analysis of the Predictive Value of Pre-embolization Patient AVM characteristics on New Neurological Deficit.

Variable	OR	95% CI	Significance
Spetzler Martin Components
Size
Medium diameter (3-6cm)	1.00
Small diameter (<3cm)	3.20	1.2-9.1	0.025
Large (>6cm)	6.23	1.0-57.0	0.048
Deep venous drainage	2.73	1.1-6.9	0.034
Eloquent location	2.38	1.0-5.7	0.050
>1 Embolization planned	2.70	1.4-8.6	0.008

^*Age and gender were not significant predictors of outcome. When adjusting for age and sex, significance was unchanged. Peri-embolization complications such as peri-procedural hemorrhage or infarction were not included in this analysis.

Variables predictive of embolization-related deficits in multivariate analysis were weighted and used to compute an AVM Embolization Prognostic Risk Score for each patient (Table 5). Patients receive one point if their treatment plan requires more than one embolization procedure, one point for small diameter (<3cm) AVM, one point for eloquent location, one point for deep venous drainage, and two points for large size (>6cm). Summation of each patient's points yields a score, range of 0-5. A score of 0 predicted no new deficits, a score of 1 predicted a new deficit rate of 6% (100% moderate/significant), a score of 2 predicted a new deficit rate of 15% (40% moderate/significant), a score of 3 predicted a new deficit rate of 21% (71% moderate/significant), and a score of 4 predicted a new deficit rate of 50% (100% moderate/significant) (Mantel-Haenszel test for linear association, p<0.0001). Based on selection criteria, no treated patients had a score of 5. Although the AVM Embolization Prognostic Risk Score predicted immediate deficits due to embolization, it did not predict long-term deficits as a significant number of patients had improvement over time.

Table 5. AVM Embolization Prognostic Score.

AVM Embolization Prognostic Score*	No. of Patients	Any Deficit (n=29)†	Moderate or Significant Deficit (n=19)⌃	Significant Deficit (n=10)Δ	Long-Term Moderate or Significant Deficit (n=4)
0	12	0%	0%	0%	0%
1	52	6%	6%	6%	2%
2	83	15%	6%	4%	1%
3	47	21%	15%	4%	4%
4	8	50%	50%	25%	0%

^*Greater than 1 embolization treatment planned, deep venous drainage, eloquent location, and small AVM size (0-3cm) are all assigned 1 point and AVM large size is assigned 2 points

^†Any deficit includes transient/minimal, moderate, and significant deficits. There is a significant association between increasing AVM embolization Prognostic Score and risk of deficit after embolization (Mantel-Haenszel test for linear association, p<0.0001)

^⌃There is a significant association between increasing AVM embolization Prognostic Score and risk of moderate or significant deficit after embolization (Mantel-Haenszel test for linear association, p<0.006)

^ΔThere is not a significant association between AVM embolization Prognostic Score and risk of significant deficit after embolization (Mantel-Haenszel test for linear association, p<0.095)

Discussion

In general, endovascular embolization is performed before surgical resection of AVMs to reduce the size of the active nidus and establish more normal blood flow patterns for brain tissue surrounding the AVM. Embolization is used for Spetzler-Martin Grade II or III lesions prior to microsurgery or radiosurgery but only used in Grade IV or V lesions in multi-modality care where the goal of treatment is complete occlusion or resection.³ Identifying factors that predict complications and poor outcome after embolization would improve patient selection and AVM treatment outcomes. Predictors of complications and outcome may also help guide patient management.

In this study we found that: 1) adjuvant embolization in the treatment of cerebral AVMs results in a low rate of neurological deficits; 2) small and large size, eloquent location, deep venous drainage, and the number of planned embolization procedures predict post-embolization neurological deficits; and 3) these variables may be weighted in the AVM Embolization Prognostic Risk Score to predict new neurological deficits after embolization, 4) a significant number of patients with deficits will improve over time; and 5) the low incidence of permanent neurological deficits underscores the utility of this technique in carefully selected patients.

Studies have demonstrated that preoperative embolization leads to decreased operative time, blood loss, and morbidity and mortality in cerebral AVM surgery.^{20, 40, 41} Embolization may be used as: 1) primary therapy, i.e. without surgical resection or radiotherapy; 2) adjuvant embolization prior to microsurgery or radiosurgery; or 3) for palliation of inoperable or otherwise uncurable AVMs. Prior studies suggest that primary embolization results in a low obliteration rate;^{7, 11, 16-19, 24, 31, 34, 35} therefore, embolization as a single modality is not generally recommended.³ In “palliative care” of inoperable AVMs, embolization may be used to treat or reverse a specific symptom due to high-flow cardiovascular impairment, venous hypertension, or arterial steal phenomenon.^{3, 19} In this study, pre-operative embolization was used only in conjunction with microsurgical resection or gamma knife treatment. Specifically, staged pre-operative embolization was used to decrease flow as a strategy to normalize cerebral blood flow, eliminate deep-feeding arteries, occlude feeding vessels not readily accessible during microsurgery, reduce volume of the active nidus, and coil occlude feeding artery aneurysms outside the operative field at the time of AVM resection.

There were 29 new post-embolization clinical deficits (8% of procedures; 14% of patients), of which 9 were moderate and 10 were significant. After a mean follow-up of 43.4±34.6 months, there were 5 persistent neurological deficits due to embolization (1% of procedures; 2% of patients) of which 4 were moderate, and one was significant. This is the first series to demonstrate that the majority of deficits following embolization are transient, with a significant number of moderate and severe deficits improving over time. In the literature there has been a wide range of reported morbidity (1-50%) and mortality (1-4%) following embolization of AVMs.^6-35 Recent and large studies have found a permanent morbidity rate of 3-14% and a mortality rate of 0-4%.^{9, 10, 14, 15, 18, 23, 35, 42-44} Rates are dependent on many factors: patient selection, embolic agents, means of delivering the occlusive agent, goals of embolization i.e., primary curative or adjuvant therapy, time of outcome assessment, and pre-existing neurological morbidity.^{6, 8-13, 27, 28, 31, 35, 36}

It is important to note inherent limitations of our outcome data. The modified Rankin scale does not measure higher cortical function and postoperative cognitive decline. This must be taken into account when considering operative morbidity.^{45, 46} Moreover, neuropsychological testing following surgical AVM excision is critical to accurate outcome assessment and should be integrated into future clinical trials.

In our study there were 10 peri-procedural or early hemorrhages following embolization (2.6% of procedures; 4% of patients), which is similar to bleeding complication rates reported in the literature (1-2% of procedures; 3-15% of patients).^{20, 47, 48} Hemorrhage was more likely to occur in small and large AVMs. Hemorrhage during or after embolization may be due to vessel perforation during catheterization, AVM rupture, intranidal aneurysm rupture, or hemodynamic changes due to alterations in feeder pressures, normal perfusion breakthrough, or venous outflow obstruction.^48-52 Further studies are necessary to elucidate the etiology of these bleeds, the factors predictive of peri- or post-embolization bleeds, and their relevance towards long-term outcome.

In multivariate analysis, greater than one embolization procedure planned, small and large diameter nidus, deep venous drainage, and eloquent location predicted new neurological deficit following embolization. The Spetzler-Martin grading scale was designed and validated to predict surgical treatment outcome, and its applicability to endovascular embolization is unclear.^{5, 53} Previous studies have found that not all elements of the Spetzler-Martin grading scale are predictive in outcomes following embolization.^{9, 10, 42} This may in part be due to the assumption that AVM size in the Spetzler-Martin grading scale has a linear relationship with outcome, as it does in microsurgery. In our series, however, patients with small AVMs and larger AVMs were significantly more likely to develop deficits after embolization. One explanation is that we have demonstrated that small and large AVMs are more likely to hemorrhage following embolization. Small AVMs may be technically more difficult to embolize than medium sized AVMs. The increased propensity of small, untreated AVMs to bleed has been demonstrated in many studies, and may be due to increased pressure within small AVMs.^54-56 Although there are relatively few patients with large AVMs in this study, size greater than 6cm is predictive in univariate and multivariate analysis. These results should be confirmed in studies of patients with large AVMs, but we expect that deficits are more likely to occur following embolization of larger AVMs as it is more difficult to achieve complete embolization resulting in an increased risk of hemorrhage versus medium sized AVMs.

These variables were weighted based on their odds ratios to create the AVM Embolization Prognostic Risk Score to facilitate the prediction of outcome. One point is assigned for each variable and two points are assigned for large AVM size (2 points). A prognosis score of 0 predicted neurological deficits in 0% of patients, a prognosis score of 1 predicted deficits in 6% of patients, a score of 2 predicted deficits in 15% of patients, a score of 3 predicted any deficit in 21%, and a prognosis score of 4 predicted deficits in 50% of patients. The AVM prognostic score is predictive of immediate deficits, but as a significant number of patients improved, it was not predictive of long-term persistent deficits. This may be a type II error due to the small number of patients with long-term deficits. The AVM prognostic score is also shaped by selection bias. We have moved towards more conservative management of patients with large (Spetzler-Martin 4-5) AVMs.^{6, 9, 10} Validation in large clinical trials is necessary.

Although the AVM Embolization Prognostic Risk Score predicts deficits due to embolization, a significant number of patients had improvement over time. Further studies are needed to determine how complications from embolization affect overall long-term outcome. The true risk and benefit profile for embolization of brain AVMs in multimodality therapy is not clearly defined. Considered beneficial by some authors,^{2, 3, 9, 41, 57} AVM embolization has reportedly resulted in excess complications which may outweigh its benefits.^{16, 58, 59} Case control studies have demonstrated a beneficial effect of embolization prior to surgery,⁴¹ but not before radiosurgery.⁵⁸

Our results are limited to patient treated with NBCA. Recent studies of patients treated with Onyx demonstrate similar obliteration rates^{21, 60-65} as those case series using NBCA, but assessment of recanalization and long-term effects are currently being assessed. Additionally many centers are using Onyx as a primary modality of therapy and microsurgery and radiosurgery as a secondary modality in patients with incomplete obliteration. This partially accounts for the higher complication rates observed in some studies. Preliminary studies with Onyx are promising, but further studies necessary. Our results would be beneficial in treatment planning of Onyx patients if confirmed in this cohort.

This study demonstrates that significant neurological complications occur in 2.7% of patients immediately following embolization, yet a majority of deficits improve or resolve over time. The combined experience and overwhelming bias of 3 microsurgeons and two neurointerventionalists support pre-operative embolization in nearly all cases whereby definitive treatment with microsurgery is deemed beneficial. Selective cases to be treated with stereotactic radiosurgery might also benefit from preoperative embolization. Our surgical experience over the last 25 years has shown a progressive decrease in operative complications, significantly decreased operative time and blood loss as endovascular techniques have improved, resulting in more complete, staged pre-operative AVM occlusion.⁶⁶

Conclusions

Adjuvant embolization in the treatment of cerebral AVMs results in a low rate of neurological deficits. Eloquent location, deep venous drainage, large size, and multiple embolization procedures are risk factors for the development of post-embolization neurological deficits. Although the AVM Embolization Prognostic Risk Score predicts complication rates in this patient population, a significant number of patients improve over time. The low incidence of permanent neurological deficits underscores the utility of this technique in carefully selected patients. Larger, multi-center trials are necessary to determine how deficits from embolization affect overall long-term outcome.