Interactive Article


Embolization of Cerebral Arteriovenous Malformations: Part I­­Technique, Morphology, and Complications

Gunnar Wikholm, M.D., Ph.D., Christer Lundqvist, M.D., Ph.D., Paul Svendsen, M.D., Ph.D.

Departments of Neurology (CL) and Interventional Radiology (GW, PS), Sahlgrenska University Hospital, Göteborg, Sweden

OBJECTIVE: The aim of this two-part study is to give a full account of all patients referred for embolization of arteriovenous malformations (AVMs) from 1987 to 1993. This article (Part I of II) presents the patient study, including angiographic features and their relation to the immediate outcome of embolization.
METHODS: Of the 192 patients referred, 150 were subsequently treated. Most patients were referred by neurosurgeons, and 85% of the AVMs were Spetzler-Martin Grade greater than or equal to 3. We have accounted for the 42 patients who did not undergo embolization.
RESULTS: Occlusion from embolization alone (total embolization) was obtained in 13% of patients. Full treatment (total embolization or embolization and then stereotactic radiation or surgery) was achieved in two-thirds of all patients (n = 100, 66%), and combined treatment with stereotactic gamma radiation was the most important part of the treatment strategy. The procedural mortality was 1.3%. The total incidence of complications after embolization was high (40%), but only 6.7% of cases were labeled severe. Of all angiographic features that were considered, large size and the presence of deep feeders were predictors of failure to achieve full treatment. Thirty-four patients with AVMs <8 cc were included in the study. These could have been irradiated as the sole treatment. In this group of small AVMs, the results of embolization were far better than in the whole group. Fourteen of the AVMs had volumes of <4 cc, and 10 of these (71%) were totally embolized. One patient had a hemianopsia. Among AVMs 4 to 8 cc in volume (n = 20), the total embolization rate was 15%, the full treatment rate in combination with gamma treatment was 75%, and 10% of the patients were operated on after embolization. Severe complications occurred in 15% of patients, but no complications occurred after November 1990.
CONCLUSION: In a series of AVMs, most of which were regarded as unsuitable for surgical excision, two-thirds were reduced to a size suitable for gamma knife treatment or totally occluded by embolization alone. The total complication rate was high, but the combined rate of death and complications affecting lifestyle was 8.0%, equal to ~3.2 years of natural history.
(Neurosurgery 39:448­459, 1996)

Key words: Arteriovenous malformations, cerebral; Arteriovenous malformations, embolization; Complications; Treatment results

In 1960, the first modern report on embolization of cerebral arteriovenous malformations (AVMs) appeared in the literature (23). There has been continuing development since then. The progress has concerned both materials and techniques as well as increased understanding of the lesions. The agents used for embolization have changed from materials like muscle, dura mater, silicon beads, polylene threads, and concentrated alcohol to the now-prevailing polyvinyl alcohol (PVA, Contour; Nycomed Ingenor, Paris, France), mainly for preoperative embolization, and the tissue adhesive N-butyl-2-cyanoacrylate (NBCA, Histacryl Bleu; B. Braun Melsungen Ag, Melsungen, Germany). Improved delivery devices have led to a change from surgical exposure, via balloons with calibrated leaks, to guidewire-guided microcatheters like Tracker (22) (Target Therapeutics, Fremont, CA) and flow-guided catheters like Minitorquer (Nycomed-Ingenor, Paris, France) and Magic (Balt, Montmorency, France) (8). The improved devices have decreased the risk of vascular damage and increased access to the lesion.

Increasingly better results with surgery and stereotactic radiation have been published (6, 16, 17, 28, 30, 34). However, the figures for frequency and level of complications and outcome are often difficult to evaluate. The purpose of this report is to present the experience from one center where embolization is the primary treatment of patients with AVMs coupled with an endeavor to use stricter criteria for evaluation of results.


The study group consisted of 150 consecutively treated patients. Referrals were made mainly made by neurosurgeons, but a few patients were referred by neurologists after consultation with local neurosurgeons. The sex distribution was 44.7% men and 55.3% women. The patients were 5 to 70 years old (35.5 ± 14.8, mean ± standard deviation [SD]) at the beginning of embolization treatment. All patients were treated by embolization within the interval from January 1, 1987 through December 31, 1993. Patients with dural arteriovenous (AV) fistulas were not included. Patients' films and records were discussed at multidisciplinary conferences (representatives from the departments of neuroradiology, neurosurgery, neuroanesthesiology, and neurology) before the patients were accepted for endovascular treatment. Forty-two patients with AVMs who were presented at the conference were not treated. The reasons for the rejections are presented in Table 1.

TABLE 1. Reasons for Not Treating Patients with Arteriovenous Malformations Who Were Presented at the Multidisciplinary Pretreatment Conference
Reason for Not Treating No. of Patients
No reply to treatment offer 7a
Patient declined 9
Recommended for gamma knife treatment 8
Recommended for surgery 1
Complicated anatomy 10
Advanced age 6
Treatment at another hospital 1
Total 42

aSix of the seven patients came from abroad.

Radiological criteria

On an angiogram, AVMs are demonstrated as tangles of tortuous vessels with evidence of shunting, seen as early venous filling. The area of abnormal vessels, where the vascular structures are beyond the capability of current imaging resolution, is called the nidus.

An AV fistula is defined as a direct communication between artery and vein, without an intervening nidus network; it is seen at angiography or fluoroscopy or is considered to be present when a microcatheter (2.2 Fr ~ 0.7 mm) is able to pass through the nidus into a draining vein. An AV fistula can exist alone or within an AVM nidus. For our study, vascular steal was labeled severe when almost no vessels except those involved in the AVMs were seen on an angiogram. Moderate steal meant less than normal contrast filling of nonfeeders.

The size of the AVMs was measured on the films. Feeding arteries and draining veins were not included. Magnification factors were always taken into account; until 1991, they were calculated by measurement of cranial size by computed tomographic (CT) scanning and after 1991 by means of 10-mm washers taped on both sides of the head. The AVMs have been approximated to ellipsoids and their volumes calculated using the formula:

V = d1 x d2 x d3 x 1/2

where d1­3 are the three diameters of the AVMs in centimeters and V is the volume in cubic centimeters (28).

Eloquent area location according to Spetzler-Martin classification was judged from anatomic data obtained mainly from angiography (32). CT and magnetic resonance imaging scans were also used when available. Cortical brain mapping and functional imaging have not been used for classification.

Embolization techniques

Three different techniques of embolization have been used.
1) The balloon with a calibrated leak, as described by Kerber (21), made cyanoacrylate injections possible within reasonably safe distances from the AVMs with flow control. This technique was used mainly from 1987 to 1989 and occasionally thereafter.

2) A microcatheter (Tracker) used over a guidewire or a purely flow-guided microcatheter (Magic) is used for rapid injection of cyanoacrylate without flow control. The Tracker catheter is also used for injection of particles (PVA).

3) A purely flow-guided microcatheter (Magic) in a wedged position (the catheter totally occluding the feeder) makes it possible to inject cyanoacrylate with flow control. In this situation, cyanoacrylate can be injected slowly in large amounts spreading throughout the AVMs (Fig. 1, A­D).

FIGURE 1. Wedged glue injection. Patient with large occipital AVMs previously embolized with NBCA in another country. The AVMs are also supplied from the middle cerebral artery (not shown). Craniotomy for clipping of an aneurysm has been performed earlier. All angiograms are vertebral artery injections in lateral projections; B and C are magnified views. A, initial angiogram showing scattered remains of partially embolized AVMs. B, a microcatheter has been successfully advanced deep into the nidus despite old feeder occlusions from glue (thick arrow indicates the catheter tip; thin arrows indicate the catheter). C, roadmap. The glue cast (black) after injection over 43 seconds with the catheter in a wedged position. The wide, irregular nidus, now filled with NBCA, is clearly seen. The white dot (arrow) indicates the tip of the microcatheter before withdrawal. D, most of the widespread AVMs are now occluded after injection of the glue. Faint filling of branches from the middle cerebral artery via the posterior communicating artery is seen. No deficit arose.

A predecessor to the Magic microcatheter was used from 1987 through 1989. It was a very floppy flow-guided catheter used for embolization in the same manner as the Magic microcatheter but without flow control (29). This catheter was injected by means of a propelling chamber. The Magic microcatheter became available in 1990 and has been widely used since 1991 for most procedures, in the wedged position when possible.

During the first years of the study, isobutyl-2-cyanoacrylate (Bucrylate; Ethicon Inc., Sommerville, NJ) was used as the main embolic material. This agent was withdrawn from the market; we have used NBCA since 1990. The cyanoacrylate concentration was varied from 30 to 100% by mixing with Lipiodol (Guerbet, Aulney-Sous-Bois, France). Tantalum powder (Nycomed Ingenor, Paris, France) was used as contrast agent in the mixture until the end of 1990. PVA has been used less frequently over time and, for the last 3 years, only in the external carotid territory. During the first 2 years, polylene threads or Viñuela cocktail (a mixture of contrast medium, ethanol, Avitene [bovine collagen; Medchem Products Inc., Woburn, MA], and PVA) (34) was used occasionally.

Pharmacological treatment

Since 1993, all patients regarded as being at increased risk of complications because of the sudden hemodynamic changes induced by embolization were prophylactically treated with a special regimen. Patients selected for this regimen were those who had angiographic evidence of large shunts with vascular steal. The regimen included arterial pressure monitoring and blood pressure control with the beta-blocking agent metoprolol (Seloken; Hässle, Göteborg, Sweden); the aim was a mean systemic arterial pressure of 70 to 80 mm Hg. The patients also received the steroid betamethasone (Betapred; Glaxo Operations UK Ltd., Greenford, Middlesex, England) according to a phase-out schedule over 1 week beginning with 8 mg. If signs of ischemic complications were observed, intravenous treatment with the calcium channel antagonist nimodipine (Nimotop; Bayer AG, Leverkusen, Germany) in a dose of 10 to 20 ml/h was initiated.

Strategy of embolization

Our strategy was to primarily close the AVMs totally. If this goal seemed unattainable, we tried to decrease the total volume by systematic embolization from the periphery inward rather than just diminishing the vascular density within the nidus, because the occlusion and complication rate after gamma knife treatment are closely correlated to the size of the irradiated volume.

When an aneurysm was present in the nidus, we tried to close it early in the treatment process. Only during the last year of the study were detachable coils available for the treatment of feeder aneurysms; we did not previously treat aneurysms endovascularly. When a hemorrhage was the presenting manifestation of AVMs with a feeder aneurysm, the symptomatic lesion was treated first (aneurysm clipping). Currently, when a feeder aneurysm is coiled, the shunts are embolized shortly after to decrease hemodynamic stress. We have only occasionally performed dedicated preoperative embolizations.

Definitions of outcome from embolization

Patients undergoing full treatment had AVMs that were eliminated by embolization alone or by supplementary elective or emergency surgery, but we also included patients with embolized AVMs that had been accepted for stereotactic radiation by independent institutions.

Patients undergoing partial treatment had AVMs that after embolization still had AV shunting, but no plans were made for further treatment with embolization, surgery, or radiation because of inaccessibility or size.

Our grading of complications is based on the National Institutes of Health (NIH) stroke scale (3). A severe complication indicates decreased function that has an influence on quality and conduct of life (>5 points on the NIH scale). A moderate complication corresponds to a deterioration of functional status but usually no influence on lifestyle (2­5 points). A slight complication (0­1 point) corresponds to a deficit that has a negligible influence on functional status.

If extravasation was observed in conjunction with a complication during angiography or was seen on a CT scan after a procedural complication, the event was classified as a hemorrhage. Other events were classified as ischemic. Occasionally, small hemorrhages may have been undetected on CT scans because of artifacts from injected glue-contrast mixture.

Radiological and clinical follow-up

For each patient, the case records, angiograms, and available CT scans were reviewed by the participating radiologists. Several different angiographic parameters were recorded as described under Results. Clinical follow-up was performed by an independent neurologist (C.L.) who investigated all files for procedural complications and events occurring <4 weeks after embolization. As the end point of the study, the neurologist performed clinical examinations of all living patients; the results from that study are presented separately (23a). However, the incidence of complications are presented here.

Statistical methods

When testing differences between proportions from unpaired samples for statistical significance, 95% confidence intervals for the difference were calculated (13). Ranges are presented with means ± SD.


One hundred and fifty consecutive patients were embolized and included in the study. In the following account, percentages are given of that number unless otherwise indicated. The presenting symptoms are displayed in Table 2.

Table 2. Presenting Symptoms of Patients Treated with Embolization for Arteriovenous Malformations
Presenting Symptom Percentage (%) No. of Patients
Cerebral hemorrhage 40.0 60
Epilepsy 36.6 55
Headache 8.7 13
Focal symptomsa 12.0 18
Pulse synchronous sound 0.7 1
Hiccup 0.7 1
Hydrocephalus 0.7 1
Accidental 0.7 1
Total 100 150

aWithout signs of cerebral hemorrhage.

The AVMs were located on the left side in 54.0% and on the right side in 42.0%, and they were crossing the midline in 4.0% of patients. The lobar distribution is presented in Table 3. The largest diameters of AVMs are presented in Figure 2 and volumes in Figure 3. Statistical measures for diameters are 10 to 120 mm (42.3 ± 17.6, mean ± SD) and for volumes are 0.5 to 280 cc (25.8 ± 31.1, mean ± SD).

TABLE 3. Anatomic Distribution of Arteriovenous Malformationsa
Lobe Percentage No. of Patients
Frontal 44.7 67
Parietal 20.7 31
Temporal 21.3 32
Occipital 19.3 29
Central 25.3 38
Infratentorial 8.0 12
Cerebellum 4.7 7

aTotals are > 100% (150), indicating that several AVMs engaged more than one lobe. The most common combination was involvement of both sides of the central sulcus.

FIGURE 2. Largest diameters (in millimeters) of embolized AVMs. Descriptive statistics: 10 to 120 mm (42.3 ± 17.6, mean ± SD).

FIGURE 3. Volumes in cubic centimeters of embolized AVMs. Descriptive statistics: 0.5 to 280 cc (25.8 ± 32.1, mean ± SD).

Eloquent areas according to the Spetzler-Martin classification (32) were involved in 72.7%, possibly involved in 6.7%, and not involved in 20.7% of patients. The distribution of Spetzler-Martin gradings is shown in Figure 4. Spetzler-Martin gradings were 1 through 5 (3.3 ± 0.9, mean ± SD).

FIGURE 4. Spetzler-Martin gradings of embolized AVMs. Descriptive statistics: Grades 1 through 5 (3.3 ± 0.9, mean ± SD).


Sixty percent of the AVMs had more than four feeders, 26% had three to four feeders, and 13% had fewer than three feeders. Selective angiograms of the external carotid artery before treatment were available in 83 patients. In 26.5% (22 of 83 patients), feeders were seen from the external carotid artery; in 73.5% (61 of 83 patients), no external carotid feeders were seen.

Perforating arteries and/or vessels to the basal ganglia were involved as feeders in 31.3% of all patients. Feeder aneurysms were found in 16% and nidus aneurysms in 12% of patients. One patient had both a feeder and a nidus aneurysm.

In 20.7% of patients, a direct fistula with a diameter >0.7 mm was seen on the scans or were evident when a microcatheter (2.2 Fr ~ 0.7 mm) went through the nidus into a draining vein. Only one patient had a large single-hole fistula without any surrounding microshunts. In all other patients, the fistulas were part of the nidus in an AVM.

Vascular steal was classified in three groups (see Patients and Methods). Severe steal was observed in 22.7%, moderate steal in 42.0%, and no steal in 35.3% of patients.

Venous drainage was purely central in 10.7%, purely cortical in 46.0%, and a combination of both in 43.3% of patients.

Venous pathological findings searched for included stenoses and ectasias. Stenoses were found in 9.3% and ectasias in 38.7% of patients. Diagnostic angiograms were never aimed at disclosing venous pathological abnormalities.

Embolization technique

The number of treatment sessions per patient is shown in Table 4. The microcatheters used in these procedures were balloons with calibrated leaks (n = 48), Tracker catheters (n = 70), and Magic catheters or, in a few instances, older flow-guided Pursil catheters (n = 86). The total of catheter types used was 204, because different catheter systems were used in several patients.

TABLE 4. Number of Treatment Sessions per Patient
No. of Procedures Percentage (%) No. of Patients
1 43.3 65
2 27.3 41
3 11.3 17
4 10.0 15
5+ (maximum 9) 8.0 12

Cyanoacrylate was used as the only embolic agent in 66.0% of patients (in 23.3% isobutyl-2-cyanoacrylate [before 1990] and in 42.7% NBCA [1990 and after]). It was used in combination with PVA in an additional 24.6% of patients. PVA only was used in 5.3% of patients. In the remaining 4.1%, cyanoacrylate or PVA was used in combination with polylene threads or Viñuela cocktail.

Complications of embolization

Two patients (1.3%) died after hemorrhage in direct connection with embolization. One death resulted from hemorrhage 8 hours after closure of a very large AV fistula. That patient did not have pharmacologically lowered blood pressure, which is currently always used in such instances. The other patient who died had undergone embolization in 9 sessions for a 107 cc AVM; the last two sessions were only 4 days apart. A total of 15 ml of NBCA had been administered by injection. During embolization, the patient developed signs of intracranial hemorrhage. An emergency CT scan showed a hemorrhage, neurosurgeons performed a craniotomy, and an attempt was made to remove the large AVM. The patient did not survive the operation.

Seventeen additional patients had hemorrhages in connection with embolization. Five patients had severe complications, six had moderate complications, three had slight complications, and three had no complications. Complications were also related to ischemic events. In all, severe complications occurred in 10 (6.6%), moderate complications in 23 (15.3%), and slight complications in 26 (17.3%) patients, as judged from the outcome 4 weeks after completed embolizations. All complications occurred ¾4 days; most were directly related to the procedure.

Regarding the reasons for complications in individual patients, we particularly searched for and found hemorrhage (14 patients), vascular damage (dissection) (4 patients), vasospasm (2 patients), untoward embolization of normal brain feeder (7 patients), thromboembolic complication on the arterial side (5 patients), and venous occlusion from glue or thrombus (0 patient). In 21 patients, no particular event was noticed that could be linked to the complication. A small hemorrhage could easily be overlooked on a CT scan because of artifacts from glue after embolization.

In patients with hemorrhages, we failed to pinpoint a probable reason for the hemorrhage, such as venous occlusion. However, during the 1980s, hemorrhages in direct conjunction with treatment were mostly caused by guidewire perforation or by a ruptured vessel from balloon overinflation. Venous compromise after embolization has always been thoroughly sought out. In 12 patients, a microcatheter was glued in, and 6 of these patients experienced complications. It was impossible to verify whether the complication was caused by the glued-in microcatheter.

Predictors for complications

Among all parameters investigated, none were found to influence the level of complication. There were no differences in rates of complications of treatment in relation to size, location, or number of treatment sessions, although there was a tendency toward lower complication rates for AVMs <30 mm in diameter (Tables 5 and 6).

Of particular note is that location of eloquent areas was not accompanied by an increased incidence of complications after embolization. The presence or absence of an aneurysm on a feeder or in the nidus did not influence the incidence of complications, nor did the level of shunting as shown by grade of steal. The presence or absence of deep venous drainage did not affect the outcome, nor did the presence of venous varices.

TABLE 5. Complications after Embolization for Different Sizes of Arteriovenous Malformationsa
<30 mm AVMs 30­60 mm AVMs >60 mm AVMs
Grade of Complication % (No. of Patients) % (No. of Patients) % (No. of Patients)
None 68 (17) 57 (59) 62 (13)
Slight 24 (6) 15 (16) 19 (4)
Moderate 4 (1) 19 (20) 10 (2)
Severe 4 (1) 7 (8) 5 (1)
Death 0 (0) 1 (1) 5 (1)
Total 100 (25) 100 (104) 100 (21)

aAVMs, arteriovenous malformations.

TABLE 6. Number of Embolizations per Level of Complication
Grade of Complication No. of Patients No. of Embolization Sessions Range (mean ± SD)a
None 89 1­7 (2.14 ± 1.37)
Slight 26 1­5 (1.71 ± 1.23)
Moderate 23 1­6 (2.5 ± 1.68)
Severe 10 1­8 (2.4 ± 1.96)
Death 2 1 and 9

aSD, standard deviation.

Predictors for partial treatment of AVMs

The rate of failure to achieve full treatment with embolization increased when thalamostriate vessels or perforating arteries were involved as feeders to an AVM. Only 40.4% of the patients with this criterion were fully treated, compared to 78.6% of patients without this type of feeder (difference, 38.2%; 95% confidence interval, 22­54%). Size also played an important role, with 100% full treatment among AVMs <30 mm, 68% among those 30 to 60 mm, and 19% among AVMs >60 mm in diameter (Table 7). Fully treated AVMs had a volume of 0.5 to 61 cc (14.5 ± 11.3, mean ± SD) as compared to 9 to 280 cc (48.3 ± 45, mean ± SD) for partially treated AVMs.

TABLE 7. Patients in Whom Total Occlusion of Arteriovenous Malformations Has Been Achieved by Embolization Alone or Could Be Expected in Combination with Gamma Knife Treatment or Surgery (Fully Treated) Divided into Size Groups
AVMa Size (mm) No. of Patients % Fully Treated No. Fully Treated
<30 25 100 25
30­60 104 68 71
>60 21 19 4

aAVM, arteriovenous malformation.

Morphological outcome and total management

Size reduction from embolization is presented in Table 8. The results of embolization and their affect on further management of the patients are presented in Table 9. One hundred patients were thus regarded as fully treated (total occlusion from embolization alone or embolization supplemented by stereotactic radiation or surgery), whereas 50 patients were partially treated. The total management of all patients in different size groups is presented in Figure 5, A and B. Numbers and percentages of the fully treated patients in different size groups are presented in Table 7.

TABLE 8. Size Reduction Attained by Embolization
% Reduction Number Percentage
100 20 13.3
75­99 31 20.6
50­74 38 25.3
25­49 23 15.3
<25 38 25.3

TABLE 9. Total Management for Patients after Completed Embolization (n = 150)
Embolization Outcome % of Patients Number
AVMsa totally occluded 13.3 20
To gamma knife treatmentb 44.0 66
Surgery after embolization 9.3 14
Clinical improvement, but not total occlusion; no further treatment performed 8.0 12
Not possible to embolize more, or patient declined further treatment 23.3 35
Procedural mortality 1.3 2
Embolization to continue in home country 0.7 1
Total 100 150

aAVMs, arteriovenous malformations.
bTwo patients (1.3%) were treated with a linear accelerator.

FIGURE 5. Final treatment for all patients with respect to the largest diameter (A) and largest volume (B) of the AVMs. No AVMs >40 mm or 30 cc were occluded by embolization alone. (Example: A, of AVMs with a diameter of 21­30 mm, 30% were occluded by embolization alone and 70% after combined therapy, for a full treatment rate of 100%. B, of AVMs with a volume of 11­20 cc, ~10% were occluded by embolization alone and ~68% after combined therapy, for a full treatment rate of 78%.)

To be able to compare the results with those of patients who undergo only gamma knife treatment, certain groupings are made in Tables 10 and 11, showing the treatment results among the smallest of all lesions.

TABLE 10. Complications for Small Arteriovenous Malformations (<4 cc, n = 14; 4­8 cc, n = 20)a
AVM Size Complications Number Year of Treatment
<4 cc None 10
Slight 2 1991 and 1993, temporary deficits
Moderate 2 1989, hemianopia
1993, temporary deficit
Total 14
4­8 cc None 11
Slight 4 1988, 1989, 1989, 1990
Moderate 2 1989
Severe 3 1987, 1988, 1990
Total 20

aAVM, arteriovenous malformation.

TABLE 11. Final Treatment for Small Arteriovenous Malformations (<4 cc, n = 14; 4­8 cc, n = 20)a
AVM Size Final Treatment Number
<4 cc Embolization 10
Radiation 4
Total 14
4­8 cc Embolization 3
Radiation 15
Operation 2
Total 20

aAVM, arteriovenous malformation.


Total surgical excision of AVMs is regarded as affording full protection against further hemorrhage. The decision to embolize or to treat with stereotactic radiation is based on the thought that total obliteration from these modalities alone or in combination gives the same protection. Nothing is currently known that contradicts this opinion (1, 36). The results of all treatment modalities must be checked by late angiography, i.e., 6 to 12 months after any treatment (15, 19).

The decision of how to treat cerebral AVMs depends heavily on knowledge of the natural history and alternative treatment possibilities. Perusal of newer literature reveals an almost unanimous opinion that the natural history for a conservatively managed patient includes an annual cumulative bleeding rate of 2 to 4%, an annual mortality of ~1%, and a combined mortality and severe morbidity rate of 2.5% (5, 7, 27). Ondra et al. (27) found no difference in the risk of hemorrhage or death, regardless of presenting symptoms, in their well-controlled series. Brown et al. (5) made the same observation and could not detect any difference in bleeding rate that was dependent on size.


Surgical treatment yields good results, as presented by a few centers regarding small AVMs or larger ones in noneloquent areas (16, 17). The outcome for AVMs in Spetzler-Martin Grades 1 through 3 was excellent or good in all instances. Because of different labeling of complications, comparison is not easy. In the series from Heros et al. (17) that included 153 patients, the complication rates were as follows: none, 56%; slight, 20%; moderate, 14%; severe, 10%; and mortality, 0.7%. The series presented by Hamilton and Spetzler (16) showed no or slight complications in 70%, moderate complications in 19%, severe complications in 10.8%, and no operative death. Both series comprised only patients with angiographically confirmed total extirpation of AVMs. From the second series, all patients with partially extirpated AVMs were excluded. In another recent series of 10 patients, comparable to our patients who underwent embolization for AVMs (volume, 20.1 ± 14.2 cc; Spetzler-Martin mean grade, 3.5 ± 0.5; left side, eloquent area), 4 patients had obvious neurological deficits after surgery (2 with hemiparesis, 1 with hemiparesis and aphasia, and 1 with hemianopsia) (14). The focus of the study was on cognitive deficits; several of the so-called normal patients had such deficits, and three were regarded as severely handicapped. One conclusion from the above series was that it is important to evaluate cognitive functions to be able to take proper supportive measures for patients. Another conclusion when comparing these results with other series may be that generally the most skilled surgeons publish their results (31).

Stereotactic radiation

The gamma knife is an efficient tool for obliterating small AVMs, with an occlusion rate of close to 90% for AVMs <1 cc (<13 mm in diameter), 80% for AVMs 1 to 3 cc (<18 in mm diameter), and 50% for AVMs 4 to 8 cc (<25 mm in diameter) (34). The rate of direct complications after treatment with the gamma knife is almost zero. A low, size-dependent incidence of neurological deficits occurs from 4 to 56 months after radiation, depending on radionecrosis. Drawbacks include size limitation and a 2-year waiting time for occlusion and protection against hemorrhage (34). A recent report (20) on 1565 patients who showed a significantly decreased risk of hemorrhage 6 months after gamma knife treatment regardless of obliteration may change this opinion.

Endovascular treatment

Regarding endovascular treatment as a primary modality for treatment of AVMs, the literature is scanty. Most articles present subgroups from the authors' experience. A few articles presenting presumably more complete series (2, 10, 26) report complete obliteration in 7 to 14%, combined treatment with a gamma knife in 20 to 30%, and partial treatment in 54 to 90% of patients. Complications reported are death in 2 to 6% and permanent deficit in 8 to 10%. How and when the clinical evaluation was performed and what levels of remaining deficit were considered significant is not clear.

Our series of patients was selected from the AVM population by choice of the referring neurosurgeon (or, in a few instances, neurologists after consultations with local neurosurgeons), who regarded active treatment as the appropriate option and considered embolization an attractive alternative to surgery or gamma knife treatment. This is illustrated by the fact that very few patients with small AVMs were referred to us. Most of the patients had high Spetzler-Martin grading (84% are Grade greater than or equal to 3; see Fig. 4), indicating increased surgical risk. This surgical selection favored left side and eloquent region location in patients referred for embolization. The large number of feeders and the large percentage of patients with perforating arteries involved as feeders to the AVMs indicated the surgical complexity of the lesions as well as the problems facing the interventionalist. We had problems achieving full treatment for patients with perforating arteries involved as feeders. The distribution of different Spetzler-Martin grades (Fig. 4) reflects the level of surgical difficulty; it is not truly valid for endovascular treatment but for the points given for size. We did not find the other Spetzler-Martin scale variables, such as deep venous drainage and, in particular, eloquent area location, predictive of either subsequent complications or failing treatment. The Spetzler-Martin gradings are included to facilitate comparison with other studies.

We found many aneurysms. This may depend on the angiographic equipment, with improved image quality, and more selective catheterizations. Intranidal aneurysms are regarded as risk factors for subarachnoid hemorrhage but are mostly presented as indicators for previous hemorrhage (12, 25). The significance of this is that if the increased risk of early recurrent hemorrhage indicated by some studies (4, 9, 11, 18) is true, the finding of an intranidal aneurysm calls for early treatment (12). We treated a patient who had had eight subarachnoid hemorrhages and in whom we found an aneurysm in centrally located AVMs. We aimed selectively for the aneurysm and occluded it with glue, leaving the rest of the AVMs untouched. That patient had no further hemorrhages during 3 years of follow-up.

In our current technique of embolization, we use a flow-guided catheter (Magic) for injection of NBCA, as described above. This catheter makes it possible to reach very distally into the intracranial circulation (Fig. 1B). The results from this type of embolization are continuously improving with increased experience. The level of complications and treatment results are not evenly distributed over the period studied, but there is an obvious tendency to improvement (37).

Continuous development with the wedged injection technique (Fig. 1, A­D) seems to further improve results, with a decreased number of complications and increased percentage of fully treated and totally embolized AVMs (preliminary data).

The techniques developed earlier and used during the earlier part of the study are still occasionally used. The balloon with a calibrated leak is at times advantageous for embolizing very large fistulas. The flow control obtained with the balloon makes it possible to have the glue stick to the vessel wall even in instances with extreme flow. The balloon we use is a detachable latex balloon (Balt). The detachability makes it possible to retrieve the catheter even if the balloon is glued.

We always use guidewire-guided catheters in the external carotid circulation, because flow-guided catheters do not function well there. In the internal carotid circulation, this kind of catheter is rarely used. It could be advantageous if too large a difference in flow exists between a main artery and the targeted feeder to allow a flow-guided catheter to enter or if a feeder leaves at an extreme angle.

The complication rate with embolization (mortality, 1.3%; severe complications, 6.7%; moderate complications, 15.3%; slight complications, 17.3%; and no complications in 59.3%) is not worse than results after surgery (16, 17). The mortality rate resulting from embolization was 1.3%; in the case of the two patients who died, circumstances possibly explaining the outcomes were present. The first patient, who had a very large fistula, had a possible parallel to the neurosurgical phenomenon of normal perfusion pressure breakthrough (33). We have since closed several large fistulas but under a regimen with pharmacologically lowered blood pressure; no hemorrhagic complications resulted. The other patient, who had an enormous AVM, had two embolization procedures only a few days apart, with a substantial affect on hemodynamics. We scheduled sessions over a short period with only 3 or 4 days' intermission. Currently, we always allow at least 4 weeks between sessions. Several more hemorrhages occurred; during the 1980s, those in direct conjunction with treatment were mostly caused by guidewire perforation or vessel rupture from balloon inflation. Hemorrhages occurring days later were probably caused by hemodynamic changes with increased feeder pressure after closure of shunts or possibly vascular wall necrosis caused by the embolic agent. Several hemorrhages were not followed by neurological sequelae.

Neurological deficits as complications are more frequent in our series than in other series of embolized patients (2, 10, 26). Several factors account for this. The AVMs were mostly large and complicated, because patients with lesions suitable for radiation or surgery were rarely referred. We probably used a more aggressive approach, resulting in approximately two-thirds of the patients having a reasonable chance of becoming free from AVMs through embolization alone or in combination with gamma knife treatment or surgery. Another factor was the strict grading of complications and the thorough clinical evaluation by a neurologist who recorded even the smallest aberration from the normal. As shown in Tables 10 and 11, the complications occurred among small AVMs concentrated to the earlier years of the study; this tendency is confirmed for most AVMs in a separate study (see Addendum, below) (37).

One finding that raises concern was the number of patients (n = 50) without fully treated AVMs. As described earlier, those patients had large AVMs that were considered impossible to treat actively with other methods. Whether embolization was of any benefit regarding further hemorrhage and death for them must be determined by continuing follow-up.

The morphological outcome and complication rate were tested against location regarding lobes and eloquent areas, size and volume, number and type of feeders, presence of aneurysms, fistulas and different levels of steal, type of venous outflow and venous pathological abnormalities, and number of treatment sessions. The only reliable predictors found were size (Table 7) and the presence of perforators or ganglionic vessels in conjunction with morphological outcome. The rate of no or slight complications was approximately the same in all size groups (Table 5).

Because the occlusion rate after gamma knife treatment is heavily dependent on the volume irradiated (34) (occlusion rate after irradiation related to AVM volume, <1 cc [88%], 1­3 cc [78%], and 4­8 cc [50%] obliteration after 2 yr), and the result of embolization of small AVMs is good (Tables 10 and 11) and continues to improve (see Addendum, below), it seems reasonable to raise the question whether embolization could be an initial treatment even for small AVMs and whether the gamma knife should be used if areas of nonobliterated nidus remain. Any size reduction would be beneficial to gamma knife treatment results.

The combined risk of death and severe complications amounts to 8.0%. This is approximately equal to 3.2 years of natural history mortality/morbidity. Whether this concentration of risk is acceptable may be debated. If the same calculations are made from results from the last 2 years of the study, the figure is far more favorable, 0% mortality and severe morbidity and a higher percentage of fully treated patients (37).


Our continuous work shows an ongoing improvement of results with essentially the same embolization technique and materials, and a steady state regarding direct treatment outcome has not yet been reached. A brief survey of all patients embolized in 1994 and the first 7 months of 1995 has shown further improved results in patients with lesions with approximately the same mean sizes, except that no very large AVMs (>80 cc or >70 mm) are included in the latter series. The AVMs were located in eloquent areas in 80%. Complications to embolization occurred in two instances among the 32 patients who underwent embolization. One complication was severe, and the other was slight. The patient with the severe complication has recovered and is now in a better state than before treatment. The treatment results were as follows: total embolization in 25% (8 of 32 patients) and combined treatment (mostly gamma knife) in 34% (11 of 32 patients), partial treatment in 3% (1 of 32 patients), and embolization to continue in 38% (12 of 32 patients) (work in progress).


Embolization of AVMs is efficient for occlusion of the nidus in small AVMs. In conjunction with gamma knife treatment, the effect is good in larger AVMs as well. The complication rate has been equal to ~3 years of natural history and is not higher than for other treatments. Whether the treatment alters the long-term natural history is still too early to state with certainty, but evidence exists that patients with AVMs totally obliterated by embolization alone or decreased in size making them small enough for supplementary gamma knife therapy will have a stable clinical course so that the long-term outcome for this patient group will be favorable. Large size and arterial supply from perforators and ganglionic vessels were negative predictors for total occlusion. No predictor for complications to treatment was identified.


This work was supported by a grant from The Göteborg Medical Society.

Received, May 24, 1995.
Accepted, March 20, 1996.
Reprint requests: Gunnar Wikholm, M.D., Section of Interventional Neuroradiology, Department of Radiology, Sahlgrenska University Hospital, S-413 14 Göteborg, Sweden.


  1. Berenstein A, Lasjaunias P: Surgical Neuroangiography. Berlin, Springer-Verlag, 1992.
  2. Berthelsen B, Löfgren J, Svendsen P: Embolization of cerebral arteriovenous malformations with bucrylate. Acta Radiol 31:13­21, 1990.
  3. Brott T, Adams HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V, Rorick M, Moomaw CJ, Walker M: Measurement of acute cerebral infarction: A clinical examination scale. Stroke 20:864­870, 1989.
  4. Brown RD Jr, Wiebers DO, Forbes GS: Unruptured intracranial aneurysms and arteriovenous malformations: Frequency of intracranial hemorrhage and relationship of lesions. J Neurosurg 73:859­863, 1990.
  5. Brown RD Jr, Wiebers DO, Forbes G, O'Fallon WM, Piepgras DG, Marsh WR, Maciunas RJ: The natural history of unruptured intracranial arteriovenous malformations. J Neurosurg 68:352­357, 1988.
  6. Colombo F, Pozza F, Chierego G, Casentini L, DeLuca G, Francescon P: Linear accelerator radiosurgery of cerebral arteriovenous malformations: An update. Neurosurgery 34:14­21, 1994.
  7. Crawford PM, West CR, Chadwick DW, Shaw MD: Arteriovenous malformations of the brain: Natural history in unoperated patients. J Neurol Neurosurg Psychiatry 49:1­10, 1986.
  8. Dion JE, Duckwiler GR, Lylyk P, Viñuela F, Bentson J: Progressive Suppleness Pursil Catheter: A new tool for superselective angiography and embolization. AJNR Am J Neuroradiol 10:1068­1070, 1989.
  9. Forster DMC, Steiner L, Håkansson S: Arteriovenous malformations of the brain: A long-term clinical study. J Neurosurg 37:562­570, 1972.
  10. Fournier D, TerBrugge KG, Willinsky R, Lasjaunias P, Montanera W: Endovascular treatment of intracerebral arteriovenous malformations: Experience in 49 cases. J Neurosurg 75:228­233, 1991.
  11. Fults D, Kelly DL Jr: Natural history of arteriovenous malformations of the brain: A clinical study. Neurosurgery 15:658­662, 1984.
  12. Garcia-Monaco R, Rodesch G, Alvarez H, Iizuka Y, Hui F, Lasjaunias P: Pseudoaneurysms within ruptured intracranial arteriovenous malformations: Diagnosis and early endovascular management. AJNR Am J Neuroradiol 14:315­321, 1993.
  13. Gardner MJ, Altman DG: Statistics with Confidence. London, BMJ, 1989, pp 28­33.
  14. Gomez-Tortosa E, Sychra JJ, Martin EM, Arteaga M, Gaviria M, Pavel DG, Dujovny M, Ausman JI: Postoperative cognitive and single photon emission computed tomography assessment of patients with resection of perioperative high-risk arteriovenous malformations. Neurosurgery 36:447­457, 1995.
  15. Guo WY, Karlsson B, Ericson K, Lindquist M: Even the smallest remnant of an AVM constitutes a risk of further bleeding. Acta Neurochir (Wien) 121:212­215, 1993.
  16. Hamilton MG, Spetzler RF: The prospective application of a grading system for arteriovenous malformations. Neurosurgery 34:2­7, 1994.
  17. Heros RC, Korosue K, Diebold PM: Surgical excision of cerebral arteriovenous malformations: Late results. Neurosurgery 26:570­577, 1990.
  18. Itoyama Y, Uemura S, Ushio Y, Kuratsu J, Nonaka N, Wada H, Sano Y, Fukumura A, Yoshida K, Yano T: Natural course of unoperated intracranial arteriovenous malformations: Study of 50 cases. J Neurosurg 71:805­809, 1989.
  19. Kader A, Goodrich JT, Stein BM, Carmel P, Michelsen WJ: Recurrent cerebral AVMs after negative postoperative angiograms. J Neurosurg 82:342A, 1995 (abstr).
  20. Karlsson B, Lindquist C, Kihlström L, Steiner L: Gamma knife surgery for AVM offers partial protection from hemorrhage prior to obliteration. J Neurosurg 82:345A, 1995 (abstr).
  21. Kerber N: Balloon catheter with a calibrated leak. Radiology 120:547­550, 1976.
  22. Kikuchi Y, Strother CM, Boyer M: New catheter for endovascular interventional procedures. Radiology 165:870­871, 1987.
  23. Luessenhop AJ, Spence WT: Artificial embolization of cerebral arteries: Report of use in a case of arteriovenous malformation. JAMA 172:1153­1155, 1960.
  24. Lundqvist C, Wikholm G, Svendsen P: Embolization of cerebral arteriovenous malformations: Part II­­Aspects of complications and late outcome. Neurosurgery 3:460­469, 1996.
  25. Lylyk P, Viñuela F, Vinters HV, Dion J, Bentson J, Duckwiler G, Lin T: Use of a new mixture for embolization of intracranial vascular malformations: Preliminary experimental experience. Neuroradiology 32:304­310, 1990.
  26. Marks MP, Lane B, Steinberg GK, Chang PJ: Hemorrhage in intracerebral arteriovenous malformations: Angiographic determinants. Radiology 176:807­813, 1990.
  27. Merland JJ, Rufenacht D, Laurent A, Guimaraens L: Endovascular treatment with isobutyl cyano acrylate in patients with arteriovenous malformation of the brain: Indications, results and complications. Acta Radiol Suppl 369:621­622, 1986.
  28. Ondra SL, Troupp H, George ED, Schwab K: The natural history of symptomatic arteriovenous malformations of the brain: A 24-year follow-up assessment. J Neurosurg 73:387­391, 1990.
  29. Pasqualin A, Barone G, Cioffi F, Rosta L, Scienza R, Da Pian R: The relevance of anatomic and hemodynamic factors to a classification of cerebral arteriovenous malformations. Neurosurgery 28:370­379, 1991.
  30. Pevsner PH: Micro-balloon catheter for superselective angiography and therapeutic occlusion. Am J Roentgenol 128:225­230, 1977.
  31. Pollock BE, Lunsford LD, Kondziolka D, Maitz A, Flickinger JC: Patient outcome after stereotactic radiosurgery for "operable" arteriovenous malformations. Neurosurgery 35:1­8, 1994.
  32. Sackett DL: Bias in analytic research. J Chronic Dis 32:51­63, 1979.
  33. Spetzler RF, Martin NA: A proposed grading system for arteriovenous malformations. J Neurosurg 65:476­483, 1986.
  34. Spetzler RF, Wilson CB, Weinstein P, Mehdorn M, Townsend J, Telles D: Normal perfusion pressure breakthrough theory. Clin Neurosurg 25:651­672, 1978.
  35. Steiner L, Lindquist C, Steiner M: Radiosurgery, in Symon L (ed): Advances and Technical Standards in Neurosurgery. Vienna, Springer-Verlag, 1992, pp 19­102.
  36. Viñuela F: Endovascular therapy of brain arteriovenous malformations. Semin Intervent Radiol 4:269­280, 1987.
  37. Wikholm G: Occlusion of cerebral arteriovenous malformations with N-butyl cyano-acrylate is permanent. AJNR Am J Neuroradiol 16:479­482, 1995.
  38. Wikholm G, Lundqvist C, Svendsen P: Transarterial embolization of cerebral arteriovenous malformations: Improvement of treatment results with experience. AJNR Am J Neuroradiol 16:1811­1817, 1995.


Wikholm et al. show a desire to contribute to the literature with an objective, well-organized critical analysis of their anatomic, technical, and clinical results in the embolization of difficult brain arteriovenous malformations (AVMs). Their openness and honesty in the evaluation of technical problems and clinical complications that they encountered in the endovascular management of this type of intracranial vascular abnormality is refreshing.

Wikholm et al. report their experience with embolization, stereotactic radiosurgery, and surgery in 150 consecutive patients with difficult AVMs, between 1987 and 1993. The article is well written and organized; it is also very thorough in the reporting of anatomic results and technical and clinical complications related to embolization techniques. The authors have used numerous delivery systems, including balloons with calibrated leaks, flow-guided microcatheters, and microcatheter / microguidewire combinations. Eighty-five percent of AVMs were Spetzler-Martin Grade 3 or higher. Occlusion for embolization alone was achieved in 13% of patients, and combined treatment with stereotactic radiosurgery was the most important part of the treatment strategy. The procedural mortality was 1.3%, and the total incidence of complications related to embolization was 40%, with 6.7% labeled as severe.

The authors did not relate their technical complications to the delivery systems that they used. It is possible that the majority of vascular perforations, dissections, and spasm occurred during the use of balloons with calibrated leaks and microcatheter/microguidewire combinations. They described a significant reduction in technical complications after 1990, and this positive outcome could be related to a change in technique and the more frequent use of flow-guided microcatheters. The authors describe an unusually high percentage (39.3%) of ischemic complications. They do not specify whether the ischemic complications also decreased with the use of flow-guided microcatheters. Were these ischemic complications related to untoward embolization of normal cerebral arteries? Or were they related to progressive retrograde thrombosis of an AVM feeder also supplying eloquent cortex? Or, were they related to the development of brain swelling caused by untoward occlusion of the AVM venous drainage with embolic material? The use of superselective amytal testing in AVM feeders helped us decrease postembolization ischemic complications resulting from delayed thrombosis of the arterial feeder also supplying proximal eloquent cortex. The motor-sensory and visual cortices can be identified by anatomic means, but the language cortex, spread in the perisylvian area, cannot be depicted by anatomic imaging.

It should be emphasized that the authors depicted numerous minor or moderate neurological ischemic complications after thorough neurological examinations performed by an unbiased neurologist. The majority of these immediate complications resolved in long-term clinical follow-up studies.

We agree with the authors that it is very difficult to clearly identify the source of postembolization hemorrhagic complications. The perforation of an arterial feeder used to be seen as the most common cause of periembolization hemorrhage, but the incidence has dramatically decreased with the use of flow-guided catheters. In our present experience, postembolization untoward occlusion of an AVM venous drainage in a partially embolized nidus is the most common cause of postembolization hemorrhagic complications. This is particularly true in large AVMs with multiple arterial feeders or in AVMs associated with high-flow arteriovenous fistulas. On occasion, we have identified rupture of a previously undetected intranidal aneurysm as the cause of postembolization hemorrhagic complication. Intranidal aneurysms in perforators such as thalamoperforators, or in lenticulostriate arteries and sometimes in anterior choroidal arteries, may be difficult to embolize because of the technical difficulties encountered in the catheterization of these arteries.

Seventy-five percent of embolized AVMs between 4 and 8 cc in volume (n = 20) were then treated with stereotactic radiosurgery, and 10% of patients were operated on after embolization. The authors highlight the point that they search not only for a decrease in angiographic density of the AVM nidus but also for a true diminution in diameter and volume of the AVM. They embolize from the periphery to the center of the AVM. Is it possible that this methodology increases the ischemic complications? The periphery of an AVM nidus is usually supplied by smaller arteries and arteries "en passage" that have "less sump effect" than centrally located feeders.

Fernando Viñuela
Interventional Neuroradiologist
Los Angeles, California

Wikholm et al. present a critical review of their experience with 150 patients with cerebral AVMs. This two-part study is an ambitious retrospective analysis of the radiological and clinical data in embolization of relatively high-grade (Grades 3­5) AVMs with liquid adhesive and polyvinyl alcohol. It reflects the world standard in regard to the technique of embolization and includes both surgical and radiosurgical treatment as well as AVMs treated with embolization alone. This careful analysis of the cases from a single center provides important information regarding the efficacy of embolization in preventing hemorrhage from AVMs and further defines complication rates for embolization of high-grade AVMs.

One hundred and fifty patients met the inclusion criteria for the study; 84% were Grade >3. Forty-four percent were referred for subsequent radiosurgery. The total procedural morbidity was close to 40%, although most of the morbidity was minor with a 1.3% overall procedural mortality. However, over the course of the study period, six patients died from AVM hemorrhage. One would expect five deaths for a cohort of 150 patients followed for 6 years, but if more than 5 patients died during the study period, it would imply that the natural history of this cohort is worsened by treatment or by the selection bias of inclusion criteria (i.e., high grade or lesion angioarchitecture).

The division of the analysis into "completely treated" and "partially treated" lesions provides important data regarding the effect of embolization on the hemorrhage rate of AVMs. Patients considered "fully treated" (100 patients) included patients with complete obliteration of the nidus by embolization alone (10%) as well as those lesions reduced to a size suitable for radiosurgery. Three postprocedural hemorrhages occurred in this group while the patients were awaiting further therapy, which implies that there was no protective effect of embolization on hemorrhage rate. Of greater concern is the group of patients with "partially treated" lesions in which nine hemorrhages after treatment occurred in the follow-up period of 4.2 years, which exceeds the expected natural history. This observation raises the serious question of whether these patients should be treated by embolization alone. Smaller AVMs are not more benign.

Scott C. Standard
L. Nelson Hopkins
Buffalo, New York

Wikholm, Lundqvist, and Svendsen are well-known individuals in the field of interventional neuroradiology and endovascular neurosurgery. I have followed their work for over 20 years. The senior author, Svendsen, has always been a highly reliable and honest individual who follows his patients carefully. Therefore, this information is of great assistance in understanding the role of embolization in the overall management of cerebral arteriovenous malformations. For the most part, the results, degree of occlusion, type of complications, and outcome are similar to others as reported. The authors' excellent ability to follow patients gives us some insight as to the efficacy of this treatment. The authors' persistence and honesty, as well as their ability to have the material reviewed by independent neurologists, helps us all in the treatment of AVMs. Important concepts are mentioned in this article that are compatible with those reported in our series (1). We also concur with the authors' analysis of the angioarchitecture and correlation with hemorrhage. Patients in whom they were able to close the intranidal, or pedicle, aneurysm were protected against hemorrhage. However, seven patients in whom they did not achieve this goal developed hemorrhages; two of the patients died. This again is compatible with our experience. Also compatible with our experience are those patients in whom hemorrhage occurred after embolization. The outcome in this group, which may be caused by a very select group of patients with very difficult malformations, seems to be worse than the outcome after spontaneous hemorrhage. We also concur with the authors that the location of the malformation is less critical from an endovascular point of view than it is from a surgical point of view. Endovascular treatment aims to close the abnormality and preserve all normal arteries in the vicinity, regardless of location in an "eloquent" or "noneloquent" area. Therefore, location as a risk factor seems to be less critical for endovascular embolization. We also concur with the authors that the preliminary angiogram, showing the size of the malformation and presence of perforators as part of the supply to the malformation, will predict success for endovascular treatment.

The authors seem to have a higher complication rate than the rates reported in other series, but it is probably related to their more honest and more thorough analysis and neurological examination. The complications of all interventional neuroradiological procedures in large series such as this one may get lost in view of the fact that these series are comprehensive and are conducted over a long period of time (7 yr). In the last several years, the use of nonballoon-tipped catheters and flow-guided catheters has markedly decreased the complication rate because vessel perforation and rupture are related to trauma from balloons or guidewires. On the other hand, microcatheter gluing has also decreased because of the decreased concentration of NBCA in the mixture. Therefore, longer injections are now possible and bonding of the glue to the microcatheter and vessel wall has also been markedly decreased. This group of investigators also follows several of our concepts; anatomic guidelines are the most important predictors of potential hazards and neurological complications, reaching the nidus and preserving all normal vessels decreases the risk, and sodium amytal was not used as a "provocative" test and the outcome was very good. Therefore, the use of sodium amytal as an excuse for poor technique, lack of superselectivity, or lack of knowledge of anatomy is clearly established in this group of patients. We also concur that the Spetzler-Martin classification is a surgical description of surgical skills and does not correlate with endovascular treatment. We do concur with the authors that we use this classification for the sake of comparison, because it is frequently requested from interventional neuroradiologists. Again, we do concur that this does not correlate with the difficulty of managing patients from an endovascular point of view. Furthermore, no matter how effective the embolization, we can only try to affect the "size" of the remaining malformations. However, no interventional procedure can change the location of the malformations, venous drainage, or, in reality, size. Therefore, the Spetzler-Martin classification is difficult to change with endovascular procedures. We also concur with the statement that in those patients in whom cure is not possible by either embolization or a combination of techniques, the aim of the embolization is a targeted approach in an attempt to close the "weakness" of the angioarchitecture of the malformation, such as intranidal aneurysms or aneurysms in the pedicle, or to reduce the pressure in the veins that have an outflow restriction (1). We agree with the technique of embolization and, for the most part, use the same approach. Based on the experience of the authors as well as ours, embolization may play a role even in small malformations that are suitable for radiosurgery to reduce the weakness of the angioarchitecture and hopefully afford protection during the latency period of the radiosurgery and to further reduce the volume (the smaller the volume, the more effective the radiosurgery).

Overall, this article is very important for the understanding of the long-term efficacy of this intervention. For the authors' benefit, the techniques available currently for intranidal embolization (Moret J, personal communication) are far superior to those available at the beginning of the study; the future outcome should be further improved because of these technical advancements. The authors also clarify the long-standing controversy concerning the use of liquid adhesive versus particles, confirming our own experience that particles are less effective and carry similar risks.

Alex Berenstein
Interventional Neuroradiologist
New York, New York

  1. Berenstein A, Lasjaunias P: Classification of brain arteriovenous malformations: Angioarchitecture, in Surgical Neuroangiography. Heidelberg, Springer-Verlag, 1993, vol 4.

Please submit your thoughts or comments to William Chandler, Internet Moderator at

Return to Table of Contents

Neurosurgery Home Page Return to the Neurosurgery Home Page

Return to the Waverly Home Page Search the Waverly Catalog

© 1996 Williams & Wilkins