Risk of Endovascular Treatment of Brain Arteriovenous Malformations
Background and Purpose— Independently assessed data on frequency, severity, and determinants of neurological deficits after endovascular treatment of brain arteriovenous malformations (AVMs) are scarce.
Methods— From the prospective Columbia AVM Study Project, 233 consecutive patients with brain AVM receiving ≥1 endovascular treatments were analyzed. Neurological impairment was assessed by a neurologist using the Rankin Scale before and after completed endovascular therapy. Multivariate logistic regression models were used to identify demographic, clinical, and morphological predictors of treatment-related neurological deficits. The analysis included the components used in the Spetzler-Martin risk score for AVM surgery (AVM size, venous drainage pattern, and eloquence of AVM location).
Results— The 233 patients were treated with 545 endovascular procedures. Mean follow-up time was 9.6 months (SD, 18.1 months). Two hundred patients (86%) experienced no change in neurological status after treatment, and 33 patients (14%) showed treatment-related neurological deficits. Of the latter, 5 (2%) had persistent disabling deficits (Rankin score >2), and 2 (1%) died. Increasing patient age [odds ratio (OR), 1.04; 95% confidence interval (CI), 1.01 to 1.08], number of embolizations (OR, 1.41; 95% CI, 1.16 to 1.70), and absence of a pretreatment neurological deficit (OR, 4.55; 95% CI, 1.03 to 20.0) were associated with new neurological deficits. None of the morphological AVM characteristics tested predicted treatment complications.
Conclusions— From independent neurological assessment and prospective data collection, our findings suggest a low rate of disabling treatment complications in this center for endovascular brain AVM treatment. Risk predictors for endovascular treatment differ from those for AVM surgery.
Treatment decisions for patients with brain arteriovenous malformations (AVMs) are based on natural-course risk estimates weighed against outcome data from invasive intervention. At present, both remain incompletely defined.1–4⇓⇓⇓
Endovascular treatment of brain AVMs aims to obliterate or reduce the size of the malformation. It is frequently followed by surgical AVM removal or radiotherapy.5–7⇓⇓ As with AVM surgery, determinants of complications from endovascular treatment are not well described.8–11⇓⇓⇓ Originally designed to predict outcome from AVM surgery, the Spetzler-Martin12 grading system is often used to categorize brain AVMs before endovascular therapy as well, but its applicability to this treatment modality remains unsettled.
Studies on AVM embolization outcome using independent neurological data collection are scarce. The purpose of this single-center study was to prospectively and independently assess treatment outcome after embolization therapy of brain AVMs with modern embolization techniques and to analyze determinants of treatment-related neurological deficits.
Patients and Methods
The New York (NY) AVM Databank is an ongoing, prospective database collecting demographic, clinical, morphological, and treatment data on consecutive patients admitted to the New York Presbyterian Hospital with brain AVM proven by brain imaging and conventional cerebral angiography. Other types of intracranial fistulas (eg, dural AV fistulas, vein of Galen–type malformations) are not included. Study patients are drawn from the New York metropolitan area and from distant referral sites. Further details on the design and methods of the data set have been described in prior publications.3,5,13⇓⇓
As a general rule, the study patients were treated with the goal of complete AVM obliteration or removal by endovascular occlusion alone or in combination with subsequent surgery or radiosurgery. The 233 consecutive patients with brain AVMs receiving ≥1 endovascular treatments between 1991 and 1998 were analyzed. Subsequent treatment was surgical in 130 patients (56%) and radiosurgical in 23 patients (10%). Neurological deficit, disability, and impairment were independently assessed by a study neurologist. Functional neurological status was documented before embolization and after each endovascular treatment session until completion of endovascular therapy by use of the modified Rankin Scale.14,15⇓ Any worsening of the patient’s preembolization Rankin score was coded as a new neurological deficit. This included worsening of patients with a preexisting neurological deficit. New neurological deficits were classified as disabling when Rankin scores were ≥3. For patients with baseline Rankin scores of ≥2, any score increase was classified as disabling. Superselective microcatheter cannulation of arteries feeding the AVM and injection of N-butyl cyanoacrylate were used to occlude the fistulae. The technique remained the same throughout the study period. One hundred patients were embolized with 1 session each; 133 patients were treated multiple times (mean, 2.4 procedures; SD, 2.8 procedures; range, 1 to 11 procedures).
Univariate statistics (χ2 or, when appropriate, Fisher’s exact test, t test) and forward stepwise multiple logistic regression procedures were used to describe the effect of demographic factors, mode of initial AVM presentation, preembolization neurological status, and morphological parameters on embolization outcome.
To test the validity of the Spetzler-Martin grading system, the univariate and multivariate (backward stepwise multiple logistic regression procedures) analyses included the total score from the grading system and its individual components: AVM size (score of 1=small size, with a maximum diameter of <3 cm; 2=medium size, with a diameter of 3 to 6 cm; 3=large size, with a diameter of >6 cm), venous drainage (score of 1=AVM with any drainage into the internal, “deep” cerebral venous system), and location of the AVM (score of 1 for AVM in functionally important, so-called “eloquent” brain regions: the sensorimotor, language, and visual cortex; hypothalamus and thalamus; internal capsule; brainstem; cerebellar peduncles; and deep cerebellar nuclei).
Demographic, clinical, and morphological information on the study sample is given in Table 1. By baseline Rankin score, 89% of the patients had no functionally relevant neurological deficit (Rankin score=0 or 1), and 94% were nondisabled (Rankin score ≤2; Table 2). A total of 33 patients [14%; 95% confidence interval (CI), 9 to 19] showed treatment-related new neurological deficits, including 5 with disabling deficits (2%; 95% CI, 0 to 4) and 2 (1%; 95% CI, 0 to 2) who died.
By Rankin scale, 200 patients (86%; 95% CI, 81 to 91) experienced no change in neurological status after treatment, and 92% were nondisabled (Table 2). Of the 200 patients who experienced no change in functional status, 41 (18%) had a preexisting neurological deficit that remained unchanged after endovascular therapy.
After the embolization session that caused a new neurological deficit, 19 of 33 patients (58%) later underwent surgery, 9 (27%) continued with the embolization therapy, 3 (9%) had achieved complete AVM obliteration at the time of the complication, and 2 (6%) died without receiving further invasive treatment.
The 2 patients who died suffered parenchymal hemorrhage, both with ventricular extension. The 5 patients with disabling complications had treatment-induced ischemic strokes (2 middle cerebral artery, 1 anterior choroidal artery, 1 anterior inferior cerebellar artery, and 1 superior cerebellar artery infarct). Clinical information on these patients is provided in Table 3.
Patients with a treatment-related new neurological deficit had significantly more embolizations and were more frequently neurologically normal at baseline (Table 4).
In the multivariate analyses, increasing age, number of embolizations, and normal neurological status at baseline were significantly associated with treatment complications (Table 5). For all other variables shown in Table 4, no significant association was found. Neither the overall score of the Spetzler-Martin grading system (P=0.163) nor its 3 components tested separately predicted treatment outcome (Table 6).
Restricting the univariate analysis to disabling neurological deficits or death, we found significant associations for increasing age [P=0.021; odds ratio (OR), 5.59; 95% CI, 1.11 to 28.14], initial presentation with hemorrhage (P=0.017; OR, 9.59; 95% CI, 1.14 to 80.98), small AVM size (P=0.005; OR, 5.30; 95% CI, 1.13 to 24.95), and presence of deep arterial feeders (P=0.021; OR, 6.60; 95% CI, 1.42 to 30.68). Of the 7 patients (43%) with disabling complications or fatal outcome, 3 had small AVMs (maximum diameter <30 mm according to Spetzler-Martin criteria). Of these, 1 was located in the cerebellar hemisphere, another in the brainstem, and the last in the basal ganglia. The low number of end points precluded meaningful multivariate models for disabling deficits or death.
Additional partial correlation analyses showed that younger patients carried larger AVMs and were treated more often than older patients.
Outcome after AVM embolization therapy has improved over the last decade. The rate of treatment-related permanent neurological deficits and death culled from the literature ranges from 4% to 18%,5 depending on the material used to obliterate the malformation, the means of delivering the occlusive agent, and patient selection.9,10,16–19⇓⇓⇓⇓⇓ Our data suggest a further reduction in treatment-related disabling morbidity and mortality. However, despite the low number of disabling complications, the overall rate of neurological deficits was 14%. Prospective and independent evaluation by a neurologist may reveal a larger number of less severe complications, as has been shown in carotid surgery outcome studies.20,21⇓ Consequently, the detection rate of mild neurological deficits may have been higher in our study than in others, but the low rate of disabling complications is accredited.
The 2 fatal complications in our series were brain hemorrhages. There was no intranidal aneurysm in either patient, and the flow-related aneurysm found in 1 patient was not considered a likely source of the hemorrhage. Also, no venous occlusion that may have led to outflow obstruction with subsequent hemorrhage was noted during or after the embolization. Autopsy was not performed in either patient; therefore, the cause of hemorrhage remains undetermined.
Some authors have suggested that most complications from embolization occur on the venous side of the AVM,6 but in our series, all nonhemorrhagic disabling events were due to arterial occlusions.
In our findings, the number of embolizations and a normal neurological status before treatment were significant predictors of treatment-related neurological deficits. The increasing risk with additional embolizations is not surprising, given the greater exposure to potential complications with each treatment. The complication risk from embolization therefore is likely to be higher in patients whose treatment plan includes multiple embolization sessions.
Normal neurological examination at baseline as a predictor of treatment-related deficits may reflect the higher likelihood of patients with no preexisting focal neurological signs to reveal new symptoms, especially a better detection rate for mild deficits. A similar effect has been observed in surgical AVM treatment,8 lending support to our observation in this study.
The reason for the higher treatment risk in older patients despite their smaller AVMs and less frequent embolizations than younger patients is less obvious. General vascular disease may be more prevalent in older patients, although none of the patients treated showed evidence of atherosclerosis. Moreover, the effect of age was marginal (OR, 1.04; lower limit of the CI, 1.01); the mean age difference between the patients who developed new deficits and those who remained unchanged was small (40±14 versus 36±13 years); and changes in hemostasis22 or nutritional problems encountered in surgery on elderly patients23 did not apply to our patients.
The low rate of disabling complications or death in our study precludes a meaningful multivariate analysis of their predictors. Trends from our univariate analyses suggest clinical and morphological predictors of poor embolization outcome that were also identified in natural-course risk estimates, ie, hemorrhage at initial presentation and small AVM size.3 The association of deep arterial feeders with the failure to achieve complete obliteration of the malformation was previously suggested.24 However, the CIs of our relative risk analyses were large, and studies on more patients are needed to better define predictors of complications.
In our analysis, neither the total score of the Spetzler-Martin grading system nor any of its 3 components predicted treatment complications. The system was originally designed and validated to predict surgical treatment outcome.12,25⇓ Although found helpful by many authors,26–28⇓⇓ our results suggest that morphological variables other than those in the grading system need to be defined to predict outcome from endovascular therapy.
It remains unproven whether endovascular AVM treatment affects the complication risk from subsequent surgery or radiosurgery. Ideally, embolization facilitates other modes of treatment, and a prior analysis showed no deleterious effect of endovascular therapy on AVM surgery outcome.8 In staged AVM treatment, complications from all treatment forms need to be considered in determinations of the total complication rate for complete AVM removal.
This work was supported in part by NIH grant RO1 NS 34949 (principal investigator, W.L. Young, MD).
- Received January 7, 2002.
- Revision received April 1, 2002.
- Accepted April 8, 2002.
- ↵Duong DH, Young WL, Vang MC, Sciacca RR, Mast H, Koennecke HC, Hartmann A, Joshi S, Mohr JP, Pile-Spellman J. Feeding artery pressure and venous drainage pattern are primary determinants of hemorrhage from cerebral arteriovenous malformations. Stroke. 1998; 29: 1167–1176.
- ↵Berenstein A, Lasjaunias P. Endovascular treatment of cerebral lesions.In: Berenstein A, Lasjaunias P, eds. Surgical Neuroangiography. New York, NY: Springer Verlag; 1992.
- ↵Hartmann A, Stapf C, Hofmeister C, Mohr JP, Sciacca RR, Stein BM, Faulstich A, Mast H. Determinants of neurological outcome after surgery for brain arteriovenous malformation. Stroke. 2000; 31: 2361–2364.
- ↵Hartmann A, Mast H, Mohr JP, Koennecke HC, Osipov A, Pile-Spellman J, Duong DH, Young WL. Morbidity of intracranial hemorrhage in patients with cerebral arteriovenous malformation. Stroke. 1998; 29: 931–934.
- ↵van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988; 19: 604–607.
- ↵de Haan R, Limburg M, Bossuyt P, van der MJ, Aaronson N. The clinical meaning of Rankin “handicap” grades after stroke. Stroke. 1995; 26: 2027–2030.
- ↵Purdy PD, Samson D, Batjer HH, Risser RC. Preoperative embolization of cerebral arteriovenous malformations with polyvinyl alcohol particles: experience in 51 adults. AJNR Am J Neuroradiol. 1990; 11: 501–510.
- ↵Wikholm G, Lundqvist C, Svendsen P. Transarterial embolization of cerebral arteriovenous malformations: improvement of results with experience. AJNR Am J Neuroradiol. 1995; 16: 1811–1817.
- ↵Rothwell PM, Slattery J, Warlow CP. A systematic review of the risks of stroke and death due to endarterectomy for symptomatic carotid stenosis. Stroke. 1996; 27: 260–265.
- ↵Boldt J, Huttner I, Suttner S, Kumle B, Piper SN, Berchthold G. Changes of haemostasis in patients undergoing major abdominal surgery: is there a difference between elderly and younger patients? Br J Anaesth. 2001; 87: 435–440.
- ↵Wikholm G, Lundqvist C, Svendsen P. Embolization of cerebral arteriovenous malformations, part I: technique, morphology, and complications. Neurosurgery. 1996; 39: 448–457;discussion 457–459.
- ↵Hamilton MG, Spetzler RF. The prospective application of a grading system for arteriovenous malformations. Neurosurgery. 1994; 34: 2–6;discussion 6–7.
- ↵Steinberg GK, Lane B, Marks MP. Endovascular therapy for intracranial vascular lesions. West J Med. 1990; 153: 542–543.