Determinants of Neurological Outcome After Surgery for Brain Arteriovenous Malformation
Background and Purpose—We sought to define determinants of neurological deficit after surgery for brain arteriovenous malformation (AVM).
Methods—One hundred twenty-four prospective patients (48% women, mean age 33 years) underwent microsurgical brain AVM resection. Patients were examined by 3 study neurologists immediately before surgery, postoperatively in-hospital, by in-person long-term follow-up, and with a structured telephone follow-up. They were classified according to the 5-point Spetzler-Martin grading system, with its 3 elements: size, venous drainage pattern, and location. The functional neurological status was classified with the modified Rankin scale. Multivariate logistic regression models were applied to test the effect of patient age, gender, and the 3 Spetzler-Martin elements on early and long-term postoperative neurological complications.
Results—Twelve patients (10%) were classified as Spetzler-Martin grade 1; 36 (29%) as grade 2; 47 (38%) as grade 3; 26 (21%) as grade 4; and 3 (2%) as grade 5. Postoperatively, in-hospital, 51 patients (41%) showed new neurological deficits (15% disabling [ie, Rankin scale score >2] and 26% nondisabling [ie, Rankin 1 or 2]). At long-term follow-up (mean follow-up time 12 months), 47 patients (38%) revealed surgery-related neurological deficits (6% disabling; 32% nondisabling). The rate of neurological complications increased by Spetzler-Martin grade. Female gender, AVM size, and deep venous drainage were significantly associated with neurological deficits at in-hospital and long-term evaluation. For patient age and AVM location, no significant association was found.
Conclusions—The findings suggest that female gender, AVM size, and AVM drainage into the deep venous system may be determinants of neurological deficit after microsurgical AVM resection.
Therapeutic decisions for brain arteriovenous malformations (AVMs) are commonly based on natural course risk estimates weighed against outcome data from invasive intervention. At present, both natural course and treatment risks remain incompletely defined.1 For the latter, the Spetzler-Martin grading system2 3 is widely used as an approximation of the risk of surgical treatment. We tested the differential effect of patient age, gender, and the 3 Spetzler-Martin score elements (AVM size, venous drainage, and brain location) on the rate of postoperative neurological deficits in the data set of the New York AVM Study project.
Subjects and Methods
The 124 brain AVM patients, enrolled prospectively between 1990 and 1998 in the Columbia–Presbyterian Medical Center AVM Databank,4 5 underwent microsurgical AVM resection. Baseline characteristics of the study sample are given in Table 1⇓.
Patients were classified according to the 5-point Spetzler-Martin grading system with its 3 elements of size (scored 1, small size, for maximum diameter of <3 cm; 2, medium size, for diameter 3 to 6 cm; and 3, large size, for diameter >6 cm), drainage (scored 1 for AVM with any drainage into the internal, “deep” cerebral venous system), and location (scored 1 for AVM in functionally important, so-called “eloquent,” brain regions: “the sensorimotor, language, and visual cortex; the hypothalamus and thalamus; the internal capsule; the brain stem; the cerebellar peduncles; and the deep cerebellar nuclei”2 ). For example, grade 1 AVMs are small, superficially draining, and located in noneloquent regions of the cortex, whereas grade 5 AVMs are large, in part or completely draining into the internal venous system, and situated in functionally important areas of the brain.
The functional neurological status was classified by 3 study neurologists (H.M., A.H., C.S.) using the 7-step modified Rankin scale (Table 2⇓). Any worsening of the patients’ preoperative Rankin scale score was coded as “new neurological deficit.” New neurological deficits were classified as “disabling” when Rankin scores were 3, 4, or 5 (for patients with preoperative Rankin scores of 2 or worse, any score increase was classified as disabling). All patients were examined immediately before surgery and postoperatively in hospital. After discharge, long-term follow-up was completed in person (n=82, 66%) and with a structured telephone follow-up (n=42, 34%). Two patients were lost to long-term follow-up. In the statistical analysis of long-term outcome, they were defaulted to their postoperative in-hospital status.
Multivariate logistic regression models were used to test the effect of patient age, gender, AVM size, drainage, and location on post-operative neurological complications early postoperatively and at long-term follow-up. The association of the Spetzler-Martin grading system with postoperative deficits was tested by using χ2 statistics. Statistical analysis was completed with SPSS for Windows, version 8.0 (SPSS, Inc).
Endovascular treatment before surgery—aimed at elimination of deep-feeding arteries, occlusion of feeding vessels not readily accessible surgically, and reduction of AVM size—was performed by superselective application of n-butyl cyanoacrylate in 79 patients (64%).
Postoperatively (in-hospital), 41% of the patients had new neurological deficits (15% disabling and 26% nondisabling). At long-term follow-up (mean follow-up time 12 [SD 14] months), surgery-related neurological deficits were found in 38% (6% disabling, 32% nondisabling; (Table 3⇓). No patient died. Two patients were lost to long-term follow-up. They showed severely disabling new neurological deficits (Rankin score of 5) at postoperative (in-hospital) evaluation. Both were defaulted to a Rankin score of 5 in the long-term follow-up analyses.
Of the 47 patients with persistent postoperative deficits at long-term evaluation, 43 had worsened from a preoperative Rankin score of zero and 4 had worsened from a score of 1. Twenty-two had a standardized telephone follow-up but no in-person evaluation. Rankin scores at baseline, postoperatively, and at long-term follow-up are shown in Table 4⇓. The clinical status of the patients with disabling permanent post-perative deficits is given in Table 5⇓.
The rate of all (disabling and nondisabling combined) neurological complications increased significantly by Spetzler-Martin grade (P<0.01, Table 6⇓). A similar effect was not shown when the analysis was restricted to disabling deficits (Table 6⇓).
There was no significant difference (P=0.26, Fisher’s exact test) in favorable outcome (Rankin 0 to 2) between patients who had preoperative AVM hemorrhage (93%) and those who presented with other symptoms (84%). The frequencies of complications by AVM size, drainage pattern, and AVM location are shown in Table 7⇓. The multivariate logistic regression analyses revealed that AVM size and deep venous drainage were significantly associated with new neurological deficits post-operatively in-hospital as well as long-term evaluation. For AVM location, no significant effect was found (Table 8⇓).
Adding patient age and gender to the above model, female gender was found to be significantly associated with complications (OR 2.5, 95% CI 1.1 to 5.9 at postoperative evaluation; OR 3.2, CI 1.4 to 7.5 at long-term evaluation). For patient age, no significant effect was found. AVM size and deep drainage remained significant.
Because of the low event rate, the separate effect on the clinically more important disabling complications could not be tested in a multivariate model. Univariate testing, however, showed no significant association of disabling deficits with AVM size, deep drainage, or eloquent location (Table 7⇑).
In an additional univariate analysis, there was no significant effect of preoperative endovascular treatment on early postoperative (P=0.18) or long-term (P=0.43) outcome. No patient experienced complications from endovascular therapy before undergoing AVM surgery.
Our finding of a 6% frequency of disabling persistent neurological deficits appears to confirm the generally reported low complication rate of AVM microsurgery. Prior reports from pooled data analysis have shown a complication rate (of disabling deficits) of 8% (95% CI 4% to 12%).6 Continuous improvement of microsurgical techniques and an ameliorating effect of staged embolization before surgery (the latter proposed in previous studies7 ) may explain the trend toward fewer disabling complications suggested by our data. Given the small number of disabling complications and the confounding effect of different morphological characteristics of AVM that can be embolized compared with those that render themselves to surgery without prior embolization, the differential effect of presurgical endovascular treatment could not be tested with sufficient statistical power in our study.
Despite the low number of severe surgical complications, the rate of all postoperative neurological deficits was higher than could be expected from the data culled from the literature.7 Independent neurological evaluation and prospective data acquisition may reveal a larger number of complications, as has been shown in vascular surgery outcome studies unrelated to AVM.8 9 Recent work has found a morbidity of natural-course AVM hemorrhage lower than that from intracranial bleeding from other causes.10 Furthermore, our outcome measure, the Rankin scale, with its low sensitivity to higher cortical function lesions,11 may underestimate important complications. In this light, treatment complications rated minor/nondisabling may still bear considerable weight in the risk balancing of AVM treatment decisions.
In our study, the Spetzler-Martin grading system was a significant predictor of neurological complication, but our findings could not confirm the low complication rates in grades 1 to 3 proposed in the work on which the scale was based or the continuous progression in the rate of complications from grades 1 through 5 that other authors12 13 have found. The grading system’s failure to predict clinically more relevant disabling complications may be due to the overall small number of such complications in our series. As in Spetzler’s and Martin’s own analysis, the grading system’s individual elements show different effect magnitudes: increasing AVM size appeared to be a stronger determinant than deep venous drainage, and eloquence of location—although significant at immediate postoperative examination—had no significant effect at long-term follow-up or in multivariate logistic regression analyses. Other studies14 have shown similar results. Further weakening the predictive value of the grading system, the 3 elements are not independent: with increasing AVM size, an eloquent brain location and a deep venous drainage component must become more frequent. Finally, the somewhat arbitrary definition of “eloquence” has been shown to result in highly variable coding when different AVM centers were compared,15 thereby suggesting a low interrater reliability of this criterion.
The association of female gender with neurological complications after AVM surgery remains unexplained. Reasons given for the influence of gender on non-AVM surgical outcome,16 such as delayed diagnosis and referral, increased comorbidity, and the effect of smaller arteries in women, are not readily transferable to AVM surgery. Interestingly, for women with brain AVM, a lower natural course risk of hemorrhage than in men has previously been described.4 Despite a currently missing explanation of these findings, our results may warrant consideration in treatment decisions.
Although independent associations between some of the tested variables and postoperative neurological deficits were shown, those found to be determinants were far from predicting all complications. Clinicians facing therapeutic decisions in AVM patients, including questions regarding treatment risks, are currently left with the rather simple notion that larger AVM may bear a higher risk of surgery than small ones. Important morphological characteristics, such as type and number of arterial AVM feeders and draining veins; the presence of arterial, nidal, or venous aneurysms; cortico-subcortical wedge-shaped morphology versus deep white matter location; and supratentorial versus infratentorial location—to name a few—have so far not been rigorously tested. Their impact on surgery outcome, however, appears plausible.
Supported in part by NIH grants RO1 NS 34949 and K24 NS 02091 (to Dr Young).
- Received March 20, 2000.
- Revision received June 30, 2000.
- Accepted June 20, 2000.
- Copyright © 2000 by American Heart Association
Spetzler RF, Martin NA. A proposed grading system for arteriovenous malformations. Neurosurgery. 1986;65:476–483.
Duong HD, 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.
DeMeritt JS, Pile-Spellman J, Mast H, Moohan N, Young WL, Hacein-Bey L, Mohr JP, Stein BM. Outcome analysis of preoperative embolization with N-butyl cyanoacrylate in cerebral arteriovenous malformations. AJNR Am J Neuroradiol. 1995;16:1801–1807.
Hartmann A, Mast H, Mohr JP, Koennecke HC, Osipov A, Pile-Spellman J, Duong H, Young WL. Morbidity of intracranial hemorrhage in patients with cerebral arteriovenous malformation. Stroke. 1998;29:931–934.
Mohr JP. Acute clinical trials: an expression of concern. Cerebrovasc Dis. 1999;9:45–50.
Schaller C, Schramm J, Haun D. Significance of factors contributing to surgical complications and to late outcome after elective surgery of cerebral arteriovenous malformations. J Neurol Neurosurg Psychiatry. 1998;65:547–554.
Hofmeister C, Stapf C, Hartmann A, Sciacca RR, Mannsmann U, terBrugge K, Lasjaunias P, Mohr JP, Mast H, Meisel J. Demographic, morphological, and clinical characteristics of 1289 patients with brain arteriovenous malformation. Stroke. 2000; 31:1307–1310.