Magnetic Resonance Imaging Assessment of Cerebral Arteriovenous Malformation Obliteration After Stereotactic Radiosurgery

Abstract

Stereotactic radiosurgery is ideal for treating small cerebral arteriovenous malformations (AVMs) that are surgically inaccessible. However, given the inherent delay of AVM obliteration and the potential for radiosurgical failure, detailed evaluation of the neurovascular architecture is necessary to monitor persistence of residual flow. Modern imaging systems such as magnetic resonance imaging (MRI) and angiography allow clinicians to assess transnidus flow after radiosurgical intervention. To determine the accuracy of an MRI diagnosis of complete thrombosis and to identify variables that affect the precision of MRI assessment. One hundred twenty patients were reviewed after receiving radiosurgery at the University of Florida from 1990 to 2010. Each patient had an MRI demonstrating AVM obliteration and an angiogram either confirming or denying AVM thrombosis. MRI correctly predicted complete AVM obliteration in 82% of patients. There was a significant correlation between AVM volume and MRI accuracy in 2 separate models. In the first model, logistic regression analysis revealed a significant linear relationship between the natural log of AVM volume and MRI accuracy. The second model showed significant evidence of a cutoff point in MRI accuracy near an AVM volume of 2.80 cm(3), above which MRI agreement with angiography is 90% and below which MRI agreement falls off sharply to remain constant at 70%. MRI is a useful diagnostic system for assessing AVM obliteration, but its accuracy is inherently linked to the nidus volume it is measuring. These results suggest that MRI may be able to take on an increasingly independent role in the evaluation of AVM regression.