The value and limitations of magnetic resonance angiography of the circle of Willis in patients undergoing carotid endarterectomy.

Magnetic resonance angiography is a useful technique to determine the patency of the circle of Willis when compared with conventional four-vessel angiography. The purpose of this study is to determine whether the integrity of the circle of Willis, assessed by magnetic resonance angiography, provides adequate collateral cerebral circulation during carotid endarterectomy and correlates with internal carotid artery back pressure. Over a recent 20-month period, 35 patients were studied preoperatively with magnetic resonance angiography of the carotid bifurcations of the circle of Willis and the vertebrobasilar system. All patients underwent standard carotid endarterectomy with intraoperative measurement of internal carotid artery back pressure. Patients with an internal carotid artery back pressure < 50 mmHg had an intraluminal shunt placed. Deficiencies in branches of the circle of Willis, the carotid bifurcation and the vertebrobasilar system determined by magnetic resonance angiography were correlated with internal carotid artery back pressure using Fisher's exact test. Only one patient had a completely intact circle of Willis. Eleven of 16 patients (69%) who had an internal carotid artery back pressure < 50 mmHg had an occluded A1 segment of the anterior cerebral artery combined with an occluded posterior communicating artery, whereas only five of 19 patients (26%) who had an internal carotid artery back pressure > 50 mmHg had similar findings (P < 0.03). Severity of occlusive disease of the contralateral internal carotid artery and the basilar artery did not independently predict internal carotid artery back pressure. An occluded anterior branch of the circle of Willis in combination with an occluded posterior branch of the circle of Willis is associated with an internal carotid artery back pressure < 50 mmHg. Although magnetic resonance angiography of the circle of Willis may provide valuable anatomic information, it is not sufficiently accurate to predict the need for carotid shunting and therefore its use cannot be justified on a routine basis.