MR Angiography for the Diagnosis of Vasospasm after Subarachnoid Hemorrhage. Is It Accurate? Is It Safe?

There are about 30 000 cases of subarachnoid hemorrhage in the United States each year. Roughly two thirds of the patients who survive the hemorrhage will develop angiographically detectable vasospasm; of these patients, approximately one half will become symptomatic. The imaging changes can be detectable as early as day 3 and are maximal at about 1 week. Symptomatic vasospasm is associated with high morbidity and mortality, each in the range of 30% (1, 2).

The pathophysiology of vasospasm is incompletely understood, but both direct vessel constriction and impairment of vasodilatation appear to be involved. The total amount of the subarachnoid hemorrhage, as well as the presence of a focal clot encasing a given segment of vessel, are known risk factors for vasospasm (2).

Fortunately, there are a number of therapies that can positively affect the grim statistics associated with vasospasm. Hypervolemic hypertensive therapy is an accepted and widely employed treatment. Calcium channel–blocking agents, balloon angioplasty, intraarterial papaverine, subarachnoid thrombolysis, and papaverine infusion also appear to have efficacy (1).

At our center, and at many centers in the United States, early treatment of ruptured aneurysms associated with subarachnoid hemorrhage is advocated to allow for aggressive therapy to prevent and treat vasospasm. Thus, by the time vasospasm develops, an aneurysm clip or endovascular coils are often present, resulting in additional challenges for diagnostic imaging.

Cerebral angiography is the standard of reference for diagnosing vasospasm; however, it is associated with a small, but definite, risk of stroke. Transcranial Doppler sonography is widely employed for screening purposes in the intensive care unit, although a number of technical problems exist; in one recent study, test results were inconclusive for the detection of angiographically significant vasospasm in about 50% of patients (3). CT angiography has shown some promise as a screening test (4). Brain perfusion examinations are also useful, and may ultimately represent the best correlation with the degree of symptomatic vasospasm (5).

In this issue of the AJNR, Grandin et al (page 1611) examine the feasibility of diagnosing vasospasm in the presence of subarachnoid hemorrhage using three-dimensional time-of-flight MR angiography (MRA) performed on a 0.5-T imaging system. MRA was compared to digital subtraction angiography by use of only maximum intensity projection images obtained within 24 hours of each other. Prevalence of angiographic vasospasm in this sample was 26%. High signal intensity ascribed to methemoglobin was observed adjacent to vessels in 16% of cases, and was felt to be limiting in about 5% of cases. The ability of MRA to detect significant vasospasm correlated highly with that of digital subtraction angiography. Nonetheless, although no false-negative evaluations were present in the anterior cerebral artery distribution, false-negative rates of about 5%, as well as reduced sensitivity, were noted when data for the internal carotid and middle cerebral arteries were analyzed separately.

This is a technically challenging application for time-of-flight MRA. Two difficulties are immediately apparent. First, since vessels are screened for vasospasm in the subacute period, the increased signal associated with methemoglobin within residual subarachnoid hemorrhage adjacent to arteries can result in false-negative results. Indeed, the vessels that lie encased in subarachnoid clot are likely to be the vessels most severely affected by vasospasm. Although Grandin et al do not feel that subarachnoid hemorrhage was a significant source of false-negative readings in their study, I continue to be concerned about this source of error. Evaluation of source images or projection schemes using integration in addition to ray projection may reduce this error.

The second difficulty with using time-of-flight MRA for the detection of vasospasm involves the presence of aneurysm clips or endovascular coils, which often results in nonvisualization of adjacent arteries on MRA images. These vessels near the ruptured aneurysm often are most severely affected by vasospasm. In this trial, only a small number of studies were performed after aneurysm clipping. This would not be the case in many centers, and the presence of artifact associated with aneurysm clips could only increase the risk of missing significant stenosis.

Vasospasm associated with subarachnoid hemorrhage is a highly prevalent condition with a potentially devastating outcome, for which a number of effective therapies exist. In this setting, time-of-flight MRA does not represent a good alternative to conventional angiography. Nonetheless, MRA, preferably performed at high field strength, and possibly in combination with MR perfusion imaging, may have a place as a screening examination if a very low false-negative rate can be demonstrated. Grandin et al have taken an important first step in assessing this potential.