Dynamic CT Perfusion in Acute Stroke

Abstract & Commentary

Synopsis: The imaging of acute stroke is undergoing rapid evolution as advances are made in MR and CT technology. Because most patients with suspected acute stroke still are evalutated with CT scanning, CT-based methods of evaluating brain tissue perfusion could significantly improve the diagnosis and management of acute stroke. Mayer and colleagues present a method of dynamic CT perfusion imaging that is feasible for any clinic with a third-generation CT scanner and discuss the results of CT perfusion analysis in 70 patients.

Source: Mayer TE, et al. Dynamic CT perfusion imaging of acute stroke. AJNR Am J Neuroradiol 2000;21:1441-1449.

The non-contrast head ct scan is the most widely used study in the setting of suspected acute stroke, and it is very useful for excluding other pathologies (such as brain hemorrhage or neoplasm) that may mimic cerebral infarction and for confirming the presence of cerebral infarction if characteristic CT findings are present. Both the diagnosis and treatment of acute stroke, however, might be improved if cerebral perfusion could be measured to identify tissue at risk and to confirm areas of infarcted tissue that might not be apparent on routine CT. Other technologies may be used to assess cerebral perfusion parameters (xenon CT, MR-based methods, positron emission tomography, etc.), but they are not necessarily available in a timely fashion or at all in many institutions. Mayer and colleagues had three aims in their study: to investigate whether the use of dynamic contrast-enhanced CT augmented by software that rapidly provides maps of perfusion parameters is feasible in cases of acute stroke, to determine which imaging parameters best detect tissue at risk, and to identify perfusion thresholds for predicting the development of infarction.

Mayer et al prospectively evaluated 70 adult patients with hemispheric symptoms of less than 12 hours’ duration whose plain CT scans showed no signs of hematoma. The initial plain CT and the perfusion maps were compared with follow-up CT scans or MR scans obtained within a week. Patients with brain stem symptoms were excluded. After the routine non-contrast CT scan was performed, dynamic contrast-enhanced CT scans were acquired by injecting 50 mL of iodinated contrast medium via a 16-gauge catheter at 10 mL/sec. Because only a single section was obtained with each bolus, one to three levels were obtained per patient, with the initial section at the level of the basal ganglia and additional rostral or caudal scans obtained at the discretion of the monitoring physician. Commercially available software was used to estimate the following cerebral perfusion parameters: cerebral blood flow (CBF), cerebral blood volume (CBV), and time to peak (TTP). Technically, the method required only a personal computer with appropriate evaluation software and a DICOM connection to the CT scanner.

The results of perfusion CT for three patients with acute stroke symptoms showed no evidence of cerebral ischemia. Further assessment led to diagnoses of seizure in one, astrocytoma in one, and radiation injury in one. Fourteen patients had symptoms that resolved within 24 hours, with 13 diagnosed with transient ischemic attack and one diagnosed with migraine. Only two out of 13 had moderate TTP delay, and none developed an infarct. Of the 53 patients in whom cerebral infarction was confirmed by follow-up studies, six had lacunar infarcts. These small ischemic foci were difficult to distinguish from artifacts in the white matter by perfusion scanning, and the actual lesion was suspected in only two out of six cases. The remaining 47 patients included 44 territorial infarcts and three infarcts of hemodynamic origin (low flow due to carotid stenosis). Forty of these infarcts were detected on CBF maps, while seven of the smaller gyral infarcts were variably detected on TTP, CBV, and CBF maps. The Table presents an abbreviated summary of results of conventional and perfusion imaging in cases of subsequently confirmed nonlacunar infarcts.

Overall, the specificity of early CT signs of infarction (swelling or hypodensity of the gray matter) was very high, but sensitivity was relatively poor. CBF maps indicating moderate or severe deficits showed true-positive results for infarcts in more than 90% of the brain territories and patients. CBV maps showed fewer definite regions of low perfusion than did CBF maps, but gave fewer false-positives. TTP maps were nearly as sensitive indicators of cerebral infarction as were CBF maps but were less specific, probably due to carotid stenosis compensated via collaterals. CBF and TTP maps together were 100% sensitive for nonlacunar cerebral infarction.

Comment by Nancy J. Fischbein, MD

With the availability of an increasingly sophisticated therapeutic armamentarium for the patient with acute ischemic infarct comes the need to clearly diagnose the presence of infarction and to define territory at risk. Since CT scanning remains far more accessible than MR-based diffusion and perfusion techniques in most medical centers, the assessment of CT-based perfusion parameters may soon become an integral part of the management of these patients. Clarification as to which parameter or combination of parameters best defines regions of frank infarction, tissue at risk, and normal tissue will come as larger studies are done in more varied patient populations. For the moment, dynamic CT perfusion imaging is feasible in any clinic with a third-generation CT scanner and should probably be attempted in any patient thought to have acute ischemic symptoms who has a normal baseline CT scan. In this study, single-section CBF maps predicted all territorial infarcts with volumes larger than 10 mL. CBF images were superior to CBV images in sensitivity and superior to TTP images in specificity. Infarction occurred in all vascular territories where loss of perfusion was more than 70% and in more than half of the cases where perfusion losses were between 40% and 70%.

There are some drawbacks to dynamic CT perfusion that should be kept in mind as we all attempt to gain experience with this technique. First, at present, only a single section or a few sections can be acquired due to limits on the amount of contrast that can be injected, as well as time, radiation dose, and other considerations. Second, the CBF image can be used only as a relative CBF map, and this can be problematic when contralateral areas of brain have baseline abnormalities that would not be known at the time of the acute CT scan. Third, low spatial resolution can cause small lesions to be overlooked, though this generally will not compromise acute management as these patients would be unlikely to benefit from thrombolytic therapy. Image quality can also be impaired in patients with poor cardiac function.