Computed Tomography

A noncontrast head CT is the initial imaging modality of choice for patients with suspected stroke for two main reasons. The foremost is the expediency with which one can obtain a CT scan because of its widespread availability, and the second is the ability of CT to exclude intracranial hemorrhage. However, in addition to differentiating ischemic stroke from hemorrhage, CT may demonstrate subtle parenchymal abnormalities indicative of early edema or infarction. It was previously believed that these changes did not occur on CT for at least 6 hours after ischemic stroke. More recent studies indicate, however, that early changes of ischemia frequently occur within a few hours of stroke onset and have been seen as soon as 1 hour after stroke.⁸ These changes include reduced attenuation in the basal ganglia,² loss of gray-white differentiation particularly in the insular region,⁹ low density in the cortex and subcortical white matter, and loss of sulcal markings, suggesting early mass effect and edema (Figure 34-1, A and B).¹⁰

Figure 34-1 A, Normal computed tomography (CT) scan of brain 2 hours after onset of aphasia and left hemiparesis. B, Repeat CT scan at 5 hours after stroke onset shows early CT changes, including basal ganglia hypodensity, loss of the insular ribbon, and slight effacement of the sulci on the left. C, CT angiogram at 5 hours after stroke onset shows complete occlusion of the left middle cerebral artery. D, Rapid reconstruction of the CT angiogram again shows occlusion of the left middle cerebral artery.

A hyperdense middle cerebral artery occurs in 20% to 37% of cases,¹¹ indicating acute thrombus within the artery. It rarely occurs without at least one other early CT abnormality. Hyperdensity in the basilar artery associated with thrombosis also has been reported.¹² In 100 patients studied within 14 hours (mean 6.4 hours) of stroke onset, multiple early CT abnormalities correlated with size of subsequent infarct and poor outcome.¹¹ In the ECASS I trial of tPA for acute stroke, early CT changes correlated with larger subsequent infarct volume and a greater likelihood of hemorrhagic conversion after tPA.¹³ Quantitative assessment of CT changes using the Alberta Stroke Program Early CT Score (ASPECTS) scale in patients treated with IV tPA also showed a relationship between early CT hypodensity (ASPECTS < 8) and hemorrhage (Table 34-2).^14,15 Based on these results, some experts recommend withholding thrombolytic therapy in patients with extensive early CT changes, particularly in patients later in the thrombolytic time window,¹⁶ although this practice is somewhat controversial. For example, subsequent analysis of the NINDS rt-PA trial data discovered that early ischemic changes did not predict symptomatic hemorrhage or response to treatment,¹⁷ and more recent evidence reports no association between early ischemic CT changes and outcome.¹⁸

TABLE 34-2 ASPECTS Measurement Tool for Early Changes on Computed Tomography

* One point is subtracted for each defined area of early ischemic change, such as focal swelling or parenchymal hypoattenuation. Score varies from 0-10.

Computed Tomography Angiography

CT angiography (CTA) can be performed using spiral CT technology, allowing for imaging of the intracranial and extracranial circulation. Optimally, CTA of the neck should include visualization of the aortic arch as well. The typical single bolus of iodine contrast material is approximately 70 mL of iodine. Owing to this injection, CTA is of limited use in patients with renal failure or contrast hypersensitivity. In acute stroke, CTA of the head and neck has been shown to be highly reliable for diagnosis of intracranial occlusions and correlates with other imaging modalities.^19,20 Three-dimensional reconstruction images can also be created using this technology and can provide additional views and information about the carotid bifurcation and carotid lesions, showing eccentric lesions or ulceration not visualized by conventional angiography (see Figure 34-1, C and D).

Computed Tomography Perfusion

In addition to imaging the brain parenchyma with a noncontrast head CT and the cerebral vasculature with CTA, CT perfusion (CTP) adds assessment of cerebral blood volume (CBV) and cerebral blood flow (CBF). Using a helical scanner during a bolus of IV contrast, the time-dependent concentration curve of contrast in each pixel can be acquired. Mean transit time (MTT) and subsequently CBF can be calculated (Figure 34-2). In patients with acute stroke, CTP has been correlated with final infarct size and outcome, particularly after recanalization.²¹ CTP maps combining CBV and CBF information identify brain tissue that progresses to infarction if not reperfused, consistent with ischemic penumbra.²² Recent evidence suggests that the inclusion of CTP in a stroke imaging protocol increases diagnostic performance.^21,23,24

Figure 34-2 Computed tomography brain perfusion scan with sequencing maps. A, Cerebral blood volume (CBV) showing no clear evidence of core infarct. B, Cerebral blood flow (CBF) showing a decrease in the right middle cerebral artery (MCA) territory. C, Mean transit time (MTT) showing delayed perfusion in the right MCA territory. These sequences together indicate a large ischemic penumbra in the right MCA territory.

Whereas CTP serves as a qualitative measure of blood flow, there have been recent investigations into using xenon CT as a quantitative measure of blood flow.²⁵ Stable xenon is an inert gas inhaled as a mixture of 27% xenon and 73% oxygen. During inhalation over a few minutes, rapid scanning is performed and pixel-by-pixel blood flow values are calculated at different brain levels (Figure 34-3). In a series of patients with middle cerebral artery (MCA) occlusion studied with xenon CT, areas of penumbra were present in all patients, and the percentage of MCA territory in the penumbral range (i.e., cerebral blood flow 8 to 20 mL/100 g/min) remained relatively constant across the group. In contrast, the percentage of MCA territory with CBF values representing infarcted tissue (i.e., cerebral blood flow <8 mL/100 g/min) varied greatly. Outcome was highly correlated with the area of infarcted MCA territory, not the amount of ischemic penumbra. These results suggest that after the first few hours, the size of the core infarcted tissue, not the amount of penumbral tissue, may be the most important imaging parameter to determine suitability for acute stroke therapy.²⁶

Figure 34-3 Xenon computed tomography blood flow study from a patient with large left hemisphere stroke 3 hours after onset of symptoms. Flow is nearly absent throughout the middle cerebral artery territory on the left.

Magnetic Resonance Imaging

Compared to CT modalities, MRI brain imaging is advantageous because it is more sensitive to cerebral infarction, especially in the brainstem and deep white matter. Typical sequences included in a MRI stroke protocol include diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) to evaluate for potential acute ischemia, multiplanar gradient-recalled (MPGR) or gradient recalled echo (GRE) to evaluate for hemorrhage, and fluid attenuated inversion recovery (FLAIR) to evaluate for important signs in both hyperacute and acute stages of stroke (i.e., assessment for absence of flow void in major cerebral arteries, which suggests occlusion or slow flow in that artery). Perfusion-weighted imaging (PWI) is also a sequence often used to determine abnormal tissue perfusion based on transit times for contrast material through brain parenchyma (Figure 34-4).

Figure 34-4 Magnetic resonance imaging of the same patient in Figure 34-2. A, Diffusion-weighted imaging (DWI) showing right basal ganglia stroke. B, Perfusion-weighted imaging (PWI) showing enhanced mean time to enhancement. These sequences together suggest a large ischemic penumbra in the right MCA territory.

DWI shows parenchymal abnormalities earlier than conventional T2-weighted images in patients with acute stroke.²⁷ DWI detects the diffusion of water in the brain and shows hyperintensity in areas of reduced diffusion (see Figure 34-4). As water moves from the extracellular to the intracellular space, there is less movement of water and loss of signal, resulting in hyperintensity.²⁸ DWI provides advantages in the evaluation of acute stroke. Early detection of lesions helps differentiate cerebral ischemia from other conditions that mimic stroke, such as seizures or toxic metabolic states. Additionally, combining DWI with PWI may identify reversibly ischemic tissue. If there is a large area of PWI abnormality indicating reduced blood flow but limited established infarction as evidenced by DWI abnormality, penumbral tissue is likely present, indicating areas of impaired flow at risk of undergoing infarction.

In stroke patients, the size of the DWI lesion and the growth of these abnormal DWI regions are strong predictors of outcome. In acute stroke, a marker of tissue viability is needed, and some investigators have suggested that the extent of mismatch between lesions on DWI and PWI could serve as this marker. The concept of DWI/PWI mismatch has been used as an inclusion criterion in several clinical trials (i.e., DIAS, DIAS-2, DEDAS, DEFUSE, EPITHET) assessing thrombolytic agents and is being employed more frequently to select patients that may ultimately benefit from reperfusion therapy.^29–34 Patients with mismatch might be more likely to respond to reperfusion therapy.³⁵ Patients with large areas of DWI abnormality or large severe PWI abnormalities may in fact be at greater risk for hemorrhage if reperfusion therapy is pursued.³⁶

Magnetic Resonance Angiography

Magnetic resonance angiography (MRA) of the head and neck offers a noninvasive method of imaging the intracranial and extracranial vasculature. MRA typically uses gadolinium contrast in appropriate patients, but important information can be obtained based on time-of-flight techniques not utilizing contrast.^37,38 Detection of dissection or occlusion in the circle of Willis and the extracranial vertebral and carotid arteries can be examined with MRA, but occlusions of small peripheral branch arteries may not be detected. Artifact may also wreak havoc in some cases by obscuring proper identification of arterial pathology. Signal dropout may occur at the site of arterial stenosis, owing to the effects of turbulent flow. If an artery is tortuous, it may extend out of the imaging section and appear occluded. MRI tends to overestimate the severity of stenosis, and evidence of severe stenosis should be confirmed with another modality. MRA is better for localizing the site of stenotic lesions than determining severity of stenosis. Similarly, differentiation between severe stenosis and occlusion is unreliable with MRI, and apparent occlusions by MRA should also be confirmed with angiography.

Intravenous Thrombolysis

Acute stroke trials using IV thrombolytic agents date back to the early 1960s with use of streptokinase,³⁹ fibrinolysin,⁴⁰ and urokinase,⁴¹ showing either no benefit or a higher mortality in patients treated with thrombolysis. These studies preceded CT imaging, so patients with hemorrhage were not excluded. The discouraging results hindered the development of more acute stroke trials until the 1980s, when several case reports showed favorable outcomes with intraarterial thrombolytic therapy within a few hours of stroke onset.^42,43 These reports resulted in small randomized trials and feasibility studies of IV thrombolytics^44,45 which ultimately gave rise to the pivotal NINDS rt-PA trial that showed for the first time a beneficial effect of thrombolytic therapy for acute stroke treatment when administered within 3 hours of symptom onset.⁴

Tissue Plasminogen Activator Within 3 Hours

The NINDS trial included more than 600 patients with acute ischemic stroke. All patients were treated within 3 hours, and half of them were treated within 90 minutes. Patients were randomly assigned to receive either IV tPA at a dose of 0.9 mg/kg to a maximum of 90 mg, or IV placebo. Primary outcome measures were favorable outcomes at 90 days measured by the NIHSS, Barthel Index, Glasgow Outcome Scale, and modified Rankin Scale (mRS). By all four measures, significantly more patients had a favorable outcome at 90 days in the tPA group compared with placebo. Treatment with tPA resulted in an 11% to 13% absolute increase in good outcomes and a minor, non-significant decrease in mortality at 3 months. The benefit was sustained at 12 months.⁴⁶ Intracerebral hemorrhage with clinical deterioration occurred in 6.4% of patients treated with tPA as compared to only 0.6% of placebo patients. Despite the increased hemorrhage rate, there was no significant increase in mortality or severe disability in the tPA group compared with placebo. When strokes were classified according to initial impression of stroke subtype, all types of strokes had more favorable outcomes with tPA. There were no clear factors that predicted response to tPA.⁴⁷ Patients with large strokes as measured by NIHSS score higher than 20 and evidence of early low density or edema on CT had a higher rate of hemorrhage after tPA.⁴⁸

On the strength of these results, in June 1996 the FDA approved IV tPA for treatment of stroke within 3 hours of onset. This recommendation was supported by the results of an analysis of patients treated within 3 hours of onset in the ATLANTIS trial.⁴⁹ A subsequent pooled analysis of NINDS rt-PA, ECASS, and ATLANTIS data showed that clinical benefit with tPA is greatest when given early, especially if started within 90 minutes (Table 34-3).⁵⁰ It was noted that not all patients recanalize with IV tPA. In a dose escalation trial of IV tPA, angiography was performed before thrombolysis in all patients, documenting the site of arterial occlusion and repeated 2 hours later. Proximal occlusions in the MCA opened less frequently than distal branch occlusions, and only 8% of carotid occlusions recanalized.⁵¹

TABLE 34-3 Odds Ratios for Modified Rankin Score 0-1 in the Combined tPA Analysis

Intraarterial Therapy

Intraarterial Thrombolysis

An alternative approach to IV thrombolysis is direct delivery of thrombolytic agents by a microcatheter embedded in the clot (Figure 34-5). The advantage of the intraarterial approach is direct visualization of the occluded artery and knowledge of the recanalization status as thrombolysis proceeds. Theoretically, delivery of the thrombolytic agent to the site of the clot should be more effective than IV infusion. The disadvantage is the additional time needed to bring the patient to the angiography suite, prepare the groin, catheterize the femoral artery, and guide the catheter from the femoral artery to the intracranial circulation before the thrombolytic agent can be administered.

Figure 34-5 A, Right carotid angiogram from a patient with embolic occlusion of the right middle cerebral artery (MCA) 4 hours after onset of symptoms. B, Angiogram from the same patient after placement of a microcatheter into the MCA clot and infusion of 120,000 U of urokinase. There is no recanalization. C, Angiogram after infusion of 1 million U of urokinase directly into the clot, showing complete recanalization of the MCA.

Urokinase was used in early studies of intraarterial thrombolysis but is no longer available.⁶⁶ Recombinant prourokinase was evaluated formally in clinical trials,^67–69 and the PROACT II study was the first acute stroke trial to show a statistically significant improvement in clinical outcome when administered within 6 hours of stroke symptom onset. The median time to treatment was 5.5 hours, and most patients were treated after 5 hours.⁶⁸ The clinical benefit was apparent despite this late time to treatment, and possibly a greater benefit would have been found had patients been treated earlier or mechanical manipulation also been allowed. Symptomatic hemorrhage occurred in 10% of patients treated with recombinant prourokinase and in 2% of controls.

Although the hemorrhage rate was higher than previous IV thrombolytic studies, the median NIHSS score of 17 indicates that the patients in the PROACT II study had more severe strokes treated at a later time interval. A higher hemorrhage rate would be expected in these scenarios. Based on factors predicting outcome in this group of patients, the treatment and control groups can be stratified according to risk. There was no differential effect of recombinant prourokinase across risk strata, indicating that all patients, regardless of risk, benefit equally from recombinant prourokinase.⁶⁹ Despite these results, prourokinase has not been FDA approved to date, and tPA tends to be more often used in cases of intraarterial thrombolysis. The exact dose, efficacy, and safety profile of intraarterial tPA is limited, but recent studies have suggested doses up to 40 mg are reasonably safe for use.⁷⁰