Brain and vascular imaging in acute ischemic stroke

Imaging in patients presenting with the symptoms of acute ischemic stroke has two objectives: assessment of the immediate pathophysiologic state of the cerebral circulation and tissue and assessment of the underlying disease. Diagnosis of the underlying disease will help in the choice of adequate secondary prophylaxis. Assessment of the brain's current pathophysiologic state is a prerequisite to choosing adequate treatment and is the more urgent of the two objectives. The imaging tool has to be available and employed quickly within the first hours after stroke onset. The imaging modality should reliably differentiate cerebral ischemia from other causes of a sudden central neural deficit. In addition, it should be able at an early stage to differentiate normal brain tissue from tissue at risk and from tissue that is already dying.

Historically, vascular imaging was developed first and provided the rationale for surgery in cases of high-grade carotid stenoses and for prophylaxis with antiplatelet agents. Brain tissue imaging with CT was developed later and this technique allowed the diagnosis of ischemic stroke and thus thrombolytic treatment, which recently has been shown to be beneficial.^1,2 A variety of new tools are now available for vascular and tissue imaging: ultrasound, CT angiography (CTA), MRI and MR spectroscopy (MRS), single photon emission tomography, and positron emission tomography.

Computed tomography and MRI/MRS are most promising for meeting the above-mentioned objectives in the clinical setting of acute stroke. Magnetic resonance perfusion imaging directly shows ischemic areas and magnetic resonance diffusion imaging immediately shows its sequelae with high sensitivity. However, the prospective value of MRI for treatment is not yet determined. Although CT is considered relatively insensitive in acute ischemic stroke by some authors, all major clinical trials are using CT before the randomization of patients. This is explained by the wide availability and high practicability of CT. This article focuses on CT and discusses whether CT is capable of distinguishing the reversibly ischemic tissue from the irreversibly injured tissue to determine which patients are most likely to benefit from thrombolysis and which are at risk from such treatment.

Ischemic cerebral edema. Brain edema plays a major role in the pathophysiology of ischemic stroke. Brain edema causes progressive microcirculatory compression, and thereby aggravates the primary ischemic insult.³ If focal cerebral blood flow (CBF) is reduced below the critical level of 10 to 15 ml/100 g/min by experimental permanent occlusion of the middle cerebral artery (MCA), cortical water content increases immediately and steadily from 80.7 to 83.0% wet weight within 4 hr.⁴ This net uptake of water is accompanied by an increase in sodium tissue concentration, a decrease in potassium tissue concentration, a shift of water from the extra- into the intracellular compartment, and a linear increase in brain volume within 4 hr of MCA occlusion.⁴ During this time period, the blood-brain barrier to serum proteins remains intact.⁴ The early water uptake without significant permeability changes of the blood-brain barrier is called "cytotoxic edema." It is caused by osmotic and ionic gradients between blood and ischemic brain tissue, pineocytosis of water in the presence of remaining blood flow (incomplete ischemia).⁴ Brain edema under such conditions does not resolve if the tissue is reperfused after 1 hour of ischemia. Moreover, reperfusion can further enhance edema in areas of dense ischemia.⁵ Above the critical flow level of 10 to 15 ml/100 g/min, brain edema does not develop.^4,6

CT can detect cytotoxic edema. The contrast on CT scans reflects different tissue X-ray attenuation, which is linearly proportional to tissue electron density, specific gravity or tissue water content.^7,8 A 1% increase in tissue water content causes a decrease in X-ray attenuation by 2.5 Hounsfield units (HU), i.e., the brain tissue becomes darker or "hypodense" in relation to normal brain parenchyma.⁹ We studied the development of ischemic brain edema after MCA occlusion in rats with CT (J. Weber et al., unpublished observations). CT was performed at 1, 2, 3, 4, and 6 hours after endovascular occlusion of the MCA using 3-mm-thick coronal slices. The changes in X-ray attenuation are shown in figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Measurement of mean X-ray attenuation after MCA occlusion in the affected and unaffected hemisphere of 10 rats. The curves are extrapolated between hour zero and one. HU, Hounsfield units.

In the unaffected hemisphere, X-ray attenuation remained stable at 50.0± 0.7 HU. In the ischemic hemisphere, X-ray attenuation significantly decreased to 48.4 ± 0.66 HU (p = 0.0011) at 1 hour after MCA occlusion, and further declined to 42.5 ± 1.6 HU at 4 hours and 40.4 ± 1.8 HU at 6 hours. The decline in X-ray absorption over 4 hours corresponds to an increase in tissue water content of 3%.⁹ This figure is in excellent agreement with the abovementioned observations in cats.⁴ We conclude from this experiment that CT is well suited to monitor ischemic cytotoxic brain edema.

Despite the significant decrease in HU at 1 hour, no hypodense area was visible at that time. The MCA territory became hypodense in relation to normal brain tissue in two of 10 rats at 2 hours and in all rats at 3 hours. That means that X-ray attenuation has to change by at least 4 HU to become visible under these circumstances. We presume that the volume of tissue with altered X-ray attenuation will also contribute to the ability to visualize parenchymal hypodensity.

These experimental findings, which prove that CT can reveal the first stages of ischemic brain edema, are in good agreement with clinical experience. In MCA trunk occlusion causing severe focal ischemia mainly in the basal ganglia, parenchymal hypodensity regularly develops within the first 6 hr after the onset of symptoms.^10-15 It is common experience that such early parenchymal hypodensities delineate a volume of tissue that will later become necrotic.^10,15,16 This can be explained by the density of focal ischemia necessary for the subsequent irreversible edema. The increased tissue water content first compromises the microcirculation and metabolism, and then the intracranial perfusion pressure, by mass effect. If the initial volume of ischemic brain edema is large, this vicious circle often ends in midbrain incarceration. We showed by a post hoc analysis that an initial hypodense tissue volume exceeding 50% of the MCA territory is associated with a 85% mortality.¹⁵ The increasing ischemic brain swelling and final midbrain incarceration are widely considered as untreatable and this situation was therefore recently called "malignant" MCA territory infarction.¹⁷

Prognostic value of CT in acute ischemic stroke. If CT identifies the extent of cytotoxic edema caused by severe ischemia, CT in acute stroke should have prognostic value. This was hypothesized for the European Cooperative Acute Stroke Study (ECASS). The patients were stratified into three groups: patients without any abnormal parenchymal hypodensity appropriate to acute stroke; patients with small hypodensities not exceeding one-third of the MCA territory; and patients with large parenchymal hypodensity. According to the protocol, the latter group of patients should have been excluded from the study, in anticipation of a poor clinical course. However, 52 patients with such extended cytotoxic edema were identified by a central CT reading panel after randomization.¹

Table 1 presents the assessment of the extent of cytotoxic edema by the CT reading panel. The first CT scan showing edema was performed at 35 minutes after the onset of symptoms. The extent of cytotoxic edema as depicted by the initial CT was not associated with the time interval from stroke onset (p = 0.522), age (p = 0.377), and gender (p = 0.757) of patients. It was correlated with neurologic performance on admission measured with the Scandinavian stroke score(p < 0.0001).

In the placebo group of the ECASS, disability and mortality at 90 days after the stroke were associated with the extent of cytotoxic edema revealed by the initial CT (p < 0.0001) (figure 2). Sixteen of 21 patients (76%, 95% CI 55-89%) with large ischemic edema had a poor clinical outcome (death or severe disability). The proportion of patients with poor outcome was 59% (CI 49-68%) with small edema and 34% (CI 27-41%) in patients without edema. One patient (0.6%) with no cytotoxic edema on admission died from extended brain edema, nine patients (8.2%) with small edema, and five patients (23.8%) with large cytotoxic edema. Two patients(1.2%) with no cytotoxic edema died from brain hemorrhage, five patients(4.5%) with small edema, and no patient with large cytotoxic edema.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Rankin scale at 3 months after stroke in the placebo group and recombinant tissue plasminogen activator (rt-PA) group in the ECASS. Rankin 0 and 1 means no disability, Rankin 2 and 3 means mild and moderate disability, Rankin 4 and 5 means severe disability. The patients are stratified according to the extent of parenchymal hypodensity in the MCA territory on the baseline CT scan. A total of 336 patients had no hypodensity, 215 patients had small hypodensities (≤33% of MCA territory), and 52 patients had large (>33% of MCA territory) parenchymal hypodensities.

Figure 2 shows the disability at 90 days for both treatment arms in patients with no, small, and large cytotoxic edema on initial CT scan. In patients with no or small cytotoxic edema, treatment with rtPA significantly increased the cure rate by 8% and 18%, respectively, and decreased the cure rate from 14% to 6% in patients with large cytotoxic edema. The proportion of patients with poor outcome was reduced by rtPA in all three subgroups.

We studied the response to rtPA in patients with no, small, and large cytotoxic edema by calculating the odds ratio (OR) (rtPA/placebo) for the following parameters: lack of disability at 90 days; overall mortality; fatal brain edema; and fatal brain hemorrhage (table 2). Treatment with rtPA significantly enhanced the chance of achieving a state without disability after stroke in patients with no or small cytotoxic edema but tended to diminish this chance in patients with large edema on their initial CT. Treatment with rtPA produced a nonsignificant increase in the risk for death within 90 days after stroke in patients with no or small cytotoxic edema and a more pronounced increase in risk in patients with large cytotoxic edema. However, the risk for death from brain edema was marginally and nonsignificantly enhanced by rtPA regardless of whether the early ischemic edema was small or large. Treatment with rtPA also increased the risk for fatal brain hemorrhage. The increase did not reach the limits of statistical significance in patients with no or small cytotoxic edema. In the rather small group of patients with large edema, fatal brain hemorrhage occurred in 5 of 31 patients after rtPA treatment (16%, 95% CI 7-33%) but in no patients receiving placebo. rtPA-treated patients with large cytotoxic edema died more often from fatal cerebral hematoma than patients with no or small hypodensity(χ² = 4.55, p < 0.05). Nine of 87 patients (10%) treated within 3 hours of stroke onset showed an ischemic edema exceeding 33% of the MCA territory. Four of the five such patients in the placebo group and three of the four patients in the rtPA group had a poor clinical outcome.

We conclude from this post hoc analysis of the ECASS data that the response to rtPA may be different in patients with no, small, or larger cytotoxic edema visible on diagnostic CT. This impression has to be confirmed in future trials on thrombolysis by prospectively stratifying patients according to their initial ischemic edema.

The potential of CT angiography. The volume of brain tissue that is prone to die as a result of severe ischemia can be small or large, depending on the site of occlusion and the capacity of collaterals. If the ischemic edema covers the entire territory of an occluded artery, this entire territory appears to be irreversibly damaged. However, if the ischemic edema is small in relation to the territory of the occluded artery, the remaining portion of this territory may be at risk from hypoperfusion. We suggested that it may be useful to perform CTA, in addition to an initial unenhanced CT, to estimate the volume of tissue compromised by hypoperfusion (Knauth et al., unpublished observations). In our experience, CTA can assess the occlusion of all large intracranial arteries with high sensitivity and specificity. For CTA, 130 ml of a nonionic contrast agent is infused into an antecubital vein at a rate of 4-5 ml/s using an injection pump. Twenty seconds later a spinal CT scan with 1.5- or 2-min section thickness is performed, covering the circle of Willis, Using a volume-rendering algorithm, the arteries are reconstructed three-dimensionally, which takes about 10 minutes. Figure 3 shows CTA of a 45-year-old-male physician 85 minutes after sudden right-sided hemiparesis and aphasia. Angiography later confirmed a dissection of the left internal carotid artery and, an embolus into the left MCA with spontaneous clot resolution associated with significant clinical improvement. We have shown experimentally that the nonionic contrast agent used does not enhance infarct volume even if a double dose is applied (Doefler et al., unpublished observations). We therefore believe that CTA combined with the initial unenhanced CT can safely define who is most likely to benefit from thrombolytic drugs and who is most at risk with such therapy.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Computed tomography angiography of a 45-year-old male physician 85 minutes after sudden right-sided hemiparesis and aphasia. The cavernous segment of the left (view from above) intracranial internal carotid (ICA) artery is missing (large arrow). The left posterior communicating artery and the A1 segment of the anterior cerebral artery are visible. Subtotal occlusion of one major left MCA branch (small arrow). Interpretation: occlusion of the left ICA by thrombus or dissection and subsequent embolism into the left MCA with partial occlusion of the MCA bifurcation.