Treatment of Acute Ischemic Stroke

Abstract

Endovascular thrombectomy for stroke has become an integral part of the management of acute ischemic stroke. This chapter discusses mechanical thrombectomy and pharmacological thrombolysis. Patient selection, technical issues, and post-procedure care are covered. The appendix covers imaging for stroke.

Appendices

Appendix 8.1: Primer on Imaging in Stroke Joel K. Curé

Imaging goals in ischemic stroke:

• 1. Confirming a diagnosis of ischemic stroke and exclusion of nonvascular (e.g., tumor) causes of the clinical ictus

• 2. Exclusion of hemorrhage and estimation of the risk of hemorrhagic transformation

• 3. Selection of patients for reperfusion therapy by distinguishing ischemic but viable (i.e., the penumbra) from infarcted tissue and excluding those for whom the therapeutic risk exceeds the anticipated benefit

• 4. Identifying large-vessel occlusion that may complicate or represent a target for therapy

CT is the most practical initial brain imaging test for stroke. This may change with increasing availability of MRI. MRI is more sensitive than CT for detection of acute infarction, is able to detect both acute and chronic hemorrhage as effectively as CT, and demonstrates higher interobserver and intraobserver reliability than CT for ischemic stroke diagnosis, even in readers with little experience [169].

Stroke location and distribution at imaging reflect mechanism [170]. Most strokes are thromboembolic and their imaging appearance reflects the territory of the occluded artery less any portion of that territory that is adequately perfused by collateral blood supply. Solitary or multiple unilateral cortical or cortical/subcortical infarcts may be secondary to either cardiogenic emboli or large arterial occlusion. Cardiogenic emboli typically account for bilateral acute infarcts in both the anterior and posterior circulations, especially in the absence of definable intracranial arterial occlusions on CTA, MRA, or transcranial Doppler. However, multiple synchronous infarcts may be encountered in patients with occlusive vasculopathies (e.g., CNS vasculitis) or coagulopathies. Lacunar infarcts, due to small arterial occlusion, are typically small (<1.5 cm) and imaging abnormalities correspond to the territory of the occluded perforating artery. These infarcts most commonly occur in the basal ganglia, thalamus, brainstem, or deep cerebellar white matter. Arterial border zone infarcts occur in regions of brain that lie between major arterial territories. These include deep cerebral hemispheric regions such as the centrum semiovale, corona radiata, and cortical zones between the ACA and MCA and the MCA and PCA territories. Arterial border zone infarcts may occur bilaterally in cases of global cerebral hypoperfusion or unilaterally in patients with severe ICA stenosis or MCA stenosis plus A1 segment hypoplasia.

1.1 Non-contrast CT Diagnosis of Acute Infarction

In CT scan interpretation the terms “hypoattenuation” and “hyperattenuation” are preferred to “hypodense” and “hyperdense.” Attenuation indicates the degree of X-ray absorption that occurs within tissue. In patients with stroke, hypoattenuating tissue tends to be edematous, and hyperattenuating tissue tends to be hemorrhagic.

Brain edema associated with hemispheric stroke may be detectable within 1–2 h of stroke onset. CT identifies ischemic lesions with a sensitivity of 65% and a specificity of 90% within 6 h of stroke onset [171]. However, the sensitivity of CT for acute ischemic stroke within the first 3 h of symptom onset has been reported to be as low as 7% [169]. CT is insensitive for small acute infarcts, especially in the posterior fossa [172], and is less sensitive than diffusion-weighted MRI in the acute setting [173]. The peak period for identifying brain ischemia on CT is 3–10 days after the ictus, well beyond the thrombolytic time window. The value of early CT in acute stroke is therefore not chiefly diagnostic, but prognostic. A large hypoattenuating area detected within 6 h of stroke onset is an indication of irreversible tissue injury [171], portends an increased risk of hemorrhagic transformation in patients treated with rTPA [174], and is associated with an increased risk of fatal brain edema [175]. ECASS-1 and subsequent analyses of its CT and patient data led to the “one-third rule.” Patients with CT-identified early ischemic changes (EIC) involving less than one-third of the MCA territory had an improved functional outcome after IV thrombolysis compared to patients with EIC in more than one-third of the MCA territory or who had no EIC on CT [176]. However, the unreliability of volume estimation with the one-third rule and lack of demonstrable evidence of an effect on treatment modification led to development of the ASPECTS scoring system (Fig. 8.3). This scoring system assigns one point each to ten regions within the MCA territory. A point is deducted for each of the ten regions demonstrating EIC. In patients undergoing intravenous thrombolysis at less than 3 h from symptom onset, a baseline ASPECTS score less than or equal to 7 predicted patients who were unlikely to achieve independent functional outcome [177]. Sensitivity of CT for acute intracranial hemorrhage approaches 100% [178].

Fig. 8.3

ASPECTS (Alberta Stroke Program Early Computed Tomography Score). This CT shows a patient with a left MCA-territory ischemic stroke with an ASPECTS score of 6. The right hemisphere is unaffected. A normal ASPECTS score (i.e., no evidence of ischemic stroke) is 10. Subtract one point for each territory showing evidence of ischemia in two standard CT cuts (level of the basal ganglia, left, and above the ganglia, right). In this case, the four areas of ischemic change in the left hemisphere are the internal capsule, lentiform nucleus, internal capsule, and M2. Abbreviation: Region of brain affected showing ischemic change: C Caudate, I Insular ribbon, IC Internal capsule, L Lentiform nucleus, M1 Anterior MCA cortex, M2 MCA, M3 Posterior MCA cortex, M4 Superior territory: Anterior MCA cortex, M5 Superior territory: Lateral MCA cortex, M6 Superior territory: Posterior MCA cortex

1.2 Early CT Findings that Suggest Infarction

• 1. Loss of gray–white differentiation may be detected within 6 h of onset of stroke symptoms in 82% of patients with MCA territory ischemia [179]. Cytotoxic edema reduces the attenuation of gray matter into the range of white matter, thereby decreasing gray–white matter contrast.

  • (a) “Insular ribbon sign”: Loss of gray–white matter differentiation in the insular cortex can be an early sign of MCA ischemia [180].

  • (b) Obscuration of the lentiform nucleus reflects cytotoxic edema within the involved basal ganglia, again decreasing gray–white matter contrast [181].

• 2. Cortical sulcal effacement due to swelling of edematous gyri.

• 3. “MCA sign.” Hyperattenuation of the M1 segment (or other intracranial arteries, e.g., the posterior cerebral artery) due to thromboembolism (Fig. 8.4).

• 4. “Sylvian dot sign.” Distal MCA (M2 or M3 branches) occlusion indicated by hyperattenuation in the Sylvian fissure [182]. Sensitivity 38%, specificity 100%, positive predictive value 100%, negative predictive value 68% [183].

Fig. 8.4

MCA sign. Hyperattenuation of the M1 segment due to thromboembolism (arrow)

The combined presence of the insular ribbon sign (Fig. 8.5), hemispheric sulcal effacement, and decreased attenuation of the lentiform nucleus is predictive of ICA occlusion [184].

Fig. 8.5

Insular ribbon sign. Insular ribbon sign. Loss of gray–white differentiation in the insula (arrowheads) can be an early sign of MCA territory ischemia

The identification of ischemic penumbra may be useful in three scenarios:

• 1. Offering treatment to patients who do not qualify for treatment under current guidelines (e.g., beyond the “time window”).

• 2. Identifying patients for whom treatment within the current time window is likely to be futile.

• 3. Identifying IV-tPA nonresponders to whom endovascular therapies might be offered [185].

1.3 CT Angiography

CT angiography (CTA) is useful in identifying large-vessel occlusion, and can complement CT perfusion. The time required for acquisition, processing, and analysis of CTA studies of patients with acute ischemic stroke averages 15 min [186]. Compared to catheter angiography, CTA has sensitivity and specificity for the detection of large-vessel occlusion of 98.4% and 98.1%, respectively [186]. CTA may be prone to false-positive results; in two series of CTA in acute stroke, a minority of patients were found to have lesions on CTA that could not be found with catheter angiography [187, 188]. CTA can be particularly useful in assessment of vertebrobasilar occlusion [128], as CT perfusion imaging of the posterior circulation territory is limited because of bone artifact. However, basilar artery lesions can be better assessed with CTA than vertebral artery lesions [189]. CTA combined with CT perfusion shows good agreement with MRI in the assessment of infarct size, cortical involvement, and intracranial cerebral artery occlusion [190]. Finally, some authors have found application of ASPECTS [177] scoring to CTA source images a robust method (and superior to ASPECT analysis of routine non-contrast brain CT) for early detection of irreversible ischemia and prediction of final infarct volume [191, 192]. Multiphase CTA is another option; [193] this technique was used to screen patients for enrollment in the ESCAPE trial [6].

1.4 CT Perfusion

CT perfusion provides quantitative data about CBF, and is becoming widely available on multidetector CT scanners. CT perfusion involves repeated (“cine”) helical CT imaging of the brain during the transit of an injected bolus of iodinated contrast through the intracranial vasculature. Measurements of the change in tissue attenuation during passage of the contrast bolus are used to generate quantitative information about CBF as well as CBV and time to peak (TTP) or mean transit time (MTT). Acquisition and processing of the data are accomplished seconds to minutes. The concept of CT perfusion was introduced more than 20 years ago [194], but had to await the development of high-speed helical CT scanners, fast computers, and software capable of rapid data analysis to make the technique clinically useful.

Normal Values of CBF and CBV

Cerebral blood flow is normally maintained within a narrow range by autoregulation. Normal CBF is approximately 80 mL per 100 g/min in human gray matter and approximately 20 mL per 100 g/min in white matter. Global CBF, as well as average CBF in the cortical mantle, which is roughly a 50:50 mix of gray matter and white matter, is approximately 50 mL per 100 g/min. Protein synthesis in neurons ceases when CBF falls below 35 mL per 100 g/min [195]. At a CBF ≤20 mL per 100 g/min, however, electrical failure occurs and synaptic transmission between neurons is disturbed, leading to loss of function of still viable neurons [196,197,198]. Metabolic failure and cell death occur at CBF ≤10 mL per 100 g/min [11]. CBV is defined as the amount of blood in a given quantity of brain tissue. Normal CBV is approximately 4–5 mL per 100 g [199, 200]. CBV can be decreased or increased during cerebral ischemia, depending on the efficacy of cerebral autoregulation and patency of collateral arterial pathways [200,201,202,203].

1.5 CT Perfusion Technique

1.5.1 Parameters

CT perfusion produces the following data:

• 1. Cerebral blood flow (CBF), measured in mL per 100 g of brain tissue per minute (mL/100 g/min) or as mL per 100 mL of brain tissue per minute (mL/100mL/min).

• 2. Cerebral blood volume (CBV), measured in mL/100 g or mL/100 mL.

• 3. Time to peak (TTP) is defined as the time delay (in seconds) between the first arrival of contrast within major arteries included in the section imaged and the peak attenuation of the brain tissue.

• 4. Mean transit time (MTT) indicates the time (in seconds) required for contrast material to pass from the arterial side to the venous side of the intracranial circulation. Blood and intravascular contrast material pass through vascular pathways of varying length and complexity in the brain’s vascular network. The average of all of these possible transit times is MTT.

  • (a) TTP and MTT are parameters unique to CBF techniques (e.g., CT perfusion and MRI perfusion) that utilize an intravascular indicator and track the passage of the indicator through the brain over the course of time to determine CBF.

1.6 Concepts

There are two commonly applied methods of CT perfusion (Table 8.4). These methods, known as the first-pass bolus tracking techniques, are based on the indicator dilution principle and provide information about CBF, CBV, MTT, and TTP. A known amount of a nondiffusible tracer (e.g., iodinated contrast material) is injected into an antecubital vein, and its concentration is repeatedly measured during its first pass through an intracranial vessel. Contrast transit through the intracranial vasculature produces a transient change in brain tissue attenuation. This change is linearly proportional to the serum concentration of the contrast agent. With helical CT scanning, these changes can be graphed as a time-attenuation curve for every voxel in a CT imaging slice.

Table 8.4 CT perfusion methods

Two different mathematical approaches are commonly used to calculate CT perfusion data from the time-attenuation curve: deconvolution and maximum slope. With deconvolution methodology, the attenuation values of an artery in the field of view (the arterial input function), such as the anterior cerebral artery, are integrated with time-attenuation information of the brain tissue on a voxel-by-voxel basis in a mathematical operation called deconvolution. In mathematical terms:

$$ {C}_t(t)=\mathrm{CBF}\cdot \left[{C}_a(t)\otimes R(t)\right] $$

where C_t(t) is the tissue time-attenuation curve; C_a(t) is the arterial time-attenuation curve; R(t) is the impulse residue function; and ⊗ is the convolution operator. The impulse residue function is an idealized tissue time-attenuation curve that would result if the entire bolus (the impulse) of contrast material was administered instantaneously into the artery supplying a given area of the brain. The plateau of impulse residue function reflects the length of time during which the contrast material (the residue) is passing through the capillary network. Both C_t(t) and C_a(t) can be measured, and the deconvolution process uses the information to calculate CBF and CBV. MTT is then derived by using the central volume principle, which relates CBF, CBV, and MTT in the following relationship:

$$ \mathrm{CBF}=\mathrm{CBV}/\mathrm{MTT} $$

The accuracy of this method depends upon an intact blood–brain barrier, as leakage of the contrast material out of the intravascular space can lead to artifactually high perfusion parameters. Accuracy can also be influenced by the choice of the reference artery [204] and recirculation of contrast material. The venous output function serves as a reference against which the CTP parametric values are normalized and scaled. Since CBV values are affected by the choice of the venous output function, the chosen ROI for the venous output function should include the voxel demonstrating the maximum area under the time-attenuation curve and the least amount of partial volume averaging [205].

In the maximum slope method, the maximum slope of the time-attenuation curve is used to calculate CBF (Fig. 8.6) [206,207,208]. Values for CBV are calculated from the maximum-enhancement ratio, which is the maximum enhancement of the time-attenuation curve in a given voxel compared to that of the superior sagittal sinus [206, 209, 210]. Software using this method reports TTP rather than MTT. The accuracy of this method depends on a rapid bolus injection of contrast material, because a delay in the appearance of contrast material in the intracranial vasculature will lead to a decrease in the maximum slope of the time-attenuation curve, and CBF will be underestimated [206, 211].

Fig. 8.6

Maximum slope method. In CT perfusion with the maximum slope method (Siemens), regions of completed infarction show up as a “black hole”—dense, black areas on CBF, CBV, and TTP images (black arrow). Adjacent areas of abnormality that are not black (white arrows) indicate regions of salvageable tissue. The CT done 2 weeks later shows the black hole region as a completed infarction

1.6.1 Validation

Quantitative CBF measurement by CT perfusion has been validated by comparison to other techniques for measuring CBF such as microspheres; [212], xenon CT [213, 214], and PET [215, 216]. CT perfusion imaging using the deconvolution technique has been shown to demonstrate little variability within individuals [215]. The use of CT perfusion in the identification of cerebral ischemia has been validated in experimental models [217,218,219]. Further validation of CT perfusion in human subjects by comparison with other brain imaging techniques in the setting of acute stroke has been extensive and is discussed below.

1.6.2 Limitations

CT perfusion imaging has several practical limitations. Brain regions close to the skull are difficult to image because of bone artifact. A peripheral IV is required for intravenous administration of the contrast material, which can be a nuisance for some intensive care unit patients. The study requires iodinated contrast, which can be problematic in patients with renal insufficiency or contrast allergy.

An important limitation concerns the use of an intravascular indicator in first-pass CT perfusion methods. As opposed to older techniques like xenon-CT and PET, in which diffusible tracers are used and only capillary perfusion is measured, all intracranial vessels are included in CT perfusion. This difference leads to an overestimation of CBF in regions that include large vessels, such as around the Sylvian fissure [220]. Moreover, this aspect of CT perfusion makes it difficult to compare CT perfusion results to CBF values obtained by the use of other methods. This situation may be ameliorated by vessel removal using threshold-based segmentation algorithms [205, 216]. Finally, variability in quantification between different CT perfusion post-processing software packages limits the ability to generalize parametric thresholds (e.g., CBF threshold representing the infarct core) between platforms [221].

1.7 Interpretation of CT Perfusion Data

Validity has been demonstrated for the commonly used mathematical techniques for CT perfusion by comparison with other CBF measurement techniques. However, each method has inherent limitations and sources of systematic error, hence the description of CT perfusion as being “semiquantitative” by some authors [222, 223]. Assessment of cerebral perfusion based on absolute values for CBF and CBV should be made with caution [224, 225].

In CT perfusion using the deconvolution method, some have found that MTT values >145% the contralateral hemisphere correlate best with tissue at risk for infarction in cases of persistent arterial occlusion, compared to DWI/FLAIR MRI [226]. Using the maximum slope method, a reduction of CBV of 60%, compared to nonischemic regions of the brain, best identified cerebral ischemia [223].

Mean transit time is prolonged in regions of cerebral ischemia. In a series of patients with acute ischemic MCA stroke, Eastwood and colleagues found average MTT to be 7.6 s in the affected MCA territories and 3.6 s in the unaffected MCA territories [227]. Areas of reduced perfusion were defined as MTT >6 s because that value represented at least three standard deviations greater than the average MTT values in unaffected MCA territories.

TTP is typically <8 s in normal brain tissue with unimpaired antegrade flow. In ischemic regions, TTP is prolonged, reflecting delayed tracer arrival through alternative pathways such as leptomeningeal vessels. TTP maps are useful for accurate identification of areas of impaired perfusion [223]. A regional TTP >8 s raises the suspicion of cerebral ischemia. However, TTP maps can provide false-positive findings when TTP is prolonged due to carotid stenosis or occlusion and regional CBF is compensated for by collateral vessels [228].

Both MTT and TTP maps can be used to identify cerebral ischemia. MTT maps offer advantages over CBF and CBV maps. MTT appears to be affected by ischemia at an earlier stage than CBF or CBV, although it is less specific [229]. Color-coded TTP and MTT maps appear to demonstrate regions of cerebral ischemia more readily than CBF and CBV maps. TTP and MTT are usually homogenous in normal areas of brain tissue, permitting easy identification of abnormal hemodynamics [229]. Moreover, CBF and CBV data are overestimated when the ROI includes major vessels, such as MCA branches [216]. In comparison, TTP and MTT do not seem to be influenced by the presence of large vessels within ROIs. The absence of regions of extended TTP or MTT is usually a reliable indication that ischemia is not present.

1.8 CT Perfusion in Ischemic Stroke

CT perfusion can be done at the same time as the initial screening CT scan in patients with acute ischemic stroke and can distinguish viable tissue from regions of completed infarction.

• 1. CT perfusion can be used to exclude poor candidates for thrombolysis, such as patients with lacunar strokes and patients with no arterial occlusions, which account for up to 25% [230] and 29% [231] of patients with acute stroke, respectively.

• 2. CT perfusion imaging can provide prognostic information because patients with profound, widespread ischemia can be expected to have poorer outcomes than those with borderline ischemia [229, 232].

CT perfusion can potentially identify salvageable tissue at risk of infarction [229, 233]. Using the deconvolution method, a mismatch between regional MTT, CBF, and CBV maps can indicate the presence of ischemic but potentially salvageable brain (penumbra) [217, 222, 228, 234]. Studies attempting to define the parameter that best identifies the infarct core have yielded different results. In a series of patients with acute stroke studied by Wintermark and colleagues, CBV <2.0 mL per 100 g best identified the irreversibly injured infarct core. Regions demonstrating MTT >145% compared to mirror-image voxels in the contralateral hemisphere optimally conformed to the ischemic region (infarct core + penumbra) [226]. A more recent study found that CBF <31% of the mean contralateral hemispheric CBF best predicted infarct core [235].

“Prognostic maps” co-demonstrating the ischemic zone (e.g., MTT > 145% contralateral mirror image voxel values = core + penumbra) in green and the infarct core (CBV <2.0 mL per 100 g) in red can be generated to provide an at-a-glance image of these parameters (Fig. 8.7) [236].

Fig. 8.7

Deconvolution method. In deconvolution CT perfusion (General Electric, Philips), the “black hole” technique is not a reliable way to identify regions of completed infarction. Threshold maps, however, can provide the same information. Here, the ischemic core (dark threshold area—red on the color image—black arrow) is defined as absolute CBV <2 mL per 100 g, and the penumbra (light threshold area—green on the color image—white arrows) is defined as the region of brain with MTT values 1.45 times the MTT values in the corresponding area of the opposite hemisphere [226]. The follow-up CT shows an infarction that corresponds to the ischemic core region

Using the maximum slope method, the relative values of CBF and CBV can be used to distinguish infarcted from ischemic tissue. In a series of patients undergoing CT perfusion studies within 6 h of stroke onset, the thresholds for best discrimination between infarcted and non-infarcted tissue were 48% of normal values for CBF, and 60% of normal values for CBV [223]. The lowest relative CBF and CBV values among brain regions not developing infarctions were 29% and 40% of normal values, respectively.

The Rapid software [237] is an automated technique to generate ischemic threshold maps. Based on results from the SWIFT PRIME trial, the following regional thresholds provide the most accurate prediction of infarct volume (values are expressed as the fraction of the values of the opposite, unaffected hemisphere [i.e., >70% reduction in regional CBF = rCBF < 0.3]): [238]

• 1. rCBF: 0.30–0.34

• 2. rCBV: 0.32–0.34

1.9 Validation of CT Perfusion in Ischemic Stroke

The deconvolution method has been validated in the diagnosis of acute ischemic stroke by comparison to CT imaging [227] and to MR T2-weighted imaging [227], diffusion imaging [229, 236], and perfusion imaging [227]. In a series of patients with acute ischemic stroke, undergoing both CT perfusion and MRI diffusion studies on admission, infarct size on CBF maps correlated highly with the size of the abnormality on the diffusion-weighted imaging (DWI) map (r = 0.968) [236]. Similarly, infarct size assessed by CT perfusion studies done on admission in patients with ischemic stroke correlated highly with infarct size measured by follow-up MRI-DWI maps obtained an average of 3 days after admission (r = 0.958) [229]. However, a recent study of treated patients who had complete early reperfusion found that CTP prognostic maps were not predictive for irreversibly or reversibly lost neurologic function [239].

The maximum slope method has been validated in acute stroke by comparison to CT, MRI, and SPECT [228]. In a series of patients with acute stroke, who underwent both CT perfusion and SPECT studies on admission, the areas of ischemia indicated by CT perfusion CBF maps correlated well with those indicated by SPECT imaging (r = 0.81) [208]. In a series in which ischemic areas on admission of CT perfusion images were compared to the follow-up CT or MR images showing final infarctions, infarction was found to develop in all patients with >70% CBF reduction and in 50% of patients with 40–70% CBF reduction [228]. Based on a threshold of CBF <60% (compared with CBF in normal vascular territories), CBF maps predicted the extent of infarction with high sensitivity (93%) and specificity (98%). Similarly, TTP >3 s predicted infarction with a sensitivity of 91% and a specificity of 93%. Notably, in the same study, a negative predictive value for TTP >3 s of 99% was found, indicating that the absence of extended TTP is usually accurate in excluding the presence of ischemia. In a series of CT perfusion studies done in patients with acute stroke <6 h after onset, and compared to the follow-up CT or MRI, threshold values of 48% and 60% of normal, for CBF and CBV, respectively, were found to discriminate best between the areas of infarction and the areas of non-infarction [223].

1.10 MRI

Magnetic resonance imaging is based on the interaction between a powerful, uniform magnetic field, radiofrequency (RF) energy, and body tissues. Protons absorb energy from pulsed RF waves (excitation) and are thereby deflected from their alignment with the main magnetic field. As the nuclei return to rest, energy is released and signals are induced in a receiver and converted into diagnostic images. During the process of energy release, spatially encoded voxel-specific relaxation constants can be obtained and, in conjunction with Fourier transform reconstruction, used to construct images that demonstrate specific tissues. A wide array of MRI imaging sequences are available (Table 8.5). Most MRI images accentuate T1 or T2 relaxation; T1 is longitudinal, or spin-lattice relaxation time and T2 is transverse, or spin-spin relaxation time. In T1-weighted images, fat has increased signal (short T1 relaxation) and water appears dark (long T1 relaxation). In T2-weighted images, water has increased signal relative to brain (long T2 relaxation). Brain tissue water content is typically increased in regions of edema, ischemia, and hemorrhage, thus changing the appearance of the tissue on MRI. T2-weighted images usually show only tissue changes caused by severe and prolonged ischemia—apparent only after some 6–24 h following stroke onset—and are therefore not optimal for evaluating acute ischemia. The hyperintense acute reperfusion marker (HARM) on FLAIR is a sign of early blood–brain barrier disruption and is caused by leakage of gadolinium into the CSF [240].

Table 8.5 MRI signal characteristics of cerebral infarction

1.11 Diffusion-Weighted Imaging

Diffusion-weighted imaging (DWI) measures the Brownian motion of water protons in tissue. Normal random motion of water protons leads to a loss of signal on diffusion-weighted images. Ischemic failure of the ATP-dependent sodium-potassium cellular membrane pumps leads to water migration from the extracellular space to the intracellular space. Random water proton motion in the constricted extracellular space is reduced. Severely ischemic brain tissue appears bright on DWI due to signal preservation in these areas of decreased Brownian motion. These changes occur within minutes after ischemic stroke (Fig. 8.8) [241, 242]. Areas of ischemia appear bright on DWI in the acute phase and become unapparent or dark after about 2 weeks. DWI images are superior to CT and conventional MRI in the detection of acute ischemia [172]. Sensitivities of 88–100% for acute stroke detection with DWI have been reported, with specificity from 86 to 100% [243]. Analyzed pooled data from several studies yielded a PPV of 100% and a NPV of 90.6% [244]

Fig. 8.8

Patterns of acute ischemic stroke on MRI. Diffusion-weighted images of an embolic stroke (A); an arterial border zone (aka watershed territory or “rosary” pattern) stroke (B); a large artery (MCA) stroke (C); a lacunar stroke (D) and posterior circulation territory embolic stroke (E)

Diffusion-weighted images are influenced by other parameters including spin density and T1 and T2 relaxation effects. Calculation of the apparent diffusion coefficient (ADC) eliminates these influences and provides pure diffusion information [245]. Two otherwise identical image sets are obtained, one with a low (but nonzero) b value and one with a b value = 1000 s/mm². The natural logarithm of signal intensity vs. b value for these two values is plotted and the slope of this line is used to determine ADC for each voxel in the image [246]. The resulting map demonstrates the calculated ADC for each pixel in the image, with signal intensity proportional to the magnitude of the ADC. Areas of acute infarction (restrained diffusion) appear bright on diffusion-weighted images and have low ADC values (appear dark) on ADC maps. Later, in the subacute phase, signal on diffusion-weighted images within infarcted areas may appear bright due to T2 shine-through, but correlation with the ADC maps will demonstrate that this is a T2 effect, and does not reflect true diffusion restraint. After about 2 weeks, diffusion becomes facilitated. Signal in the infarcted area decreases on DWI and increases on ADC maps as a result. Decreased ADC values indicate with good sensitivity (88%) and specificity (90%) that an infarct is less than 10 days old [247]. Venous infarctions, in contrast, cause an increase in ADC values in the acute phase because of vasogenic edema, although in later stages the ADC map appearance becomes complex because of the coexistence of cytotoxic and vasogenic edema and the presence of hemorrhage [248].

Diffusion-weighted imaging can be useful in the workup of patients with TIA [249]. “Dots of hyperintensity” on DWI, indicating microinfarctions that are too small to cause permanent neurological symptoms, are found in some 40–50% of patients with traditionally defined TIA [250]. This has led to a recommendation for a change of the definition of TIA from a clinical to a tissue-based definition, specifically: “A transient episode of neurological dysfunction caused by focal brain, spinal cord, or retinal ischemia, without acute infarction.” [251] Patients with clinically transient neurological events in whom asymptomatic diffusion abnormalities are discovered have a high risk of early completed stroke [252].

Diffusion-weighted imaging and ADC maps are dynamic. Areas of ischemic injury may enlarge by 43% in the first 52 h after onset [253], although in most patients lesion size appears to reach a maximum by 24 h [254]. Conversely, all areas demonstrating DWI hyperintensity and ADC map hypointensity may not necessarily be infarcted, as bright regions on DWI can be reversed by reperfusion. In a series of patients with acute stroke, 19.7% demonstrated “normalization” of ADC abnormalities after reperfusion [255]. Tissues with 75–90% of ADC values in normal brain are likely to proceed to infarction, whereas tissues with ADC values >90% of normal are more likely to recover [256]. Nevertheless, DWI hyperintensity is a necessary stage on the path to infarction [249], and the volume of DWI abnormalities does correlate with clinical severity [253, 257].

1.12 Perfusion Imaging

MRI perfusion imaging employs a first-pass tracking technique and deconvolution method for calculating the brain perfusion parameters. MRI perfusion uses the same deconvolution technique as CT perfusion, which is described in detail above. In MRI perfusion, a bolus of gadolinium is injected rapidly into a peripheral vein, and tissue- and arterial-input curves are used to generate CBF, CBV, TTP, and MTT images. The information is not quantitative because the MR signal change after IV administration of gadolinium is not proportionally related to the plasma concentration of gadolinium. MRI perfusion is subjected to many of the limitations of CT perfusion, such as the dependence of lesion volume on arterial input function selection [258], and controversy about the optimal perfusion parameters for the identification of affected tissue [259]. The value of MRI perfusion imaging lies in the perfusion–diffusion mismatch hypothesis. This holds that abnormal regions on perfusion images which appear normal on diffusion-weighted imaging are equal to the penumbra, and represent potentially salvageable tissue. A perfusion–diffusion mismatch pattern is present in some 70% of patients with anterior circulation stroke scanned within 6 h of onset [260], is strongly associated with proximal MCA occlusion [260], and resolves on reperfusion [261, 262]. A recent study reported low favorable clinical responses and high mortality rates in a small group of patients (N = 8) with matched perfusion and diffusion abnormalities, especially those with large infarctions [263], who underwent attempted intra-arterial thrombolysis.

The penumbra on perfusion imaging has been defined as regions where DWI is normal and TTP >4 s [264], although, for practical purposes, any region that is abnormal on perfusion imaging but normal on DWI may represent salvageable tissue [265]. Among MRI perfusion parameters, CBF, MTT, and TTP appear to best identify all affected tissue (and thereby distinguish penumbra when compared to DWI) [266, 267], and CBV seems to best predict the final infarct volume [268]. Compared to final infarct volumes imaged on MRI, the sensitivities of CBF, CBV, and MTT for detection of perfusion abnormalities were 84%, 74%, and 84%, respectively, and the specificities were 96%, 100%, and 96%, respectively [268].

Together, perfusion imaging and DWI can identify tissue that is at the risk of infarction but amenable to salvage with revascularization [269]. In a series of patients receiving IV thrombolytics for acute ischemic stroke and imaged both before and after 2 h of treatment, 78% of patients had complete resolution of perfusion lesions and 41% had resolution of DWI lesions [270]. Perfusion–diffusion imaging has been used in clinical trials to select patients for thrombolysis. Intravenous desmoplasia was given only to patients with a DWI-PWI mismatch ≥20%, and the drug was found to be potentially effective in improving clinical outcomes [271, 272].

1.13 MR Angiography

MRA techniques fall into three categories:

• 1. Time of flight: Very common MRA technique.

  • (a) Depends on a strong signal from blood flowing into a plane where stationary tissue signal has been saturated.

    Advantage: No contrast agent is used.

    • (i) Disadvantages: Spin dephasing in areas of turbulent flow or magnetic susceptibility (near paramagnetic blood products, ferromagnetic objects, and air/bone interfaces) causes signal loss that may lead to overestimation of stenosis.

• 2. Phase contrast: Not often used.

  • (a) Image contrast from the differences in phase accumulated by moving spins in a magnetic field gradient. Stationary spins accumulate no net phase.

    • (i) Advantages: No contrast agent is used. Less likely to confuse fresh clot for flowing blood as it is strictly flow dependent.

    • (ii) Disadvantages: Acquisition times are relatively long.

• 3. Contrast-enhanced MRA: Common MRA technique.

  • (a) Based on a combination of rapid 3D imaging and the T1-shortening effect of IV gadolinium.

    • (i) Advantages: High signal-to-noise ratios, robustness irrespective of blood flow patterns or velocities, and fast image acquisition, allowing for the evaluation of larger anatomic segments (from the aortic arch to the circle of Willis).

    • (ii) Disadvantage: Requires IV gadolinium, which carries a small risk of complications, particularly in patients with renal insufficiency (see below).

Gadolinium and Nephrogenic Systemic Fibrosis

Gadolinium is a chemical element with an atomic number of 64. It has seven unpaired electrons in its outer shell which hasten T1 relaxation and increase signal in the area of interest. Gadolinium alone is toxic, but not when combined with a chelating agent. Several FDA-approved gadolinium preparations are available. A study of high-dose gadolinium administration in a population with a high prevalence of baseline renal insufficiency showed no renal failure associated with its administration [273]. The rate of anaphylactic reactions is also very low; in a survey of >700,000 patients receiving gadolinium, the rate of serious allergic reactions was <0.01% and most reactions were limited to mild nausea or urticaria [274].

Nephrogenic systemic fibrosis (aka nephrogenic fibrosing dermopathy) is strongly associated with gadodiamide (Omniscan^TM; GE Healthcare, Princeton, NJ) [275, 276]. Although most patients have a history of exposure to gadodiamide, other gadolinium-based agents have been implicated [277]. It appears to occur only in patients with renal insufficiency, generally in those requiring dialysis [277], and is dose dependent [276]. The condition consists of thickening and hardening of the skin of the extremities, due to increased skin deposition of collagen. The condition may develop rapidly and result in wheelchair dependence within weeks. There may also be involvement of other tissue such as the lungs, skeletal muscle, heart, diaphragm, and esophagus [278]. The mechanism is not understood. An estimate of the incidence of this syndrome comes from an Internet-based medical advisory originating in Denmark, which reported that, of about 400 patients with severely impaired renal function, 5% were subsequently diagnosed with nephrogenic systemic fibrosis [279].

Management consists of correction of renal function (usually dialysis), which may result in a cessation or reversal of symptoms [280].

1.14 Identification of Hemorrhage on MRI

MRI is as sensitive for acute hemorrhage as CT [169]. The appearance of intracranial hemorrhage changes with time as the hemoglobin moiety changes from non-paramagnetic oxyhemoglobin through the paramagnetic forms (deoxyhemoglobin, methemoglobin, and hemosiderin). Subacute blood appearing hyperintense on T1-weighted images is in the methemoglobin form. The characteristic T1 shortening here is due to a phenomenon known as “dipole-dipole relaxation enhancement (PEDDRE).” T2 shortening (and the associated signal loss) depends on the presence of an intact cell membrane sequestering paramagnetic hemoglobin moieties from the extracellular space and thereby establishing a local magnetic gradient. Red blood cells usually undergo lysis in the subacute phase (i.e., methemoglobin) of parenchymal hemorrhage evolution. Early in the pre-lysis phase, blood appears bright on T1 (PEDDRE) and dark on T2-weighted images (paramagnetic effect). After RBC lysis, methemoglobin-dominant hematomas still appear bright on T1 (again, PEDDRE), but become bright on T2-weighted images due to disruption of the paramagnetic effect by RBC lysis. Deoxyhemoglobin (acute) and hemosiderin (chronic) share similar appearances on T1 (isointense to gray matter) and T2 (hypointense to gray matter) MRI. However, acute hemorrhage is typically associated with vasogenic edema, while chronic hemorrhage is not (the latter may be associated with cavitation, gliosis, and focal atrophy). The recurrence of T2 shortening in chronic (hemosiderin) hematomas long after RBC lysis is due to the ingestion of hemosiderin by macrophages [281].

Acute hemorrhage characteristics on MRI are summarized in Table 8.6. Susceptibility-weighted MRI can help identify acute cerebral hemorrhage, microbleeds, and intravascular clot [282]. Asymptomatic microbleeds are caused by hypertension and amyloid angiopathy, and are found in up to 6% of elderly patients and 26% of patients with prior ischemic stroke [249]. The finding of microbleeds in patients with acute ischemic stroke may predict an increased risk of hemorrhage transformation after thrombolysis. In a study of patients undergoing IA thrombolysis for acute ischemic stroke, microbleeds were found in 12% of patients prior to treatment [283]. Symptomatic hemorrhages occurred in 20% of patients with an evidence of prior microbleeds, compared to 11% of patients without prior microbleeds. The Bleeding Risk Analysis in Stroke Imaging Before Thrombolysis (BRASIL) study found that the risk of intracranial hemorrhage attributable to microbleeds was small and unlikely to exceed the benefits of thrombolytic therapy. This study could not draw conclusions about the risk of hemorrhage in patients with multiple microbleeds, however [284].

Table 8.6 MRI signal characteristics of cerebral hemorrhage

Appendix 8.2: NIH Stroke Scale

The National Institutes of Health Stroke Scale (NIHSS) is widely used and it provides important prognostic information [285,286,287,288].

A detailed description of the NIHSS can be downloaded at www.ninds.nih.gov/disorders/stroke/strokescales.htm.

Higher scores indicate greater stroke severity (Tables 8.7 and 8.8). A score of ≥16 predicts a high probability of death or severe disability whereas a score of ≥6 predicts a good recovery [288]. Some 60–70% of acute ischemic stroke patients with a baseline NIHSS score <10 will have a favorable outcome after 1 year, compared to only 4–16% of patients with a score >20 [289].

Table 8.7 NIH stroke scale

Table 8.8 NIH stroke scale score severity