Imaging and Treatment of Patients with Acute Stroke: An Evidence-Based Review

CT.

Although it is accepted that CT is sensitive in detecting acute ICH, surprisingly few studies have been conducted to support this belief.¹ One of the original studies on a first-generation scanner in the 1970s identified 66 hemorrhages with CT, but the data lacked postmortem confirmation. Subsequently, in a postmortem series of 79 patients, in 4 of 17 patients, CT missed ICHs—all of which were located in the brain stem.² While there is little doubt that the third-generation scanners are far superior today, the precise accuracy of CT for the detection of ICH is unknown.

MR Imaging.

MR imaging was reported to detect ICH within 6 hours and as early as 23 minutes from symptom onset.^3,4 One study evaluated 62 patients with ICH within 6 hours of onset and 62 controls, with 3 experienced readers using CT as the reference standard.⁵ The readers, blinded to clinical and CT results, identified all acute hemorrhages on MR imaging, yielding 100% sensitivity and specificity. A subsequent study comparing CT and MR imaging for detection of hemorrhagic stroke within 6 hours of onset, using the discharge diagnosis as the reference standard, found 86% sensitivity and 100% specificity for both CT and MR imaging. Of 29 acute hemorrhages on CT, 25 were identified on MR imaging (in 3 of 4 cases, blood was misclassified as chronic instead of acute).⁶ Similarly, 4 cases of ICH identified on MR imaging were not seen on CT (all were ischemic infarcts with hemorrhagic transformation). Therefore, it appears that rare cases of early intracranial hemorrhage may be missed on either CT or MR imaging. However, susceptibility-weighted sequences have improved sensitivity in the detection of cerebral microbleeds that are not otherwise detected on CT. Thus far, studies suggest that neither the presence nor number of microbleeds is associated with an increased risk of hemorrhagic transformation in either tPA-treated or untreated patients.^7⇓–9

Summary of Evidence.

In the evaluation of patients with acute stroke for IV thrombolysis, CT has become the technique of choice for exclusion of ICH, based on randomized controlled trials (strong evidence). Recent studies indicate that the accuracy of MR imaging in detecting ICH is likely equivalent to that of CT even in the hyperacute setting (within 6 hours of ictus) and is even more accurate for detection of chronic cerebral microbleeds (moderate evidence).

CT.

The diagnosis of acute stroke is primarily based on the clinical history and physical examination. At times, patients may present with strokelike symptoms due to nonstroke etiologies. Stroke mimics may account for up to 19% of patients initially diagnosed with stroke and up to 14% of patients treated with IV tPA.^10,11 An additional goal of imaging is to confirm diagnosis and to evaluate the extent of ischemic changes. Large ischemic strokes may demonstrate early signs of ischemia on CT, including loss of gray-white differentiation, hyperattenuated clot in the proximal MCA, sulcal effacement, and local mass effect. Such signs were found in 31% and 81% of patients within 3 and 5 hours after symptom onset, respectively.^12,13 Early CT signs involving more than one-third of the MCA distribution have been associated with large strokes, increased risk of hemorrhagic transformation, and poor outcome.^14⇓–16 The Alberta Stroke Program Early CT Score (ASPECTS), a 10-point scoring system to detect early ischemic changes on NCCT, was developed as a tool that would be more reliable than the MCA rule.¹⁷ While ASPECTS was found to have superior interobserver agreement, it only modestly improved accuracy for predicting functional outcome and performed the same as the MCA rule for predicting symptomatic hemorrhage.^17,18

MR Imaging.

DWI is more sensitive for detecting ischemic changes within minutes of stroke onset and ischemic lesions as small as 4 mm in diameter compared with CT.^19⇓–21 As time from symptom onset increased, the sensitivity of DWI for the diagnosis of ischemic stroke also increased 73%, 81%, and 92% for <3 hours, 3–12 hours, and >12 hours, respectively, whereas CT had only 12%, 20%, and 16% sensitivity at these respective time intervals. The sensitivity for MR imaging to detect ischemia within 6 hours of onset is 81%–91%; therefore, the absence of a DWI lesion does not completely exclude ischemic stroke. DWI may be falsely negative in very early stroke and in patients with small subcortical or vertebrobasilar infarctions or in patients with low stroke severity (NIHSS score < 4).^{21⇓⇓–24} Therefore, stroke timing, location, size, and perfusion status contribute to the visualization of a DWI lesion.

Summary of Evidence.

DWI is superior to CT for identifying acute ischemic stroke within the first 12 hours of symptom onset (strong evidence). However, MR imaging does not necessarily have a greater influence than CT on decision-making or clinical outcomes in the acute setting, given its limited availability and the need for patient screening. Early CT signs of ischemia involving more than one-third of the MCA distribution have been associated with large strokes, increased risk of hemorrhagic transformation, and poor outcome (strong evidence).

MR Imaging.

An emerging goal of acute stroke imaging is to identify patients who may benefit from treatment outside the current therapeutic window for IV thrombolysis. Thus, identifying tissue that may be rescued with therapy (ie, the ischemic penumbra) has been a major focus of clinical research. Using multimodal MR imaging, DWI is postulated to delineate irreversibly injured tissue (infarct core), and PWI is postulated to delineate critically hypoperfused tissue that will evolve into infarction if not reperfused in a timely manner (ischemic penumbra). When DWI and PWI are superimposed, the nonoverlapping “mismatched” region has been proposed to represent the penumbra.^25,26 Consistent with the idea of an evolving ischemic core, the DWI lesion may grow into a final infarct approximating the early PWI abnormality.²⁷ The DWI-PWI mismatch has been rapidly translated into clinical trials under 3 categories of study design: 1) to select patients for enrollment in the therapeutic trial, 2) to validate DPM as a biomarker of the penumbra, and 3) both (Table 2).

Within the first category, the Desmoteplase in Acute Ischemic Stroke trial (DIAS),²⁸ the Dose Escalation for Ischemic Stroke trial (DEDAS),²⁹ and the DIAS-II³⁰ have been completed. These trials examined a novel thrombolytic agent, desmoteplase, compared with placebo in patients demonstrating DPM between 3 and 9 hours after stroke onset. While the initial phase-2 studies, DIAS and DEDAS, demonstrated safety and early signs of efficacy for desmoteplase, the phase-3 study, DIAS-II, showed unexpectedly high mortality and no overall clinical benefit. Unfortunately, these negative results did not permit an analysis of which component of the study failed—the therapeutic agent or the DPM criteria.

Within the second category, the Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution (DEFUSE) trial aimed to validate DPM by comparing clinical outcomes in patients with DPM and reperfusion with outcomes in patients without DPM and reperfusion; patients in both groups were treated with tPA between 3 and 6 hours of stroke onset.³¹ Of 74 patients enrolled, 75% had DPM and 25% did not. In patients with DPM, reperfusion was associated with better clinical outcome than no reperfusion. However, due to the small number of patients without DPM (n = 11), the converse that nonmismatch patients did not improve with reperfusion could not be proved. The high ratio of mismatch to nonmismatch patients made it clear that the definition for DPM was too liberal, resulting in a relatively unselected group of patients with stroke.

On the basis of the DEFUSE results, the DPM definition was refined for the DEFUSE-2 study.^32,33 However, a subset of patients with DPM with “target mismatch” was selected, which excluded patients with a “malignant profile” (large baseline DWI and PWI lesions) because DEFUSE indicated that these patients have worse outcomes. Thus far, selection of patients with the “target mismatch” definition demonstrated improved clinical outcome and reduced infarct growth compared with the nontarget mismatch group.

Studies giving information about both the utility of the imaging biomarker and the efficacy of treatment include the Echo-Planar Imaging Thrombolytic Evaluation Trial (EPITHET)³⁴ and MR and Recanalization of Stroke Clots Using Embolectomy (MR-RESCUE).³⁵ In EPITHET, patients with stroke after 3 hours from symptom onset were serially imaged at 3 time points: 1) before administration of tPA versus placebo, 2) at 3–5 days, and 3) at 90 days after treatment. With its similarity to the DEFUSE protocol, most (86%) of enrolled patients were categorized as having DPM, making conclusions regarding nonmismatch patients difficult. The primary end point showed no difference in geometric mean infarct growth between the tPA and placebo groups. MR-RESCUE,³⁵ an ongoing study, aims to test both a similar DPM protocol in combination with treatment by using the Merci retriever (Concentric Medical, Mountain View, California) versus placebo up to 8 hours after stroke onset.

Taking advantage of the endogenous susceptibility of deoxyhemoglobin on T2*-weighted images is an active investigation into imaging oxygen metabolism by using MR imaging (eg, MR-oxygen metabolic index),^36⇓–38 a parameter similar to PET-derived cerebral metabolic rate of oxygen. If these physiologic measures of oxygen metabolism accurately delineate the ischemic penumbra, they may yield ischemic thresholds that are time-invariant,³⁸ unlike diffusion and perfusion thresholds, which vary with elapsed time between stroke onset and imaging.^19,39 These relatively new imaging approaches will require validation in well-designed trials.

CT.

Similar to PWI, CTP is capable of providing maps of tissue perfusion. In a prospective multicenter study, patients with acute stroke were imaged <12 hours from stroke onset with CT and MR imaging.⁴⁰ The CTP parameter most accurately reflecting the ischemic core (compared with DWI) was absolute CBV < 2 mL/100 g, while the parameter most accurately reflecting the penumbra was a relative mean transit time of >145% of the contralateral hemisphere. However, in more recent studies, relative CBF was found to be more predictive of the ischemic core and final infarct volume than absolute CBV.^41⇓–43 Recently, CTP mismatch with PWI used for both core (CBF) and penumbra (MTT, CBV) in combination with intracranial vessel occlusion on CTA was used to select patients for entry into a randomized controlled trial of IV tenecteplase versus IV tPA administered within 6 hours of stroke onset.⁴⁴ This trial was positive for its co-primary end points, improvement in the NIHSS score and percentage reperfusion at 24 hours. However, it is currently unknown whether tenecteplase is superior to tPA in unselected patients because treatment without CTA and CTP screening would save time and perhaps broaden the applicability of tenecteplase.

Summary of Evidence.

Determination of tissue viability by using advanced imaging has the potential to individualize therapy and extend the therapeutic time window for some patients with acute stroke. Several imaging modalities, including MR imaging, CT, and PET, have been examined in this role. Operational hurdles have limited the use of some of these modalities in the acute stroke setting (eg, PET), while others such as MR imaging have been studied in large clinical trials. Randomized controlled trials have not demonstrated a benefit of thrombolytic treatment in patients who are selected by using MR imaging–based criteria such as DPM (moderate evidence); however, studies are ongoing. Clinical trials with positive results will be needed before the use of penumbral imaging techniques in routine clinical decision-making.

CTA.

An advantage of CTA is that it can be performed immediately after the prerequisite NCCT in patients with stroke. In several small case series, the sensitivity and specificity of CTA for trunk occlusions of the circle of Willis were 83%–100% and 99%–100%, respectively, compared with DSA.^{45⇓⇓–48} A study with 2 blinded observers comparing CTA with DSA measured 475 short segments of intracranial arteries in 41 patients.⁴⁷ For detection of ≥50% stenosis, CTA had 97.1% sensitivity and 99.5% specificity. A similar study of 672 vessel segments in 28 patients found excellent sensitivity and specificity for intracranial stenosis of 98% and 98%, respectively, with 100% sensitivity and 100% specificity for intracranial occlusion.⁴⁸ This study also compared time-of-flight MRA with CTA and found CTA to have significantly higher sensitivity and positive predictive value compared with MRA (see below).

MRA.

While MRA appears to be a useful tool for measuring stenosis in large extracranial vessels, its sensitivity decreases for smaller caliber intracranial vessels. For time-of-flight, complete or partial signal voids in regions of high and/or turbulent flow may occur, leading to an overestimation of the extent of stenosis. Therefore, cautious interpretation of the lumen caliber is warranted. In a study of intracranial disease comparing CTA and MRA with DSA in 28 patients (in 672 vessel segments), time-of-flight MRA had a sensitivity of 70% and 81% and a specificity of 99% and 98% for intracranial stenosis and intracranial occlusion, respectively.⁴⁸ The Stroke Outcomes and Neuroimaging of Intracranial Atherosclerosis (SONIA) trial was a prospective multicenter study comparing the diagnostic accuracy of transcranial Doppler and MRA with DSA.⁴⁹ The SONIA study found that both transcranial Doppler and MRA have high negative predictive values (86% and 91%, respectively) but low positive predictive values (36% and 59%, respectively). Sensitivity and specificity could not be obtained in this study because not every patient underwent DSA.⁴⁹

Summary of Evidence.

CTA and MRA have high diagnostic accuracy for detecting large-vessel occlusion compared with DSA (strong evidence). However, evidence supporting these imaging modalities in acute stroke management is lacking (limited evidence). Prospective studies examining the accuracy of acute noninvasive vascular imaging and whether it alters clinical outcome after stroke are needed.

Summary of Evidence.

IV thrombolysis is the standard-of-care treatment for acute ischemic stroke in the anterior circulation within 4.5 hours of symptom onset (strong evidence). Larger more proximal clots (ICA, M1 MCA) exhibit lower recanalization rates with IV thrombolytic treatment, and such patients tend to have worse prognosis (strong evidence).

Summary of Evidence.

IA thrombolysis may be used at a qualified stroke center to treat patients with major stroke due to MCA occlusion in <6 hours who cannot receive IV alteplase (strong evidence). The FDA has approved several mechanical devices for intracranial clot removal in appropriately selected patients up to 8 hours from symptom onset (strong evidence). Although considered appropriate for emergency stroke treatment, the effect on outcomes has not been established.