Physiologic imaging in acute stroke: Patient selection

Abstract

Treatment of acute stroke is changing, as endovascular intervention becomes an important adjunct to tissue plasminogen activator. An increasing number of sophisticated physiologic imaging techniques have unique advantages and applications in the evaluation, diagnosis, and treatment-decision making of acute ischemic stroke. In this review, we first highlight the strengths, weaknesses, and possible indications for various stroke imaging techniques. How acute imaging findings in each modality have been used to predict functional outcome is discussed. Furthermore, there is an increasing emphasis on using these state-of-the-art imaging modalities to offer maximal patient benefit through IV therapy, endovascular thrombolytics, and clot retrieval. We review the burgeoning literature in the determination of stroke treatment based on acute, physiologic imaging findings.

Introduction

The cornerstone to acute ischemic stroke evaluation and treatment is neuroimaging. Current potential treatments, whether intravenous tissue plasminogen activator (IV-tPA), intra-arterial (IA) thrombectomy, or burgeoning alternative therapies, all rely on some form of brain imaging to ensure appropriate patients are selected, and to avoid potentially disastrous complications—i.e. the administration of IV-tPA to a patient with an intracerebral hemorrhage. As a result, neuroimaging has become an essential element of any stroke treatment regimen. The goals of neuroimaging are to, at a minimum, rule out stroke mimics such as infection or tumors and exclude hemorrhage. Additionally, many modalities are now used to delineate the extent of vessel occlusion, assess the salvageability of ischemic tissue, evaluate vessel patency, and ascertain the underlying cause of the stroke event.¹ Each neuroimaging modality has its own set of specific strengths, weaknesses, and limitations, giving each method its own specific subset of patient presentations in which it may be a suitable choice for use. We review the strengths, weaknesses, and possible indications for various stroke imaging techniques.

Pathophysiologic considerations

Within minutes after cerebral vessel occlusion, the area of the brain that is subjected to the most severe blood flow reduction undergoes necrotic cell death. The infarct core refers to the part of the brain that is irreversibly injured. Fulminant cell death within the infarct core occurs by a variety of mechanisms including oxidative stress, excitotoxicity, peri-infarct depolarization, and apoptosis. The area surrounding the infarct core, known as the penumbra, is threatened by hypoperfusion. This poorly perfused region is thus hypofunctioning, but remains metabolically viable.

Within the penumbra, hypoperfusion leads to failure of membrane Na⁺/K⁺ ion pumps. Water moves rapidly into the cell, leading to cytotoxic edema and restricted extracellular diffusion of water.² This cellular swelling reduces extracellular volume, restricting diffusion of water.^2,3 The viability of this penumbra depends largely on collateral circulation. With time, the infarct core may grow into the penumbra.² The penumbra is considered to be an important predictor of outcome after stroke, leading some experts to call for consideration of penumbra size in recanalization decisions.^4–7 However, recent studies have shown that the penumbra is predictive of stroke outcome only when paired with successful recanalization.^8,9

I. Computed tomography (CT)

Owing to its ubiquity, speed, and compatibility with implantable metals, CT imaging has become one of the most common stroke imaging modalities, providing crucial anatomic and occasionally physiologic information. Important CT imaging modalities for acute stroke include non-contrast CT (NCCT), CT angiography (CTA), and CT perfusion (CTP). The former two modalities provide predominantly anatomic information while the latter gives physiologic information.

(a) NCCT

NCCT is the mainstay of the initial assessment of acute stroke. It accurately and rapidly alerts the clinician if an intracranial hemorrhage has occurred, drastically altering intervention options, and reveals stroke mimics. NCCT also provides early imaging signs indicating a large ischemic stroke; these changes on NCCT occur within six hours of symptoms onset in up to 82% of patients.¹ These imaging signs include the hyperdense artery sign for clot detection, insular ribbon sign, obscuration of lentiform nucleus sign, sulcal effacement, and global loss of gray-white differentiation.^10,11

Extensive hypoattenuation has been associated with an eight-fold increased risk of hemorrhagic transformation (HT) after thrombolytic therapy.^10,12 Therefore, consideration of extensive hypoattenuation is used by many to determine the likely benefit of thrombolytic therapy, although this has been debated in the literature.

Despite these data, NCCT performs poorly in the detection of acute infarction, particularly in the posterior fossa. In one study, the sensitivity for detecting acute ischemic stroke on NCCT ranged from 57% to 71%.¹³ Early signs of acute infarctions are subtle, especially for smaller arterial occlusions and in the hyperacute stage of ischemic stroke, leading to limited sensitivity and poor interobserver reliability.^14,15 Moreover, the relationship between early ischemic changes on NCCT and poor outcomes after tPA is equivocal.¹⁰ Thus, although NCCT may provide information that has prognostic value in large strokes, its chief utility is to exclude hemorrhage and stroke mimics.

One promising area regarding NCCT is thin-slice reconstruction, using 1.25 mm to 2.5 mm slices.^16–18 Riedel and Jensen have highlighted the importance of the hyperdense middle cerebral artery (MCA) sign revealed by thin-slice CT in showing clots.^16–18 In their studies, using thin-slice CTs correlated with area under receiver-operating curves ranging from 0.94 to 0.97.¹⁸ The same authors demonstrated that in 138 patients with acute MCA strokes receiving IV thrombolysis, clot size was successfully measured after immediate thin-slice NCCT. No thrombus greater than 8 mm was successfully recanalized.¹⁷ This rapid, high diagnostic yield with thin-cut NCCT holds great potential for the future.

(b) CTA

Improvement in multi-detector CT technology has made CTA a valid alternative to catheter angiography in the evaluation of intracranial and extracranial vasculature.^19,20 In CTA, a bolus of iodinated contrast is injected intravenously and time-optimized CT scanning is activated.¹ Collateral circulation, thrombus location, and thrombus extent can be evaluated with CTA. Knowing the exact location of the occlusion allows early planning of the best method of management. Because images from the aortic arch to the circle of Willis can be obtained with excellent spatial resolution, patency or dissection of carotid, vertebral, and intracranial vessels can be evaluated.¹⁰ The literature demonstrates three unique benefits of CTA in acute ischemic stroke evaluation.

First, highlighting its benefit in pre-surgical planning, the Interventional Management of Stroke (IMS) III trial demonstrated that patients receiving a pre-intervention CTA benefited from quicker times from groin access to thrombectomy.²¹ CTA may facilitate improved pre-procedure planning, allowing for expedited treatment. Second, patients with a CTA revealing either proximal MCA occlusion or significant thrombus burden may be poor candidates for IV thrombolytics; these patients may be appropriate for IA or mechanical thrombolysis/thrombectomy.²² Finally, evaluating data from the Keimyung Stroke Registry, Nambiar et al. found that patients with recanalization and good or intermediate leptomeningeal collaterals by CTA benefited from IA therapy, whereas those patients with poor collaterals did not.²³ CTA imaging depends on proper timing, sound technical planning, and adequate cardiac output.¹ Thus disturbance of these conditions may yield poor images with inadequate diagnostic yield. However, if these limitations are overcome, CTA can prove essential in selecting the patients who are most likely to benefit from IA therapy.

(c) CTP

Focal vessel occlusion triggers global or regional vasodilation and increased oxygen extraction to meet tissue demands.^24,25 Vasodilation increases cerebral blood volume (CBV) and low flow velocities, or cerebral blood flow (CBF). Thus, it takes longer for blood to traverse an ischemic area, leading to elevated mean transit times (MTT). However, when the ischemic insult is severe or prolonged, auto-regulatory mechanisms fail to maintain adequate perfusion for neuronal viability, even with maximal vasodilation. Vessel caliber changes cannot accommodate the needed increased CBV, thus CBV is decreased and tissue death ensues. In the area of irreversible damage, CBF is also low and MTT is high (CBF = CBV/MTT).^1,26 Areas of increased CBV and high MTT values represent viable tissue under threat of infarction (ischemic penumbra) while areas of decreased CBV and prolonged MTT represent irreversibly damaged tissue (infarct core).¹⁰

CTP imaging provides a way to characterize the area of ischemic penumbra and infarct core. Modern helical CT scanners track injected iodinated contrast as it enters the arterial system, passing through capillaries into veins.²⁷ The linear relationship between the concentration of contrast and the signal intensity is used to generate time vs. concentration of contrast curves for each pixel. The above physiologic parameters of cerebral circulation, CBV (normal 4–6 ml/100 g), CBF (normal 50–60 ml/100 g/min), and MTT (normal four seconds), are quantitatively assessed after deconvolution analysis of arterial and venous time-attenuation curves.^1,10,26 Comparison among these parameters can differentiate potentially reversible tissue damage from permanently damaged tissue, thus allowing for selection of patients who may benefit from reperfusion interventions.¹

The CTP parameter that has been suggested to most accurately describe the ischemic penumbra is MTT, while low absolute CBV value describes the infarct core.²⁸ The mismatch between these two parameters indicates an area at risk of infarction with or without delayed thrombolytic therapy. An advantage of CTP is the ability for it to be obtained rapidly, and sometimes concomitant with, CTA data. Additionally, the same study can be used to assess collateral flow based on temporal resolution.²⁸

Although CTP provides a quantitative assessment of the physiologic state of the brain region affected by an ischemic event, its accuracy is debated.¹ Assumptions underlying deconvolution algorithms have been questioned.^29–32 Additionally, CTP images are spatially limited to 2–4 consecutive sections with coverage of 20–40 mm thickness, which means they may underestimate the full extent of brain perfusion.¹ The limited spatial coverage is beginning to be resolved by the use of multidetector CT scanners, such as the 320-detector scanner, which can cover the entire brain with a single rotation.¹ Moreover, because of its spatial limitation, CTP cannot detect small lacunar vessels, but it has 95% accuracy in delineation of supratentorial strokes.^26,33 CTP images continue to be limited in their interobserver reliability, variation in thresholds used to define infarct core and penumbra, and relatively high ionizing radiation doses due to tracking of contrast bolus.¹ However, CTP remains a worthwhile tool to obtain valuable physiologic data that NCCT and CTA do not provide.

II. Magnetic resonance imaging (MRI)

MRI provides improved sensitivity for detecting ischemic and infarcted tissue and avoids exposure to ionizing radiation and iodinated contrast. However, MRI often has limited availability, requires prolonged acquisition times, and is incompatible with some specific models of coronary stents and defibrillators. Important MRI sequences in stroke assessment include gradient echo (GRE), fluid-attenuated inversion recovery imaging (FLAIR), MR angiography (MRA), diffusion-weighted imaging (DWI), and perfusion-weighted imaging (PWI). Conventional T1- and T2-weighted sequences are relatively unrevealing in acute stroke.^1,34

(a) GRE and FLAIR

GRE is highly sensitive for detecting acute hemorrhage, including clinically silent microbleeds undetectable by NCCT. Meanwhile, FLAIR imaging detects subtle subarachnoid hemorrhages that may also be missed by CT imaging. The prognostic significance of these clinically silent microbleeds is debatable.^1,14,15 Thus GRE and FLAIR provide similar and more detailed information compared to NCCT imaging in the initial assessment of a patient with acute neurological changes. However, these sequences often do not change management of the acute stroke patient.

(b) MRA

Like CTA, MRA is an excellent noninvasive assessment of intracranial and extracranial vasculature for stenosis, occlusion, intravascular thrombi, and luminal narrowing.^10,35 MRA is particularly useful when evaluating large, proximal arteries. Time-of-flight MRA (TOF-MRA) is often used to assess intracranial vasculature while contrast-enhanced MRA better assesses extracranial vasculature.¹⁰ Reported ranges of the sensitivity and specificity of MRA in detection of cervical and intracranial stenosis range from 70% to 100%.¹⁴ With information obtained from MRA demonstrating the patency of proximal vessels, intervention strategies can be enhanced. TOF-MRA, though dependent on flow rates, is especially attractive because it requires no additional contrast dosage, thus minimizing nephrotoxicity and risk of allergic reaction.¹⁰

Patients presenting with stenosis of large vessels, such as the MCA, are often considered candidates for IA therapy.³⁶ Therefore, accurate detection of large vessel occlusions becomes essential for a preferred imaging modality. Many attempts have been made to compare MRA to CTA in regards to visualization of occlusion and stenosis of the major cerebral vessels. In one study, image readers were blinded to patient history, clinical data, and estimated degree of stenosis of large vessels. The study showed that CTA had a higher degree of sensitivity for both stenosis and occlusion of major vessels (98% and 100%, respectively) than MRA (70% and 87%, respectively).³⁷ However, alternative studies have shown that MRA may still be an acceptable imaging technique to show complete occlusion of major vessels.^38–40

(c) MR diffusion

DWI depends on the molecular motion of water. The net effect of water moving from the extra- to intra-cellular space is an overall reduction in water mobility due to intra-cellular structural and molecular components acting as barriers to free motion. This is captured as hyperintensity on DWI and hypointensity on apparent diffusion coefficient (ADC) maps.

DWI is a linearly T2-weighted and exponentially DWI while ADC is a linearly DWI without a T2-weighted component. DWI is interpreted in the context of ADC because of the possibility of T2 signal “shine-through” on DWI, which will be hyperintense in ADC images. The changes on both DWI and ADC evolve over time with the evolution of the ischemia.

After ischemic stroke, ADC hypointensity progresses until it reaches its nadir in 1–4 days. This hypointensity returns to baseline in 1–2 weeks.^41–43 Following this time period, there is a progressive increase in intensity of ADC as the ischemic tissue is degraded and cleared, leaving a cerebrospinal fluid (CSF)-filled cavitary lesion. Variability of DWI with stroke age is more varied and may not correspond to that of ADC. It can be mildly hyperintense, isointense, or hypointense depending on the strength of the T2 effects and diffusion component.⁴¹ Factors such as infarct type, patient age, and reperfusion status may affect the evolution of signal changes on DWI.^44,45

DWI is regarded as the most reliable method for early detection of cerebral ischemia, delineation of the infarct core, and the identification of stroke mimics.^46–48 In fact, the reported sensitivity and specificity of detecting acute ischemia within 12 hours on DWI is high, 81–100% and 86–100%, respectively.^2,41,46–50 This sensitivity and specificity are even higher in specialized stroke centers.⁴¹ DWI allows visualization of ischemia as early as 11 minutes after symptom onset and correlates remarkably well with clinical measures of stroke severity.^14,51,52 It is held to be the best method of identifying the infarct core and tissues likely to progress to infarction.^53–55

A substantial number of DWI lesions persist even after thrombolytic reperfusion. However, reversal of DWI-indicated infarction is known to occur and has been published in multiple series. In one study, 19% of patients had reversal of a DWI hyperintense region on a follow-up image after treatment with IA thrombolysis.⁵⁶ The Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution (DEFUSE) trial achieved 43% median reversal rate of DWI changes at 30 days after treatment, which is higher than most other studies. In a more recent publication in Stroke, Luby et al. demonstrated that, in a cohort of 71 patients who all received IV tPA, 8% of patients experienced sustained DWI reversal on subsequent imaging.⁵⁷

DWI is also predictive of clinical outcome. Results of the DEFUSE trial show that early recanalization contributes to both partial reversibility of small to moderately-sized DWI lesions and better clinical outcomes.⁵⁸ Large DWI lesions predicted poor outcomes regardless of recanalization.^59–62 In a study by Yoo et al., 54 patients with acute stroke were treated with heparin or IV tPA.⁶² Those with DWI lesions greater than 72 ml had poor outcomes, defined by modified Rankin score 3–6.⁶¹ A separate analysis by the same investigators found that even with a 50% recanalization rate, patients with DWI lesions greater than 70 ml had consistently poor outcomes.⁶² The analysis of 98 patients from the Echoplanar Imaging Thrombolytic Evaluation (EPITHET) trial by Parsons and colleagues found that patients with DWI lesions greater than 25 ml gained little benefit from IV tPA treatment.⁵⁹ These findings support the fact that DWI lesions are an indicator of an area that is irreversibly damaged or destined for infarction, and correlate well with National Institute of Health Stroke Scale (NIHSS), Glasgow Outcome Scale, and Rankin/modified Rankin scale.^54,63–65

Although patients with large DWI lesions tend to have poor outcomes regardless of recanalization, whether DWI or ADC images can be used to predict adverse events such as HT is not clear. In one study, 645 patients with anterior circulation strokes were treated with IV or IA thrombolytics, and increasing size of DWI lesion was associated with an increase in symptomatic HT.⁶⁶ The investigators observed that the subgroup with DWI lesions of 100 ml or greater had the highest risk for HT. In another study where 29 patients were treated with IV thrombolytics, the absolute number of voxels (a volume unit used in two-dimensional (2-D) imaging) with ADC ≤ 550 × 10⁶ mm²/s correlated with HT.⁶⁷ Several other studies have found correlation between ADC area and symptomatic HT.^68,69 However, other studies have found no correlation between ADC size and HT, and some have discerned MR perfusion better predicts HT than DWI.^70,71

Despite the many advantages discussed above, DWI can produce false positive and false negative results. False positives may occur because of T2 “shine-through”. This necessitates the interpretation of DWI images in conjunction with ADC or exponential maps. Non-ischemic lesions, such as demyelinating diseases, cause acute neurologic symptoms and decreased perfusion, and can be mistaken for infarcts. Interpretation of DWI in conjunction with conventional MR imaging such as FLAIR, contrast-enhanced MRI, and T2 can help distinguish these conditions from infarcts.⁴¹ Conversely, false negatives can be due to small punctate infarcts, especially in the medulla.^68,71

(d) MR perfusion (PWI)

Like CTP, the primary goal of MR perfusion imaging in acute stroke is to identify threatened but salvageable tissue. There are two main ways to obtain MR perfusion imaging: arterial spin labeling (ASL) and gadolinium-based contrast. ASL relies on paramagnetic labeling endogenous arterial water and does not require the use of exogenous gadolinium-based contrast agent. However, ASL is not in widespread use for three primary reasons. Arterial blood is delayed in reaching ischemic penumbra; therefore, the short ASL tracer half-life may obscure sensitive assessment of CBF because of the loss of spin tag labeling over this delayed time course.^72,73 A low signal-to-noise ratio, limited visualization of collaterals also due to loss of spin tag labeling, and long imaging times have also limited its widespread use.^24,73 ASL strategies using transit-time insensitive labels would be useful to advance this approach.

Gadolinium-based dynamic susceptibility contrast (DSC) imaging is the technique of choice in most clinical centers because it enables the improved visualization of small arteries, in contrast to conventional T1 imaging.²⁴ DSC traces contrast as it arrives in large arteries, through smaller vessels, and finally into large intracranial veins. Important measurements derived from DSC include CBV, CBF, MTT, time to peak concentration (TTP) and time at which the deconvolved function reaches its maximum value (T_max).²⁴ These measurements are useful in estimating the overall area of ischemia, which includes the infarct core, the penumbra, and benign oligemia.

Like with CTP, DSC parameters are useful when used in conjunction with DWI to obtain a mismatch. Several studies have used DWI-TTP mismatch to estimate an area threatened by ischemia.^74–76 Likewise, MTT and TTP have been used to identify the area threatened by hypoperfusion.^53,61,77 Large mismatches between MR perfusion maps and DWI have also been shown to correlate with the risk of infarct growth.^53,74,75,78

Finally, the widespread use of perfusion imaging remains stymied by a lack of standard definitions for key imaging parameters like PWI/DWI mismatch and MTT prolongation. For example, a prospective study was conducted recruiting 32 patients with acute ischemic stroke who received DWI and PWI at baseline and follow-up T2-weighted imaging at one month.⁷⁹ Ten perfusion parameters were calculated (six MTT, three CBF, one CBV, seven relative, and three quantitative) to measure lesion and mismatch volumes. Median lesion volume and PWI/DWI mismatch varied significantly, leading to widely disparate estimates of “tissue at risk,” depending on the perfusion parameter used. These data point to the urgency of standardizing methods for processing perfusion data, the authors highlight. The acceptance of standard metrics for perfusion imaging and their subsequent validation are needed to promote wide use of this tool.^26,79–81

IV. Comparison of MRI and CT

Acute ischemic stroke is a time-sensitive event requiring immediate recognition and intervention. The goal of early imaging is to exclude hemorrhage and stroke mimics, identify ischemic areas, estimate the risk of hemorrhage, and delineate extent of vascular occlusion, infarct core, and penumbra. In any acute stroke evaluation, NCCT is absolutely necessary to evaluate for hemorrhage. Below, we will discuss the advantages and disadvantages in CT and MR after the initial NCCT.

Multimodal CT imaging for acute stroke includes NCCT, CTA, and CTP. When combined, these imaging modalities provide both qualitative and quantitative assessment of intracranial hemorrhage, intracranial and extracranial circulation, as well as the estimation of the ischemic tissue area. The drawbacks of CT imaging are the use of iodinated contrast and the radiation exposure involved. To illustrate the importance of this, a Canadian study demonstrated a mean effective radiation dose for baseline 64-slice NCCT brain scan to be 2.7 mSv; by adding CTA and CTP, this value increased to 13 mSv.⁸² Iodinated contrast agents used with CT can cause nephrotoxicity in patients with baseline renal dysfunction. CTP typically also suffers from limited spatial brain coverage, even with 64 multiple detector (MD) CT scanners. This is being increasingly addressed in modern CT detectors and toggling table techniques.¹⁰

On the other hand, MRI is devoid of ionizing radiation, provides improved anatomical detail, and can identify earlier ischemic changes, small infarcts, infarcts in the posterior fossa, and microbleeds. MRI is much more sensitive in the diagnosis of acute stroke (83%) compared to NCCT (26%).^34,46 GRE and FLAIR have excellent accuracy for detecting intracranial hemorrhage.¹⁰ MRA allows the assessment of intracranial and extracranial circulation and will show areas of vessel occlusion. T2 images can also detect loss of arterial signal void in occluded vessels within minutes of the stroke onset.¹⁰ T2 and FLAIR allow assessment of older cerebral infarct and extent of concomitant small vessel disease. MRI provides what is generally considered to be the most reliable estimates of the infarct core characterized by hyperintensity in DWI with associated reduction in ADC signal.^2,83 MR perfusion uses gadolinium tracer to estimate CBV, MTT, and CBF from which infarcted area can be estimated. The mismatch between DWI and PWI images provides a surrogate for penumbral tissue.²

However, MRI tends to take longer than CT, is less but increasingly available, and is contraindicated in patients with some pacemakers, cardiac prostheses, and who are deemed medically unstable. MR is also more sensitive to motion artifacts, which can be a significant limitation in patients with acute neurological changes.

Both CT and MRI modalities provide enough information to allow treatment decisions to be made after acute ischemic stroke. The choice of imaging will depend on the institutional preference, equipment availability, and clinical urgency. Many stroke centers are standardizing protocols to facilitate quicker evaluation and maximizing the efficient use of limited resources.

V. Imaging and patient selection

The landmark 1995 National Institute of Neurological Disorders and Stroke (NINDS) trial showed benefit of tPA when patients were treated within three hours after symptom onset.¹² The European Cooperative Acute Stroke Study (ECASS III) showed tPA success within 4.5 hours in carefully selected patients.⁸⁴ Several studies using tPA at five hours and six hours have failed to show a benefit from attempts at thrombolysis.^85,86 Thus, the current recommendation is to treat patients with IV tPA if they present within 4.5 hours of symptom onset. As a result of the strict time limit, only 1%–7% of patients receive thrombolytic therapy because most patients cannot determine with a good degree of certainty the exact time of symptom onset.^87–90 Modern imaging techniques have shown that there is considerable variation between individual stroke patients in terms of collateral circulations, ischemic core, and penumbra. It is likely that some patients who present within the treatment window may have no salvageable tissue left, while others may have a significant degree of penumbra even several hours past the recommended treatment window. There is a tremendous opportunity to customize stroke treatment by using imaging to carefully select patients who might benefit from intervention outside the current treatment window, thus making treatment benefits available to many more patients. Table 1 summarizes several patient series correlating imaging with patient selection.

Table 1.

The effect of imaging modality on treatment decision-making.

Study citation	n	Imaging	Design	Results
Lima et al., 2014103	126	CTA	P	In adult patients with CTA demonstrating anterior circulation ischemic stroke with symptoms within 24 hours of onset, these treatment modalities were chosen: --IV-tPA: 51 (23.7%) --IA-tPA: 10 (4.6%) --IV and IA-tPA: 16 (7.4%) --No treatment: 138 (64.2%)
Chung et al., 2013104	57	FLAIR	P	In adult patients with acute ischemic stroke who present with 2.5 hours of onset of symptoms, have MRI/MRA-confirmed persistence of occlusion after IV-tPA, and were afterward treated with an IA form of thrombolysis, these treatment modalities were chosen based on the presence of FLAIR-hyperintense lesions: --“Chemical thrombolysis”: 29 (90.6%) --Mechanical thrombolysis: 28 (87.5%) --Balloon angioplasty: 12 (37.5%) --PENUMBRAa: 4 (12.5%) --Stenting: 5 (15.6%)
Kidwell et al., 2013105	66	Multimodal MRI, NCCT	P	In adult patients with acute ischemic stroke demonstrated by multimodal MRI or CT demonstrated in anterior large-vessel circulation with therapy initiated within eight hours of symptoms, recanalization TIMI of 2–3, and follow-up imaging within seven days, these treatment modalities were chosen: Multimodal MRI (34 patients): --IA-tPA: 12 (35.3%) --Bridging IV-tPA to IA-tPA: 11 (32.4%) --Mechanical thrombectomy: 11 (32.4%) CT (32 patients) --IV-tPA: 3 (9.4%) --Mechanical thrombectomy: 1 (3.1%) --Bridging IV-tPA to endovascular: 12 (37.5%) --Combined mechanical thrombectomy and IA-tPA: 16 (50%)
Kang et al., 2012106	430	MRI	P	In adult patients with unclear-onset stroke with PWI/DWI mismatch >20%, negative or subtle FLAIR changes, these treatment modalities were chosen: --Reperfusion therapy given: 83 (19.3%) • IA-tPA: 57 (68.7%) • IV-tPA + IA-tPA: 17 (20.5%) • IV-tPA: 9 (10.8%) --No reperfusion therapy: 347 (80.7%)
Salottolo et al., 2011107	108	Multimodal CT (CTA, PCT, NCCT)	R	In adult patients with acute ischemic stroke with symptom onset to hospital arrival <2.5 hours, when multimodal CT is added to the diagnostic workup, the following effects on tPA administration were noted: --Median time from arrival to tPA was shorter with multimodal CT than for those evaluated with only NCCT (55 vs. 78 minutes, p = 0.02) --No significant difference in odds ratio of receiving timely tPA (<60 minutes) for those evaluated with multimodal CT in addition to NCCT
Thomas et al., 2011108	207	CTA	R	In adult patients with acute ischemic stroke presenting within 24 hours of symptom onset who underwent emergent CTA, these treatment modalities were chosen: --IV-tPA: 25% --IA-tPA: 2.4% --Mechanical thrombectomy: 6.8% --Surgery: 3.3% --Admitted to neuroscience ICU: 52%
Hassan et al., 2010109	164	CTP	R	In adult patients with acute ischemic stroke treated with endovascular approaches based on either CTP or time interval between symptom onset and presentation, these treatment modalities were chosen: Time-Guided treatment (103 patients): --IV-tPA: 47 (37%) --IA-tPA: 84 (66%) --Mechanical thrombectomy: 41 (32%) --Angioplasty: 37 (29%) CTP-Guided treatment (61 patients): --IV-tPA: 32 (46%) --IA-tPA: 50 (72%) --Mechanical thrombectomy: 27 (39%) --Angioplasty: 26 (38%)

^aPENUMBRA: Food and Drug Administration (FDA)-approved device for clot aspiration, debulking, and retrieval. CTA: computed tomography angiography; NCCT: non-contrast computed tomography; DWI: diffusion-weighted imaging; PWI: perfusion-weighted imaging; TCD: transcranial Doppler ultrasonography; IV-tPA: intravenous tissue plasminogen activator; IA-tPA: intra-arterial tissue plasminogen activator; MRI: magnetic resonance imaging; MRA: magnetic resonance angiography; FLAIR: fluid-attenuated inversion recovery; TIMI: Thrombolysis in Myocardial Infarction score.

Patient selection is based on two factors: First, the ability to identify the irreversibly damaged tissue (infarct core); second, the ability to identify and distinguish the viable but threatened tissue (penumbra) from the infarct core. Both MR and multimodal CT imaging detect the site of vessel occlusion and characterize the area of infarct core and the penumbra. DWI estimates the volume of the infarct core while PWI gives an estimate of an area of overall ischemia. The mismatch between the two correlates with the penumbra including areas of physiological oligemia, which are not in danger of turning into an infarct.¹⁰ If the clinical outcome depends on the reperfusion of the salvageable penumbral tissue, then the time window for thrombolytic therapy can be widened, provided it is given to patients with significant DWI/PWI mismatch to warrant a net clinical benefit.

Several trials have suggested a clinical benefit in patients with significant DWI/PWI mismatch treated outside the recommended time window. One study was a non-placebo-controlled trial in which patients who were found to have at least a 50% DWI-TTP mismatch were treated with IV tPA between three and six hours after onset of symptoms.⁹¹ The rate of recanalization, HT, and neurologic improvement were similar between patients treated at an extended time window (n = 43) and those who were treated according to the recommended time window of 0–3 hours (n = 79).

The Desmoteplase In Acute Stroke (DIAS) trial, a double-blind, placebo-controlled phase II study, enrolled patients who presented between three and nine hours after onset of stroke symptoms, scored from 4 to 20 on the NIHSS, and notably showed at least a 20% PWI/DWI mismatch on initial imaging.⁹² Patients were randomized to receive weight-adjusted doses of desmoteplase or placebo. Reperfusion was obtained significantly more often with desmoteplase (71.4% vs. 19.2%, p = 0.012). Favorable functional 90-day outcomes were seen in 60% of patients treated with the highest desmoteplase dose compared to 22.2% in the placebo group. The Dose Escalation of Desmoteplase for Acute Ischemic Stroke (DEDAS) trial showed similar results with these PWI/DWI parameters. Two other trials corroborate the value of significant mismatch PWI (at least 20% more than DWI lesion) in assessing the likelihood of a positive clinical benefit.^93,94 Only one trial, DIAS-2, failed to show clinical benefit for treatment of patients presenting three to nine hours after symptom onset.⁹⁴ It was speculated that these results might be due to low absolute mismatch volumes compared to other trials, milder stroke population, and low prevalence of occlusion on MR or CTA in this sample.

Finally, there is a 5.9% risk of HT in patients treated with IV-tPA compared to 1.1% of controls after acute ischemic stroke. Acute insult to the blood-brain barrier (BBB) is important for facilitating HT. T2-MRI permeability imaging has recently been shown to be valuable in assessing the disruption to the BBB, predicting which patients are most at risk for HT.^95,96 While this is a relatively new approach with mostly small retrospective studies documenting its utility, using permeability imaging to select out patients from being treated with tPA could be a powerful tool in the future.

How imaging changes patient management and treatment course

Device development has recently exploded for novel endovascular therapies. However, large randomized controlled trials (RCTs) have called into question the superiority of endovascular therapy against tPA alone. As ischemic stroke is a heterogeneous process affecting a diverse set of risk groups, advanced imaging could provide for patient-tailored treatment approaches.⁹⁷

Few studies exist that investigate the benefit of a given imaging modality on decision-making surrounding patient management. With increasingly sophisticated diagnostic imaging becoming part of standard care, understanding how a given test adds value and changes treatment decision-making is important.⁹⁸ As discussed here, these new techniques may add value to existing imaging protocols, assist in efficiently using limited resources in comprehensive stroke centers (CSCs), or replace existing time-based thrombolytic therapy with specific imaging biomarkers.

No formal guidelines exist for identifying which patients should be transferred from a primary stroke center (PSC) to CSCs with advanced endovascular treatment capabilities.⁹⁹ Transfers should be reserved for those patients who would benefit most. Thomas et al. recently proposed using CTA as a tool to identify such patients.⁹⁹ In their historical cohort study of 207 patients presenting with acute ischemic stroke, those with CTA-demonstrated proximal occlusions were more likely to require neurointerventional services (26% vs. 2%, p < 0.001), mechanical intervention (19% vs. 0%, p < 0.0001), and IA-tPA (9% vs. 0%, p = 0.001).⁹⁹ Understanding the added value of CTA-based resource use is especially important in this group because clinical evaluation by grading systems like the NIHSS are poorly predictive of proximal artery and posterior circulation ischemia.^100,101

Because patients presenting with unclear-onset strokes represent a large share of those excluded from time-based thrombolysis, DWI/PWI mismatch could potentially identify candidates for various thrombolytic therapies. In a prospective, multicenter study of 430 patients with unclear-onset stroke with DWI/PWI mismatch >20%, negative or subtle FLAIR changes, reperfusion therapy was given in 83 (19.3%). Most were given IA-tPA or IA-tPA + IV-tPA (68.7% and 20.5%, respectively).¹⁰² Altogether, those 19.3% treated had better outcomes as defined by modified Rankin scale score of 0 to 2 after adjusting for age, sex, and baseline NIHSS in logistic regression analysis (odds ratio (OR): 2.25, 95% confidence interval (CI): 1.14–4.49).

In conclusion, acute stroke likely will evolve to benefit significantly from multimodal imaging, allowing the treating physician to identify infarcted core and any viable penumbra. Emergent imaging selection should be thoroughly evaluated in any emergency department setting, with established protocols in place to facilitate rapid patient assessment and treatment according to consistently applied evidence-based care. We hope this review of the current literature regarding acute stroke imaging will aid centers in establishing that system of care that they believe is best indicated according to the literature and their own resource availability.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

None declared.