Treatment and imaging of intracranial atherosclerotic stenosis: current perspectives and future directions

Abstract Background and Purpose Intracranial atherosclerosis is a common cause of stroke worldwide. It results in ischemic stroke due to different mechanisms including artery‐to‐artery embolism, in situ thrombo‐occlusion, occlusion of perforating arteries, and hemodynamic failure. In this review, we present an overview of current treatment and imaging modalities in intracranial atherosclerotic stenosis. Methods PubMed was searched for relevant articles in English that evaluated the treatment and imaging of intracranial atherosclerotic stenosis (ICAS). Results Aggressive medical management, consisting of dual antiplatelet therapy and intensive risk factor management, is important in patients with ICAS because of a substantial risk of recurrent stroke, approximately 20% in the first year, in patients on aspirin or warfarin alone. Recent trials have suggested that, aggressive medical therapy results in better outcome as compared with intracranial stenting. However, the question remains what the optimal treatment strategy would be in patients with recurrent strokes in the setting of failed aggressive medical therapy. Moreover, controversy exists whether a subgroup of patients with symptomatic ICAS could benefit from intracranial stenting if selection is based on radiological evidence of hemodynamic failure. With regard to imaging, transcranial Doppler ultrasound and magnetic resonance angiography are useful screening tests for exclusion of ICAS, but need confirmation by other imaging modalities when stenosis is suggested. Computed tomography angiography has a high positive and negative predictive value for detection of intracranial luminal stenosis of 50% or higher, but performs worse than digital subtraction angiography with regard to establishing the exact degree of luminal stenosis. Novel imaging techniques including high‐resolution CT and MRI better identify plaque characteristics than conventional imaging methods. Conclusions Currently, aggressive medical management remains the standard of care for patients with ICAS. Further research is needed to identify high‐risk subgroups and to develop more effective treatments for ICAS patients.

Additional papers were identified through searches based on the references of relevant studies.

| STROKE MECHANISMS IN INTRACRANIAL ATHEROSCLEROTIC STENOSIS
The underlying mechanism of ischemic stroke in ICAS is typically inferred by infarct pattern on neuroimaging (Holmstedt, Turan, & Chimowitz, 2013). Typical patterns are border zone infarctions as a result of hypoperfusion due to a highly stenotic artery, and territorial infarctions as a result of artery-to-artery embolism. One of the possible causes of a lacunar infarction in ICAS is plaque extension over small penetrating artery ostia (also known as branch occlusive disease; Holmstedt et al., 2013).
With regard to localization, intracranial atherosclerosis in the anterior circulation was more often associated with artery-to-artery embolism (52% vs. 34%) and less often with perforator branch occlusion (12% vs. 40%) than intracranial atherosclerosis in the posterior circulation (Kim, Nah, et al., 2012).
Determining whether intracranial artery stenosis is symptomatic or asymptomatic may not be straightforward, since at least 19% of recurrent strokes in ICAS could have been caused by other coexisting van den Wijngaard et al. | (3 of 16) e00536 mechanisms such as cardioembolism and extracranial large artery disease (Famakin, Chimowitz, Lynn, Stern, & George, 2009). Moreover, studies with microembolic signal monitoring by transcranial Doppler indicate that a combined embolism-hypoperfusion mechanism could be common in symptomatic MCA stenosis (Leung et al., 2015;Wong et al., 2002). In a prospective study of 30 patients with symptomatic MCA stenosis, TCD monitoring showed microembolic signals in eight out of 16 patients with border zone infarcts (Wong et al., 2002).
Hemodynamic compromise in conjunction with multiple small arteryto-artery emboli may result in border zone infarctions because of failure to clear emboli in a poorly perfused brain area (Wong et al., 2002). Also, for stroke recurrence in ICAS, it was demonstrated that artery-to-artery thrombo-embolism (as demonstrated with TCD) in combination with impaired washout at border zones (as demonstrated with watershed infarction on MRI DWI) was a common mechanism (Leung et al., 2015).

| Imaging stroke mechanism in lacunar stroke
In patients with intracranial atherosclerosis, those with lacunar and nonlacunar presentations have similarly high risks of recurrent stroke (18% vs. 22% over a mean follow-up of 1.8 years in the Warfarin-Aspirin Symptomatic Intracranial Disease [WASID] trial) (Khan, Kasner, Lynn, & Chimowitz, 2012). The differentiation of subtypes of lacunar ischemic stroke caused by focal macroatheroma (known as branch occlusive disease) or fibrohyalinosis (attributed to hypertension, microatheroma and endothelial failure) could be important because patients without macroatheroma could get less long-term benefit from antiatheromatous treatments (Benavente et al., 2012;Wardlaw, Smith, & Dichgans, 2013). On conventional imaging, the perforating arteriolar lumen and atheroma in the perforating vessel are difficult to identify.
Recently, 7-Tesla MRI has become available, which has the potential to improve our understanding of small vessel disease (SVD) by visualizing the vascular pathology itself as well as parenchymal markers which previously could only be examined postmortem (Benjamin, Viessmann, MacKinnon, Jezzard, & Markus, 2015). With 7-Tesla MRA, differences in the lenticulostriate arteries of patients with SVD, hypertension, and previous stroke have been reported with a reduced number of branches compared with healthy controls (Benjamin et al., 2015). Direct visualization of the affected lenticulostriate artery in a specific lacunar infarct has also been performed in a few selected number of cases (Kang et al., 2012). Vessel wall imaging with 7-Tesla MRI for small vessel disease is a promising technique since it can be used to visualize basal intracranial vessel wall disease which may be useful in determining the role of intracranial atheroma in the pathogenesis of lacunar infarction (Benjamin et al., 2015).

| Vascular lumen and degree of stenosis
The method currently used in clinical practice to grade atherosclerosis is measuring the degree of stenosis expressed in percentage of the total vessel lumen. However, a seemingly normal vascular lumen in patients with intracranial atherosclerosis does not necessarily indicate a healthy vessel segment since the lumen can remain normal for a long period of time because of vascular remodeling (Glagov, Weisenberg, Zarins, Stankunavicius, & Kolettis, 1987). Positive (outward) remodeling and large enhancing plaques are associated with unstable plaque morphology, resulting in plaque rupture, which has been demonstrated with high-resolution imaging for the embolic subtype of intracranial atherosclerotic disease (Ryoo, Lee, Cha, Jeon, & Bang, 2015). Therefore, simply measuring the vessel lumen as a measure of intracranial atherosclerosis might underestimate the thromboembolic complication risk of these patients (Ma et al., 2010). However, the WASID trial showed that the degree of luminal stenosis did affect the risk of stroke recurrence. At a median follow-up of 1.8 years, risk of stroke in the territory of the stenotic artery was 11% for patients with 50-69% stenosis, 18% for patients with 70-79% stenosis, 30% for patients with 80-89% stenosis while 9% in the 90-99% stenosis group (Kasner et al., 2006).
The severity of ICAS is also related to stroke mechanism (Jiang et al., 2006). In 80 symptomatic ICAS patients, stroke with the branch occlusive disease subtype had a milder degree of stenosis in comparison with nonbranch occlusive disease subtypes of ICAS patients (41% vs. 75%) (Ryoo et al., 2015). Although the severity of ICAS is related to the risk of border zone infarction (Wong et al., 2002), the risk of branch occlusive stroke is not associated with the severity of ICAS stenosis (Lopez-Cancio et al., 2014).
Next to reduction in the lumen diameter, length of the stenosis also seems relevant since the risk of ipsilateral ischemic stroke one year after stenting is 8% for lesions of ≤5 mm, 12% for lesions 5-10 mm in length, and 56% in patients with ICAS length of >10 mm (Mori, Fukuoka, Kazita, & Mori, 1998).

| Collateral status
Most of the infarcts in patients with MCA stenosis are smaller than the large cortical infarcts caused by acute MCA occlusion. This can be explained by the gradual development of sufficient leptomeningeal anastomosis between anterior, posterior, and middle cerebral arteries in the process of atherosclerotic stenosis of the MCA (Wong et al., 2002). More severe stenoses generally exhibit greater degrees of compensatory collateral flow as measured with DSA (Liebeskind et al., 2012). Collaterals were subsequently categorized as none (grade 0), poor (grades 1 or 2), or good (grades 3 or 4). Good collaterals demonstrated a protective effect on averting territorial stroke risk in severe e00536 (4 of 16) | van den Wijngaard et al.
Nevertheless, the authors conclude that isolated measures of the degree of stenosis in an artery may be inadequate for identifying hemodynamic or embolic significance (Liebeskind et al., 2012). (20%) in medically treated patients and in 11/46 (24%) of stented patients in the group with partial recanalization and poor collaterals (Liebeskind et al., 2012).
In a retrospective cohort study in which composite vascular assessment in intracranial atherosclerosis was used, good collateral compensation in patients with compromised antegrade flow was associated with a more favorable outcome (modified Rankin scale 0-2 at 3 months: 92 vs. 57%) (Lau et al., 2012). Collateral status is increasingly being used for patient selection in ICAS studies (Gonzalez et al., 2015a;Miao et al., 2015), but the exact value of collateral status in ICAS needs further study before it can be used for decision-making in clinical practice.

| Plaque components
Various components of atherosclerotic plaques might also be an important indicator of thromboembolic risk. Atherosclerotic plaques can be defined as being either stable or unstable. Unstable plaques are soft, lipid-rich, and have an inflamed thin fibrous cap. These unstable plaques are more vulnerable to plaque disruption than collagenrich plaques with a thick fibrous cap, which are considered stable plaques (Drouet, 2002;Libby, 1995;Loree, Kamm, Stringfellow, & Lee, 1992).
For intracranial atherosclerosis, these plaque characteristics could determine the likelihood of future ischemic events suggesting that even mild intracranial stenosis of <50% might be clinically relevant in the presence of an unstable plaque. In 259 autopsies of patients with ischemic stroke, 43% had at least one intracranial plaque inducing luminal stenosis graded ≥30%, which was considered to be the cause of the stroke in 6% of cases; among these, 73% were severe stenoses and 27% were stenoses between 30% and 75%. The authors conclude that stenosis graded 30-75% could be associated with occurrence of ischemic stroke and therefore be of clinical significance (Mazighi et al., 2008). A recent study investigating large cerebral arteries from 15 autopsy cases with noncardioembolic brain infarcts, found that 20% of advanced atherosclerotic lesions (i.e., thin fibrous cap atheroma and fibrocalcific plaques) only showed stenosis of <40% of the vessel lumen (Gutierrez et al., 2015). These findings suggest that the degree of stenosis does not fully account for cerebral atherosclerosis burden.
However, the exact relevance of these findings is yet to be determined since patients with ICAS of <50% were excluded from the randomized controlled ICAS trials (Chimowitz et al., 2005Zaidat et al., 2015). Further characterization of plaque morphology in vivo, not only in patients with severe ICAS but also in patients with milder degree of stenosis will provide more insight into the mechanism leading to ischemic stroke (Ritz et al., 2014).

| Medical management
Optimal treatment for symptomatic ICAS is still evolving (Gao et al., 2015;Holmstedt et al., 2013;Turan et al., 2014). The current treatment of patients with ischemic events attributable to intracranial stenosis is based on a combination of antiplatelet drugs and optimization of blood pressure and LDL cholesterol values through lifestyle modification and drug treatment (Qureshi & Caplan, 2014). In recent years, the preference for dual antiplatelet therapy in patients with symptomatic ICAS has increased . Support for aggressive medical management (i.e., intensive risk factor management and combined aspirin plus clopidogrel) comes from the lower early recurrent stroke rates in the SAMMPRIS trial (Derdeyn et al., 2014;Zaidat et al., 2015) compared with historical controls from the WASID trial taking either aspirin or warfarin (Chimowitz et al., 2005;Turan et al., 2014). After adjustment for difference in baseline characteristics, patients from the WASID trial had an almost twofold higher risk of the SAMMPRIS trial primary outcome (12.6% vs. 21.9% for any stroke or death within 30 days after enrolment or ischemic stroke in the territory of the qualifying artery beyond 30 days of enrolment). This finding supports the hypothesis that the lower rate of the primary outcome in the medical arm of SAMMPRIS compared with WASID patients was a result of the aggressive medical management used in the SAMMPRIS trial (Chaturvedi et al., 2015).
Box 1 ASITN/SIR Grading system for collaterals with digital substraction angiography.

0
No collaterals visible to the ischemic site 1 Slow collaterals to the periphery of the ischemic site with persistence of some of the defect 2 Rapid collaterals to the periphery of the ischemic site with persistence of some of the defect and to only a portion of the ischemic territory 3 Collaterals with slow but complete angiographic blood flow of the ischemic bed by the late venous phase 4 Complete and rapid collateral blood flow to the vascular bed in the entire ischemic territory by retrograde perfusion van den Wijngaard et al.

| (5 of 16) e00536
Further support for dual antiplatelet therapy comes from the CLAIR (Clopidogrel plus aspirin vs. aspirin alone for reducing embolization in patients with acute symptomatic cerebral or carotid artery stenosis) trial, which showed that combined aspirin and clopidogrel decreased the number of microembolic signals on transcranial Doppler ultrasound compared with aspirin alone in patients with symptomatic ICAS (31% vs. 54%) Wong et al., 2010).

| Stenting versus medical management
The SAMMPRIS trial has shown that patients with a transient ischemic attack or stroke due to 70-99% stenosis of a major intracranial artery had greater benefit from aggressive medical management alone than with percutaneous transluminal angioplasty and stenting with the Wingspan stent plus aggressive medical management (Derdeyn et al., 2014). Intracranial stenting was associated with an increased risk of recurrent ischemic events or death when compared with medical therapy in patients treated for symptomatic intracranial stenosis (15% vs. 6% at 30 days and 23% vs. 15% at 32 months follow-up). In the VISSIT (the Vitesse Intracranial Stent Study for Ischemic Stroke Therapy) trial, worse outcomes after stenting were also shown with a different (i.e., balloon expanding) stent in comparison with medical therapy in symptomatic intracranial atherosclerotic stenosis (24% vs. 9% at 30 days, 36% vs. 15% at 12 months follow-up) (Compter et al., 2015;Derdeyn et al., 2014;Zaidat et al., 2015). Despite the disappointing performance of intracranial stenting in both VISSIT and SAMMPRIS, it is possible that improvement of operator experience and periprocedural factors such as blood pressure management may lead to improved outcomes after stenting in future trials Zaidat et al., 2015). Newer stent technology could possibly also enhance the safety and success of future endovascular procedures (Zaidat et al., 2015). Device selection based on arterial access and lesion morphology is currently implemented in a large Chinese study , in which self-expanding stents are preferably used in patients with tortuous arterial access, while balloon-expandable stents are preferred in patients with smoother access.
Currently, aggressive medical management remains the standard of care for patients with ICAS (Kernan et al., 2014). This is further supported by a recent study which showed that a majority of symptomatic high-grade intracranial plaques regress or remain quiescent by 1 year under intensive medical therapy (Leung et al., 2015).

| Treatment of high-risk subgroups
It has been argued that a high-risk subgroup of patients who do not respond to medical management might possibly benefit from stenting (Abou-Chebl & Steinmetz, 2012; Gao et al., 2015). In previous trials (Compter et al., 2015;Derdeyn et al., 2014;Zaidat et al., 2015), patients were enrolled based on stenosis grade without further use of lesion-based risk models or distinction between the different subtypes of ischemia (perforator occlusion, in situ thrombosis, hemodynamic stroke, and embolic stroke). Hemodynamic infarctions that relate to a low flow situation based on critical stenosis with insufficient leptomeningeal supply are from a pathophysiological point of view better suited for endovascular therapy than other ICAS stroke mechanisms (Lutsep et al., 2015). In a recent post hoc analysis of the SAMPRISS trial, a distinction was made between ICAS patients with and without hypoperfusion symptoms defined as symptoms related to change in position, effort on exertion, or recent change in antihypertensive (Lutsep et al., 2015 (Lutsep et al., 2015). However, in our view, a beneficial effect of stenting for patients with hypoperfusion symptoms cannot be ruled out either.
Since patients presenting with perforator ischemia possibly have an excessive risk of periprocedural stroke (due to occluding stenosed perforators by plaque shifting), these patients are excluded in a new multicenter randomized clinical trial (Gao et al., 2015).
In a prospective registry study that evaluates stenting of symptomatic ICAS , imaging features are used for patient selection in addition to the degree of stenosis. For inclusion, hypoperfusion has to be demonstrated by means of reduced blood flow on CT perfusion or SPECT, poor collaterals on DSA, hemodynamic ischemic lesion on MRI, or high peak systolic velocities on TCD . The first results of intracranial stenting in patients with severe ICAS and radiological evidence of hemodynamic failure look promising, with a low rate of recurrent stroke or TIA (4% at 30 days follow-up) and a high technical success rate (97%) (Miao et al., 2015). Whether preprocedural hemodynamic imaging (such as CT perfusion) to demonstrate hypoperfusion will result in a different long-term outcome of intracranial stenting in comparison with aggressive medical therapy remains to be proven (Gao et al., 2015).

| Other therapies
In the EC-IC (extracranial-intracranial) bypass trial, patients with intracranial stenotic lesions consistently had worse outcome following direct revascularization in comparison with medical therapy alone (The EC/ IC Bypass Study Group, 1985

| IMAGING OF INTRACRANIAL ATHEROSCLEROSIS
In general, timely diagnosis is important in ICAS because time is a predictor of stroke recurrence, that is, a higher recurrent stroke risk in patients within 2.5 weeks after the first ischemic event than later (hazard ratio: 1.7; 95% CI 1.1-2.7) (Kasner et al., 2006). An important feature of intracranial atherosclerosis is that plaques are not fixed stenoses, but are dynamic and subject to remodeling under medical therapy (Famakin et al., 2009).

| Radiologic mimics of intracranial atherosclerosis
While anatomic diagnosis of arterial narrowing is made with reasonable accuracy, ascribing the etiology for the stenosis remains challenging in clinical practice (Prabhakaran & Romano, 2012 for ascribing different stroke etiologies (Dieleman et al., 2014a).

| Digital subtraction angiography
Digital subtraction angiography is considered the gold standard for quantification of stenosis and assessment of collateral flow (Kasner et al., 2006;Qureshi & Caplan, 2014). Such precise information is most valuable in patients in whom intracranial angioplasty, stenting, or EDAS is an option (Qureshi & Caplan, 2014). DSA has been used extensively to assess ICAS because of the inherent high spatial resolution leading to high imaging quality ( Figure 1). However, disadvantages include high costs, limited availability, and a small risk (<1%) of serious periprocedural complications (Cloft, Joseph, & Dion, 1999;Cloft, Lynn, Feldmann, & Chimowitz, 2011;Willinsky et al., 2003).
Although DSA is the criterion standard, it is usually not performed in the acute setting when diagnostic imaging in acute ischemic stroke patients is required. DSA is only used in the acute setting in selected patients when intraarterial recanalization procedures are indicated, and not for determining the presence of ICAS per se.  (Netuka et al., 2010;Safain et al., 2014). Vessel stenosis due to plaque occurs in three dimensions and with 2D-DSA these spatial features are flattened. Measuring a 3D phenomenon such as stenosis on a 2D imaging modality likely has inherent error (Safain et al., 2014;Streifler et al., 1994). A recent study with 18 patients with intracranial atherosclerosis has shown superior performance of high-resolution CT angiography over 2D-DSA and 3D-rotational angiography in characterization of plaque morphology (such as ulceration, calcification, dissection of the plaque, and core protrusion into the vessel lumen). Compared to highresolution CTA, such lesions are missed in 61% (n = 11) when evaluation of intracranial atherosclerosis is done with 2D-DSA, and in 50% of patients (n = 9) with 3D-rotational angiography (Safain et al., 2014).
Digital subtraction angiography can also provide valuable information about collateral flow. In a recently published prospective study in a Chinese population, patient selection for intracranial stent placement was based on the presence of hemodynamic failure as demonstrated on DSA with an ASITN/SIR collateral score (See Box 1) of <3.
This selection of patients for hemodynamic failure in ICAS resulted in a <5% 30-day rate of stroke, TIA, or death after intracranial stenting (Miao et al., 2015), substantially lower than the 15% and 24% 30day rates in the SAMMPRIS and VISSIT trials Zaidat et al., 2015). Finally, DSA is the most accurate modality to evaluate ICAS after stent placement because CTA and MRA are limited by stent-generated artifacts.

| Transcranial Doppler ultrasound
Transcranial Doppler ultrasound is safe, inexpensive, and easily applied in clinical practice (Markus, 1999). Transcranial Doppler ultrasound provides real-time flow information, can detect microembolic signals, and can provide information about cerebral autoregulation (Topcuoglu, 2012). However, its use in clinical practice is limited by a high operator dependency and the requirement of suitable temporal bone acoustic windows (which can be absent in up to 20% of patients) (Alexandrov, Demchuk, Wein, & Grotta, 1999;Markus, 1999;Meseguer et al., 2010;Seidel, Kaps, & Gerriets, 1995).
One early study, including 130 acute ischemic stroke patients, reported a sensitivity of 87.5%, specificity of 88.6%, positive predictive value (PPV) of 87.5%, and a negative predictive value (NPV) of 88.6% of TCD for the detection of stenosis or occlusion compared with DSA (Alexandrov et al., 1999 (Feldmann et al., 2007). This study was limited to stenosis only and excluded vessel occlusions, probably explaining the lower reported PPV.
Transcranial color-coded duplex ultrasound (TCCS), a technique that combines flow velocity measurements with imaging of parenchymal structures (Zipper & Stolz, 2002), overcomes this problem partially, but it would still require confirmatory testing with DSA. Recent studies using TCCS, reported a sensitivity of 72.9-88.9%, specificity of 82.9-94.8%, PPV of 51.1-79.4%, and NPV of 77.3-99.3% for the identification of stenosis or occlusion (Hou et al., 2009;Roubec et al., 2011). However, these ultrasound techniques are not often used in clinical practice for evaluation of ICAS because of the high operator dependency.

| Noncontrast CT
In general, the initial imaging study in stroke patients is noncontrast CT (NCCT). Calcification of coronary arteries as measured on NCCT has been strongly associated with the risk of future cardiac events (Criqui et al., 2014). Multiple studies aimed to determine a similar correlation between calcification in the intracranial ICA and ischemic stroke (Bos et al., 2012;Chung et al., 2010). Although calcification did not predict ischemic stroke, the degree of calcification in the intracranial ICA as assessed with NCCT was related to stenosis as shown with DSA (Sohn, Cheon, Jeon, & Kang, 2004;Taoka et al., 2006;Woodcock, Goldstein, Kallmes, Cloft, & Phillips, 1999). Calcification volume measured on NCCT appeared to be a risk factor for stroke (Bos et al., 2014). However, a direct causal link between intracranial calcification and ischemic stroke has not yet been proven (Chimowitz & Caplan, 2014). Another major disadvantage of measuring calcification on NCCT is that this imaging modality cannot assess the distal intracranial arteries, since they are too small and only large calcified plaques can be detected (Sohn et al., 2004). Measuring calcification on NCCT might therefore probably be not enough to provide insight into intracranial atherosclerosis and the subsequent stroke risk.

| CT angiography
CT angiography is a useful screening tool because it is only minimally invasive, fast, and more widely available in clinical practice than MRA and DSA. CT angiography has a high interoperator reliability when assessing stenosis grade (Bash et al., 2005) and it is less susceptible to motion artifacts compared with MRI techniques (Bash et al., 2005;Nguyen-Huynh et al., 2008). Disadvantages are the exposure to radiation and the necessity of iodinated contrast material use, which may lead to allergic reaction and nephropathy in some cases (Bash et al., 2005). Computed tomography angiography is not suitable for depiction of vessels with a diameter smaller than 0.7 mm (Skutta, Furst, Eilers, Ferbert, & Kuhn, 1999;Villablanca et al., 2007), due to limited spatial resolution. Concern with branching arteries of <2 mm exists since slight differences in measurements might lead to a large difference in final estimation of degree of stenosis (Nguyen-Huynh et al., 2008). Also, dense and extensive mural calcifications may reduce the accuracy of measuring the degree of stenosis with CTA (Marquering, Nederkoorn, Bleeker, van den Berg, & Majoie, 2013). Several studies reported problems of CTA imaging with the vertebral artery and the ICA in the region of the skull base (Graf, Skutta, Kuhn, & Ferbert, 2000;Skutta et al., 1999). However, studies with more advanced CTA postprocessing techniques have shown the possibility to reliably depict vessels close to the skull base (Bash et al., 2005;Johnson, Heath, Kuszyk, & Fishman, 1996;Kuszyk, Heath, Johnson, Eng, & Fishman, 1999;Kuszyk et al., 1995). An overview of studies comparing CTA with DSA for identification of intracranial stenosis is presented in Table 1. Overall, studies show that CTA is useful as a screening tool for identification of ICAS and intracranial occlusion. It may therefore be used to exclude cases of ICAS, replacing the use of DSA in many cases. However, since some studies report a lower PPV of CTA compared with DSA, these authors suggest that DSA might be required for confirmation of CTA findings (Duffis et al., 2013;Graf et al., 2000;Liebeskind, Kosinski, Saver, & Feldmann, 2014;Nguyen-Huynh et al., 2008;Roubec et al., 2011).

| Time-of-flight MRA
Time-of-flight (TOF) MRA is a noninvasive technique that does not use any radiation nor contrast material to visualize arteries (Nederkoorn et al., 2002) (Figure 2). The main disadvantage is the high susceptibility of TOF-MRA to flow-related artifacts. Complete absence of MR signal in an artery can occur even though the vessel is not entirely occluded. Severe stenosis not only causes these flow-related artifacts in most cases but also in <70% stenosis these artifacts are reported (Nederkoorn et al., 2002), which influences the reported sensitivity and specificity values of MRA compared with DSA (Table 2).
Postprocessing techniques have been optimized throughout the years and MR scanners with increasingly higher magnetic field strengths (and thus increased spatial resolution) have been developed for use in clinical practice (Korogi et al., 1997;Willinek et al., 2004).
These optimized techniques are thought to underlie the improved ability of TOF-MRA to detect intracranial stenosis. Early studies with 1.5 T TOF-MRA compared with DSA showed sensitivity values ranging from 85% to 88% and specificity values from 86% to 99% for the detection of steno-occlusive disease (Furst et al., 1996;Korogi et al., 1994;Stock, Radue, Jacob, Bao, & Steinbrich, 1995). Although more recent studies report improved values of TOF-MRA compared with DSA in the detection of intracranial stenosis, other studies failed to show this. These discrepancies result in varying reported test characteristics, with sensitivity values ranging from 70% to 95%, specificity values from 95% to 99%, PPV from 59% to 84%, and NPV from 91% to 99% (Bash et al., 2005;Choi et al., 2007;Feldmann et al., 2007;Sadikin et al., 2007).
Even with newer postprocessing techniques, DSA is still required occasionally to confirm TOF-MRA findings (Sadikin et al., 2007).
Studies comparing TOF-MRA and CTA for its ability to identify intracranial stenosis and occlusion compared with DSA conclude that CTA is superior to TOF-MRA; CTA had higher sensitivity than MRA (98% compared with 70%) and a higher PPV than MRA (93% compared with 65%) (Bash et al., 2005).
The combination of CTA and MRA for detection of intracranial stenosis and occlusion compared with DSA has also been studied.
Although MRA alone had a sensitivity of 92% and a specificity of 91% for the detection of stenosis ≥50%, adding CTA increased these values to 100% and 99%, respectively, and resulted in a predictive value of  93% (Hirai et al., 2002). These are promising results for the identification of ICAS in ischemic stroke patients, indicating that a combination of different modalities could be able to replace DSA in the imaging of ICAS. However, more research is needed to validate the combined use of CTA and MRA in clinical practice.

| Contrast-enhanced MRA
Limited spatial resolution and sensitivity for correct timing of imaging after contrast administrations have been described as disadvantages of contrast-enhanced (CE) MRA (Wutke et al., 2002). Techniques to time the arrival of contrast are used to optimize the quality of CE MRA. Spatial resolution of CE MRA has been improved with newer techniques, including more efficient coil systems. Studies with these improved techniques report a sensitivity and specificity ranging from 90% to 100% and 76% to 99%, respectively (Nederkoorn et al., 2003;Willinek et al., 2005;Wutke et al., 2002). These values do not seem to indicate an advantage of CE MRA over optimized TOF-MRA (Nederkoorn et al., 2003).

| NOVEL TECHNIQUES
More detailed visualization of intracranial vessels is possible with advanced technological developments, for example, increased spatial resolution and improved postprocessing techniques. With high-resolution imaging, the focus is shifted toward more detailed pathological characterization of intracranial atherosclerosis, in addition to measurements of the vessel lumen (Arenillas, 2011;Chen, Wong, Lam, Zhao, & Ng, 2008).

| High-resolution computed tomography angiography
A new CTA technique called ultra-high resolution cone-beam CTA (CB-CTA) has been used for the evaluation of intracranial atherosclerotic stenosis. CBCT-A is an invasive catheter-based modality with a radiation dose similar to that of conventional CTA. CB-CTA did not differ significantly in assessing the absolute percent stenosis of lesions when compared with other 3D modalities such as 3D rotational angiography (Safain et al., 2014). However, both CB-CTA and 3D rotational angiography differed in the measurement of percent stenosis when compared with traditional 2D-DSA (Safain et al., 2014). Since intracranial stenosis is a three-dimensional phenomenon, it is possible that CB-CTA and 3D rotational angiography are more accurate than standard 2D-DSA. However, these findings from a single study (Safain et al., 2014) need further validation with DSA in a prospective manner.
In addition, CB-CTA provided greater resolution when compared with DSA. New information about plaque morphology was identified in more than 60% of patients when CB-CTA was used following standard DSA (Damaskos, Aartman, Tsiklakis, van der Stelt, & Berkhout, 2015;Safain et al., 2014). Whether the plaque morphology evaluated on CB-CTA has any additional benefit in comparison with conventional CTA and MRA still needs to be proven.
Promising results of intracranial vessel wall imaging have been published recently, in which 7T MRI was compared with histopathology. Areas of foamy macrophages were generally seen as proton attenuation-, T2-, and T2*-hypointense areas, whereas areas of increased collagen content showed more ambiguous signal intensities (van der Kolk, Zwanenburg, Denswil, et al., 2015). Several HR MRI studies described differences between symptomatic and asymptomatic patients with intracranial atherosclerotic plaques, thereby identifying plaque characteristics that are probably related to the development of symptoms (Chung et al., 2012;Ryu et al., 2009;Xu et al., 2010).
Further prospective studies with larger sample sizes are needed to confirm these promising results of HR MRI in intracranial atherosclerosis in recent studies. Although further validation is needed, it appears that HR MRI is likely the most promising technique so far to achieve detailed assessment of intracranial atherosclerosis imaging.

| Directions for future research
Further characterization of prognostic radiological features as well as differentiation of hemodynamic, embolic, and perforator lesion patterns could play an important role in patient selection and outcome in currently ongoing and future clinical trials. New trials are also necessary to assess the optimal therapy in patients with symptomatic ICAS and recurrent strokes in the setting of failed aggressive medical therapy. Identification of stroke mechanisms of intracranial atherosclerosis in relation to future ischemic events, could well lead to the identification of a high-risk subgroup of ICAS patients who might benefit from more aggressive treatment approaches.
F I G U R E 3 High-resolution MRI of vertebral artery stenoses with plaque components Panels A-D show T2weighted and T1 postcontrast images (panels C and D have plaque components marked) of a crosssection of a vertebral artery plaque with a thick, intact, fibrous cap (gray) and lipid core (white with black asterisk). Panels E-H show T2-weighted and T1 postcontrast images (panels G and H have plaque components marked) of a cross section of a vertebral artery plaque with a ruptured fibrous cap (gray) and lipid core (white with black asterisk), which enhances with contrast (white asterisk) and is also indicative of plaque rupture. The solid white line shows the outside vessel wall and the dashed white line the lumen "Reprinted from Holmstedt et al. (2013) with permission from Elsevier"

CONFLICTS OF INTEREST
None.