- Email: [email protected]
Assessment of collateral flow in patients with cerebrovascular disorders
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Journal of Neuroradiology (2013) xxx, xxx—xxx
Available online at
ScienceDirect www.sciencedirect.com
ORIGINAL ARTICLE
Assessment of collateral flow in patients with cerebrovascular disorders Joseph Donahue , Suna Sumer , Max Wintermark ∗ Division of Neuroradiology, University of Virginia, Charlottesville, VA, United States
KEYWORDS Collateral circulation; Cerebrovascular circulation; Perfusion Imaging; Cerebrovascular accident
Summary The ability to maintain cerebral parenchymal perfusion during states of acute or chronic ischemic insult depends largely on the capacity of the cerebral collateral circulation. Perfusion techniques, including perfusion-CT and arterial spin labeling, may not only describe the overall status of the collateral network, but can also quantify the pathophysiologic collateral reserve, which is occult to conventional imaging techniques. The following review details advanced imaging modalities capable of resolving pathophysiologic collateral circulation in a functional and dynamic manner, with regards to the evaluation of both acute ischemic penumbra and chronic cerebral vascular reserve. Specifically, the applications of perfusion-CT, arterial spin labeling MRI techniques, and transcranial Doppler are reviewed in the context of collateral circulation with emphasis on perfusion techniques and proposed clinical utility. © 2013 Published by Elsevier Masson SAS.
Introduction In general terms, cerebral collateral circulation is the network of vascular anastomoses providing supplemental blood flow in states of acute and/or chronic principal perfusion pathway insufficiency. Collateral circulation pathways encompass a number of entities which may be broadly categorized into large structural collateral vessels (primary and secondary) and pathophysiological collateral vessels [1].
Abbreviations: CTA, computed tomography angiography; MRA, magnetic resonance angiography; MTT, mean transit time; rCBF, regional cerebral bold flow; rCBV, regional cerebral blood volume; ASL, arterial spin labeling. ∗ Corresponding author. Radiology and Medical Imaging P.O. Box 800170 [FedEx: 1215 Lee Street] Charlottesville, VA 22908, United States. Tel.: +1 434 243 9312. E-mail address: [email protected] (M. Wintermark).
The Circle of Willis constitutes the primary structural collateral circulation pathway between the hemispheres, as well as between the anterior and posterior circulations. The Circle of Willis, and its well-known variations or deficiencies, are readily detailed by conventional imaging modalities such as CTA/MRA and direct catheter angiography [1,2]. Secondary structural collateral pathways include both extracranial and intracranial anastomoses, which are expected normal anatomic vascular structures that may become hypertrophied with ischemic stimulus. The largest and most clinically relevant example of these secondary structural collateral pathways is the intracranial leptomeningeal anastomoses between the distal cortical branches of the intracranial arteries. These vessels are resolvable with CTA and MRA techniques [1,3,4]. Pathophysiologic collateral vessels are potential anastomotic pathways, which are recruited over time by ischemic parenchyma when principle arterial pathways are insufficient to provide stabilized tissue perfusion [1]. Although not
0150-9861/$ – see front matter © 2013 Published by Elsevier Masson SAS. http://dx.doi.org/10.1016/j.neurad.2013.11.002
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
2
J. Donahue et al.
necessarily derived from a specific and anatomically constant arterial plexus, the pathophysiologic collaterals may be of significant importance to the overall cerebrovascular reserve capacity. Cerebral collateral circulation is a dynamic entity on a number of levels. Within seconds of a large artery occlusion, changes in the direction and velocity of flow within the primary structural collaterals are mediated by changes in blood pressure gradients and reflexive vasodilation. Such changes are noted to occur within several heart beats during carotid endarterectomy. An incompletely understood complex interplay of metabolic and hemodynamic factors are proposed to initiate and sustain recruitment of secondary structural and pathophysiological collateral vessels in states of chronic hypoperfusion. Additionally, various comorbidities such as hypertension may variably decelerate or endanger incipient collateral network growth and the overall tissue perfusion capacity [2]. The dynamic nature of these cerebrovascular reserve components emphasizes the need for a multifaceted interrogation of collateral vessel status. CTA, MRA, and direct catheter angiography are all well recognized and accurate modalities for the evaluation of structural primary and secondary collateral vessels [5]. However, the parenchymal perfusion capacity of the pathophysiologic collateral pathways may not necessarily be inferred from the status of primary and secondary structural collateral vessels [6]. As such, during periods of ischemia, tissue fate may depend on the status of these hemodynamically significant pathophysiologic collateral pathways that are occult to conventional vascular imaging modalities [7]. With this in mind, the following discussion is intended to explore the role of perfusion-CT in the assessment of cerebral collateral circulation with regards to two specific applications: acute ischemic stroke and chronic cerebrovascular disease.
Assessment of cerebral collateral circulation in acute ischemic stroke The adequacy of collateral circulation in acute ischemic stroke is an important determinant of the fate of the ischemic tissue and the prognosis for vascular recanalization [8,9]. The blood flow supplied by the collateral arterial network helps maintain misery perfusion in areas of ischemic penumbra. Robust collaterals also likely facilitate the ‘‘wash-out’’ of thrombi fragments and mitigates against infarct progression. Similarly, collateral arterioles may also help deliver thrombolytic agent to more distal thrombotic regions, thereby assisting in more rapid recanalization.
Functional imaging of cerebral collateral circulation Perfusion-CT In states of hemodynamic perturbation, blood may arrive to parenchyma via collateral pathways in antegrade or retrograde direction; however, this is difficult to distinguish in practice. Perfusion-CT complements angiographic techniques by offering a dynamic assessment of parenchymal
Figure 1 Interpretation of perfusion-CT maps. The green and red boxes highlight key differences in perfusion characteristics of endangered yet salvageable ischemic penumbra and irreversible ischemic core.
perfusion, regardless of occlusion site or directionality of blood flow [10]. This technique allows for the assessment of the pathophysiologic collateral vessels as inferred from the ischemic penumbra estimate [6,10]. Perfusion-CT monitors the parenchymal wash-in and wash-out of an iodinated contrast material bolus with serial acquisitions, usually at several supratentorial levels. In the post-processing stage, the investigator or the processing software select a reference artery input and vein output functions. A series of time-concentration curves are then generated for each voxel, and perfusion-CT parameters are generated that describe these curves in each voxel: mean transit time (MTT), regional cerebral blood volume (rCBV), and regional cerebral blood flow (rCBF). MTT designates the average time required for blood cell passage through the parenchymal capillary network. rCBV reflects the volume of blood per unit of parenchyma, mL/100 g. rCBF is a rate unit derived from rCBV/MTT. MTT is the most sensitive perfusion indicator of ischemia, with changes in cerebral blood flow and cerebral blood volume are more specific for distinguishing penumbra from ischemic core [9]. As detailed in Fig. 1, MTT is elevated in all ischemic cerebrovascular events. Parenchymal autoregulatory mechanisms initially compensate with reactive vasodilation and collateral flow recruitment, thus increasing rCBV. Again, as rCBF is derived from rCBV/MTT, a proportional increase in both the numerator and denominator will result in an unchanged, normal rCBF value. This is a pattern typically observed during a transient ischemic attack. With worsening and more prolonged ischemia, MTT becomes more prolonged and rCBF eventually decreases. Where collateral circulation is robust enough to maintain rCBV, ischemic penumbra exists; that is, tissue at risk for infarction but that is still currently salvageable. When the autoregulatory mechanisms are overwhelmed such that the collateral circulation is inadequate to maintain threshold tissue blood flow, the rCBV falls and an irreversible infarct develops [10—12]. These principles are illustrated in Fig. 2 with the perfusion-CT map of an acute right MCA territory infarct. In the subcortical areas of ischemic penumbra, the leptomeningeal collateral circulation is adequate to maintain rCBV. However, the caudate head and putamen, perfused by an inadequately collateralized lenticulostriate system, demonstrate decreased rCBV and thus constitute the ischemic core.
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders
3
Figure 2 Perfusion-CT map of an acute left MCA territory infarct. Corpus striatum infarct core (bracketed on rCBV panel) corresponding to areas of inadequate lenticulostriate collateral capacity. The majority of the penumbra (bracketed on MTT and rCBF) is within the perfusion territory of robust leptomeningeal collaterals.
Arterial spin labeling MRI Arterial spin labeling is an advanced noncontrast MRI technique in which selective radiofrequency labeling of arterial serum protons produce endogenous vascular contrast signal that is subtracted from a control image. Generally, proton labeling is performed in a labeling plane upstream from the region of interest (i.e. extracranial vasculature labeled in a plane parallel to the skull base), and a post labeling delay is inserted prior to read out, allowing labeled blood to reach the parenchyma. Multiple acquisitions may be performed utilizing different post labeling delays to characterize arterial transit time and more closely interrogate cerebral blood flow parameters [13,14]. As the overall signal differences between control and labeled series are typically low (0.5—2%), multiple subtraction data sets are obtained and averaged to generate perfusion maps [15]. There are several generations of MR spin labeling techniques: continuous (CASL), pulsed (PASL), and pseudocontinuous (pCASL). CASL is a first generation technique that utilizes a continuous (1—2 seconds) RF pulse applied to the extracranial vasculature utilizing surface coils and dedicated scanner hardware [16]. This method has lost favor in clinical utility due to relatively low labeling efficiency, in part relating to often irregular vessel geometry and imperfect labeling plane perpendicularity [17]. PASL is a labeling-efficient and reproducible second generation technique in which a large volume of blood is prepared with a short (∼10 ms) radiofrequency pulse within a thick labeling slab that is selected for each major vascular territory [17]. However, labeling slab planning requires advanced knowledge of arterial anatomy, and often necessitates a large field of view time of flight (TOF) planning angiogram to minimize contamination from adjacent non-target arterial structures. Misestimates of perfusion parameters may be encountered with PASL slab labeling contamination in which extracranial vessel tortuosity precludes a pure vascular territory sample. This is potentially confounding in states of chronic cerebrovascular disease in which extracranial collateralization is often present. pCASL is a third generation hybrid of earlier techniques employing a train of short RF labeling pulses that achieve target artery specificity with gradient induced variations in vessel labeling efficiency. Superselective ASL is a pCASL derived technique that utilizes rotational modulation of
gradients to produce a focal labeling point with a high spatial discriminatory capacity sufficient to interrogate individual vessels above the Circle of Willis [16]. A limitation inherent to ASL has particular importance to parenchymal perfusion mapping. All ASL techniques utilize selective arterial proton spin inversion as the basis of endogenous vascular contrast resolution (i.e. no intravenous gadolinium based contrast agent is used). This however necessarily means that contrast signal decays with the T1 relaxation time of blood, approximately 1.6 seconds at 3 Tesla. As such, perfusion of parenchymal regions beyond this arterial transit time may be underestimated [15]. This limitation is especially pertinent to the evaluation of arterial collateralization in which upstream steno-occlusive conditions may artifactually suggest decreased CBF and collateral reserve insufficiency. Of the strategies currently available to address this limitation, the utilization of higher field strength systems is most intuitive, as the contrast decay is somewhat mitigated by prolonged T1 relaxivity. Additionally, Velocity Selective (VS-ASL) is a technique without post labeling delay, in which a velocity sensitive pulses invert blood proton spins traveling in excess of a designated cutoff value, and read out is performed using those spins that decelerate to below the cutoff velocity value; in this fashion the issue of signal decay due to post labeling delay is theoretically minimized. Nonetheless, ASL is a non-invasive non-contrast technique capable of obtaining quantitative perfusion data with specificity for individual vascular territories. ASL CBF measurements generally correlate most closely with dynamic susceptibility contrast MRI mean transit time and time to peak parameters [18]. However, in part due to the aforementioned limitations, ASL tends to overestimate the volume of perfusion abnormalities and has lower specificity for mismatch status than dynamic contrast susceptibility perfusion weighted imaging (DCS-PWI) [19,20]. Transcranial Doppler Initially introduced as a non-invasive bedside technique to detect vasospasm following subarachnoid hemorrhage, transcranial Doppler demonstrates utility in describing intracranial arterial patency and collateralization status. Although generally not as sensitive to intracranial stenoses as CTA, transcranial Doppler offers the advantage or real
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
4
J. Donahue et al.
Figure 3 Operational penumbra is an imaging biomarker of tissue that is at risk for infarction (green text), but not yet irreversibly damaged (i.e. tissue that will recover with and infarct without reperfusion).
time evaluation and specificity of flow directionality. Identification of flow diversion, a low-resistance high velocity pattern, in the presence of known steno-occlusive disease correlates well with the presence of leptomeningeal collaterals. Limitations of the technique, as with all ultrasound, include operator variability and acoustic window availability. There are four commonly utilized acoustic windows in adults, the temporal, suboccipital, orbital, and submandibular approaches. In some inpatient settings, these acoustic windows may be obscured by bandaging, or otherwise inaccessible due to anatomic variability [21—24].
The ischemic penumbra and collateral circulation The utility of perfusion imaging in regards to the assessment of cerebral collateral circulation stems from the many similarities that collateralization shares with the ischemic penumbra. Ischemic penumbra is a multifactorial continuum of various states of parenchymal viability. The severity and reversibility of neuronal metabolic and electrical derangements, the integrity of parenchymal capillary beds, and progression of apoptotic pathways are all determinants of tissue fate. These complexities are of course not individually detailed; however, perfusion imaging provides a practical, operational depiction of salvageable tissue. Operational penumbra is a trichotomous imaging biomarker used to distinguish endangered tissue that will recover with, and infarct without reperfusion from the infarct core and surrounding normal parenchyma (Fig. 3). The operational concept of ischemic penumbra is informed by the perfusion threshold values for tissue at risk, which were derived through outcomes observation of patients presenting with acute ischemic stroke that subsequently did or did not attain reperfusion [25]. Thus, operational penumbra is influenced by the method and efficacy of revascularization and may vary depending of locoregional tissue conditions such as temperature, metabolic susceptibility, genetics, etc. Although not 100% accurate in predicting tissue fate given these variables, the operational penumbra concept is effective for guiding treatment decisions. As less than 40% of patients suffering from acute ischemic stroke present within the established time windows for systemic thrombolysis, an imaging determinant of ischemic penumbra is useful in instances of unclear time of symptom onset, such as ‘‘wake up stroke.’’ Establishing the functionality of arterial collaterals and operation penumbra with perfusion imaging is facilitating a promising paradigm shift from a ‘‘time is brain’’ to a ‘‘penumbra or imaging is brain’’ approach to acute ischemic stroke triage and management [11,26].
Figure 4 Examples of two patients, both with right M1 occlusion. Ischemic core is designated in red; ischemic penumbra shown as green. Despite the same level of large vessel occlusion, the absolute size of the ischemic area and the relative proportions of ischemic core to ischemic penumbra vary considerably due to the differing leptomeningeal collateral arterial adequacy. Patient B, who did not achieve recanalization, suffered penumbra evolution towards infarct [7].
Collateral blood flow shares similarities with, but is distinct from, the ischemic penumbra. Penumbra measurements provide only an indirect quantitation of the effectiveness of collateral circulation. In Fig. 4, CTA maximum intensity projection images and perfusion-CT data overlay of two patients with right M1 MCA occlusions are shown. In patient A, with good leptomeningeal collaterals, the MCA occlusion translates to a smaller total ischemic area and smaller ischemic penumbra. In patient B, with poor leptomeningeal collaterals, the occlusion translates to a large ischemic area and large ischemic penumbra, which without recanalization progressed to a large territory infarct. These two cases illustrate the fact that the site of occlusion, by itself, does not allow prediction of the infarct volume or proportion of penumbra [7]. This variability is due to differences in the adequacy of collateralization. Fig. 5 furthers this important distinction between penumbra and structural arterial collateralization. Perfusion-CT derived ischemic penumbra and infarct core volumes from
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders
5
Figure 5 Infarct core (red) and ischemic penumbra (green) volume plots in 165 patients with acute M1 or ICA occlusions illustrating the highly variable penumbra volume for any given ischemic core volume [6].
165 patients who sustained acute proximal large vessel occlusion, either M1 or ICA terminus, are plotted. This graph illustrates the extreme variability of observed penumbra volume for any given infarct core volume. Similar variability is also observed when penumbra volume is plotted against volume of total ischemic area [6]. This illustrates two important concepts. First, good collateral circulation does not necessarily translate to a large penumbra; low volume of penumbra may exist precisely because the collateral circulation is robust enough to maintain relatively normal rates of perfusion within the tissue adjacent to an infarct core. Secondly, there is no fixed relationship between site of occlusion, ischemic core volume, and penumbra volume. Even given similar sites of large vessel occlusion, the volume and relative proportion of penumbra to ischemic core is highly variable. Again, this variability is due to the fact that collateral circulation differs between patients. However, even with the combination of clinical data, noncontrast head CT, and CTA based structural collateral vessel data, penumbra is still not predicted accurately [4,6]. This suggests that interrogation of structural arterial collateralization solely with CTA or other anatomic modalities is insufficient to accurately infer the status of pathophysiologic collaterals and the overall parenchymal perfusion capacity of the collateral network. In other words, critical and predictive information regarding the functional capacity of the collateral vessels that sustain ischemic penumbra is lost when evaluation of the collateral network is limited to structural scoring [6,25,27]. To this end, arterial spin labeling offers potentially complementary information in regards to parenchymal collateralization status during acute ischemic events. The value that ASL may add in characterizing collateral perfusion networks is not solely in redemonstrating areas of oligemia and infarct. Rather, territorial ASL is uniquely capable of noninvasively mapping the individual perfusion regions subtended by the major vasculature, even in states of extensive collateralization [28,29]. In distinction to conventional perfusion CT and MR, which cannot precisely determine the origin or direction (antegrade or retrograde) of collateral flow, ASL may describe the presence and degree of
collateralization contributed to the ischemic penumbra by each individual vessel [30,31]. The precision of ASL for delineating distal collateral perfusion territory is comparable to digital subtraction angiography when combined with time of flight MRA, even in the presence of advanced steno-occlusive disease and Circle of Willis deficiencies [29,32]. When given the knowledge of actual vascular territory boundaries, the importance of watershed or borderzone physiology between differentially collateralized regions becomes evident. Hendrikse et al. found that diffusion abnormalities localize to perfusion territory interfaces significantly more frequently than the currently estimated 2—5% that is predicted by structural anatomic imaging [28,33]. Thus with ASL, classification of infarcts may potentially be more precisely subdivided into thromboembolic or hypoperfusional etiologies [30,31]. Beyond describing the anatomic boundaries of collateralized perfusion territories, ASL can offer insight into pathophysiologic collateral function. In general, the fast decay of ASL contrast signal proportional to T1 relaxation, among various other technical parameters, results in the oversensitivity of the technique to even benign oligemic perfusion delays, and typically results in the over estimation of perfusion abnormality volume in the acute stroke setting [18,20]. This however, is the basis of the high negative predictive value of ASL for the presence of perfusional abnormalities [20]. Not surprisingly, ASL demonstrates high sensitivity for perfusion anomalies corresponding to transient ischemic attack symptoms, likely exceeding the sensitivity of combined contrast bolus perfusion MR and MR angiographic evaluations [34—36]. Thus ASL can help localize incipient deficiencies in parenchymal collateralization that may herald future infarct.
Perfusion-CT based assessment of cerebral collateral circulation in chronic cerebrovascular disease A major aim in the evaluation of patients with chronic cerebrovascular conditions is to accurately predict a patient’s individual risk of future stroke. The relevance of
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
6
J. Donahue et al.
perfusion-CT in this regard is related to the importance of a refined assessment of collateral circulation capacity. The recruitment of collaterals to serve areas of chronic ischemia is protective not only due to stabilized parenchymal perfusion capacity, but also for the ability to help ‘‘wash-out’’ thrombi from the small vessel circulation. As discussed below, the dynamic element that perfusion-CT adds to the imaging work up of such patients enhances future stroke risk stratification.
Concept of parenchymal perfusion delay During perfusion-CT post-processing, an arterial and venous reference vessel is selected prior to the generation of time concentration curves. The arterial reference vessel serves as the primary input function for the generation of the MTT, while the selected reference vein is the principle input for the rCBV calculation. The vessel designated as the arterial input function is assumed to represent the only vascular supply to a tissue voxel. It is assumed that the contrast bolus that is delivered by the reference artery arrives to the tissue without delay. Selection of a reference artery with a hemodynamically significant upstream stenosis may produce appreciable regional perfusion abnormalities characterized by increased MTT and decreased rCBF. In such cases, this could potentially be confused with acute ischemia [37,38]. Additionally, the artifact produced by the delay of the contrast bolus arrival may be compounded by the dispersion of the contrast material throughout multiple small collateral vessels proximal to the site of stenosis. Multiple delay correction strategies have been devised to normalize the parenchymal time concentration curves to account for bolus delay and dispersion. Some methods utilize various deconvolution algorithms that relax the assumptions regarding the exclusivity of reference vessel contrast material delivery to a tissue voxel, while other utilize more distal arterial input functions that are closer to each displayed parenchymal territory [38]. Fig. 6 presents a perfusion-CT data set of a patient with a chronic high-grade stenosis of the right middle cerebral artery, with and without application of the delay correction program. As seen in the setting of acute ischemia, the parenchyma downstream to the arterial stenosis demonstrates decreased rCBF and increased MTT on uncorrected data sets. The rCBV, which is a function of venous outflow, remains normal. With the application of delay correction, the perfusion parameters of the affected territory are normalized and displayed as equivalent to neighboring territories with normal arterial vascular supplies. The potential for introducing artifact during postprocessing has traditionally been considered a limitation of perfusion-CT. In fact, much debate exists in the literature regarding the standardization of reference vessel selection precisely due to the potential variability in the perfusion maps as a function of reference vessel laterality and patency. However, the mismatch between the corrected and uncorrected perfusion maps contains important data, and ought not to be automatically discarded. The delay in tissue perfusion depicted in the corresponding time-concentration curve plots (Fig. 7) demonstrates the basis of quantitative collateral assessment.
Figure 6 Perfusion-CT data in patient with a high-grade chronic right MCA stenosis. In the first column, the ACA has been selected as the arterial input function; the right MCA territory demonstrates decreased MTT and rCBF secondary to the right MCA stenosis. In the second column, each vascular territory has an individually selected arterial input function and has been processed with a delay correction algorithm. The perfusion values of the right MCA territory have been normalized to account for contrast bolus delay and dispersion related to the right MCA stenosis. Note that rCBV is a function of venous output, and is unaffected by the arterial stenosis.
Figure 7 Time-concentration curve plot of the same patient with a high-grade chronic right MCA stenosis. The ACA and left MCA demonstrate early contrast material arrival and prompt time to peak, consistent with early wash-in and wash-out. The high-grade right MCA stenosis causes delayed parenchymal contrast density. The delay is a quantitative reflection of the net collateral arterial capacity available to the right MCA territory [37].
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders
7
Figure 8 Correlation of leptomeningeal collateral score and Circle of Willis collateral scores with MTT delay based physiologic collateral assessment. Bars represent medians, boxes represent interquartile ranges, and error bars represent 95% CIs. MTT indicates mean transit time [37].
The delay in perfusion between the territory supplied by the stenotic artery and the remainder of the unaffected vascular territories reflects a sum of the collateral pathways available to the right MCA territory. Although the directionality of the collateral arterial flow remains indeterminate (i.e. may be antegrade through the stenotic artery or retrograde via leptomeningeal collaterals), the perfusion delay is an independent quantitation of net arterial collateral capacity [38]. In other words, the delay in MTT, which is related to contrast bolus delay and dispersion, provides a dynamic assessment of total collateral circulation and cerebrovascular reserve [37,38].
Clinical significance of parenchymal perfusion delay This parenchymal perfusion delay is an independent variable with prognostic clinical utility that may not be captured by traditional structural evaluation methods. Even in patients without a history of prior stroke, MTT delay of greater than 0.5 seconds between MCA territories was highly predictive of new, incident infarcts [37]. The prognostic value of MTT delay for occurrence of subsequent infarct is not necessarily inferable from CTA leptomeningeal collateral score. Although certain malignant CTA profiles have demonstrated prognostic value for later infarct, as shown in Fig. 8, the correlation of MTT delay with leptomeningeal collateral scores is marginal [7,37]. Similarly, the patency and completeness of the Circle of Willis shows weak correlation with MTT delay. As expected, the less complete the Circle of Willis or the lower the leptomeningeal collateral score of the affected hemisphere, the greater the MTT delay [37]. When taken together, this data suggest that pathophysiologic collaterals indeed play a pivotal role in the disease process of chronic cerebrovascular disease. However, the relatively coarse and ordinal data provided by CTA based structural assessments generally lack the fine discriminatory capabilities to create refined patient risk stratification models. Perfusion-CT has a unique ability among CT applications to characterize the functional status
of the pathophysiologic collateral pathways with a continuous and quantitative data [7,10,37]. As with the interrogation of collateralization status in the acute ischemic setting, ASL provides complementary information regarding pathophysiologic collateral function in the setting of chronic cerebrovascular disease. Again, in cases of significant steno-occlusive disease, collateralization may result in deviation of actual perfusion territories from expected anatomic vascular regions. Territorial ASL technique has demonstrated that approximately 10% of cortical infarcts may be misclassified by conventional anatomic designations due to collateralization, meaning that when a perfusion delay is identified, there is a substantial possibility of misidentifying the responsible vessel [28]. ASL can accurately discriminate between not only vertebrobasilar and internal carotid artery perfusion territories, but also delineate external carotid artery collateral contributions and quantitate the function of extra-anatomic bypass grafts with superselective vessel mapping techniques [39—41]. Moreover, when a perfusion delay is recognized, ASL may help characterize the cerebrovascular reserve capacity, further stratifying the risk of subsequent infarct or likely response to intervention [42,43]. As ASL is a noncontrast technique, same session pre- and post-acetazolamide challenge assessment may be performed to determine the vasodilatory reserve of the pathophysiologic collateral pathways [15,44]. In this manner, the anatomic specificity of MRI may be combined with a functional assessment of cerebrovascular reserve capacity, a variable with significant prognostic value that has been validated in studies utilizing transcranial Doppler, perfusion CT, and other modalities [44,45].
Summary Cerebral collateral circulation encompasses the entirety of the vascular networks that supply parenchymal blood flow, from the primary structural large vessels to the inconstant and variably recruited pathophysiological collaterals. Although no single imaging modality fully characterizes the
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
8
J. Donahue et al.
complexity and pathophysiological significance of cerebral collateral circulation, the capacity of perfusion-CT and arterial spin labeling MRI to depict the functional status of the collateral network has relevant applications in both acute and chronic ischemic states. More specifically, the unique ability of these modalities to resolve pathophysiologic collateral circulation in a functional and dynamic manner informs evaluation of both acute ischemic penumbra and chronic cerebral vascular reserve. Though not fully reflecting the intricacies of the complex metabolic determinants of ischemic tissue fate, perfusion-CT provides a practical context of operational penumbra in the evaluation of acute stroke. This imaging biomarker is a quantitative proxy for the effectiveness of net collateral arterial function. Similarly, the characterization of parenchymal perfusion delay that perfusion-CT offers yields a continuous and quantitative metric basis for future ischemic event risk stratification models. Arterial spin labeling MRI provides a quantitative parenchymal perfusion assessment with the capacity to discriminate between vascular territories, thus identifying the anatomic extent of collateralization and the respective watershed territories. Additionally, arterial spin labeling MRI furthers the assessment of pathophysiologic collateral function as determined by the cerebrovascular reserve capacity.
Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.
Acknowledgements Many thanks to Drs. Tan, Keedy, Wintermark, and Srinivasan for the gracious use of figures and tables from the cited references.
References [1] Liebeskind DS. Collateral circulation. Stroke 2003;34(9): 2279—84. [2] Romero JR, et al. Cerebral collateral circulation in carotid artery disease. Curr Cardiol Rev 2009;5(4):279—88. [3] Souza LC, et al. Malignant CTA collateral profile is highly specific for large admission DWI infarct core and poor outcome in acute stroke. AJNR Am J Neuroradiol 2012;33(7):1331—6. [4] Tan IY, et al. CT angiography clot burden score and collateral score: correlation with clinical and radiologic outcomes in acute middle cerebral artery infarct. AJNR Am J Neuroradiol 2009;30(3):525—31. [5] Smith WS, et al. Significance of large vessel intracranial occlusion causing acute ischemic stroke and TIA. Stroke 2009;40(12):3834—40. [6] Zhu G, et al. Computed tomography workup of patients suspected of acute ischemic stroke: perfusion computed tomography adds value compared with clinical evaluation, non-contrast computed tomography, and computed tomography angiogram in terms of predicting outcome. Stroke 2013;44(4):1049—55. [7] Tan JC, et al. Systematic comparison of perfusion-CT and CT-angiography in acute stroke patients. Ann Neurol 2007;61(6):533—43.
[8] Liebeskind DS, et al. Collaterals dramatically alter stroke risk in intracranial atherosclerosis. Ann Neurol 2011;69(6): 963—74. [9] Gonzalez RG, et al. The Massachusetts General Hospital acute stroke imaging algorithm: an experience and evidence based approach. J Neurointerv Surg 2013;5(Suppl. 1):pi7—12. [10] Wintermark M, et al. Prognostic accuracy of cerebral blood flow measurement by perfusion computed tomography, at the time of emergency room admission, in acute stroke patients. Ann Neurol 2002;51(4):417—32. [11] Leiva-Salinas C, et al. Tissue at risk in acute stroke patients treated beyond 8 h after symptom onset. Neuroradiology 2013;55(7):807—12. [12] Dankbaar JW, et al. Dynamic perfusion-CT assessment of early changes in blood brain barrier permeability of acute ischaemic stroke patients. J Neuroradiol 2011;38(3):161—6. [13] MacIntosh BJ, et al. Assessment of arterial arrival times derived from multiple inversion time pulsed arterial spin labeling MRI. Magn Reson Med 2010;63(3):641—7. [14] Bokkers RP, et al. Arterial spin-labeling MR imaging measurements of timing parameters in patients with a carotid artery occlusion. AJNR Am J Neuroradiol 2008;29(9):1698—703. [15] Hendrikse J, Petersen ET, Golay X. Vascular disorders: insights from arterial spin labeling. Neuroimaging Clin N Am 2012;22(2):259—69, x-xi. [16] Hartkamp NS, et al. Mapping of cerebral perfusion territories using territorial arterial spin labeling: techniques and clinical application. NMR Biomed 2013;26(8):901—12. [17] Chen Y, Wang DJ, Detre JA. Test-retest reliability of arterial spin labeling with common labeling strategies. J Magn Reson Imaging 2011;33(4):940—9. [18] Niibo T, et al. Arterial spin-labeled perfusion imaging to predict mismatch in acute ischemic stroke. Stroke 2013;44(9): 2601—3. [19] Nael K, et al. Quantitative analysis of hypoperfusion in acute stroke: arterial spin labeling versus dynamic susceptibility contrast. Stroke 2013. [20] Zaharchuk G, et al. Comparison of arterial spin labeling and bolus perfusion-weighted imaging for detecting mismatch in acute stroke. Stroke 2012;43(7):1843—8. [21] Bathala L, Mehndiratta MM, Sharma VK. Cerebrovascular ultrasonography: technique and common pitfalls. Ann Indian Acad Neurol 2013;16(1):121—7. [22] Bathala L, Mehndiratta MM, Sharma VK. Transcranial doppler: technique and common findings (Part 1). Ann Indian Acad Neurol 2013;16(2):174—9. [23] Kim Y, et al. Relationship between flow diversion on transcranial Doppler sonography and leptomeningeal collateral circulation in patients with middle cerebral artery occlusive disorder. J Neuroimaging 2009;19(1):23—6. [24] Kirsch JD, et al. Advances in transcranial Doppler US: imaging ahead. Radiographics 2013;33(1):E1—14. [25] Zhu G, et al. Does perfusion imaging add value compared with plain parenchymal and vascular imaging? J Neurointerv Surg 2012;4(4):246—50. [26] Michel P, et al. Perfusion-CT guided intravenous thrombolysis in patients with unknown-onset stroke: a randomized, double-blind, placebo-controlled, pilot feasibility trial. Neuroradiology 2012;54(6):579—88. [27] McVerry F, Liebeskind DS, Muir KW. Systematic review of methods for assessing leptomeningeal collateral flow. AJNR Am J Neuroradiol 2012;33(3):576—82. [28] Hendrikse J, et al. Relation between cerebral perfusion territories and location of cerebral infarcts. Stroke 2009;40(5):1617—22. [29] Wu B, et al. Collateral circulation imaging: MR perfusion territory arterial spin-labeling at 3T. AJNR Am J Neuroradiol 2008;29(10):1855—60.
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders [30] Petcharunpaisan S, Ramalho J, Castillo M. Arterial spin labeling in neuroimaging. World J Radiol 2010;2(10):384—98. [31] van Laar PJ, van der Grond J, Hendrikse J. Brain perfusion territory imaging: methods and clinical applications of selective arterial spin-labeling MR imaging. Radiology 2008;246(2):354—64. [32] Chng SM, et al. Territorial arterial spin labeling in the assessment of collateral circulation: comparison with digital subtraction angiography. Stroke 2008;39(12):3248—54. [33] Momjian-Mayor I, Baron JC. The pathophysiology of watershed infarction in internal carotid artery disease: review of cerebral perfusion studies. Stroke 2005;36(3):567—77. [34] MacIntosh BJ, et al. Multiple inflow pulsed arterial spinlabeling reveals delays in the arterial arrival time in minor stroke and transient ischemic attack. AJNR Am J Neuroradiol 2010;31(10):1892—4. [35] Zaharchuk G, et al. Arterial spin labeling imaging findings in transient ischemic attack patients: comparison with diffusion- and bolus perfusion-weighted imaging. Cerebrovasc Dis 2012;34(3):221—8. [36] Zaharchuk G. Arterial spin label imaging of acute ischemic stroke and transient ischemic attack. Neuroimaging Clin N Am 2011;21(2):285—301. [37] Keedy AW, et al. Contrast delay on perfusion CT as a predictor of new, incident infarct: a retrospective cohort study. Stroke 2012;43(5):1295—301.
9
[38] Konstas AA, et al. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: Theoretic basis. AJNR Am J Neuroradiol 2009;30(4):662—8. [39] Helle M, et al. Superselective arterial spin labeling applied for flow territory mapping in various cerebrovascular diseases. J Magn Reson Imaging 2013;38(2):496—503. [40] Dang Y, et al. Quantitative assessment of external carotid artery territory supply with modified vessel-encoded arterial spin-labeling. AJNR Am J Neuroradiol 2012;33(7): 1380—6. [41] Wintermark M, et al. Comparative overview of brain perfusion imaging techniques. J Neuroradiol 2005;32(5): 294—314. [42] Golay X, Hendrikse J, Van Der Grond J. Application of regional perfusion imaging to extra-intracranial bypass surgery and severe stenoses. J Neuroradiol 2005;32(5):321—4. [43] de Castro-Afonso LH, et al. Carotid artery stenting performed with a flow-reversal technique: improved technical performance. J Neuroradiol 2013;40(1):29—37. [44] Bokkers RP, et al. Cerebrovascular reactivity within perfusion territories in patients with an internal carotid artery occlusion. J Neurol Neurosurg Psychiatry 2011;82(9):1011—6. [45] Silvestrini M, et al. Impaired cerebral vasoreactivity and risk of stroke in patients with asymptomatic carotid artery stenosis. JAMA 2000;283(16):2122—7.
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
ARTICLE IN PRESS
Journal of Neuroradiology (2013) xxx, xxx—xxx
Available online at
ScienceDirect www.sciencedirect.com
ORIGINAL ARTICLE
Assessment of collateral flow in patients with cerebrovascular disorders Joseph Donahue , Suna Sumer , Max Wintermark ∗ Division of Neuroradiology, University of Virginia, Charlottesville, VA, United States
KEYWORDS Collateral circulation; Cerebrovascular circulation; Perfusion Imaging; Cerebrovascular accident
Summary The ability to maintain cerebral parenchymal perfusion during states of acute or chronic ischemic insult depends largely on the capacity of the cerebral collateral circulation. Perfusion techniques, including perfusion-CT and arterial spin labeling, may not only describe the overall status of the collateral network, but can also quantify the pathophysiologic collateral reserve, which is occult to conventional imaging techniques. The following review details advanced imaging modalities capable of resolving pathophysiologic collateral circulation in a functional and dynamic manner, with regards to the evaluation of both acute ischemic penumbra and chronic cerebral vascular reserve. Specifically, the applications of perfusion-CT, arterial spin labeling MRI techniques, and transcranial Doppler are reviewed in the context of collateral circulation with emphasis on perfusion techniques and proposed clinical utility. © 2013 Published by Elsevier Masson SAS.
Introduction In general terms, cerebral collateral circulation is the network of vascular anastomoses providing supplemental blood flow in states of acute and/or chronic principal perfusion pathway insufficiency. Collateral circulation pathways encompass a number of entities which may be broadly categorized into large structural collateral vessels (primary and secondary) and pathophysiological collateral vessels [1].
Abbreviations: CTA, computed tomography angiography; MRA, magnetic resonance angiography; MTT, mean transit time; rCBF, regional cerebral bold flow; rCBV, regional cerebral blood volume; ASL, arterial spin labeling. ∗ Corresponding author. Radiology and Medical Imaging P.O. Box 800170 [FedEx: 1215 Lee Street] Charlottesville, VA 22908, United States. Tel.: +1 434 243 9312. E-mail address: [email protected] (M. Wintermark).
The Circle of Willis constitutes the primary structural collateral circulation pathway between the hemispheres, as well as between the anterior and posterior circulations. The Circle of Willis, and its well-known variations or deficiencies, are readily detailed by conventional imaging modalities such as CTA/MRA and direct catheter angiography [1,2]. Secondary structural collateral pathways include both extracranial and intracranial anastomoses, which are expected normal anatomic vascular structures that may become hypertrophied with ischemic stimulus. The largest and most clinically relevant example of these secondary structural collateral pathways is the intracranial leptomeningeal anastomoses between the distal cortical branches of the intracranial arteries. These vessels are resolvable with CTA and MRA techniques [1,3,4]. Pathophysiologic collateral vessels are potential anastomotic pathways, which are recruited over time by ischemic parenchyma when principle arterial pathways are insufficient to provide stabilized tissue perfusion [1]. Although not
0150-9861/$ – see front matter © 2013 Published by Elsevier Masson SAS. http://dx.doi.org/10.1016/j.neurad.2013.11.002
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
2
J. Donahue et al.
necessarily derived from a specific and anatomically constant arterial plexus, the pathophysiologic collaterals may be of significant importance to the overall cerebrovascular reserve capacity. Cerebral collateral circulation is a dynamic entity on a number of levels. Within seconds of a large artery occlusion, changes in the direction and velocity of flow within the primary structural collaterals are mediated by changes in blood pressure gradients and reflexive vasodilation. Such changes are noted to occur within several heart beats during carotid endarterectomy. An incompletely understood complex interplay of metabolic and hemodynamic factors are proposed to initiate and sustain recruitment of secondary structural and pathophysiological collateral vessels in states of chronic hypoperfusion. Additionally, various comorbidities such as hypertension may variably decelerate or endanger incipient collateral network growth and the overall tissue perfusion capacity [2]. The dynamic nature of these cerebrovascular reserve components emphasizes the need for a multifaceted interrogation of collateral vessel status. CTA, MRA, and direct catheter angiography are all well recognized and accurate modalities for the evaluation of structural primary and secondary collateral vessels [5]. However, the parenchymal perfusion capacity of the pathophysiologic collateral pathways may not necessarily be inferred from the status of primary and secondary structural collateral vessels [6]. As such, during periods of ischemia, tissue fate may depend on the status of these hemodynamically significant pathophysiologic collateral pathways that are occult to conventional vascular imaging modalities [7]. With this in mind, the following discussion is intended to explore the role of perfusion-CT in the assessment of cerebral collateral circulation with regards to two specific applications: acute ischemic stroke and chronic cerebrovascular disease.
Assessment of cerebral collateral circulation in acute ischemic stroke The adequacy of collateral circulation in acute ischemic stroke is an important determinant of the fate of the ischemic tissue and the prognosis for vascular recanalization [8,9]. The blood flow supplied by the collateral arterial network helps maintain misery perfusion in areas of ischemic penumbra. Robust collaterals also likely facilitate the ‘‘wash-out’’ of thrombi fragments and mitigates against infarct progression. Similarly, collateral arterioles may also help deliver thrombolytic agent to more distal thrombotic regions, thereby assisting in more rapid recanalization.
Functional imaging of cerebral collateral circulation Perfusion-CT In states of hemodynamic perturbation, blood may arrive to parenchyma via collateral pathways in antegrade or retrograde direction; however, this is difficult to distinguish in practice. Perfusion-CT complements angiographic techniques by offering a dynamic assessment of parenchymal
Figure 1 Interpretation of perfusion-CT maps. The green and red boxes highlight key differences in perfusion characteristics of endangered yet salvageable ischemic penumbra and irreversible ischemic core.
perfusion, regardless of occlusion site or directionality of blood flow [10]. This technique allows for the assessment of the pathophysiologic collateral vessels as inferred from the ischemic penumbra estimate [6,10]. Perfusion-CT monitors the parenchymal wash-in and wash-out of an iodinated contrast material bolus with serial acquisitions, usually at several supratentorial levels. In the post-processing stage, the investigator or the processing software select a reference artery input and vein output functions. A series of time-concentration curves are then generated for each voxel, and perfusion-CT parameters are generated that describe these curves in each voxel: mean transit time (MTT), regional cerebral blood volume (rCBV), and regional cerebral blood flow (rCBF). MTT designates the average time required for blood cell passage through the parenchymal capillary network. rCBV reflects the volume of blood per unit of parenchyma, mL/100 g. rCBF is a rate unit derived from rCBV/MTT. MTT is the most sensitive perfusion indicator of ischemia, with changes in cerebral blood flow and cerebral blood volume are more specific for distinguishing penumbra from ischemic core [9]. As detailed in Fig. 1, MTT is elevated in all ischemic cerebrovascular events. Parenchymal autoregulatory mechanisms initially compensate with reactive vasodilation and collateral flow recruitment, thus increasing rCBV. Again, as rCBF is derived from rCBV/MTT, a proportional increase in both the numerator and denominator will result in an unchanged, normal rCBF value. This is a pattern typically observed during a transient ischemic attack. With worsening and more prolonged ischemia, MTT becomes more prolonged and rCBF eventually decreases. Where collateral circulation is robust enough to maintain rCBV, ischemic penumbra exists; that is, tissue at risk for infarction but that is still currently salvageable. When the autoregulatory mechanisms are overwhelmed such that the collateral circulation is inadequate to maintain threshold tissue blood flow, the rCBV falls and an irreversible infarct develops [10—12]. These principles are illustrated in Fig. 2 with the perfusion-CT map of an acute right MCA territory infarct. In the subcortical areas of ischemic penumbra, the leptomeningeal collateral circulation is adequate to maintain rCBV. However, the caudate head and putamen, perfused by an inadequately collateralized lenticulostriate system, demonstrate decreased rCBV and thus constitute the ischemic core.
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders
3
Figure 2 Perfusion-CT map of an acute left MCA territory infarct. Corpus striatum infarct core (bracketed on rCBV panel) corresponding to areas of inadequate lenticulostriate collateral capacity. The majority of the penumbra (bracketed on MTT and rCBF) is within the perfusion territory of robust leptomeningeal collaterals.
Arterial spin labeling MRI Arterial spin labeling is an advanced noncontrast MRI technique in which selective radiofrequency labeling of arterial serum protons produce endogenous vascular contrast signal that is subtracted from a control image. Generally, proton labeling is performed in a labeling plane upstream from the region of interest (i.e. extracranial vasculature labeled in a plane parallel to the skull base), and a post labeling delay is inserted prior to read out, allowing labeled blood to reach the parenchyma. Multiple acquisitions may be performed utilizing different post labeling delays to characterize arterial transit time and more closely interrogate cerebral blood flow parameters [13,14]. As the overall signal differences between control and labeled series are typically low (0.5—2%), multiple subtraction data sets are obtained and averaged to generate perfusion maps [15]. There are several generations of MR spin labeling techniques: continuous (CASL), pulsed (PASL), and pseudocontinuous (pCASL). CASL is a first generation technique that utilizes a continuous (1—2 seconds) RF pulse applied to the extracranial vasculature utilizing surface coils and dedicated scanner hardware [16]. This method has lost favor in clinical utility due to relatively low labeling efficiency, in part relating to often irregular vessel geometry and imperfect labeling plane perpendicularity [17]. PASL is a labeling-efficient and reproducible second generation technique in which a large volume of blood is prepared with a short (∼10 ms) radiofrequency pulse within a thick labeling slab that is selected for each major vascular territory [17]. However, labeling slab planning requires advanced knowledge of arterial anatomy, and often necessitates a large field of view time of flight (TOF) planning angiogram to minimize contamination from adjacent non-target arterial structures. Misestimates of perfusion parameters may be encountered with PASL slab labeling contamination in which extracranial vessel tortuosity precludes a pure vascular territory sample. This is potentially confounding in states of chronic cerebrovascular disease in which extracranial collateralization is often present. pCASL is a third generation hybrid of earlier techniques employing a train of short RF labeling pulses that achieve target artery specificity with gradient induced variations in vessel labeling efficiency. Superselective ASL is a pCASL derived technique that utilizes rotational modulation of
gradients to produce a focal labeling point with a high spatial discriminatory capacity sufficient to interrogate individual vessels above the Circle of Willis [16]. A limitation inherent to ASL has particular importance to parenchymal perfusion mapping. All ASL techniques utilize selective arterial proton spin inversion as the basis of endogenous vascular contrast resolution (i.e. no intravenous gadolinium based contrast agent is used). This however necessarily means that contrast signal decays with the T1 relaxation time of blood, approximately 1.6 seconds at 3 Tesla. As such, perfusion of parenchymal regions beyond this arterial transit time may be underestimated [15]. This limitation is especially pertinent to the evaluation of arterial collateralization in which upstream steno-occlusive conditions may artifactually suggest decreased CBF and collateral reserve insufficiency. Of the strategies currently available to address this limitation, the utilization of higher field strength systems is most intuitive, as the contrast decay is somewhat mitigated by prolonged T1 relaxivity. Additionally, Velocity Selective (VS-ASL) is a technique without post labeling delay, in which a velocity sensitive pulses invert blood proton spins traveling in excess of a designated cutoff value, and read out is performed using those spins that decelerate to below the cutoff velocity value; in this fashion the issue of signal decay due to post labeling delay is theoretically minimized. Nonetheless, ASL is a non-invasive non-contrast technique capable of obtaining quantitative perfusion data with specificity for individual vascular territories. ASL CBF measurements generally correlate most closely with dynamic susceptibility contrast MRI mean transit time and time to peak parameters [18]. However, in part due to the aforementioned limitations, ASL tends to overestimate the volume of perfusion abnormalities and has lower specificity for mismatch status than dynamic contrast susceptibility perfusion weighted imaging (DCS-PWI) [19,20]. Transcranial Doppler Initially introduced as a non-invasive bedside technique to detect vasospasm following subarachnoid hemorrhage, transcranial Doppler demonstrates utility in describing intracranial arterial patency and collateralization status. Although generally not as sensitive to intracranial stenoses as CTA, transcranial Doppler offers the advantage or real
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
4
J. Donahue et al.
Figure 3 Operational penumbra is an imaging biomarker of tissue that is at risk for infarction (green text), but not yet irreversibly damaged (i.e. tissue that will recover with and infarct without reperfusion).
time evaluation and specificity of flow directionality. Identification of flow diversion, a low-resistance high velocity pattern, in the presence of known steno-occlusive disease correlates well with the presence of leptomeningeal collaterals. Limitations of the technique, as with all ultrasound, include operator variability and acoustic window availability. There are four commonly utilized acoustic windows in adults, the temporal, suboccipital, orbital, and submandibular approaches. In some inpatient settings, these acoustic windows may be obscured by bandaging, or otherwise inaccessible due to anatomic variability [21—24].
The ischemic penumbra and collateral circulation The utility of perfusion imaging in regards to the assessment of cerebral collateral circulation stems from the many similarities that collateralization shares with the ischemic penumbra. Ischemic penumbra is a multifactorial continuum of various states of parenchymal viability. The severity and reversibility of neuronal metabolic and electrical derangements, the integrity of parenchymal capillary beds, and progression of apoptotic pathways are all determinants of tissue fate. These complexities are of course not individually detailed; however, perfusion imaging provides a practical, operational depiction of salvageable tissue. Operational penumbra is a trichotomous imaging biomarker used to distinguish endangered tissue that will recover with, and infarct without reperfusion from the infarct core and surrounding normal parenchyma (Fig. 3). The operational concept of ischemic penumbra is informed by the perfusion threshold values for tissue at risk, which were derived through outcomes observation of patients presenting with acute ischemic stroke that subsequently did or did not attain reperfusion [25]. Thus, operational penumbra is influenced by the method and efficacy of revascularization and may vary depending of locoregional tissue conditions such as temperature, metabolic susceptibility, genetics, etc. Although not 100% accurate in predicting tissue fate given these variables, the operational penumbra concept is effective for guiding treatment decisions. As less than 40% of patients suffering from acute ischemic stroke present within the established time windows for systemic thrombolysis, an imaging determinant of ischemic penumbra is useful in instances of unclear time of symptom onset, such as ‘‘wake up stroke.’’ Establishing the functionality of arterial collaterals and operation penumbra with perfusion imaging is facilitating a promising paradigm shift from a ‘‘time is brain’’ to a ‘‘penumbra or imaging is brain’’ approach to acute ischemic stroke triage and management [11,26].
Figure 4 Examples of two patients, both with right M1 occlusion. Ischemic core is designated in red; ischemic penumbra shown as green. Despite the same level of large vessel occlusion, the absolute size of the ischemic area and the relative proportions of ischemic core to ischemic penumbra vary considerably due to the differing leptomeningeal collateral arterial adequacy. Patient B, who did not achieve recanalization, suffered penumbra evolution towards infarct [7].
Collateral blood flow shares similarities with, but is distinct from, the ischemic penumbra. Penumbra measurements provide only an indirect quantitation of the effectiveness of collateral circulation. In Fig. 4, CTA maximum intensity projection images and perfusion-CT data overlay of two patients with right M1 MCA occlusions are shown. In patient A, with good leptomeningeal collaterals, the MCA occlusion translates to a smaller total ischemic area and smaller ischemic penumbra. In patient B, with poor leptomeningeal collaterals, the occlusion translates to a large ischemic area and large ischemic penumbra, which without recanalization progressed to a large territory infarct. These two cases illustrate the fact that the site of occlusion, by itself, does not allow prediction of the infarct volume or proportion of penumbra [7]. This variability is due to differences in the adequacy of collateralization. Fig. 5 furthers this important distinction between penumbra and structural arterial collateralization. Perfusion-CT derived ischemic penumbra and infarct core volumes from
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders
5
Figure 5 Infarct core (red) and ischemic penumbra (green) volume plots in 165 patients with acute M1 or ICA occlusions illustrating the highly variable penumbra volume for any given ischemic core volume [6].
165 patients who sustained acute proximal large vessel occlusion, either M1 or ICA terminus, are plotted. This graph illustrates the extreme variability of observed penumbra volume for any given infarct core volume. Similar variability is also observed when penumbra volume is plotted against volume of total ischemic area [6]. This illustrates two important concepts. First, good collateral circulation does not necessarily translate to a large penumbra; low volume of penumbra may exist precisely because the collateral circulation is robust enough to maintain relatively normal rates of perfusion within the tissue adjacent to an infarct core. Secondly, there is no fixed relationship between site of occlusion, ischemic core volume, and penumbra volume. Even given similar sites of large vessel occlusion, the volume and relative proportion of penumbra to ischemic core is highly variable. Again, this variability is due to the fact that collateral circulation differs between patients. However, even with the combination of clinical data, noncontrast head CT, and CTA based structural collateral vessel data, penumbra is still not predicted accurately [4,6]. This suggests that interrogation of structural arterial collateralization solely with CTA or other anatomic modalities is insufficient to accurately infer the status of pathophysiologic collaterals and the overall parenchymal perfusion capacity of the collateral network. In other words, critical and predictive information regarding the functional capacity of the collateral vessels that sustain ischemic penumbra is lost when evaluation of the collateral network is limited to structural scoring [6,25,27]. To this end, arterial spin labeling offers potentially complementary information in regards to parenchymal collateralization status during acute ischemic events. The value that ASL may add in characterizing collateral perfusion networks is not solely in redemonstrating areas of oligemia and infarct. Rather, territorial ASL is uniquely capable of noninvasively mapping the individual perfusion regions subtended by the major vasculature, even in states of extensive collateralization [28,29]. In distinction to conventional perfusion CT and MR, which cannot precisely determine the origin or direction (antegrade or retrograde) of collateral flow, ASL may describe the presence and degree of
collateralization contributed to the ischemic penumbra by each individual vessel [30,31]. The precision of ASL for delineating distal collateral perfusion territory is comparable to digital subtraction angiography when combined with time of flight MRA, even in the presence of advanced steno-occlusive disease and Circle of Willis deficiencies [29,32]. When given the knowledge of actual vascular territory boundaries, the importance of watershed or borderzone physiology between differentially collateralized regions becomes evident. Hendrikse et al. found that diffusion abnormalities localize to perfusion territory interfaces significantly more frequently than the currently estimated 2—5% that is predicted by structural anatomic imaging [28,33]. Thus with ASL, classification of infarcts may potentially be more precisely subdivided into thromboembolic or hypoperfusional etiologies [30,31]. Beyond describing the anatomic boundaries of collateralized perfusion territories, ASL can offer insight into pathophysiologic collateral function. In general, the fast decay of ASL contrast signal proportional to T1 relaxation, among various other technical parameters, results in the oversensitivity of the technique to even benign oligemic perfusion delays, and typically results in the over estimation of perfusion abnormality volume in the acute stroke setting [18,20]. This however, is the basis of the high negative predictive value of ASL for the presence of perfusional abnormalities [20]. Not surprisingly, ASL demonstrates high sensitivity for perfusion anomalies corresponding to transient ischemic attack symptoms, likely exceeding the sensitivity of combined contrast bolus perfusion MR and MR angiographic evaluations [34—36]. Thus ASL can help localize incipient deficiencies in parenchymal collateralization that may herald future infarct.
Perfusion-CT based assessment of cerebral collateral circulation in chronic cerebrovascular disease A major aim in the evaluation of patients with chronic cerebrovascular conditions is to accurately predict a patient’s individual risk of future stroke. The relevance of
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
6
J. Donahue et al.
perfusion-CT in this regard is related to the importance of a refined assessment of collateral circulation capacity. The recruitment of collaterals to serve areas of chronic ischemia is protective not only due to stabilized parenchymal perfusion capacity, but also for the ability to help ‘‘wash-out’’ thrombi from the small vessel circulation. As discussed below, the dynamic element that perfusion-CT adds to the imaging work up of such patients enhances future stroke risk stratification.
Concept of parenchymal perfusion delay During perfusion-CT post-processing, an arterial and venous reference vessel is selected prior to the generation of time concentration curves. The arterial reference vessel serves as the primary input function for the generation of the MTT, while the selected reference vein is the principle input for the rCBV calculation. The vessel designated as the arterial input function is assumed to represent the only vascular supply to a tissue voxel. It is assumed that the contrast bolus that is delivered by the reference artery arrives to the tissue without delay. Selection of a reference artery with a hemodynamically significant upstream stenosis may produce appreciable regional perfusion abnormalities characterized by increased MTT and decreased rCBF. In such cases, this could potentially be confused with acute ischemia [37,38]. Additionally, the artifact produced by the delay of the contrast bolus arrival may be compounded by the dispersion of the contrast material throughout multiple small collateral vessels proximal to the site of stenosis. Multiple delay correction strategies have been devised to normalize the parenchymal time concentration curves to account for bolus delay and dispersion. Some methods utilize various deconvolution algorithms that relax the assumptions regarding the exclusivity of reference vessel contrast material delivery to a tissue voxel, while other utilize more distal arterial input functions that are closer to each displayed parenchymal territory [38]. Fig. 6 presents a perfusion-CT data set of a patient with a chronic high-grade stenosis of the right middle cerebral artery, with and without application of the delay correction program. As seen in the setting of acute ischemia, the parenchyma downstream to the arterial stenosis demonstrates decreased rCBF and increased MTT on uncorrected data sets. The rCBV, which is a function of venous outflow, remains normal. With the application of delay correction, the perfusion parameters of the affected territory are normalized and displayed as equivalent to neighboring territories with normal arterial vascular supplies. The potential for introducing artifact during postprocessing has traditionally been considered a limitation of perfusion-CT. In fact, much debate exists in the literature regarding the standardization of reference vessel selection precisely due to the potential variability in the perfusion maps as a function of reference vessel laterality and patency. However, the mismatch between the corrected and uncorrected perfusion maps contains important data, and ought not to be automatically discarded. The delay in tissue perfusion depicted in the corresponding time-concentration curve plots (Fig. 7) demonstrates the basis of quantitative collateral assessment.
Figure 6 Perfusion-CT data in patient with a high-grade chronic right MCA stenosis. In the first column, the ACA has been selected as the arterial input function; the right MCA territory demonstrates decreased MTT and rCBF secondary to the right MCA stenosis. In the second column, each vascular territory has an individually selected arterial input function and has been processed with a delay correction algorithm. The perfusion values of the right MCA territory have been normalized to account for contrast bolus delay and dispersion related to the right MCA stenosis. Note that rCBV is a function of venous output, and is unaffected by the arterial stenosis.
Figure 7 Time-concentration curve plot of the same patient with a high-grade chronic right MCA stenosis. The ACA and left MCA demonstrate early contrast material arrival and prompt time to peak, consistent with early wash-in and wash-out. The high-grade right MCA stenosis causes delayed parenchymal contrast density. The delay is a quantitative reflection of the net collateral arterial capacity available to the right MCA territory [37].
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders
7
Figure 8 Correlation of leptomeningeal collateral score and Circle of Willis collateral scores with MTT delay based physiologic collateral assessment. Bars represent medians, boxes represent interquartile ranges, and error bars represent 95% CIs. MTT indicates mean transit time [37].
The delay in perfusion between the territory supplied by the stenotic artery and the remainder of the unaffected vascular territories reflects a sum of the collateral pathways available to the right MCA territory. Although the directionality of the collateral arterial flow remains indeterminate (i.e. may be antegrade through the stenotic artery or retrograde via leptomeningeal collaterals), the perfusion delay is an independent quantitation of net arterial collateral capacity [38]. In other words, the delay in MTT, which is related to contrast bolus delay and dispersion, provides a dynamic assessment of total collateral circulation and cerebrovascular reserve [37,38].
Clinical significance of parenchymal perfusion delay This parenchymal perfusion delay is an independent variable with prognostic clinical utility that may not be captured by traditional structural evaluation methods. Even in patients without a history of prior stroke, MTT delay of greater than 0.5 seconds between MCA territories was highly predictive of new, incident infarcts [37]. The prognostic value of MTT delay for occurrence of subsequent infarct is not necessarily inferable from CTA leptomeningeal collateral score. Although certain malignant CTA profiles have demonstrated prognostic value for later infarct, as shown in Fig. 8, the correlation of MTT delay with leptomeningeal collateral scores is marginal [7,37]. Similarly, the patency and completeness of the Circle of Willis shows weak correlation with MTT delay. As expected, the less complete the Circle of Willis or the lower the leptomeningeal collateral score of the affected hemisphere, the greater the MTT delay [37]. When taken together, this data suggest that pathophysiologic collaterals indeed play a pivotal role in the disease process of chronic cerebrovascular disease. However, the relatively coarse and ordinal data provided by CTA based structural assessments generally lack the fine discriminatory capabilities to create refined patient risk stratification models. Perfusion-CT has a unique ability among CT applications to characterize the functional status
of the pathophysiologic collateral pathways with a continuous and quantitative data [7,10,37]. As with the interrogation of collateralization status in the acute ischemic setting, ASL provides complementary information regarding pathophysiologic collateral function in the setting of chronic cerebrovascular disease. Again, in cases of significant steno-occlusive disease, collateralization may result in deviation of actual perfusion territories from expected anatomic vascular regions. Territorial ASL technique has demonstrated that approximately 10% of cortical infarcts may be misclassified by conventional anatomic designations due to collateralization, meaning that when a perfusion delay is identified, there is a substantial possibility of misidentifying the responsible vessel [28]. ASL can accurately discriminate between not only vertebrobasilar and internal carotid artery perfusion territories, but also delineate external carotid artery collateral contributions and quantitate the function of extra-anatomic bypass grafts with superselective vessel mapping techniques [39—41]. Moreover, when a perfusion delay is recognized, ASL may help characterize the cerebrovascular reserve capacity, further stratifying the risk of subsequent infarct or likely response to intervention [42,43]. As ASL is a noncontrast technique, same session pre- and post-acetazolamide challenge assessment may be performed to determine the vasodilatory reserve of the pathophysiologic collateral pathways [15,44]. In this manner, the anatomic specificity of MRI may be combined with a functional assessment of cerebrovascular reserve capacity, a variable with significant prognostic value that has been validated in studies utilizing transcranial Doppler, perfusion CT, and other modalities [44,45].
Summary Cerebral collateral circulation encompasses the entirety of the vascular networks that supply parenchymal blood flow, from the primary structural large vessels to the inconstant and variably recruited pathophysiological collaterals. Although no single imaging modality fully characterizes the
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
8
J. Donahue et al.
complexity and pathophysiological significance of cerebral collateral circulation, the capacity of perfusion-CT and arterial spin labeling MRI to depict the functional status of the collateral network has relevant applications in both acute and chronic ischemic states. More specifically, the unique ability of these modalities to resolve pathophysiologic collateral circulation in a functional and dynamic manner informs evaluation of both acute ischemic penumbra and chronic cerebral vascular reserve. Though not fully reflecting the intricacies of the complex metabolic determinants of ischemic tissue fate, perfusion-CT provides a practical context of operational penumbra in the evaluation of acute stroke. This imaging biomarker is a quantitative proxy for the effectiveness of net collateral arterial function. Similarly, the characterization of parenchymal perfusion delay that perfusion-CT offers yields a continuous and quantitative metric basis for future ischemic event risk stratification models. Arterial spin labeling MRI provides a quantitative parenchymal perfusion assessment with the capacity to discriminate between vascular territories, thus identifying the anatomic extent of collateralization and the respective watershed territories. Additionally, arterial spin labeling MRI furthers the assessment of pathophysiologic collateral function as determined by the cerebrovascular reserve capacity.
Disclosure of interest The authors declare that they have no conflicts of interest concerning this article.
Acknowledgements Many thanks to Drs. Tan, Keedy, Wintermark, and Srinivasan for the gracious use of figures and tables from the cited references.
References [1] Liebeskind DS. Collateral circulation. Stroke 2003;34(9): 2279—84. [2] Romero JR, et al. Cerebral collateral circulation in carotid artery disease. Curr Cardiol Rev 2009;5(4):279—88. [3] Souza LC, et al. Malignant CTA collateral profile is highly specific for large admission DWI infarct core and poor outcome in acute stroke. AJNR Am J Neuroradiol 2012;33(7):1331—6. [4] Tan IY, et al. CT angiography clot burden score and collateral score: correlation with clinical and radiologic outcomes in acute middle cerebral artery infarct. AJNR Am J Neuroradiol 2009;30(3):525—31. [5] Smith WS, et al. Significance of large vessel intracranial occlusion causing acute ischemic stroke and TIA. Stroke 2009;40(12):3834—40. [6] Zhu G, et al. Computed tomography workup of patients suspected of acute ischemic stroke: perfusion computed tomography adds value compared with clinical evaluation, non-contrast computed tomography, and computed tomography angiogram in terms of predicting outcome. Stroke 2013;44(4):1049—55. [7] Tan JC, et al. Systematic comparison of perfusion-CT and CT-angiography in acute stroke patients. Ann Neurol 2007;61(6):533—43.
[8] Liebeskind DS, et al. Collaterals dramatically alter stroke risk in intracranial atherosclerosis. Ann Neurol 2011;69(6): 963—74. [9] Gonzalez RG, et al. The Massachusetts General Hospital acute stroke imaging algorithm: an experience and evidence based approach. J Neurointerv Surg 2013;5(Suppl. 1):pi7—12. [10] Wintermark M, et al. Prognostic accuracy of cerebral blood flow measurement by perfusion computed tomography, at the time of emergency room admission, in acute stroke patients. Ann Neurol 2002;51(4):417—32. [11] Leiva-Salinas C, et al. Tissue at risk in acute stroke patients treated beyond 8 h after symptom onset. Neuroradiology 2013;55(7):807—12. [12] Dankbaar JW, et al. Dynamic perfusion-CT assessment of early changes in blood brain barrier permeability of acute ischaemic stroke patients. J Neuroradiol 2011;38(3):161—6. [13] MacIntosh BJ, et al. Assessment of arterial arrival times derived from multiple inversion time pulsed arterial spin labeling MRI. Magn Reson Med 2010;63(3):641—7. [14] Bokkers RP, et al. Arterial spin-labeling MR imaging measurements of timing parameters in patients with a carotid artery occlusion. AJNR Am J Neuroradiol 2008;29(9):1698—703. [15] Hendrikse J, Petersen ET, Golay X. Vascular disorders: insights from arterial spin labeling. Neuroimaging Clin N Am 2012;22(2):259—69, x-xi. [16] Hartkamp NS, et al. Mapping of cerebral perfusion territories using territorial arterial spin labeling: techniques and clinical application. NMR Biomed 2013;26(8):901—12. [17] Chen Y, Wang DJ, Detre JA. Test-retest reliability of arterial spin labeling with common labeling strategies. J Magn Reson Imaging 2011;33(4):940—9. [18] Niibo T, et al. Arterial spin-labeled perfusion imaging to predict mismatch in acute ischemic stroke. Stroke 2013;44(9): 2601—3. [19] Nael K, et al. Quantitative analysis of hypoperfusion in acute stroke: arterial spin labeling versus dynamic susceptibility contrast. Stroke 2013. [20] Zaharchuk G, et al. Comparison of arterial spin labeling and bolus perfusion-weighted imaging for detecting mismatch in acute stroke. Stroke 2012;43(7):1843—8. [21] Bathala L, Mehndiratta MM, Sharma VK. Cerebrovascular ultrasonography: technique and common pitfalls. Ann Indian Acad Neurol 2013;16(1):121—7. [22] Bathala L, Mehndiratta MM, Sharma VK. Transcranial doppler: technique and common findings (Part 1). Ann Indian Acad Neurol 2013;16(2):174—9. [23] Kim Y, et al. Relationship between flow diversion on transcranial Doppler sonography and leptomeningeal collateral circulation in patients with middle cerebral artery occlusive disorder. J Neuroimaging 2009;19(1):23—6. [24] Kirsch JD, et al. Advances in transcranial Doppler US: imaging ahead. Radiographics 2013;33(1):E1—14. [25] Zhu G, et al. Does perfusion imaging add value compared with plain parenchymal and vascular imaging? J Neurointerv Surg 2012;4(4):246—50. [26] Michel P, et al. Perfusion-CT guided intravenous thrombolysis in patients with unknown-onset stroke: a randomized, double-blind, placebo-controlled, pilot feasibility trial. Neuroradiology 2012;54(6):579—88. [27] McVerry F, Liebeskind DS, Muir KW. Systematic review of methods for assessing leptomeningeal collateral flow. AJNR Am J Neuroradiol 2012;33(3):576—82. [28] Hendrikse J, et al. Relation between cerebral perfusion territories and location of cerebral infarcts. Stroke 2009;40(5):1617—22. [29] Wu B, et al. Collateral circulation imaging: MR perfusion territory arterial spin-labeling at 3T. AJNR Am J Neuroradiol 2008;29(10):1855—60.
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002
+Model NEURAD-437; No. of Pages 9
ARTICLE IN PRESS
Assessment of collateral flow in patients with cerebrovascular disorders [30] Petcharunpaisan S, Ramalho J, Castillo M. Arterial spin labeling in neuroimaging. World J Radiol 2010;2(10):384—98. [31] van Laar PJ, van der Grond J, Hendrikse J. Brain perfusion territory imaging: methods and clinical applications of selective arterial spin-labeling MR imaging. Radiology 2008;246(2):354—64. [32] Chng SM, et al. Territorial arterial spin labeling in the assessment of collateral circulation: comparison with digital subtraction angiography. Stroke 2008;39(12):3248—54. [33] Momjian-Mayor I, Baron JC. The pathophysiology of watershed infarction in internal carotid artery disease: review of cerebral perfusion studies. Stroke 2005;36(3):567—77. [34] MacIntosh BJ, et al. Multiple inflow pulsed arterial spinlabeling reveals delays in the arterial arrival time in minor stroke and transient ischemic attack. AJNR Am J Neuroradiol 2010;31(10):1892—4. [35] Zaharchuk G, et al. Arterial spin labeling imaging findings in transient ischemic attack patients: comparison with diffusion- and bolus perfusion-weighted imaging. Cerebrovasc Dis 2012;34(3):221—8. [36] Zaharchuk G. Arterial spin label imaging of acute ischemic stroke and transient ischemic attack. Neuroimaging Clin N Am 2011;21(2):285—301. [37] Keedy AW, et al. Contrast delay on perfusion CT as a predictor of new, incident infarct: a retrospective cohort study. Stroke 2012;43(5):1295—301.
9
[38] Konstas AA, et al. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: Theoretic basis. AJNR Am J Neuroradiol 2009;30(4):662—8. [39] Helle M, et al. Superselective arterial spin labeling applied for flow territory mapping in various cerebrovascular diseases. J Magn Reson Imaging 2013;38(2):496—503. [40] Dang Y, et al. Quantitative assessment of external carotid artery territory supply with modified vessel-encoded arterial spin-labeling. AJNR Am J Neuroradiol 2012;33(7): 1380—6. [41] Wintermark M, et al. Comparative overview of brain perfusion imaging techniques. J Neuroradiol 2005;32(5): 294—314. [42] Golay X, Hendrikse J, Van Der Grond J. Application of regional perfusion imaging to extra-intracranial bypass surgery and severe stenoses. J Neuroradiol 2005;32(5):321—4. [43] de Castro-Afonso LH, et al. Carotid artery stenting performed with a flow-reversal technique: improved technical performance. J Neuroradiol 2013;40(1):29—37. [44] Bokkers RP, et al. Cerebrovascular reactivity within perfusion territories in patients with an internal carotid artery occlusion. J Neurol Neurosurg Psychiatry 2011;82(9):1011—6. [45] Silvestrini M, et al. Impaired cerebral vasoreactivity and risk of stroke in patients with asymptomatic carotid artery stenosis. JAMA 2000;283(16):2122—7.
Please cite this article in press as: Donahue J, et al. Assessment of collateral flow in patients with cerebrovascular disorders. J Neuroradiol (2013), http://dx.doi.org/10.1016/j.neurad.2013.11.002