The information obtained from intracranial vascular imaging can influence clinical decision making in patients with carotid stenosis. In patients with carotid territory TIAs or stroke in association with tandem arterial stenosis affecting the extracranial and intracranial ICA, many clinicians would advise treatment of the extracranial stenosis first with either carotid endarterectomy or endovascular therapy. If TIAs or stroke recur in the absence of extracranial ICA restenosis, especially if one can detect emboli in the ipsilateral MCA on transcranial Doppler imaging and no other cause for the symptoms is identified, then one may conclude that the intracranial stenosis is causing the recurrent symptoms. This could lead to further intensification of best medical therapy or perhaps endovascular intervention, although it must be stressed that no randomised studies comparing these interventions have been published to date.
In the North American Symptomatic Carotid Endarterectomy Trial (NASCET), the presence of intracranial atherosclerotic disease on catheter an-
giography significantly influenced the risk of subsequent stroke only in patients with 85-99% extracranial ICA stenosis who were treated with best medical therapy alone; intracranial stenosis did not significantly impact on stroke risk in patients with < 50%, 50 to 69%, or 70 to 84% stenosis . The three-year risk of ipsilateral stroke distal to an 8599% extracranial ICA stenosis was significantly higher in those with versus those without intracranial atherosclerosis (45.7% vs. 25.3%, relative risk 1.8 [95% confidence interval 1.1-3.2]) . In the same study, the three-year risk of ipsilateral stroke in surgically treated patients with 85-99% extracranial ICA stenosis was similar in those with and those without intracranial atherosclerosis (8.6% vs. 10%, relative risk 0.9, [95% confidence interval 0.2-3.0]). These data can be of use when counselling this subgroup of patients before carotid endarterectomy. Patients who have combined very severe extracranial and intracranial atherosclerosis have a very high risk of stroke if they are treated with best medical therapy alone, but the presence of intracranial stenosis does not appear to increase the hazard associated with surgery if they proceed to carotid endarterec-tomy.
A further study revealed that the presence of in-tracranial collateral supply to the symptomatic hemisphere in medically treated patients with symptomatic 70-99% extracranial ICA stenosis reduced the two-year absolute risk of ipsilateral hemispheric stroke by 16.5% (11.3 vs. 27.8%, p = 0.005) and the risk of ipsilateral hemispheric TIA by 17% (19.1 vs. 36.1%, p = 0.008) compared with patients who did not have collaterals . Surgically treated patients with symptomatic 70-99% extracranial ICA stenosis who had collateral supply to the affected hemisphere had a lower peri-operative stroke risk in the 30 days after surgery compared with those patients without collaterals (1.1 vs. 4.9%) . However, the presence of collaterals did not provide significant added protection against TIA or stroke in these surgically treated patients during prolonged follow-up at 2 years . The beneficial effect of collateral supply was not evident in either the medically or surgically treated patient with near occlusion of the ICA. These data may also be useful in risk stratifying pa-
Fig. 7. a Selective intra-arterial carotid DSA (lateral projection) showing luminal irregularity with segmental narrowing and widening of branches of both the middle and anterior cerebral arteries. The patient presented with a haemorrhagic stroke after taking an unknown mixture of illicit substances, thought to include 'ecstasy'and cocaine. These peripheral arterial branches are rarely clearly visualised with current intracranial CTA or MRA imaging techniques; b High power view of the same patient showing stenotic and dilated segments of the angular branch of the middle cerebral artery (arrows); c I-ADSA following a left common carotid injection shows occlusion of the cervical internal carotid artery with refilling of the cavernous and supraclinoid internal carotid artery by reversed flow in the ophthalmic artery; the ophthalmic artery is supplied by branches of the external carotid artery. Large arrow, point of reconstitution of the intracranial cavernous ICA; Small arrow, reversed flow in the ophthalmic artery.
One of the limitations of I-ADSA is it does not directly image the vessel wall itself or allow one to identify the exact site of disease within the affected vessel wall. Furthermore, if peri-procedural complication rates are not kept to a minimum, the potential benefit from intervention in a patient with extracranial ICA stenosis, especially in those with asymptomatic severe carotid stenosis, may be outweighed by the combined short-term risks of angiog-raphy and revascularisation.
'Non-invasive' intracranial angiography
Intracranial CTA and MRA currently offer lower spatial resolution and less clear visualisation of the intracranial vessels than I-ADSA, but the peri-procedural morbidity associated with CTA or MRA is substantially lower. However, the adverse impact of the lower accuracy of these relatively 'non-invasive' imaging studies of the intracranial circulation in patients with extracranial ICA stenosis is uncertain.
CTA involves the intravenous administration of a non-ionic iodinated contrast agent, that is typically injected at a rate of 3ml per second . The patient is transported through the gantry of a spiral multi-detector CT scanner by a motorised table, and the rotating detector ring of the scanner acquires 3D data at a time that coincides with the phase of high intra-arterial contrast enhancement (first pass imaging). Post-processing of the acquired data sets allows one to construct 3D CT angiographic images, on which one can visualise the intracranial vessels and quantify the degree of stenosis of an intracranial artery. More rapid table movement obviously allows a larger field of imaging to be covered, but leads to more movement blurring artefact in the acquired images (Z-axis blur). Industry developments are focusing on producing wider blocks of detectors that will sample a larger area than that covered by the currently available spiral CT scanners, thus minimising the need for table movement with the goal of improving imaging quality.
The advantages of CTA include the fact that it can be performed in conjunction with standard CT brain imaging, it is a relatively non-invasive test which can be performed on an out-patient basis, and current 'leading-edge' machines can cover a region of interest from the aortic arch to the intracranial middle cerebral arteries.
However, there are some limitations and disadvantages of CTA as an intracranial vascular imaging technique. Patients undergoing CTA probably have approximately a 0.04% risk of experiencing a severe adverse reaction, and a 0.004% risk of experiencing a very severe adverse reaction to the iodinated contrast agent . Patients who are treated with metformin should discontinue therapy for 48 hours peri-procedur-ally, and an iso-osmolar contrast agent needs to be used in elderly diabetic patients with elevated serum creatinine levels to minimise the risk of contrast-induced renal failure. Therefore, clinicians must follow the guidelines for patient' preparation laid down by the local Radiology department.
CTA does not allow one to visualise sequential vessel opacification over time, thus rendering investigation of collateral intracranial blood supply more dif ficult than with catheter angiographic studies. This is because filling of collateral vessels may be delayed, and although patent, these collateral channels may not be opacified with CTA. Therefore, one cannot exclude the possibility that collateral channels exist, even if they are not identified with intracranial CTA. Because CT is very sensitive at detecting calcium, heavy calcification of a stenosed intracranial artery may also impede visualisation of the residual lumen of the vessel and hinder grading of the severity of an intracranial stenosis . CTA may not always clearly visualise the lumen of the intra-cavernous portion of the ICA because the degree of enhancement of the cavernous sinus may be similar to the enhancement of the adjacent intra-cavernous carotid artery; this may obscure visualisation of the artery . It may also be difficult to isolate the petrosal intra-osseous segment of the intracranial ICA from the surrounding bone using automated reconstruction techniques, and time-consuming post-processing of the images from this region is required on an angiographic workstation .
One small prospective series reported that in-tracranial CTA was reliable at identifying intracra-nial ICA or MCA trunk occlusion that was subsequently confirmed on intra-arterial DSA . However, a subsequent small retrospective study that compared the sensitivity of intracranial CTA with I-ADSA suggested that CTA may be a sensitive test for identifying severe (70 to 99%) stenosis or occlusion of the intracranial MCA or ACA, but not for identifying severe stenosis of the intracranial ICA . Therefore, larger prospective studies are required to adequately assess the sensitivity and specificity of intracranial CTA in identifying intracra-nial anterior circulation stenosis or occlusion in patients with extracranial carotid artery stenosis.
Despite these technical limitations, CTA is likely to play an increasingly important role in imaging the intracranial circulation in patients with extracranial ICA stenosis in future, especially in view of the ready availability of this test.
Magnetic Resonance Angiography (MRA)
In carotid stenosis patients without a contraindication to MRI, the combination of MRI and
MRA allows the comprehensive evaluation of the brain structure, perfusion, and vasculature. There are three main MR techniques used to image the intracranial vessels, and all of these techniques can also visualise the extracranial vessels during the same imaging session:
- Contrast enhanced MRA (CE-MRA)
Each technique has its own limitations and produces artefacts which the interpreting Radiologist must be familiar with. Some are common to all techniques, such as movement artefact, whereas others are technique-specific. These are summarised below:
PC-MRA applies a magnetic field to create a phase shift in protons within a predetermined range of flow velocities . The magnitude of the phase shift is used to create an image, with high signal representing flowing protons within the predetermined velocity range. Background structures are then suppressed to allow construction of a PC-MR angio-gram . PC-MRA may allow depiction of slow flow within a large intracranial artery and flow within smaller intracranial vessels.
One of the limitations of PC-MRA is that the technique provides images with poor spatial resolution. Furthermore, PC-MRA only detects phase shift caused by flow within a predetermined velocity range. If an extracranial or intracranial ICA stenosis causes an alteration in flow velocity above or below the predetermined velocity range, the distal intracranial vessel segment will not be detected and visualised. Therefore, incorrect velocity encoding at the beginning of the imaging protocol may lead to the false impression that an intracranial artery is occluded, when in fact, flow may only be reduced within the patent vessel. The technique is also limited by being insensitive to the direction of flow, and although this may be useful when one is imaging flow in a tortuous intracranial artery, venous blood flow may potentially obscure images of the arterial anatomy. For these reasons, we do not routinely use PC-MRA in the investigation of in-tracranial arterial stenosis, and mainly reserve this technique for the investigation of cerebral venous sinus thrombosis.
Time of flight ("in-flow") MRA is a gradient-echo short T1 sequence in which protons in flowing blood that enter the magnetic imaging field generate a strong signal, whereas protons in stationary background tissue are suppressed, thus generating a high intensity signal within blood vessels . With 2D ToF MRA, the image is built up from thin slices which have been sequentially acquired, whereas 3D ToF MRA acquires the data for the whole imaging slab at the same time. In comparison with 2D ToF MRA, the 3D technique is less susceptible to degradation by movement artefact, there is an improved signal-to-noise ratio, and shorter radio-frequency pulses can be used leading to better suppression of signal from background structures, such as fat within the orbit and dorsum sella. For these reasons, the 3D ToF MRA technique is usually preferred to 2D ToF MRA for visualisation of the intracranial arterial circulation.
Current ToF-MRA techniques provide familiar angiogram type images that can easily be reconstructed and manipulated by the reporting Radiologist. One can visualise the intracranial ICA, the MCA, the ACA and their first branches only. This is useful in identifying stenosis or occlusion of the ICA, proximal MCA or ACA, but peripheral branch occlusion will not be reliably identified with this technique (Fig. 8a). Although 3D ToF-MRA still has relatively poor spatial resolution in comparison with I-ADSA, the new generation of 3 Tesla MRI scanners should significantly improve the spatial resolution of the acquired images compared with those obtained on a 1.5 Tesla MRI scanner. Current demonstrations suggest that at least one higher order of peripheral vessels will be visualised with 3 Tesla MR imaging.
ToF-MRA has a number of methodological limitations that one must be aware of:
Fig. 8. a Intracranial ToF-MRA showing a proximal left MCA stenosis (large white arrow) in a patient with a left hemispheric MCA infarct of undetermined aetiology. The minor irregularity of the left ophthalmic artery (arrowhead) suggests that an ICA embolus fragmented and partially occluded both vessels; b Intracranial ToF MRA in a patient with a recent left MCA territory intracranial haemorrhage. The haema-toma appears hyperintense on this T1-weighted imaging sequence (T1 shortening), thus obscuring adequate visualisation of this distal branches of the left MCA (long black arrow) compared with the normal distal branches of the right MCA. One cannot exclude either an aneurysm or a focal stenosis of the distal left MCA with this technique at this stage after symptom onset. This image also shows hyperintesnity from fat within the petrous apex (arrowhead) and the orbits (dashed arrow) which can easily be removed by postprocessing of the images.
- 'In plane flow saturation': ToF MRA is most sensitive at detecting flow perpendicular to the plane of data acquisition, and the signal obtained from blood flowing in vessels aligned at lesser angles to the acquisition plane is less intense and must not be erroneously interpreted as indicating stenosis of that vessel. In addition, vessels which run parallel to the plane of data acquisition may not have detectable flow within them at all, thus resulting in absent arterial segments which must not be interpreted as being occluded. This most commonly occurs in the MCA and its branches.
- 'In slab saturation': ToF-MRA data are acquired in 'slabs' or 'chunks'. When the flowing blood traverses one of these slabs, the signal intensity from the flowing blood at the periphery of the slab temporarily decreases, resulting in an abrupt change in signal intensity and an inhomogeneous signal return from the artery being studied. This may also give the false impression of a focal stenosis in the vessel, and hence the importance of studying the source images that are used to generate the ToF-MRA.
- 'Susceptibility artefact': ToF-MRA is intrinsically vulnerable to inhomogeneity in the local magnetic environment. The interfaces between air and soft tissue at the skull base are a frequent source of this artefact, and one should not over-interpret loss of signal in the ICA in this region as indicating stenosis or occlusion.
- 'T1 shortening': Tissue with intrinsically short T1 signal will be seen on the acquired ToF MRA images. Subcutaneous and orbital fat can easily be removed by post processing techniques, but blood products, e.g., within an intracerebral haematoma or a haemorrhagic infarct, will remain on the reconstructed image and may obscure precise visualisation of local vascular anatomy (Fig. 8b).
- Overestimation of stenosis severity: This has been reported to occur especially in the intra-ca-vernous portion of the intracranial ICA .
Gadolinium chelates produce signal change on T1-
weighted MRI sequences (T1 shortening) in pro portion to their concentration in blood. Unlike PC-MRA or ToF-MRA, contrast enhanced intracranial MRA simply depends on local intravascular T1 shortening caused by the presence of gadolinium within a vessel, and does not rely on velocity-dependent or inflow phase shifts. Therefore, movement and flow artefacts are minimised, and even vessels which run in the plane of imaging are well visualised. For these reasons, CE-MRA is usually the in-tracranial MR angiographic imaging modality of choice to visualise the intracranial vessels when in-tracranial arterial stenosis is suspected.
The timing from the injection of contrast to the onset of image acquisition is critical to the success of CE-MRA . During the study, an initial small 'timing bolus' of 3 ml of gadolinium may be administered manually or via an automated pump, and the time taken for the bolus to reach the intracranial circulation is calculated. Alternatively, one can use an automated 'bolus detection' and 'scan triggering' scheme to precisely begin image acquisition once the bolus arrives in the arteries being studied . Typically, a 40 ml bolus of gadolinium is injected into a proximal forearm vein, followed by 40 ml of saline at a rate of 3 ml/s. Image acquisition begins as the contrast arrives in the intracranial arterial circulation, so that visualisation of intracerebral veins is avoided. A 3D volume ultrashort T1-weighted acquisition sequence is used, and the imaging parameters are set to minimise the signal return from background tissues, and to focus on the profoundly short T1 signal change from the gadolinium bolus within the artery. If one is imaging the intracranial circulation in a patient with extracranial ICA stenosis, it is important to appreciate that CE-MRA has poorer spatial resolution than I-ADSA, although the image quality is often better that that seen with non-enhanced MRA. However, if the image acquisition is not precisely timed to coincide with the arrival of the injected gadolinium bolus, then the arteries may be under-filled or the images obscured by 'venous phase contamination'. Furthermore, CE-MRA is also susceptible to TI shortening effects, as outlined above.
Therefore, if the clinical history, or the results of extracranial imaging studies suggest that in-
tracranial vascular imaging could facilitate clinical decision making in a patient with asymptomatic or symptomatic extracranial ICA stenosis, the options are to screen the patient with I-ADSA, CTA or MRA. Although I-ADSA is the gold standard vascular imaging technique, MRA (especially CE-MRA or ToF-MRA) or CTA are safer and allow one to obtain information about brain structure and vascular anatomy in one sitting. If further information is required that is likely to influence clinical management, e.g. delineation of collateral arterial blood supply in a patient with critical ICA stenosis, or the exclusion of intracranial vasculopa-thy involving smaller arterioles that one cannot visualise with current non-invasive arterial imaging techniques, one should proceed to I-ADSA. The imaging technique of choice is clearly determined by local availability and local expertise, and it is important to realise that certain CT and MRI systems provide better quality vascular imaging than others. When the Neuroradiologist is informed about the precise information that the clinician requires from the test, he/she can decide which non-invasive imaging test should be performed or whether one should proceed directly to catheter angiography. Advances in intracranial CTA and MRA techniques in future are likely to reduce the need for I-ADSA in many carotid stenosis patients in whom intracranial vascular imaging is needed.
Was this article helpful?