Evaluation of carotid artery stenosis

INTRODUCTION — Computed tomographic (CT) scanning and magnetic resonance imaging (MRI)
are useful for evaluating the question of cerebral infarction which may result from carotid artery
stenosis. Infarctions related to internal carotid artery stenosis may be deep, subcortical, or cortical.
However, carotid stenosis may exist in the absence of infarction on MRI and CT.

The definition of asymptomatic or symptomatic carotid artery stenosis is based upon the history and
physical examination, depending upon whether or not there are symptoms or signs of carotid
territory ischemia. In the large clinical trials addressing the management of carotid artery stenosis,
the detection of "silent" infarcts on CT or MRI did not qualify the stenosis as symptomatic. In clinical
practice, however, radiographic evidence of ischemia in the territory of a stenotic internal carotid
artery may affect management.

Four diagnostic modalities are used to directly image the internal carotid artery:
• Cerebral angiography

• Carotid duplex ultrasound

• Magnetic resonance angiography

• Computed tomographic angiography

This topic will review the clinical use of these different techniques and their unique advantages and
disadvantages. In addition, we will review the different methods of measuring the degree of carotid
stenosis used with angiography.

STENOSIS MEASUREMENT — The methods of evaluating the degree of angiographic stenosis

vary in technique and accuracy. If the results of clinical trials are to be generalized, there is a need
for uniformity in measurement [1].
The two major randomized clinical trials evaluating the utility of endarterectomy in symptomatic
patients used different methods to measure carotid stenosis (show figure 1) [2]. (See "Carotid
endarterectomy in symptomatic patients").
Currently, three methods (NASCET, ECST and CC) predominate worldwide. Although all three were
originally devised for use with conventional contrast angiography, these methods can also be used
with magnetic resonance (MR) and computed tomography (CT) angiography.
NASCET method — The North American Symptomatic Carotid Endarterectomy Trial (NASCET)
method measures the residual lumen diameter at the most stenotic portion of the vessel and
compares this to the lumen diameter in the normal internal carotid artery distal to the stenosis [3].
ECST method — The European Carotid Surgery Trial (ECST) method measures the lumen
diameter at the most stenotic portion of the vessel and compares this to the estimated probable
original diameter at the site of maximum stenosis [4].
CC method — The common carotid (CC) method measures the residual lumen diameter at the most
stenotic portion of the vessel and compares this to the lumen diameter in the proximal common
carotid artery [2,5].
Comparison — The maximum stenosis is generally in the carotid bulb, a wider portion of the artery
than the distal segment. As a result, the same degree of stenosis is quantified as a higher
percentage stenosis when measured by the ECST or CC methods than when measured by the
NASCET method. The ECST methodology also requires an assumption of the true lumen, which
increases the risk of interobserver variability. (Show figure 1).
Despite these differences, the results of all three methods have a nearly linear relationship to each
other and provide data of similar prognostic value [2]. Equivalent measurements for the three
methods have been determined [2,6]:
 • A 50 percent stenosis with the NASCET method is equivalent to a 65 percent stenosis for
both the ECST and CC methods.
• A 70 percent stenosis with the NASCET method is equivalent to an 82 percent stenosis for
both the ECST and CC methods.
CONVENTIONAL CEREBRAL ANGIOGRAPHY — Cerebral angiography is the gold standard for
imaging the carotid arteries. The development of intraarterial digital subtraction angiography (DSA)
reduces the dose of contrast, uses smaller catheters, and shortens the length of the procedure.
Although there is lower spatial resolution, DSA has largely replaced conventional angiography [7].
The quality of the angiogram depends upon selective catheterization of the carotid artery with at
least two unimpeded views. Aortic arch injections alone are inadequate; suboptimal studies can lead
to misinterpretations as an irregular stenosis can be either underestimated or overestimated in a
single projection.
Advantages — Cerebral angiography permits an evaluation of the entire carotid artery system,
providing information about tandem atherosclerotic disease, plaque morphology, and collateral
circulation which may affect management [8]. In particular, the presence of associated intracranial
atherosclerotic disease may identify a group that is likely to benefit from carotid endarterectomy [9].
However, pathologic evaluation of the plaque specimen provides the most useful data on plaque
composition, which may have a bearing on prognosis [10,11]. (See "Carotid endarterectomy in
symptomatic patients").
Disadvantages — The disadvantages of angiography include its invasive nature, high cost, and risk
of morbidity and mortality. In a review of prospective studies using cerebral angiography, the risk of
all neurologic complications was approximately 4 percent and the risk of serious neurologic
complications or death was approximately 1 percent (range 0 to 6 percent) [7]. The risk of morbidity
is increased with cerebrovascular symptoms, advanced age, diabetes, hypertension, elevated serum
creatinine, and peripheral vascular disease. The size of the catheter, amount of contrast and
procedure duration also affect the likelihood of complications [10]. One study found that embolic
events following angiography occur more frequently than the apparent neurologic complication rate
[12]; the clinical significance of this finding is unclear.
Although often considered the "gold standard" of carotid neurovascular imaging methods,
conventional DSA has the disadvantage of a limited number of projections, typically two or three,
depicting the carotid artery and bifurcation. This limitation could lead to an underestimation of the
degree of carotid stenosis in arteries that have asymmetrical rather than concentric stenotic lumens
[13,14]. Rotational angiography provides 16 to 32 projections and is far less subject to this problem,
but it is seldom used in practice.

ULTRASOUND
Carotid duplex ultrasound — Carotid duplex ultrasound (CDUS) uses B-mode ultrasound imaging
and Doppler ultrasound to detect focal increases in blood flow velocity indicative of high grade
carotid stenosis [15-17]. The peak systolic velocity is the most frequently used measurement to
gauge the severity of the stenosis (show radiograph 1), but the end-diastolic velocity, spectral
configuration, and the carotid index or peak internal carotid artery velocity-to-common carotid artery
velocity ratio provide additional information [18,19]. Color Doppler flow imaging may improve the
efficiency of the test, but it has not been shown to improve accuracy [15,17,20,21].
We examined the correlation between Doppler velocities and the residual lumen diameters of
internal carotid arteries from surgical pathological specimens to establish Doppler criteria for residual
lumen diameter, independent of the percent stenosis [22]. Peak systolic velocity (PSV), end-diastolic
velocity (EDV), and carotid index (peak internal carotid artery [ICA] velocity ÷ common carotid artery
[CCA] velocity) correlated with the residual lumen diameter.
By adjustment of velocity criteria, we found that CDUS can be either highly specific or highly
sensitive for detecting a residual lumen diameter of <1.5 mm [22]:
• A specificity of 100 percent was found for PSV >440 cm/sec, EDV >155 cm/sec, or carotid
index >10. The sensitivity for these measures was 58 percent, 63 percent, and 30 percent,
respectively. By combining these criteria, the sensitivity increased to 72 percent.
• A sensitivity of 96 percent was found for PSV >200 cm/sec combined with either an EDV
>140 cm/sec or a carotid index >4.5. The specificity for these combined measures was 61
percent.
A meta-analysis published in 2006 concluded that CDUS compared with intraarterial cerebral
angiography for the diagnosis of 70 to 99 percent carotid stenosis had a sensitivity and specificity of
0.89 (95% CI 0.85-0.92) and 0.84 (95% CI 0.77-0.89), respectively [23]. An earlier systematic review
concluded that the sensitivity and specificity of CDUS compared with digital subtraction angiography
for diagnosing complete carotid occlusion were 96 and 100 percent [24].
Although limited, CDUS has utility in obtaining information about plaque composition and intraplaque
hemorrhage, which may increase the risk of embolism and impact on prognosis [11,15-17,25,26]. In
one study, patients with echolucent atherosclerotic plaques had a significantly increased risk of
ischemic cerebrovascular events compared with patients with echogenic plaques [27].
Advantages — CDUS is a noninvasive, safe, and relatively inexpensive technique for evaluation the
carotid arteries. It is 81 to 98 percent sensitive and 82 to 89 percent specific in detecting a significant
stenosis of the internal carotid artery [15-17,28]. Data from the NASCET trial showed that a carotid
index (peak ICA velocity ÷ CCA velocity) above 4.0 provided the highest accuracy (sensitivity 91
percent, specificity 87 percent, overall accuracy 88 percent) for predicting a high grade (70 to 99
percent) stenosis amenable to surgery [29].
Using published outcome data and receiver operator characteristic analysis, test criteria can be
developed that maximize patient outcome for a specific clinical scenario [30,31]. However, as
discussed below, the positive predictive value of CDUS for identifying appropriate asymptomatic
candidates for carotid intervention may be lower in community centers.
Disadvantages — The absence of flow in the internal carotid artery may be due to occlusion, but
hairline residual lumens can be missed on CDUS [32]. In addition, several studies have found that
CDUS tends to overestimate the degree of stenosis [28,33].
CDUS is less precise in determining stenoses of less than 50 percent compared with stenoses of
higher degrees [15,16], but this rarely impacts on its clinical utility. CDUS may also be less accurate
in determining stenoses in the range of 50 to 69 percent compared with 70 percent or greater [28].
However, this too rarely impacts on its clinical utility because most patients are not considered for
endarterectomy unless the degree of stenosis is 70 percent or higher. (See "Carotid endarterectomy
in symptomatic patients", section on Summary and Recommendations).
CDUS imaging may be limited by features such as calcific carotid lesions, tortuous or kinked carotid
arteries, and patient body habitus. Another limitation of CDUS is that only the cervical portion of the
internal carotid artery can be evaluated, although transcranial Doppler may provide some information
about downstream vessels. (See "Transcranial Doppler" below).
The limitations of CDUS are illustrated in studies performed in community-based centers which
showed that performing carotid endarterectomy on the basis of CDUS alone would result in a
significant number of unnecessary surgeries [34,35]. (See "Carotid endarterectomy: Preoperative
evaluation; surgical technique; and complications", section on Role of other imaging studies, for a
more complete discussion of this issue.)
Finally, the accuracy of CDUS relies heavily upon the experience and expertise of the
ultrasonographer [17,36]. Measurement threshold properties may vary widely between laboratories,
and the magnitude of the variation is clinically important [37]. There may be substantial variability in
interpretation even when the same scanner and same criteria for carotid stenosis are used [36].
Although important, it may be difficult for the clinician to know the accuracy of his/her local
ultrasound laboratory. Accreditation by the Intersocietal Commission for the Accreditation of Vascular
Laboratories (ICAVL), a multidisciplinary group assures that the ultrasound data meet certain criteria,
including correlation against the gold standard of conventional angiography.
Transcranial Doppler — As an adjunct to CDUS, transcranial Doppler (TCD) examines the major
intracerebral arteries through the orbit and at the base of the brain. TCD is often used in conjunction
with CDUS to evaluate the hemodynamic significance of internal carotid artery (ICA) stenosis, and it
can be used to improve the accuracy of CDUS in identifying surgical carotid disease [38].
TCD can evaluate the intracranial hemodynamic consequences of high grade carotid lesions, such
as the development of collateral flow patterns in the circle of Willis, reversal of flow in the ophthalmic
and anterior cerebral arteries, absence of ophthalmic or carotid siphon flow, and reduced MCA flow
velocity and pulsatility [39,40].
An assessment of TCD by the American Academy of Neurology (AAN) concluded that TCD is
possibly useful for the evaluation of severe extracranial ICA stenosis or occlusion, but in general
carotid duplex and MRA are the tests of choice [41]. The AAN report noted that the clinical utility of
TCD to detect impaired cerebral hemodynamics distal to high grade extracranial ICA stenosis or
occlusion and assist with stroke risk assessment requires evaluation and confirmation in randomized
clinical trials.
We examined the sensitivity and specificity of TCD criteria in detecting a hemodynamically significant
stenosis (residual lumen diameter <1.5 mm) at the origin of the ICA [42].
• For the transorbital approach, the strongest indicators of a residual lumen diameter <1.5
mm were reversed flow in the ipsilateral ophthalmic artery (OA) and a >50 percent peak
systolic velocity difference between the carotid siphons (distal ICAs) in patients with
unilateral ICA origin stenosis. These findings were 100 percent specific and 31 percent
and 26 percent sensitive respectively.
• For the transtemporal approach in patients with a unilateral stenosis, a >35 percent
difference in ipsilateral middle cerebral artery (MCA) peak systolic velocity relative to the
contralateral MCA, or a >50 percent difference in contralateral anterior cerebral artery
(ACA) peak systolic velocity relative to the ipsilateral ACA were 100 percent specific for
identifying a residual lumen diameter of <1.5 mm. Sensitivities were 32 percent and 43
percent respectively. Irrespective of contralateral stenosis, a >35 percent difference in
ipsilateral MCA peak systolic velocity relative to the ipsilateral posterior cerebral artery
(PCA) had a 100 percent specificity and a 23 percent sensitivity for detecting a <1.5 mm
minimal residual lumen diameter.
TCD can also be used for detection of middle cerebral artery microemboli that arise from the heart or
carotid artery [43]. These are visualized as high intensity signal transients (HITS) within the Doppler
spectrum. Although preliminary evidence in some studies suggests that detection of asymptomatic
emboli associated with carotid stenosis may predict stroke risk [44], other studies have not confirmed
a significant relationship [45]. Thus, the clinical utility of microembolic signal detection by TCD is not
yet established [46]. A larger multicenter prospective cohort study is underway to determine whether
TCD embolus detection in the MCA is useful as a predictor of stroke risk [47,48].
Additional modalities — Newer modalities such as 3-dimensional ultrasound and compound B-
mode ultrasound may offer improved carotid plaque imaging compared with CDUS. If so, they may
provide a means of assaying carotid plaque features that are markers for different stages and
phenotypes of atherosclerosis.
3D ultrasound — Three dimensional (3D) ultrasound improves 3D visualization [49]. Advantages
compared with B-mode ultrasound include the potential for quantitative monitoring of plaque volume
changes in all three directions (circumferentially as well as length and thickness) rather than one or
two directions [50]. This in turn could allow measurement of plaque volume change, which may be a
more sensitive marker of plaque progression than measurements of plaque area, intima-media
thickness, and carotid stenosis.
Disadvantages of 3D ultrasound include a tendency for underestimation of vessel stenosis and
difficulty imaging areas of calcification [51].
Compound ultrasound — Compound ultrasound utilizes a technique called compounding to
average several images taken from different perspectives [52]. Advantages compared with B-mode
ultrasound include improved visualization of plaque texture and surface, as well as reduction of
artifacts [53]. In addition, reproducibility in the evaluation of plaque morphology appears to be good,
and interobserver agreement of plaque echogenicity is higher than with B mode [53]. Advances in
computational power have made real time compound imaging available for clinical practice.

MR ANGIOGRAPHY — The magnetic resonance angiography (MRA) techniques most often

employed for evaluating the extracranial carotid arteries utilize either two or three dimensional time-
of-flight (TOF) MRA or gadolinium-enhanced MRA (contrast enhanced MRA or CEMRA).
MRA produces a reproducible three dimensional image of the carotid bifurcation with good sensitivity
for detecting high grade carotid stenosis (show radiograph 2). In earlier studies, MRA was found to
generally overestimate the degree and length of stenosis [16,54]. However, a later study of three
dimensional TOF MRA found that it did not overestimate the degree of stenosis when corresponding
MRA and digital subtraction angiography (DSA) projections were compared [55].
Contrast enhanced magnetic resonance angiography (CEMRA) offers several advantages over
traditional TOF techniques. The use of a paramagnetic agent acting as a vascular contrast allows for
higher quality images that are less prone to artifacts.
Both TOF MRA and CEMRA are accurate for the identification of high-grade carotid artery stenosis
and occlusion, but appear to be less accurate for detecting moderate stenosis [56]. The sensitivities
of either MRA technique for the identification of carotid artery occlusion or severe stenosis were
similar and ranged from 91 to 99 percent, while specificities ranged from 88 to 99 percent.
Compared with carotid duplex ultrasound, MRA is less operator dependent and does produce an
image of the artery. However, MRA is more expensive and time-consuming than carotid duplex
ultrasound and is less readily available. Furthermore, MRA may not be performed if the patient is
critically ill, unable to lie supine, or has claustrophobia, a pacemaker or ferromagnetic implants [16].
In different series, up to 17 percent of MRA studies are incomplete because the patient could not
tolerate the study or could not lie still enough to produce an image of adequate quality for
interpretation [57].

CT ANGIOGRAPHY — Computed tomography angiography (CTA) provides an anatomic depiction

of the carotid artery lumen and allows imaging of adjacent soft tissue and bony structures. Three
dimensional reconstruction allows relatively accurate measurements of residual lumen diameter.
CTA may be particularly useful when CDUS is not reliable (eg, severe kinking, severe calcification,
short neck, or high bifurcation) or when an overall view of the vascular field is required [58].
A meta-analysis published in 2006 concluded that CTA compared with intraarterial cerebral
angiography for the diagnosis of 70 to 99 percent carotid stenosis had a sensitivity and specificity of
0.77 (95% CI 0.68-0.84) and 0.95 (95% CI 0.91-0.97), respectively [23].
An earlier systematic review and meta-analysis that compared CTA with arteriography or digital
subtraction angiography (DSA) concluded that CTA is an accurate method for detection of severe
carotid artery disease, particularly for detection of carotid occlusion, where CTA had a sensitivity and
specificity of 97 and 99 percent, respectively [59].
CT angiography requires a contrast bolus comparable to that administered during a conventional
angiogram. As a result, impaired renal function is a relative contraindication for its use, particularly in
patients with diabetes or congestive heart failure. (See "Pathogenesis, clinical features, and
diagnosis of radiocontrast media-induced acute kidney injury (acute renal failure)").

DIAGNOSIS OF COMPLETE OCCLUSION — No surgical treatment has been proven to be of

benefit for preventing a subsequent stroke in patients with complete carotid artery occlusion. Thus, it
is important to adequately distinguish between completely occluded vessels and those with some
remaining flow since the latter group may benefit from carotid endarterectomy.
In earlier reports, MRA and CDUS both had difficulties distinguishing very severe stenosis from
occlusion. False-positive and false-negative results occurred [60,61]. Combined techniques of 3D
and 2D TOF seemed to improve the performance of MRA but were not routinely performed
[16,54,62,63]. However, later studies indicated that improved recognition of near and total occlusion
is possible with CDUS [24] and MRA [24,64]. In addition, one report found that CTA was 100 percent
sensitive and specific for detection of near and total carotid occlusion [65].
Nevertheless, in current practice the combination of MRA and CDUS is probably sufficient for
identifying patients with carotid artery occlusion. This was illustrated in a study of 274 patients in
whom angiography identified 37 total and 21 near occlusions [66]. Ultrasound adequately identified
all angiographically determined total occlusions, but 3 of 21 near occlusions (14 percent) were
reported as totally occluded on ultrasound, and one of the three patients was a candidate for carotid
endarterectomy. MRA correctly identified 34 of 37 near total occlusions (92 percent) and all total
occlusions. The authors concluded that no further imaging is necessary when complete occlusion is
suspected on the basis of an initial CDUS study and confirmed on MRA.

CHOICE OF IMAGING TEST — Conventional cerebral angiography has been considered the gold
standard for the evaluation of internal carotid artery stenosis [67]. However, angiography is
associated with a small but real risk of stroke, which makes it ill suited to be used as a screening
test. In addition, most patients with ischemic symptoms referable to the carotid vascular territory do
not have severe carotid stenosis [68,69]. In one series of 261 patients with carotid ischemic strokes
and 813 patients with carotid TIA, carotid disease was absent in 55 and 64 percent, respectively
(and in 69 and 77 percent of those without a carotid bruit) [69].
As a result, patients are generally selected for angiography using one of the noninvasive tests
(CDUS, TOF MRA, CEMRA, and CTA). These noninvasive tests have essentially replaced
conventional cerebral angiography in the presurgical evaluation of carotid stenosis.
In a meta-analysis of 41 studies and 2541 patients published in 2006 that assessed different
noninvasive imaging methods, the following observations were made [23]:
• CDUS, MRA, CEMRA and CTA all have high sensitivities and specificities for diagnosing
70 to 99 percent carotid stenosis in patients with ipsilateral carotid territory ischemic
symptoms
 • CEMRA may be marginally more accurate than the other noninvasive methods, but this
technique is relatively new and the published studies included in the meta-analysis came
from research environments as opposed to routine clinical practice environments
• The accuracy of the noninvasive tests for 50 to 69 percent carotid stenosis appears to be
substantially reduced compared with 79 to 99 percent stenosis. However, the data are
sparse.
The combination of carotid ultrasound and MRA may obviate the need for conventional angiography
in the presurgical assessment of patients with carotid artery disease, particularly when the tests
agree [33,70,71]. Some have reported that the combination of ultrasound and MRA is cost-effective
[72,73] and results in an overall error rate that is comparable to the interobserver reliability when two
radiologists are presented with the same conventional angiogram revealing carotid artery disease
[74]. (See "Carotid endarterectomy: Preoperative evaluation; surgical technique; and
complications").
Bypassing angiography before surgery requires that noninvasive tests be highly specific as well as
sensitive. TCD may be beneficial in this setting, increasing the specificity of carotid duplex ultrasound
in detecting a <1.5 mm residual lumen diameter [42].

Conclusions — Our general approach to patients with suspected carotid stenosis is to first perform
carotid duplex ultrasound. Those with stenoses less than 50 percent are followed with serial
examinations, usually on an annual basis to determine if there is progression. If there is greater than
50 percent stenosis suspected, the patient is evaluated with transcranial Doppler examination and
MRA. CTA is performed in lieu of MRA if there is a contraindication to magnetic resonance imaging
and in cases where the duplex ultrasound and MRA do not agree.
Conventional angiography is rarely performed; indications include patients who cannot tolerate an
MRA and in whom the risk of dye is sufficient to warrant bypassing CTA in favor of the gold standard
examination. Angiography is also done if nonatherosclerotic disease is suspected (eg, dissection,
vasculitis).