Two randomized controlled trials, the ACAS in 19951 and ACST in 20042 reported that in patients with asymptomatic ICA stenosis greater than 60-70% (NASCET) carotid endarterectomy reduced the risk of stroke from 2% to 1% per year. In these trials carotid endarterectomy was associated with a 2-3% perioperative rate of stroke or death. However, medical intervention which was left to the discretion of the local teams was suboptimal in relation to current practice. For example, in the ACST study, statins and antiplatelet agents were administered to only 25% and 80% of patients respectively.2
Current evidence indicates that the average annual risk of ipsilateral cerebral and any territory stroke among patients with asymptomatic moderate-severe ICA stenosis receiving medical intervention alone has fallen to approximately 1%3,4 making routine carotid endarterectomy unjustified. However, if patient subgroups with sufficiently higher than average risk, despite current optimal medical intervention, could be reliably identified, then carotid surgery may still be justified. This is now possible using a number of texture features from normalized carotid plaque images obtained with ultrasound.5,6t
Ultrasonic texture features associated with stroke
Several factors that have been proposed to predict stroke in asymptomatic carotid stenosis and thus contribute to risk stratification, include severity of stenosis7, echolucent (hypoechoic) plaque,8-11 low gray scale median (GSM),12,13 increased percentage of echolucent plaque components,13 clinical risk factors,12 the presence of discrete white areas (DWA) without acoustic shadowing12 and the presence and size of a juxtaluminal black plaque without a visible echolucent cap.6
It should be noted that image normalization is essential for accurate reproducibility of the above texture features. Image normalization is performed by using two reference points: blood (gray scale = 0) and adventitia (gray scale = 190). (The gray scale in a computer rages from 0 which is absolute black to 255 which is absolute white).
Histology studies have shown that in unstable, symptomatic plaques the necrotic core is twice as close to the lumen compared with asymptomatic plaques.14 These findings correspond to subsequent observations made by cross sectional studies using ultrasound, showing an association between juxtaluminal black (hypoechoic) area (JBA) and the presence of patient neurological symptoms.15-17 A JBA in ultrasonic images of asymptomatic carotid artery plaques is associated with a lipid core close to the lumen on histologic examination of carotid endarterectomy specimens.18 A cross-sectional study of patients with symptomatic and asymptomatic plaques has demonstrated that a JBA greater than 8mm2 in the absence of a visible echogenic cap was associated with a high prevalence of symptomatic plaques in all grades of stenosis.15
Further correlations between ultrasonic texture features and plaque histology have been made more recently.
A low GSM is associated with a histologically unstable plaque,19 less calcification, 20 low collagen content, a large lipid core, a thin fibrous cap,19,21 increased inflammation and neovascularization.21 In symptomatic patients, a low GSM (< 25) is associated with recurrent symptoms.22
DWA in a hypoechoic plaque are associated with intraplaque haemorrhage inflammation and inflammation in one study20 and with neovascularization, increased number of macrophages and intraplaque haemorrhage in others 21,23
A JBA near the luminal portion of the plaque (without a visible cap) is associated with a necrotic core located close to the lumen on histology,18 macroscopic plaque ulceration,20 decreased numbers of smooth muscle cells, large lipid core, thion fibrous cap and plaque rupture.21
Stroke risk stratification in patients with asymptomatic carotid stenosis
The potential of combinations of ultrasonic plaque texture features for stratifying the risk of ipsilateral stroke in patients with asymptomatic carotid stenosis has been investigated in the Asymptomatic Carotid Stenosis and Risk of Stroke (ACSRS) Study, which was a multicentre cohort study of 1121 patients with asymptomatic ICA stenosis undergoing medical intervention alone, followed up for a mean of 4 years.5,6
In two independent analyses and publications of the ACSRS material,5,6 stenosis, gray scale median, plaque size, presence of discrete white areas in hypoechoic plaques, size of a juxtaluminal black area and history of contralateral TIAs have been found to be independent predictors of risk and could be used in combination to stratify stroke risk. The annual stroke risk varied from 0.4% to 10%. In the presence of greater than 70% diameter stenosis, the annual risk of stroke was <1% in 722 patients, 1-1.9% in 110, 2-3.9% in 162, 4-5.9% in 70 and 6-10% in 57.
A key message that emerged is that not a single ultrasonic feature can identify all the potentially unstable and high risk plaques. This is because there are several mechanisms that result in embolization. Some plaques produce emboli because of a thrombus on their surface, some because they rupture having a large lipid core and a thin fibrous cap (not visible on ultrasound), while others rupture because of mechanical forces irrespective of their structure. This is why a combination of plaque features performs better than a single feature alone.
Implications on current practice
In the light of current knowledge, it is no longer appropriate to operate on patients with asymptomatic carotid stenosis without stratifying patients into different risk categories. Already, several vascular surgeons and neurologists expect risk stratification to be part of a patient’s vascular work-up.
Even if we assume a 50% reduction in stroke rates by modern medical therapy which includes use of statins with an LDL cholesterol target of < 70 mg/dl, there will still be many patients with an annual stroke risk greater than 2%.
The ESVS 2017 Guidelines24 recommend the following: “In ‘average surgical risk’ patients with an asymptomatic 60-99% stenosis, carotid endarterectomy should be considered in the presence of one or more imaging characteristics that may be associated with an increased risk of late ipsilateral stroke, provided documented perioperative stroke/death rates are <3% and the patient’s life expectancy exceeds 5 years (level of evidence B, Class IIa)”.
Imaging characteristics that may be associated with an increased risk of late ipsilateral stroke as quoted by the ESVS 2017 Guidelines are: Silent infarction on CT/MRI, stenosis progression, large plaque area, large juxta-luminal black area (JBA) on computerised plaque analysis, plaque echolucency, intra-plaque haemorrhage on MRI, impaired cerebral vascular reserve (CVR), and spontaneous embolisation on transcranial Doppler (TCD) monitoring.
Reproducible identification and measurements of the ultrasonic texture features quoted, large plaque area (> 80 mm2), large juxta-luminal black area (JBA) (> 8 mm2) without visible echogenic cap, plaque echolucency (GSM <15) and presence of DWA (not quoted in the guidelines) can only be made after image normalisation 5,6,15 (see below).
Software for image analysis
Reproducible measurements of plaque texture features such as GSM and size of a juxtaluminal black plaque area can only be obtained if ultrasound images are normalized. This is because when ultrasound is performed in a dimly lit room the gain is reduced and when performed in a brightly lit room the gain is increased. Image normalization so that the gray scale value of blood in the vessel lumen is zero and of adventitia 190 has overcome this problem. Normalization is performed by the software produced by LifeQ using the above two reference points and linear scaling. In addition, the software incorporates the ACSRS algorithm for risk stratification5,6,15 and provides an automatic report on the key texture features including a plaque image and the 5 year risk of stroke (also annual stroke risk).
Image normalization which is essential prior to image analysis and measurement of texture features including JBA size using our software is relatively simple on a laptop adding an extra 10-15 minutes to the conventional ultrasound examination.
Instructions for the ultrasonographer for optimal equipment settings such as dynamic range, TGC, overall gain adjustments, having the image beam at right angles to the vessel wall etc 5 are provided in the Appendix of the user’s manual. Capturing the appropriate views of plaque images using ultrasound is within the capability of every vascular ultrasonographer adding very little extra time to the conventional velocity measurements for grading the degree of stenosis.
A one day training course on a range of 25 standard images is included in the software price. This will ensure that the operator and his team can obtain reproducible measurements of texture features after image normalization.