The clinical implications of adopting new criteria for the grading of internal carotid artery stenosis

Abstract

Introduction

A patient's eligibility for carotid endarterectomy (CEA) is determined primarily by the degree of carotid stenosis detected on duplex ultrasound. The Australasian Society for Ultrasound in Medicine (ASUM) criteria are widely used to grade carotid stenoses in many practices throughout Australasia.

Methods

We sought to investigate the potential impact on the grading of carotid artery stenosis if practitioners switched from the ASUM criteria to the United Kingdom's joint recommendation (UKJR) criteria by reviewing 100 patients with a haemodynamically significant carotid artery stenosis.

Results

We found agreement between the criteria in 100% of cases for stenoses <50%, in 80% of cases for stenoses 50–69%, in 89% of cases for stenoses ≥70% and in 100% of cases for stenoses ≥80%. While there was variation in grading of stenoses in 16% of cases, reclassification resulted in no change in the number of patients eligible for CEA. The UKJR guideline enabled more precise categorisation of haemodynamically significant stenosis into deciles.

Discussion

Because the UKJR guideline is more comprehensive, we believe that adopting this guideline would enable the ultrasound practitioner to grade carotid stenoses more precisely, better understand the nuances of carotid duplex imaging and more successfully navigate and interpret complex carotid examinations, without impacting the number of patients eligible for CEA.

Introduction

Stroke is a leading cause of mortality and disability in both low‐income and high‐income countries worldwide,1 and it has been estimated that extracranial carotid artery disease accounts for 11.5% of all ischemic strokes.2 Unlike other causes of ischemic stroke, carotid arterial disease is readily treatable, as carotid endarterectomy (CEA) has been shown to decrease the risk of recurrent stroke in patients with significant carotid stenosis.3, 4, 5

The severity of internal carotid artery (ICA) stenosis is the major determinant of treatment choice. In symptomatic patients with >70% ICA stenosis, there is strong evidence of benefit with CEA, and more modest benefit in symptomatic patients with 50–69% stenosis provided the CEA is performed within 2 weeks of the presenting event.6 In asymptomatic patients with ICA stenosis >70%, CEA may be beneficial if performed in an institution with low perioperative complication rates.7 Duplex ultrasound is the modality of choice for diagnosing and classifying the severity of ICA stenosis and is often the only imaging modality used prior to intervention. Therefore, accurate classification of stenosis severity is crucial for surgical decision‐making.

Since 2006, the Australasian Society for Ultrasound in Medicine (ASUM) has provided a basic guideline for the classification of carotid stenosis severity (Appendix S1). In September 2015, ASUM briefly adopted new criteria based on the United Kingdom's joint recommendations (UKJR) for reporting carotid ultrasound investigations,8 before returning to the original ASUM guidelines, which are currently under revision. The considerations that led to the change in the criteria and the subsequent return to the original ASUM criteria have not been disclosed. Compared with the original ASUM criteria, the UKJR criteria contain marked differences when classifying patients with ≥50% ICA stenosis. In particular, the peak systolic velocity (PSV) for categorising a ≥70% stenosis varies between >270 cm/s (ASUM) and >230 cm/s (UKJR). There are also newly introduced categories of ‘>90% stenosis’ and ‘near occlusion’. Finally, the St Mary's ratio (ICA_PSV/CCA_{end‐diastolic velocity (EDV)}) was introduced to allow classification of patients with haemodynamically significant ICA stenosis into stenosis deciles (50–59%, 60–69%, 70–79%, 80–89%, >90% but less than near occlusion).

This study was undertaken to compare the ASUM criteria with the UKJR criteria8 when classifying the severity of carotid stenosis and to determine the implications of changing criteria on the number of patients receiving CEA.

Methods

Carotid ultrasound scans performed prior to 1 September 2015 in the Waikato Vascular Laboratory were identified through the picture archiving and communication system (PACS) retrospectively until 100 unique eligible patients with a haemodynamically significant (≥50%) ICA stenosis, as per the ASUM criteria, were identified for analysis.

All ultrasound scans were performed on either the GE Logiq 9 (GE Healthcare, Milwaukee, WI, USA) or the Philips IU22 (Philips Healthcare, Bothell, WA, USA) by qualified vascular sonographers or by trainee vascular sonographers under direct supervision.

Exclusion criteria included:

• The absence of haemodynamically significant ICA stenosis bilaterally (<50% stenosis) as per the previous ASUM criteria (Appendix S1)

• Previous carotid interventions such as CEA or stenting

• Bilateral ≥70% ICA stenosis or unilateral occlusion based on reported results as this can cause overestimation of stenosis in the less severe side8

• Multiple repeat scans on the same patient within the study period (only the most recent study was included)

The data collected for each eligible case included:

• Date of the scan

• Patient's national health index (NHI) number, age and gender

• The PSV and EDV of the distal common carotid artery (CCA)

• The PSV and EDV at the point of the highest ICA stenosis

• Reported degree of stenosis

• Presence or absence of a cerebrovascular event and the timing of this in relation to the ultrasound scan

• Whether the stenosis was on the symptomatic side (e.g. contralateral side for motor symptoms or ipsilateral for transient monocular vision loss)

• If surgical intervention occurred

• Lead vascular consultant

• Referral source

Patient outcomes were obtained by reviewing the electronic records in the Clinical Results Viewer. The degree of carotid stenosis was recategorised using both the ASUM and UKJR criteria.

This study was a retrospective audit of clinical data approved and registered by the Waikato District Health Board Clinical Audit Support Group.

Results

We retrospectively identified 100 eligible patients (200 carotid arteries) with a haemodynamically significant ICA stenosis out of a total of 574 carotid duplex examinations performed on 566 unique patients at Waikato Hospital's Vascular Laboratory between February 2015 and September 2015 as per the inclusion and exclusion criteria (Figure 1).

Figure 1.

Exclusion Criteria of Patients Scanned Between 25.02.2015 and 31.08.2015 to Obtain 100 Eligible Patients.

The prevalence of haemodynamically significant (≥50%) ICA disease (unilateral or bilateral) in patients referred for carotid duplex examinations was 32% (181/566); conversely, 68% (385/566) of patients had no significant stenosis. One percentage of patients were excluded (6/566) due to bilateral >70% ICA stenosis and 5% (27/566) due to an ICA occlusion.

Table 1 outlines the demographics and clinical presentation of the 100 patients with ≥50% ICA stenoses. The majority of these scans were performed on patients who had experienced recent neurological or visual symptoms. Scans performed on asymptomatic patients were for surveillance scans, pre‐surgical workups (such as prior to coronary artery bypass) or to investigate carotid bruits. The number of all ICAs in each stenosis grade when applying both the ASUM and UKJR criteria is presented in Table 2/Figure 2.

Table 1.

Demographic and clinical presentation of 100 patients with >50% ICA stenoses

Age, mean (range)	75 (50–93)
Sex
Male	64%
Female	36%
Cerebrovascular event?
None	39%
Present	61%
Primary symptomatic event (n, %)
Ocular only	6 (10%)
TIA	31 (51%)
Stroke	24 (39%)
Time since last event (n, %)
<2 weeks	34 (56%)
2–4 weeks	5 (8%)
>4 weeks	22 (36%)

Table 2.

Re‐classifying the grade of stenosis in 200 ICAs using UKJR criteria

ASUM criteria % stenosis	n	UKJR criteria % stenosis			n
		Decreased	No change	Increased
<50%	76		<50%		76
50–69%		<50%			11
	88		50–69% (combined)		70
			50–59%		31
			50–69%		17
			60–69%		22
				70–79%	6
				70–89%	1
70–79%	23	50–69%			1
		60–69%			3
			70–79% (combined)		9
			70–79%		8
			70–89%		1
				80–89%	5
				>90%	5
80–99%	13		80–99% (combined)		13
			80–89%		2
			>90%		1

The left two columns shows the number of ICAs in each stenosis category when applying the ASUM criteria.The right four columns shows the number of cases after reclassification using the UKJR guideline.

Figure 2.

Reclassification of 125 Internal Carotid Arteries with >50% Stenoses According to the ASUM Criteria (dark blue circles) and the UKJR (Yellow Triangles). Blue Arrows Depict a Decrease in Stenosis or Non‐Significant Increase in Degree of Stenosis. Red Arrows Show an Increase in Degree of Stenosis as per the ASUM Criteria.

ASUM stenoses <50%

There is perfect agreement between the ASUM and UKJR criteria for the threshold of a haemodynamically significant stenosis (50%). Specifically, no ICA graded as <50% stenosis by ASUM criteria was recategorised as a >50% stenosis when applying the UKJR criteria. However, some variation in grading stenoses was observed with stenoses >50%. When the UKJR criteria were applied, 16% (32/200) of all stenoses shifted into a different category.

ASUM stenoses 50–69%

In 88 patients with a 50–69% stenosis, 70 (79.5%) remained unchanged when applying the UKJR criteria, 11 (12.5%) shifted downwards becoming a <50% and 7 (8.0%) shifted upwards becoming >70%.

ASUM stenoses 70–79% and all ASUM stenoses ≥70%

In 23 patients with a 70–79% stenosis, 8 (34.8%) remained unchanged when applying the UKJR criteria, 4 (17.4%) shifted downwards becoming a 50–59% or 60–69% stenosis and 11 (47.8%) shifted upwards becoming >80%. As the threshold for CEA is commonly considered a high‐grade ICA stenosis of ≥70%, we also analysed all patients in this broader category. In 36 patients with high‐grade ICA stenoses ≥70% (ASUM categories 70–79% and 80–99%), 32 (88.9%) remained unchanged and only 4 (11.1%) shifted downwards becoming a 50–59% or 60–69% stenosis.

ASUM stenoses 80–99%

In 13 patients with an 80–99% stenosis, none shifted into a different category; 2 became classified as subcategory decile 80–89% stenosis and 11 as >90% but less than near occlusion.

Surgical intervention and potential changes in patient management

Twenty‐one patients underwent CEA and had the following ASUM stenosis grading; five 50–69% stenoses; seven 70–79% stenoses and nine 80–99% stenoses. Six of these stenoses (29%) increased to a higher category when applying the UKJR criteria. Such reclassification would have strengthened the case for carotid CEA in these patients. In one patient (5%) who received carotid CEA, the degree of stenosis decreased following UKJR reclassification from 70–79% stenosis to a 60–69% stenosis. This patient presented within 2 weeks of the symptomatic event and would have therefore likely still received CEA despite being downgraded.

The majority of carotid CEA were performed on symptomatic patients. Of the 24% (5/21) of patients operated on for asymptomatic carotid stenoses, all were carried out on ASUM and UKJR stenoses >70%.

There is strong evidence of benefit for CEA in >70% ICA stenosis, and more modest benefit in patients with a 50–69% stenosis provided the CEA is performed within 2 weeks of the presenting event.6 When applying this strategy with both the ASUM and UKJR criteria, an identical number, 28% (28/100), of patients were eligible for intervention.

Criteria agreement

When using the UKJR criteria, all cases had two or more diagnostic criteria in agreement. All three diagnostic criteria were in agreement in 64% (64/100) of cases.

Discussion

Carotid artery duplex is the most appropriate first‐line imaging modality for the assessment of carotid artery disease.9 For the last 15–20 years, duplex ultrasound had been used safely and effectively as the only imaging modality prior to CEA in the vast majority of patients.10

Unlike digital subtraction angiography (DSA) which allows for direct measurements of luminal stenosis, duplex ultrasound relies primarily on indirect quantification of stenosis by assessing flow velocities (PSV, EDV) and velocity change ratios (ICA/CCA ratio or St Mary's ratio).8, 11 These parameters are subject to considerable intraobserver and interobserver variabilities. It is widely recognised that ultrasound is a highly operator‐dependent imaging modality. The technique and experience of the operator may significantly affect velocity measurements as well as the final interpretation of the examination.12, 13, 14 A particularly concerning disagreement in the classification of carotid stenosis has been shown between initial outpatient scans vs. specialist vascular laboratory scans15 prompting the recommendation to rescan all patients referred for specialist opinion with an outside diagnosis of ICA stenosis.

Apart from the broader issues of interobserver variability, a wide range of very specific factors may influence the operator's ability to accurately acquire, measure and interpret flow velocities, including (i) anatomic considerations (vessel geometry, luminal geometry, width of ICA bulb, stenosis length), (ii) physiologic factors (cardiac output, aortic valvular disease, hypertension, loss of vessel compliance, presence of coexisting disease at other sites, presence of high‐grade contralateral disease, presence and recruitment of collateral vessels) and (iii) technical factors (angle uncertainty in cases of shadowing or non‐laminar flow, intrinsic spectral broadening caused by non‐uniform beam geometry16, 17 as well as spectral Doppler gain settings8). Given the range of nuances associated with the performance and interpretation of carotid duplex ultrasound, the need for reliable, validated and highly specific scanning techniques and interpretation criteria cannot be overemphasised.

With the initiation and eventual publication of the major carotid surgery trials in the 1990s (NASCET, ECST, ACAS), a complex and dynamic debate took place in the literature regarding the use of various ultrasound criteria for the characterisation of ICA stenosis. A wide variety of criteria were developed by numerous groups often with marked interinstitutional differences.18, 19, 20, 21, 22, 23, 24, 25, 26, 27 For instance, the value of ICA PSV for the diagnosis of a ≥70% ICA stenosis varied from ≥130 cm/s21 to ≥325 cm/s.28 At the same time, it was becoming increasingly recognised that flow velocity measurements may not be absolute, but may indeed be operator and machine dependent 29, 30 even when performed under highly controlled conditions.31 More importantly, however, the performance of ultrasound criteria for the diagnosis of ICA stenosis tended to drift and appeared to vary between individual operators, between different laboratories, between different ultrasound systems and even within the same laboratory and the same machine over time.14, 32, 33, 34, 35, 36 Achieving a set of consistent ultrasound criteria that could be rolled out and broadly adopted on a regional, national and international scale appeared to be an unattainable and elusive goal. The difficulty in locking in universal carotid criteria led most authors to conclude that the only way forward was for each laboratory to internally validate criteria they had adopted from others or to develop their own validated criteria.14, 33, 36, 37, 38, 39, 40, 41

While validation mechanisms are well established, these primarily rely on the comparison of duplex ultrasound with a gold standard or on less robust methods such as second reader, peer review of duplex findings or correlation with surgical findings.38, 42 Unfortunately, the requirement for rigorous validation brings with itself a great number of challenges. Firstly, the best form of validation is by DSA but this relies on ready access to an angiography suite, is expensive and carries a significant risk of complications including a small risk of permanent disability.43 With the increasing uptake of ultrasound as the sole imaging modality prior to CEA, DSA is nowadays rarely performed. Consequently, the role of DSA as an appropriate validation tool has diminished. Validation by other modalities such as computed tomography angiography (CTA) or magnetic resonance angiography (MRA), while less ideal, is feasible, but is also associated with significant limitations including additional costs (CTA and MRA), limitations in the use of nephrotoxic contrast (CTA), technical limitations associated with dense calcific plaques (CTA), interpretation of flow voids (MRA) as well as local, regional or national limitations in the access to MRA. Neither CTA nor MRA is considered a gold standard for the characterisation of ICA stenosis. CTA and MRA are better suited as adjuncts to carotid duplex in complex cases and are more commonly employed to evaluate the state of the extracranial circulation beyond the sites accessible of duplex ultrasound such as the aortic arch and proximal vessels, as well as the distal ICA, circle of Willis and intracranial vessels. A further challenge to meaningful validation is the fact that the number of patients presenting with surgically significant ICA stenosis (≥70%) is relatively modest. Smaller centres that perform a fraction of the carotid duplex work compared to our institution would struggle to recruit sufficient number of patients for a meaningful validation programme or such programme would have to span a long course of time during which changes in staffing and equipment may render validation problematic and of questionable clinical value. Outpatient centres, such as smaller private radiology practices, would not have the ability to validate their findings at all due to even smaller number of patients, lack of access to confirmatory imaging and/or difficulty in gaining access to records of confirmatory imaging performed at regional hospitals due to roadblocks associated with privacy legislation. A less robust validation by a second reader from static images obtained by an initial operator suffers from inherent biases of the examination already performed. Only a real‐time rescan by a blinded peer reviewer may provide a valid form of an audit, but this may still be prone to selection bias and does not offer a gold standard. Finally, validation against surgical findings is feasible, but imprecise and difficult to quantify. In summary, validation of carotid duplex is a goal that is idealistic and only achievable in high‐volume expert‐level institutions with easy access to confirmatory imaging and vascular surgery. The unacknowledged clinical reality is that the vast majority of practitioners who perform carotid duplex ultrasound do not engage in a formal validation process. Indeed, there appears to be a degree of complacency with regard to the application of carotid Doppler criteria so that the majority of users simply adopt some form of historical criteria with no functional knowledge of how the criteria were derived or where they come from.44 Even some published guidelines are unreferenced.45, 46

It follows intuitively that the application of different criteria to the same observations changes the proportion of high‐grade carotid stenoses diagnosed.47 The flow‐on effect is that institutional differences in carotid duplex criteria may result in significant variability in classification of ICA stenosis, and this may lead to striking discrepancies in the access to CEA, resource allocation and the care of patients.8, 48, 49 Attempts to reverse the frustrating and often impracticable trend for ongoing validation have been made with the use of several expert‐lead consensus statements in the US,23 UK,8 Germany50 and Poland.51 When adopted in clinical practice, the techniques and interpretation criteria from a consensus approach work well in clinical practice, particularly for the key threshold of a surgically significant carotid stenosis of ≥70%.52, 53 While the consensus statements do vary in some aspects, there is generally good agreement on the main technical and interpretive points. It may be feasible and indeed desirable in future to aim for an international consensus on the performance of and criteria for the interpretation of carotid duplex ultrasound.54

Our study demonstrates that an adjustment in diagnostic criteria from the ASUM criteria to UKJR criteria results in: (i) perfect agreement on the threshold of a haemodynamically significant stenosis in 100% of cases, (ii) only a modest change in the classification of haemodynamically significant stenoses in the 50–69% category, (iii) good agreement on the threshold of a surgically significant stenosis ≥70% in 89% of cases and (iv) perfect agreement on the threshold of a critical (≥80%) stenosis in 100% of cases. Most importantly, our study shows that reclassification of carotid stenoses resulted in no change in the number of patients eligible for carotid CEA despite variation in grading of stenoses in 16% of cases.

Our study had several limitations. Despite a review of a large number of examinations (574), the number of ICA stenoses in each category was relatively modest. This was particularly the case in the high‐grade stenosis 70–79% (23 stenoses) and critical stenosis (13 stenoses). We also excluded from our analysis those patients who presented with bilateral stenoses of which at least one was high‐grade as well as patients with unilateral occlusion to avoid the complex and sometimes somewhat subjective discussion of how stenoses should be classified in the presence of high‐grade contralateral disease. The number of these patients was very small (6, 1%). Finally, as the above discussion regarding carotid duplex validation suggests, the results of the carotid duplex studies included in this study were not formally validated against DSA or any other imaging. However, in no patient receiving CEA did the surgical findings contradict the ultrasound impression.

Conclusion

Changing from the current ASUM carotid guideline to the UKJR guideline would have negligible impact on the characterisation of the major carotid stenosis categories while allowing more precise characterisation of stenoses into deciles. There would be no appreciable impact on the access of patients to CEA. We propose that in comparison with the ASUM guideline, the clinical advantage of the UKJR guideline lies in its comprehensive technical and physiological information, concrete examples of correct sampling and measurement techniques, disclosure of common pitfalls and rigorous academic summary of the relevant issues with accompanying references. We believe that adopting the UKJR guideline would enable to the ultrasound practitioner (sonographer, radiologist, vascular surgeon) to better understand the nuances of carotid duplex imaging and more successfully navigate and interpret complex carotid examinations.

Disclosure

The authors have no conflict of interests to declare.

Authorship declaration

The authors certify that this manuscript is original, has been written by the stated authors, has not been published previously and is not under consideration for publication by another journal. Moreover, each author certifies that they have participated to a sufficient degree to take public responsibility for the work and believes that the manuscript describes truthful facts. They declare that they shall produce the data on which the manuscript is based for examination by the editors or their assignees, should it be requested.

Supporting information

Appendix S1. Guidelines, policies and statements