Near-Occlusion is a Common Variant of Carotid Stenosis: Study and Systematic Review

ABSTRACT: Background: Symptomatic carotid near-occlusion is often described as rare. Recent studies have shown that near-occlusions are overlooked, especially near-occlusion without full collapse (with a small but normal-appearing distal internal carotid artery). Objective: To assess the prevalence of near-occlusion among symptomatic ≥50% carotid stenosis, incidence of symptomatic near-occlusion, and review the literature. Methods: Prospective controlled single-center cross-sectional study. Consecutive cases with symptomatic ≥50% carotid stenosis were examined with computed tomography angiography (CTA). The CTAs were assessed for near-occlusion by two observers. A systematic literature review was performed with emphasis on how study design affects prevalence estimate. Results: Totally, 186 patients with symptomatic ≥50% carotid stenosis were included, 34% (n = 63, 95% CI 27, 41) had near-occlusion. The incidence of symptomatic near-occlusion was 3.4 (95% CI 2.5, 4.2) per 100,000 person-years. Inter-rater κ was 0.71. The average prevalence of near-occlusion among symptomatic ≥50% carotid stenosis was higher in studies with good design (30%, range 27%–34%) than studies without good design (9%, range 2%–10%). Conclusions: Near-occlusion is common variant of symptomatic ≥50% carotid stenosis, both in the current study and in all previous studies of good design. Studies that suggest that near-occlusion is rare have had methodological issues.


INTRODUCTION
Carotid near-occlusion is a severe variant of carotid stenosis in which the artery distal to the stenosis is reduced in size. [1][2][3] This is thought to be a physiological reaction to the stenosiscaused reduction in blood volume and blood pressure beyond a stenosis. Near-occlusions can be subdivided into full collapse (threadlike distal artery, Figure 1A) and without full collapse (small but normal-appearing distal artery, Figure 1B). [1][2] Only stenosis without near-occlusion (conventional stenosis) should be assessed as percent when using NASCET grading. 4 In patients with symptomatic stenosis, near-occlusion should be separated from conventional stenosis as recommended management differs. [5][6] Carotid ultrasound has poor sensitivity for near-occlusion (13%). 7 Although near-occlusions can be assessed by computed tomography angiography (CTA), only 20% of near-occlusions are detected when CTAs are assessed in routine practice. 8 Many seem to apply the criteria for NASCET stenosis quantification with percent to all cases, but forget that NASCET called for assessment for nearocclusion before any measurements be done because whenever the ICA lumen is reduced from near occlusion, percent ratio calculations will be fallacious. 4 Thus, near-occlusion seems to be often overlooked in routine practice, especially those without full collapse. [7][8] As near-occlusion with full collapse less often undergoes revascularization than conventional stenosis, studies requiring treatment, such as large randomized trials, will likely underestimate the prevalence of near-occlusion. 3,9 Therefore, near-occlusions might not be 'rare', as they are often described to be. [10][11][12][13][14][15] Systematic reviews and meta-analyses of near-occlusion either do not present a prevalence estimate, 1-2 present near-occlusion as 'rare', [11][12]15 'relatively rare', [15][16][17] as ≤10% of 'cases' (of varying definition), [15][16][17][18][19][20] or as 20% of ≥70% stenosis. 17 In consecutive studies assessing near-occlusion prognosis, the prevalence of near-occlusion among symptomatic ≥50% carotid stenosis has been 27%-31%, but with possible selection bias. 9,21 No study has aimed to estimate the prevalence of symptomatic near-occlusion among patients with symptomatic ≥50% carotid stenosis. No systematic review has assessed how such prevalence is affected by study design (selection, modality, and near-occlusion definition). No estimate of incidence of symptomatic near-occlusion has been made.

Aims
The aims of this study were to assess the prevalence of symptomatic near-occlusion among symptomatic ≥50% carotid stenosis, the incidence of symptomatic near-occlusion, and perform a systematic review focusing on the impact of study design on near-occlusion prevalence estimate.

Prospective Study
A prospective controlled single-center study performed between February 2018 and March 2020. Cases with suspected symptomatic carotid stenosis (local admissions or by referral) assessed at Umeå Stroke Centre were included. All patients in Northern Sweden with suspected carotid stenosis aimed or revascularization are preoperatively assessed at this center, with 11 referring hospitals. Based on population data from the Swedish Bureau of Statistics, 1,869,135 person-years transpired during the study period in the study area. The study was approved by the regional ethics board in Umeå. All patients provided informed consent, written whenever possible.
Clinical exclusion criteria (applied before approaching patients for consent) were severe co-morbidity or major stroke as presenting event, making carotid revascularization clearly unfeasible. Only obviously unfeasible cases were excluded in this fashion, i.e., cases that underwent revascularization or were reasonable candidates were included, even when revascularization was not performed. CTA exclusion criteria were kidney failure and contrast allergy. Of included patients, analyses were restricted to those with symptomatic ≥50% stenosis on CTA (including near-occlusion, but not occlusion). Symptomatic was defined as having a recent (<6 months) ipsilateral ischemic stroke, TIA, amaurosis fugax, or retinal artery occlusion.
CTA was performed using various protocols at the referring hospitals or the department of radiology at the University hospital of Northern Sweden. All CTAs were reviewed by one observer (EJ). All cases with suspected near-occlusion and controls were reviewed by a second observer (AF) blinded to first reviewer. Both observers had near-occlusion expertise. Cases of disagreement handled by consensus discussion. As in previous studies, near-occlusion was sought prior to measurements for stenosis quantification. 4 Near-occlusion was diagnosed when a severe carotid stenosis was the most reasonable cause for a small distal internal carotid artery (ICA). 9,22 This was assessed by systematic interpretation of residual stenosis diameter, distal ICA diameter, distal ICA diameter compared to contralateral ICA (ICA ratio) and distal ICA diameter compared to ipsilateral external carotid artery (ECA ratio). 9,22 ICA asymmetry associated with Circle of Willis variation that mimic near-occlusion was recognized. 22 A conservative approach was used, diagnosing near-occlusion only when sufficiently certain. Among near-occlusions, those with a threadlike appearance of the distal artery were considered as full collapse. Occlusion was defined as no contrast seen beyond the stenosis. Conventional stenoses were graded with percent according to NASCET method, i.e., comparing smallest stenosis lumen diameter with the distal ICA well beyond the bulb. 4 Very severe stenoses, with less dense lumen opacity than the lumen proximal and distal (likely artery diameter ≤0.6 mm, the voxel size), were arbitrarily assigned to have a 0.5 mm diameter when visible and 0.2 mm when not visible.
We recognized that the use of ultrasound alone, 7 omitting near-occlusion without full collapse 1-3 and requiring treatment 9 or clinical selection to multiple exams, would likely affect near-occlusion prevalence estimate. Therefore, we defined good design as a consecutive sample without additional selection criteria (such as revascularization or requiring multiples exams), use of an angiographic technique and that near-occlusion without full collapse was clearly included in the definition of near-occlusion. Prevalence was analyzed separately for those with and without good design and by share of symptomatic cases. Articles with overlapping and/or pooled data were handled so that no patient was assessed several times for the same modality. However, patients examined with several modalities were assessed for each separate modality whenever feasible. The systematic review was performed according to PRISMA guidelines.
None of the four studies with mixed (48%) 30 or no data reported on share of symptomatic cases 7,23,28 had good design. The prevalence of near-occlusion among ≥50% stenosis was Cases with asymptomatic stenosis also studied and presented separately, excluded in this analysis.

THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES
31% when the only design flaw was requiring two exams and ranged 5%-13% when for other design flaws (Table 3).

DISCUSSION
The main findings of this study were a high prevalence of near-occlusion among symptomatic ≥50% carotid stenosis, and that similar high prevalence has been reported in all previous studies with good design, but was lower in studies without good design. Additionally, we made the first estimate of the incidence of symptomatic near-occlusion.

Main Findings
Our prevalence finding is likely much higher than what many perceive the prevalence of near-occlusion to be. This contrast between perceived and actual prevalence has several plausible causes. Near-occlusions are often overlooked on both ultrasound and when CTA are assessed in routine practice. [7][8] As presented in a previous literature review, many of the included and other studies have only included near-occlusion with full collapse (with a threadlike distal ICA, by some called 'string sign') in the definition of nearocclusion. 1-2 By doing so, near-occlusions without full collapse (with a small but normal-appearing distal ICA) are overlooked. To overlook near-occlusion without full collapse does not seem rational since 94% of near-occlusions were without full collapse in the large trials, 3 trial on which we base our guidelines. [5][6] Overlooking nearocclusion without full collapse does not seem to be by design, but a misunderstanding of how near-occlusion was diagnosed in the large trials. 2 Near-occlusion prevalence can also be underestimated by selection bias if studies require treatment 3,[25][26][27]29 or several exams 7 for entry. The low share (6%) of full collapse among near-occlusions in large trials 3 compared to consecutive series (40%-42%) 9,21 was likely in part explained by requiring treatment in the trials. In our analysis, we defined good design in order to separately assess studies with and without these potential sources of bias. In studies with good design, the prevalence of near-occlusion was markedly higher than in studies without good design. Some of the studies in the review were designed for other aims but were also assessable for prevalence of near-occlusion, even if they did not include the more inclusive near-occlusion definition of both part and full collapse. All three studies with good design had similar outcome. In the previous two studies, CTA was done by clinical selection (in 32% 9 and 71% 21 of respective populations), introducing the possibility of selection bias. The current study was designed in order to overcome this possible bias by ensuring that all eligible patients were examined with CTA. However, the current study also had possible selection bias as of 239 patients with symptomatic carotid stenosis, 78% were included, 15% were excluded by not providing informed consent, 6% were excluded due to clearly not candidates for revascularization, and 2% were excluded as angiography was not possible. With similar findings in all three studies, despite different types of selection, it seems unlikely that the high prevalence estimates were caused by selection bias.

Diagnostic Uncertainty in General
Separating near-occlusion and conventional stenosis is sometimes difficult. We used the state-of-the art approach, based on feature interpretation, similar to previous prognostic studies. 3,9,21 This approach has drawbacks as it requires experience. Although a conservative approach was used, some accentuation of nearocclusion cannot be excluded. Several conditions with small distal ICA similar to near-occlusion were accounted for. However, underestimation of near-occlusion is possible as there were many cases with unclear cause of small distal ICA. A previous study found similar issues with unclear cases. 22 We had reasonable inter-rater reliability, similar to a previous study. 9 This was possibly and underestimation as most cases without near-occlusion suspicion were not assessed by both observers, and such cases would likely have above average agreement. More importantly, as this study was performed by collaborating experts, though in separate location, the assessed reliability was likely higher than when applied in routine practice.
Better diagnostic approaches for separating conventional stenosis and near-occlusion are warranted as near-occlusion is currently overlooked, 7-8 current state-of-the art has feasibility issues for routine practice use, and recommended management differs. [5][6] The high risk of recurrent stroke in near-occlusion with full collapse also warrants further clarification. 9,21 These method improvements could include physiological approaches, such as velocity distal to the stenosis on carotid ultrasound 31 or phase contrast MRI. 32 That near-occlusions are not rare but common should reasonably increase the priority of these studies.

Study Strengths and Weaknesses
The strengths of this study were prospective design situated at a stroke unit doing preoperative evaluations (not a revascularization clinic), steering cases to CTA, dedicated to nearocclusion analysis, and state-of-the art diagnostic approach. The sample size was moderate, but the 95% CI still excluded the possibility of near-occlusions being rare. As we did not assess cases that were clearly not eligible for revascularization, our findings are foremost applicable to cases eligible for revascularization. Some cases in referring hospitals might not have been sent and there was a moderate rate of refusing study participation. Only if such cases were more or less often near-occlusions than those included (which is unknown) would it would affect the prevalence estimate. However, some symptomatic nearocclusions in the study area during the study period might not have been included, why the incidence estimate was likely an underestimation.

Summary
Near-occlusion is a common variant of carotid near-occlusion. Previous assessments of near-occlusion being rare can be explained by suboptimal methodology.

FUNDING
This study was funded by Knut and Alice Wallenberg Foundation, Region Västerbotten, the research fund for neurological research at the university hospital of northern Sweden, the Jeansson foundation, the Swedish Stroke fund, and the Northern Swedish Stroke fund. Disclosures: The authors have no conflicts of interest.

DISCLOSURES
The authors have no conflicts of interest.

STATEMENT OF AUTHORSHIP
EJ: Idea, concept, data collection, funding, analyses, and manuscript draft. AJF: Data collection, manuscript edit.