Evaluation of ocular blood flow in the assessment of symptomatic carotid stenosis

Abstract

Background and Purpose

The degree of internal carotid artery (ICA) stenosis determined by criteria from the North American Symptomatic Carotid Endarterectomy Trial (NASCET) is not the most accurate index to assess distal flow compromise. Distal ICA perfusion is also determined by factors such as tandem carotid stenosis and collateral circulation. Quantification of end-organ ocular perfusion using non-invasive laser speckle flowgraphy (LSFG) may provide insights into distal ICA flow. This study prospectively assessed the degree of ICA flow using LSFG.

Methods

Eighteen patients with symptomatic carotid stenosis underwent LSFG evaluation. LSFG was used to extract ocular blood flow metrics recorded simultaneously in the retina, choroid, and optic nerve head. The following ocular flow parameters were measured with LSFG: mean blur rate (MBR), flow acceleration index (FAI), and rising rate (RR). Syngo iFlow perfusion imaging was used to objectively quantify contrast flow in the ICA and brain parenchyma during digital subtraction angiography. Time to peak (TTP) and contrast delay were extracted from seven different regions of interest (ROIs).

Results

MBR, FAI, and RR were correlated with NASCET degree of stenosis. FAI and RR also improved after stenting. TTP improved after stenting in three ROIs. A moderate negative correlation was observed between FAI and contrast delay.

Conclusions

LSFG non-invasively quantifies end-organ blood flow distal to the ICA origin. LSFG metrics have the potential to quantify end-organ perfusion and determine if a proximal carotid stenosis is symptomatic.

Introduction

Internal carotid artery (ICA) atherosclerosis is a major risk factor for stroke. ¹ The North American Symptomatic Carotid Endarterectomy Trial (NASCET) allows for stratification of severity of stenosis based on two-dimensional measurements of luminal diameters.^2–4 Strata derived from NASCET include severe (70−99%), high moderate (50−69%), and low moderate stenosis (<50%).^3,5 However, the clinical significance of these measurements is confounded by compensatory mechanisms, such as collateral flow and the presence of tandem stenosis along the ICA course. ⁶ This can lead to a mismatch between symptoms and the degree of stenosis. Analysis of the distal ICA may yield additional information about distal blood flow and end-organ perfusion. Blood flow in the ophthalmic artery is sensitive to perfusion changes in the ICA.^7,8 Hence, the quantification of ocular tissue perfusion may provide a more accurate assessment of distal ICA flow and account for compensatory mechanisms such as collateral flow.

Laser speckle flowgraphy (LSFG) assesses retinal perfusion using the speckle pattern produced by a laser scattered by blood cells moving in the ocular fundus. ⁹ LSFG is a non-invasive technique that can measure and visualize real-time ocular circulation. It does not require contrast. It has been documented in a case report that LSFG correlated with improved ocular blood flow after reconstructive vascular bypass surgery in the setting of an ICA occlusion. ¹⁰ To our knowledge, no studies have evaluated the accuracy of LSFG in assessing ocular perfusion in patients with symptomatic carotid stenosis treated with stenting. The aim of this pilot study is to evaluate the potential of LSFG in quantifying distal ICA flow. Furthermore, assessing distal ICA perfusion using LSFG may provide objective measurements of collateral flow and its impact on ocular perfusion.

Methods

Patients with symptomatic carotid stenosis were prospectively recruited for the study after approval by the institutional review board at our institution. All participants gave informed consent. Patients with severe ICA stenosis (≥ 70%) and high moderate ICA stenosis (50–69%) were included in the study.^3,5 Symptomatic stenosis was defined as ICA stenosis resulting in symptoms of amaurosis fugax or neurological deficits localized to the ipsilateral vascular territory of the ICA. ¹¹ If possible, LSFG scans were performed before and after stenting. An objective measure of cerebral perfusion was obtained with syngo iFlow (iFlow, Siemens). iFlow was used to quantify an index of contrast velocity in the cerebral vasculature using digital subtraction angiography (DSA) images.

LSFG

LSFG data were acquired with LSFG-NAVI^® (Softcare Co., Ltd, Fukuoka, Japan). Blood flow maps of the retinal vasculature were created. These blood flow maps were then analyzed using Cobitos software (v.1.0.58.0, Softcare Co., Ltd). The degree of fundus pigmentation was corrected by the software to account for influence of the pigmentation on laser speckle blood flow. Blood flow at the optic nerve head was isolated with an elliptical “rubber band” region of interest (Figure 1). This elliptical region (STIN) was further broken down into quadrants (superior, temporal, inferior, and nasal) for more detailed spatial analysis. The software separated pixels captured in these regions between and within the branches of the central retinal artery and vein passing over the optic nerve head. The pixels in the optic nerve head between branches of the retinal arterioles and venules represent the prelaminar and underlying laminar optic nerve capillary beds. The pixels within the retinal arterioles and venules represent blood flow to the retina.

Figure 1.

Laser speckle flowgraphy (LSFG) images of optic nerve head with associated waveforms before and after stenting. (A) Selection of several “rubber band” regions (1, 2, 3) of LSFG-NAVI^® images obtained prior to stenting. The elliptical region (STIN) over the optic nerve head is divided into four quadrants from which LSFG metrics can be derived. Within each quadrant, metrics can be further broken down by different tissue areas (Varea, Tarea, Aarea). (B) Pulse waveform of mean blur rate (MBR) in arbitrary units (AUs) extracted prior to stenting. (C) Selection of elliptical region over the optic nerve head after stenting. This region highlights a change in intensity of the blood flow map when compared to the pre-stenting intensity. (D) Pulse waveform of MBR extracted after stenting. The waveforms display different periods and amplitudes from those derived prior to stenting.

Mean blur rate (MBR), an index of ocular blood velocity, was extracted by the software and averaged across 120 video image frames collected over four seconds. ¹² Minimum and maximum MBR, as well as MBR range, were calculated. Additionally, parameters serving as indices of ocular blood acceleration were computationally derived from MBR using pulse waveform analysis. These parameters included flow acceleration index (FAI) and rising rate (RR).

Iflow measurements

iFlow allows for quantification of cerebral perfusion by translating DSA images into perfusion heatmaps. ¹³ iFlow images display the intensity of the contrast agent in the cerebral vasculature changing over time. ¹⁴ Two investigators (MJ and RP) analyzed DSAs using iFlow. Seven regions of interest (ROIs) were sampled. Two ROIs included pre- and post-ICA stenosis (Figure 2). The other five ROIs consisted of the junction of the high cervical and petrous segments of the ICA (hcICA), the distal cavernous segment of the ICA (cavICA), the ophthalmic artery (OPHT), the retinal blush (RB), and frontal lobe parenchyma (PAR). iFlow quantified the time to peak (TTP), which is the time from the injection of contrast to the time at which the contrast intensity of the selected ROI reached a maximum value. ¹⁵ Contrast delay was defined as the difference between the TTPs extracted from the ROIs proximal and distal to the stenosis or stent. A ratio of TTP between the cavICA and hcICA ROIs (TTP cavICA/TTP hcICA) was obtained for normalization and to account for the variability in contrast injection rate. TTPs obtained from each rater were averaged for further analysis.

Figure 2.

Syngo iFlow (iFlow) representation of digital subtraction angiography (DSA) images. (A) DSA image of carotid artery stenosis captured prior to stenting. An area of stenosis in the proximal internal carotid artery (ICA) is noted (black arrow). (B) iFlow image of contrast intensity captured prior to stenting. The stenosis is observed as a region where a change of flow is seen in the heatmap (yellow arrow). Two markers captured time to peak (TTP), one proximal to the region of stenosis and one distal to the region of stenosis (yellow rectangles). Contrast delay was calculated from the difference in these TTPs. (C) DSA image of carotid artery captured after stenting (black arrowhead). (D) iFlow images of contrast intensity captured after stenting. Contrast intensity in the location of the stent is noted (yellow arrowhead). Markers placed are analogous to those placed prior to stenting (yellow rectangles). (E) iFlow image of cerebral perfusion. The region distal to carotid stenosis is displayed with regions of interest (ROIs) noted (white rectangles and circles). ROIs displayed correspond to brain/ocular perfusion ROIs (1 = hcICA, 2 = cavICA, 3 = OPHT, 4 = RB, 5 = PAR). TTP was extracted for all ROIs.

Collateral flow from the circle of Willis or branches of the external carotid artery was adjudicated by an experienced investigator. The investigator was blinded of any clinical information, and, after reviewing all the vascular imaging available for each patient (DSA, CTA, and/or MRA), determined the presence of collateral flow in the ipsilateral vascular territory of ICA stenosis. It was considered that collateral flow was present in the distal ICA if there was contrast opacification of the anterior communicating artery or ipsilateral posterior communicating artery on DSA, CTA, and/or MRA. Collateral flow through the branches of the external carotid artery was also identified in DSA.

Statistical analysis

Statistical analysis was performed using R statistical software (v.4.2.0). Intraclass correlation coefficients (ICCs) for the two raters’ TTPs extracted from iFlow were calculated. The coefficients were assessed according to a scale of slight (less than 0.2), fair (0.2 to 0.4), moderate (0.4 to 0.6), substantial (0.6 to 0.8), and almost perfect (greater than 0.8 to 1.00). Based on a Shapiro-Wilk test of normality, either Pearson or Spearman correlation coefficients were calculated to analyze correlation between iFlow parameters, LSFG metrics, and NASCET degree of stenosis. Pre- and post-stenting iFlow metrics were compared using either a two-sided paired t-test or a two-sided paired Wilcoxon signed-rank test. Potential differences in LSFG metrics between baseline and follow-up scans were also assessed. Differences were assessed for LSFG metrics extracted from the eye ipsilateral to the treated carotid artery stenosis. The choice of a paired t-test or paired Wilcoxon signed-rank test was based on a Shapiro-Wilk test of normality in differences among matched pairs of metrics. Plots were also created for LSFG metrics with associated descriptive statistics (means). Statistical tests were conducted with an alpha of 0.05.

Results

Patient characteristics

Eighteen of twenty eligible patients were prospectively enrolled in the study (Figure 3). The median age was 69.5 years old (IQR, 61–76.5), 61 percent (n = 11) were female, and 72 percent (n = 13) presented with stroke (Supplementary Table 1). Fourteen patients underwent successful stenting. Collateral flow was present in 11 of the 14 patients who underwent stenting. Of these 11 patients, collateral flow was present through the anterior communicating artery (n = 4), ophthalmic artery (n = 6), and posterior communicating artery (n = 1). LSFG was performed in 18 patients at baseline and five after stenting. Thirteen patients were discharged before a second LSFG could be obtained or refused it. In one patient, LSFG data were excluded due to poor imaging quality from cataract. iFlow data were obtained in 18 patients at baseline and 14 after stenting.

Figure 3.

Progression of patients through study workflow. Diagrammed workflow of study is depicted. Eligible stenosis was defined as NASCET degree of stenosis of 70 percent or greater, or symptomatic and NASCET degree of stenosis of 50 percent or greater. Laser speckle flowgraphy (LSFG) and syngo iFlow (iFlow) data were collected both before and after stenting. LSFG data were excluded from one patient due to cataract.

Table 1.

Correlations of laser speckle flowgraphy (LSFG) metrics with North American symptomatic carotid endarterectomy trial (NASCET) degree of stenosis and comparison of LSFG metrics pre- and post-stenting.

LSFG Metric	Correlation With NASCET At Baseline	Pre- and Post-Stenting Comparison (N = 5) a
	Rho b	Pre-Stenting c	Post-Stenting c	Difference d	P-value e
MBR (inferior: Aarea)	0.51	13.5 (± 6.1)	16.0 (± 8.9)	2.54 (18.52%)	0.3
Minimum MBR (inferior: Aarea)	0.52	10.0 (± 4.6)	10.4 (± 6.5)	0.4 (4.0%)	0.7
Maximum MBR (inferior: Aarea)	0.49	16.0 (± 7.0)	22.0 (± 12.0)	5.28 (37.5%)	0.2
MBR Range (temporal: Varea)	−0.59	6.2 (± 1.2)	11.3 (± 5.6)	5.1 (82.3%)	0.11
FAI (inferior: Varea)	−0.51	2.30 (± 0.71)	3.58 (± 1.18)	1.28 (55.65%)	0.032**
RR (superior: Tarea)	−0.56	10.38 (± 1.47)	12.54 (± 0.90)	2.16 (20.81%)	0.031**

^aN = 4 for MBR range (temporal: Varea) due to a missing value in pre-stenting data.

^bSpearman correlation coefficient, p < 0.05

^cMean (standard deviation).

^dMean difference (percent change from pre-stenting to post-stenting).

^ePaired t-test.

MBR, mean blur rate; FAI, flow acceleration index; RR, rising rate; Aarea, all-tissue area; Varea, vessel area with correction for extravascular influence; Tarea, tissue area; NASCET, North American Symptomatic Carotid Endarterectomy Trial.

**p < 0.05.

LSFG

Moderate correlations with NASCET degree of stenosis were observed for ocular blood acceleration metrics (FAI and RR) and ocular blood velocity metrics (MBR) among all patients with baseline LSFG scans (n = 17, Table 1). In patients who underwent stenting (n = 5), intravascular FAI improved when compared to baseline (mean difference of 1.28, percent change = 55.65%, p = 0.032). Tissue RR also improved (mean difference of 2.16, percent change = 20.81%, p = 0.031). In the analysis of patients with collateral flow (n = 4, Supplementary Figure 1), FAI and RR also improved after stenting. Patients with ophthalmic collateral flow who underwent stenting (n = 2) versus the patient without any collateral flow (n = 1) were analyzed. FAI and RR metrics also improved in this subset of patients. MBR metrics decreased after stenting in patients with ophthalmic collateral flow. However, in the patient without collateral flow, two of these same MBR metrics increased after stenting (Supplementary Figure 2).

Iflow

The ICC for TTP measurements was almost perfect (ICC = 0.94, p < 0.001, Supplemental Table 3). Correlations of iFlow metrics with LSFG and NASCET degree of stenosis showed moderate negative correlation of contrast delay with intravascular FAI (inferior quadrant) before stenting (rho = -0.61, p = 0.021). No iFlow metrics were significantly correlated with NASCET degree of stenosis.

iFlow at baseline and after stenting (n = 14, Supplemental Table 4) showed that contrast delay decreased after stenting compared to baseline (mean difference = -0.37, percent change = -50%, p = 0.004). TTPs decreased post-stenting in three regions: the junction of the high cervical and petrous ICA segments (mean difference = -0.54, percent change = -14.21%, p = 0.035), the cavernous ICA segment (mean difference = -0.69, percent change = -16.55%, p = 0.024), and frontal lobe parenchyma (mean difference = -1, percent change = -16.31%, p = 0.044).

Discussion

NASCET criteria for assessing the severity of luminal carotid stenosis is the gold standard in clinical practice.^3,16,17 However, the severity of stenosis can be overestimated, and measurements involved in calculations are subjective.^3,18 Additionally, degree of stenosis may not correlate with downstream distal perfusion, as collateral flow can compensate for proximal flow-limiting stenosis. ⁶ This may explain why iFlow metrics were not significantly correlated with NASCET degree of stenosis in our cohort. The quantification of ocular perfusion through LSFG could potentially provide indirect data on the status of distal ICA flow to end-organs. In this study, we described the status of ocular perfusion in symptomatic patients with carotid stenosis assessed using NASCET criteria.

MBR is an index of ocular blood velocity and represents the relative velocity index of erythrocytes. ¹⁹ MBR is the most-reported LSFG metric and has a very high reproducibility. ²⁰ FAI and RR are ocular blood flow acceleration indices. FAI measures maximum acceleration at a time when ocular blood velocity is increasing to its peak. ¹² RR measures how suddenly ocular blood velocity increases before the blood reaches peak velocity. ¹² FAI and RR had a negative correlation with NASCET degree of stenosis, corroborating clinical observations that more severe stenosis leads to decreased ocular perfusion. The effectiveness of carotid stenting in improving distal ICA flow was also corroborated when FAI and RR improved after stenting. The patient cohort was too small to identify significant differences in ocular perfusion metrics when collateral flow was present. A study by Shinohara et al. demonstrated that in seven patients with ocular ischemic symptoms and ipsilateral ICA occlusion, MBR in the affected eye was compromised and correlated with decreased choroidal blood flow. ²¹ Ismail et al. have also demonstrated in rabbits that obstruction of the ICA leads to significant variations in MBR. ²²

LSFG metrics can be influenced by systemic factors, like age or gender, and ocular circulatory factors, such as optic disc area. ²³ In the study mentioned earlier by Ismail et al., the abrupt ICA occlusion led to a transient increase of MBR, which later decreased. ²² FAI has also been correlated with systemic variables, such as blood pressure, and quantitative indices of ocular circulation. ¹² Previous studies have correlated RR with systemic circulation parameters, such as heart rate and left ventricular ejection fraction.^12,24,25 Hence, MBR, FAI, and RR provide an insight into systemic, regional, and local parameters that affect ocular perfusion. Since these metrics are obtained at the end-organ, collateral flow to the distal ICA is also reflected in these parameters.

iFlow was used to objectively quantify distal blood flow before and after stenting. However, iFlow metrics did not correlate with NASCET degree of stenosis. This was most likely because of changes in distal flow due to the presence of collaterals from the external carotid artery and the circle of Willis. A significant correlation was observed between FAI and contrast delay quantified with iFlow. LSFG metrics may correlate with cerebral blood flow in patients who do not have flow-limiting stenosis beyond the ophthalmic artery origin. This could be explored in future studies. In this cohort, no patients had tandem stenosis beyond the ophthalmic artery origin.

Our exploratory pilot study is limited by various factors. The main limitation of this study was the relatively small sample size. However, this was an exploratory study aimed at determining the feasibility of LSFG in patients undergoing carotid stenting. In addition, more research is needed to assess effects of systemic and local confounders on LSFG metrics. Anatomical variants of vasculature supplying the eye, such as contribution from the external carotid artery, may influence end-organ measurements of blood flow. Further research might also address the relative significance of specific ocular regions (retina, choroid, and optic nerve head) in assessing end-organ perfusion in patients with proximal carotid stenosis.

Conclusions

The results of this pilot study suggest that LSFG metrics can quantify end-organ ocular perfusion in patients with proximal carotid stenosis. This information provides additional clinical information on severity of stenosis. MBR, FAI, and RR had significant correlation with NASCET degree of stenosis; FAI and RR also improved after carotid stenting.

Abbreviation List

ICA

Internal carotid artery

DSA

Digital subtraction angiography

NASCET

North American Symptomatic Carotid Endarterectomy Trial

ICC

Intraclass correlation coefficient

CI

Confidence interval

CTA

Computed tomography angiography

MRA

Magnetic resonance angiography

iFlow

syngo iFlow

ROI

Region of interest

TTP

Time to peak

hcICA

Junction of high cervical and petrous segments of ICA

cavICA

Cavernous segment of ICA

OPHT

Ophthalmic artery

RB

Retinal blush

PAR

Frontal lobe parenchyma

LSFG

Laser speckle flowgraphy

MBR

Mean blur rate

AU

Arbitrary unit

FAI

Flow acceleration index

RR

Rising rate

STIN

Area of elliptical region placed by investigators analyzing LSFG images

Varea

Vessel area with correction for extravascular influence

Tarea

Tissue area

Aarea

All-area