High-resolution image-guided WEB aneurysm embolization by high-frequency optical coherence tomography

Abstract

Background

High-frequency optical coherence tomography (HF-OCT) is an intra-vascular imaging technique capable of assessing device-vessel interactions at spatial resolution approaching 10 μm. We tested the hypothesis that adequately deployed Woven EndoBridge (WEB) devices as visualized by HF-OCT lead to higher aneurysm occlusion rates.

Methods

In a leporine model, elastase-induced aneurysms (n=24) were treated with the WEB device. HF-OCT and digital subtraction angiography (DSA) were performed following WEB deployment and repeated at 4, 8, and 12 weeks. Protrusion (0-present, 1-absent) and malapposition (0-malapposed, 1-neck apposition >50%) were binary coded. A device was considered ‘adequately deployed’ by HF-OCT and DSA if apposed and non-protruding. Aneurysm healing on DSA was reported using the 4-point WEB occlusion score: A or B grades were considered positive outcome. Neointimal coverage was quantified on HF-OCT images at 12 weeks and compared with scanning electron microscopy (SEM).

Results

Adequate deployment on HF-OCT correlated with positive outcome (P=0.007), but no statistically significant relationship was found between good outcome and adequate deployment on DSA (P=0.289). Absence of protrusion on HF-OCT correlated with a positive outcome (P=0.006); however, malapposition alone had no significant relationship (P=0.19). HF-OCT showed a strong correlation with SEM for the assessment of areas of neointimal tissue (R²=0.96; P<0.001). More neointimal coverage of 78%±32% was found on ‘adequate deployment’ cases versus 31%±24% for the ‘inadequate deployment’ cases (P=0.001).

Conclusion

HF-OCT visualizes features that can determine adequate device deployment to prognosticate early aneurysm occlusion following WEB implantation and can be used to longitudinally monitor aneurysm healing progression.

INTRODUCTION

Endovascular techniques have been widely used and accepted for aneurysm treatment.¹ Excluding the aneurysm sac from circulation and protecting the parent vessel are the main targets. There are two main categories of endovascular aneurysm treatment: intrasaccular and endoluminal procedures. The Woven EndoBridge (WEB, Microvention, Aliso Viejo, CA) device is a new-generation intrasaccular aneurysm embolization device that enables a stable flow-disruption and thrombosis within the aneurysm with a high-attenuation metallic mesh across the aneurysm neck promoting neointimal proliferation and exclusion of the aneurysm.² The success of these devices has changed the understanding of aneurysm healing and has refocused attention to the aneurysm neck.³ One of the most significant advantages of these devices is elimination of the need for post-procedural dual antiplatelet therapy, which may be associated with complications.⁴

Optical coherence tomography (OCT) is a catheter-based imaging technique that is increasingly being used in interventional cardiology.⁵ The use of OCT has been proposed for the diagnosis and management of cerebrovascular pathologies,⁶ and recent studies have shown the benefits of imaging for cerebrovascular aneurysms.^7–9 Using near-infrared light, OCT acquires three-dimensional information and captures remarkable details about the vessel wall characteristics, the presence of intraluminal thrombus,¹⁰ and the progression of post-treatment aneurysm healing.¹¹ OCT provides highly detailed procedural guidance for an ideal device deployment that can potentially increase the success rates of achieving complete, early aneurysm occlusion.^8,12,13 However, OCT imaging devices designed for use in coronary arteries, are unsuitable for routine delivery and imaging through highly tortuous intracranial anatomy.⁶

High-frequency OCT (HF-OCT) is a new generation of OCT technology developed specifically for neurovascular use in tortuous intracranial anatomy.¹⁴ In this study, we hypothesized that periprocedural assessment of device apposition and lack of protrusion by HF-OCT is predictive of early aneurysm occlusion, and that device characteristics assessed by HF-OCT could be an important predictor of aneurysm healing. Furthermore, HF-OCT offers unprecedented visualization of neointimal tissue growth over the device at the level of the aneurysm neck, which provides crucial information for patient management and may allow customized therapies in the near future.¹⁵

MATERIALS AND METHODS

Experimental procedures

All animal research procedures were approved by our university’s Institutional Animal Care and Use Committee. Elastase-induced aneurysms were created in a New Zealand White Rabbit (NZWR) model (n=24) (sex: male; weight, 3.8–4.16kg).¹⁶ Aneurysms were allowed to mature for a minimum of 3 weeks before WEB device placement.

All procedures were performed under general anesthesia. All animals were pre-anesthetized by glycopyrrolate (0.01 mg/kg) and given buprenorphine sustained release (0.15 mg/kg) for pain management before the procedures. Anesthesia was induced by intramuscular injection of ketamine (35 mg/kg) and xylazine (5 mg/kg) and maintained with mechanical ventilation of 1%–3% isoflurane. The heart rate, respiration rate, oxygen saturation level, end-tidal CO₂ level, and rectal temperature were continuously monitored throughout the procedure. Intramuscular atipamezole (0.35 mg/kg) was used for anesthesia reversal. No antiplatelet therapy was given before or after the procedure.

Once anesthetized, a 6F introducer sheath was inserted over a guidewire through a right transfemoral approach. Each rabbit was given 100IU/kg heparin following arterial access. A Navien-072 guide catheter was positioned in the brachiocephalic trunk. DSA was acquired for the measurements of the aneurysm to determine the required device size. The WEB devices (SL W4–5-4, n=14; SL W4–6-5, n=7; and SLS W4–6-S, n=2) were deployed under fluoroscopic guidance. Following manufacturer recommendation, the WEB was slightly oversized to ensure a stable position inside the aneurysm. Following deployment, HF-OCT imaging was performed using an intravascular catheter (Vis-M, Gentuity LLC, Sudbury MA) compatible with an 0.017” microcatheter.¹⁴ To clear arterial blood during HF-OCT data acquisition, iodinated contrast agent was injected via a power injector at 4 mL/s for 4s through the guide catheter positioned at the origin of the innominate artery. DSA data were subsequently acquired. DSA and HF-OCT imaging were repeated on half of the animals at 4 weeks (n=12), and the other half were imaged at 8 weeks (n=12): all animals (n=24) were imaged at 12 weeks following implantation, to assess healing of the aneurysm and analyze conformational changes of the device. At 12 weeks, an overdose of sodium pentobarbital (150 mg/kg) was used for euthanasia. Perfusion under physiological pressures was obtained with saline followed by 4% paraformaldehyde in distilled water. The right subclavian artery was resected en bloc and placed over an overnight immersion fixation in 2.5% glutaraldehyde and then rinsed in 0.1 M cacodylate buffer. The longitudinal sections were processed under a dissecting microscope to expose the neck of the aneurysm, and then dehydrated through a graded series of ethanol concentrations (up to 100%) followed by critical point drying in carbon dioxide. The samples were mounted onto aluminum stubs, grounded with silver conductive paste, sputter coated with gold/ palladium, and imaged with scanning electron microscopy (SEM) (Quanta 200 MKII FEG; FEI, Hillsboro, OR).

Image analysis

Aneurysm healing was assessed at baseline on DSA images following implantation, and at 4, 8, and 12 weeks using the 4-point Web Occlusion Scale (WOS, Grade A: Complete aneurysm occlusion, Grade B: Complete occlusion with recess filling, Grade C: Aneurysm neck remnant, Grade D: Aneurysm remnant).^17,18 At 12 weeks, a WOS grade of A or B was considered as positive outcome.

HF-OCT data sets were analyzed using ImageJ (NIH). A device is considered ‘protruding’ when struts herniated from the aneurysm neck into the parent artery (online supplementary figure 1). The extent of the protrusion is measured on cross-sectional HF-OCT slices. The HF-OCT endoscopic imaging probe has an imaging field of view of approximately 14 mm, allowing visualization within the aneurysm. However, the device struts attenuate the optical signal resulting in shadows beyond the struts. Interrogation of the tissue is possible between the struts, where the cells serve as an optical window. The wall-apposition status on HF-OCT was evaluated according to a dichotomous outcome, noted as either malapposed or well-apposed (ie, neck apposition >50%) depending on the presence or absence of a gap between the device struts and aneurysm neck (online supplementary figure 1). A binary code was used to assess protrusion (0-present, 1-absent) of the device struts (not just the detachment zone), and malapposition (0-malapposed, 1-neck apposition >50%). A device was considered adequately deployed if it received a score of 1 on both metrics. The same criteria were used for the post-implant assessment of device implantation on DSA images. Three-dimensional HF-OCT renderings were generated using software Osirix (Pixmeo SARL, Geneva, Switzerland).

Twelve-week HF-OCT images were examined for the presence of neointimal tissue covering the device struts. HF-OCT data sets were analyzed slice by slice using ImageJ (NIH), and the areas of the neointimal tissue and uncovered device struts were manually traced. Multispectral SEM images (both backscatter and secondary electron imaging) were combined for clear delineation of uncovered struts. The images manually segmented with ImageJ. Specifically, neointimal tissue and uncovered strut areas were drawn by a manual tracing method on SEM images acquired for displaying aneurysm ostium. The neointimal tissue coverage was calculated by dividing the neointimal tissue area by total area representing the aneurysm neck. (online supplementary figure 2). Twelve-week HF-OCT, DSA, and SEM images were examined for the partial or complete intra WEB thrombosis.

Statistical analyses were performed by using SPSS 22 (IBM Corporation, Armonk NY). Comparisons between groups for continuous variables were evaluated by using t-tests and the Fisher exact test or Pearson χ2 for binary categoric variables. Aneurysm occlusion rates at different time points were evaluated with the χ2 test. Fisher’s exact t-test was used to determine if HF-OCT or DSA ‘adequate deployment’ can predict the aneurysm occlusion assessed with 12 weeks’ DSA. The extent of protrusion was compared with a non-parametric Mann–Whitney U test. A P-value<0.05 was considered statistically significant.

RESULTS

This study included 24 NZWR aneurysms treated with an equal number of WEB devices, and no animals were excluded from the analysis or had any adverse events to the procedures. In vivo DSA and HF-OCT data sets were collected after device deployment and imaging was repeated at 4, 8, and 12 weeks. The mean neck diameter was 2.7±0.74 mm, aneurysm height was 8.2±1.74 mm, and aneurysm width was 3.5±0.54 mm. An example of three-dimensional HF-OCT imaging of a WEB device is illustrated in figure 1.

Figure 1.

DSA (A) and corresponding three-dimensional HF-OCT imaging of a WEB device at 12-week follow-up (B). The inset shows the same device at a different angulation from the same HF-OCT acquisition. Corresponding cross-sectional HF-OCT images (C, D). Lack of coverage is indicated in red color (C), red star; (D), red arrow): a thin layer of coverage is observed on both the three-dimensional HF-OCT and cross-sectional images (D), white arrow). HF-OCT has the ability to visualize the device apposition in vivo, as well as detailed interaction with the parent artery and the aneurysm neck, and neointimal tissue coverage progression. Scale bars are equal to 1 mm.

After deployment, all cases were rated as WOS grade D based on initial DSA images. Half of the cases (n=12) were imaged at 4 weeks, and grade A, C, and D occlusions were found in two (17%), four (33%), and six (50%) cases, respectively. The remaining animals (n=12) were imaged with DSA at 8 weeks, and grade A, B, C, and D occlusions were found in one (8%), two (17%), four (33%), and five (42%) cases, respectively. All cases (n=24) were imaged at 12 weeks following implantation, and grade A, B, C, and D occlusions were found in three (12.5%), two (8.5%), eight (33%), and 11 (46%) cases, respectively (table 1). Angiographically successful (grade A & B) occlusion rates at 4, 8, and 12 weeks were not statistically different at 17% (n=2), 25% (n=3), and 21% (n=5), respectively (χ2 0.257, P=0.881).

Table 1.

Aneurysm occlusion rates at different time points quantified by DSA

Web Occlusion Scale	Grade A	Grade B	Grade C	Grade D
Post-implantation (n=24)	0	0	0	24 (100%)
Four-weeks (n=12)	2 (17%)	0	4 (33%)	6 (50%)
Eight-weeks (n=12)	1 (8%)	2 (17%)	5 (42%)	4 (33%)
Twelve-weeks (n=24)	3 (12.5%)	2 (8.5%)	8 (33%)	11 (46%)

The mean neck size was 2.18±0.25 mm on the five cases with angiography success, and 2.83±0.79 mm for the aneurysms that failed to occlude based on initial DSA measurements. There was no statistically significant difference in neck sizes across the groups (P=0.085). The mean aneurysm height was 9.6±2.6 mm for successful outcome group and 7.8±1.3 for unsuccessful outcome group (P=0.20).

Device apposition and protrusion were evaluated on baseline DSA images. A total of 33% of cases (n=8) were classified as ‘adequately deployed’. No statistically significant relationship was found between positive outcome and adequate deployment based on DSA images (P=0.29).

Similarly, wall apposition and protrusion status were evaluated on initial HF-OCT images. A total of 30% of cases (n=8) had malapposition, and 58% of cases (n=14) had protrusion. Five (5) aneurysms (21%) treated with the WEB were classified as adequately deployed based on initial HF-OCT images. Fisher’s exact test showed an association between adequate deployment as observed on HF-OCT and successful aneurysm occlusion after 12 weeks (P=0.008). Based on HF-OCT images, the absence of protrusion was related to the positive outcome (P=0.007), however, malapposition alone had no significant correlation with positive outcome (P=0.19). The maximal extension of protrusion was measured on HF-OCT images, and the negative outcome group was found to have a greater extension than the positive outcome group (0.345 mm IQR=0.483 mm vs 0.0 mm IQR=0 mm respectively; P=0.008, Mann–Whitney U test).

Discrepancies between HF-OCT and DSA device assessments were found. HF-OCT classified as adequately deployed a total of five cases, whereas eight cases were classified as adequately deployed by DSA images. Acceptable WOS grade at 12 weeks was seen in 21% of cases (n=5), and 80% of those cases (n=4) were classified as adequately deployed by HF-OCT: however, only 60% of those cases (n=3) were classified as adequate by DSA.

The sensitivity and specificity for adequate deployment evaluated on baseline HF-OCT were 66.6% (95% CI, 22.8 to 95.7), and 94.4% (95% CI, 71.7 to 99.8), respectively, for the prediction of negative vs positive angiographic outcome at 12 weeks’ DSA. The overall accuracy for adequate deployment evaluation on post-treatment HF-OCT was as high as 87.5% (95% CI, 67.6 to 97.34) for the prediction of positive vs negative outcome at 12 weeks (online supplementary table 1).

Neointimal tissue assessment by SEM and HF-OCT at 12 weeks were compared. Areas of neointimal tissue quantified by HF-OCT showed a strong correlation with SEM assessment (R²=0.96; P<0.0001) (figure 2). An example of HF-OCT visualization of a small area of incomplete endothelization qualitatively compared with SEM is shown in figure 3. Additionally, we found a mean neointimal coverage of 78%±32% for adequate deployment cases (HF-OCT assessment), whereas a mean neointimal coverage of 31%±24% was found for inadequate deployment cases (P=0.002) (figure 4). Furthermore, in a case where DSA image shows contrast filling into the aneurysm neck, HF-OCT identifies neointimal coverage and partial healing of the aneurysm (figure 5).

Figure 2.

Scatter diagram of HF-OCT and SEM measurements of neointimal coverage (%). The linear model accounts for most of the variation in the data (R²=0.96). Solid line is the result of linear regression modeling (slope is 1.0 and intercept −0.05) and the dotted lines represent the 95% CI.

Figure 3.

SEM showing device neointimal coverage including uncovered struts on one side of the device (A), (C), (red arrows). Corresponding in vivo HF-OCT data collected at 12 week follow-up show tissue growth over the device (B) (white arrow), as well as areas of uncovered struts, strongly correlating with SEM findings (D), (red arrow). Scale bars on HF-OCT images are equal to 1 mm.

Figure 4.

Examples of a ‘not adequately deployed’ device based on HF-OCT assessment (A–D). Baseline HF-OCT (A), and 12 week terminal DSA with negative outcome (WOS-C), SEM, and HF-OCT (C–D, respectively) for a case showing device protrusion and absence of malapposition (white arrowin A). A large area of uncovered device struts was identified on both SEM (C) and HF-OCT images (red arrow inD). Example of ‘adequate deployment’ (E–H). Baseline HF-OCT (E), and 12 week DSA (F), SEM image (G), and 12 weeksHF-OCT (H), for a case without protrusion or malapposition. A positive outcome (WOS-A) and complete neointimal coverage is observed on SEM (G). Similarly, the device struts are completely covered by neointimal tissue as observed by HF-OCT (white arrowin H). Scale bars on HF-OCT images are equal to 1 mm.

Figure 5.

Scanning electron microscopy (SEM) image shows an area of uncovered struts at the distal aneurysm location (A, red arrow). DSA image at 12 weeks shows limited contrast filling at the proximal neck area that may be misdiagnosed as neck remnant (B, red arrow). In agreement with SEM, in vivo HF-OCT at 12 weeks shows neointimal coverage over the proximal neck location (C, white arrow), and a cluster of uncovered struts over the distal portion of the aneurysm (D,red arrow). Scale bars on HF-OCT images are equal to 1 mm.

Intra WEB thrombosis assessment at 12 weeks on DSA, OCT, and SEM images demonstrated five cases of total thrombosis on DSA, three cases on HF-OCT, and two cases on SEM images. Fisher’s exact test showed statistically significant association between DSA and HF-OCT, DSA and SEM, HF-OCT and SEM assessment for intra WEB thrombosis (P=0.005, P=0.04, P=0.01)

At 12 weeks, mean neointimal coverage as assessed by SEM was 55%±35% for eight apposed cases vs 34%±27% for 16 malapposed cases (P=0.045, t-test). Based on HF-OCT protrusion assessment at 12 weeks, mean neointimal coverage as assessed by SEM for 15 protruded cases was 23%±18% and 70%±28% for the nine cases where protrusion was absent (P<0.001, t-test). As expected, mean neointimal coverage for the positive outcome group was higher than the negative outcome group: 88%±11% and 29%±22% respectively (P<0.001, t-test). Case level data are presented in online supplementary table 2.

DISCUSSION

The WEB device has been shown to be a safe and effective treatment for intracranial aneurysms. However, despite an excellent safety profile, there is a known risk of recanalization, and potential thromboembolic events in the case of protrusion, with a potential need of retreatment.¹⁹ In this study, we have demonstrated that HF-OCT is an excellent tool to assess WEB deployment. It allows for a precise evaluation of the device struts and its relationship to the parent vessel and the aneurysm neck. Combined with the WEB retrievable design, real-time in vivo assessment of these features at high resolution enables informed procedural guidance for satisfactory device deployment.

Endosaccular flow disrupters are a relatively new class of devices designed for the treatment of aneurysms. The biomechanics and radiologic appearance of aneurysms treated with endosaccular devices differs from the ones treated with other traditional methods, such as coiling. DSA is widely used for the evaluation of treated aneurysms, and the Raymond Roy,²⁰ the O’Kelly–Marotta,²¹ and the WOS^17,18 grading scales are widely used in clinical trials for the assessment of treated aneurysms. However, due to a lower resolution and its two-dimensional projection nature, DSA may not be always able to reliably characterize aneurysm occlusion. On the contrary, HF-OCT can precisely delineate the location of small areas of incomplete endothelization that can be used to provide procedural guidance at follow-up and re-treatment. In this study, a detailed image analysis showed cases where DSA indicated areas with contrast filling (figure 1), but HF-OCT identified neointimal coverage (ie, healing).

Such a mechanism of neo-endothelization could explain a lower occlusion rate in cases with device protrusion. In a preclinical in vivo model, it has been shown that the parent artery acts as the main source for neointimal cells.²² Over a flow-diverter surface, endothelial cells growth relies on an underlying smooth muscle cell substrate. For these type of devices, it has been hypothesized that neo-endothelialization plays a more significant role than thrombus formation in the aneurysm cavity for the complete occlusion of an aneurysm.²² With this study, we suggest that the migration and adherence of endothelial and smooth muscle cells over WEB devices at the level of the neck is affected by the protrusion of the device into the parent artery. In case of protrusion, the base part of the device is exposed to different hemodynamic factors such as high shear stress. Moreover, the radiographic appearance of the device protrusion from the aneurysm may lead to the continuation of dual antiplatelet therapy, thereby potentially increasing the risk of delayed aneurysm occlusion.²³ These factors might delay the formation of a durable thrombus that is followed by smooth muscle cell migration. In multiple cases analyzed in this study, a large thrombus formation or neointimal cells were not present on the protruded part of the devices. For the most severe protrusions, the neointimal tissue initially covering the periphery of the device failed to cover its most prominent portion, and instead showed a growth pattern that covered its inner surface (online supplementary figure 3). We can hypothesize that this is due to a delay in thrombus formation and smooth cell layer growth, inhibited by the direct exposure of the protruded part to the blood flow of the parent artery.

A previous study showed a high post-treatment mean aneurysm inflow was associated with the failure of aneurysm occlusion treated with WEB devices.²⁴ The study also suggested that the change in inflow rate inbetween pre- and post-treatment flow rate could predict treatment success. Our findings indicate that treatment failure is mostly related to poor device deployment that causes insufficient flow disruption and migration, and the attachment of neointima over the device struts. Proper device placement is the likely key for the maximal flow reduction and tissue remodeling. Proper placement allows the low-porosity regions such as the device base to overcome strong inflow jets. Being able to assess device deployment is valuable for evaluating the procedure and is potentially helpful for treatment guidance.

Although the apposition of flow diverters appears to be essential for early aneurysm occlusion in animal models,^8,25,26 our data suggested that neck apposition alone was not independently predictive of early occlusion with the WEB device. A combination of apposition and device protrusion were predictive of early occlusion in this aneurysm model, with protrusion into the parent vessel being the leading mechanism of delayed healing. These data may shed light and inform optimal techniques for the use of WEB for aneurysm treatment.

In addition, we have shown that assessment of neointimal tissue by HF-OCT strongly correlated with SEM findings. To the best of our knowledge, there have been no previous reports of neointimal tissue growth over implanted WEB devices assessed by HF-OCT imaging in vivo.

This study has some limitations. First, we used criteria for the classification of wall apposition and assessment of device protrusion that were not previously established. This was intentionally selected due to the different hemodynamic properties of intrasaccular flow disrupters. A second limitation is that only a single reader analyzed the electron microscopy samples. However, to prevent bias, this reader was blinded to both DSA and HF-OCT outcomes. Furthermore, this study focused on the analysis of protrusion and malapposition as key factors influencing aneurysm healing; however, some other criteria such as intra-aneurysmal thrombosis may play an important role. The study was conducted over a shorter follow-up period as compared with clinical practice without evidence of WEB compaction. However, this is mitigated using an animal model with a rapid healing response to endovascular devices. Finally, HF-OCT uses near-infrared light that, when compared with other non-invasive modalities, results in a limited tissue penetration ultimately unable to visualize the dome of treated aneurysms and the apex of intrasaccular devices.

CONCLUSION

HF-OCT is a useful diagnostic tool to image device protrusion and malapposition in aneurysms treated by the latest generation of intrasaccular flow disrupters. The absence of protrusion was associated with subsequent, early aneurysm occlusion; however, malapposition alone did not show an effect on the rate of aneurysm healing. HF-OCT offers an unprecedented visualization of the neointimal tissue covering the device struts. It allows for a precise assessment of the device relationship with the parent artery and the aneurysm neck. Results of this study indicate the potential value of HF-OCT for both procedural guidance as well as follow-up imaging to thoroughly understand healing mechanisms.

Data availability statement

Data presented in the manuscript.