Creation of a predictive calculator to determine adequacy of occlusion of the woven endobridge (WEB) device in intracranial aneurysms—A retrospective analysis of the WorldWide WEB Consortium database

Abstract

Background

Endovascular treatment with the woven endobridge (WEB) device has been widely utilized for managing intracranial aneurysms. However, predicting the probability of achieving adequate occlusion (Raymond–Roy classification 1 or 2) remains challenging.

Objective

Our study sought to develop and validate a predictive calculator for adequate occlusion using the WEB device via data from a large multi-institutional retrospective cohort.

Methods

We used data from the WorldWide WEB Consortium, encompassing 356 patients from 30 centers across North America, South America, and Europe. Bivariate and multivariate regression analyses were performed on a variety of demographic and clinical factors, from which predictive factors were selected. Calibration and validation were conducted, with variance inflation factor (VIF) parameters checked for collinearity.

Results

A total of 356 patients were included: 124 (34.8%) were male, 108 (30.3%) were elderly (≥65 years), and 118 (33.1%) were current smokers. Mean maximum aneurysm diameter was 7.09 mm (SD 2.71), with 112 (31.5%) having a daughter sac. In the multivariate regression, increasing aneurysm neck size (OR 0.706 [95% CI: 0.535–0.929], p = 0.13) and partial aneurysm thrombosis (OR 0.135 [95% CI: 0.024–0.681], p = 0.016) were found to be the only statistically significant variables associated with poorer likelihood of achieving occlusion. The predictive calculator shows a c-statistic of 0.744. Hosmer–Lemeshow goodness-of-fit test indicated a satisfactory model fit with a p-value of 0.431. The calculator is available at: https://neurodx.shinyapps.io/WEBDEVICE/.

Conclusion

The predictive calculator offers a substantial contribution to the clinical toolkit for estimating the likelihood of adequate intracranial aneurysm occlusion by WEB device embolization.

Introduction

The woven endobridge (WEB; Microvention/Terumo, Aliso Viejo, CA, USA), is specifically designed to address the challenges of treating wide neck bifurcation aneurysms (WNBAs), has revolutionized care in endovascular neurosurgery.^1–3 Although it has been in use in Europe since 2011, it was not until 2018 that the device received FDA approval in the USA.^1–3 Functioning as an intrasaccular flow disruptor, the WEB device is a self-expanding mesh sphere that initiates intrasaccular thrombosis, effectively isolating the aneurysm from the circulation. ⁴ This mechanism contributes to its minimal morbidity and low complication rates. ⁴

The WEB device has been approved for managing both ruptured and unruptured wide-necked aneurysms of the anterior communicating artery (Acomm), middle cerebral artery (MCA), internal carotid artery (ICA), and basilar artery bifurcations.^5–7 It has also proven its efficacy and safety in managing recurring and sidewall aneurysms in the ICA, particularly when the neck angle is suitable to proper device deployment.^5,6

The WEB device was the first widely adopted tool for intrasaccular flow disruption and has been extensively studied for the treatment of intracranial aneurysms. ⁴ A number of independent trials in Europe and the United States have provided evidence for its safe and effective application. These include the WEB clinical assessment of intrasaccular aneurysm therapy (WEBCAST) ⁸ and its subsequent WEBCAST-2 trial, ⁹ the French Observatory study, ¹⁰ and the WEB-intrasaccular therapy (WEB-IT) investigation. ¹¹ Additionally, the WorldWide WEB consortium has produced multiple studies further affirming the device's utility in the clinical setting.^7,12–16

As the number of aneurysms treated using endovascular techniques continues to rise, there is a growing demand for a reliable system to predict aneurysm occlusion clinically. ¹⁷ The modified Raymond–Roy classification is a widely accepted tool used to evaluate the extent of aneurysm occlusion after embolization. ¹⁸ This grading system, which is based on a three-point scale, provides an intuitive understanding of the completeness of endovascular occlusion: Grade 1 refers to complete occlusion; Grade 2 indicates a residual neck; and Grade 3 signifies persistent aneurysm flow. ¹⁸ Our study used the term “adequate” to express aneurysms achieving grade 1 or 2 Raymond–Roy.

In clinical practice, having reliable predictors of adequate occlusion is crucial to optimize treatment plans and patient counseling, highlighting the need for decision-support tools capable of providing accurate predictions based on comprehensive data sets. Accordingly, we aimed to create a predictive model using multi-institutional data to predict the last follow-up adequate (Raymond–Roy 1 or 2) occlusion of the WEB device. Our goal was not to replace but to augment the nuanced decision-making process inherent in patient care, thus developing a tool that fits seamlessly into the clinical workflow.

Methods

Patient selection

This study leveraged the WorldWide WEB Consortium's database, a collective repository from 30 academic institutions spread across North America, South America, and Europe. We conducted a retrospective examination of intracranial aneurysm cases managed with the WEB device between January 2011 and December 2022. Our study cohort consisted of consecutive adult patients, 18 years and older, presenting with both ruptured and unruptured saccular aneurysms treated with the WEB in any location. We excluded patients presenting with aneurysm morphologies unsuitable for WEB device placement, such as fusiform and blister aneurysms. Aneurysm appropriateness for WEB device treatment, based on size and neck angle, was independently assessed by interventionalists without any influence from the study protocol. Data collection involved a standardized datasheet encompassing patient demographics, aneurysm characteristics, procedural details, complications, angiographic results, and functional outcomes. Given that the study exclusively used anonymized patient data, informed consent was deemed unnecessary. All participating centers secured institutional review board approval.

Outcome evaluation

We evaluated angiographic outcomes through digital subtraction angiography, MR angiography, or CT angiography. The Raymond–Roy three-point occlusion scale served as the measure for categorizing post-treatment and final follow-up aneurysm occlusion into complete occlusion (RR1), neck remnant (RR2), and aneurysm remnant (RR3). Adequate occlusion was defined as complete occlusion or neck remnant without the presence of an aneurysm remnant. Functional outcome was assessed using mRS at the last follow-up. Independent functional status was defined with a mRS score of 0–2.

Thromboembolic complications occurring from the date of the procedure up to the last follow-up were recorded. Intraprocedural thromboembolic complications were identified on DSA as either thrombus formation, slow filling of a previously normal filling vessel, or complete vessel occlusion. Post-procedural thromboembolic complications were identified using a combination of clinical and radiographic findings. Post-procedural imaging was performed at the discretion of the individual institutions. Routine screening for clinically silent infarcts was not consistently performed. An ischemic complication was considered symptomatic if there were patient-reported symptoms or clinical signs attributable to thromboembolism; this included transient or resolving signs and symptoms. Hemorrhagic complications were identified intraoperatively as contrast extravasation on DSA or post-procedure imaging. Hemorrhages were counted as symptomatic if the patient-reported symptoms or demonstrated signs attributable to hemorrhage. Other complications included intraprocedural device deployment issues, air embolism, and vascular access complications. Complications were considered permanent if still present at a 3-month follow-up.

Statistical analyses

Statistical analyses were performed using R Studio version 4.2.3. Patient demographics and clinical characteristics were summarized using descriptive statistics. Continuous variables were presented as mean ± standard deviation (SD) and categorical variables as number (percentage).

Bivariate logistic regression was used to assess the association between each potential predictor and the outcome of achieving adequate occlusion following WEB device deployment. A multivariate logistic regression model was then used to further investigate the significance of the variables. These variables included age (elderly [≥65 years old] vs non-elderly [<65 years old]), sex, current smoking status, Hunt–Hess score, mRS score, bifurcation aneurysms, multiple aneurysms, maximum aneurysm diameter, aneurysm neck diameter, aneurysm height, presence of a daughter sac, partial intrasaccular thrombosis, WEB device shape, and use of an adjunctive device.

The adequacy of the model fit was evaluated using the Hosmer–Lemeshow goodness-of-fit test. Calibration and discrimination of the model were assessed with calibration plots and the c-statistic, respectively. The variance inflation factor (VIF) was calculated for each predictor in the model to check for multicollinearity. A p-value less than 0.05 was considered statistically significant.

Results

Patient demographics and characteristics

The study encompassed a total of 356 patients, with their demographics and clinical characteristics outlined in Table 1. Demographically, 124 (34.8%) were male, 108 (30.3%) were elderly (≥65 years old), and 118 (33.1%) were current smokers. The distribution of Hunt–Hess and mRS scores is provided in Table 1, illustrating the varying degrees of aneurysm severity and clinical outcomes. Most aneurysms (90.7%) were classified as bifurcation types, and 161 patients (45.2%) had multiple aneurysms. The mean maximum diameter of the aneurysms was 7.09 mm (SD: 2.71).

Table1.

Patient demographics and characteristics.

Variable		Overall (n = 356)	No adequate occlusion (n = 46)	Adequate occlusion (n = 310)
Male		124 (34.8%)	18 (39.1%)	106 (34.2%)
Elderly		108 (30.3%)	16 (34.8%)	92 (29.7%)
Current smoker		118 (33.1%)	15 (32.6%)	103 (33.2%)
Hunt–Hess score	0	183 (51.4%)	21 (45.7%)	162 (52.3%)
	1	88 (24.7%)	10 (21.7%)	78 (25.2%)
	2	28 (7.9%)	3 (6.5%)	25 (8.1%)
	3	34 (9.6%)	7 (15.2%)	27 (8.7%)
	4	15 (4.2%)	3 (6.5%)	12 (3.9%)
	5	8 (2.2%)	2 (4.3%)	6 (1.9%)
mRS score	0	225 (63.2%)	26 (56.5%)	199 (64.2%)
	1	59 (16.6%)	8 (17.4%)	51 (16.5%)
	2	16 (4.5%)	2 (4.3%)	14 (4.5%)
	3	20 (5.6%)	3 (6.5%)	17 (5.5%)
	4	13 (3.7%)	4 (8.7%)	9 (2.9%)
	5	13 (3.7%)	2 (4.3%)	11 (3.5%)
Bifurcation aneurysms		323 (90.7%)	41 (89.1%)	282 (91.0%)
Multiple aneurysms		161 (45.2%)	16 (34.8%)	145 (46.8%)
Mean maximum diameter (SD)		7.09 (2.71)	8.33 (3.89)	6.91 (2.44)
Presence of daughter sac		112 (31.5%)	17 (37.0%)	95 (30.6%)
Use of adjunctive device		29 (8.1%)	6 (13.0%)	23 (7.4%)

Length of follow-up

Overall mean follow-up was 16.09 [IQR: 5.80–20.40]. Mean follow-up for non-occluded aneurysms were 12.94 [IQR: 6.00–17.40]. Mean follow-up for occluded aneurysms were 16.56 [IQR: 5.65–22.00].

Bivariate regression analyses

In bivariate analysis, modified Rankin scale (MRS) score of 4 was associated with less likelihood of occlusion when compared to MRS score of 0 (OR 0.825 [95% CI: 0797–1.162], p = 0.047) (Table 2). Increasing maximum diameter (OR 0.978 [95% CI: 0.966–0.991], p < 0.001), increasing height (OR 0.977 [95% CI: 0.963–0.990], p < 0.001), increasing neck size (OR 0.949 [95% CI: 0.926–0.972], p < 0.001) and partial aneurysm thrombosis (OR 0.616 [95% CI: 0.502–0.757], p < 0.001) was associated with decreased likelihood of occlusion.

Table 2.

Bivariate regression of predictive variables for adequate occlusion.

Variable		OR	2.5%	97.5%	p
Male		0.976	0.907	1.050	0.513
Elderly		0.973	0.902	1.050	0.484
Hunt–Hess score		0.977	0.951	1.003	0.085
mRS score	0	Reference	Reference	Reference	Reference
	1	0.980	0.890	1.080	0.685
	2	0.991	0.835	1.175	0.914
	3	0.966	0.828	1.127	0.662
	4	0.825	0.683	0.996	0.047
	5	0.962	0.797	1.162	0.691
Bifurcation aneurysms		1.025	0.909	1.156	0.689
Multiple aneurysms		1.056	0.985	1.133	0.128
Increasing maximum diameter		0.978	0.966	0.991	<0.001
Increasing maximum height		0.977	0.963	0.990	<0.001
Presence of daughter sac		0.968	0.898	1.043	0.391
Use of adjunctive device		0.919	0.809	1.044	0.194
Neck size		0.949	0.926	0.972	<0.001
Partially thrombosed		0.616	0.502	0.757	<0.001
WEB SL (compared to WEB SLS)		0.938	8.857	1.027	0.165

Bold values indicate significance where p < 0.05.

Multivariate regression analyses

The multivariate analysis identified increasing aneurysm neck size (OR 0.706 [95% CI: 0.535–0.929], p = 0.13) and partial aneurysm thrombosis (OR 0.135 [95% CI: 0.024–0.681], p = 0.016) as being the only statistically significant variables associated with poorer likelihood of achieving occlusion (Table 3).

Table 3.

Multivariate regression of predictive variables for adequate occlusion.

Variable		OR	2.5%	97.5%	p
Male		0.852	0.409	1.820	0.673
Elderly		0.653	0.302	1.430	0.280
Hunt–Hess score		0.806	0.605	1.086	0.145
mRS score	0	Reference	Reference	Reference	Reference
	1	0.716	0.298	1.867	0.470
	2	1.947	0.384	18.316	0.484
	3	0.923	0.233	5.178	0.916
	4	0.358	0.090	1.603	0.153
	5	0.927	0.157	7.964	0.938
Bifurcation aneurysms		0.874	0.216	2.744	0.832
Multiple aneurysms		1.572	0.734	3.487	0.252
Increasing maximum diameter		1.010	0.815	1.279	0.928
Increasing maximum height		0.955	0.753	1.198	0.696
Presence of daughter sac		0.947	0.456	2.037	0.887
Use of adjunctive device		0.705	0.249	2.281	0.531
Neck size		0.706	0.535	0.929	0.013
Partially thrombosed		0.135	0.024	0.681	0.016
WEB SL (compared to WEB SLS)		0.706	0.209	1.942	0.532

Bold values indicate significance where p < 0.05.

Hosmer–Lemeshow goodness-of-fit test

The predictive calculator demonstrated a discriminative ability with a c-statistic of 0.744. The Hosmer–Lemeshow goodness-of-fit test revealed a p-value = 0.431, indicating a satisfactory model fit. All variance influence factors were <5. This model is available as an online predictive calculator at: https://neurodx.shinyapps.io/WEBDEVICE/

Discussion

Our study presents a novel predictive calculator for estimating the probability of achieving adequate occlusion (Raymond–Roy classification 1 or 2) using the WEB device for intracranial aneurysms. Utilizing data from 356 patients, we identified various factors and their associations with the likelihood of adequate occlusion.

We identified many variables that may influence the likelihood of adequate occlusion. These include mRS score, increasing maximum diameter, increasing maximum height, increasing neck size and partial thrombosis. Our findings echo previous research that has underscored the importance of many of these variables in predicting aneurysm treatment outcomes.^19–28

Interestingly, our bivariate and multivariate analysis found that larger aneurysms were less likely to be adequately occluded by the WEB device. This supports previous research, which found that larger aneurysms might be more challenging to treat and have lower occlusion rates. ²⁸

Our predictive model demonstrated a reliable discriminative ability with a c-statistic of 0.744, indicating that our model can differentiate between patients who will and will not achieve adequate occlusion approximately 74.8% of the time. This is a valuable tool, but it is vital to emphasize that it should be used to supplement, not replace, clinical judgment.

There are several limitations to our study. As a retrospective analysis, it is subject to inherent biases. Additionally, the study's generalizability might be limited due to the specific patient population and centers included in the WorldWide WEB Consortium. Due to database limitations, variables such as irregularity of aneurysm shape, oversized WEB device, or undersized WEB device could not be retrieved. We recommend future research to validate our model in different populations and clinical settings. Several patients in this study had to be excluded due to incomplete data. Second, cases were included beginning in 2011. Therefore, different device iterations and initial experiences likely do not reflect present results most accurately.

Conclusion

Our study offers a nuanced tool for clinicians to estimate the likelihood of achieving adequate occlusion when using the WEB device for intracranial aneurysms. By considering various factors, this tool can aid clinicians in making informed decisions about patient management. Future research should focus on validating our model in different populations and clinical settings to ensure its widespread applicability.