Unruptured, intracranial aneurysms affect approximately 2–4% of the general population.1 2 Since the introduction of detachable platinum coils almost three decades ago, the need for open surgical clipping of aneurysms has been substantially decreased due to the use of endovascular coil embolization. Up to 22 000 aneurysms are treated via endovascular therapy annually.3 Wide necked aneurysms (WNA), defined as a neck diameter ≥4.0 mm or dome-to-neck ratio <2.0, present a unique therapeutic challenge and often require an adjunctive device.

The well established stent assisted coiling technique, in which intracranial stents reconstruct the aneurysm neck to facilitate coil packing, has improved the outcome of WNA treatment over standalone coiling.4 5 The new generation of stents include the eNeuroform Atlas stent system (Stryker Neurovascular) and the low profile visualized intraluminal stent (LVIS) and LVIS Jr devices (Microvention Terumo). Another device specifically designed for WNAs is the Woven EndoBridge (WEB) Aneurysm Embolization System (Microvention Terumo), an intrasaccular flow diverter.

Few studies have compared treatment of aneurysms with different devices due to the heterogeneity of trial design, patient selection, and device limitations.6 7 We present a comparative analysis between the Neuroform Atlas, LVIS, and WEB devices to assess for differences in primary effectiveness, safety, and retreatment rates.

The study was designed to compare objective, prospectively collected, core laboratory adjudicated data in subjects undergoing treatment for WNA with the Neuroform Atlas stent, LVIS, and WEB. The study selection criteria were trials that reported on the primary effectiveness rate of complete (ie, 100%) aneurysmal occlusion without significant (>50%) parent artery stenosis, with central core laboratory adjudicated outcomes, at the 12 month follow-up using DSA. The comparative analysis aimed to assess differences across the following endpoints: (1) primary effectiveness (defined as complete aneurysmal occlusion without retreatment/>50% parent vessel stenosis), (2) primary safety (exact definitions varied between the studies), (3) complete aneurysmal occlusion (angiographic obliteration (ie, Raymond class I),8 and (4) retreatment rates of target WNA (outcomes evaluated at the 12 month follow-up). Matching adjusted indirect comparison (MAIC) analysis was used to compare outcomes.

Device instruction for use and published clinical trial results were reviewed for extraction of objective, prospectively collected, core laboratory adjudicated data. Patient demographics, aneurysm morphology, and data corresponding to primary effectiveness (complete occlusion without retreatment or >50% parent vessel stenosis), safety, complete occlusion (Raymond Class I), and retreatment rates were obtained from: (1) the ATLAS (Safety and Effectiveness of the Treatment of Wide Neck, Saccular Intracranial Aneurysms With the Neuroform Atlas Stent System) study publications9 10 for the Neuroform Atlas, and the FDA summary of safety and effectiveness data and study publications for LVIS (Pivotal Study of the Low Profile Visualized Intraluminal Support)11 and WEB-IT (Woven EndoBridge Intrasaccular Therapy Study).12

We classified data for ATLAS study patients according to the inclusion/exclusion criteria of each comparator device, and only compared data from ATLAS study patients that met the inclusion/exclusion criteria of studies of the specified study. All subjects harbored wide necked aneurysms as defined by a neck width of ≥4 mm or dome-to-neck ratio <2. ATLAS subjects who met the following criteria were included in the ATLAS/LVIS subgroup: (1) aneurysm size of 4–20 mm and (2) patients aged 18–75 years. ATLAS subjects who met the following criteria were included in the ATLAS/WEB-IT subgroup: (1) target WNA location was middle cerebral artery bifurcation, anterior communicating artery, internal carotid artery bifurcation/terminus, or basilar apex; (2) aneurysm size 3–10 mm; (3) patients aged 18–75 years; and (4) modified Rankin Scale score of 0 or 1.

As described previously, inclusion/exclusion criteria used by each comparator study were applied to ATLAS data to create two subgroups, one for LVIS and another for WEB. Crude (unadjusted) estimates were described for the ATLAS/LVIS and ATLAS/WEB-IT subgroups. The MAIC method13 was used to compare patient level outcomes from the ATLAS study with the aggregate outcomes reported in the literature for the LVIS and WEB studies. First, this method weights the individual patients in the ATLAS/LVIS subgroup to match the LVIS study on patient age and aneurysm characteristics (ie, location, dome height/width, neck measurements, and dome-to-neck ratio) to generate balanced trial cohorts. Second, the outcomes were compared between the ATLAS/LVIS subgroup and the LVIS study by applying the weights to the ATLAS/LVIS subgroup. Both crude (unadjusted) and weighted values are presented in the tables. This same process was followed to compare ATLAS/WEB versus WEB-IT. Crude descriptive analyses were calculated using SAS 9.4 (SAS Institute, Cary, North Carolina, USA). The MAIC analyses were performed using the maic package in R (R Statistical Software, V4.1.3; R Core Team 2021). P values <0.05 were considered statistically significant.

This study constituted the first comparative analysis between the three most used FDA approved endovascular devices to aid in the treatment of WNA using prospective independently adjudicated data. Following the application of our inclusion criteria, three trials were found eligible for inclusion (other prospective trials, eg, CLinical Assessment of WEB device in Ruptured aneurYSms (CLARYS) and WEB Clinical Assessment of Intrasaccular Aneurysm Therapy (WEBCAST) were excluded; table 3). Our data showed that the Neuroform Atlas provided statistically significant improvement in complete occlusion rates compared with the WEB and LVIS, improvement in the primary efficacy endpoint compared to the WEB-IT, and a lower retreatment rate for ATLAS compared with the WEB-IT cohort. Safety events rates were not significantly different between ATLAS/LVIS, and between ATLAS/WEB-IT, with a trend for lower rates in the WEB-IT cohort.

Primary safety event definitions in the three trials

WNA are known to pose unique therapeutic challenges but their exact definitions varied considerably between the three included trials. While the WEB-IT trial was exclusive to bifurcation aneurysms,12 both the ATLAS and LVIS trials included side wall aneurysms (about 41.4% of aneurysms in the LVIS11 trial were sidewall aneurysms,11 vs 45.1% and 24.1% of aneurysms in the ATLAS anterior9 and posterior circulation10 cohorts, respectively). To account for this, we compared ATLAS patients who would have met the inclusion criteria of each corresponding device trial to the original trial cohort. When comparing ATLAS with WEB-IT, we excluded side wall aneurysms from the analysis. As the LVIS trial allowed side wall aneurysms, a similar distribution of side wall aneurysms was included in the adjusted ATLAS comparison cohort (table 2). Nonetheless, the high rates of complete aneurysm occlusion among subjects treated with the Neuroform Atlas at 12 months in this study (87.5%) are in line with three recent retrospective studies that reported complete occlusion rates at~12 month follow-up of 85.7% at 6–12 months14, 82%,15 and 92%.6 In contrast, the retreatment rates were relatively higher (3.8–6.9%) compared with the no retreatment outcome reported in two studies using the Neuroform Atlas to treat wide necked aneurysm at 12 months.6 15 The higher retreatment rates of this study are likely influenced by the prospective and rigorous follow-up in the humanitarian device exemption/pre-market approval trials compared with the inherent bias of uncontrolled low-quality retrospective studies.

The WEB device is deployed within the aneurysmal sac, similar to coils, and ideally obviating the need for parent artery treatment. A cumulative analysis of the cohort from three prospective multicenter studies, the French Observatory,16 the WEB Clinical Assessment of Intrasaccular Aneurysm Therapy (WEBCAST),17 and the WEBCAST-2,18 showed a complete aneurysm occlusion rate of 52.9% at 12 months.19 This occlusion rate was similar to the results of safety and effectiveness data,20 the WEB-IT trial.12 The retreatment rate for the WEB pooled cohort was 7.2% between the index procedure and the 12 month follow-up,21 which was close to the rate of the WEB-IT study, thus validating the data we used for comparison between devices.

Long term data with both the WEB/Atlas are currently pending given the recency of both devices. Moreover, while the differences between WEB embolization and coil embolization of aneurysms remain to be elucidated, the current best data stem from previous literature (eg, CARAT study22 on aneurysms coiling). While most neck residuals are likely benign,23 a higher risk of aneurysm rerupture is associated with neck residuals,22 24 therefore underscoring the importance of achieving complete angiographic occlusion which would favor the Atlas with higher occlusion rates. Nonetheless, the 5 year results of the WEB-IT trial, recently published, reported complete occlusion and adequate occlusion rates of 58.1% and 87.2% of patients, respectively.25 No rebleeding events from the index aneurysms were encountered throughout the study follow-up, regardless of occlusion status or initial rupture status, pointing towards a benign natural history of neck remnants with the WEB device which is further in line with the 5 year follow-up of the WEBCAST/WEBCAST-2 combined data.26 Furthermore, 76.8%25 of the aneurysms’ occlusion grade in the 5 year WEB-IT remained stable or improved, again in line with the French data (87.1%).26

In this study, the primary focus of the paper was to compare efficacy across the devices given the significant heterogeneity in the definitions of safety endpoints across the trials. Regardless, despite lower primary effectiveness rates, the WEB-IT trial trended towards having a comparatively lower reported safety events rate compared with the ATLAS trial, although this did not reach statistical significance in the analysis (table 2). This could be attributed, partially at least, to the different safety endpoint definitions, with lower thresholds for inclusion of transient adverse events as safety events in the ATLAS trial compared with the WEB-IT and LVIS trials (table 4; definitions obtained from the summary of safety and effectiveness data of each trial).

Summary of main endpoints in the three included trials versus other trials

As with many prospectively collected data analyses and comparative studies, bias may result from the lack of a direct comparator for the same study. Despite opting to compare the ATLAS results using only aneurysms that would meet the inclusion criteria for that trial and using the MAIC analysis to minimize bias, some heterogeneity remains difficult to account for, including the potential differences between the central core laboratory adjudications of each trial. The strict inclusion criteria for our analysis resulted in excluding other prospective trials which might potentially limit the generalizability of the results.

This study demonstrated that wide necked intracranial aneurysms treated with the Neuroform Atlas had higher occlusion rates and similar or reduced retreatment rates than patients treated with the LVIS or WEB. Comparison of prospectively collected, core laboratory adjudicated data can provide meaningful comparisons between studies and devices. The onus is on the clinicians to determine if this translates into clinically measurable benefits for the patient.