Article Text

Download PDFPDF
Virtual simulation with Sim&Size software for Pipeline Flex Embolization: evaluation of the technical and clinical impact
  1. Lorenzo Piergallini1,2,
  2. Federico Cagnazzo2,
  3. Giorgio Conte3,
  4. Cyril Dargazanli2,
  5. Imad Derraz2,
  6. Pierre-Henri Lefevre2,
  7. Gregory Gascou2,
  8. Daniel Mantilla2,
  9. Carlos Riquelme2,
  10. Alain Bonafe2,
  11. Vincent Costalat2
  1. 1 Neuroradiology, Hospices Civils de Lyon, Lyon, France
  2. 2 Neuroradiology, CHU Montpellier, Montpellier, France
  3. 3 Neuroradiology, La Fondazione IRCCS Ca' Granda Ospedale Maggiore di Milano Policlinico, Milano, Italy
  1. Correspondence to Dr Lorenzo Piergallini, Neuroradiology, Hospices Civils de Lyon, Lyon 69002, France; lorenzo.piergallini{at}gmail.com

Abstract

Introduction During flow diversion, the choice of the length, diameter, and location of the deployed stent are critical for the success of the procedure. Sim&Size software, based on the three-dimensional rotational angiography (3D-RA) acquisition, simulates the release of the stent, suggesting optimal sizing, and displaying the degree of the wall apposition.

Objective To demonstrate technical and clinical impacts of the Sim&Size simulation during treatment with the Pipeline Flex Embolization Device.

Methods Consecutive patients who underwent aneurysm embolization with Pipeline at our department were retrospectively enrolled (January 2015–December 2017) and divided into two groups: treated with and without simulation. Through univariate and multivariate models, we evaluated: (1) rate of corrective intervention for non-optimal stent placement, (2) duration of intervention, (3) radiation dose, and (4) stent length.

Results 189 patients, 95 (50.2%) without and 94 (49.7%) with software assistance were analyzed. Age, sex, comorbidities, aneurysm characteristics, and operator’s experience were comparable among the two groups. Procedures performed with the software had a lower rate of corrective intervention (9% vs 20%, p=0.036), a shorter intervention duration (46 min vs 52 min, p=0.002), a lower median radiation dose (1150 mGy vs 1558 mGy, p<0.001), and a shorter stent length (14 mm vs 16 mm, p<0.001).

Conclusions In our experience, the use of the virtual simulation during Pipeline treatment significantly reduced the need for corrective intervention, the procedural time, the radiation dose, and the length of the stent.

  • aneurysm
  • flow diverter
  • device
  • brain

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Flow diverter stents emerged as an option to treat complex intracranial aneurysms, such as wide-necked or fusiform aneurysms through blood flow reorganization and endothelialization over the stent structure, but the choice of length and diameter and their deployment may represent a challenging step. The final length and positioning of the stent inside the parent artery are difficult to predict beforehand and are nevertheless critical for multiple reasons.

Implantation of an undersized stent may cause poor wall apposition with the risk of endoleak development. Conversely, oversizing the flow diverter may cause ‘fishmouthing’ and ‘lips’ of the stent, potentially leading to an incomplete distal opening during deployment, as well as endoleaks. Oversizing may also be responsible for a ‘transitional zone’ of decreased metal coverage1 at the level of the aneurysm neck. Evidence exists that optimal wall apposition favorably impacts endothelialization of the stent, influencing the probability of aneurysm occlusion.2

The conventional approach to stent sizing relies on two-dimensional digital subtraction angiography (2D-DSA) measurements of the parent artery. The operator has to anticipate areas of stent elongation or foreshortening, depending on the length and diameter of the parent vessel, size of the stent, and the width of the aneurysm neck. In addition, in order to ensure an optimal wall apposition, the proximal and distal target landing zone must be positioned in a straight segment of the parent artery. For this approach, the experience of the operator plays a major role in achieving proper sizing.

Automated optimization methods for flow diverter simulations have been developed to improve the stent deployment and to optimize treatment outcomes, initially only as research tools.3 4

The recent, commercially available software Sim&Size (Sim&Cure, Grabels, France) is able to perform virtual sizing before stent deployment using per-operative three-dimensional rotational angiography (3D-RA) acquisition. The software simulates the mechanical behavior of the flow diverter, predicting the final stent position and the vessel wall apposition with a high degree of accuracy. The simulation software helps to standardize the selection of the size of the stent, avoiding undersizing or oversizing, and, theoretically, preventing adjunctive maneuvers.

This study aimed to determine if the use of the Sim&Size software during endovascular embolization of intracranial aneurysms with the Pipeline Flex Embolization Device (Medtronic, Irvine, CA, USA) may translate into a measurable procedural advantage, compared with conventional sizing methods.

Methods

This monocentric, retrospective cohort study was approved by the Institutional Review Board of our institution; due to the retrospective nature of this study, the requirement for informed patient consent was waived. Data were retrieved from a prospectively maintained database, including interventions from January 2015 to December 2017. We included all consecutive patients who underwent intracranial aneurysm embolization with the use of the Pipeline Flex Embolization Device (‘Pipeline’ hereafter). Cases were excluded in which multiple Pipeline constructs were planned in advance, in all cases involving giant or fusiform dissecting aneurysms that would have required more than one flow diverter stent to achieve sufficient coverage. These cases were excluded because the Sim&Size software cannot simulate the apposition of multiple, overlapping stents.

Data collection

Data collection included: demographics, aneurysm characteristics, ruptured/unruptured status, vessel diameter, adjunctive coiling, experienced versus less experienced operator, treatment duration, radiation dose, need for corrective intervention due to poor device positioning, stent length and diameter, and complications. Aneurysm dimensions were measured on the 3D acquisition. Ruptured status was defined if flow diversion was performed as the first treatment within 72 hours from the hemorrhage. Large vessel aneurysms were defined as those of the internal carotid, vertebral, or basilar arteries, while small vessel aneurysms were those of the middle cerebral, anterior cerebral, and anterior communicating arteries. An experienced operator was defined as an interventionalist who had performed more than 100 flow diverter embolization treatments. Treatment duration was measured in minutes, derived from the time on the first and last series. The first series was acquired after the positioning of the long sheath and the intermediate catheters in the internal carotid or the vertebral arteries. The end of the procedure was considered the final control run. This calculation was intended to eliminate the time required for the access, from the femoral to the internal carotid/vertebral artery, on which the application of the software does not have an impact. When adjunctive coiling was performed, the duration of coiling was removed in order to achieve the time relative only to the stent deployment. The radiation dose was recorded from the automatic report of the angiographic unit saved on the PACS, in terms of Air Kerma, in mGy. Low-dose 15 images/second fluoroscopy was used throughout the intervention, with series with three images/second in the arterial phase. The need for corrective intervention due to poor stent positioning was defined as a composite endpoint: (1) need to discard the stent due to insufficient or excessive length and deploy a new one (with different length and/or diameter), (2) need to deploy a second stent due to either insufficient coverage of the aneurysm neck or poor proximal apposition, or (3) need to balloon the stent due to poor wall apposition.

Complications were defined as peri-procedural or long-term adverse events that required therapy or a change in the management, even if eventually regressed with no sequelae. Femoral access complications were evaluated and excluded, as the use of the software Sim&Size could not impact on their occurrence.

Pipeline deployment technique

All interventions were performed in a standardized fashion throughout the 3 years of the study: under general anesthesia, via a transfemoral approach, with a triaxial system. A 6 French inner diameter long sheath, either a Destination (Terumo, Tokyo, Japan) or, less commonly, a Shuttle (Cook Medical, Bloomington, IN, USA), was negotiated up to the origin of the internal carotid or the vertebral artery. An intermediate catheter Navien 6 French (Stryker Neuroendovascular, Fremont, CA, USA) was then advanced in the petrous horizontal segment of the internal carotid artery or in the V3-V4 segment of the vertebral artery. The microcatheters used were either a Headway27 (MicroVention Terumo, Tustin, CA, USA) or a Phenom 27 (Medtronic, Minneapolis, MN, USA); the flow diverter stent deployed in all cases was a Pipeline. After deployment, an arterial VasoCT (Philips Healthcare, Best, Netherlands) with 10% diluted contrast medium was performed to evaluate wall apposition.

Conventional sizing method

Conventionally, the operator selected a target distal landing zone as well as a proximal landing zone in a straight part of the parent vessel, in order to ensure a good wall apposition. The first step was to select the diameter of the implant based on the 2D-DSA measurements of the parent artery in the target proximal landing zone. The operator then tried to anticipate areas of stent elongation, especially distally to the aneurysm, as well as of potential foreshortening, mainly depending on the length of the aneurysm neck. The parent vessel centerline was measured between the distal target landing zone and proximal landing zone using 3D-RA reconstruction of the angiosuite to select the stent length. All these measurements contributed to the final choice of stent diameter and length.

Sim&Size software sizing method

The commercially available Sim&Size software, CE marked and Food and Drug Administration (FDA) approved, is a simulation tool that predicts the physical behavior of endovascular devices such as the Pipeline (see figure 1). The software reconstructs the 3D arterial geometry using the data from the 3D-RA acquisition on a local personal computer, without the need of data sent to the cloud. The accuracy of the reconstruction can be manually optimized, if necessary. Once the trajectory of the microcatheter has been selected into the parent vessel, the distal and proximal landing zones of the stent are defined manually by the operator, targeting straight vessel segments. A first recommendation, in terms of device dimensions, is then immediately proposed by the software, to achieve the best deployment within the chosen landing zones, with an appropriate wall apposition. The degree of wall apposition is simultaneously displayed as an interactive color-scale along the stent length. Different landing zones, device dimensions, and the amount of ‘push’ applied during device deployment can all be assayed in real time by the operator, modifying the final position and the behavior of the device.5 6

Figure 1

Technical passages of the simulation with Sim&Size software. Relapse after coiling of a posterior communicating artery aneurysm of the internal carotid artery. (A) Selection of the volume of interest from the 3D acquisition. (B) Optimization of the accuracy of the reconstruction by manually selecting the desired threshold for the arterial walls. (C) Selection of the trajectory of the microcatheter inside the parent vessel (green arrow). (D) After the selection of distal and proximal landing zones in straight segments of the parent artery, the software suggests a Pipeline stent of 4×18 mm, with a good degree of wall apposition, as displayed by the color-scale (an area of poor apposition at the aneurysm neck is always present). The push during deployment is set at 20% (the recommended standard). (E) With a pipeline stent of 3.5×20 mm, a proximal zone of poor wall apposition in shown (in red). (F) With a Pipeline stent of 4×14 mm, wall apposition is good but the proximal landing zone of the stent appears to be in the carotid siphon curvature.

In the present study, immediately after the 3D acquisition, performed as the first series of the intervention, the first operator scrubbed out, downloaded the 3D data and performed the virtual simulation, while the second operator arranged the antero-posterior, latero-lateral, and working projections. The whole process of data extraction and processing took approximately 3–6 min. For an example of simulation on one patient of the present series, see figure 2.

Figure 2

Example of the application of the simulation software for one patient of the series, presenting with an aneurysm at the origin of the ophthalmic artery on the internal carotid artery. (A) Lateral oblique internal carotid digital subtraction angiography acquisition from a Navien 6F, shown in the bottom of the picture. (B) Measurements 2D of the proximal and distal parent artery. (C) Simulation with the Sim&Size software, suggesting a Pipeline stent 4.25×12 mm, with optimal wall apposition and aneurysmal neck coverage, that was deployed immediately after the simulation. (D), (E) Non-subtracted images of the stent deployed in the internal carotid artery. (F) Thick maximal intensity projection reconstruction of the VasoCT acquisition with diluted contrast, showing good wall apposition of the stent.

Statistical analysis

Descriptive statistics were used to characterize patients and interventional procedures: counts and percentages are reported for categorical variables; continuous variables are summarized as median and IQR.

The Kolmogorov-Smirnov test was used to assess the normality of continuous variables. In the univariate analysis, the Mann-Whitney test was used to compare non-parametric continuous variables between ‘with software’ and ‘without software’ groups, unpaired Student t-test for parametric continuous variables, and Fisher’s exact test for binomial variables. In the multivariate analysis, a binary logistic regression model was used to compare ‘with software’ and ‘without software’ groups, with these two groups’ membership considered as the dependent variable, as previously described,7 including all variables with p<0.05 at univariate analysis. Model datasets were constructed in order to assess possible interactions between variables based on theory. The Wald test was performed to assess if the explanatory variables in the logistic regression model were significant. P values <0.05 indicated statistical significance. All statistical analyses were performed with SPSS v.25 (IBM, Armonk, NY, USA).

Results

From January 1, 2015 to December 31, 2017, 198 consecutive patients underwent embolization of a cerebral aneurysm with a Pipeline. Nine patients were excluded, in which multiple Pipeline constructs had been planned beforehand for giant or fusiform aneurysms.

Of the included 189 cases, 95 were treated without and 94 with the Sim&Size simulation software. In 2015, all the 52 patients underwent embolization without the use of the software Sim&Size, in 2016, 36/79 cases were treated using the Sim&Size software, depending on the availability of the Sim&Cure team for the initial formation of the operators, and in 2017, all 58 cases were treated with the software. Overall, 68% of the procedures (65% and 71% of the treatments with and without simulation, respectively) were performed by an experienced operator.

In seven patients, the radiation doses were missing, as well as the duration of the intervention for two patients.

Baseline patient and aneurysm characteristics, details of treatment, and outcomes allocated according to ‘with software’ and ‘without software’ status, are presented in table 1.

Table 1

Baseline patient and aneurysm characteristics, details of treatment, and outcomes allocated according to ‘with software’ and ‘without software’ status

In the univariate analysis, compared with interventions without Sim&Size use, procedures with the software had a lower rate of need for corrective intervention (9% vs 20%, p=0.036). Among the non-simulation group, 19 patients required adjunctive maneuvers: 10 had the stent discarded with positioning of a new one, five had a second stent overlapped on the first one, and five underwent balloon angioplasty (one patient required both a second stent implantation and balloon angioplasty). Among the simulation group, eight patients require adjunctive maneuvers: three had the stent discarded with positioning of a new one and five underwent balloon angioplasty.

The simulation group in the univariate analysis also had a shorter median intervention duration (46 min vs 52 min, p=0.002), a lower median radiation dose (1150 mGy vs 1558 mGy, p<0.001), and a shorter median flow diverter length (14 mm vs 16 mm, p<0.001).

The duration and the radiation dose of the intervention showed moderate linear correlation (Spearman’s correlation coefficient=0.42, p<0.001).

Overall, in the group without simulation, five patients (5%) experienced complications: one small intraparenchymal hematoma due to a perforator artery rupture, likely caused by the guidewire of the stent; two minor strokes (diffusion-positive spot in the lenticular and corona radiata, respectively) with light motor symptoms completely regressed; one modest subarachnoid hemorrhage after procedure likely due to the arterial stretching; and one small intraparenchymal hematoma together with modest subarachnoid hemorrhage, likely due to perforator artery rupture. In the group with simulation, three patients (3%) experienced complications: one embolic partial occlusion of the right central sulcus (rolandic) artery with hemiparesis that partially resolved; one minor stroke with a transient deficit of the leg; and one asymptomatic internal carotid artery dissection. All complications but one (the embolic central artery partial occlusion in the simulation group, in which leg weakness persisted after discharge) fully resolved without sequelae.

Multivariate analysis, conducted with a binary regression logistic model, assessed the influence of the virtual simulation on the investigated outcomes, revealing possible interaction between them. The analysis showed that the use of the Sim&Size software was independently associated with a lower rate of need for corrective intervention (Wald test=4.92, OR=0.35 (95% CI 0.14 to 0.89), p=0.027), with a shorter intervention duration (Wald test=5.03, ß=−0.023, p=0.025), with a lower radiation dose (Wald test=8.95, ß=−0.001, p=0.003), and with a shorter stent length (Wald test=16.17, ß=−0.223, p<0001].

Discussion

The present study is the first assessing the advantages of the virtual simulation by the Sim&Size software in two retrospective cohorts of patients. The use of virtual simulation was associated with a significant reduction of the rate of corrective maneuvers after stent deployment, of the intervention duration, of the radiation dose, and of the length of the stent deployed.

Need for corrective maneuvers

Sim&Size virtual simulation software has been previously described and has been associated with shorter stent lengths compared with standard calculations on 2D/3D images, but the study was not designed to evaluate other technical outcomes and clinical events among subjects treated with and without simulation.6

In the present study, the reduced rate of corrective interventions (9% vs 20%) can be explained by the more accurate sizing of the device, improving the stent wall apposition, and reducing the possibility of an inadequate proximal landing zone opening. Accordingly, we observed a lower rate of balloon angioplasty in the Sim&Size group, with rates in the two cohorts (with and without simulation) in general aligned or lower than previously reported.8 9 In addition, virtual simulation lowered the need either to discard the stent for incorrect dimensions, or to deploy a second device because of the insufficient neck coverage, or the termination of the proximal landing zone in a curve. These events were rarely quantified in previous studies, but it was assumed that they represent a rare but significant cause of longer procedural time and unproductive expenses in real-setting practice, as well as a potential source of complications.

Duration of the treatment and radiation dose

The reduction in duration of the interventions with the use of the software may seem paradoxical, as time for the virtual simulation is added to the standard workflow. The reduction of the duration with the software is likely multifactorial and is independent of the need for corrective interventions at the multivariate analysis. Often, in complex anatomy cases, after the beginning of the stent deployment, the operator has to partially re-sheath the stent multiple times with trial-and-error maneuvers to modify the proximal and distal landing zones. An accurate virtual simulation can minimize these maneuvers, and the operator can proceed more smoothly and rapidly, as the deployment is ‘planned’ ahead. Furthermore, the time needed for conventional sizing methods is not negligible, as the operator has to anticipate the behavior of the stent (foreshortening, and elongation) in different segments of the parent vessel after multiple measurements in 2D-DSA and 3D-RA.

A longer intervention duration has been associated with a higher rate of ischemic events after endovascular treatment of intracranial aneurysms10 or diagnostic angiography.11 12 Hahnemann and colleagues reported that stent-assisted coiling procedures longer than 120 min were significantly associated with silent ischemic events. Another recent study of WEB device associated with stent showed that the mean intervention time among patients with and without complications was 100 min and 76 min, respectively (even though the difference did not reach statistical significance, p=0.09).13 In our study, the complication rate was slightly lower in the simulation group, although the difference was not significant. However, it seems reasonable that shorter procedural time of flow diverter treatment may decrease the risk of symptomatic or silent ischemic events.

Another important point of our study was the reduction of the radiation dose that was linearly correlated with the procedure duration. Accordingly, a lower radiation dose reflects the decreased procedural time thanks to the preplanned virtual simulation. The median reduction in radiation dose (408 mGy) may be estimated as equivalent to an effective dose of approximately 5 milliSievert for a cerebral angiography procedure,14 which represents 2.5 times the average effective dose of a head non-contrast computed tomography scan.15 Although not formally evaluated in the present study, it is possible that the use of the Sim&Size software may allow lower interventions’ costs, as there were significatively less flow diverter stents discarded, the need for adjunctive stents was lower, and the duration of the interventions with the software was shorter.

Stent length

The length of the flow diverter is another parameter that should be accurately selected based on the target vessel characteristics. The reduction in stent length with virtual simulation is another significant finding, as the median of the Pipeline length passed from 16 mm without software to 14 mm with Sim&Size. Sixteen millimeters, without use of the software, represents a relatively short size, and reflects our attempt at deploying the shortest stent as possible, even before the software implementation. The software allowed an additional decrease of approximately 15% of the length of the stent. The use of shorter stents, together with accurate diameter and optimal wall apposition, could also theoretically lower the risk of ischemic complications, though the study was not designed to detect a difference in the two cohorts.

The final length of the stent can be also related to the pushing force applied by the operator, influencing the stent compaction and the surface metal density: the pushing force can be quantitatively simulated with the software, enhancing the precision of the deployment.

Furthermore, shorter stents with a more precise deployment can minimize the coverage of arterial side branches and perforators. Perforator infarctions are not uncommon, ranging from 3%16 up to 14% in the posterior circulation,17 and can lead to devastating consequences.16 A recent meta-analysis showed an incidence of 3% of new visual symptoms after deployment of flow diverters covering the ophthalmic artery.18 However, when a formal ophthalmologist evaluation is undertaken, newly developed visual abnormalities, though mostly asymptomatic, rise up to 40%.19 20 Another recent meta-analysis showed a symptomatic rate of 1% for coverage of the anterior choroidal artery.21 Accordingly, although it was not evaluated in the present study, due to the intrinsic limitation of the retrospective design, it seems cautious to avoid unnecessary coverage of arterial branches or perforators. In addition, if arterial side branch coverage cannot be avoided, giving the possibility to evaluate the stent compression during the simulation time, we can minimize the risk of excessive metal coverage, balancing the compression of the device, and reducing the risk of vessel occlusion.

Finally, the complication rate was slightly lower among the simulation group but the difference was not statistically significant (three in the simulation cohort versus five without simulation), and it was mainly related to the low occurrence in both groups. Assessing the impact of the Pipeline virtual simulation on the adverse events and clinical outcome of the patient would require a larger sample size and prospectively designed studies.

Limitations

The present study has limitations due to its retrospective design. First, patients treated with or without software were not randomly assigned. Indeed, this study involves consecutive patients over 3 years, with an increasing use of the Sim&Size software: in 2015, the software had not been adopted yet, in 2016, it was used for less than 50% of the cases, and in 2017, for all cases. Accordingly, the utilization of the software in 2016 was not random, depending on the availability of the Sim&Cure team. Nonetheless, it is likely that virtual simulation and the presence of the Sim&Cure engineers were required in 2016 for the more challenging cases.

Another limitation, due to the progressively increasing rate of software use over time, is the growing experience of the operators from 2015 to 2017, potentially impacting the measured outcomes. Even though most of the interventions (71% without and 65% with, p=0.44) were performed by the senior operators, who were assumed to be already in the ‘plateau’ phase of the learning curve for Pipeline deployment, it is possible that the increasing experience of the operators during the study period might have influenced the measured outcomes.

Finally, only the duration of the intervention, but not the radiation dose, could be corrected by eliminating the duration of adjunctive coiling, as the Air Kerma dose was calculated automatically by the angiographic unit at the end of each intervention. Nonetheless, the incidence of coiling in the two cohorts was not different.

Conclusions

In our experience, Sim&Size virtual simulation software is a useful tool that can assist the interventionalist during the sizing and the planification of the delivery of the Pipeline Flex Embolization Device. Virtual simulation-assisted procedures were associated with a lower rate of corrective interventions for poor wall apposition, a reduced duration of the treatment and radiation dose, and with a shorter stent length.

References

Footnotes

  • Contributors LP: conceptualization, methodology, statistical analysis, writing-original draft preparation. FC: methodology, writing-reviewing and editing. GC: statistical analysis, methodology. CD: data curation, methodology. PHL: data curation, methodology. GG: data curation, methodology. ID: data curation, methodology. CR: data curation, methodology. AB: data curation, methodology. VC: reviewing and editing, methodology, supervision.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests VC declares having stocks of Sim&Cure.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request to the corresponding author.