Evaluation of Electrocardiogram-Gated Computed Tomography Angiography to Quantify Changes in Geometry and Dynamic Behavior during the Cardiac Cycle of the Nellix Endovascular Sealing System

Background The Nellix endovascular sealing system (EVAS) was a unique concept with regard to its sealing concept that failed, related to high migration rates. We investigated the changes in aortoiliac morphology during the cardiac cycle before and after EVAS using electrocardiography (ECG)-gated CT. Methods Eight patients scheduled for EVAS were prospectively enrolled. ECG-gated CT scans were made pre- and postoperatively. Measurements were performed in the mid-systolic and mid-diastolic phases. Endpoints were changes in infrarenal aortoiliac morphology postoperatively compared to preoperatively and their changes in the cardiac cycle. Results Both pre- and postoperatively, there were no changes during the cardiac cycle. EVAS caused an increase in neck diameter and surface in both phases (p < 0.001). EVAS increased the luminal AAA volume (p < 0.001), with a decrease in thrombus volume (p < 0.001) in both phases and an increase in total volume (p < 0.001) in the systolic phase. During follow-up, one patient presented with >5 mm migration. There were no differences in the movements of this patient compared to the remaining patients. Conclusion The cardiac cycle had a very limited effect on the aortoiliac dynamics before and after EVAS and, therefore, there is probably not a role for ECG-gated CT in enhanced surveillance programs. EVAS itself has a significant impact on anatomy, particularly the neck diameter, length, and volumes of the AAA.


Introduction
After its commercial introduction in 2013, the Nellix endovascular sealing system (EVAS) was rapidly adopted in the vascular community, but late complications such as proximal endoleaks and proximal migration were observed. Tis frst led to a refned IFU which signifcantly reduced the applicability of the technique [1] and later the fnal withdrawal from the market in May 2022. Te system was distinct from other systems by sealing the aneurysm with polymer-flled endobags and thus fxating inside the aneurysm sac itself. Lateral bending due to incomplete endobag flling in AAAs with large thrombus volumes was suggested as a driving force behind distal migration. Te polymer-flled endobags increased the rigidity of the stents and could thus reduce the risk on migration [1,2].
Excessive motion of the aorta, in both the longitudinal and lateral directions, has been suggested as a risk factor for endograft migration after conventional EVAR [3]. Tis could also be an important factor after EVAS, where downward forces on the polymer shell and lateral forces through curves in the stent frames may play a role. Te current gold standard for both preoperative sizing and follow-up after EVAR/EVAS is static CT imaging. Tis method relies on static imaging of a dynamic process; hence, the accuracy of these measurements is uncertain.
A previous study has examined the translation and aortic curvature of EVAS during the cardiac cycle, but no quantitative anatomical parameters have been examined previously [4]. Tis study aimed to get more insight into the efect of EVAS on the motion of the aortoiliac tract by measuring changes in ECG-gated computed tomography (ECG-gated CT) before and after EVAS. Tis is still clinically relevant, as many patients that were treated in the past are still in enhanced surveillance programs [5].

Study Design.
Tis was a single-center prospective observational study. Eight patients who were scheduled for EVAS were prospectively enrolled in this study. Each patient underwent an ECG-gated CT preoperatively and six weeks after EVAS. Te preoperative CT was used for procedural planning. Patient demographics as well as procedural data were collected and all pre-and postoperative imaging was studied. Te study was conducted according to the principles of the Declaration of Helsinki (64 th WMA General Assembly, Fortaleza, Brazil, 2013) and in accordance with the applicable guidelines, regulations, and acts. For this prospective study, approval by an independent institutional review board was obtained, and each patient had given written informed consent before inclusion in this study. Patient scan information and data were stored anonymously. Te study was registered on clinicaltrials.gov with ID: NCT02438605.

2.2.
Imaging. An (electrocardiography) ECG-gated CT provides images at various stages of the cardiac cycle. By analyzing these images, it is possible to measure the movements of the stent graft and the arteries during the diastole and systole cycles of the heart. A Phillips Brilliance iCT 256-slice scanner was used to perform the ECG-gated CT scans in all cases, with dose modulation (automatic), using a standardized protocol, and during a single breath hold. Radiation exposure was tube voltage 120 kV, reference mAs 400 mAs, and pitch 0.34. Te scan was made at 10 time points during the cardiac cycle and at the highest atrial pressure (20%) and lowest atrial pressure (78%), which were the chosen phases for further assessment. Te studied slice thickness was 0.9 mm, and Xenetix 300 (Guerbert GmbH, Sulzbach, Germany) was used as iodine contrast medium and adapted according to the patient's weight. Te scan was started using bolus triggering at the height of the aortic arch at a threshold of 150 HU over baseline.

Analysis.
All images were analyzed using 3Mensio (Pie Medical Imaging, Maastricht, Te Netherlands). Measurements were performed based on anatomical landmarks, after creating a central lumen line (CLL) with the workfow assistant in 3Mensio. Te measurement protocol was partially based on reporting standards and previous studies regarding CTA-imaging after EVAR [6]. In order to create the CLL, a 3D reconstruction and semiautomatic segmentation of the abdominal aorta, renal, and iliac arteries were needed in two phases of the cardiac cycle, 20% and 78% phases, per patient. Te CLL was required to be able to view the images perpendicular to the vessel/arterial axis. Te CLL was created semiautomatically, but manually checked and confrmed, by sliding through every plane on snake view and correcting where necessary.
Measurements on morphology were divided into three areas, including the infrarenal neck, the AAA sac, and the common iliac arteries. Te start of the neck was defned as the lowest point of the inferior renal artery. Te level with 15% increase in neck diameter compared to the diameter at the level of the lowest renal artery was defned as the end of the neck. Te start of the aneurysm sac was defned as the end of the neck, and the end was defned as the frst slice cranial to the apex of the aortic bifurcation. Te α-angle was defned as the angle between the suprarenal aorta and the neck of the aneurysm. Te β-angle was defned as the angle between the neck and the aneurysm sac.
All measurements were performed by two independent researchers (L.H. and T.L.) and tested for interobserver variability. Outliers and diferences were remeasured for accuracy and corrected if necessary.

Statistical Analysis.
Data collection and analysis were performed using IBM SPSS Statistics version 21.0 (IBM Corporation, Armonk, NY, USA). All data were checked for outliers and reevaluated for correctness. Te Wilcoxon signed rank test was used to test for diferences in anatomy both preoperatively and postoperatively as well as during the cardiac cycle. Te values p < 0.01 were considered statistically signifcant. All measurement data are presented using medians and interquartile ranges (IQR, Q25-Q75).

Patient Data. Patient demographics are presented in
Te study included 8 patients suitable for elective AAA repair using the Nellix endovascular sealing system (Endologix, Irvine, CA, USA), of which 6 (75%) were male, with a median age of 75.5 (Q25-Q75: 62.2-83.1) years. Te maximum AAA diameter was 58.5 mm (range 42.0-75.1 mm). A median volume of 83 mL (range 50-135 mL) of the polymer was used to reach a mean fll pressure of 185 mmHg (range 166-220 mmHg). During implantation, one patient that had a type Ia endoleak caused by underflling of the endobags had a secondary fll, resolving the issue. In another patient, concomitant angioplasty and stenting of the left superfcial femoral artery were performed using a self-expanding nitinol stent (Smart Flex, Cordis Corp., Bridgewater, NJ, USA) because of a fowlimiting stenosis.
Te median procedure time was 75 minutes (range 49-101 min), and patients were admitted for a median of 2 days (range 2-4 days). One patient underwent a successful reintervention for a common femoral artery occlusion one day after surgery. Using intra-arterial thrombolysis, the occlusion was dissolved, and the patient was discharged four days after surgery and had a further uneventful follow-up. One patient developed an occlusion of the left popliteal artery seven days after the operation which was successfully treated with thrombolysis. All other patients underwent an uneventful recovery, and no major adverse events were seen at discharge and after 6 weeks of follow-up. Te median follow-up was 25 months (range 6-36 months). Within this period no patients died, and there was one patient that presented with a distal migration of ≥5 mm without clear signs of a type Ia endoleak.

Interobserver Variability.
Te median ICC for continuous variables to assess interobserver variability was 0.68 (range 0.50-0.78) and was therefore classifed as good according to Landis and Koch interpretation of ICC.

Neck Anatomy.
Te changes in neck anatomy are displayed in Table 2.
Tere were no signifcant changes observed in the infrarenal neck area during the cardiac cycle in both the preand postoperative CT scans. Almost all variables were higher in the diastolic heart phase except for neck length and the lumen volume. With regard to neck angulation, there are some slight changes in both alpha and beta angles and the angulation of the renal arteries during the cardiac cycle, without statistical signifcance, and with a wide variability in both the pre-and postoperative scans.
When comparing the pre-and postoperative CT scans, a signifcant increase in neck diameter and surface was observed after EVAS in both the diastolic and systolic phases. In addition, the neck length signifcantly decreased after EVAS between both phases. Te sealing length, defned as the start of the endobag to the start of the aneurysm, was 23.0 mm (range 4.8-34.1 mm) at the peak diastolic phase and 18.6 mm (range 6.3-36.3 mm) at the peak systolic phase (p � 0.674).

AAA Sac Anatomy.
In preoperative CT scans, there were no signifcant changes during the cardiac cycle in the aneurysm sac (Table 3). In the postoperative scan, we observed a slight but signifcant change of the luminal volume in the cardiac cycle that was larger in the peak diastolic phase (±6 mL).
When comparing the pre-and postoperative scans, there was a signifcant increase in the luminal volume after EVAS with a signifcant decrease in thrombus volume in both the systolic and diastolic phases. Te total sac volume was signifcantly increased in the peak systolic phase. In this phase, the AAA length was also increased in contrast to the diastolic phase.

Iliac Artery Anatomy.
During the cardiac cycle, there were no signifcant changes observed in the anatomy of the iliac arteries in either preoperative or postoperative scans (Table 4), except for the diameter and area of the distal right common iliac artery.
When comparing the pre-and postprocedural imaging, there were also no signifcant diferences except for the diameter of the right common iliac artery, which was larger after implantation when measured in the diastolic phase. Te total angulation of both iliac arteries was lower after endovascular sealing as compared to preoperative, albeit nonsignifcant.

Discussion
Te current study aimed to investigate the dynamic behavior of aortic anatomy in patients undergoing EVAS. Measurements were performed according to current reporting standards for conventional CT scans and adapted for dynamic CT scanning based on studies performed in a comparable study setting [7,8]. Te current data contribute to the knowledge on this technology, as they suggest that the use of dynamic CT scanning is not indicated in the follow-up of patients that were treated with Nellix in the past and also identifed a decrease in neck length after placement as a potential risk factor.
In May 2022, Endologix announced in a targeted communication that the production of Nellix devices and its accessories had stopped (Endologix, Nellix End of Life Communication, 10 May 2022). In a recent meta-analysis including 703 patients from seven studies with a mean follow-up of more than two years (range 24-72 months), the pooled estimated incidence of type 1 endoleak, migration, and re-intervention were 25%, 22%, and 27%, respectively [9]. Although the technology is no longer used in clinical practice, many patients are still under surveillance. Recently, the European Society for Vascular Surgery came out with a focused update on their guidelines, recommending that patients should be enrolled in enhanced follow-up using conventional CTA and duplex ultrasound. A potential role of ECG-gated CTA was not discussed [5].
Te data presented in this study suggest that the EVAS reconstruction was stable during the cardiac cycle in these patients, as we found in general no signifcant changes       during the cardiac cycle in both the pre-as well as the postoperative state. Tese results, however, were measured in the postoperative period and, as a consequence do not ensure the longitudinal stability of the reconstruction. It is known that in patients, particularly when treated outside the IFU, stent migration may occur, leading to a proximal endoleak [9]. Root cause analysis of migration had indicated lateralization of the stents as the driving factor of migration, leading to a refnement in the instructions for use in 2018. Te maximum aneurysm sac/maximum fow lumen diameter ratio became a parameter in order to increase the stifness of the reconstruction [1]. In the current study, the reconstruction lead to a decrease in neck length compared with the CT before surgery. Whether this may have contributed to the risk on migration cannot be concluded from the current data. Te current population had a median follow-up of 2-years and within that timeframe one patient presented with a migration of more than 5 mm. In this patient, there was no signifcant motion during the cardiac cycle. Terefore, this may suggest that an ECG-gated CT scan in the direct postoperative phase is not predictive for the occurrence of migration. To date, no longitudinal studies have been performed using ECG-gated CT scanning, but this would certainly be of value as increased movement during the cardiac cycle might provide an early sign of instability. Te Nellix device itself caused signifcant changes in the aortic anatomy, particularly with regard to an increased neck diameter, surface, and length, and to the aortic volumes, including the luminal and thrombus volumes. Previous studies which have examined the post-EVAS changes of the anatomy confrm these fndings [10]. It has been shown that implantation of the stents leads to changes to the iliac anatomy and the volumes of the aneurysm and thrombus, consistent with the current observations.

Radiology Research and Practice
All of these changes are likely to be caused by the polymer-flled endobags. Te pressurization of these endobags may squeeze the thrombus, declining its volume and increasing the fow volume, and could exert a pressure on the arterial wall, leading to an increase in neck size and AAA volume. Tese observations emphasize the diference of the EVAS technique when compared to standard EVAR. Te implications of these observations, however, are yet unknown. As an unexpected observation, Berg et al. described a lower incidence of the postimplant syndrome after EVAS when compared to EVAR [11]. Whether this is related to the absence of fresh thrombus and/or the decrease in thrombus volume remains to be investigated, but a biological impact of thrombus cannot be excluded.
Te current study has limitations. First, the sample size was relatively small. Second, we used only two series of the ECG-gated CT scan. Te scans were performed with a total of ten phases during the cardiac cycle. An analysis of all ten phases might have provided more data. However, to attain for the anatomy in its' maximum range, the diastolic and systolic pressures are chosen as the best phases for analysis as they are the phases with both the minimal and maximum changes during the heart cycle. Moreover, CT scans were only available in the early postoperative phase. Tis would have been relevant, for example, to see if the decrease in neck length after EVAS would be stable in time or would be progressive. Te interobserver variability of this study was small, as little to no changes had to be made after secondary measurement of all variables. Software automatization of the CLL is mostly to credit for the accuracy, as measurement can be reproduced in a consistent way.

Conclusion
Tis study shows that the cardiac cycle has little to no efect on the morphology of the arteries both before and shortly after EVAS, and therefore there is probably not a role for ECG-gated CTA in enhanced surveillance programs. EVAS itself has a signifcant impact on anatomy, particularly the neck diameter, length, and volumes of the AAA.

Data Availability
Te datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.