Stent-graft surface movement after endovascular aneurysm repair: baseline parameters for prediction, and association with migration and stent-graft-related endoleaks

Objectives To evaluate the influence of baseline parameters on the occurrence of stent-graft surface movement after endovascular aneurysm repair (EVAR) and to investigate its association with migration and stent-graft-related endoleaks (srEL). Methods In this retrospective, cross-sectional study, three-dimensional surface models of the stent-graft, delimited by landmarks using custom-built software, were derived from the pre-discharge and last follow-up computed tomography angiography (CTA). Stent-graft surface movement in the proximal anchoring zone between these examinations was considered significant at a threshold of 9 mm. The Cox proportional hazards model was used to determine baseline variables associated with the occurrence of stent-graft surface movement. The association between migration and srEL with stent-graft surface movement was tested with the chi-square and the Fisher exact test, respectively. Results Stent-graft surface movement was observed in 54 (28.9%) of 187 patients. Multivariate analysis revealed that age ([HR] 1.05; p = 0.017), proximal neck diameter ([HR] 5.07; p < 0.001), infrarenal aortic neck angulation ([HR] 1.02, p = 0.002), and proximal neck length ([HR] 0.62, p < 0.001) were significantly associated with the occurrence of stent-graft surface movement. Migration and srEL occurred in 17 (31.5%) and 5 (9.3%) patients, with and 11 (8.3%) and 2 (1.5%) without stent-graft surface movement (p < 0.001, p = 0.022). Conclusions Age, neck diameter, infrarenal neck angulation, and proximal neck length were significantly associated with the occurrence of stent-graft surface movement. Apart from possible use of adjunctive sealing systems, concerned patients may benefit from regular CTA surveillance, enabling timely diagnosis of subtle changes of stent-graft position. Key Points • Stent-graft surface movement, demonstrating subtle, three-dimensional changes in stent-graft position in the proximal anchoring zone, can be derived from CTA examinations. • Age, proximal neck diameter, and infrarenal neck angulation were significantly associated with an increased incidence of stent-graft surface movement. Stent-graft surface movement is significantly more frequent in patients with stent-graft migration and stent-graft-related endoleaks. • Consideration of risk factors for stent-graft surface movement may help to identify patients who might benefit from regular CTA surveillance and timely diagnosis of subtle changes of stent-graft position, enabling re-interventions to prevent migration and srEL. Electronic supplementary material The online version of this article (10.1007/s00330-019-06282-w) contains supplementary material, which is available to authorized users.


Introduction
The robustness of the overlapping area between the aortic wall and the stent-graft is a determining factor for the long-term durability of endovascular aortic repair (EVAR) [1,2]. Stentgraft migration has a reported prevalence ranging from 1.1 to 28% [2][3][4][5]. It is responsible for the majority of late complications after EVAR, including late stent-graft-related endoleaks, resulting in aneurysm sac enlargement, and even rupture [6,7]. Different mechanisms, such as the radial forces of selfexpandable stent-grafts due to oversizing [8,9] and the pulsatile forces of blood flow [10][11][12], were suggested to be associated with continuous changes in stent-graft position and decreasing apposition of the stent-graft surface, consequentially causing migration over time. In addition, disease progression could trigger and accelerate both mechanisms [13,14]. Stentgraft dynamics over time are complex and three-dimensional [15][16][17], and not only limited to a one-dimensional, most commonly caudal displacement of the stent-graft, as defined for migration [18]. Longitudinal displacement may occur simultaneously or consecutively with stent-graft surface movement in other directions [16,19]. Focusing only on caudal migration, stent-graft surface movement in other directions, as a result of subtle aneurysm neck changes, may be overlooked. Diagnosis of these subtle changes on regular computed tomography angiography (CTA) images is difficult, even with centerline reconstructions on a vascular workstation [16,19]. On the contrary, the calculation of three-dimensional (3D) surface models, derived from CTA images, enables a simplified visualization and assessment of stent-graft surface movement [15,19], with the potential of a more sensitive determination of changes of the stent-graft position within the infrarenal aortic neck. Therefore, knowledge of the predisposing factors associated with the occurrence of stent-graft surface movement may influence the pre-operative planning in terms of patient selection for EVAR, the possible use of adjunctive sealing systems, and surveillance strategy. Detection of stent-graft surface movement during CTA surveillance, particularly in patients at risk, may allow for planning of timely prophylactic re-interventions to prevent stent-graft migration and consecutive sealing loss.
The primary objective of this study was to evaluate the influence of baseline clinical and morphological parameters on the probability of stent-graft surface movement in patients undergoing EVAR. The second aim was to investigate the association of stent-graft surface movement with migration and stent-graft-related endoleaks (type 1 and 3), representing the leading causes of rupture after EVAR [20,21].

Study design
This was a retrospective single-center, cross-sectional study of patients with abdominal aortic aneurysm (AAA) who were followed after elective EVAR. The institutional review board (No. 1843/2018) approved the study protocol and waived written, informed consent.

Study population
Our institutional database was screened for patients who underwent elective endovascular AAA repair between May 2002 and July 2014 and who met the following inclusion criteria: (1) EVAR with bifurcated stent-grafts without intraprocedure implantation of proximal adjunctive sealing devices and (2) available standardized CTA, (a) pre-procedure, within 3 months prior to intervention, (b) pre-discharge within 1 week after intervention, and (c) after a time interval of at least 6 months.
All patient data were anonymized and de-identified prior to analysis. One hundred eighty-seven patients (14 female) with a mean age of 73.2 ± 8.0 (range, 52-90) years were available for final analysis. Details of patient selection and patient characteristics are given in Fig. 1 and Table 1, respectively. A part of the study population was included in a previously described cohorts [15,46]. In contrast to our study, analyses referred on the comparison of stent-graft surface movement in patients with and without a type 2 endoleak, and clinical and morphological parameters for the prediction of late stent-graft related endoleaks, respectively.

CTA examination protocol and stent-graft surface movement
CTAs were performed in the caudo-cranial direction using a 16-slice (Somatom Sensation 16, Siemens Healthineers; n = 122) or a dual-source (Somatom Definition Flash; Siemens Healthineers; n = 65) scanner. The institutional standard protocol included an arterial phase, ranging from the celiac trunk to the groin solely for pre-interventional imaging, followed by a late phase limited to the extent of the stent-graft in case of postinterventional imaging. Acquisition parameters are given in Table 2.
Stent-graft surface movement was assessed as described by Nolz et al [15]. In short, three-dimensional (3D) surface models of the stent-graft, delimited by landmarks using custom-built software, were derived from arterial CTA images. The software calculated a vector for each point of the stent-graft surface, describing the surface movement between the postprocedure and final follow-up CTA, which was given in millimeters. Based on results of previous studies [15,22], where significantly higher rates of EVAR failure were reported in patients with a stent-graft movement more than 9 mm, this threshold was considered significant.

Measurements and definitions
The influence of baseline clinical and morphological parameters on the probability of stent-graft surface movement > 9 mm in the proximal anchoring zone was analyzed. As previously reported [15], this zone was defined as the proximal 3 cm of the covered stent-graft.
Baseline clinical parameters included age (at the time of EVAR), sex, and body mass index (BMI). Baseline morphological parameters included anatomical variables measured in the pre-procedural CTA scans, device-dependent variables collected from the pre-interventional CTA and/or the stentgraft procedure, and early postinterventional variables assessed on the pre-discharge CTA. Measurements were performed with the syngo.via imaging software (Siemens Healthineers).
Anatomical variables determined in accordance with the Society for Vascular Surgery standards for EVAR [23] included (1) aneurysm sac diameter, (2) maximum diameter of the proximal anchoring zone, (3) iliac sealing diameters, and (4) proximal and distal neck lengths. The supra-and infrarenal aortic neck angulations were measured based on the 2D methodology described by van Keulen et al [24], and adapted for measuring along a semi-automatically drawn center lumen line (CLL). Assessment of the psoas muscle area (PMA) was performed as described by Indrakusuma et al [25]. The area of both psoas muscles was added and corrected for patient height using the formula (left PMA + right PMA)/ (height 2 ) [26]. This variable was determined as heightcorrected PMA.
The presence of a thrombus with > 2 mm thickness at the circumference of the proximal anchoring zone was evaluated and classified as follows: (1) no thrombus, (2) thrombus < 25%, (3) thrombus 25-50%, or (4) thrombus ≥ 50%. The presence of calcifications was categorized according to the same classification. On the basis of gathered anatomical variables, compliance requirements with the manufacturer's instructions for use (IFU) were analyzed, and rated as within or outside the IFU.   (5) presence of any active fixation mechanism (e.g., hooks, anchoring pins, or dull barbs).
Early postinterventional variables included (1) the presence of a type 2 endoleak (T2EL) and (2) the renal artery to stent-graft distance (RSD). RSD was measured along a semi-automatically drawn CLL, as described by Bastos Goncalves et al [27]. RSD was defined as the distance between the lowest renal artery to the lowest stent-graft fabric marker, representing the level of circumferential stent-graft covering of the aortic wall. RSD changes between both examinations indicate stentgraft migration, which was defined as an RSD increase of > 0.5 cm [28]. Routinely performed CTA follow-up examinations were retrospectively screened for the presence of a stent-graft-related endoleak, type 1 or type 3, as described by Chaikof et al [18].

Statistical analysis
Normally distributed, continuous data were presented as the mean ± standard deviation. Potential differences between groups were compared using the t test. Non-normally distributed data were described by medians and interquartile ranges (IQRs). Possible differences between groups were tested with the Wilcoxon-Mann-Whitney U test. Dichotomous variables were described in absolute numbers and percentages, and possible differences between groups were tested by the chi-square test or the Fisher exact test, as appropriate. The univariate Cox proportional hazards model, with calculation of hazard ratios with 95% confidence intervals, was used to investigate the influence of baseline clinical and morphological parameters on the probability of stent-graft surface movement-free survival. A multivariate Cox proportional hazards model was conducted by using a backward selection of parameters, with a limit of p < 0.1 required to enter and to stay in the model. The proportional hazards assumption was tested for each variable individually using the time-dependent covariate method [29]. No relevant violations of the assumption were found (see supplementary material). The effect of adding interaction terms was assessed using the partial likelihood ratio test. No statistically significant interactions were determinable. Additionally, receiver operating characteristics (ROC) analysis was performed in variables, significantly associated with stent-graft surface movement in the multivariate analysis. Cutoff values were defined with the Youden J statistics. The Kaplan-Meier life table method was used to determine freedom from stent-graft surface movement, migration, and stentgraft-related endoleaks. For these survival analyses, all observations were censored at the time of the patient's last CTA. All tests were two-sided; significance was assumed at p < 0.05. All statistical analyses were performed using SPSS for Windows (version 24.0; IBM Corporation).
For identification of patients with stent-graft surface movement, the optimal cut-off points for age, neck diameter, neck length, and infrarenal neck angulation were 72.

Discussion
The present study analyzed the influence of baseline clinical and morphological parameters on the occurrence of stent-graft surface movement after endovascular aneurysm repair. A multivariate Cox model revealed a relation between stent-graft surface movement and patient age, the proximal aneurysm neck diameter as well as the infrarenal aortic neck angulation, whereas proximal neck length proved to be significantly Furthermore, stent-graft migration and srEL occurred significantly more frequently in patients with stent-graft surface movement.
EVAR should not be considered to be a single-step procedure, based on preoperative anatomical features alone. During pre-procedure planning, one should be aware of the ongoing process of conformational geometrical changes of the stentgraft and the diseased aorta [30,31]. In this context, an intact proximal sealing zone is of crucial importance for stent-graft stability and integrity, preventing distal migration, stent-graftrelated endoleaks, and, ultimately, aortic rupture. However, detection of these subtle changes on regular CTA images is difficult, even with centerline reconstructions on a vascular workstation [16,19]. A recent study by Schuurmann et al [19], using a semiautomatic software, observed significant changes in proximal stent-graft dimension and apposition on regular CTA scans prior to failure, when patients with and without later stent-graft migration were compared. This observation was reflected in our results, where subtle changes of the stent-graft surface could be detected in 28.9% of patients using a custom-made semi-automatic software. Consistently, migration and srEL were significantly more frequent in patients with stent-graft surface movement.
The incidence of stent-graft migration in the present study was 15.0%, which is in line with percentages (1.1-28%) previously reported [2][3][4][5]. Migration bears the risk of sealing loss followed by late stent-graft-related endoleaks, which were observed in 3.7% of our patients. This incidence is lower compared with the percentages (3.9-10.6%) previously reported [32][33][34], which may be explained by the exclusion of patients with early srEL and re-intervention within 6 months.
Different anatomical aspects of proximal neck morphology were found to be associated with the incidence of migration. In this context, the aortic neck diameter was found to be a risk factor for stent-graft migration [3,4,35], which is in accordance with our results where an increasing neck diameter was  Fig. 4 a-d 83-year-old male patient with evident angulation and wide diameter of the infrarenal neck on pre-procedural CTA-maximum intensity-projection images (a). In comparison to the pre-discharge CTA (b), there is an obvious lateralization of the superior mesenteric artery and the abdominal aorta with consecutive compression of the inferior vena cava after 4 years of follow-up (c). Further, the aortic neck diameter increased from 29 to 36 mm with consecutive sealing loss at the dorsal circumference of the proximal anchoring zone and occurrence of a type 1a endoleak (arrow). Significant stent-graft surface movement (> 9 mm) at the proximal anchoring zone is visualized on 3D matched surface models of postprocedural and final follow-up CT (d) related to stent-graft surface movement in the multivariate analysis. Shorter neck length is a well-known risk factor for migration [36,37], which is in line with our results, demonstrating the protective effect of longer neck lengths in terms of stent-graft surface movement. Another feature of hostile neck anatomy is angulation, which decreases the necessary pulldown force causally responsible for dislodgement of the stent-graft [38]. Consequently, neck angulation was found to be a risk factor for early migration [32] and type 1 endoleaks [39]. This mechanism is supported by our results, where neck angulation was found to be significantly associated with stentgraft surface movement in the multivariate analysis. Displacement forces and stent-graft stability are also related to the iliac fixation [14,37]. Through distal fixation, different authors have identified the docking area of the modular stent graft limb as a vulnerable region, which seems to be exposed to the highest mechanical stress [31] and surface movement [15]. Taking into account that the modular limb was predominantly left-sided (78%) in our cohort, the left iliac sealing diameter was significantly associated with an increased stent-graft surface movement in the univariate, but not in the multivariate, analysis. Another mechanism with a negative influence on wall compliance and aneurysm morphology is atherosclerotic disease progression [40], which particularly affects elderly patients [41]. Several authors reported an association between early stent-graft complications, including migration [42,43], and increasing age. These findings were supported by our data, where increasing age was significantly associated with stent-graft surface movement in the multivariate analysis.
Different approaches were made to streamline the surveillance protocol proposed by the Society of Vascular Surgery [18], designed to reduce radiation exposure and the total costs of EVAR. As outlined by Hoel and Schanzer [44,45], there should be a balance between adequate disease detection and unnecessary use of resources in postoperative surveillance. Knowledge of the predisposing factors associated with the occurrence of stent-graft surface movement may help to identify patients who benefit from regular CTA surveillance. Detection of subtle changes of the stent-graft position in these patients may allow for planning of timely prophylactic reinterventions to prevent stent-graft migration and consecutive sealing loss [19].

Limitations
Our study has several limitations. First, data were evaluated in a retrospective manner and we had to exclude 50 patients with early re-interventions for stent-graft failure, rendering the assessment of stent-graft surface movement impossible. Since we have no information on possible late complications of these patients, we cannot completely rule out a possible selection bias. Second, inclusion of second-generation devices and rudimentary planning in the early period of our study may have negatively affected the outcome in certain patients. Third, comparing the pre-discharge and the last CTA followup resulted in a heterogeneous time interval for measurements. As a consequence, the study design does not allow a conclusion about the time point for the occurrence of stent-graft surface movement and migration, nor on its chronological sequence. Fourth, due to the relatively small sample size and heterogeneity by means of implanted devices, we are not able to draw a reasonable statement on the performance of different devices.

Conclusion
Older patients, those with larger aneurysm neck diameters, and those with increased infrarenal aortic neck angulation experienced higher rates of stent-graft surface movement. Consideration of these risk factors may help to identify patients who benefit from possible use of adjunctive, prophylactic sealing systems and a regular CTA surveillance or even open repair as a primary intervention. Detection of stentgraft surface movement during CTA surveillance may allow for planning of timely prophylactic re-interventions to prevent stent-graft migration and stent-graft-related endoleaks.
Funding Information Open access funding provided by Medical University of Vienna.

Compliance with ethical standards
Guarantor The scientific guarantor of this publication is Richard Nolz.

Conflict of interest
The authors declare that they have no competing interests.
Statistics and biometry No complex statistical methods were necessary for this paper.
As suggested by the reviewers, an expert in statistics was consulted for revision of the statistics section.
Pascal Baltzer, MD Medical University of Vienna, Austria, Department of Bio-medical Imaging and Image-guided Therapy, Division of Cardiovascular and Interventional Radiology.
Pascal Baltzer, MD was added as one of the authors. Ethical approval Institutional Review Board approval was obtained.
Study subjects or cohorts overlap Some study subjects or cohorts have been previously reported in the European Journal of Vascular and Endovascular Surgery by Nolz et al [15,46].

Methodology
• retrospective • cross-sectional study • performed at one institution This retrospective observational study followed the checklist of items as published in the STROBE guidelines.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.