Abstract
Objectives
The results of branched endovascular repair of thoracoabdominal aneurysms are mainly dependent on durability of the graft used. The purpose of this study was to evaluate postoperative aortic main body and bridging stent-graft remodeling, and their impact on bridging stent-graft instability at one year.
Methods
Computed tomoangiographies of 43 patients (43 aortic main body mated with 171 bridging stent-grafts) were analyzed before and after branched endovascular repair as well as after a follow-up of 12 months. Primary endpoint was aortic main body remodeling (migration >5 mm, shortening >5 mm, scoliosis >5° or lordosis >5°). Shortening was defined as a reduced length in the long axis, scoliosis as left-right curvature, and lordosis as antero-posterior curvature. Aortic main body remodeling, aneurysm sac changes, and bridging stent-graft tortuosity were evaluated to study their correlations and the impact on the bridging stent-graft instability.
Results
At 12 months, aortic main body remodeling was observed in 72% of the cases, migration in 39.5% (mean 5.21 mm), shortening in 41.9% (mean 5.79 mm), scoliosis in 58.1%, (mean 10.10°), lordosis in 44.2% (mean 5.78°). Migration, shortening, and scoliosis were more frequent in patients with larger aneurysms (p = .005), while scoliosis was significantly more frequent in type II thoracoabdominal aneurysm (p = .019). Aortic main body remodeling was significantly associated to bridging stent-graft remodeling (r: 0.3–0.48). The bridging stent-graft instability rate was 9.3%. Despite a trend toward significance (p = .07), none of the evaluated aortic main body and bridging stent-graft changes were associated with bridging stent-graft instability at 12 months.
Conclusions
Aortic main body remodeling is frequent especially in large and extended thoracoabdominal aneurysm aneurysms. Aortic main body and bridging stent-graft remodeling was significantly correlated. While these geometric changes had no significant impact on bridging stent-graft instability at one year, a close long-term follow-up after branched endovascular repair could predict bridging stent-graft failures.
Keywords
Introduction
Endovascular treatment of thoracoabdominal aortic aneurysms (TAAA) with branched or fenestrated endovascular aortic repair (BEVAR or FEVAR, respectively) is a less invasive alternative to open repair with acceptable results.1–3 In both endovascular techniques, an aortic main graft (AMB) is mated with bridging stent grafts (BSGs) to achieve aneurysm exclusion while maintaining target vessels (TVs) perfusion.
Due to anatomic variations, many branched devices need to be manufactured based on patient-specific anatomy. Manufacturing of these custom-made devices however needs time. Solutions such as off-the-shelf devices (e.g. T-branch®, Cook Medical, Bloomington, IN, USA) were introduced to offer branched endovascular repair also in urgent cases.4–6
The clinical outcome of BEVAR and FEVAR is mainly affected by bridging stent-graft complications such as occlusions, stenoses, and endoleaks, which are also coined as bridging stent instability and can require reinterventions.7–9 It still remains unclear which factors drive bridging stent-graft instability. One interesting observation of Mastracci et al. was that BSG instability was more frequent in BEVAR. 10
Severe tortuosity, evolution of aortic disease, and graft material deterioration could be significant predictors for branch occlusion. A better understanding of how AMB and BSG remodeling over time could prevent the occurrence of complications by timely reintervention and suggest which technology implementation is needed. 11 , 12
At present, studies evaluating the behavior of devices remodeling over time in terms of structural three-dimensional changes are scarce. The aim of this study was to determine geometric changes of both AMBs and BSGs after BEVAR and their clinical impact on bridging stent grafts instability.
Methods
Study design
The data of consecutive patients treated with BEVAR from January 2017 to June 2018 for TAAA at a single institution were prospectively entered into a dedicated database and retrospectively evaluated. Patients who received computed tomography angiographies (CTAs) preoperatively, postoperatively (within one month), and at 12 months were included in the study. All included CTAs were high-resolution CTs (0.75–1 mm slice thickness) and performed with the same protocol at our institution. For the analysis, the arterial phase acquired in bolus tracking technique with 120 ml Ultravist 350 (Bayer Vital, Leverkusen, Germany) at 4 ml/s, at the end of deep expiration was utilized.
Both elective and high-risk patients were treated with the following indications: maximum aneurysm diameter >60 mm in men and >55 mm in women, >5 mm aneurysm growth during a six-month period, presence of a saccular aneurysm or an anastomotic false aneurysm. Symptomatic patients containing rupture or aneurysms >80 mm were considered urgent cases. Post dissection aneurysms were not included in this study, as a small true lumen was considered a potential bias for AMB structural changes. All demographics are included in Table 1. All patients had provided written informed consent for the inclusion of their data in the database. The study was performed in accordance with the rules of the ethical review board of our institution.
Patients’ characteristics.
ASA: American Society of Anesthesiologists; TAAA: thoracoabdominal aortic aneurysm.
Devices
All thoraco-abdominal branched endografts (AMB) were based on the Zenith stent-graft platform (Cook Medical) and generally contained four cuffs. Two types of side-branched devices were used: a custom-made-device and an off-the-shelf side-branched device (T-Branch®, Cook Medical). Depending on the clinical conditions, custom-made devices were used for elective cases, whereas T-Branch was mainly used for urgent indications as previously described. 4 All AMBs were manufactured with caudally directed cuffs. Thoracic aortic endografts were used for proximal grafting in case of type II and III TAAA.
The thoraco-abdominal side-branched devices were connected with the target vessels (TVs) using balloon-expandable covered stents as main bridging stent-graft (BSG), distally extended with self-expanding bare stents in case of angulated anatomy or to increase the distal landing zone maintaining the patency of side-branches. The most used covered stents were Advanta V12-iCast (Getinge Maquet, Rastatt, Germany) mated with Viabahn (W.L. Gore & Associates, Flagstaff, AZ, USA) and the Viabahn-VBX (W.L. Gore & Associates). The most used self-expanding bare metal stent was the SmartFlex (Cardinal Health, Dublin, OH, USA) as distal BSG mainly for visceral arteries.
Sizing and procedural technique
The graft plan was composed and reviewed by an experienced vascular surgeon. The AMB oversizing rate relative to the native aorta was about 15–20%, while the BSGs were oversized 5–10% based on the ideal planned distal landing zone. All interventions were performed in a hybrid room (Axiom Artis FA; Siemens Healthineers, Forchheim, Germany) in general anesthesia. A percutaneous femoral access was used for the main endografts and eventually contralateral iliac limb implantations. Surgical cut-down of the left axillary artery was used for target vessel cannulation and deployment of bridging stents. Since 2015 a new spinal cord ischemia prevention protocol was in place, abandoning preoperative cerebrospinal fluid drainage as a preemptive measure. This approach required an early blood flow restoration immediately after main grafts implantation by removing the sheaths and closing the percutaneous femoral access. 13 , 14 This technique aims at restoring the arterial perfusion to the pelvic and peripheral district as soon as possible. During target vessel cannulation, bridging stents delivery and implantations, the activated clotting time was maintained between 250 and 300 s.
For elective type II and III TAAA repair, a staged approach is followed. The thoracic aorta is usually treated first, followed by implantation of the abdominal device six to eight weeks later. In urgent type II and III TAAA repair, the “open branch” technique was performed, bridging the celiac trunk cuff with a covered stent at least two weeks after EVAR. 15 A single-stage treatment was used for type IV TAAA bridging all target vessels with covered stents.
Other details of procedural technique and postoperative surveillance were previously published. 5 , 7
Imaging, measurements, and definitions
The preoperative CTA was used to define the extension of TAAA using the Crawford classification and to measure the preoperative maximum aneurysm sac diameter in the centerline. Sac change, AMB, and BSG remodeling were measured comparing the postoperative and follow-up CTAs, and delta values were considered. BSG remodeling was measured by branch tortuosity index (BTI), including both the bridging device and the distal native vessel (up to main bifurcation and/or collateral).
Bridging stent-graft instability was defined as any BSG-related reintervention or BSG occlusion without reintervention. 16
Aortic main body (AMB) remodeling included four different types of three-dimensional alterations of the stent-graft: migration as a craniocaudal movement, shortening as a reduction of craniocaudal stent-graft length, scoliosis meaning an increasing lateral angulation and lordosis representing increasing antero-posterior angulation.
AMB remodeling was defined as at least one AMB alteration ≥5 mm or ≥5°.
Migration
Migration was defined as the AMB displacement measuring the caudal displacement relative to an adjacent vertebral body bone landmark. It was measured by the distance between two parallel lines touching the upper gold marker of the most proximal cuff (generally for celiac trunk) and the upper edge of the more distal thoracic vertebra (T12) in sagittal view. It was measured in millimeters (Figure 1).

AMB migration. Sagittal view of maximum intensity projection (MIP) including the branched portion at postoperative CTA (a). Twin window measuring a migration of 13.2 mm at one year CTA follow-up (b). It was measured by the distance between two parallel lines touching the upper gold marker of the most proximal cuff (arrow, celiac trunk) and the upper edge of the more distal thoracic vertebra (T12, dotted line).
Shortening
Shortening was defined as the change in AMB length in its long (i.e. craniocaudal) axis. It was measured by the distance between two parallel lines touching the proximal and distal extremity of AMB in coronal view. It was measured in millimeters (Figure 2).

AMB shortening. Coronal view of maximum intensity projection (MIP) including the whole AMB at postoperative CTA (a). Twin window measuring a shortening of 16.6 mm at one year CTA follow-up (b). It was measured by the distance (arrow) between two parallel lines touching the proximal and distal extremity of AMB.
Scoliosis
Scoliosis was defined as AMB left-right curvature in the coronal plane. It was measured by the higher longitudinal angle of AMB branched portion in coronal view, based on the ideal CLL. It was measured analogous to Cobb’s angle used for vertebral scoliosis in degrees (Figure 3).

AMB scoliosis. Coronal view of maximum intensity projection (MIP) including the branched portion at postoperative CTA (a). Twin window measuring a scoliosis of 40.1° at one year CTA follow-up (b). It was measured by the higher longitudinal angle of AMB branched portion in coronal view, based on the ideal CLL.
Lordosis
Lordosis was defined as the antero-posterior AMB curvature. It was measured by the higher longitudinal angle of AMB branched portion in sagittal view, based on the ideal CLL. It was measured analogous to Cobb’s angle in degrees (Figure 4).

AMB lordosis. Sagittal view of maximum intensity projection (MIP) including the branched portion at postoperative CTA (a). Twin window measuring a lordosis of 16° at one year CTA follow-up (b). It was measured by the higher longitudinal angle of AMB branched portion in sagittal view, based on the ideal CLL.
Sac change
Sac change was defined as the delta value of the largest aortic diameter measured perpendicular to the aortic CLL. It was measured in millimeters.
Branch tortuosity index
Branch tortuosity index (BTI) was defined as the tortuosity index of each branch, including the BSG and native vessel. It was measured by the ratio between the actual length of bridging target vessel along the CLL and the shortest distance between the beginning of the stent and the end of the vessel. It is a ratio, e.g. Branch (curved length)/Branch (straight length) = TI; 82 mm/57 mm = 1.43.
All CTAs were evaluated using Aquarius iNtuition software (version 4.4.11; TeraRecon, Foster City, Calif) and Jivex Diagnostic Client (version 5.1.0.19, VISUS Health IT GmbH, Germany).
All measurements and CLLs were manually performed and adjusted. For a subset of 10 patients (23.3%), all measurements were performed by two readers (S.F. and R.M.) to assess inter-observer reliability.
Statistical analysis
Continuous variables are presented as the mean ± standard deviation or median and interquartile range, depending on normal or non-normal distribution. Correlations between variables were evaluated with Pearson’s correlation coefficients. Comparison of different subgroups was performed using the Mann–Whitney U test. Values of p < .05 were considered to be statistically significant. Inter-observer variability was tested with the intraclass correlation coefficient (ICC). An ICC > 0.9 was required for each observer to begin with the study measurements. The statistical tests were performed using SPSS, version 25 (IBM Corp, Armonk, NY).
Results
Aortic main body characteristics
Forty-three patients (43 AMBs mated with 171 BSGs) met the inclusion criteria. Thirty-nine (90%) AMBs were 4-branches configuration; the remaining were 2 > 4 and 2 < 4 cuffs, respectively. AMBs were 31 (72%) custom-made devices and 12 (28%) T-branch, respectively. Five of 176 antegrade cuffs have been closed by plug placement.
Primary outcome
Aortic main body remodeling at one year occurred in 72% (31/43) of the cases. It was not significantly related to any different AMB configuration (CMD, T-branch, number of branches).
Migration
The mean AMB migration was 5.21 ± 4.71 mm. A migration ≥5 mm was observed in 39.5% and ≥10 mm in 16.2% of the cases, respectively. The migration was strongly related to shortening (Pearson correlation: r = .7, p < . 00001) and scoliosis (r = .72, p < .00001). Furthermore, it was significantly more often observed in patients with aneurysm sac larger than 65 mm (p = .005). Table 2 describes subgroup analysis.
AMB remodeling and aneurysm sac change (subgroups analysis).
aII + III type TAAA were compared with IV type TAAA.
bDichotomized at the median value.
Note: Boldface values indicate statistical significance (p < .05).
Shortening
The mean AMB columnar shortening was 5.79 ± 5.40 mm. A shortening ≥5 mm was observed in 41.9% and ≥10 mm in 18.6% of the cases, respectively. The shortening was strongly related to scoliosis (r = .78, p < .00001) and significantly more often observed in patients with aneurysm sac larger than 65 mm (p = .005).
Scoliosis
The mean AMB scoliosis change was 10.10° ± 9.28. A scoliosis ≥5° was observed in 58.1% and ≥10° in 37.2% of the cases, respectively. Scoliosis was significantly more pronounced in larger (>65 mm diameter, p = .002) and more extensive aneurysms (type II, p = .019).
Lordosis
The mean AMB lordosis change was 5.78 ± 4.53°. A lordosis ≥5° was observed in 44.2% and ≥10° in 14% of the cases, respectively. No statistically significant correlations were observed for this factor. In the postoperative CTA, the AMB curvature (≥5°) was anterior in 26 (60%) cases or posterior in 11 (26%) cases, respectively. In the follow-up CTA, the grade of lordosis decreased in 19 (44%) cases.
Sac change
The mean sac change was 5.48 ± 5.17 mm. At one-year follow-up, the aneurysm size remained stable, showed shrinkage (≤5 mm) or expansion (≥5 mm) in 55.8%, 37.2%, and 7% of the cases, respectively. The mean maximum sac diameter was 65 ± 10 mm postoperatively and 61 ± 12 mm at one year (p = .023). Sac changes were significantly greater in larger aneurysms (p = .024).
BSG remodeling
The branch tortuosity index (BTI) was higher in renal arteries compared to visceral arteries (1.46 vs. 1.11) within 30 days. The tortuosity increase at one year was significantly more pronounced in the renal arteries (+7%) than in the visceral arteries (+2.5%) (p ≤ 0.0001). The mean BTI for post-operative vs. one year later were: celiac trunk 1.16 vs. 1.19, superior mesenteric artery 1.08 vs. 1.08, right renal artery 1.49 vs. 1.57, left renal artery 1.42 vs. 1.44. None of these changes were significant. At one-year, overall BSG remodeling was moderately related to migration and scoliosis (r = .43, p < .00001; r = .48, p < .00001, respectively), and mildly related to shortening (r = .3, p = . 00008).
Remodeling and BSG instability
The overall BSG remodeling showed a trend toward a significant correlation with BSG instability (mean/patient: +17.3% vs. +30.3% with instability, p = .07).
Most of the AMB remodeling phenomena did not significantly increase in case of BSG stenosis/occlusions compared to type IC endoleaks. Contrariwise, sac change was significantly related to type IC endoleaks (p ≤ .003). In particular, shrinkage significantly increased in patients treated for type IC endoleaks (10.3 vs. 4 mm, p = .009). No significant differences were found between subgroups of different BSGs in term of remodeling and instability.
Interobserver agreement
For a random sample of 10 patients, repeat measurements were performed by a blinded observer. The intraclass correlation coefficient was 0.973 (95% CI: 0.958–0.983), revealing consistent interobserver agreement.
Patients’ demographics and main clinical results
Forty-three patients with a mean age of 70 ± 6.71 years were included. The mean clinical follow-up was 16.2 months. Technical success was achieved in all cases. Mortality during the follow-up was 4.6%; two patients died at 19 and 21 months due to non TAAA-related reasons.
The GFR decreased during follow-up (pre-operative vs. last GFR was 83 ± 18 vs. 74 ± 19 mL/min/1.73 m2 (p = .02). Six cases (14%) with postoperative signs of SCI with mild neurologic impairment (SCI Greenberg Classification type 4) were reported pre-discharge, of which four fully regressed. No patients suffered from bowel ischemia during follow-up.
Two (4.6%) reinterventions occurred at the level of aortic components (one proximal disease degeneration and one type III endoleak). Except lordosis, AMB remodeling and sac change were higher in these cases (p=ns). Eleven cases (25.5%) of type II endoleak and five of type IB endoleak (all treated for secondary procedures) were detected during follow-up. The sac expansion (≥5 mm) occurred in three cases, one with type III and two with type IB endoleak, respectively. Aortic-related reinterventions and endoleaks (II, III, IB) were not statistically associated to BSG instability.
BSG instability
Sixteen cases of BSG instability were detected (overall BSG instability 9.3%, visceral BSG 5.2% vs. renal BSG 4.1%, n.s.). The BSG reintervention and occlusion rate were 8.2% and 1.1%, respectively (see Table 3). Two cases with multiple BSG instability were observed (triple distal edge stenosis; triple stenosis and one occlusion). Fourteen BSGs were treated with secondary procedures, bar the two BSG asymptomatic occlusions (one celiac trunk and one renal artery).
Results: BSG instability and secondary procedures.
BSG: bridging stent graft; EL: endoleak; CT: celiac trunk; LRA: left renal artery; SMA: superior mesenteric artery.
All type of IC endoleaks were detected at postoperative CTA and subsequently treated with distal relining. No cases of BSG fractures and/or disconnections were detected.
BSG instability occurred in four Advanta, four Advanta/Viabahn, and eight VBX, respectively.
Discussion
To our knowledge, this is the first study focusing on multidimensional geometric changes of the aortic stent-graft after BEVAR. In the current literature, there are numerous works analyzing geometric changes after BEVAR and FEVAR. These studies, however, focus on the changes of the bridging stent-grafts. 12 The underlying hypothesis of this study was that bridging stent instability originates in geometrical changes of the aortic stent-graft.
The central finding of our study is that aortic stent-grafts are subject not only to migration, but also to three-dimensional deformation. Especially in case of type II thoracoabdominal aortic aneurysms, the long segments of stented aorta tended to shorten by more than 9 mm in one year and develop significant scoliosis. This phenomenon has not been described before, but it can be hypothesized that a shortening of the aortic stent-graft might have similar effects as migration, as the branch origins shift their positions with the aortic stent-graft in either case.
England et al. analyzed the degree and effect of migration in fenestrated endografts in 154 patients. 17 The authors demonstrated in this multicenter study that in 21% of the patients, a migration of > 6 mm was evident, mainly diagnosed in the first two years after implantation. The same group published single center results of 55 patients, reporting 10 migrations > 4 mm of which 5 had BSG failure. 18 In this study, migration occurred 11.3–84.7 months post procedure. It is not known whether these results are transferrable to BEVAR or whether other forms of aortic stent-graft remodeling might affect BSG instability. In the recent geometric analysis by de Niet et al., migration leads to BSG failure in two out of three cases, one after BEVAR and one after FEVAR. 12
Another phenomenon described in our study is the deviation of the stent-graft from its longitudinal axis, forming lateral and ventral bends. Alluding to the vertebral column, we used the terms scoliosis and lordosis for these alterations in stent row alignment. Until now, aortic stent-graft scoliosis meaning lateral kinking has not been reported in the literature. A mean scoliosis change of 10° was observed in this study. This became especially pronounced in type II TAAA repair with a mean scoliosis change of 15° vs. 5.4° in type IV repair. Large aortic aneurysms were also associated with a greater degree of scoliosis with a mean scoliosis change of 14° in aneurysms > 65 mm vs. 8° in aneurysms < 65 mm. Thus, it can be hypothesized that with greater aneurysm diameter and extension, the aortic stent-graft has more space to change its shape. In this context, all endografts had the same configuration with outer branches and tapered portion at this level; this segment could represent a weakness point in terms of columnar strength, conforming the stent struts to the anatomy.
Whether an increasing scoliosis will lead to increased kinking of the branches in the long run is unclear. In the present patient cohort, migration and scoliosis were significantly associated with an increased branch remodeling. It is well-known that renal branches with severe tortuosity and angulation are prone to BSG instability and renal failure. 11 Recently, the risk of BSG tortuosity after F-BEVAR was evaluated in a single-center study comparing post-operative CTA and BSG failures; the authors concluded that a higher angle of curvature leads to a higher hemodynamic force that results in a higher rate of thrombosis. 19
In the current study, AMB and/or BSG remodeling did not significantly affect bridging stent instability. However, a trend toward significance (p = .07) was found and the number of bridging stent complications at one year was not high enough for statistical analysis. In a recent work on BSG-related endoleaks in fenestrated and branched EVAR, Kärkkäinen et al. conclude that late BSG-related endoleaks occur as a result of aortic stent-graft migration, sac remodeling, and stent displacement. 20 Therefore, a close long-term follow-up after BEVAR is recommended in order to prevent BSG failures.
Limitations
This study has several limitations. Being a single-center report on a relatively small number of patients, an evaluation of a larger cohort is mandatory before any general conclusions may be drawn. The patients included were all treated with a single aortic stent-graft platform, rendering the results not applicable to other platforms. On the other hand, the heterogeneity among patients thus is limited. This report lacks external validation. We have sought to offset this by employing inter-observer tests. Furthermore, the measurement technique used is simple to reproduce and no specific software was needed to perform the measurements.
Another possible limitation not taken into account is anatomic changes like vertebral body fractures or scoliosis. Furthermore, cardiac and respiratory cycles are affecting the aorta and its branches. During inspiration and expiration, the angles of renal arteries do vary. 21 The CTAs were obtained during expiration, and therefore the angles during inspiration were not recorded. Also, the CTAs were not ECG-gated. Aortic expansion during systole is however not relevant in the stented segment.
Conclusions
Branched aortic endografts undergo three-dimensional geometric changes over time. In this study, a remodeling of the aortic stent-graft including shortening, scoliosis, and lordosis are demonstrated. Especially in extensive aortic disease and large aneurysm sacs, these alterations are remarkable.
A significant association of AMB remodeling phenomena and a correlation between AMB and BSG remodeling have been demonstrated. While these geometric changes had no significant clinical impact at one year, it seems prudent to assess these changes in mid- and long-term follow-up to identify potential risk factors for bridging stent-graft failure.
Footnotes
Acknowledgement
Presented as abstract at the VEITH Symposium, New York, USA, 19–23 November 2019.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
