Thrombotic complications after aortic arch replacement with frozen elephant trunk stent-graft: A 10-year United Kingdom institutional experience

Abstract

Objective

Postoperative intraluminal thrombosis after frozen elephant trunk replacement has been reported to occur with a frequency of 6% to 17% and is associated with poor outcomes. The purpose of this institutional review is to analyze thrombosis rate, predisposing patient and operative factors, and assess different anticoagulation regimens.

Methods

This retrospective cohort study includes 174 patients operated on over 10 years. one hundred forty-five of these underwent elective aortic arch replacement; 29 had the procedure for Type A aortic dissection repair.

Results

Sixteen elective (11%) and 3 dissection patients (10%) had radiographic evidence of intraluminal thrombus. There were no statistical differences in demographic or intraoperative characteristics between the 2 groups. Of the 16 elective patients with thrombus, 12 (75%) had aneurysmal disease. Central graft position is associated with a higher incidence of intraluminal thrombus formation than eccentric position in both cohorts, 17% versus 7% in elective patients and 15% versus 0% in the dissection group. Patients with intraluminal thrombosis had significantly lower 6-month survival in both cohorts (69% vs 92% and 66% vs 88%; P = .0037) and this was also true for the elective group (69% vs 96%; P = .0001). Of several anticoagulation regimens employed over the study period, early introduction of warfarin proved superior.

Conclusions

The incidence of thrombus formation is higher in patients with aneurysmal disease and when the graft is positioned centrally. Early anticoagulation with warfarin appears to be protective. We advocate the creation of a registry to help improve outcomes after this complex surgery.

Intraluminal thrombus 5 days after FET aortic arch repair.

Central Message.

Intraluminal thrombosis after frozen elephant trunk replacement of the aortic arch is a common and serious complication. Appropriate and early anticoagulation may help reduce the incidence.

Perspective.

In keeping with previous publications, our institutional experience shows an 11% incidence of intraluminal thrombosis after frozen elephant trunk surgery. What constitutes appropriate anticoagulation is subject to debate and pooled institutional or registry data are likely to be required to generate the evidence needed to guide therapy.

Since its introduction 15 years ago, frozen elephant trunk (FET) replacement of the aortic arch has become a common procedure for thoracic aortic pathologies. The indications for FET range from ascending aorta aneurysm, multiple aneurysmal pathology, and penetrating aortic ulcers to acute or chronic aortic dissections. As experience with this technique grows globally, more potential complications and postoperative sequelae are finding their way into the medical literature. Intraluminal thrombosis (ILT) has been described by several groups.1, 2, 3, 4 Due to a described incidence of up to nearly 17% and because of the potentially life-threatening consequences, ILT is a serious complication that warrants further investigation.

Although to date there is a paucity of evidence, the currently available literature indicates 2 categories of risk factors.

• •Patient-related, such as advanced age, female sex,⁴ relatively large aorta size compared with body⁵ (ie, aortic size index or ratio of aortic size to body surface area⁶), or autoimmune disease⁷; and
• •Operation-related, such as size of intraluminal stent,³ ratio of stent to aortic diameter,⁵ or central versus eccentric graft position.¹

Details of how to manage this complication are still being established and reviews of institutional experience are currently the best available evidence.

ILT carries a significant mortality burden, up to 30%, and a high morbidity burden, such as embolic events, in up to 27% of patients.¹ To date, there has been no review of prophylactic anticoagulation after FET placement. The purpose of this article is to compare the experience in our institution with the available literature scrutinize different postoperative anticoagulation regimens and their success in preventing ILT.

Materials and Methods

The study was approved by the Royal Papworth Hospital Institutional Review Board (reference Non-HRA0025; January 4, 2023), which waived the need for informed written consent due to its retrospective nature. The study was designed utilizing the Strengthening the Reporting of Observational Studies in Epidemiology checklist for observational studies.⁸

We conducted a single-center, retrospective, cohort study of adult patients undergoing elective aortic arch replacement with an FET at an academic, quaternary referral hospital in the United Kingdom over a 10-year period from January 2014 to December 2023. Patients who died before the first postoperative imaging study were excluded from the analysis. Anesthesia and intensive care data were retrieved from the electronic monitoring system (Metavision), whereas surgical, demographic and outcome data were retrieved from the inhouse patient database (CARDS II). All patient imaging was reviewed on PACS (Insight Zero; Intelerad Medical Systems).

Two hundred fourteen cases were screened, 185 elective and 29 acute type A dissections (TADs). Of these we excluded 40: 24 due to early mortality and no available imaging and 16 due to incomplete data. All available postoperative imaging of the included final 174 patients was independently reviewed by a specialist cardiothoracic radiologist, a cardiac surgeon, and a cardiac anesthesiologist.

We predominantly use Thoraflex hybrid plexus grafts (Terumo Aortic), although we have some experience with the E-vita Open Neo hybrid stent-graft system (Artivion Inc). In the elective patient group, we used 133 Thoraflex and 12 E-vita Neo grafts, whereas all TAD patients received Thoraflex stent-grafts when required. Most of the procedures were performed under moderate hypothermic arrest, aiming for a core temperature of 25 to 28 °C. Our current surgical protocol, which was used for the majority of the cases in this series, involves bilateral axillary artery and central venous cannulation and fashioning an extra-anatomic bypass of the left subclavian artery and direct anastomosis of the other arch branches. Most of the FETs were deployed in aortic zone 0 or 1.

The following formula was used was used for sizing the graft:

$\raisebox{1ex}{$Longdiameteroftruelumen+shortdiameteroftruelumen$}\!\left/ \!\raisebox{-1ex}{$2$}\right.+$ 10%

In cases of aortic aneurysm, the FET graft was oversized by 10% where possible. Alternatively, the largest diameter available FET device was deployed into the aneurysmal distal aorta. The length of the FET was 100 or 120 mm in most cases. The longer FET devices were used when needed to cover a distal aortic tear.

Radiological evidence of thrombotic material within the lumen of the stent-graft was chosen as a primary outcome measure. Categorical differences between the groups were evaluated using the χ² or Fisher exact test, as appropriate. Continuous data were tested using the Shapiro-Wilk test of normality and analyzed using a paired Student t test for parametric data and the paired samples Wilcoxon test for nonparametric data. Statistical analyses were performed using Medcalc software (Medcalc Software Ltd) and a P value < .05 was considered statistically significant.

Results

A total of 10.9% of the included patients developed ILT: 11% in the elective group and 10.3% in the TAD group. There was no statistically significant difference in patient demographics, risk score. or intraoperative data between ILT and non-ILT patients in the elective or the emergency group. There was also no difference in postoperative blood loss, postoperative interval before anticoagulation therapy was started, or ICU length of stay. Hospital length of stay in the group who experienced ILT was longer, although this did not reach significance. There was no significant difference in the rate of clinically apparent thrombotic events; however, mortality at 30 days and at 6 months postoperatively was significantly higher in the group that had radiological evidence of thrombus (Tables 1 and 2).

Table 1.

Demographic, intra- and postoperative variables in elective patients

	Thrombus no (n = 129)	Thrombus yes (n = 16)	P value
Age	68 (67-70)	72 (63-75)	.2
Male sex	83 (64)	8 (50)	.26
Height	171 (170-175)	171.5 (159-176)	.27
Weight	83 (80-85.3)	81.5 (62-90)	.38
EuroSCORE	17.2 (15.1-20.1)	19.6 (12.6-32.7)	.35
Redo aortic operation	30 (23)	3 (19)	.68
Emergency	61 (47)	7 (44)	.79
CPB (min)	248 (238-253)	230 (201-290)	.63
Crossclamp (min)	95 (76-119)	104 (54-150)	.77
12-h blood loss (mL)	315 (275-361)	290 (231-514)	.91
Anticoagulant started postoperative d	2 (1.8-2.8)	1.8 (1.7-2.0)	.18
ICU LOS (d)	5 (4-7)	5 (2.6-14.1)	.87
Hospital LOS (d)	16 (14-18.8)	21 (14.3-26.8)	.2
Other thromboembolic events	14 (11)	2 (12.5)	.69
Alive at 30 d	125 (96)	11 (69)	.0001
Alive at 6 mo	119 (92)	11 (69)	.0037

Values are presented as n (%) or median (interquartile range). EuroSCORE, European System for Cardiac Operative Risk Evaluation; CPB, cardiopulmonary bypass; ICU, intensive care unit; LOS, length of stay.

Table 2.

Demographic, intra- and postoperative variables in type A dissection patients

	Thrombus no (n = 26)	Thrombus yes (n = 3)	P Value
Age	65 (56-70)	74	.14
Male sex	19 (73)	2 (66)	.82
Height	172 (170-176)	170	.59
Weight	85 (79-90)	80	.62
Redo aortic operation	0	0
CPB (min)	279 (257-301)	248	.89
Crossclamp (min)	117 (71-166)	106	.94
12-h blood loss (mL)	350 (264-494)	275	.69
Anticoagulant started postoperative d	2.7 (1.8-3.8)	2.5	.38
ICU LOS (d)	6 (4-10)	4	.97
Hospital LOS (d)	17 (15-20.3)	19	.7
Other thromboembolic events	0	0	0
Alive at 30 d	24 (92)	3 (100)	.63
Alive at 6 mo	23 (88)	2 (66)	.31

Values are presented as n (%) or median (interquartile range). EuroSCORE, European System for Cardiac Operative Risk Evaluation; CPB, cardiopulmonary bypass; ICU, intensive care unit; LOS, length of stay.

The highest incidence of ILT related to type of pathology in elective patients was found in aortic arch aneurysm repair with an incidence of 14%. We found that 75% of patients who developed thrombus had aneurysmal disease. Other pathologies had an 8% to 10% risk of thrombus; however, there was no statistical significance associated with these findings. Notably, there was a higher incidence of thrombus presence in central FET positioning (ie, central position of the stent in the native aorta [16% vs 7%], which was also found in TAD patients (17% vs 0%) (Tables 3 and 4). A total of 13% of patients receiving a Thoraflex graft developed ITL compared with 17% who received an E-vita Neo graft.

Table 3.

Incidence of intraluminal thrombus depending on pathology, graft position, and anticoagulation regimen in elective patients

Indication	Thrombus no (n = 129)	Thrombus yes (n = 16)	P value
Aortic arch aneurysm	54 (86)	9 (14)	.89
Dilated aortic arch	31 (91)	3 (9)
Non-A non-B dissection	5 (100)	0
Previous type A dissection	27 (90)	3 (10)
Type B dissection	12 (91)	1 (8)
FET position
Eccentric FET	85 (93)	7 (7)	.08
Central FET	44 (83)	9 (17)
Anticoagulation
Enoxaparin	67	10
Tinzaparin	48	2
Warfarin	9	0
Aspirin	3	1
Aspirin + clopidogrel	0	2
Argatroban	1	0
None	1	1

Values are presented as n (%) or n. FET, Frozen elephant trunk.

Table 4.

Frozen elephant trunk (FET) position and anticoagulation in type A dissection patients

	Thrombus no (n = 26)	Thrombus yes (n = 3)
FET position
Eccentric FET	9	0
Central FET	17	3
Anticoagulation
Enoxaparin	14	3
Tinzaparin	11	0
Warfarin	1	0

Values are presented as n.

Anticoagulation therapy practice changed in our institution over the years. Except for warfarin, every regimen has seen some cases of thrombosis. No anticoagulation, aspirin, and aspirin with clopidogrel have seen especially high rates of thrombosis. The trend toward a high ILT rate in patients treated with enoxaparin is maintained in patients with TAD (Tables 3 and 4).

Discussion

The development of the FET stent-graft has revolutionized aortic arch surgery. Although traditional methods of repair still have a role to play in aortic pathology, the FET technique offers better outcomes as well as better stabilization of the distal aorta and a landing zone for subsequent abdominal aortic intervention in cases involving the descending aorta.⁹ Despite all the potential advantages of FET surgery, ILT is a common complication and is associated with significant morbidity and worse outcomes. Our results with 11% of patients developing ILT and an associated 30-day and 6-month mortality of more than 30% is in keeping with previously published data.^1,5 Although the difference in hospital length of stay between patients with and without ILT did not reach statistical significance, the longer stay in the ILT group is likely to be a function of the higher morbidity burden. For these reasons, ILT should be addressed as part of the ongoing evolution of perioperative management of patients undergoing FET aortic arch replacement.

Attempts to characterize ILT have focused on risk factors such as patient-, device-, or procedure-related characteristics.⁷ Our institutional experience shows similar thrombus rates to those described elsewhere.1, 2, 3, 4, 5 However, counter to previous publications, we did not find a significant difference in thrombus incidence between sexes, where Benk and colleagues⁴ identified female sex as a significant risk factor. Our study supports previous findings that degenerative aortic aneurysm as index pathology and central FET position are risk factors for ILT formation.^1,5

The precise mechanism behind ILT formation is currently unclear. Major aortic surgery is likely to disrupt all 3 factors described by Virchow as determining thrombus formation—blood flow velocity, vessel intimal lining perturbation, and coagulation status. Not being able to achieve a distal seal for the hybrid graft is likely to be a contributing factor. Using the largest-diameter FET stent carries the potential risk of thrombus generation by decreasing blood velocity compared with the graft. In our practice, minimizing the volume of residual aneurysmal aorta and providing a better landing zone for the subsequent thoracic endovascular aortic repair have been beneficial. Attempts to characterize the mechanisms underlying ILT formation have so far been inconclusive and have at times yielded contradictory information.⁷ This is likely due to the limited number of studies, mostly describing single-center experiences.

Martens and colleagues¹ report an increased risk of ILT with increased stent diameter to body size, whereas Misfield and colleagues⁵ found that a lower stent-to-aneurysm diameter ratio is associated with increased thrombosis.^1,5 Large aneurysmal descending aorta, large dimension of the stent relative to aorta or patient size, and incomplete aneurysmal stent coverage have also been associated with an increased rate of ILT.^1,5 Despite the contradictions, the common theme in these findings is that they pertain to the blood flow disruption aspect of Virchow's triad.

The formation of pockets in areas of relative excess of graft material, such as at the small curvature of the aorta, promotes low-flow, stasis or turbulent-flow, all of which may lead to thrombus formation.^1,7 Other areas of similarly abnormal flow may occur at the distal end of the graft when there is a significant size mismatch between graft and native aorta or in case of incomplete aneurysmal coverage. The disruption of laminar flow in these areas has been evaluated by means of magnetic resonance imaging, computed tomography (CT) imaging, or transesophageal echocardiography.¹

To minimize the risk of spinal cord ischemia, we use proximalization of the distal aortic anastomosis. Using 10- to 12-cm stents and anastomosing the arch in zone 0 or 1 often leads to placement in the arch rather than in the descending aorta. This is likely to influence the flow characteristics within the stent portion with a greater difference in blood velocity between the inner and outer curvatures compared with a zone 3 placement for which the devices were originally designed.

Based on our findings and those of others, it is reasonable to assume that a high proportion of aneurysm disease, more proximal stent deployment, use of large-diameter stents, and failure to achieve a distal seal predict a greater tendency to intraluminal stent thrombosis.

Unsurprisingly, coagulation disorders have been associated with higher rates of ILT. For example, heparin-induced thrombocytopenia and factor V Leiden mutation have been associated with clot formation.^1,2 Managing postoperative bleeding conservatively and at the expense of high transfusion requirements has been associated with increased risk of ILT.⁵ It is unclear if this is due to increased transfusion, the hypotension that is often associated with excessive postoperative bleeding, or delayed commencement of anticoagulation therapy. Due to the large geographical catchment area of our center, the majority of TAD patients referred to us arrive with a severe coagulopathy and receive more blood and blood products than elective patients. The fact that the TAD patients included in the present study show the same ILT rate as elective patients makes higher transfusion requirements as a reason for ILT formation less likely.

The majority of ILT after FET seems to occur within the first 7 to 10 days after surgery.1, 2, 3 This stands in contrast to the formation of graft thrombus in thoracic endovascular aortic repair or endovascular aortic repair cases, where a median time to thrombus has been described at 10 months.^7,10 The question of the time frame in which ILT develops after FET is important because it determines when to schedule the first routine postoperative imaging study. The 1 study that so far has commented on the timing of imaging found that the authors’ imaging strategy was heterogeneous and ranged from before discharge to during follow-up appointments. The majority of patients included in the study had their first scan on postoperative day 15.¹ This is in keeping with our practice, which up to now has been equally heterogeneous.

CT angiography (CTA) remains the highest yield investigation and is recommended in the long-term postoperative management of aortic disease.¹¹ Transesophageal echocardiography has also been shown to be effective in diagnosing intrastent thrombosis,^1,12 but it is debatable if patients will accept this as a long-term screening tool or even in the management of acute ILT.

With an incidence of up to 16.8%⁵ and the associated high mortality and morbidity burden, effective anticoagulation therapy is required to prevent ILT as much as possible. Most of the available literature focusses on incidence, risk factors, and consequences. Only a limited number of articles describe mostly institutional practice. Benk and colleagues⁴ used aspirin only—unless there were clinical reasons to opt for a different anticoagulation regimen—with a reported ILT rate of 5%. Ibrahim and colleagues³ only mention that anticoagulation therapy was administered during the patients’ hospital stay, without specifying the pharmacological agent(s), and report an incidence if 6.25%.³ Walter and colleagues² report initiating an activated partial thromboplastin time-targeted continuous heparin infusion 6 hours postoperatively without specifying if and how anticoagulation therapy was continued after stopping the infusion, with an ILT rate of 6%. Martens and colleagues¹ describe a strategy of using aspirin with prophylactic low-dose heparin and a rate of ILT of 8.2%. This group only uses therapeutic anticoagulation, either with a direct oral anticoagulant agent or warfarin, if there is another indication to do so.¹

A number of measures to prevent ILT have been suggested.⁷ Amongst them are anticoagulation therapy with warfarin for up to 3 months until a follow-up CTA is performed, although no recommendation for a post-CTA anticoagulation strategy is made. Additionally, the authors suggest this only for patients belonging to risk groups and those with altered blood flow in the stent or with coagulation disorders.¹ Risk groups, as they have emerged from a small number of studies, are summarized in Table 5. Other recommendations include performing routine surveillance CTA scans in all patients early within the first postoperative week.⁷

Table 5.

Risk groups for intraluminal thrombus identified so far in the literature

Patients in risk group
Patient factors	Age2,4 Female2 (Degenerative) aortic aneurysm
Operation factors	Increased stent graft diameter index1 Anticipated type Ib endoleak1 Incomplete aneurysmal stent coverage5 Central frozen elephant trunk position∗ Large descending aorta1 Stent graft diameter <34 mm5 Conservative management of major postoperative bleeding5 150-mm stent graft length3 Concomitant aortic valve/root surgery3

^∗Current study.

Our practice has seen multiple anticoagulation regimens. Based on our present findings, it has become departmental standard to start warfarin on postoperative day 2 for 3 months followed by a lifelong direct oral anticoagulant therapy. All FET patients now undergo their initial surveillance CT imaging study before hospital discharge, but usually within the first postoperative week, and again 6 months after that. Since changing our practice, we have not seen any further cases of ILT in the past 8 months, either at initial or follow-up CTA.

Our study has several limitations. We present single-center data in this retrospective and descriptive study. The fact that our postoperative anticoagulation regimens have undergone several changes and are determined by individual surgeon preference gives us the unique opportunity to compare outcomes associated with different practices while also mitigating potential bias. We acknowledge that our study is not generalizable due to its single-center nature.

The number of patients who underwent FET aortic arch replacement for emergency TAD repair is small. Most TAD patients receive an interposition graft or, more recently, an AMDS graft (Artivion Inc). Although the number is too small to contribute to the statistical analysis, treating preexisting coagulopathy, with is part of the disease process, does not seem to increase the incidence of ILT. The choice of anticoagulation agent is very heterogeneous in our cohort. Although it does not allow comparative analysis, it shows useful trends that have allowed us to modify our practice with good outcome.

Our experience in using the E-vita Neo graft is still in its infancy and, compared with the Thoraflex graft, the numbers are small. The higher percentage of patients developing ITL after E-vita Neo aortic arch replacement should be noted with caution.

Conclusions

Given the high incidence of ILT and the high risk of poor patient outcome, we advocate early postoperative CTA and a more aggressive anticoagulation regimen with warfarin initially for 3 months and followed by a lifelong direct oral anticoagulation therapy. Despite being established practice, the evidence base for perioperative and long-term management of patients with FET is thin. Collaborative research, preferably in the form of large clinical registries involving different institutions and different disciplines, is necessary to provide the data that can shape practice guidelines. In addition, industry partners should be called upon to continue improving stent-graft design by working closely with clinicians. Only such a combined approach is likely to significantly influence the incidence of this serious complication.

Conflict of Interest Statement

The authors reported no conflicts of interest.

The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which there may be a conflict of interest. The editors and reviewers of this article have no conflicts of interest.