1. Introduction

Cerebral embolization during transcatheter aortic valve replacement (TAVR) is frequent and causes stroke in up to 2.3%–9.1% of patients [1–5]. Current data from an all-comers analysis investigating 1919 TAVR patients over a 10-year decade revealed a 30-day stroke incidence of 3.4%. In this study, neuroimaging reports detected more often left-sided cardioembolic strokes (45.6%) than right-sided (25%) and bilateral (13.2%) infarctions [6]. In general, left-hemispheric ischemic strokes have been reported to be associated with a worse clinical outcome than right-hemispheric strokes [7, 8].

Theories behind the predilection for cerebral embolization of the left-brain hemisphere attribute an important role to the anatomical and rheological characteristics of the aortic arch [9–12]. Volumetric flow studies showed that cardioembolic particles have the tendency to slide along the outer curvature of the ascending aorta and to enter the first ostium of the aortic arch branches [10, 13]. Geometrical variations such as a common origin of the brachiocephalic and left common arteries (bovine arch anatomy) or a sharp angulation of the aortic arch might raise the likelihood of embolic particles entering the left cerebral circulation [10, 11, 13, 14]. Catheter tracking and manipulation during the TAVR procedure might additionally increase the risk of left-sided strokes caused by embolization of mobilized debris in such anatomical constellations [15–17]. Studies investigating the impact of aortic arch variations on stroke localization in TAVR are lacking. We therefore sought to investigate the impact of the aortic arch anatomy on the incidence of stroke laterality during transcatheter aortic valve implantation by multislice computed tomography (MSCT) analysis.

2. Materials and Methods

2.1. Patients

We conducted a retrospective, MSCT-based analysis of all consecutive patients who developed a symptomatic, right- or left-sided periprocedural ischemic stroke after TAVR at the Department of Cardiovascular Surgery at the German Heart Centre Munich between June 2007 and August 2022. The inclusion criteria for this study were the availability of brain imaging, pre-TAVR aortograms, and manifestation of stroke within 30 days after the procedure. The exclusion criteria were the presence of incomplete or poor-quality MSCT data sets of the aortic arch and its supraaortic branches, aortic dissection, usage of cerebral embolic protection devices, participation in the ongoing prospective randomized PROTECT TAVI trial (clinicaltrials.gov NCT02895737 328), significant carotid artery stenosis (>70%) ipsilateral to the stroke, conversion to surgical aortic valve replacement, intraoperative resuscitation, and bilateral and hemorrhagic stroke (Figure 1).

This study complied with the Declaration of Helsinki and was approved by the Local Ethics Committee of the Technical University of Munich (approval reference number: 725/21 SKK).

2.3. MSCT Measurements of the Aortic Arch

All MSCT measurements were performed with an automated software for 3-dimensional CT reconstruction (3mensio Structural Heart version 10.2, Pie Medical Imaging, Maastricht, the Netherlands) as described previously by the authors [5, 19]. The following parameters were assessed.

2.3.1. Aortic Arch Anatomy

Aortic arch anatomy was defined as follows:

• (i)

  Normal aortic arch is the separated origin of the brachiocephalic, left common carotid, and left subclavian arteries

• (ii)

  Bovine arch type I is the common origin of the brachiocephalic and left common carotid artery

• (iii)

  Bovine arch type II is the common origin of the brachiocephalic and left common carotid artery, with the left common carotid artery bifurcating at an average distance of ≤1 cm from the origin [13].

2.3.2. Aortic Arch Configuration

Arch configuration according to the outer and inner curvature of the aortic arch was assessed as described previously by the authors [5, 20] and categorized as follows [5]:

• (i)

  Aortic arch type I: all supraaortic branches originate from the aortic arch at the level of the first line (Figure 2(a))

• (ii)

  Aortic arch type II: at least one of the supraaortic branches originates between the two lines (Figure 2(b))

• (iii)

  Aortic arch type III: at least one of the supraaortic branches originates at the level or below the second line (Figure 2(c)).

2.3.3. Aortic Arch Curvature

Aortic arch curvature was assessed by determining the maximal aortic arch angulation (AAA) and the aortic arch tortuosity index (TI). 3-mensio imaging analysis initially included automatic extraction of a three-dimensional (3D) center line along the vascular course of the aortic arch, which was manually adjusted to obtain the most accurate measurement.

For evaluation of AAA, two reference points were selected along the center line at the level of the bifurcation of the pulmonary trunk and at the level of the fourth thoracic vertebra body upper edge according to Boufi et al. [21] (Figure 3). Then, two planes were generated perpendicular to the center line at the marked points. The maximal measured angle between these two planes was defined as the aortic arch angulation (Figure 4).

The TI of the aortic arch was determined by calculating an aortic arch distance factor: [(center-line distance)/(straight-line distance) − 1] × 100 [5, 21]. The center-line distance reflects the incremental true length of the aorta along the previously defined 3D center line between the marked reference points, while the straight-line distance indicates the linear distance between the landmarks (Figure 5).

2.3.4. Take-Off Angles of the Supraaortic Arteries

Measurement of the take-off angles of the supraaortic arteries included the determination of the following angles as described previously [5]:

• (i)

  Angle between the aorta and the brachiocephalic artery (AA/BA angle)

• (ii)

  Angle between the aorta and the left common carotid artery (AA/LCC angle)

Interobserver and intraobserver reliability for all computer tomographic measurements was assessed as described previously in detail by the authors [5, 22]. The measurements were performed by two independent cardiac surgeons with long-standing expertise in cardiovascular imaging.

3. Results

A total of 4,590 patients underwent TAVR between June 2007 and August 2022 in our department. Among these, 124 patients developed a periprocedural stroke. Forty-seven patients were excluded from the analysis either due to conversion to surgical aortic valve replacement (n = 2), the use of cerebral embolic protection devices (n = 5), intraoperative and early postoperative resuscitation (n = 6), poor-quality MSCT scans (n = 2), hemorrhagic stroke (n = 8), and bilateral stroke manifestation (n = 24). Therefore, the study cohort comprised 77 patients of whom 66.2% (n = 51) suffered a left-sided stroke and 33.8% (n = 26) a right-sided stroke (p = 0.006). The mean age at the time of TAVR was 79.8 ± 7.6 years, and the median EuroSCORE II and STS Score were 3.4 [1.1–22] and 3.0 [1.2–19.5], respectively. Further baseline characteristics are displayed in Table 1, showing no difference between the study groups (right-vs. left-sided stroke).

In total, 46 self-expandable (SEV) and 31 balloon-expandable (BEV) transcatheter heart valves were implanted via transfemoral (n = 65), transaortal (n = 4), transapical (n = 7), and transaxillary (n = 1) access. TAVR valve type, access route, and balloon predilatation and postdilatation did not differ between TAVR patients with right-sided and left-sided stroke manifestation (Table 2). In 41.6% of the patients (n = 32/77), post-TAVR stroke occurred within 24 h after the index procedure, whereas 58.4% suffered from a subacute periprocedural stroke.

MSCT measurements revealed no significant difference regarding the anatomy and configuration of the aortic arch in TAVR patients with right-sided and left-sided ischemic strokes (Table 3). Normal aortic arch anatomy and type II arch configuration were the most frequently observed morphologies in patients with right- and left-sided strokes, with 65.4% and 72.5% and 69.2% and 76.5%, respectively. Determination of the supraaortic take-off angles did not show any difference between right- and left-sided brain infarctions (AA/BA angle: p = 0.620; AA/CCA: p = 0.337). Calculation of aortic arch angulation (right-sided 95.5° [63–141] vs. left-sided 88.0° [27–132]; p = 0.267) and aortic arch TI (right-sided 15.8 [8.3-45.7] vs. left-sided 14.6 [0.4–41.5]; p = 0.0526) was also similar in both groups (Table 3).

Assessment of interobserver and intraobserver reproducibility revealed an overall mean ICC of 0.982 (inter-) and 0.970 (intra-). The detailed results of the calculation for each MSCT parameter are included in Supplementary Materials (Supplement Tables 1 and 2).

4. Discussion

Since 2002 TAVR has emerged as an established treatment option in patients with aortic stenosis [23, 24]. However, TAVR-associated embolic strokes remain one of the most feared complications, significantly affecting survival and quality of life [25, 26]. Recent data from an all-comers analysis (n = 1919) suggest a hemispheric predilection for stroke associated with TAVR [6]. While it seems intuitively conceivable that anatomical variations like a severely angulated arch or a complex bovine arch configuration might affect the direction of embolic material, there are no clinical studies that objectively characterize this hypothesis in TAVR patients.

Prior computer-based volumetric flow models of the aortic arch suggested a correlation of arch morphology and corresponding stroke laterality [10, 27, 28]. However, this could not be confirmed clinically in our study. Based on a 3-dimensional MSCT analysis, aortic arch anatomy in our patients had no influence on hemispheric distribution of post-TAVR stroke. In line with previous studies [6, 11], a higher propensity of left-sided ischemic infarctions could be detected in our patients. We initially hypothesized that bovine arch configuration might be an important determinant for embolic stroke localization [13, 14]. Previous biomechanical flow studies indicated that embolic particles have the tendency to slide along the outer curvature of the ascending aorta entering the first ostium of the aortic arch branches [8, 10]. In the standard aortic arch, the brachiocephalic artery represents the first branch, which runs upwards and parallel to the direction of the ascending aorta, raising the likelihood of embolic particles entering the right cerebral circulation [13, 14]. In case of the bovine arch, the brachiocephalic trunk and the left common carotid artery share the same origin resulting in a more oblique vessel orientation of the left common carotid artery. This, in turn, might be more aligned with the trajectories of arriving embolic material [11, 13]. Despite a bovine arch prevalence of 29.9% (23/77) in our study cohort, which is higher than the general population-based incidence of 8–25% [7, 9, 29], no difference in stroke laterality could be observed. Similar to our findings, Gold et al. also showed no significant difference in right-left propensity of cerebral infarcts comparing standard with the bovine arch in 119 patients suffering from cardioembolic, non-TAVR related stroke [14].

In contrary, a recently published study by Mataka and colleagues supported a correlation of aortic arch anatomy and hemispheric laterality in a large investigation of 615 patients with acute cardioembolic brain infarcts due to atrial fibrillation [13]. MSCT measurements in this study revealed a significant propensity for left-sided cerebral cardioembolic infarcts in patients with bovine configuration [13]. One reason for the inconsistent study findings might be the difference in the analyzed cohorts. While the study population by Mataka et al. focused on patients suffering from spontaneous, non-perioperative strokes, our study cohort compromised procedure-related cerebral infarctions. Embolic clot motion in our patients might be different. Acute stroke following TAVR predominantly is due to embolization arising from particulate matter during advancement of THV devices and THV deployment [30]. Intravascular devise manipulations might induce flow disturbances potentially affecting clot trajectory and right-left stroke propensity.

Among the branching pattern of the aortic arch, arch steepness is also thought to be associated with stroke laterality [11, 27, 28]. Choi et al. performed fluid dynamic simulations using patient-dependent aorta models suggesting that aortic steepness might be an important risk factor for propensity of stroke distribution [28]. In case of a steep arch morphology, the left common carotid artery runs parallel to the inlet flow stream which might result in more emboli entering this side [27]. Arch steepness in our TAVR analysis was represented by evaluation of aortic arch configuration (types I–III) and arch angulation. We found a type II steep aortic arch to be the most common configuration in our patients (74.0%), with a median arch angulation of 90°degree. However, arch steepness in our clinical analysis was no predictor for stroke laterality in TAVR-associated strokes. Thus, the assumptions made in the biomechanical studies could not be confirmed in our clinical study setting. Also, Elsaid and colleagues found no significant correlation between arch angulation and left-sided stroke propensity in 192 patients with cardioembolic cerebral infarction [11].

5. Limitations

Our study reports the results of a retrospective, single-center investigation of limited size. As the primary outcome was the relationship between stroke laterality and the underlying aortic arch anatomy in TAVR-associated stroke, stroke severity and associated disability were not assessed. In addition, study results might be confounded by procedural factors such as operators’ experience, TAVR access routes, and advances in techniques over the study period, as well as postprocedural factors such as atrial fibrillation and initiation of sufficient anticoagulation.

6. Conclusion

Our data add evidence that periprocedural stroke after TAVR is associated with left-hemispheric predilection. However, there was no correlation between aortic arch morphology and stroke laterality. Based on the anatomy of the aortic arch, it is therefore not possible to predict the risk of periprocedural TAVR-associated stroke.

Conflicts of Interest

The authors declare that there are no conflicts of interests.