Cerebrovascular events appear in 2 different scenarios in patients undergoing TAVI: a) during the acute phase related to the procedure, and (b) during the chronic phase, with a constant rate of new episodes during follow-up. Although the pathogenesis of the acute phase of stroke or TIA following TAVI is likely multifactorial, given the arterial distribution and temporal pattern of peri-procedural brain infarctions, embolization is likely to be the dominant mechanism.16–18 Nombela-Franco et al.17 found that 50% of acute phase stroke/TIAs appear within the first 24hours post-TAVI, suggesting that catheter manipulation in the setting of atherosclerotic-laden aortas and severely calcified aortic valves might play an important role. The remainder of strokes in the acute phase appears after the first 24hours of the procedure, with a vulnerable period of up to 2 months. In this setting, thromboembolism originating directly from the native-transcatheter heart valve complex per se, or as a result of chronic or new atrial fibrillation, likely further contributes to strokes early post-TAVI. Vascular complications, and an increased atherosclerotic burden are other proposed mechanisms. On the other hand, the stroke rate incidence during the chronic phase post-TAVI is similar in patients treated surgically or medically, suggesting that the risk of neurological events in this period is determined by the underlying baseline risk profile of patients with aortic stenosis rather than by factors associated with the procedure.18–22

Clinically overt stroke simply represents the tip of the iceberg in terms of cerebral embolization. The spectrum of cerebrovascular damage following TAVI also involves subclinical cerebral infarcts detected by diffuse weighted magnetic resonance imaging (DWI), which have been related to a potential cognitive decline in the long-term.23–31

Cerebral DWI is a technique providing in vivo insights allowing the differentiation of acute from chronic stroke as well as nonspecific white-matter lesions. It is sensitive to changes in the mobility of water molecules from the extracellular to intracellular compartment seen in the early phases of the cascade of ischemic tissue changes. Therefore, DWI provides temporal information, because acute lesions become increasingly hyperintense during the first few days but then attenuate during the subsequent weeks (Figure 1). In addition, DWI has high sensitivity in identifying small ischemic lesions. Moreover, certain imaging patterns, such as multiple cortical or/and subcortical acute infarctions affecting both hemispheres, or anterior and posterior cerebral circulations, strongly suggest a cardiac or aortic source of embolism. However, it is important to keep in mind that such findings do not necessarily prove cerebral ischemia, given that other mechanisms, such as prothrombotic, inflammatory or infectious processes could also provide a similar distribution.23

Figure 1.

A: Post-TAVI cerebral diffuse weighted magnetic resonance imaging in an 82-year-old patient following successful TAVI, demonstrating acute ischemic lesions (hyperintense images) distributed across both hemispheres and within the anterior and posterior territories, suggesting an embolic etiology. B: Origin of embolization following TAVI. Aortic plaque or valve disruption during catheter and guidewire passages, thrombus formation during the procedure, and subacute thromboembolism originating directly from the native-transcatheter heart valve complex per se, atherosclerotic burden, or caused by chronic or new atrial fibrillation, are the main sources of emboli associated with TAVI. TAVI, transcatheter aortic valve implantation. From Kahlert et al.27 and Fanning et al.18 with permission.

(0.23MB).

Several studies have demonstrated the near ubiquitous presence of subclinical or silent cerebral lesions detected by DWI following TAVI with a cerebral pattern highly suggestive of an embolic process (Table 1).24–30 Similarly, transcarotid Doppler studies have shown a high incidence of high-intensity transient signal (HITS) during TAVI, especially during valve positioning and implantation, thus suggesting the importance of embolization during these procedural steps.31,32 For these reasons, some authors have suggested a likely common origin of these cerebral DWI lesions with those generated by macroemboli that usually result in a clinically overt stroke. Importantly, the scanner potency used (ie, 1.5 T or 3 T), as well as the time point for performing DWI post-TAVI could affect the sensitivity for detecting silent cerebral infarcts, as these lesions tend to disappear over time, being totally absent at 30 days following the procedure.23,31–33 Therefore, in embolic protection device (EPD) studies, change in the cerebral DWI ischemic total lesion volume (TLV) has been proposed as a surrogate endpoint of clinical stroke.34,35 However, even though the biological plausibility and cost-rationale of this assumption, it should be confirmed in larger studies. In fact, the current American Heart Association guidelines do not recommend the use of silent cerebral infarcts detected on brain imaging as a surrogate endpoint for stroke studies, unless all study patients undergo standardized imaging at specific time points according to the study protocol.36,37

Gaseous embolization, an inflammatory state related to cardiac instrumentation, thrombus, or embolization of debris from the aortic wall or the aortic valve have been proposed as possible sources of silent DWI cerebral defects detected following cardiac procedures. Keeley et al.38 detected embolic debris in 51% of patients undergoing catheterization in a prospective study involving more than 1000 patients. Van Mirghem et al.19,39 studied the histopathological characteristics of debris traveling to the brain captured by the Claret filter in 81 patients undergoing TAVI. Thrombotic material was found in 74% of patients, and interestingly, tissue derived debris originating from the aortic wall, valve or myocardium was found in 63% of patients (Figure 2). These findings may relate to the still poorly understood pathophysiology of CVE after aortic valve manipulation. Disruption of the aortic intima, native valve, or myocardium following catheter manipulation during TAVI could lead to microthrombosis and consequent embolic complications and subsequent new silent cerebral lesions.40

Figure 2.

Debris material travelling to the brain captured by the Claret embolic protection device during transcatheter aortic valve implantation. Histologic illustrations of captured debris of a degenerated aortic leaflet fragment, heart muscle, and arterial wall. Co, collagenous tissue; En, endothelial lining; Fi, fibrin. From Van Mieghem et al.19,39 with permission.

(0.58MB).

The relationship between subclinical cerebral infarcts and cognitive decline during the follow-up period post-TAVI remains uncertain. Recent studies performed in TAVI patients failed to find an association between new cerebral lesions and cognitive decline during follow-up.30,41 By contrast, some surgical series of patients undergoing aortic valve replacement found a relationship between new DWI cerebral lesions after the intervention and cognitive decline during follow-up.42 Similarly, silent brain infarcts detected with DWI in healthy patients are associated with a greater risk of dementia.43 Several issues need to be highlighted to better understand these apparent discrepancies. Firstly, there are currently no validated models for assessing neurocognitive status in patients undergoing TAVI. Interstudy differences could be partially explained by either different types of cognitive test battery or variations among definitions of cognitive decline. Secondly, a baseline degree of cognitive impairment is usually documented in a significant proportion of patients undergoing TAVI. In this situation, the floor effect (the incapacity to detect changes due to low baseline scores) needs to be taken into account when interpreting the results of the neurocognitive battery tests, especially in studies with negative results.44 In fact, part of the mild improvement in test scores observed in some studies that repeat the test at several time points may be related to a “learning effect” in the patients completing the test.32 Thirdly, cognitive assessment is affected by a high prevalence of inter- and intraobserver variability, and consequently the risk of bias is high in the absence of a well-matched control group.37

To avoid these prior concerns, some authors have proposed the use of normative test data well matched with the patient group of interest across demographic, health, and language normative domains, as well as the inclusion of study-specific well-matched controls and test-retest reliability adjustments for improving the generalizability and comparability of results across studies.30,32,44–46 Of note, none of these recommendations have yet to be followed in the published EPD studies.

The Embrella Embolic Deflector device (ED) (Edwards Lifesciences; Irvine, California, United States) is a mechanical system designed to deflect debris travelling to the brain originating from aortic valve or ascending aorta during TAVI. The ED is deployed into the aortic arch and is designed to cover the ostium of the brachiocephalic trunk and left common carotid artery, without systematic covering of the ostium of the left subclavian artery. The ED has 2 heparin coated polyurethane membranes with 100μm pore size called petals. The petals are mounted on a self-expandable nitinol frame attached to a 110cm long nitinol shaft. The petals are oriented in opposite directions to cover the supra-aortic arteries (the device extends over a length of 58mm with a width of 25mm). The system can be delivered through a 6 Fr delivery sheath introduced from the right radial or right brachial artery. Three radiopaque markers help fluoroscopic-guided deployment (Figure 3A, 3B, 3C and Table 3).48

Two studies have explored the efficacy of the ED; the ProTAVI-C study and a study performed by Samim et al. (Table 2).32,55 None of these studies were randomized. The ProTAVI-C was the first study to explore the efficacy of an EPD during TAVI, and to date is the most complete study evaluating this device. The study enrolled 52 patients undergoing TAVI across 6 hospitals. The ED was implanted in 41 patients while the remaining 11 patients comprised a control group. All events were adjudicated by an independent blinded core laboratory. Patients enrolled in this study were at intermediate preoperative risk, the median age was 83 years (range, 79-86 years) and patients with prior stroke/TIA were excluded (Table 4). The feasibility and safety of the ED was shown (Table 3). Patients undergoing TAVI with the ED did not present with significant additional complications. The ED was successfully deployed at the level of the greatest aortic curvature in all patients. Device success was achieved in all but 1 patient (97.6%), and the additional procedural time due to ED deployment was 2 min [interquartile range, 1-3 min]. There were no cases of membrane device rupture, and there were 2 complications related to the device procedure: 1 radial thrombosis without consequences and 1 brachial arterial pseudoaneurysm that required surgical repair. Transcarotid Doppler imaging was performed in all patients. The presence of HITS was detected during every step of the TAVI procedure, the most significant being while crossing the native aortic valve and positioning the transcatheter valve. Interestingly, the ED group showed a higher number of HITS than controls (632 [347-893] in ED group vs 279 [0-505] in the control group, P < .001), suggesting that ED manipulation may also represent a potential source of embolic debris. At a clinical level, there were 2 strokes and 1 TIA, all occurring within the ED group (the rate of clinical CVEs in ED group was 7.3%, all occurring after the procedure). The incidence of silent ischemic cerebral lesions detected by DWI was not reduced with the use of the device. All 40 patients who had baseline and postprocedure DWI (34 in the ED group and 6 in the control group) had new DWI defects, with most of these patients showing multiple defects. However, the use of an ED system appeared as a protective factor to reduce the ischemic TLV when compared with controls (Table 5). All but 1 of the neurocognitive tests failed to reveal any significant differences between treatment groups, with the Montreal Cognitive Assessment battery (MoCA) demonstrating a mild improvement in the ED group (score of 24 [21-27] in ED group vs score of 25 [23-28] in controls; P < .001). However the low number of patients, specifically in the control group, and the lack of randomization were the principal limitations of the study.

The TriGuard (TG) EPD (Keystone Heart Ltd., Herzliya, Israel) is a mechanical system conceived to deflect debris travelling to the brain originating from the aortic valve or ascending aorta during TAVI. The TG device is the only device designed to cover all 3 major arteries in the aortic arch (brachiocephalic trunk, left common carotid artery, left subclavian artery). The device has 5 functional parts: a dual frame nitinol frame, a thin nitinol coated mesh filter with pore size of 250μm (or 130μm for the second generation device), upper and lower stabilizers, and the tail end (distal to the heart) attached to the plunger during the procedure. The coated mesh has chemical and physical properties to reduce the likelihood for thrombus formation. The stabilizers are anchored within the ostium of the innominate artery and in the inner curvature of aortic arch respectively to facilitate the positioning of the filter. The upper stabilizer requires a minimum diameter of 8×11mm into the innominate artery at 15mm of the ostium. The lower stabilizer requires an aortic arch of 35mm or less to anchor properly. Irregularities in the innominate artery anatomy or an innominate ostium diameter less than recommended may result in suboptimal positioning and stability of the device. Special care should be taken with a severely calcified aorta, specifically in the vicinity of innominate ostium. The upper stabilizer angle of the innominate should not deviate from being perpendicular to the arch by more than 45°. The system can be delivered through a 75cm long 9 Fr delivery sheath introduced from the femoral artery (Figure 3D, 3E, 3F and Table 3).49

The DEFLECT III trial is the most relevant study evaluating the TG device. DEFLECT III is a single-arm blinded randomized clinical trial.53 The study enrolled 85 patients (39 controls) across 13 centers to evaluate the feasibility, safety, and efficacy of the TG device in patients undergoing TAVI (Table 2). The trial included patients considered at intermediate surgical risk (Table 4). Technical success (defined as complete 3-vessel cerebral coverage) was achieved in the 88.9% (40 of 45) of patients (Table 3). Time to complete deployment was 13min in the DEFLECT I study, but was not shown in the DEFLECT III trial. In terms of clinical efficacy, there was no between-group difference in the primary safety endpoint (including death, stroke, life-threatening or disabling bleeding, stage 2 or 3 acute kidney injury, or major vascular complications) (Table 5). In the intention-to-treat analysis, the use of the TG device was associated with nonsignificant freedom from of new cerebral DWI lesions (21.2 vs 11.5%), and a nonsignificant reduction in “new neurologic impairment” (a post hoc endpoint defined as worsening in National Institutes of Health Stroke Scale score from baseline with DWI evidence of ischemia). Although the authors noted a reduction in the cerebral TLV in the TG group when TLV was categorized into small, medium and large sizes, the proportion of patients with large ischemic volume was similar between groups (46% in the TG group vs 48% in controls). Although cognitive assessment via the MoCA and the Cogstate Research battery did not show significant between-group differences at discharge and at the 30-day evaluation, on the International Shopping List Test (a measure of episodic memory [delayed recall]), significant differences were observed when patients were evaluated at discharge favoring the interventional arm (65.4% vs 30.4%, P = .022). The main limitation of DEFLECT III was the high rate of loss to follow-up (31% of patients were lost to postinterventional DWI and 26% of patients were lost to the postinterventional cognitive and neurologic assessments).

The Claret embolic protection device (CD) (Claret Medical, Inc.; Santa Rosa, California, United States) is the only mechanical system designed to capture and not to deflect debris travelling to the brain. The CD is deployed into the ostia of brachiocephalic trunk and left common carotid artery. The system consists of 2 polyurethane filters with 140μm diameter pores fixed in a flexible nitinol radiopaque frame attached within a 100cm long catheter. The proximal filter is delivered into the brachiocephalic artery (right side) and allows apposition within vessels measuring 9-15mm in diameter. The system enables the delivery of the distal filter (left-sided filter) into the left common carotid artery. The entire system is positioned through the right radial or right brachial artery using a 6 Fr sheath. The CD is deployed before passing the TAVI delivery system into the aortic arch and is withdrawn following removal of the TAVI delivery system (Figure 3G, 3H, 3I, and Table 3).50

There are 2 randomized clinical trials exploring the efficacy of the CD: the Claret Embolic Protection and TAVI trial (CLEAN-TAVI, NCT01833052) with the Medtronic CoreValve system, and the MISTRAL-C trial (Table 2) with the Edwards-SAPIEN 3 system. To date, none of these studies have been published. CLEAN-TAVI is the most relevant trial. The main objective was to evaluate the impact of the use of the CD on the number of cerebral lesions in patients undergoing TAVI. Patients underwent DWI at day 2 and at day 7 post-TAVI, and serial neurological and cognitive assessments were performed. The initial results were encouraging (unpublished data, Dr. A Linke et al., oral presentation, TCT 2014). One hundred patients were enrolled (50 with CD and 50 controls) between April 2013 and June 2014 at the University Hospital of Leipzig (Germany). In the CD group, patients were mainly at intermediate preoperative risk (mean Society of Thoracic Surgeons score of 5.6%), there was a relatively high proportion of women (60%), and a high incidence of prior atrial fibrillation (32%) (Table 4). Device deployment/success was achieved in 96% of participants (Table 3). The causes of unsuccessful device deployment were severe tortuousity of the supra-aortic trunks (left common carotid artery in 1 case and subclavian artery in the other). In addition, there was an accidental dislocation of a correctly deployed filter by a pigtail catheter. Interestingly, National Institutes of Health Stroke Scale evaluation revealed a substantial incidence of neurological deficits at 2 days post-TAVI in both the control and device groups (28 vs 13%, P = 0.8), mainly because an unusually high incidence of ataxia was documented (24% of controls vs 9% in the CD group; relative risk = 1.56 [1.08-2.21], P < .05). However, at 7- and 30-days post-TAVI, the incidence of any neurological symptom was lower, with no between-group differences (day 7: 10 vs 13%; day 30: 12 vs 12%, respectively). In terms of reducing subclinical cerebral injury, the CD failed to reduce the incidence of new DWI cerebral lesions (98% of incidence of new DWI lesions at 2 days post-TAVI). However, in the filter group, there was a 50% reduction in the number of DWI cerebral lesions (day 2: median of 8 lesions in filter group vs median of 16 lesions in control group, P = .023; day 7, median of 5 lesions in filter group vs median of 10 lesions in the control group, P = .012). In addition, there were also a 41% reduction of TLV in the CD group (471 vs 800μL, P = .024) in the magnetic resonance imaging (MRI) performed at day 2, and 53% (220 vs 472μL, P = .013) in the MRI performed within 7 days post-TAVI. These changes were observed largely within cerebral territories that were protected by the filter (day 2 post-TAVI: 246 vs 527μL, P = .002 within cerebral anatomic regions protected by the CD) (Table 5). To date, there are no published data of neurocognitive assessments.54

The MISTRAL-C is a multicenter double-blind randomized trial enrolling 63 patients with symptomatic severe aortic stenosis with a 1:1 randomization to TAVI with or without the use of second generation of Claret Sentinel EPD (Table 2). 56 The use of this device was associated with a trend toward fewer new brain lesions on the 5-day post-TAVI MRI, especially within the cerebral regions corresponding to the areas offered protection by the device (unpublished data, Van Mieghem et al., oral presentation, TCT 2015). However, these results should be interpreted with caution, given that only 57% of the randomized patients underwent MRI.