Role of Cerebral Embolic Protection Devices in Patients Undergoing Transcatheter Aortic Valve Replacement: An Updated Meta‐Analysis

Abstract

Background

Cerebral embolic protection devices (CEPD) capture embolic material in an attempt to reduce ischemic brain injury during transcatheter aortic valve replacement. Prior reports have indicated mixed results regarding the benefits of these devices. With new data emerging, we performed an updated meta‐analysis examining the effect of CEPD during transcatheter aortic valve replacement on various clinical, neurological, and safety parameters.

Methods and Results

A comprehensive review of electronic databases was performed comparing CEPD and no‐CEPD in transcatheter aortic valve replacement. Primary clinical outcome was all‐cause stroke. Secondary clinical outcomes were disabling stroke and all‐cause mortality. Neurological outcomes included worsening of the National Institutes of Health Stroke Scale score, Montreal Cognitive Assessment score from baseline at discharge, presence of new ischemic lesions, and total lesion volume on neuroimaging. Safety outcomes included major or minor vascular complications and stage 2 or 3 acute kidney injury. Seven randomized controlled trials with 4016 patients met the inclusion criteria. There was no statistically significant difference in the primary clinical outcome of all‐cause stroke; secondary clinical outcomes of disabling stroke, all‐cause mortality, neurological outcomes of National Institutes of Health Stroke Scale score worsening, Montreal Cognitive Assessment worsening, presence of new ischemic lesions, or total lesion volume on diffusion‐weighted magnetic resonance imaging between CEPD versus control groups. There was no statistically significant difference in major or minor vascular complications or stage 2 or 3 acute kidney injury between the groups.

Conclusions

The use of CEPD in transcatheter aortic valve replacement was not associated with a statistically significant reduction in the risk of clinical, neurological, and safety outcomes.

Nonstandard Abbreviations and Acronyms

CEPD

cerebral embolic protection devices

TAVR

transcatheter aortic valve replacement

Clinical Perspective.

What Is New?

• Our meta‐analysis provides a summary of available data from all major randomized controlled trials evaluating potential benefit of cerebral embolic protection devices in patients undergoing transcatheter aortic valve replacement, specifically exploring various clinical, neurocognitive, neuroimaging, and safety end points.

What Are the Clinical Implications?

• As stroke remains one of the most feared complications related to transcatheter aortic valve replacement, our study highlights the potential benefit of cerebral embolic protection devices in preventing disabling stroke.

• With cerebral embolic protection devices technology evolving, further studies are required to test this hypothesis of relative benefit of cerebral embolic protection devices in transcatheter aortic valve replacement.

Transcatheter aortic valve replacement (TAVR) has emerged as a treatment option for patients with aortic stenosis across the spectrum of surgical risk. ¹ , ² Despite the recent technological advances and improvement in operators' experience, periprocedural stroke remains an unpredictable and devastating complication of TAVR, affecting close to 2.5% of patients. ³ , ⁴ Covert or silent cerebral infarctions detected by diffusion‐weighted magnetic resonance imaging (DW‐MRI) are reported in 58% to 100% of patients and have been associated with an increased risk of postprocedural delirium, dementia, and long‐term neurocognitive decline. ⁵ , ⁶ Because embolization during valve manipulation during TAVR is one of the main mechanisms of stroke in these patients, cerebral embolic protection devices (CEPD) have been developed to prevent debris released during the procedure from embolizing to the cerebral circulation. These devices have been shown to successfully capture debris in 99% of patients ⁷ and have been associated with a numerical reduction in the number of DW‐MRI related end points. ⁸ However, larger population data on overall efficacy and safety of these devices are lacking as most randomized controlled trials (RCT)s have been underpowered for various clinical end points. Therefore, we conducted an updated meta‐analysis to investigate the effect of CEPD during TAVR on clinical, neurological, and safety outcomes.

Methods

This meta‐analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines (Figure 1; Figure S1). ⁹ The authors declare that all supporting data are available within the article (and its online supplementary files). This meta‐analysis did not require institutional review board approval.

Figure 1. PRISMA flow chart for this study.

PRISMA indicates Preferred Reporting Items for Systematic Reviews and Meta‐Analyses.

Publications Search Strategy

Two independent reviewers (A.K. and A.D.) performed a literature search of studies comparing outcomes of CEPD versus no‐CEPD (or control) in patients undergoing TAVR from inception to October 5, 2022, using PubMed, EMBASE, and Cochrane databases. We also performed manual searches of reference lists of other meta‐analyses along with abstracts presented at international or national conferences to identify eligible studies. The key words included “transcatheter aortic valve replacement” or “transcatheter aortic valve implantation” and “embolic protection device” (Data S1).

Selection Criteria

We selected RCTs that (1) included adults aged ≥18 years, (2) compared the outcomes of CEPD versus no‐CEPD in patients undergoing TAVR, and (3) reported our preferred outcomes as an end point. We excluded (1) studies that did not compare CEPD with no‐CEPD or control, (2) editorials, (3) case reports, and (4) duplicate studies.

Data Extraction and Appraisal

Two reviewers (A.K. and A.D.) independently extracted data on study characteristics and baseline patient characteristics from the included studies. Disagreements were resolved by discussion and by opinion of a third reviewer (S.S.). Study quality was assessed using Cochrane Collaboration Risk of Bias 2.0 tool for RCTs. ¹⁰ Intention‐to‐treat model was used for the included studies.

Outcomes

Our primary clinical end point was all‐cause stroke at the longest duration of follow‐up defined by the clinical trials. For all but 1 trial stroke was assessed at 30 days. ⁷ , ⁸ , ¹¹ , ¹² , ¹³ , ¹⁴ The PROTECTED‐TAVR (Stroke Protection With Sentinel During Transcatheter Aortic Valve Replacement) trial reported clinical events up to 3 days postprocedure. ¹⁵ Secondary clinical end points included disabling stroke and all‐cause mortality. Neurocognitive and neuroimaging end points included worsening of the National Institutes of Health Stroke Scale (NIHSS) and the Montreal Cognitive Assessment (MoCA) scores from baseline at discharge, presence of new ischemic lesions, and total lesion volume on DW‐MRI, respectively. Safety outcomes included any major or minor vascular complications as defined by the included trials and stage 2 or 3 acute kidney injury (AKI). In the included studies all‐cause stroke and disabling stroke was defined according to Valve Academic Research Consortium‐2 ⁷ , ⁸ , ¹¹ , ¹² , ¹³ , ¹⁴ and Neurologic Academic Research Consortium definitions. ¹³ , ¹⁴ , ¹⁵ Other secondary end points in the included studies were defined according to Valve Academic Research Consortium‐2 definitions. Six of the included studies ⁷ , ⁸ , ¹¹ , ¹² , ¹³ , ¹⁴ reported major or minor vascular complications related to the procedure, but the definition of a vascular complications in PROTECTED‐TAVR was limited to the CEPD access, ¹⁵ hence we did not include it in the analysis for major or minor vascular complications.

Based on the mechanism of protection of CEPD, we performed subgroup analysis for all the included outcomes by categorizing the CEPD into 2 major groups: filter devices and deflection devices. The filter devices Sentinel cerebral embolic protection device (Claret Medical/Boston Scientific, Marlborough, MA) and Claret Montage Dual Filter System (Claret Medical Inc., Santa Rosa, CA) share a similar device design and are both positioned in the brachiocephalic and the left common carotid arteries, with the only exception being that the third generation Sentinel device has a new ergonomic device handle and additional 5‐cm catheter working length compared with its precursor the second generation Claret Montage device. ¹⁶ Thus filter group included studies ⁷ , ⁸ , ¹¹ , ¹⁵ that used these as their CEPD. The deflection devices TriGUARD 3 and TriGuard HDH (Keystone Heart, Tampa, FL) are positioned in the aortic arch across the brachiocephalic, left common carotid, and subclavian arteries; the deflection group included studies ¹² , ¹³ , ¹⁴ using these as their CEPD.

Statistical Analysis

A random‐effects model using the Mantel Haenszel method was used to calculate risk ratio (RR) and standardized mean difference with 95% CIs for all categorical and continuous variables. ¹⁷ To account for small counts, exact meta‐analysis method was used. ¹⁸ Continuous variables reported as medians with interquartile range were converted to means and SDs using the method described by Hozo et al. ¹⁹ Results were reported using forest plots. In the statistical model, the heterogeneity was quantified by the variance of the random effect. Heterogeneity across studies was evaluated using the Higgins I ² statistic (25%–50%=mild heterogeneity, 50%–75%=moderate heterogeneity, and >75%=severe heterogeneity). Funnel plots were used to detect publication bias. A P value <0.05 indicated statistical significance. Sensitivity analysis of primary outcome was performed by excluding REFLECT I (A Randomized Evaluation of the TriGuard Embolic Deflection Device to Reduce the Impact of Cerebral Embolic Lesions After Transcatheter Aortic Valve Implantation) to assess robustness of results. This pooled analysis was performed using Revman 5.4.1.

A random‐effects meta‐regression model was used to evaluate the association of baseline mean age, diabetes, prior history of cerebrovascular accident, Society of Thoracic Surgeons score, atrial fibrillation, sex, and type of heart valve on the primary outcome using STATA.

Results

Our initial literature search yielded 1392 results, out of which 664 papers were excluded on initial review for screening for duplicates. Of the remaining 728 articles, 708 articles were excluded by the application of exclusion criteria described (Figure 1) and 20 full text articles were selected for detailed assessment. Seven RCTs met inclusion criteria and were included in this analysis. ⁷ , ⁸ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ A summary of the included studies is depicted in Table 1.

Table 1.

Summary of Included Studies

Study author, y	Trial	Study design	Follow‐up, d	Group (n)	TAVR valve type	Device success, %	Procedure success, %	Mean fluoroscopy time, min	Procedure time, min	Stroke definition	Device type	Primary outcome
Lansky et al (2015) 13	REFLECT‐ I	RCT	30	CEPD (141)	CoreValve (32.4%), Sapien (62.5%)	93.40%	89.70%	NR	NR	VARC‐2, NeuroARC	TriGuard	Hierarchical composite of (1) all‐cause mortality or any stroke, (2) National Institutes of Health Stroke Scale score worsening or Montreal Cognitive Assessment worsening, and (3) total volume of cerebral ischemic lesions
				Control (63)	CoreValve (35.5%), Sapien (59.7%)	NA	90.30%	NR	NR
Haussig et al (2016) 11	CLEAN‐TAVI	RCT	30	CEPD (50)	Medtronic CoreValve	92%	90%	17.4	72.1	VARC‐2	Claret Montage Dual Filter	Numerical reduction in new positive postprocedural DW‐MRI brain lesions relative to baseline at 2 days after TAVR
				Control (50)		NA	NA	14.4	54.1
Van Mieghem et al (2016) 8	MISTRAL‐C	RCT	30	CEPD (32)	Edwards SAPIEN 3 (54%), Edward SAPIEN XT (15%), Medtronic CoreValve (25%)	93.80%	NR	NR	NR	VARC‐2	Sentinel	New cerebral lesions by DW‐MRI 5–7 days after TAVR
				Control (33)		NA	NR	NR	NR
Kapadia et al (2017) 7	Sentinel	RCT	30	CEPD (121)	Edwards SAPIEN 3 (55.8%), Edwards SAPIEN XT (17.5%), CoreValve Evolut R (24.2%)	NR	NR	16	80.75	VARC‐2	Sentinel	Occurrence of MACCE at 30 days compared with a historical performance goal
				Control (119)	Edwards SAPIEN 3 (53.4%), Edwards SAPIEN XT (16.9%), CoreValve Evolut R (23.7%)	NR	NR	15	68
Nazif et al (2021) 14	REFLECT II	RCT	30	CEPD (157)	Edwards SAPIEN (60.5%), Medtronic CoreValve (35.3%)	71%	69.70%	NR	NR	VARC‐2, NeuroARC	TriGuard	Composite of all‐cause mortality, stroke, life‐threatening or disabling bleeding, stage 2 or 3 acute kidney injury, coronary artery obstruction requiring intervention, major vascular complication, and valve‐related dysfunction requiring intervention and was compared with a historical performance goal
				Control (57)	Edwards SAPIEN (62.4%), Medtronic CoreValve (36.9%)	NA	NR	NR	NR
Lansky et al (2021) 12	DEFLECT III	RCT	30	CEPD (46)	Edwards SAPIEN/XT/3 (63.5%), Medtronic CoreValve (31%)	88.90%	100	28.4	NR	VARC‐2	TriGuard	Composite of the following MACCE: all‐cause mortality, all stroke (disabling and nondisabling), life‐threatening (or disabling) bleeding, acute kidney injury (stage 2 or 3), and major vascular complications
				Control (39)		NA	100	18.8	NR
Kapadia et al (2022) 15	PROTECTED‐ TAVR	RCT	3	CEPD (1501)	Edwards SAPIEN 3 (64.3%)	94.40%	NR	NR	NR	NeuroARC	Sentinel	Stroke within 72 hours after TAVR or before discharge (whichever came first) in the intention‐to‐treat population
				Control (1499)	Edwards SAPIEN 3 (63.7%)	NA	NR	NR	NR

CEPD indicates cerebral embolic protection device; CLEAN‐TAVI, Claret Embolic Protection and TAVI (Transcatheter Aortic Valve Implantation); DW‐MRI, diffusion‐weighted magnetic resonance imaging; MACCE, major adverse cardiac and cerebrovascular events; MISTRAL‐C, MRI (Magnetic Resonance Imaging) Investigation in TAVI with Claret; NA, not applicable; NeuroARC, Neurologic Academic Research Consortium; NR, not reported; PROTECTED‐TAVR; Stroke Protection With Sentinel During Transcatheter Aortic Valve Replacement; RCT, randomized controlled trial; REFLECT; A Randomized Evaluation of the TriGuard Embolic Deflection Device to Reduce the Impact of Cerebral Embolic Lesions After Transcatheter Aortic Valve Implantation; TAVR, transcatheter aortic valve replacement; and VARC‐2, Valve Academic Research Consortium‐2.

Pooled analysis from these 7 studies resulted in a total of 4016 patients with a mean age of 81 ± 1.98 years. The duration of follow‐up ranged from 3 days to 30 days with a mean follow‐up duration of 26.1 days and a median follow‐up of 30 days. Three RCTs ⁶ , ⁹ , ¹⁴ used the Food and Drug Administration‐approved Sentinel cerebral embolic protection device, 3 RCTs ¹² , ¹³ , ¹⁴ used the TriGuard embolic deflection devices (TriGUARD 3 and TriGuard HDH), and 1 RCT ¹¹ used the Claret Montage Dual Filter System as their CEPD. The baseline characteristics of the included cohorts are listed in Table 2. Publication bias using funnel plots and the risk of bias assessment according to Cochrane Risk of Bias are described in Figures S2–S11.

Table 2.

Baseline Characteristics of the Included Cohorts

Study author, y	Trial	Group, n	Mean age, y	Male, %	BMI	Hypertension, %	Diabetes, %	A.Fib, %	History of stroke, %	History of CAD, %	History of CKD, %	EuroScore II	STS score
Lansky et al (2015) 13	REFLECT‐I	CEPD (141)	79.8	56.7	NR	NR	42.9	33.1	13.1	NR	19.9	4.8	4.6
		Control (63)	81.5	66.7	NR	NR	31.7	25.8	11.3	NR	17.7	5.5	4.8
Haussig et al (2016) 11	CLEAN‐TAVI	CEPD (50)	80	42	NR	88	40	NR	2	52	NR	NR	5.6
		Control (50)	79.3	44	NR	94	50	NR	6	50	NR	NR	5.2
Van Mieghem et al (2016) 8	MISTRAL‐C	CEPD (32)	82	53	NR	66	13	29	19	NR	NR	NR	4.6
		Control (33)	82	51	NR	70	27	27	18	NR	NR	NR	5.8
Kapadia et al (2017) 7	Sentinel	CEPD (121)	83.1	47.9	27	NR	40.5	34.7	4.1	50.4	NR	NR	5.6
		Control (119)	85	51.3	27	NR	37.8	30.3	5	55.5	NR	NR	6.6
Nazif et al (2021) 14	REFLECT II	CEPD (157)	80.3	54.8	NR	NR	39.1	28	10.8	NR	NR	3.76	NR
		Control (57)	78.1	61.4	NR	NR	40.4	29.8	3.5	NR	NR	3.59	NR
Lansky et al (2021) 12	DEFLECT III	CEPD (46)	82.5	43.5	NR	80.4	21.7	21.7	13.3	NR	23.9	10.1	6.3
		Control (39)	82.3	48.7	NR	71.8	23.1	35.9	17.9	NR	25.6	7.2	7.4
Kapadia et al (2022) 15	PROTECTED‐TAVR	CEPD (1501)	78.9	58	NR	87.1	33.4	34.1	7.6	56.9	NR	NR	3.3
		Control (1499)	78.9	62.2	NR	87.6	34.8	31.4	8.2	58.9	NR	NR	3.4

AFib indicates atrial fibrillation; BMI, body mass index; CAD, coronary artery disease; CEPD, cerebral embolic protection device; CKD, chronic kidney disease; CLEAN‐TAVI, Claret Embolic Protection and TAVI (Transcatheter Aortic Valve Implantation); EuroScore II, European System for Cardiac Operative Risk Evaluation II; MISTRAL‐C, MRI (Magnetic Resonance Imaging) Investigation in TAVI with Claret; NA, not applicable; NR, not reported; PROTECTED‐TAVR; Stroke Protection With Sentinel During Transcatheter Aortic Valve Replacement; REFLECT; A Randomized Evaluation of the TriGuard Embolic Deflection Device to Reduce the Impact of Cerebral Embolic Lesions After Transcatheter Aortic Valve Implantation; and STS, Society of Thoracic Surgeons.

All‐cause stroke was reported in all 7 studies. For the primary clinical end point of all‐cause stroke, the RR with CEPD was 15% lower than without CEPD (RR, 0.85 [95% CI, 0.61–1.18]; P=0.34; I ² = 0%; Figure 2). Although not statistically significant, this indicates that the RR of CEPD to no‐CEPD in this sample was 0.85, with the true value represented in a range of 0.61 (risk decreasing 39%) to 1.18 (risk increasing 18%) in the patient population. Sensitivity analysis after excluding REFLECT I showed similar results (Figure S12). For the secondary clinical end point of disabling stroke, although not statistically significant, the RR with CEPD was 42% lower than without CEPD (RR, 0.58 [95% CI, 0.29–1.13]; P=0.11, I ²=0%; Figure 3), with the true value represented in a range of 0.29 (risk decreasing 71%) to 1.13 (risk increasing 13%). For all‐cause mortality, the RR with CEPD was 3% higher than without CEPD, with true value represented in a range of 0.49 (risk decreasing 51%) to 2.17 (risk increasing 117%) (RR, 1.03 [95% CI, 0.49–2.17]; P=0.93, I ²=0%; Figure 4), although this did not reach statistical significance. Worsening of NIHSS score from baseline at discharge was reported in 5 RCTs. The RR with CEPD was 21% higher than without CEPD (RR, 1.21 [95% CI, 0.77–1.91]; P=0.41; I ²=0%; Figure 5). Although not statistically significant, this indicates that the RR of CEPD to no‐CEPD in this sample is 1.21, with the true value represented in a range of 0.77 (risk decreasing 23%) to 1.91 (risk increasing 91%). Worsening of MoCA score from baseline at discharge was reported in 4 RCTs. The RR with CEPD was 10% lower than without CEPD (RR, 0.90 [95% CI, 0.58–1.38]; P=0.63; I ²=28%; Figure 6). Although not statistically significant, this indicates that the RR of CEPD to no‐CEPD in this sample is 0.90, with the true value represented in a range of 0.58 (risk decreasing 42%) to 1.38 (risk increasing 38%). Also, there was no statistically significant difference in the presence of new ischemic lesions (RR, 0.99 [95% CI, 0.95–1.04]; P=0.70; I ²=0%; Figure 7) and total lesion volume (standardized mean difference, ‐0.53 [95% CI, −1.20 to 0.14]; P=0.12; I ²=95%; Figure 8) seen on DW‐MRI in those with and without CEPD. Safety outcomes including major or minor vascular complications and stage 2 or 3 AKI did not show any statistically significant differences between the CEPD and no‐CEPD groups. The RR for major or minor vascular complications (RR, 1.17 [95% CI, 0.61–2.24]; P=0.63, I ²=56%) was 17% higher in the CEPD group (Figure 9) and the RR for stage 2 or 3 AKI groups (RR, 0.96 [95% CI, 0.44–2.11]; P=0.93, I ²=0%; Figure 10) was almost similar between the 2. However, it must be noted that both these outcomes had a wide range of CIs. For major or minor vascular complications, the CEPD group showed an RR of 1.17 with true value ranging from 0.61 (risk decreasing 39%) to 2.24 (risk increasing 124%). Similarly for stage 2 or 3 AKI, the RR of CEPD to no‐CEPD was 0.96, with true value ranging from 0.44 (risk decreasing 56%) to 2.11 (risk increasing 111%).

Figure 2. Pooled risk ratio and 95% CI for all‐cause stroke.

CEPD indicates cerebral embolic protection devices; and M‐H, Mantel–Haenszel.

Figure 3. Pooled risk ratio and 95% CI for disabling stroke.

CEPD indicates cerebral embolic protection devices; and M‐H, Mantel–Haenszel.

Figure 4. Pooled risk ratio and 95% CI for all‐cause mortality.

CEPD indicates cerebral embolic protection devices; and M‐H, Mantel–Haenszel.

Figure 5. Pooled risk ratio and 95% CI for NIHSS Worsening.

CEPD indicates cerebral embolic protection devices; M‐H, Mantel–Haenszel; and NIHSS, National Institutes of Health Stroke Scale.

Figure 6. Pooled risk ratio and 95% CI for MoCA Worsening.

CEPD indicates cerebral embolic protection devices; M‐H, Mantel–Haenszel; and MoCA, Montreal Cognitive Assessment.

Figure 7. Pooled risk ratio and 95% CI for presence of new ischemic lesions.

CEPD indicates cerebral embolic protection devices; and M‐H, Mantel–Haenszel.

Figure 8. Standardized mean difference and 95% CI for total lesion volume on DW‐MRI.

CEPD indicates cerebral embolic protection devices; DW‐MRI, diffusion‐weighted magnetic resonance imaging; and M‐H, Mantel–Haenszel.

Figure 9. Pooled risk ratio and 95% CI for major or minor vascular complications.

CEPD indicates cerebral embolic protection devices; and M‐H, Mantel–Haenszel.

Figure 10. Pooled risk ratio and 95% CI for Stage 2 or 3 acute kidney injury.

CEPD indicates cerebral embolic protection devices; and M‐H, Mantel–Haenszel.

Subgroup analysis by mechanism of CEPD protection did not show any statistically significant difference in all‐cause stroke, all‐cause mortality, NIHSS score worsening, MoCA score worsening, new ischemic lesions, total lesion volume, major or minor vascular complications, and stage 2 or 3 AKI. However, we did find statistically significant difference in disabling stroke in filter based CEPD group (RR, 0.42 [95% CI, 0.20–0.88]; P=0.02, I ²=0%) over defection devices (RR, 2.20 [95% CI, 0.48–10.11]; P=0.31, I ²=0%; Figure 3). This indicates that the filter‐based CEPD showed 58% reduction in disabling stroke, with the true value represented in a range of 0.88 (risk decreasing 12%) to 0.20 (risk decreasing 80%) in this population. Based on the incidence of disabling stroke in the CEPD group (0.56%) versus the no‐CEPD group (1.4%), we calculated the number needed to treat to prevent an additional disabling stroke to be 123. Interestingly, the subgroup analysis also showed contrasting trends between filter and deflection devices specifically for all‐cause stroke, total lesion volume on DW‐MRI, and safety outcomes for major or minor vascular complications and stage 2 or 3 AKI. To account for small counts, the results using exact method meta‐analysis are reported separately in Table S1.

Random‐effects meta‐regression models demonstrated no significant interaction of baseline mean age, diabetes, prior history of cerebrovascular accident, Society of Thoracic Surgeons score, atrial fibrillation, sex, and type of heart valve with risk of all‐cause stroke among patients receiving CEPD (Figures S13–S20).

Discussion

Periprocedural stroke after TAVR continues to be an unpredictable and devastating complication leading to a prolonged hospital stay, a reduced chance of returning to one's baseline level of independence, and an increased mortality rate. ²⁰ It not only increases the cost of index hospitalization but also poses extra economic burden on annual health care costs owing to the long‐term chronic needs in these patients. ²¹ Thus, stroke prevention is critical in patients undergoing TAVR and becomes even more imperative as TAVR expands to the asymptomatic and younger population carrying a lower surgical risk. To that end, CEPDs were developed to mitigate the risk of various cerebrovascular events, including clinically overt disabling and nondisabling strokes. Given the prior evidence that confirms debris capture by CEPD, ²² it was expected that CEPD would reduce clinical complications and mortality rates in patients undergoing TAVR; however, these trials have been underpowered for various clinical and neurological outcomes. We conducted an extensive literature search to identify randomized trials evaluating CEPD use in TAVR and combined results from 7 trials involving >4000 patients in this meta‐analysis.

Our meta‐analysis, which included the recently published, largest RCT of CEPD, showed no statistically significant difference between CEPD and no‐CEPD groups in terms of (1) clinical outcomes including all‐cause stroke, disabling stroke, and all‐cause mortality; (2) neurocognitive end points such as worsening of the NIHSS and MoCA from baseline to discharge; or (3) neuroimaging parameters (presence of new ischemic lesions and total lesion volume on DW‐MRI). Importantly, our analysis showed that the use of CEPD during TAVR was safe and was not associated with an increased risk of procedural complications (ie, major or minor vascular complications or worsening AKI). Although several propensity‐matched analyses ²³ , ²⁴ and the largest RCT ¹⁵ using filter‐based devices demonstrated a reduction in disabling stroke in patients with CEPD compared with those without CEPD, we did an exploratory subgroup analysis combining studies that used the filter‐based mechanism. Our analysis showed a statistically significant reduction in disabling stroke using filter devices; however, these findings should be interpreted with caution as these results are primarily driven by the PROTECTED‐TAVR trial, which had the largest sample size compared with other RCTs. Additionally, disabling stroke was one of multiple secondary end points in most of these RCTs included in our analysis. Thus, this observation can act as a hypothesis that should be explored in future studies. Our results support the finding of contemporary TVT registry analysis by Butala et al that showed no significant association between CEPD use and in‐hospital stroke in their primary instrumental variable analysis; however, they observed modest reduction in risk of stroke in secondary propensity‐weighted analysis. ²⁵

Although there are several theories that explain the failure of CEPD devices, in our opinion, factors that may contribute to the absence of notable clinical benefit with these devices can be largely categorized into patient‐related, periprocedural, trial‐design related, and device‐related factors. Individual patient factors like age, female sex, chronic kidney disease, and history of previous cerebrovascular disease have been associated with increased periprocedural stroke in TAVR. ²⁶ , ²⁷ , ²⁸ Although these factors may help us risk stratify our patients at risk for stroke, studies have been unable to identify consistent clinical predictors or a high‐risk population that may benefit from the selective use of CEPD, because many of these factors are inherently associated with higher risk of stroke outside of procedure‐related embolic phenomenon. So, even if we try to reduce embolism‐related stroke events through the use of CEPD, these residual factors may still contribute to stroke risk on short‐ or long‐term follow‐up. ²⁹ In addition, delayed periprocedural events that increase the risk of stroke including new onset atrial fibrillation, valve thrombosis, and plaque embolization would be unlikely to be affected by the intraprocedural use of CEPD. ³⁰ Furthermore, in real‐world practices unfavorable anatomy may preclude CEPD use in some patients especially those with poor radial access or torturous great vessels with high atherosclerotic burden. Therefore, it is possible that many of the patients who stand to benefit from CEPD the most might not even be candidates for these devices.

From a trial design perspective, it is important to note that sample size that was estimated to prove superiority of CEPD for one of the biggest trials (PROTECTED‐TAVR) was based on the assumption that the incidence of stroke was 4% in the control group and 2% in the CEPD group. But larger registry studies show that the rates of in‐hospital and 30‐day strokes after TAVR have decreased in the past decade to 1.6% and 2.3%, respectively. ³ Thus, to prove statistical benefit of these devices we need studies with greater event rates or larger sample sizes. The ongoing BHF PROTECT‐TAVI (British Heart Foundation Randomized Trial of Routine Cerebral Embolic Protection in Transcatheter Aortic Valve Implantation ISRCTN16665769) aims to recruit 7730 patients and is expected to be completed by 2026. The results from this large‐scale RCT may offer more insight on the benefit of CEPD in preventing strokes and improving mortality.

Finally, device‐related factors like improper sizing relative to the aortic arch, pore size larger than microemboli, unsuccessful positioning, and incomplete protection of certain brain areas could explain limited benefit of CEPD. The filter devices like Sentinel and Claret Montage Filter System cover brachiocephalic artery and the left common carotid artery, leaving the left vertebral artery exposed. ¹⁵ Theoretically, this problem could be solved by deploying an additional device in the left vertebral artery. The concomitant use of Wirion filter (Allium Medical, Inc.; Caesaria, Israel), which was used as an embolic protection system during carotid artery stenting, had been proposed in the literature to provide coverage to the left vertebral artery. ³¹ Although this particular device was recalled due to safety concerns, there are other filters that could be sized for the left vertebral system. The TriGuard deflection devices were designed to provide full brain coverage as these are deployed across the aortic arch covering all the major aortic branches. Based on the deployment site, they may offer the added benefit of avoiding atherosclerotic or calcific disease along the greater curve of the arch or at the ostia of the arch branch vessels. Yet in the REFLECT I trial, only 57.3% of TriGuard HDH recipients showed complete cerebral coverage for the entirety of the TAVR procedure. Furthermore, the device interfered with the TAVR delivery system in approximately 9% of cases. ¹³ Other design‐associated challenges of TriGuard devices include relatively larger access site compared with the Sentinel embolic protection device, ³² requirement of a dedicated stabilizer in some of these devices, and wider surface area in some of the newer devices. ³³ These drawbacks of TriGuard devices might explain why there were no statistically significant differences in the clinical outcomes, including disabling stroke, in our exploratory analysis.

Besides the Sentinel and TriGuard devices, several new CEPD are currently under development and are presently either in preclinical or feasibility testing stage. Several of these devices like the Emblok, Point‐Guard, Emboliner, ProtEmbo, Embolisher, and Fliterlex would offer protection to all supra‐aortic arteries and would have additional features like dual layer mesh filters and ability to deflect microparticles as small as 60 microns. The Emblok device in particular has been designed with an integrated pigtail catheter to aid for precise valve implantation and to minimize contrast injection. The Fliterlex device has been designed to provide full‐body embolic protection as it extends beyond aortic arch into the descending aorta. ³⁴ It should be noted that the results of our analysis cannot be generalized to these newer devices as they were not studied in any of the included trials.

Although our study did not find a reduction in the primary clinical end point of all‐cause stroke, there may still be hope for cerebral protection during TAVR. Our study found that CEPD was safe and, in the case of the filter devices, may be associated with a lower rate of major disabling stroke. Furthermore, the subgroup analysis according to device type demonstrates the impact of device technology on outcomes of CEPD as shown by favorable trends for filter‐based devices when evaluating all cause stroke, total lesion volume on DW‐MRI, vascular complications, and AKI. Thus, there is potential to improve CEPD efficacy by evolving technologies particularly for filter‐based devices. As CEPD technology evolves and we await more evidence on their efficacy, the decision to use CEPD in current scenario should be carefully weighed against the potential risks, benefits, device safety, and device cost.

Limitations

The main limitation of our analysis is the limited number of published studies. Therefore, we need to be cautious about drawing a true conclusion or association. However, the statistical power can be improved using a certain level of information borrowing when historical or similar studies are available. Second, some of the results of this analysis, particularly the reporting of disabling stroke in the subgroup of studies that used the filter devices, are predominantly driven by the PROTECTED‐TAVR trial, the largest RCT in this analysis. Third, the included studies varied by device type, which could potentially explain the high heterogeneity for major or minor vascular complications. Fourth, one of the included studies (REFLECT I) was suspended prematurely due to discontinuation of TriGuard HDH. We performed a sensitivity analysis after excluding results from REFLECT I, which did not change the primary outcomes of our analysis. Also, we acknowledge the lack of registration of our analysis as one of the other limitations. Finally, though a random effects model was used to account for heterogeneity, it does not eliminate the subtle differences in various trials including but not limited to different types of TAVR devices or their implantation techniques, differences in baseline characteristics like comorbid conditions, background therapy including antiplatelet therapy or anticoagulation strategies, degree of atherosclerosis, social characteristics, and difference in definition of similar sounding outcomes or different outcomes.

Conclusions

In conclusion, our updated meta‐analysis encompassing the totality of the randomized clinical trial data including the most recent PROTECTED‐TAVR study showed CEPD use in TAVR was not associated with a statistically significant benefit in clinical, neurocognitive, or neuroimaging outcomes.

Sources of Funding

None.

Disclosures

None.

Supporting information

Data S1.