Transcatheter aortic valve replacement (TAVR) has become the most widely used treatment in low to high-risk patients with aortic valve disease, exceeding all other forms of Surgical Aortic Valve Replacement (SAVR) in 2019 [1]. With advances in device technology and increased operator experience, TAVR has become a proven safe and effective treatment method for aortic valve replacement (AVR) [2]. Although these advances have reduced many of the early complications seen with TAVR, such as vascular complications and paravalvular leaks, concern persists over the increased risk of cerebrovascular accidents secondary to embolization of aortic and aortic valve debris during TAVR. Stroke remains an important cause of morbidity and mortality in patients undergoing TAVR [3, 4], in spite of procedure and device evolution [5].

Post-TAVR stroke remains a serious complication associated with increased mortality [6, 7, 8]. The 30-day risk of stroke following TAVR ranges between 1 to 5.5% [3, 9], and one-year stroke rates between 4.3 to 8.2% [8, 9], with the occurrence of stroke associated with a 3.5-fold increase in the risk of death in the first month post-TAVR [3, 5, 10, 11]. Even with the evolution of new TAVR devices, review of the Society of Thoracic Surgeons (STS)/American College of Cardiology (ACC) Transcatheter Valve Therapy (TVT) Registry has shown only slight declines in perioperative and 30-day stroke rates [1], and has further demonstrated stroke risk to be independent of TAVR physician experience [12, 13].

To mitigate TAVR stroke risk, cerebral protection devices were developed to help protect the brain from embolic strokes. These devices deploy temporary filters in the aortic arch and/or great vessels (i.e., brachiocephalic and left common carotid arteries) to capture embolic debris dislodged during TAVR. Among these devices, the Sentinel Cerebral Protection System (Sentinel CPS) is the only Food and Drug Administration (FDA) approved device for use with TAVR in the USA [14, 15], with conflicting evidence surrounding its benefit in stroke reduction [3, 5, 14].

With procedural success rates over 90%, the Sentinel CPS has an excellent safety profile, with low rates of mortality, neurological events, and major adverse cardiac and cerebrovascular events (MACCE) in randomized clinical trials (RCT) [3, 5, 15, 16, 17, 18]. However, the use of this device in real-world settings has not been well-studied, and no guidelines exist on CPS device usage and patient selection in TAVR. Thus, the main objective of this study was to evaluate and compare the clinical characteristics of patients who receive the Sentinel device to concurrent patients who do not in a real-world setting. The primary analysis objective of this study is to evaluate the association of Sentinel CPS use in TAVR on perioperative cerebrovascular accidents (CVA), as well as the occurrence of CVA and a composite outcome of death, CVA, and bleeding/vascular events in consecutive patients undergoing TAVR at 1 year.

This is a real-world study, performed at a single center on a consecutive series of patients undergoing TAVR. With no existing guidelines on CPS device use and patient selection in TAVR, the main objective of our study was to compare the characteristics of patients who tend to receive Sentinel CPS devices with those who do not in a “real-world” setting. We also examined the occurrence of post-TAVR complications between groups, namely in-hospital and one-year stroke. In this study, we found patients who received a Sentinel CPS device were more likely to be healthier and deemed less complex when compared to concurrent patients who did not receive a device. In addition, we demonstrated no statistically significant association between device use and decreased in-hospital/30-day post-TAVR stroke or in the occurrence of our composite endpoint (death, stroke, and bleeding/vascular events) at one-year.

With RCT demonstrating similar [27] or superior [28] clinical outcomes of TAVR to SAVR in low surgical risk patients, the extension of TAVR to young and low-operative risk patients has rapidly grown [29, 30]. These findings highlight the necessity to ensure TAVR remains safe, with a low risk of adverse clinical events, especially in a younger population with longer life expectancies. Specifically, CVA post-TAVR is a devastating complication and remains non-negligible ranging up to 5.5% at 30-days and 8.2% at 1 year [3, 8, 9, 31]. Major CVA post-TAVR is a known independent predictor of morbidity and mortality [15, 32, 33], with moderate to severe permanent disability leading to dependency in up to 40% of survivors and a further 80% facing social isolation and significant financial stress [29, 34, 35]. These patients represent a significant societal healthcare burden and contribute to increased financial strain on healthcare systems.

More than half of post-TAVR CVA are due to procedure-related emboli [29], which led to the development of embolic protection devices like the Sentinel CPS. To date, clinical studies have not conclusively determined the efficacy of these devices in preventing CVA [29, 36, 37]. The Sentinel CPS is the most studied embolic protection device in TAVR, and the most commonly used device worldwide [32]. Reviewing results from the three largest RCTs on this device totaling 528 patients (MISTRAL-C, CLEAN-TAVI, and SENTINEL), none were able to demonstrate a reduction in the clinical endpoint of mortality and/or stroke [15, 16, 18]. Lacking statistical power secondary to small sample sizes and a low incidence of post-TAVR CVA, none of these trials were able to use CVA as a primary endpoint [29, 32, 38]. Several meta- and propensity matched-analyses have also tried to examine Sentinel CPS efficacy [29], with available meta-analyses yielding similar results (due to significant study overlap) and failing to demonstrate device efficacy [14, 39, 40]. Propensity matched analyses using data from multiple large registries have also been unable to consistently demonstrate significant reductions in stroke with Sentinel CPS use [36, 37, 41, 42].

Although our study was small and not randomized, our findings are consistent with current literature on the lack of Sentinel CPS efficacy on stroke reduction, and representative of real-world TAVR data. A total of 2 in-hospital (0.8%) and 4 one-year (1.7%) CVA occurred post-TAVR in our study. Despite a lower numerical incidence of CVA in the Sentinel CPS group at both time points, unadjusted analyses demonstrated no statistically significant difference between groups (0 vs. 2; p = 0.198 and 1 vs. 3; p = 0.233, respectively). To account for baseline differences and non-randomization, a propensity score was developed using all factors in Table 1 and Table 2, which was then used in multivariate analysis. Even after adjustment, the adjusted hazard ratio for the CPS group still showed no difference in stroke incidence at 1 year (p = 0.132), with no variables included in the analysis reaching statistical significance. Like other studies though, our small sample size and low incidence of CVA, reduced our statistical power and ability to demonstrate any reduction in stroke. Our overall one-year stroke rates (1.7%) were even lower than 30-day stroke rates ($>$2.3%) reported by large TAVR registries like the TVT [6, 35].

While unadjusted analysis of our composite endpoint demonstrated a significantly lower occurrence in the CPS group (p = 0.034), after adjustment this was no longer true (p = 0.289, Table 3). Interestingly, being on immunosuppressant therapy and dialysis were both statistically significant major risk factors of the composite endpoint in TAVR patients (Table 3). This is not surprising, considering our composite endpoint included death and stroke. Overall mortality in dialysis patients is 10–20 times higher than in the general population, with a one-year survival rate of 30% in adults $>$65, almost doubling to 54% in patients $>$85 [43]. Furthermore, chronic kidney disease and dialysis are both known risk factors for cerebrovascular disease, and are associated with higher stroke rates, worse stroke severity, outcomes, and mortality [44]. Similarly, patients on immunosuppressant therapy have increased mortality rates [45], secondary to increased rates of infection [46] and malignancy [47, 48]. While low BMI (20 kg/m${}^2$) has been shown to be an independent predictor of increased short and long-term mortality in real-world patients undergoing TAVR [49], this finding was not demonstrated in our study as the mean BMI of all patients in our study was 29.2 $\pm$ 7 with no statistically significant difference in distribution between groups.

With no benefit in stroke reduction or the occurrence of our composite endpoint demonstrated with Sentinel CPS use, we focused on exploring why some patients were selected to receive the device while some were not. Comparing baseline and pre-TAVR procedural characteristics/patient status between groups, patients receiving the Sentinel CPS device were overall healthier at baseline and deemed less complex (Tables 1,2). No Sentinel CPS patients had significantly higher incidences of dialysis (p = 0.016), chronic lung disease (p = 0.012), prior TIA (p = 0.020) and pacemaker (p = 0.022), and were more likely to be on immunosuppressant therapies (p 0.001) when compared to the Sentinel CPS group. Pre-TAVR, no Sentinel CPS patients also had lower hemoglobin (p = 0.001) and albumin (p 0.0001) levels – both potent markers of frailty and mortality following TAVR [50], as well as higher creatinine levels (p = 0.016). Examining patient complexity, patients receiving Sentinel CPS also had significantly lower STS Risk Scores (p 0.0001) and TAVR operative risk assessments (p 0.0001).

With no guidelines available to aid decision making on CPS device use and patient selection, the data presented here suggest that at our institution, Sentinel CPS was used more frequently in men (p = 0.006) and patients who were healthier and less complex. These findings are not surprising, since as patient complexity increases, patients tend to develop more contraindications or relative contraindications to Sentinel CPS use. For example, patients with COPD may have had multiple arterial punctures for blood gases in the past, and thus have a difficult radial artery to access for device use. Alternatively, patients on dialysis or those with a history of prior TIA may be more likely to have significant carotid disease or aortic arch and great vessel calcification, precluding the use/deployment of the Sentinel CPS. Finally, men typically have larger radial arteries and less tortuosity, facilitating the deployment of Sentinel CPS when compared to women.

While the final decision rests with the operator, during the time period of this study our Valve team’s policy was to consider Sentinel CPS in all patients. Investigating the possible driving factors behind patient selection in our study revealed those receiving a device were likely chosen simply based on the ease and safety of device placement. Operators were more likely to deploy the Sentinel CPS in healthier patients, who inherently have lower baseline stroke risks, and were least likely to have derived any benefit from device placement. While the mechanism of benefit for Sentinel CPS has very strong theoretical plausibility, recent randomized data is consistent with our findings, suggesting the net effect on objective clinical endpoints such as stroke is non-significant [15, 51]. In the PROTECTED TAVR study, a multi-center, multi-country RCT comparing the effects of Sentinel use on the incidence of perioperative stroke reduction totaling 3000 patients, Kapadia et al. [51] also demonstrated no differences in stroke (Sentinel 2.3% vs. No Sentinel 2.9%, p = 0.30) or mortality (0.5% vs. 0.3%) at 72 hours or before discharge.

With accumulating evidence against device benefit, our institution has moved towards a “do no harm” approach first when selecting patients for device use. If a patient has anatomic or clinical factors that make device use a risk for complication, no device is used or even attempted. Instances where we recommend against Sentinel CPS use include: (1) Patients who have weakly palpable or occluded radial arteries that preclude arterial access; (2) Severe tortuosity in the subclavian precluding the ability to adequately deploy the device; (3) Patients with severe atherosclerosis or calcification in the brachiocephalic or left common carotid arteries, where manipulations in this area may cause greater risk than benefit; and (4) patients for whom the right radial is used for the pigtail catheter or a guide catheter for coronary protection due to inadequate femoral access.

In summary, with no definitive answer on device effectiveness on stroke reduction, or clear evidence of clinical benefit [29, 51] to date, the utility of Sentinel CPS use in TAVR remains questionable. Associated with substantial morbidity, longer hospital stays, increased mortality, readmission rates, and hospital costs [3, 6, 7, 8, 9, 52, 53], CVA post-TAVR are devastating complications and adversely affect the healthcare system. Despite these implications, it is unlikely Sentinel CPS use will become a clinical standard based on current evidence. With an already low occurrence of stroke post-TAVR, and significant device-associated costs (~$\$$3000 USD) that are not recuperated and compounded by a high number needed to treat (NNT: 125 to prevent 1 disabling stroke) [51], Sentinel CPS use may not be the answer to lowering stroke incidence in TAVR. Until further evidence supporting efficacy in stroke reductions becomes available, device use will likely remain operator and patient anatomy dependent, with some consideration to patient request and/or institutional budgets.

While our study was not randomized for patient assignment to Sentinel, several statistical control methods were used to allow comparison between groups. However, other potentially significant unmeasured confounders may still exist. Patient frailty represents one such confounder, as no formal measure of frailty was available in our study, and may have been an important unaccounted difference between groups. While our sample size was small, and therefore our results not adequately powered to compare event rates between patients in the Sentinel CPS group with those who did not receive one, our results do represent a consecutive series in a real-world clinical setting with control of the operators performing TAVR. Lastly, no routine neurologist assessment or neuroimaging surveillance was performed in our study to evaluate for CVA post-TAVR. Although this likely resulted in a lower perioperative CVA detection rate, it is unlikely to have significantly affected results of our study endpoints, especially considering the already exceedingly low occurrence of stroke (1.7%). Furthermore, all strokes reported in our study were diagnosed (and verified) using the recognized standardized VARC-3 definition criteria for stroke, eliminating clinical outcome definition variability.

In conclusion, our study demonstrates that the Sentinel CPS device is more likely to be used in patients who are less sick and complex compared with other TAVR patients. This is likely a consequence of healthier patients having safer and more amenable anatomy for device use. Device use was not significantly associated with a lower occurrence of perioperative stroke or reduction in the composite endpoint (death, stroke, bleeding, and vascular events) at one-year following TAVR. Further research involving randomization and larger multicenter studies is required to understand how CPS is utilized in the “real-world” and to identify its potential role in reducing stroke after TAVR.

The datasets generated and/or analyzed during the current study are not publicly available due to privacy/ethical restrictions, but are available from the corresponding author on reasonable request.

HJ, RS, and JG designed the research study. HJ, RS, LG, and HR performed the research and interpreted the data. RT, FK and AZ interpreted the data. All authors contributed to the writing of this manuscript. All authors contributed to editorial changes and revisions to the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

There is no Ethics approval number for this study. It is retrospective. The local database was approved by the Western Institutional Review Board (WIRB), and informed consent was waived for this study.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.