Despite substantial device iteration and procedural refinements over the past decade, the incidence of stroke after TAVI remains 2-3% and is one of the most feared complications of TAVI. In recent years, several cerebral embolic protection devices have been introduced in an effort to reduce embolic debris to the brain and, as a consequence, to reduce the incidence of procedure-related stroke. Although these devices are intuitively appealing (and have been shown to capture embolic debris in many cases), there remains considerable uncertainty as to whether they truly improve clinical outcomes.

Currently, the only approved cerebral embolic protection device in the U.S. is the Sentinel Device—a dual filter device that is delivered via the right radial artery and deploys 2 nitinol filters in the innominate and left common carotid arteries. To date, this device has been studied in 4 randomized trials. The first 2 trials (CLEAN-TAVI, SENTINEL Pivotal) were small, mechanistic studies that focused on demonstrating that the Sentinel device can reduce the extent of new embolic lesions by MRI. Although CLEAN-TAVI demonstrated benefit, no benefit was seen in the larger, multicenter Sentinel Pivotal trial.

More recently, 2 randomized trials that were designed to test clinical benefit have been performed. The PROTECTED-TAVR trial, a 3000 patient RCT performed in the United States, randomized 3000 U.S. patients to Sentinel vs. no Sentinel and failed to demonstrate a benefit for the primary endpoint- any stroke at 72 hours (2.2% vs. 2.8%, p=NS), but there was a suggestion of benefit for the secondary endpoint of disabling stroke, which occurred in 0.5% of the CEPD group and 1.3% of the no CEPD group (p=0.02). The BHF PROTECT-TAVI trial was an even larger trial, conducted in the UK. In that trial, there was no difference between Sentinel and no Sentinel for any stroke (the primary endpoint) or disabling stroke (1.2% vs. 1.4%). A pooled analysis of both trials also failed to demonstrate benefit on either endpoint.

One of the main issues with these trials is that stroke (and disabling stroke) are rare events. As such, both trials (and even the pooled analysis) were ultimately underpowered to demonstrate modest benefits. To address this concern, Butala and colleagues performed an observational study using an instrumental variable analysis (an approach that can provide true causal inference) using data from >400,000 patients from the US TVT registry. In that study, which used instrumental variable analysis to provide true causal inference, there did appear to be a benefit of Sentinel on disabling stroke although the magnitude of benefit was small (absolute risk difference 0.12%, p=0.05; NNT ~800). There was no benefit on non-disabling stroke, however. Subgroup analysis demonstrated several subgroups that seemed to derive substantial benefit from Sentinel including patients with prior stroke and patients undergoing valve-in-valve TAVR (absolute risk differences 0.6-0.7%; NNT ~160).

Based on these findings, it appears that cerebral protection with the Sentinel device does reduce disabling stroke, but the magnitude of benefit is very small for the typical TAVI patient. What is unclear is why the benefit of the Sentinel device (which captures debris in most patients is so small). Potential explanations include device-induced harm, protection of only 3 of 4 great vessels, or the fact that the mechanism of benefit is conversion of larger strokes to smaller strokes, making any stroke a suboptimal endpoint. What is clear is that there is a clear need for additional embolic protection devices that overcome some of the limitations of Sentinel. Several of these devices are currently being tested in mechanistic trials in the United States, and we anxiously await their results to see whether they are important steps forward in improving the safety and long-term outcomes of TAVI.