CardioExchange Archive

CardioExchange’s Harlan Krumholz interviews Jay Giri and Robert W. Yeh about their research group’s analysis of patient-registry data on embolic protection devices that are commonly used in carotid artery stenting. The study is published in JACC: Cardiovascular Interventions.

THE STUDY

Researchers used the NCDR’s CARE registry to analyze data from 12,135 consecutive carotid stent procedures performed from January 2007 through March 2012. Stents were used in conjunction with their corresponding FDA-approved embolic protection device (EPD) in 78% of cases. The rate of in-hospital death or stroke was nonsignificantly higher with the Precise/Angioguard system (2.5%) than with the Xact/Emboshield system (1.9%) or the Acculink/Accunet system (1.8%); the differences remained nonsignificant in adjusted analyses. These three devices were used in 73% of all procedures.

THE INTERVIEW

Krumholz: How much weight do you put on comparative effectiveness research that is derived from registry data?

Giri: Registries provide a valuable resource for comparative effectiveness research, particularly regarding questions that require large patient volumes to answer and that are unlikely to be addressed in randomized clinical trials. In those circumstances, registries (despite biases inherent in observational analyses) may provide the only available evidence and have the advantage of not being limited to highly selected centers and patient populations. Therefore, their conclusions may be more generalizable to everyday practice. Finally, registries can serve an important exploratory role in providing information for power estimates and patient enrollment numbers for RCTs.

Yeh: Observational comparisons of devices and therapies from large registries complement RCTs in several ways: (1) They yield data on the safety and effectiveness of new treatments in diverse populations, including groups traditionally underrepresented in RCTs; (2) they provide adequate sample sizes to detect small signals that RCTs are not powered to study; and (3) they can address comparisons that are important to physicians and patients but are less likely to be funded by industry or federal agencies.

Of course, registry analyses are susceptible to confounding if they do not try to identify and account for the various factors that may influence physicians’ selection of a particular treatment. I believe that the weight one places on any particular observational comparison from registry data should depend on the following factors: What is the nature of the comparison? Do prognostic patient characteristics strongly influence treatment selection in the given clinical setting? Does the registry collect data on those factors? Do the methods adequately address the lack of randomization? Thus, the validity of registry data must be judged on a case-by-case or a study-by-study basis. With respect to our study, the selection of carotid stenting systems is probably governed more by institutional contracts and by physician training and comfort than by prognostically important, unmeasured patient characteristics. We believe that this allows for a robust observational comparison of carotid stenting systems.

Krumholz: Should we be doing head-to-head trials of the different EPD devices?

Giri: Given that the devices were associated with quite small differences, I do not think a randomized trial comparing them would be a good use of limited resources for clinical investigation. Besides, given the in-hospital event rates we found, a head-to-head trial of the compared devices is very unlikely to occur, as it would take well over 10,000 patients. With the current reimbursement situation for carotid stenting, device makers have waning interest in even continuing their own postmarketing registries, let alone supporting a much more expensive RCT that runs the risk of reflecting poorly on their device.

Yeh: Traditional post-approval randomized comparisons of these devices that do not have substantive mechanistic differences don’t seem to be a good use of resources, particularly in the field of carotid artery disease, where the role of carotid stenting is still not well defined. There is much promise in doing more pragmatically designed randomized comparisons that can leverage existing registries to compare devices cost-effectively, similar to what the TASTE trial did for aspiration thrombectomy in STEMI patients. On the pre-approval side, there is also an important question regarding whether “me too” devices that effectively do the same thing as existing devices must demonstrate more than “substantial equivalence” or “noninferiority” to gain FDA approval.

Krumholz: Are in-hospital outcomes sufficient to conclude that the devices are similar?

Giri: In this case, on the question of procedural safety, I would say yes. Our primary endpoint was procedure-related adverse events (stroke/death) associated with use of the different devices. We used in-hospital outcomes as the primary endpoint to avoid the selection bias associated with our 30-day data, which were available in only about three quarters of patients. Nevertheless, the proportions of adverse events between devices were nearly identical with the available 30-day data, further supporting the conclusions from our primary analysis. Notably, we focused on the procedural safety of carotid stenting but not procedural efficacy (i.e., primary or secondary stroke prevention over several years), which would have to be evaluated with long-term follow-up. We cannot comment on whether one carotid stent type is associated with greater efficacy than another with respect to long-term stroke prevention, given the data contained within the NCDR CARE registry.

Yeh: For EPDs, whose purpose is to prevent procedural complications and which are removed before the procedure is complete, in-hospital outcomes seem to be adequate. Data out to 30 days definitely provide even more reassurance. Efficacy and safety of the stent component of these systems would benefit from longer-term follow-up data. We have discussed the possibility of linking the CARE registry with long-term administrative claims data, as has been done for the NCDR CathPCI registry, but we have yet to do that. Such a linkage could help provide longer-term comparative data on these devices.

Krumholz: What are the implications of your findings for clinical practice?

Giri: Operators do not mix and match carotid stents and EPDs very often. This is a very different device-usage pattern from what we see in other coronary and peripheral interventions, where operators use their judgment to select equipment tailored to given anatomic situations. This practice may be driven, in a large proportion of cases, by the current carotid-stenting reimbursement requirements for industry-sponsored post-marketing surveillance study participation.

In addition, the three most commonly used carotid stenting systems are associated with a similar procedural safety profile, so operators should feel confident in using the devices they feel most comfortable with for a given clinical situation. A tremendous amount of time and energy has been spent theorizing about various technical considerations of carotid stents and EPDs that might make one superior to another. The current study effectively argues that continued focus on these technical issues is unlikely to significantly improve stroke rates around the time of carotid stenting. Nevertheless, we should still search for novel ways to prevent stroke in this population.

Yeh: Operators often choose among different devices that are intended to achieve the same goal. In interventional cardiology, we’ve seen substantial variation in the use of devices, more often governed by physicians’ preferences rather than by patient characteristics. In the case of carotid stenting systems, our registry data seem to provide reassurance that operators can continue to use their preferred devices according to their comfort level and experience, without adversely affecting patient outcomes.

How do the comments from Dr. Giri and Dr. Yeh affect your view of the carotid stenting systems that are commonly used to treat carotid artery stenosis?