Clinical Trial Design for Second Generation TAVI - Academic View William S. Weintraub, MD John H. Ammon Chair in Cardiology Christiana Care Health System Professor of Medicine Thomas Jefferson University Disclosures: Grants or Consulting from Sanofi-Aventis, Merck, Pfizer, Otsuka, Gilead, Abbott, GSK, Astra-Zeneca, Lilly, BMS, Bayer, Shionogi

TAVI Works! TAVI reduced mortality and repeat hospitalizations compared to medical therapy in PARTNER B cohort However, stroke, vascular complications and bleeding were increased with TAVI These results cannot be generalized to other patient groups RCT comparison to medical therapy no longer possible

Usual Questions For the next trial: Primary and secondary endpoints Appropriate patient populations Comparator selection Power – sample size Statistical design Sites and recruitment

Primary and Secondary Endpoints All cause mortality seems to be the endpoint for the moment Comparisons between devices may require different endpoints Consider hospitalizations for heart failure, syncope, symptoms Durability Paravalvular leak Health status – may need new instruments Cost-effectiveness should be part of all trials

What are the Potentially Appropriate Populations? Non-surgical candidates Borderline surgical candidates Current surgical candidates

The Comparator For patients not appropriate for surgery, historical controls or first generation device Expanded indications, perhaps standard surgery

Power and Sample Size For similar devices to the current, similar populations, studies compared to historical controls will be relatively small, similar to PARTNERS Expanded indications will require larger sample sizes

Statistical Design Comparison to historical controls will be simple superiority Comparison of new devices to current or surgery will most probably be non-inferiority, will require large sample sizes and will be a considerable challenge.

Sites and Recruitment The number of centers performing TAVI is still small This will increase considerably over the next several years As the trials become more complex, more sites will be needed to achieve adequate sample size, especially for non-inferiority compared to active control

Call for a National Registry Only a national registry will provide sufficient numbers to fully describe efficacy, safety and cost A national registry will permit detailed assessment of outcomes in subgroups This may be an approach for comparative effectiveness studies between different devices, approaches A registry will permit benchmarking of outcomes Properly constructed, a registry in an institution permits data collection to be incorporated in workflow

Conclusions It is unusual for a comparison to medical therapy to offer such a straightforward outcome Comparing new devices may be or become more difficult Defining the population will be a challenge A national registry is a high priority