Reply [to: Factors Affecting Mortality in Transcatheter Aortic Valve Implantation / Iscan et al.]

To the Editor:
In their letter Iscan and colleagues conclude that because of the limited sample size and inequalities in the baseline characteristics, the results of our study are unreliable. Transcatheter aortic valve implantation (TAVI) has become an established treatment option for patients at prohibitive or high risk to undergo surgical aortic valve replacement (SAVR). Transfemoral aortic valve implantation (TF-AVI) has become the access option of first choice in many centers because of its minimally invasive nature compared with transapical aortic valve replacement (TA-AVI). However, at this moment no randomized data comparing access techniques are available. For this reason, we sought to compare short-term and long-term outcomes after TF-AVI versus TA-AVI using the comprehensive data from the PRAGMATIC Registry.

We agree that the limited sample size in the TA-AVI group is a limitation of this analysis. Yet, with respect to the comments of Iscan and colleagues [1], we would like to point out that TF-AVI is the default strategy in many centers world wide and that, therefore, the distribution of patients between both groups in our analysis reflects the current state of TAVI, which is confirmed in several large registries in both Europe and the United States (ie, FRANCE-2 and TVT Registry). Second, differences in baseline characteristics are inevitable because of the nonrandom allocation of patients in an observational study. To address this bias, we conducted a multivariable analysis adjusting all outcomes for these differences. Both all-cause mortality and cardiovascular in-hospital mortality were significantly higher in the TA-AVI group. Iscan and colleagues rightfully indicate an uneven distribution between the sizes and types of prosthetic valves resulting from the fact that only the Edwards SAPIEN Valve is available for TA-AVI. Nonetheless, multiple studies have shown that there is no effect of valve type on long-term outcomes. We therefore believe that the conclusions of our analysis hold while we consider the limitations of a post-hoc analysis in a prospective cohort study that were highlighted in the discussion and that they may stimulate the discussion on treatment strategy in clinical practice and further research addressing the respective role of various TAVR modalities.