Author + information
- Received March 4, 2011
- Revision received October 11, 2011
- Accepted October 17, 2011
- Published online January 17, 2012.
- Ruediger Lange, MD, PhD,
- Sabine Bleiziffer, MD,
- Domenico Mazzitelli, MD,
- Yacine Elhmidi, MD,
- Anke Opitz, MD,
- Marcus Krane, MD,
- Marcus-Andre Deutsch, MD,
- Hendrik Ruge, MD,
- Gernot Brockmann, MD,
- Bernhard Voss, MD,
- Christian Schreiber, MD,
- Peter Tassani, MD, PhD and
- Nicolo Piazza, MD, PhD⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Nicolo Piazza, German Heart Center, Lazarette Strasse 36, 80636 Munich, Germany
Objectives The purpose of this study was to investigate the evolution of patient selection criteria for transcatheter aortic valve implantation (TAVI) and its impact on clinical outcomes.
Background Anecdotal evidence suggests that patient selection for TAVI is shifting toward lower surgical risk patients. The extent of this shift and its impact on clinical outcomes, however, are currently unknown.
Methods We conducted a single-center study that subcategorized TAVI patients into quartiles (Q1 to Q4) defined by enrollment date. These subgroups were subsequently examined for differences in baseline characteristics and 30-day and 6-month mortality rate. The relationship between quartiles and mortality rate was examined using unadjusted and adjusted (for baseline characteristics) Cox proportional hazard models.
Results Each quartile included 105 patients (n = 420). Compared with Q4 patients, Q1 patients had higher logistic EuroSCORES (25.4 ± 16.1% vs. 17.8 ± 12.0%, p < 0.001), higher Society of Thoracic Surgeons scores (7.1 ± 5.5% vs. 4.8 ± 2.6%, p > 0.001), and higher median N-terminal pro–B-type natriuretic peptide levels (3,495 vs. 1,730 ng/dl, p < 0.046). From Q1 to Q4, the crude 30-day and 6-month mortality rate decreased significantly from 11.4% to 3.8% (unadjusted hazard ratio [HR]: 0.33; 95% confidence interval [CI]: 0.11 to 1.01; p = 0.053) and from 23.5% to 12.4% (unadjusted HR: 0.49; 95 CI: 0.25 to 0.95; p = 0.07), respectively. After adjustment for baseline characteristics, there were no significant differences between Q1 and Q4 in 30-day mortality rate (adjusted HR ratio: 0.29; 95% CI: 0.08 to 1.08; p = 0.07) and 6-month mortality rate (HR: 0.67; 95% CI: 0.25 to 1.77; p = 0.42).
Conclusions The results of this study demonstrate an important paradigm shift toward the selection of lower surgical risk patients for TAVI. Significantly better clinical outcomes can be expected in lower than in higher surgical risk patients undergoing TAVI.
The Medtronic CoreValve (Medtronic, Minneapolis, Minnesota) and Edwards SAPIEN (Edwards LifeSciences, Irvine, California) prostheses obtained Conformité Européenne mark approval in March 2007 and November 2007, respectively, to treat high or prohibitive surgical risk patients with severe aortic stenosis. Similar to coronary stent use for off-label indications (1–6), there is some evidence to suggest that patient selection criteria for transcatheter aortic valve implantation (TAVI) are evolving away from the pre-market inclusion and exclusion criteria (7). This could mean that transcatheter aortic valves are implanted in higher or lower risk patients than those originally included in the initial safety and feasibility trials. Anecdotally, the paradigm shift appears to be in the direction of treating less sick patients with TAVI. The extent of this shift and its impact on clinical outcomes, however, are still uncertain. The Medtronic CoreValve SURTAVI (Surgical Replacement and Transcatheter Aortic Valve Implantation) trial and Edwards PARTNER 2 (Placement of AoRTic TraNscathetER Valve Trial) SAPIEN XT trial, although in their early stages of development, will be concerned with randomizing intermediate to high surgical risk patients to TAVI or surgical aortic valve replacement. These trials provide us with yet another indication that TAVI is being directed at the treatment of lower and lower surgical risk patients.
To investigate the evolution of patient selection criteria for TAVI , we conducted a single-center study that subcategorized a total of 420 TAVI patients into quartiles and subsequently examined the temporal changes in baseline characteristics and the impact of TAVI on survival.
Between June 2007 and June 2010, 420 patients underwent TAVI at the German Heart Center Munich. High surgical risk patients with severe aortic valve stenosis were referred for TAVI after a dedicated team of cardiac surgeons, cardiologists, and anesthesiologists reached a consensus that TAVI was in the best interest of these patients. Informed consent was subsequently obtained from all patients.
Suitability for TAVI was confirmed by a combination of imaging modalities: transesophageal echocardiography, multislice computed tomography, and angiography. Our approach to prosthesis type (Medtronic CoreValve or Edwards SAPIEN) and size selection and vascular access route was described in detail elsewhere (8). Briefly, patients were considered for TAVI if the aortic valve annulus diameter was 18 to 27 mm. Our strategy for vascular access is based on minimal invasiveness. Thus, we first consider a transfemoral CoreValve approach, followed by subclavian CoreValve, and finally transapical Edwards SAPIEN.
Device description and procedure
Details of the Medtronic CoreValve and Edwards SAPIEN device, and technical aspects of the procedure, were previously published (8). Procedures were performed in a hybrid surgical suite. Depending on underlying comorbidities, patients received either general or local (with mild sedation) anesthesia. With increasing experience, there was a conscious decision to use local anesthesia, especially during transfemoral valve implantations. Transfemoral vascular access was obtained either percutaneously (pre-closed with a 10-F Prostar XL and a 6-F Perclose ProGlide, Abbott Vascular, Abbott Park, Illinois) or by surgical cutdown.
Post-implantation dilation or a second transcatheter valve may have been required in cases of significant paravalvular aortic regurgitation.
Definition of outcomes
The primary endpoint is all-cause mortality rate at 30-day and 6-month follow-up. To ensure adequate 6-month follow-up, the enrollment period ended on June 31, 2010. Secondary endpoints included stroke/transient ischemic attack, vascular complications requiring any percutaneous or surgical correction, and need for a new permanent pacemaker within 14 days of the procedure. Post-implantation aortic regurgitation estimated by echocardiography was categorized as none/trivial (0 to 1), mild (2), or moderate/severe (3 to 4).
The 420 patients were subdivided into equal quartiles (each quartile with 105 patients). The 4 subgroups were subsequently analyzed for differences in baseline characteristics and clinical outcomes.
Continuous variables were tested for normality using the Shapiro-Wilks goodness-of-fit test; normally distributed variables are presented as mean ± SD and non-normally distributed variables are presented as median (interquartile range). Dichotomous or nominal variables are described as numbers and percentages.
One-way analysis-of-variance was used to compare means across quartiles; post hoc pairwise comparison was done with Bonferroni correction. In cases of nonparametric distribution or ordinal data, the Kruskal-Wallis analysis of ranks was used; post hoc comparison was done using the Mann-Whitney test (Wilcoxon rank sum test) with Bonferroni correction. Categorical variables were compared by the Pearson chi-square test or, when indicated, the Fisher exact test.
A Cox proportional hazard regression model provided event rate graphs for 30-day and 6-month mortality rate for quartiles 1 to 4. Two models were used: an unadjusted model and an adjusted model that included baseline characteristics associated with mortality rate in a univariable analysis. Because of its skewed distribution, the natural logarithm of pro–B-type natriuretic peptide was used in the Cox regression models. Although a p value <0.10 was used for the inclusion of covariates into the univariable model, a p value <0.05 was used as the criterion for statistical significance in the multivariable model.
All statistical analyses were performed using SPSS version 17.0 (SPSS Inc., Chicago, Illinois). Unless otherwise specified, a 2-tailed p value of ≤0.05 was considered statistically significant.
Baseline and procedural characteristics
Patients undergoing TAVI in the first quartile had significantly higher logistic EuroSCORES than those in the second, third, or fourth quartiles (Q1: 25.4 ± 16% vs. Q2: 18.8 ± 10% vs. Q3: 18.3 ± 11% vs. Q4: 17.8 ± 12%, analysis of variance p < 0.001). The Society of Thoracic Surgeons (STS) score was significantly lower in Q4 than in Q1, Q2, or Q3 (Q1: 7.1 ± 5.4% vs. Q2: 6.2 ± 3.5% vs. Q3: 5.8 ± 3.9% vs. Q4: 4.8 ± 2.6%, analysis of variance p < 0.001).
The ratio of use of the CoreValve to Edwards SAPIEN decreased from 87 (13%) in Q1 to 59 (41%) in Q4 (p < 0.001). Correspondingly, the number of transfemoral cases decreased by 36% from Q1 to Q4 (85% vs. 49%, p < 0.001). There was a 30% reduction in the use of general anesthesia over time (Q1: 100% vs. Q4: 69%, p < 0.001) and less contrast was used in Q4 than in Q1 (156 ± 69 ml vs. 104 ± 41 ml, p < 0.001).
Association of baseline characteristics with 30-day and 6-month mortality rate
Univariable analysis revealed that age, logistic STS score, atrial fibrillation, previous cardiac surgery, and previous stroke were associated with 30-day mortality rate (p < 0.10) (Table 3).
Univariable analysis revealed that age, logistic EuroSCORE, STS score, left ventricular ejection fraction, atrial fibrillation, previous cardiac surgery, chronic obstructive pulmonary disease, and N-terminal pro–B-type natriuretic peptide (p < 0.10) were associated with 6-month mortality rate (Table 4).
From Q1 to Q4, the crude 30-day mortality rate significantly decreased from 11.4% to 3.8% (chi-square test, p = 0.03) (unadjusted HR: 0.33; 95% confidence interval [CI]: 0.11 to 1.01; p = 0.051) and the 6-month mortality rate significantly decreased from 23.5% to 12.4% (chi-square test, p = 0.031) (unadjusted HR: 0.49; 95% CI: 0.25 to 0.95; p = 0.072) (Fig. 1A). There were no significant differences in mortality rate observed between Q1 and Q4 after adjustment for baseline characteristics at 30-day and 6-month follow-up (30-day mortality rate adjusted HR: 0.29; 95% CI: 0.08 to 1.08; p = 0.07) (6-month adjusted mortality rate HR: 0.67; 95% CI: 0.25 to 1.77; p = 0.42) (Fig. 1B).
There were no significant differences in 30-day and 6-month mortality rates in patients undergoing transfemoral TAVI and transapical TAVI (30-day HR: 1.07; 95% CI: 0.54 to 2.11; p = 0.83) (6-month HR: 1.13; 95% CI: 0.71 to 1.80; p = 0.60). Similar findings were noted within quartiles.
Secondary clinical endpoints
Clinical outcomes at 30 days and 6 months are shown in Table 7. Transfemoral vascular complications decreased by 17% from Q1 (26.9%) to Q4 (9.5%) (p = 0.008). Although the absolute percentages trended downward, there were no significant differences observed in the rate of stroke/transient ischemic attack or need for a permanent pacemaker from Q1 to Q4.
In the present study, we found that the crude 30-day and 6-month mortality rates improved from Q1 to Q4. Although no significant differences in mortality were observed between Q1 and Q4 after adjustment for baseline characteristics, an important signal between quartiles remained. With growing experience, we observed a shift toward the treatment of younger patients with fewer comorbidities and lower surgical risk scores (logistic EuroSCORE and STS score). The results of this study suggest that improved survival outcomes can be expected in lower surgical risk patients undergoing TAVI.
Possible explanations for the progressive decrease in mortality observed in the present study include patient-, operator/procedure-, and device-related factors. Although we would expect improved outcomes in treating less complex patients with TAVI, this has not been previously demonstrated. From Q1 to Q4, we observed a significant mortality reduction at both 30-day (11.4% to 3.8%) and 6-month (23.5% to 12.4%) follow-up. After adjusting for baseline characteristics (patient-related factors), however, there were no statistically significant differences in 30-day or 6-month mortality among quartiles. These results suggest that underlying comorbidities play an important role in acute and intermediate-term survival after TAVI. These findings do not exclude the potential contribution of a learning curve, changes in procedural techniques, or device improvements. Notwithstanding their importance, these latter factors were not included in the present analysis because the information was not collected or it was difficult to quantify/record such phenomena.
Previous publications have alluded to an improvement in outcomes with growing experience (9–11). Without controlling for confounding, these reports suggested that a learning curve (operator experience) and device iterations accounted for the reduction in adverse events over time. Nonetheless, the authors did not provide adjusted mortality figures to investigate the possible contribution of changing baseline characteristics over time.
The baseline characteristics observed in Q1 (e.g., mean age 81 years; STS score 7%) are suggestive of those found in contemporary TAVI studies (9–17). In contrast, and to our knowledge, patient characteristics similar to those in Q4 (e.g., mean age 78 years; STS score, 4%) have not been previously reported in a TAVI-related study. This change in patient selection pattern is likely fueled by increasing operator/team experience and improving clinical outcomes. At our center, patient selection is performed by a team of cardiac surgeons, cardiologists, and anesthesiologists. Furthermore, clinical judgment, as opposed to any surgical risk algorithm, plays a fundamental role in the selection of patients for TAVI.
Recent TAVI publications report a 30-day mortality rate ranging from 6.7% to 13% (13–18). Our overall 30-day mortality rate of 9.5% is well within this range. However, in our last 105 patients with a mean age of 78 years, an STS score of 4.6%, a and logistic EuroSCORE of 17.8%, the 30-day mortality was 3.8%. Keeping this in mind, Grossi et al. (19) reported a 30-day mortality of 7.8% after surgical aortic valve replacement in a cohort of 731 high-risk patients with a mean logistic EuroSCORE of 17.2%. In that study, 43% were seputagenarians and 44% were octogenarians or nonagenarians.
The Bern-Rotterdam collaboration recently published a propensity score-matched analysis comparing TAVI and surgical aortic valve replacement (SAVR) (20). Compared with the unmatched TAVI cohort, the matched TAVI cohort were younger (mean age 79 vs. 84 years, p < 0.001), had lower logistic EuroSCORES (16% vs. 22%, p < 0.001), had less severe symptoms, and were less likely to report a history of coronary artery bypass surgery, atrial fibrillation, stroke, peripheral vascular disease, or pulmonary hypertension. Based on pre-specified methods to control for confounding, the authors concluded that 30-day survival rates may be similar among selected TAVI and SAVR patients. The authors considered that the matched TAVI cohort could be eligible for a randomized comparison of TAVI and SAVR. It is likely that future randomized trials comparing TAVI and SAVR (e.g., SURTAVI and PARTNER 2) will enroll patients with baseline characteristics similar to those of the matched TAVI cohort from the Bern-Rotterdam study and the Q4 cohort in the present study.
We observed an important reduction in the rate of femoral complications over time. Likely explanations for this include a learning curve associated with the use of pre-closure devices and establishing a lower threshold to use subclavian or transapical access in the setting of aortoiliac disease (calcification, tortuosity, or borderline vessel size). Table 2 demonstrates that the number of nontransfemoral TAVI cases increased from Q1 to Q4. Thus, vascular access site selection is based ultimately on patient safety; a femoral access route may be feasible in many patients, but a safer access route may exist. Furthermore, in patients with some degree of femoral vessel calcification undergoing transfemoral TAVI, we prefer to perform a surgical cutdown. This avoids potential failure of the pre-closure devices and allows the operator to puncture an area of the vessel devoid of calcification. During our initial experience, pre-closure was performed using a single 10-F Prostar device. Later in our experience, we implemented a double pre-closure technique using a 10-F Prostar device and a 6-F Proglide. The 6-F Proglide provides additional closure in cases in which the 10-F Prostar fails. Other centers have reported using 2 or 3 6-F Proglide devices for pre-closure (21). The important point is that centers are accustomed to an effective, safe, and reproducible strategy.
As with any multivariable analysis, the present study can be limited by unmeasured confounding. Because we only included baseline characteristics in our model, it is possible that procedure-, operator-, and device-related factors also have an influence on 30-day and 6-month mortality. Our goal was to investigate changing patient selection patterns and its influence on mortality.
The results of this study demonstrate an important paradigm shift toward the selection of lower surgical risk patients for TAVI. Significantly better clinical outcomes can be expected in lower than in higher surgical risk patients undergoing TAVI. As TAVI becomes more routine and widely available, operators may be tempted to implant the device in younger patients with less comorbidities. Uncertainties about the long-term durability, in addition to the unresolved issues of paravalvular aortic regurgitation and conduction abnormalities, should be cautiously weighed against the immediate benefits being widely reported.
Dr. Lange is a consultant and advisor for Medtronic. Drs. Bleiziffer and Mazzitelli are consultants and proctors for Medtronic and Edwards LifeSciences. Dr. Voss is a instructor for Medtronic and receives financial compensation. Dr. Piazza is a consultant and proctor for Medtronic and CardiAQ. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- hazard ratio
- surgical aortic valve replacement
- Society of Thoracic Surgeons
- transcatheter aortic valve implantation
- Received March 4, 2011.
- Revision received October 11, 2011.
- Accepted October 17, 2011.
- American College of Cardiology Foundation
- Lemos P.A.,
- Lee C.H.,
- Degertekin M.,
- et al.
- Ong A.T.,
- Serruys P.W.,
- Aoki J.,
- et al.
- Hoye A.,
- Tanabe K.,
- Lemos P.A.,
- et al.
- Rao S.V.,
- Shaw R.E.,
- Brindis R.G.,
- Klein L.W.,
- Weintraub W.S.,
- Peterson E.D.
- Grines C.L.
- Piazza N.,
- Otten A.,
- Schultz C.,
- et al.
- Bleiziffer S.,
- Ruge H.,
- Mazzitelli D.,
- et al.
- Webb J.G.,
- Altwegg L.,
- Boone R.H.,
- et al.
- Himbert D.,
- Descoutures F.,
- Al-Attar N.,
- et al.
- Grube E.,
- Buellesfeld L.,
- Mueller R.,
- et al.
- Piazza N.,
- Grube E.,
- Gerckens U.,
- et al.
- Moynagh A.M.,
- Scott D.J.,
- Baumbach A.,
- et al.
- Tamburino C.,
- Capodanno D.,
- Ramondo A.,
- et al.
- Lefevre T.,
- Kappetein A.P.,
- Wolner E.,
- et al.
- Eltchaninoff H.,
- Prat A.,
- Gilard M.,
- et al.,
- FRANCE Registry Investigators
- Bleiziffer S.,
- Ruge H.,
- Mazzitelli D.,
- et al.
- Piazza N.,
- van Gameren M.,
- Jüni P.,
- et al.