Ten-year Durability, Hemodynamic Performance, and Clinical Outcomes after Transcatheter Aortic Valve Implantation Using a Self-expanding Device

Since September 2007, all patients underwent TAVI procedures at the Heart Centre, Segeberger Kliniken, Bad Segeberg, Germany, have been included in a prospective registry (Post-TAVI registry: NCT03192774) that is approved by the local ethics committee (Ärztekammer Schleswig–Holstein) and conforms to the Declaration of Helsinki. All patients included in this study have provided written consent for enrollment and for systematic follow-up.

Out of 302 patients who received the CoreValve transcatheter heart valves (THV) between September 2007 and February 2015, 267 patients had a complete follow-up (88.4%) and have been included in the present analysis. For descriptive purposes, the total cohort was divided into two groups: (a) survivors at 10 years (n = 17 patients) and (b) non-survivors, who died before completing 10 years after valve implantation (n = 250 patients).

Clinical and echocardiographic follow-up was routinely performed post-TAVI before discharge, at 6 months, and at 1, 2, and 5 years. Additionally, for the sake of the present analysis, all patients surviving beyond 5 years after TAVI were approached and personally interviewed (at the institution or through home visits) for clinical and echocardiographic examinations.

Clinical data, quality of life, and medical history were collected using standardized forms and stored at the Centre for Clinical Research of the Heart Centre, Bad Segeberg, through a secure server in accordance with good clinical practice guidelines. Clinical and echocardiographic follow-up was completed either through an outpatient visit or—in the case of impaired patient mobility—a home visit.

The main objectives of the study were to evaluate the 10-year hemodynamic performance (as assessed by echocardiography) and bioprosthetic valve failure (BVF) including structural valve deterioration (SVD) according to the endpoint definitions proposed by the Valve Academic Research Consortium 3 (VARC 3) [12].

According to these definitions, bioprosthetic valve dysfunction (BVD) can be due to SVD, non-SVD, endocarditis, or valve thrombosis, and may be associated with no, moderate, or severe hemodynamic valve deterioration. Severe hemodynamic deterioration is defined as (a) an increase in mean transvalvular gradient ≥ 20 mmHg to ≥ 30 mmHg with a concomitant decrease in effective orifice area (EOA) compared with the early post-TAVI measurements, OR (b) new occurrence, or increase of ≥ 2 grades of intraprosthetic aortic regurgitation (AR) resulting in severe AR.

BVF was defined as any of the following: (i) BVD leading to death; (ii) BVD leading to aortic valve re-intervention (i.e., TAVI-in-TAVI, paravalvular leak closure, or SAVR); or (iii) BVD leading to severe hemodynamic deterioration.

All follow-up echocardiographic image quality was ensured according to a detailed acquisition protocol. Image interpretation was based on a detailed analysis protocol according to current guidelines and standardized VARC-3 endpoint definitions [12].

Evaluation of THV performance included serial assessments of aortic valve (AV) leaflet morphology, mobility, peak velocity, AV peak gradient, AV mean gradient, aortic valve area (AVA), peak velocity in the left ventricular outflow tract (LVOT), and time velocity integrals (TVI). Color and spectral Doppler echocardiography were used to assess severity and origin of AR.

Categorical variables are summarized as frequencies and percentages, while continuous variables are presented as mean ± SD or median [25th–75th quartiles], depending on distribution. Inter-group comparisons were conducted using Student’s t or Mann–Whitney U test for continuous variables, and by chi-square or Fischer’s exact test for categorical variables. The Kruskal–Wallis H test was used to compare serial echocardiographic data. Survival curves were constructed using the Kaplan–Meier method to assess the cumulative rates of all-cause and cardiac mortality. Data analysis was performed using SPSS V.24.0 (IBM Corp., Armonk, NY, USA). Freedom from BVF and SVD were assessed with the Kaplan–Meier method (actuarial analysis) and the cumulative incidence method (actual analysis) adjusted for the competing risk of all-cause mortality. This analysis was performed using STATA 17 software.

At the time of TAVI, the mean patient age was 80.4 ± 7.0 years, and 55.8% were female. The mean Society of Thoracic Surgeons (STS) score was 6.1 ± 5.2%, and 21.9% were classified as high risk, 37.4% as intermediate, and 40.7% as low surgical risk. TAVI was performed mostly via trans-femoral access (97.8%). The Medtronic CoreValve was used in all cases, most often using valve size of 29 mm (59.4%) or 26 mm (34.6%). Further baseline and procedural characteristics are listed in Tables 1 and 2.

At 10 years post-TAVI, 17 patients were still alive while 250 had died. Patients who died before completing the follow-up showed a trend toward higher baseline pro-BNP levels (2376 [1001–6087] vs. 2598 [296–3175]; p = 0.080) and smaller aortic valve area (0.69 ± 0.26 vs. 0.86 ± 0.45; p = 0.016). Tables 1 and 2 summarize the comparison of baseline and procedural characteristics of survivors and non-survivors up to 10 years.

Table 3 summarizes the 30-day post-procedural outcome, and we observed more acute kidney injury post-TAVI in the group who died before completing the 10 years (0 [0.0] vs. 52 [20.8%]; p = 0.036), and more post-TAVI AR > II (14 [5.8%] vs. 0 [0.0%]; p = 0.043). The temporal trend in the incidence of 30-day complications post-TAVI through four time intervals (2007/2008, 2009/2010, 2011/2012, and 2013–2015) revealed that all-cause mortality, cardiac mortality, and permanent pacemaker rates did not differ significantly among the different time intervals (p = 0.783, p = 0.418, and p = 0.630, respectively), while life-threatening bleeding and major vascular complications declined remarkably at the last interval from 2013 to 2015 (p = 0.020 and p = 0.008, respectively) (Supplementary Figure S1, see supplementary material).

Clinical follow-up was available in 99.3%, 98.6%, 96.7%, and 88.4% of our patients at 3, 5, 7, and 10 years, respectively. The cumulative rates of all-cause mortality at 3, 5, 7, and 10 years were 23.7%, 40%, 65.8%, and 89.8%, while corresponding rates of cardiac mortality were 13.1%, 24.5%, 35.5%, and 49.5%, respectively (Supplementary Figure S2, see supplementary material). Supplementary Table S1 summarizes univariable and multivariable predictors of all-cause mortality.

Transthoracic echocardiography before discharge and at 5-, 7-, and 10-year follow-up revealed AVA of 1.94, 1.87, 1.69, and 1.98 cm², respectively (p = 0.236) (Fig. 1a). Mean and peak transvalvular gradient before discharge and at 5-, 7-, and 10-year follow-up were nearly the same (8.3, 9.0, 8.2, and 10.1 mmHg; p = 0.796; and 15.05, 15.5, 14.6, and 17.6 mmHg; p = 0.493, respectively) (Fig. 1b). The corresponding rates of mild and ≥  ARIII were 47.9%, 45.5%, 46.3%, and 50% and 0.0%, 1.5%, 4.9%, and 10% (p = 0.090), respectively (Supplementary Figure S3, see supplementary material).

From the total cohort, 30 patients developed BVD, including 11 patients with SVD, 14 with non-SVD, three with infective endocarditis, and two with valve thrombosis (Supplementary Figure S4, see supplementary material).

Among those with documented SVD, two patients developed moderate aortic valve stenosis, three had mild and moderate transvalvular aortic insufficiency, and six developed severe SVD. Among those with severe SVD, three patients developed severe bioprosthetic valve stenosis at 72, 104, and 120 months, and the first patient was treated with TAVI-in-TAVI, while the latter two died suddenly. Two patients developed severe transvalvular AR: one at 64 months, treated with TAVI-in-TAVI; and one at 120 months, treated conservatively. One patient had, at 61 months, combined severe stenosis and insufficiency and was managed with TAVI-in-TAVI. The actual rate of freedom from SVD (using the cumulative incidence function) was 99.8%, 98.9%, 98.8%, 98.8%, 98.5%, and 97.9% at 5, 6, 7, 8, 9, and 10 years, respectively. The corresponding actuarial rates (using the Kaplan–Meier method) were 99.2%, 97.6%, 97.6%, 97.6%, 94.6%, and 80.9%%, respectively (Fig. 2a).

Non-structural valve dysfunction was documented in 14 patients: four with mild paravalvular leakage (PVL), seven with moderate PVL, and three patients with severe PVL documented at 9, 54, and 75 months); all severe PVL were treated with TAVI-in-TAVI. BVD due to infective endocarditis was diagnosed in three patients, while valve thrombosis without hemodynamic significance was diagnosed in two patients.

Eleven patients had evidence of BVF. Six patients developed hemodynamically severe SVD, three had significant non-SVD, and two had infective endocarditis. Actual rates of freedom from BVF were 98.6%, 97.9%, 97.2%, 97.1%, 96.8%; and 96.1% at 5, 6, 7, 8, 9; and 10 years, respectively. The corresponding actuarial rates were 97.6%, 95.9%, 95% 94.9%, 91.1%; and 78.8%, respectively (Fig. 2b).

Experience with early-generation surgical bioprosthetic valves showed that SVD commonly begins 8 years after implantation, with a greatly increased rate of SVD after 10 years [15, 16]. Studies on the performance of surgical valves during the first decade following valve implantation have reported rates of freedom from SVD at 10 years > 85% [17‐19]. Currently, long-term outcomes with second-generation porcine Hancock II valve (Medtronic) documented survival rates without SVD at 10, 15, and 20 years of 95%, 75%, and 49%, respectively [20]. The present study evaluated the early-generation transcatheter self-expanding CoreValve system and showed an actual survival rate without BVF and SVD at 10 years of 96.1% and 97.9% (actual rate), respectively.

As TAVI has only been widely available since 2007, data concerning long-term durability are limited [9]. A large FRANCE-2 registry, which included 4201 patients after THV implantation, reported a cumulative rate of severe and moderate SVD at 5-year follow-up of 2.5% and 13.3%, respectively [21]. Another study which included only patients who underwent TAVI with the CoreValve system reported a 5-year rate of SVD of 1.4% [6]. Our study reported a lower actuarial rate of SVD at 5 and 6 years compared with the previously mentioned studies, with rates of 0.8% and 2.4%, respectively.

The incidence of SVD 5 to 10 years post-TAVI was described in the UK TAVI registry, where severe SVD occurred in 0.4% at a mean of 5.3 years after implantation, while moderate SVD was reported in 8.7% of patients [22]. Sathananthan et al. performed a 10-year follow-up in high-risk patients who were treated during the early experience of TAVI and found that the rate of SVD at 4, 6, 8, and 10 years was 0.4%, 1.7%, 4.7%, and 6.5%, respectively [23]. Compared with the previously mentioned studies (which used the cumulative incidence to assess the SVD), the present study observed a lower actual rate of SVD and BVF at 10 years of 2.1% and 3.9%, respectively. This difference could be attributed to the implantation of different THV, as a balloon-expandable valve was implanted in 98.3% of patients in the previous study.

The NOTION (Nordic Aortic Valve Intervention) trial is the first to compare only low- risk patients with severe AS who were treated with TAVI using first- and second-generation CoreValve systems versus surgical aortic valve replacement [24]. NOTION reported a low 10-year cumulative incidence of severe SVD of 1.5%, and a 10-year cumulative incidence of BVF of 9.7% [24]. Our study reported an actual rate of severe SVD at 10 years of 2.1% and BVF of 3.2%. We observed a higher rate of severe SVD than that reported in the NOTION trial but a lower BVF rate. We included nearly double the number of patients, which may explain the higher rate of severe SVD in our analysis. The higher rate of BVF reported in the NOTION trial was mainly driven by the higher valve-related death (5%) than in ours. Despite the high all-cause mortality in our study, we reported only three valve-related deaths. The NOTION trial enrolled low-risk patients, in contrast to our study, which included 59.3% of patients with intermediate and high surgical risk.

These findings together raise the question of whether the continuous improvement of newer-generation valves leads to a lower rate of degeneration and failure. In the PARTNER 2A trial, the rate of SVD with the older-generation SAPIEN XT was high (9.5%), but with the newer-generation SAPIEN 3, the 4-year SVD was only 2.5% [9]. It is also recognized that differences in durability exist between different bioprosthetic surgical aortic valve designs and generations [16‐19]. As newer generations of THV have anti-calcification treatment of leaflets, improvement in leaflets and frames, and the addition of a skirt at the lower part of most valves [9], we assume that the rate of SVD will continue to decline. However, more data are still needed to confirm these results, particularly in low-risk patients.

Our study reported a nonsignificant difference in transvalvular gradient and effective orifice area during the 10 years of follow-up. Although these findings were similar to what was documented by the NOTION trial and the study by Sathananthan et al., we should interpret these results with caution due to the high mortality rate [23, 24]. The current analysis reported a high rate of mortality at 10 years (89.8%), which is not surprising because the cohort of our study consisted of a population in their 80s with a relatively limited further life expectancy and multiple comorbidities. We compared the baseline and periprocedural data between patients who were still alive at 10 years and who had died, and we found a trend toward higher pro-BNP, more severe aortic valve stenosis, and a higher rate of significant post-TAVI AR among non-survivors. The higher baseline peak pressure gradient, reduced LV EF, high baseline pro-BNP, and SVD were predictors for all-cause mortality at univariable analysis, while only post-TAVI acute kidney injury was the independent predictor of all-cause mortality. It is notable that our data do not reflect the contemporary outcomes, as they came from early experience. Comparing the periprocedural complications among different time intervals revealed improved rates of life-threatening bleeding and major vascular complications. Despite the improvement in periprocedural complications, the mortality rates did not decline. These findings together indicate that the mortality in the study population was multifactorial and related to the multiple comorbidities.

Although the follow-up in the current study was 88.4%, the analysis is limited by the high mortality rates at 10 years in an elderly high-risk population. We tried to overcome this problem by computing the cumulative incidence to compete for high all-cause mortality. The main cohort of our study consisted of a population in their 80s with a relatively limited further life expectancy, so the results would be less useful in a younger population with lower procedural risk and fewer comorbidities. Additionally, the echocardiographic follow-up was relatively limited, which might underestimate the incidence of SVD (echocardiography was available in 69.4%, 77.8%, and 82.4% of patients at 5, 7, and 10 years).