Key Points
Question
What is the durability of transcatheter heart valves and is this similar to surgical bioprostheses?
Findings
In this analysis of data from the PARTNER Trial, population hemodynamic trends showed early favorable hemodynamic changes after transcatheter aortic valve replacement implantation, with midterm stability at a median follow-up of 3.1 (maximum, 5) years, and no significant changes in surgical aortic valve replacement. Reintervention and severely abnormal hemodynamics on echocardiograms were infrequent and not associated with excess death in patients after either transcatheter aortic valve replacement or surgical aortic valve replacement.
Meaning
There is longitudinal durability of both transcatheter and surgical aortic valve replacements, with rare findings of adverse hemodynamics or valve deterioration.
This analysis of data from the PARTNER Trial determines the midterm hemodynamic performance of balloon-expandable transcatheter heart valves.
Abstract
Importance
Use of transcatheter aortic valve replacement (TAVR) for severe aortic stenosis is growing rapidly. However, to our knowledge, the durability of these prostheses is incompletely defined.
Objective
To determine the midterm hemodynamic performance of balloon-expandable transcatheter heart valves.
Design, Setting, and Participants
In this study, we analyzed core laboratory–generated data from echocardiograms of all patients enrolled in the Placement of Aortic Transcatheter Valves (PARTNER) 1 Trial with successful TAVR or surgical AVR (SAVR) obtained preimplantation and at 7 days, 1 and 6 months, and 1, 2, 3, 4, and 5 years postimplantation. Patients from continued access observational studies were included for comparison.
Interventions
Successful implantation after randomization to TAVR vs SAVR (PARTNER 1A; TAVR, n = 321; SAVR, n = 313), TAVR vs medical treatment (PARTNER 1B; TAVR, n = 165), and continued access (TAVR, n = 1996). Five-year echocardiogram data were available for 424 patients after TAVR and 49 after SAVR.
Main Outcomes and Measures
Death or reintervention for aortic valve structural indications, measured using aortic valve mean gradient, effective orifice area, Doppler velocity index, and evidence of hemodynamic deterioration by reintervention, adverse hemodynamics, or transvalvular regurgitation.
Results
Of 2795 included patients, the mean (SD) age was 84.5 (7.1) years, and 1313 (47.0%) were female. Population hemodynamic trends derived from nonlinear mixed-effects models showed small early favorable changes in the first few months post-TAVR, with a decrease of −2.9 mm Hg in aortic valve mean gradient, an increase of 0.028 in Doppler velocity index, and an increase of 0.09 cm2 in effective orifice area. There was relative stability at a median follow-up of 3.1 (maximum, 5) years. Moderate/severe transvalvular regurgitation was noted in 89 patients (3.7%) after TAVR and increased over time. Patients with SAVR showed no significant changes. In TAVR, death/reintervention was associated with lower ejection fraction, stroke volume index, and aortic valve mean gradient up to 3 years, with no association with Doppler velocity index or valve area. Reintervention occurred in 20 patients (0.8%) after TAVR and in 1 (0.3%) after SAVR and became less frequent over time. Reintervention was caused by structural deterioration of transcatheter heart valves in only 5 patients. Severely abnormal hemodynamics on echocardiograms were also infrequent and not associated with excess death or reintervention for either TAVR or SAVR.
Conclusions and Relevance
This large, core laboratory–based study of transcatheter heart valves revealed excellent durability of the transcatheter heart valves and SAVR. Abnormal findings in individual patients, suggestive of valve thrombosis or structural deterioration, were rare in this protocol-driven database and require further investigation.
Trial Registration
clinicaltrials.gov Identifier: NCT00530894
Introduction
Transcatheter aortic valve replacement (TAVR) is a safe and effective treatment for severe aortic stenosis. However, given the relatively recent introduction of transcatheter heart valves (THV), to our knowledge, data on their durability are sparse. Late deterioration of surgically implanted bioprosthetic valves is well described, with the reduced need for anticoagulation compared with mechanical valves seen as an acceptable tradeoff for potential reintervention. While this risk vs benefit equation may be especially appealing in the elderly population currently receiving TAVRs, reports of early leaflet thickening and possible valve thrombosis demand further investigation.
As the first and largest of the randomized TAVR trials, the Placement of Aortic Transcatheter Valves (PARTNER) 1 Trial provides a rich, independently adjudicated data resource to address these issues. This report will analyze these data to determine the longitudinal hemodynamic profile of the SAPIEN THV (Edwards) and assess the incidence and consequences of any adverse changes as detected by echocardiography. Similar analyses are also performed in patients with surgical bioprosthetic aortic valve replacement (SAVR) in the PARTNER 1 Trial.
Methods
The PARTNER 1 Trial consisted of 3 cohorts: PARTNER 1A, or high risk, which randomized patients to either TAVR or SAVR (TAVR, n = 348; SAVR, n = 351); 1B, or inoperable, which randomized to either medical treatment or TAVR (n = 179); and continued access (TAVR, n = 2040). The trial protocol (Supplement 1) was approved by all participating institutional review boards, and all patients provided written informed consent (eAppendix in Supplement 2). Patients received the first-generation SAPIEN THV device. Trial details have been previously published.
This report includes all successful TAVRs (device implantation as randomized/assigned) in all 3 groups (n = 2482) and all successful SAVRs (n = 313) (eFigure 1 in Supplement 2). Protocol-specified echocardiograms were obtained at baseline and postimplantation intervals of approximately 7 days, 1 and 6 months, and 1, 2, 3, 4, and 5 years. Patients without any postimplantation echocardiographic data were excluded from hemodynamic analysis (TAVR, n = 78; SAVR, n = 21), and others were censored at the time of aortic valve reintervention, death, or end of follow-up (Figure 1). All echocardiogram data were analyzed by a single core laboratory following best practices for randomized trials, with demonstrated excellent reproducibility in this data set.
Figure 1. STROBE Flowchart Showing Patient Flow.
SAVR indicates surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement.
aSD: 45 days.
Echocardiograms were analyzed according to standard practices as previously described and included peak and mean aortic valve gradients, effective orifice area (EOA), and Doppler velocity index (DVI) to evaluate valve systolic hemodynamics and transvalvular aortic regurgitation (AR). The relevant clinical end points are all-cause death and aortic valve reintervention due to structural deterioration.
Statistical Analysis
Average temporal trends in postprocedure and follow-up transthoracic echocardiograms were analyzed longitudinally, accounting for both the temporal pattern and the variability both among patients and within each patient’s sequence of measurements. A temporal decomposition nonlinear mixed-effects model was used to resolve a number of time phases on mean response and to estimate shaping parameters for each phase. We assumed that there was no informative censoring (ie, missing data were missing at random). Longitudinal nonlinear regression for repeated measurements (SAS PROC NLMIXED; SAS Institute) was used to implement the temporal decomposition model. We used bootstrap method to estimate the SE of the estimate of the mean response at selected times. Estimates at different points were then compared using z test.
To descriptively assess the association between echocardiogram measures and composite event (death or reintervention), we considered average observed echocardiogram measurements stratified by presence or absence of composite event within 6-month intervals. Because not all patients had echocardiogram measurements just before an event within a given interval, for this analysis, only the last echocardiogram observation was carried forward to the end of each 6-month interval. This is similar to the last observation carried forward approach generally applied when one encounters missing longitudinal observations in longitudinal data analyses, which assumes that the hemodynamic measurements would have been stable during this period.
Continuous variables are summarized as mean (SD). Comparisons were made using Wilcoxon rank sum nonparametric test. Categorical data are described using frequencies and percentages. Comparisons were made using χ2 test or Fisher exact test when frequency was less than 5. Patient-specific echocardiogram profiles are graphed with straight lines connecting observed contiguous echocardiographic estimates (eFigure 2 in Supplement 2). Asymmetric 95% CIs of the estimated temporal trend of longitudinal responses were obtained using bootstrap percentile method. All analyses were performed using SAS statistical software version 9.4 (SAS Institute) and R software version 3.2.3 (The R Foundation). All P values were 2-tailed, and statistical significance was set at P < .05.
Results
Among the 2482 patients who received TAVR, 2404 patients (96.9%) had at least 1 postimplantation aortic valve mean gradient recorded, with a total of 10 746 echocardiograms. Their mean (SD) age was 84.5 (7.2) years, and 1179 (47.5%) were female (Table 1). Comorbidities were common, and all patients had severe aortic stenosis, with a mean (SD) aortic valve peak gradient of 71.3 (22.4) mm Hg, an aortic valve mean gradient of 44.0 (14.3) mm Hg, and an aortic valve area of 0.65 (0.19) cm2. A total of 1277 of 2449 patients (52.1%) received a 23-mm diameter THV and 1172 (47.9%) received a 26-mm diameter THV; the implantation technique was transapical in 1056 of 2482 patients (42.5%) and transfemoral in 1426 (57.5%). Baseline characteristics were similar in the SAVR cohort (Table 1). Median follow-up in the TAVR cohort was 3.1 years, and the mean (SD) follow-up was 2.9 (1.8) years. Overall survival and freedom from reintervention in the TAVR cohort at 5 years was 34% by nonadjusted parametric analysis. Median follow-up in SAVR was 3.2 years, and the mean (SD) follow-up was 2.9 (1.9) years. Overall survival and freedom from reintervention in SAVR at 5 years was 37% by nonadjusted parametric analysis.
Table 1. Transcatheter Aortic Valve Replacement (TAVR) and Surgical Aortic Valve Replacement (SAVR) Population Descriptions.
| Characteristic | TAVR Cohort, No./Total No.a (%) | SAVR Cohort, No./Total No.a (%) |
|---|---|---|
| PARTNER 1 cohort | ||
| Cohort A | 321/2482 (12.9) | 313/313 (100) |
| Cohort B | 165/2482 (6.6) | NA |
| Randomized continuing access cohort B | 39/2482 (1.6) | NA |
| Nonrandomized continuing access cohort A | 1742/2482 (70.2) | NA |
| Nonrandomized continuing access cohort B | 215/2482 (8.7) | NA |
| Baseline characteristics | ||
| Age, mean (SD), y | 84.5 (7.15) | 84.5 (6.33) |
| Female | 1179/2480 (47.5) | 134/313 (42.8) |
| White | 2342/2479 (94.5) | 294/313 (93.9) |
| Hypertension | 2280/2479 (92.0) | 294/313 (93.9) |
| Diabetes | 917/2482 (36.9) | 128/313 (40.9) |
| History of renal disease | 448/2482 (18.0) | 69/313 (22.0) |
| History of pulmonary disease | 1550/2478 (62.6) | 211/313 (67.4) |
| History of coronary artery disease | 1928/2479 (77.8) | 241/313 (77.0) |
| Presence of ≥1 preprocedure coronary artery stenosis >50% | 2216/2414 (91.8) | 264/288 (91.7) |
| History of myocardial infarction | 638/2467 (25.9) | 90/310 (29.0) |
| NYHA class 3 or 4 heart failure symptoms | 2354/2479 (95.0) | 297/313 (94.9) |
| Peripheral arterial disease | 1737/2482 (70.0) | 210/313 (67.1) |
| STS risk score >10 | 1552/2473 (62.8) | 231/313 (73.8) |
| Baseline echocardiogram data | ||
| LV mass, mean (SD), g | 250 (76.3) | 278 (86.2) |
| Visual ejection fraction, mean (SD), % | 52.5 (12.9) | 53.4 (12.4) |
| Doppler stroke volume index, mean (SD), mL/m2 | 34.9 (10.9) | 34.0 (10.7) |
| Peak AV gradient, mean (SD), mm Hg | 71.3 (22.4) | 73.3 (24.1) |
| Mean AV gradient, mean (SD), mm Hg | 44.0 (14.3) | 43.4 (14.3) |
| AV area, mean (SD), cm2 | 0.65 (0.19) | 0.64 (0.19) |
| Doppler velocity index, mean (SD), m/s | 0.20 (0.06) | 0.20 (0.05) |
| Aortic regurgitation (moderate or severe) | 261/2424 (10.8) | 43/302 (14.2) |
| Mitral regurgitation (moderate or severe) | 534/2419 (22.1) | 59/299 (19.7) |
| Procedural and postprocedural data | ||
| Access site for implantation: transapical | 1056/2482 (42.5) | NA |
| Prosthesis size, mm | ||
| 19 | NA | 37/310 (11.9) |
| 21 | NA | 124/310 (40.0) |
| 23 | 1277/2449 (52.1) | 109/310 (35.2) |
| 25 | NA | 37/310 (11.9) |
| 26 | 1172/2449 (47.9) | NA |
| 27 | NA | 3/310 (1.0) |
| Intraprocedural valve-in-valve | 56 (2.3) | NA |
| Prosthesis-patient mismatch, Y/N | ||
| Moderate | 717/2376 (30.2) | 88/285 (30.9) |
| Severe | 448/2376 (18.9) | 94/285 (33.0) |
Abbreviations: AV, aortic valve; LV, left ventricular; NA, not applicable; NYHA, New York Heart Association.
Patients with data available.
Five-year echocardiographic data were available in 424 patients who received TAVR. Comparison of paired, serial echocardiogram data in 399 patients showed favorable changes compared with baseline on first postimplantation study, with slight adverse changes at 5 years (Table 2). Paired baseline and late echocardiographic data in the 49 patients with 5-year echocardiograms in the SAVR cohort (serial echocardiogram data in 42 patients) are shown in eTable 1 in Supplement 2.
Table 2. Patients Who Received Transcatheter Aortic Valve Replacement With Available Paired Echocardiographic Data at Baseline, First Postimplantation, and 5 Years.
| Measurement | No.a | Mean (SD) | P Valueb | ||
|---|---|---|---|---|---|
| Preimplantation | First Postimplantation | 5 y | |||
| LV diameter, cm | |||||
| Diastolic | 309 | 4.55 (0.77) | 4.51 (0.80) | 4.36 (0.75) | .02 |
| Systolic | 272 | 3.36 (0.94) | 3.21 (0.88) | 3.24 (0.85) | .69 |
| LV mass, g | 309 | 251.0 (76.8) | 247.4 (76.3) | 199.7 (63.7) | <.001 |
| Ejection fraction, % | 392 | 52.7 (12.9) | 54.5 (11.5) | 54.0 (10.0) | .12 |
| Stroke volume, mL | 360 | 67.9 (19.5) | 67.2 (19.8) | 68.0 (20.6) | .59 |
| Stroke volume index, mL/m2 | 357 | 36.4 (10.5) | 36.0 (10.5) | 36.5 (11.5) | .54 |
| AV gradient, mm Hg | |||||
| Peak | 393 | 76.3 (24.4) | 20.5 (8.66) | 19.0 (11.2) | <.001 |
| Mean | 393 | 47.7 (15.9) | 11.0 (4.77) | 10.3 (6.40) | <.001 |
| AV area, cm2 | 360 | 0.66 (0.20) | 1.66 (0.50) | 1.57 (0.45) | .06 |
| Doppler velocity index, m/s | 370 | 0.20 (0.06) | 0.52 (0.14) | 0.50 (0.13) | .27 |
| Prosthesis-patient mismatch, No. (%), Y/N | |||||
| Moderate | 370 | NA | 115 (31.1) | 109 (29.5) | .54 |
| Severe | 370 | NA | 67 (18.1) | 79 (21.4) | .54 |
| Regurgitation (moderate or severe), No. (%) | |||||
| Total aortic | 387 | 38 (9.8) | 31 (8.0) | 44 (11.4) | .11 |
| Paravalvular | 395 | NA | 24 (6.1) | 19 (4.8) | .43 |
| Transvalvular | 395 | NA | 3 (0.8) | 16 (4.1) | .004 |
Abbreviations: AV, aortic valve; LV, left ventricular; NA, not applicable.
Patients with available data at all 3 times.
First postimplantation vs 5 years. Comparisons (continuous data) were made using Wilcoxon rank sum nonparametric test. Comparisons (categorical data) were made using χ2 test or Fisher exact test when frequency was less than 5.
Modeled Population Hemodynamic Trends
The overall TAVR population trend in aortic valve mean gradient revealed small changes, including a reduction from 12.1 (95% CI, 11.6-12.5) to 9.2 (95% CI, 9.06-9.41) mm Hg (P < .001) in the first few months postimplantation with a slight increase thereafter to 10.1 (95% CI, 9.80-10.5) mm Hg (P < .001) at 5 years (Figure 2A) (Table 3). Similar changes were noted in EOA and DVI (Figure 2B and C) (Tables 3). Patients with 23-mm diameter and 26-mm diameter prostheses displayed similar patterns (albeit with higher gradients and smaller DVIs and EOAs in the smaller THVs), as did those with transapical vs transfemoral implantation routes (Table 3). The SAVR population did not show any significant hemodynamic changes over time (Table 3) (eFigure 3 in Supplement 2).
Figure 2. Temporal Trends in Transcatheter Aortic Valve Replacement.
Solid lines indicate the parametric estimate of the temporal trend after procedure; dotted lines indicate 95% CIs. Filled circles indicate grouped data without regard to the repeated measurements based on time intervals, provided here as a crude verification of the model fit. DVI indicates Doppler velocity index.
Table 3. Hemodynamic Trends.
| Model | Estimate (95% CI) | ||
|---|---|---|---|
| Mean Gradient, mm Hg | EOA, cm2 | DVI | |
| Early hemodynamic trendsa | |||
| TAVR (first postimplantation to 3 mo) | |||
| All patients (2387 mean gradient, 2361 EOA, and 2372 DVI measurements) | 12.1 (11.6-12.5) to 9.2 (9.06-9.41) |
1.63 (1.61-1.64) to 1.72 (1.69-1.75) |
0.509 (0.504-0.515) to 0.537 (0.530-0.545) |
| P valueb | <.001 | <.001 | <.001 |
| Transapical TAVR (1002 mean gradient, 990 EOA, and 994 DVI measurements) | 11.3 (10.6-12.3) to 8.6 (8.31-8.75) |
1.59 (1.54-1.61) to 1.72 (1.68-1.76) |
0.495 (0.478-0.504) to 0.539 (0.530-0.546) |
| P valueb | <.001 | <.001 | <.001 |
| Transfemoral TAVR (1385 mean gradient, 1371 EOA, and 1378 DVI measurements) | 12.5 (12.0-12.9) to 9.7 (9.50-9.91) |
1.65 (1.62-1.67) to 1.71 (1.67-1.77) |
0.516 (0.510-0.521) to 0.533 (0.522-0.547) |
| P valueb | <.001 | .03 | .005 |
| 23-mm Diameter THV (1218 mean gradient, 1203 EOA, and 1208 DVI measurements) | 13.1 (12.5-14.6) to 9.8 (9.61-10.1) |
1.50 (1.48-1.52) to 1.61 (1.57-1.67) |
0.505 (0.497-0.511) to 0.543 (0.531-0.552) |
| P valueb | <.001 | <.001 | <.001 |
| 26-mm Diameter THV (1137 mean gradient, 1126 EOA, and 1132 DVI measurements) | 11.1 (10.7-11.5) to 8.6 (8.38-8.79) |
1.76 (1.74-1.78) to 1.84 (1.79-1.90) |
0.515 (0.508-0.523) to 0.533 (0.517-0.543) |
| P valueb | <.001 | .004 | .04 |
| SAVR (first postimplantation to 3 mo) | |||
| All patients (286 mean gradient, 282 EOA, and 284 DVI measurements) | 10.3 (5.44-18.9) to 10.8 (10.2-11.4) |
1.46 (1.40-1.50) to 1.64 (1.44-1.88) |
0.486 (0.457-0.498) to 0.486 (0.476-0.510) |
| P valueb | .85 | .09 | >.99 |
| Midterm hemodynamic trendsa | |||
| TAVR (3 mo to 5 y) | |||
| All patients (1911 mean gradient, 1900 EOA, and 1901 DVI measurements) | 9.2 (9.06-9.41) to 10.1 (9.80-10.5) |
1.72 (1.69-1.75) to 1.63 (1.61-1.64) |
0.537 (0.530-0.545) to 0.510 (0.505-0.515) |
| P valueb | <.001 | <.001 | <.001 |
| Transapical TAVR (781 mean gradient, 778 EOA, and 778 DVI measurements) | 8.6 (8.31-8.75) to 9.9 (9.45-10.4) |
1.72 (1.68-1.76) to 1.59 (1.56-1.62) |
0.539 (0.530-0.546) to 0.500 (0.494-0.506) |
| P valueb | <.001 | <.001 | <.001 |
| Transfemoral TAVR (1130 mean gradient, 1122 EOA, and 1123 DVI measurements) | 9.7 (9.50-9.91) to 10.4 (10.0-10.8) |
1.71 (1.67-1.77) to 1.65 (1.62-1.67) |
0.533 (0.522-0.547) to 0.516 (0.510-0.521) |
| P valueb | .003 | .03 | .007 |
| 23-mm Diameter THV (974 mean gradient, 969 EOA, and 970 DVI measurements) | 9.8 (9.61-10.1) to 11.2 (10.5-11.8) |
1.61 (1.57-1.67) to 1.50 (1.48-1.52) |
0.543 (0.531-0.552) to 0.506 (0.500-0.512) |
| P valueb | <.001 | <.001 | <.001 |
| 26-mm Diameter THV (912 mean gradient, 906 EOA, and 906 DVI measurements) | 8.6 (8.38-8.79) to 9.0 (8.58-9.28) |
1.84 (1.79-1.90) to 1.77 (1.74-1.78) |
0.533 (0.517-0.543) to 0.515 (0.509-0.523) |
| P valueb | .05 | .004 | .04 |
| SAVR (3 mo to 5 y) | |||
| All patients (206 mean gradient, 204 EOA, and 204 DVI measurements) | 10.8 (10.2-11.4) to 9.9 (8.97-10.9) |
1.64 (1.44-1.88) to 1.46 (1.41-1.50) |
0.486 (0.476-0.510) to 0.486 (0.463-0.498) |
| P valueb | .20 | .09 | .89 |
Abbreviations: DVI, Doppler velocity index; EOA, effective orifice area; SAVR, surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement; THV, transcatheter heart valve.
The 95% CIs in this table are not necessarily symmetric and are equivalent to 2 SEs rather than SD. The P values shown test the difference of the mean estimates between the 2 times and indicate statistical but not necessarily clinical significance.
Estimates at different points were compared using z test.
To address the possible clinical significance of these hemodynamic trends in TAVR, we assessed vital status and freedom from reintervention at each 6-month period throughout follow-up and then compared hemodynamic data from last trial echocardiogram (eFigure 4 in Supplement 2). The relationship between the last mean gradient and event-free survival changed over time, such that mean gradients were higher among survivors for up to 3 years of follow-up but similar in survivors and nonsurvivors thereafter. In contrast, there was no time-varying association between either last EOA or DVI and event-free survival. Because these findings suggested a relationship between adverse events and low-flow/low-output states, we also compared ejection fraction and stroke volume index by event-free survival. Ejection fraction showed the same time-varying relationship as mean gradient, while stroke volume index was uniformly higher in event-free survivors than in those with either death or aortic valve reintervention. The SAVR cohort was too small to perform a similar analysis.
Transvalvular Aortic Regurgitation
Eighty-nine patients (3.6%) in the TAVR cohort had moderate transvalvular AR and 6 patients (0.2%) had severe transvalvular AR on at least 1 echocardiogram during the trial, for a total of 89 patients (3.7%) with moderate or severe AR. Of the 6 with severe AR, 4 died (3 of noncardiovascular causes and 1 of an unknown cause), including the 3 patients included in the abnormal hemodynamics count. Of those with echocardiograms at postimplantation and 5 years, 16 patients had moderate or severe AR at 5 years compared with 3 at postimplantation. No patients in the SAVR cohort had severe transvalvular AR during follow-up (2 patients had moderate AR).
Aortic Valve Reinterventions
Twenty patients (0.8%) in the TAVR cohort and 1 (0.3%) in the SAVR cohort met the protocol definition of aortic valve reintervention. In the TAVR group, 9 had SAVR, 8 had late valve-in-valve TAVR, and 3 had balloon aortic valvuloplasty. While most reinterventions were performed for paravalvular leak, 5 (25%) were performed for structural THV indications, including aortic stenosis in 1 patient, valve thrombosis in 1 patient, and transvalvular AR in 3 patients. Ten of 20 patients (50%) showed no appreciable or consistent hemodynamic changes, although the last available echocardiogram data in 9 of these 10 preceded the reintervention by more than a month and up to several years. Of the remaining 10, 1 showed a high initial mean gradient postimplantation that did not change, and 5 had no postimplantation echocardiogram data. Only 4 (20%) showed the classic pattern of increasing gradients and decreasing EOA and DVI consistent with prosthetic stenosis; just 1 of the patients with a structural indication for reintervention was in this group.
Hemodynamic Outliers and Transvalvular Regurgitation
Because the extreme risk in the PARTNER 1 population may have precluded reintervention even in the presence of adverse hemodynamics, we evaluated the incidence of large changes between subsequent echocardiograms as well as outlier values noted on any postimplantation echocardiogram. Both methods are recommended by the Valve Academic Research Consortium 2 (VARC-2) to assess time-related valve safety. In applying the VARC-2 criteria for mild aortic stenosis, we found that large numbers of patients in the TAVR and SAVR cohorts met these criteria for hemodynamic outliers, from 3.5% to 49.0% of successful TAVRs and 1.6% to 45.0% of successful SAVRs, making these end points impractical for case reviews (eTable 2 in Supplement 2). Thus, we also considered additional criteria for severe stenosis, which identified the most severely abnormal hemodynamic findings.
Using these more extreme cut points, an absolute mean gradient of 40 mm Hg or greater occurred during follow-up in 11 patients (0.5%) in the TAVR cohort, of whom 8 died (2 of cardiovascular causes, 3 of noncardiovascular causes, and 3 of unknown causes), and 1 had reintervention (82% death /reintervention). This is proportionally comparable with the numbers of patients in the overall cohort experiencing these events. In addition to the 20 reinterventions, 1566 patients (63.1%) died, of whom 509 patients died of cardiovascular causes, 566 of noncardiovascular causes, and 491 of unknown causes. An absolute value of DVI of 0.25 or less occurred in 46 patients (1.9%), of whom 22 died (5 of cardiovascular causes, 5 of noncardiovascular causes, and 12 of unknown causes), and none had reintervention, for a death/reintervention incidence of 50%. A between-echocardiogram increase in aortic valve mean gradient of 20 mm Hg or greater occurred in 10 patients (0.5%), 6 of whom died (3 of cardiovascular causes, 2 of noncardiovascular causes, and 1 of unknown causes), and none had reintervention, for an event incidence of 60%.
Applying these same criteria to the 292 patients in the SAVR cohort showed that 1 (0.3%) had an absolute mean gradient of 40 mm Hg or greater and died of noncardiovascular causes, and no patient had a between-echocardiogram increase in mean gradient of 20 mm Hg or greater. One patient had a valve-in-valve TAVR placed. An absolute value of DVI of 0.25 or less occurred in 6 patients (0.2%), of whom 5 died (1 of cardiovascular causes, 2 of noncardiovascular causes, and 2 of unknown causes).
Discussion
This report of 2482 patients who received TAVR in the PARTNER 1 Trial is the largest, core laboratory–based study of transcatheter heart valves performed to date to our knowledge. Population modeling of hemodynamic trends showed 2 phases in TAVR, with consistency across variables (mean gradient, DVI, and EOA), reflecting early favorable changes in the first few months with minimal longitudinal changes out to 5 years. Similarly, severely abnormal hemodynamic data in individual patients, which might be suggestive of valve thrombosis or stenosis, were rare in this protocol-driven database. However, echocardiogram data suggest a slight hemodynamic deterioration and increasing AR prevalence. Nevertheless, an early signal for deterioration also noted in prior reports does not appear to be more frequent in patients after TAVR than in patients after SAVR. Together, these data demonstrate excellent durability of THV, comparable with surgical prostheses and consistent with reports in smaller cohorts (including a small subset of the current patients) or those with shorter follow-up.
Because of the very large population, we were able to create statistically robust models of longitudinal hemodynamic trends in the TAVR cohort. The early favorable changes have been reported previously in patients after SAVRs and were also seen in this study. These changes suggest the possibility of early remodeling of the valves and/or postprocedural changes in loading conditions or physiology, such as left ventricular function. The possibility of very early valve thrombosis within days of implantation, as has been reported, followed by short-term resolution, cannot be excluded given our data’s limited temporal resolution. The observation of relative stability to 5 years is reassuring. Nonetheless, further evaluation for a longer follow-up period is warranted, as most reports suggest that bioprosthetic dysfunction is minimal until 8 to 10 years after implantation.
An important word of caution arises from the Valve-in-Valve International Data Registry, which reported a time to intervention for bioprosthetic valve failure of only 9 (interquartile range, 6-12) years. Structural valve deterioration is not typically an acute process, and the development of stenosis (40% of patients), transvalvular regurgitation (30% of patients), or both stenosis and regurgitation (30% of patients) would presumably begin well before the need for intervention. Thus, the small but significant changes in EOA and DVI as well as the increase in the incidence of moderate/severe transvalvular AR between discharge and 5 years is important, albeit probably expected for any tissue bioprosthesis, and mandates continued surveillance.
Although the patients in the SAVR cohort in the PARTNER Trial cannot be compared directly with the pooled TAVR data because these groups were not randomized, the not-dissimilar findings are reassuring, especially given the differences between the procedures (retention of leaflets in TAVR) and the devices (presence of sewing rings in SAVR). Further, the longitudinal changes we observed are similar in magnitude and direction to that reported in 2 of the few other core laboratory–based studies of surgical bioprostheses, both conducted in elderly populations with a high prevalence of comorbid conditions.
Much attention has focused on recent reports of valve thrombosis and leaflet thickening, with incidence estimates varying greatly depending on the timing and type of assessments. While this appears to occur early after implantation, it may be associated with later leaflet degeneration and/or pannus ingrowth, leading to structural failure. Our finding of infrequent hemodynamic deterioration is reassuring, as it minimizes concerns regarding possible longer-term consequences of valve thrombosis. In contrast, the increasing incidence of transvalvular regurgitation in TAVR is of concern.
Finding associations between clinical events and preceding adverse hemodynamic changes is difficult, given the periodic nature of clinical trial surveillance assessments. Although causal hemodynamic changes must precede an event, they may occur nearly simultaneously or at least during the same interval between assessments, making such changes difficult to discern when comparing surveillance hemodynamic data and subsequent events. To minimize this problem, we compared the last hemodynamic measures obtained in an event-free cohort vs one with death or reintervention, finding that event-free survival was associated with higher mean gradient and ejection fraction up to 3 years, but there was no relationship between DVI and EOA and events at any point during follow-up. If events were caused by valve dysfunction, mean gradients would be lower in survivors and DVI and EOA higher, relationships which were not observed. Instead, the constant association between events and lower stroke volume index suggests that low-flow/low-output states may be more closely associated with morbidity than adverse valve hemodynamics. The association of these distinct hemodynamic patterns with death/reintervention cannot be proven but likely reflects a primary causative role for underlying or concomitant disease rather than valve deterioration.
Reintervention was quite rare in this TAVR cohort, became less frequent over time, and was usually not due to structural deterioration of THV. Unfortunately, protocol-driven echocardiogram data were temporally distant and uninformative, highlighting the critical importance of real-time clinical data. Collection of such information in future studies could aid considerably in finding associations between hemodynamic changes and events.
Valve durability is an important and ongoing question, and consensus guidelines have been developed to aid in its detection. To our knowledge, the current VARC-2 cut points for time-varying safety and mild aortic stenosis have not been previously applied in a large patient cohort. We found that they were commonly met by patients in both the TAVR and SAVR cohorts, suggesting that they may not adequately discriminate adverse changes. This is especially true for decreases in EOA noted on subsequent echocardiograms, which were found in one-third to one-half of patients. While the known difficulty in accurately measuring left ventricular outflow track diameter (required for calculation of valve area using the continuity equation) may introduce substantial variability, the use of DVI, a dimensionless ratio of the integral of flow velocities above and below the valve that avoids the need for outflow tract measurement and controls for high flow states such as regurgitation, offered only slight improvement. While other investigators have offered varying cut points, including greater than 50% or a 10–mm Hg increase in mean gradient, variability in echocardiogram data may be hard to distinguish from true hemodynamic changes; either or both can arise from physiologic changes (eg, anemia or reduced ventricular function), deficiencies in image acquisition, and/or inaccurate analysis (which should have been minimized in this core laboratory–based study with excellent reproducibility). Future proposed parameters assessing valvular dysfunction should be tested empirically in the PARTNER Trial or other large data sets to ensure that they provide the desired discrimination.
The VARC-2 criteria for severe stenosis were rarely met by patients in either the TAVR or SAVR cohorts and were not associated with an excess incidence of death or reintervention compared with the overall trial results. Importantly, these criteria were designed to detect mild AS (from any cause) during postimplantation surveillance and not to define early structural valve deterioration. In this respect, large changes in hemodynamics between echocardiograms were infrequent and of unclear clinical significance. Whether this is caused by inherent variability in echocardiogram findings, the temporal distance of adverse hemodynamics from clinical events, absent hemodynamic changes in patients with valve dysfunction (as has been reported with valve thrombosis), or some other reason, our findings suggest that annual surveillance echocardiography after TAVR may not always provide the expected early warning for valve deterioration in the first 5 years after implantation.
Limitations
The PARTNER 1 Trial represents a unique and carefully curated large data set, which, despite low event-free survival (34% at 5 years), yielded a robust patient cohort at 5 years, representing the entire trial rather than a substudy population. Nevertheless, this is a small proportion of those enrolled, largely owing to high mortality in this aged cohort. Although survivorship bias may have affected our results, with some deaths perhaps due to undetected valve deterioration, the lack of adverse population trends in valve hemodynamics or excess events in those individuals with outlier echocardiogram findings mitigates these concerns. As TAVR use broadens to younger and less comorbid populations, long-term durability becomes increasingly important. To this end, the timely collection and reporting of data establishing long-term durability beyond the 5 years reported in this article is still needed.
As noted above, the protocol-mandated surveillance echocardiograms were performed only at infrequent times that were not necessarily related to clinical events. While not ideal, this would be true in any prospective study design and can be mitigated by rigorous clinical data collected at the time of adverse events. The hemodynamic trend data were extrapolated from the actual dates of echocardiogram performance, which varied by patient regardless of the protocol-specified duration postimplantation. Further, early changes from implantation to 3 months were modeled on population trends over this interval, as only 30-day and 6-month echocardiograms were mandated.
Although all echocardiograms were interpreted centrally with strict quality controls, our results may still be limited by poor image quality and/or completeness, as late follow-up echocardiograms were often obtained in the community rather than at the certified enrolling sites. The small number of patients with hemodynamic deterioration precluded modeling of predictors or associated factors. Additionally, paravalvular leak is a significant complication of TAVR, especially in the first generation of valves, with significant prognostic implications. Finally, these data apply only to the first-generation SAPIEN prostheses and not to later, more current iterations of this device or to other THV designs. In addition, only 23-mm- and 26-mm-diameter valves were available, and device choice was made based on single-plane linear systolic annular dimensions; thus, the effect of sizing parameters cannot be assessed accurately given the current methods of sizing the THV.
Conclusions
This large, core laboratory evaluation of 2482 patients receiving TAVR and 313 receiving SAVR in the PARTNER 1 Trial demonstrates excellent longitudinal durability of the SAPIEN THV using both population hemodynamic trends as well as case reviews of reintervention and patients with large adverse changes between echocardiograms. Abnormal findings in individual patients suggestive of valve thrombosis or structural deterioration were rare in this protocol-driven database and require further investigation.
Trial protocol.
eAppendix. Institutional review board approval.
eFigure 1. CONSORT diagram showing patient flow.
eFigure 2. Hemodynamic data.
eFigure 3. Temporal trends in surgical aortic valve replacement.
eFigure 4. Associations between last mean gradient, Doppler velocity index (DVI), ejection fraction (EF) and stroke volume index (SVI) by vital status/reintervention during each 6 month interval of follow-up.
eTable 1. Surgical aortic valve replacement patients with available paired echocardiographic data at baseline, first postimplantation, and 5 years.
eTable 2. Incidence of VARC-2 and other selected cut points for severe aortic stenosis.
References
- 1.Mack MJ, Leon MB, Smith CR, et al. ; PARTNER 1 Trial Investigators . 5-year outcomes of transcatheter aortic valve replacement or surgical aortic valve replacement for high surgical risk patients with aortic stenosis (PARTNER 1): a randomised controlled trial. Lancet. 2015;385(9986):2477-2484. [DOI] [PubMed] [Google Scholar]
- 2.Kapadia SR, Leon MB, Makkar RR, et al. ; PARTNER Trial Investigators . 5-year outcomes of transcatheter aortic valve replacement compared with standard treatment for patients with inoperable aortic stenosis (PARTNER 1): a randomised controlled trial. Lancet. 2015;385(9986):2485-2491. [DOI] [PubMed] [Google Scholar]
- 3.Deeb GM, Reardon MJ, Chetcuti S, et al. ; CoreValve US Clinical Investigators . 3-Year outcomes in high-risk patients who underwent surgical or transcatheter aortic valve replacement. J Am Coll Cardiol. 2016;67(22):2565-2574. [DOI] [PubMed] [Google Scholar]
- 4.Brennan JM, Edwards FH, Zhao Y, et al. ; DEcIDE AVR (Developing Evidence to Inform Decisions about Effectiveness–Aortic Valve Replacement) Research Team . Long-term safety and effectiveness of mechanical versus biologic aortic valve prostheses in older patients: results from the Society of Thoracic Surgeons Adult Cardiac Surgery National Database. Circulation. 2013;127(16):1647-1655. [DOI] [PubMed] [Google Scholar]
- 5.Zhao DF, Seco M, Wu JJ, et al. . Mechanical versus bioprosthetic aortic valve replacement in middle-aged adults: a systematic review and meta-analysis. Ann Thorac Surg. 2016;102(1):315-327. [DOI] [PubMed] [Google Scholar]
- 6.Grover FL, Vemulapalli S, Carroll JD, et al. ; STS/ACC TVT Registry . 2016 annual report of the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy Registry. Ann Thorac Surg. 2017;103(3):1021-1035. [DOI] [PubMed] [Google Scholar]
- 7.Makkar RR, Fontana G, Jilaihawi H, et al. . Possible subclinical leaflet thrombosis in bioprosthetic aortic valves. N Engl J Med. 2015;373(21):2015-2024. [DOI] [PubMed] [Google Scholar]
- 8.Del Trigo M, Muñoz-Garcia AJ, Wijeysundera HC, et al. . Incidence, timing, and predictors of valve hemodynamic deterioration after transcatheter aortic valve replacement: multicenter registry. J Am Coll Cardiol. 2016;67(6):644-655. [DOI] [PubMed] [Google Scholar]
- 9.Dangas GD, Weitz JI, Giustino G, Makkar R, Mehran R. Prosthetic heart valve thrombosis. J Am Coll Cardiol. 2016;68(24):2670-2689. [DOI] [PubMed] [Google Scholar]
- 10.Fearon WF, Kodali S, Doshi D, et al. ; PARTNER Trial Investigators . Outcomes after transfemoral transcatheter aortic valve replacement: a comparison of the randomized PARTNER (Placement of AoRTic TraNscathetER Valves) trial with the NRCA (Nonrandomized Continued Access) registry. JACC Cardiovasc Interv. 2014;7(11):1245-1251. [DOI] [PubMed] [Google Scholar]
- 11.Dewey TM, Bowers B, Thourani VH, et al. . Transapical aortic valve replacement for severe aortic stenosis: results from the nonrandomized continued access cohort of the PARTNER trial. Ann Thorac Surg. 2013;96(6):2083-2089. [DOI] [PubMed] [Google Scholar]
- 12.Leon MB, Smith CR, Mack M, et al. ; PARTNER Trial Investigators . Transcatheter aortic-valve implantation for aortic stenosis in patients who cannot undergo surgery. N Engl J Med. 2010;363(17):1597-1607. [DOI] [PubMed] [Google Scholar]
- 13.Smith CR, Leon MB, Mack MJ, et al. ; PARTNER Trial Investigators . Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med. 2011;364(23):2187-2198. [DOI] [PubMed] [Google Scholar]
- 14.Douglas PS, DeCara JM, Devereux RB, et al. ; American Society of Echocardiography Standards; American College of Cardiology Foundation . Echocardiographic imaging in clinical trials: American Society of Echocardiography Standards for echocardiography core laboratories: endorsed by the American College of Cardiology Foundation. J Am Soc Echocardiogr. 2009;22(7):755-765. [DOI] [PubMed] [Google Scholar]
- 15.Douglas PS, Waugh RA, Bloomfield G, et al. . Implementation of echocardiography core laboratory best practices: a case study of the PARTNER I trial. J Am Soc Echocardiogr. 2013;26(4):348-358.e3. [DOI] [PubMed] [Google Scholar]
- 16.Rajeswaran J, Blackstone EH. A multiphase non-linear mixed effects model: an application to spirometry after lung transplantation [published online 2014]. Stat Methods Med Res. doi: 10.1177/0962280214537255 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Diggle PJ, Heagerty P, Liang KY, Zeger SL. Analysis of Longitudinal Data. 2nd ed New York, NY: Oxford University Press; 2002. [Google Scholar]
- 18.Blackstone EH. Breaking down barriers: helpful breakthrough statistical methods you need to understand better. J Thorac Cardiovasc Surg. 2001;122(3):430-439. [DOI] [PubMed] [Google Scholar]
- 19.Efron B, Tibshirani RJ. An Introduction to the Bootstrap. New York, NY: Chapman and Hall/CRC; 1998. [Google Scholar]
- 20.Verneke G, Molenberghs G. Linear Mixed Models for Longitudinal Data. New York, NY: Springer-Verlag; 2001. [Google Scholar]
- 21.Kappetein AP, Head SJ, Généreux P, et al. ; Valve Academic Research Consortium (VARC)-2 . Updated standardized endpoint definitions for transcatheter aortic valve implantation: the Valve Academic Research Consortium-2 consensus document (VARC-2). Eur J Cardiothorac Surg. 2012;42(5):S45-S60. [DOI] [PubMed] [Google Scholar]
- 22.Dvir D, Webb JG, Bleiziffer S, et al. ; Valve-in-Valve International Data Registry Investigators . Transcatheter aortic valve implantation in failed bioprosthetic surgical valves. JAMA. 2014;312(2):162-170. [DOI] [PubMed] [Google Scholar]
- 23.Daubert MA, Weissman NJ, Hahn RT, et al. . Long-term valve performance of TAVR and SAVR: a report from the PARTNER I Trial. JACC Cardiovasc Imaging. 2016;S1936-878X(16)30895-6. [DOI] [PubMed] [Google Scholar]
- 24.Szeto WY, Svensson LG, Rajeswaran J, et al. ; Placement of Aortic Transcatheter Valves Trial Investigators . Appropriate patient selection or health care rationing? lessons from surgical aortic valve replacement in the Placement of Aortic Transcatheter Valves I trial. J Thorac Cardiovasc Surg. 2015;150(3):557-568.e11. [DOI] [PubMed] [Google Scholar]
- 25.Asch FM, Heimansohn D, Doyle D, et al. . Mitroflow aortic bioprosthesis 5-year follow-up: North American prospective multicenter study. Ann Thorac Surg. 2012;94(4):1198-1203. [DOI] [PubMed] [Google Scholar]
- 26.Goldman S, Cheung A, Bavaria JE, Petracek MR, Groh MA, Schaff HV. Midterm, multicenter clinical and hemodynamic results for the Trifecta aortic pericardial valve. J Thorac Cardiovasc Surg. 2017;153(3):561-569.e2. [DOI] [PubMed] [Google Scholar]
- 27.Egbe AC, Pislaru SV, Pellikka PA, et al. . Bioprosthetic valve thrombosis versus structural failure: clinical and echocardiographic predictors. J Am Coll Cardiol. 2015;66(21):2285-2294. [DOI] [PubMed] [Google Scholar]
- 28.Kodali S, Pibarot P, Douglas PS, et al. . Paravalvular regurgitation after transcatheter aortic valve replacement with the Edwards sapien valve in the PARTNER trial: characterizing patients and impact on outcomes. Eur Heart J. 2015;36(7):449-456. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial protocol.
eAppendix. Institutional review board approval.
eFigure 1. CONSORT diagram showing patient flow.
eFigure 2. Hemodynamic data.
eFigure 3. Temporal trends in surgical aortic valve replacement.
eFigure 4. Associations between last mean gradient, Doppler velocity index (DVI), ejection fraction (EF) and stroke volume index (SVI) by vital status/reintervention during each 6 month interval of follow-up.
eTable 1. Surgical aortic valve replacement patients with available paired echocardiographic data at baseline, first postimplantation, and 5 years.
eTable 2. Incidence of VARC-2 and other selected cut points for severe aortic stenosis.