Incidence and Outcomes of Surgical Bailout During TAVR: Insights From the STS/ACC TVT Registry
Focus on Complex Transcatheter Aortic Valve Replacement Cases
Central Illustration
Abstract
Objectives:
The aim of this study was to evaluate the incidence and outcomes of surgical bailout during transcatheter aortic valve replacement (TAVR).
Background:
The incidence and outcomes of unplanned conversion to open heart surgery, or “surgical bailout,” during TAVR are not well characterized.
Methods:
Data from the Society of Thoracic Surgeons/American College of Cardiology TVT (Transcatheter Valve Therapy) Registry was analyzed with respect to whether surgical bailout was performed during the index TAVR procedure. A Cox proportional hazards models was used to evaluate 1-year mortality and major adverse cardiovascular events.
Results:
Between November 2011 and September 2015, a total of 47,546 patients underwent TAVR. Surgical bailout during TAVR was performed in 1.17% of the cases (n = 558); the most frequent indications were valve dislodgement (22%), ventricular rupture (19.9%), and aortic valve annular rupture (14.2%). The incidence of surgical bailout significantly decreased over time (first tertile 1.25%, second tertile 1.43%, third tertile 1.04%; p = 0.0088). The 30-day and 1-year incidence of major adverse cardiovascular events (54.6% vs. 7.4% [p < 0.0001] and 63.92% vs. 20.29% [p < 0.0001]) and all-cause mortality (50.00% vs. 4.98% [p < 0.0001] and 59.79% vs. 17.06% [p < 0.0001]) were significantly higher in those who underwent bailout. Independent predictors of surgical bailout included female sex, hemoglobin, left ventricular ejection fraction, nonelective cases, and nonfemoral access. Body surface area was the only independent predictor of survival after surgical bailout.
Conclusions:
In a large, nationally representative registry, the need for surgical bailout in patients undergoing TAVR is low, and its incidence has decreased over time. However, surgical bailout after TAVR is associated with poor outcomes, including 50% mortality at 30 days.
Introduction
Transcatheter aortic valve replacement (TAVR) is an established therapy for patients with severe aortic stenosis at increased risk for surgery. However, TAVR still carries risk for major intraoperative complications, which may require emergent conversion to open heart surgery, or “surgical bailout.” TAVR clinical trials have focused on procedural success and selected clinical outcomes, and therefore data on the incidence of surgical bailout and its clinical outcomes are limited. Conversion to open heart surgery during TAVR has been reported in up to 7.7% of cases and appears to be associated with increased morbidity and mortality (1–4). Available data suggest an important learning curve with TAVR, with significant improvements over time in the rates of procedural success and complications (2,5,6), and it is possible that the need for surgical bailout may have also decreased over time. Therefore, we sought to evaluate the overall incidence of surgical bailout during TAVR, as well as the trend in its incidence over time, its post-operative outcomes, and the possible predictors of need for and survival after surgical bailout from the National Cardiovascular Data Registry Society of Thoracic Surgeons/American College of Cardiology TVT (Transcatheter Valve Therapy) Registry.
Methods
The TVT Registry is a collaborative initiative that collects data of all commercial TAVR cases in the United States. Registry activities have been approved by the Duke University Institutional Review Board, and quality checks are performed at the Duke Clinical Research Institute and the National Cardiovascular Data Registry data warehouse. The TVT Registry is linked with Centers for Medicare and Medicaid Services (CMS) claims to capture clinical events requiring hospitalization. TVT Registry characteristics, data acquisition, event detection, quality control, and analysis have been described previously (7).
Study population
All patients with severe aortic stenosis from 396 sites who underwent TAVR from November 1, 2011, to September 30, 2015, were included in our study population. Patients were excluded from this cohort if additional procedures (transcatheter mitral valve repair or replacement) were performed at the time of TAVR (n = 101), no or both valve systems (balloon- and self-expandable) were used (n = 41), or surgical bailout status was missing (n = 122). Only the first admission for a TAVR patient was included. The incidence of surgical bailout was defined as the need for emergent conversion to open heart surgery for any reason during the index procedure. The need for emergent vascular surgery is not included in this analysis.
Definitions and endpoints
Clinical endpoints were assessed at hospital discharge, 30 days, and 1 year after TAVR. In-hospital outcomes included all-cause mortality, myocardial infarction (MI), stroke, major vascular and major bleeding complications, any bleeding, access-site bleeding, new permanent pacemaker implantation, new atrial fibrillation, new need for dialysis, and aortic valve reintervention. The Valve Academic Research Consortium-2 definitions of all-cause mortality, MI, stroke, and major vascular and bleeding complications were used for the in-hospital outcomes. Major adverse cardiovascular events (MACE) were defined as the composite of all-cause mortality, MI, and stroke. Net adverse cardiovascular events were defined as the composite of all-cause mortality, MI, stroke, and major vascular or bleeding complications. Thirty-day and 1-year outcomes were identified from CMS-linked claims data using International Classification of Diseases-Ninth Revision codes (Online Table 1).
Trends in surgical bailout over time
The incidence of surgical bailout and adverse events over time was assessed by stratifying patients into 3 tertiles on the basis of their procedure date. The first period (tertile) was from November 1, 2011, to February 22, 2013, the second from February 23, 2013, to June 13, 2014, and the third from June 14, 2014, to September 20, 2015, and included 5,933, 13,316, and 28,297 patients, respectively. The trends in the use of nonfemoral access and the number of new TAVR sites, defined as a hospital not participating in TAVR clinical trials and performing its first TAVR case during a particular time period, were analyzed in an attempt to elucidate possible reasons associated with the change in the incidence of surgical bailout over time.
Statistical analysis
Baseline clinical and procedural characteristics, as well as clinical outcomes, were compared according to surgical bailout status. Categorical variables are reported as frequencies and percentages and were compared using the Pearson chi-square test. Continuous variables are presented as median and interquartile range and were compared using the Wilcoxon-Mann-Whitney test. Pairwise comparisons were carried out among the tertiles using the Bonferroni multiple comparison test. One-year all-cause mortality and MACE rates were calculated using the Kaplan-Meier method and compared using the log-rank test. Cox proportional hazards regression modeling was performed to assess the unadjusted and adjusted associations between surgical bailout and 1-year all-cause mortality and MACE, as well as for the 30-day landmark analysis of mortality and MACE. For all time-to-event analysis, the marginal model approach was used to account for the clustering effect within each hospital. Missing data among the baseline and procedural characteristics were handled using multiple imputations. A logistic regression model was carried out to identify variables associated with need for and survival after surgical bailout. Only data from the latest 2 tertiles by procedure date (February 23, 2013, to September 30, 2015) were used, to avoid the learning curve during the early experience and excluding patients who had the valve system type missing. Baseline and procedural characteristics were assessed for significance in a univariate fashion, and statistically significant covariates (p < 0.10) were entered in the final model. Additionally, a logistic regression was performed to identify predictors of each reason for surgical bailout with more than 25 cases. Unadjusted and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) are presented. The generalized estimating equation was used to account for the clustering effect within each hospital. A p value < 0.05 was considered to indicate statistical significance. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina).
Results
Study population
A total of 48,161 patients underwent TAVR at 396 sites in the United States between November 1, 2011, and September 30, 2015, from which 47,546 were eligible for this analysis. Emergent conversion to open heart surgery, or surgical bailout, was performed in 558 patients (1.17%). The most common single reason for surgical bailout was ventricular rupture (19.89%), followed by prosthetic valve dislodgement into the left ventricle (19.35%), annular rupture (14.16%), aortic dissection (8.24%), and coronary occlusion (6.09%) (Figure 1A). The baseline clinical and procedural characteristics of the patients grouped by surgical bailout status are presented in Tables 1 and 2, respectively. Patients who required surgical bailout were older, were more likely to be female, had a smaller median body surface area, and a higher median left ventricular ejection fraction. More patients in the surgical bailout group had general anesthesia, were on inotropes at the time of TAVR, underwent emergent or salvage procedures, and had longer fluoroscopy times. Nonfemoral access was more commonly used in the surgical bailout cohort, driven primarily by increased use of transaortic access. Additionally, the total institutional TAVR volume was lower in the surgical bailout group.
Reasons for Surgical Bailout and In-Hospital All-Cause Mortality
(A) Percentage of patients (n = 558) requiring surgical bailout for each reason. (B) In-hospital all-cause mortality rates by each reason for surgical bailout and need for cardiopulmonary bypass. AO = aorta; LV = left ventricle.
| Overall (N = 47,546) | No Bailout (n = 46,988) | Bailout (n = 558) | p Value | |
|---|---|---|---|---|
| Age, yrs | 83 (77–88) | 83 (77–88) | 85 (80–88) | <0.0001 |
| Female | 22,900 (48.2) | 22,538 (48.0) | 362 (64.9) | <0.0001 |
| Race white | 44,720 (94.1) | 44,193 (94.1) | 527 (94.4) | 0.697 |
| Race black | 1,823 (3.8) | 1,802 (3.8) | 21 (3.8) | 0.930 |
| Body mass index, kg/m2 | 26.9 (23.6–31.3) | 26.9 (23.6–31.3) | 26.5 (23.2–30.6) | 0.051 |
| Body surface area, m2 | 1.84 (1.67–2.01) | 1.84 (1.68–2.01) | 1.74 (1.60–1.92) | <0.0001 |
| STS mortality risk score, % | 6.58 (4.33–10.3) | 6.58 (4.33–10.3) | 6.66 (4.42–9.89) | 0.718 |
| NYHA functional class III/IV | 38,640 (81.3) | 38,195 (81.3) | 445 (79.8) | 0.214 |
| GFR, ml/min/1.73 m2 | 60 (44–76) | 60 (44–76) | 59 (44–75) | 0.783 |
| ESRD on dialysis | 1,968 (4.1) | 1,959 (4.2) | 9 (1.6) | 0.003 |
| Diabetes mellitus | 17,790 (37.4) | 17,628 (37.5) | 162 (29.0) | <0.0001 |
| Peripheral artery disease | 14,669 (30.9) | 14,509 (30.9) | 160 (28.7) | 0.254 |
| Carotid artery stenosis | 8,944 (18.8) | 8,867 (18.8) | 77 (13.8) | 0.002 |
| Prior stroke | 9,001 (18.9) | 8,897 (18.9) | 104 (18.6) | 0.858 |
| Prior myocardial infarction | 11,882 (25.0) | 11,783 (25.1) | 99 (17.7) | <0.0001 |
| Prior permanent pacemaker | 7,671 (16.1) | 7,588 (16.1) | 83 (14.9) | 0.414 |
| Prior ICD | 2,100 (4.4) | 2,083 (4.4) | 17 (3.1) | 0.112 |
| Prior PCI | 16,791 (35.3) | 16,634 (35.4) | 157 (28.1) | 0.0004 |
| Prior CABG surgery | 13,909 (29.3) | 13,840 (29.5) | 69 (12.4) | <0.0001 |
| Prior aortic valve intervention | 7,000 (14.7) | 6,939 (14.7) | 61 (10.9) | 0.011 |
| Prior nonaortic valve intervention | 1,217 (2.6) | 1,214 (2.6) | 3 (0.5) | 0.002 |
| Previous cardiac surgeries | ||||
| None | 32,348 (68.0) | 31,880 (67.9) | 468 (83.9) | <0.0001 |
| 1 | 12,555 (26.4) | 12,493 (26.6) | 62 (11.1) | |
| 2 or more | 1,969 (4.1) | 1,950 (4.2) | 19 (3.4) | |
| Current smoker | 2,598 (5.5) | 2,574 (5.5) | 24 (4.3) | 0.222 |
| Severe chronic lung disease | 6,510 (13.7) | 6,442 (13.7) | 68 (12.2) | 0.305 |
| Home oxygen | 5,968 (12.6) | 5,908 (12.6) | 60 (10.8) | 0.190 |
| Hostile chest | 3,799 (8.0) | 3,771 (8.0) | 28 (5.0) | 0.009 |
| Porcelain aorta | 2,963 (6.2) | 2,930 (6.2) | 33 (5.9) | 0.748 |
| Atrial fibrillation or flutter | 19,476 (41.0) | 19,263 (41.0) | 213 (38.2) | 0.177 |
| Conduction defect | 16,056 (33.8) | 15,875 (33.8) | 181 (32.4) | 0.521 |
| KCCQ score | 38.5 (22.9–57.6) | 38.5 (22.9–57.3) | 40.6 (25.0–59.9) | 0.082 |
| Hemoglobin, g/dl | 11.8 (10.5–13.0) | 11.8 (10.5–13.0) | 12.0 (10.6–13.2) | 0.013 |
| Platelet count, ×109/l | 191 (152–237) | 191 (152–237) | 196 (152–239) | 0.463 |
| Angiography | ||||
| Triple-vessel coronary disease | 12,611 (26.5) | 12,536 (26.7) | 75 (13.4) | <0.0001 |
| Left main coronary artery disease >50% | 4,893 (10.3) | 4,862 (10.4) | 31 (5.6) | 0.0002 |
| Proximal LAD disease >70% | 9,287 (19.5) | 9,220 (19.6) | 67 (12.0) | <0.0001 |
| Echocardiography | ||||
| LV ejection fraction, % | 57 (45–63) | 57 (45–63) | 60 (53–65) | <0.0001 |
| Degenerative aortic stenosis | 44,845 (94.3) | 44,323 (94.3) | 522 (93.6) | 0.342 |
| Tricuspid aortic valve | 42,846 (90.1) | 42,339 (90.1) | 507 (90.9) | 0.564 |
| Moderate or severe AR | 9,595 (20.2) | 9,489 (20.2) | 106 (19.0) | 0.466 |
| Moderate or severe MR | 13,781 (29.0) | 13,626 (29.0) | 155 (27.8) | 0.518 |
| Moderate or severe TR | 11,368 (23.9) | 11,238 (23.3) | 130 (23.3) | 0.693 |
| Overall (N = 47,546) | No Bailout (n = 46,988) | Bailout (n = 558) | p Value | |
|---|---|---|---|---|
| Heart team reason for TAVR | ||||
| Patient preference | 61 (0.1) | 59 (0.1) | 2 (0.4) | 0.0011 |
| Inoperable: technical | 2,605 (5.5) | 2,574 (5.5) | 31 (5.6) | |
| Prohibitive risk: comorbidities | 9,121 (19.2) | 9,011 (19.2) | 110 (19.7) | |
| Prohibitive risk: debilitated | 2,755 (5.8) | 2,718 (5.8) | 37 (6.6) | |
| Inoperable: extreme risk | 7,803 (16.4) | 7,739 (16.5) | 64 (11.5) | |
| High risk (30-day mortality >8%) | 22,744 (47.8) | 22,464 (47.8) | 280 (50.2) | |
| Intermediate risk (30-day mortality 4%–8%) | 1,518 (3.2) | 1,500 (3.2) | 18 (3.2) | |
| Low risk (30-day mortality <4%) | 294 (0.6) | 292 (0.6) | 2 (0.4) | |
| Procedure location | ||||
| Cardiac catheterization laboratory | 5,182 (10.90) | 5,121 (10.90) | 61 (10.93) | 0.7680 |
| Hybrid suite/cardiac catheterization laboratory | 12,501 (26.29) | 12,362 (26.31) | 139 (24.91) | |
| Hybrid operating room suite | 29,664 (62.39) | 29,309 (62.38) | 355 (63.62) | |
| Other | 165 (0.35) | 162 (0.34) | 3 (0.54) | |
| Type of anesthesia | ||||
| Moderate sedation | 3,542 (7.5) | 3,530 (7.5) | 12 (2.2) | <0.0001 |
| General anesthesia | 43,686 (91.9) | 43,146 (91.8) | 540 (96.8) | |
| Epidural | 18 (0.04) | 18 (0.04) | 0 (0) | |
| Combination | 195 (0.4) | 189 (0.4) | 6 (1.1) | |
| Inotropes | 20,697 (43.5) | 20,276 (43.2) | 421 (75.5) | <0.0001 |
| Acuity | ||||
| Elective | 41,714 (87.7) | 41,252 (87.8) | 462 (82.8) | <0.0001 |
| Urgent | 3,961 (8.3) | 3,912 (8.3) | 49 (8.8) | |
| Emergency | 1,642 (3.5) | 1,603 (3.4) | 39 (7.0) | |
| Salvage | 229 (0.5) | 221 (0.5) | 8 (1.4) | |
| Access site | ||||
| Femoral | 34,104 (71.7) | 33,767 (71.9) | 337 (60.4) | <0.0001 |
| Nonfemoral | 13,183 (27.7) | 12,979 (27.6) | 204 (36.6) | |
| Transapical | 9,046 (19.0) | 8,936 (19.0) | 110 (19.7) | |
| Transaortic | 2,850 (6.0) | 2,773 (5.9) | 77 (13.8) | |
| Axillary | 122 (0.3) | 121 (0.3) | 1 (0.2) | |
| Subclavian | 563 (1.2) | 555 (1.2) | 8 (1.4) | |
| Transiliac | 189 (0.4) | 187 (0.4) | 2 (0.4) | |
| Transseptal | 15 (0.03) | 15 (0.03) | 0 (0) | |
| Transcarotid | 68 (0.14) | 68 (0.14) | 0 (0) | |
| Other | 330 (0.7) | 324 (0.7) | 6 (1.1) | |
| Access method | ||||
| Percutaneous | 21,568 (45.4) | 21,372 (45.5) | 196 (35.1) | <0.0001 |
| Cutdown | 13,945 (29.3) | 13,781 (29.3) | 164 (29.4) | |
| Mini-thoracotomy | 9,353 (19.7) | 9,236 (19.7) | 117 (21.0) | |
| Mini-sternotomy | 2,107 (4.4) | 2,055 (4.4) | 52 (9.3) | |
| Other | 277 (0.6) | 265 (0.6) | 12 (2.2) | |
| Valve type | ||||
| Self-expanding | 10,682 (22.5) | 10,603 (22.6) | 79 (14.2) | <0.0001 |
| Balloon expandable | 36,044 (75.8) | 35,639 (75.9) | 405 (72.6) | |
| Missing | 820 (1.7) | 746 (1.6) | 74 (13.3) | |
| Valve-in-valve | 1,449 (3.1) | 1,437 (3.1) | 12 (2.2) | 0.215 |
| Valve size, mm | ||||
| 23 | 14,498 (30.5) | 14,592 (30.4) | 206 (36.9) | <0.0001 |
| 26 | 19,891 (41.8) | 19,706 (41.9) | 185 (33.2) | |
| 29 | 8,822 (18.6) | 8,749 (18.6) | 73 (13.1) | |
| 31 | 3,421 (7.2) | 3,411 (7.3) | 10 (1.8) | |
| Contrast volume, ml | 106 (70–158) | 106 (70–158) | 110 (69–166) | 0.326 |
| Fluoroscopy time, min | 18 (13–26) | 18 (13–25) | 21 (16–31) | <0.0001 |
| Postop AV gradient, mm Hg | 9 (6–12) | 9 (6–12) | 10 (6–13) | 0.025 |
| Institution total TAVR volume | 189 (110–303) | 189 (110–303) | 176 (109–278) | 0.01 |
In-hospital clinical outcomes
The all-cause in-hospital mortality was significantly higher in patients who required surgical bailout (49.64% vs. 3.52%; p < 0.0001). With the exception of new permanent pacemaker implantation, the incidence of all nonfatal complications was significantly higher in the surgical bailout cohort (Table 3). Among the reasons for surgical bailout, ventricular rupture had the highest in-hospital all-cause mortality (66.7%) (Figure 1B). After adjusting for need for cardiopulmonary bypass, ventricular rupture was associated with increased risk for in-hospital death compared with valve dislodgement to the left ventricle (OR: 4.08; 95% CI: 2.28 to 7.29; p < 0.0001), valve dislodgment to the aorta (OR: 3.35; 95% CI: 1.18 to 9.49; p = 0.0227), aortic dissection (OR: 2.21; 95% CI: 1.05 to 4.63; p = 0.0367), and other reasons (OR: 2.18; 95% CI: 1.27 to 3.73; p = 0.0043). Annular rupture had an increased all-cause mortality only when compared with valve dislodgement to the left ventricle (OR: 2.42; 95% CI: 1.34 to 4.38; p = 0.0035). Additionally, patients who required cardiopulmonary bypass had significantly increased all-cause mortality compared with those who did not require it (OR: 2.02; 95% CI: 1.36 to 2.98; p = 0.0005) (Figure 1B).
| Overall (N = 47,546) | No Bailout (n = 46,988) | Bailout (n = 558) | p Value | |
|---|---|---|---|---|
| All-cause mortality | 1,930 (4.06) | 1,653 (3.52) | 277 (49.64) | <0.0001 |
| Myocardial infarction | 204 (0.43) | 195 (0.41) | 9 (1.61) | <0.0001 |
| Stroke | 989 (2.08) | 957 (2.04) | 32 (5.73) | <0.0001 |
| MACE | 2,898 (6.10) | 2,603 (5.54) | 295 (52.87) | <0.0001 |
| Major vascular complication | 3,374 (7.10) | 3,258 (6.93) | 116 (20.79) | <0.0001 |
| Major bleeding | 2,794 (5.88) | 2,716 (5.78) | 78 (13.98) | <0.0001 |
| Any bleeding | 4,518 (9.52) | 4,332 (9.23) | 186 (33.33) | <0.0001 |
| Access-site bleeding | 875 (1.84) | 850 (1.81) | 25 (4.48) | <0.0001 |
| New permanent pacemaker implantation | 4,025 (8.47) | 3,997 (8.51) | 28 (5.02) | 0.003 |
| Atrial fibrillation | 2,578 (5.42) | 2,516 (5.35) | 62 (11.11) | <0.0001 |
| New need for dialysis | 698 (1.47) | 663 (1.41) | 35 (6.27) | <0.0001 |
| Aortic valve reintervention | 149 (0.31) | 140 (0.30) | 9 (1.61) | <0.0001 |
| MACE plus any bleeding | 5,316 (11.2) | 4,981 (10.6) | 335 (60.0) | <0.0001 |
| NACE | 6,734 (14.2) | 6,306 (13.42) | 428 (76.7) | <0.0001 |
| Length of hospital stay, days | 5 (3–9) | 5 (3–9) | 9 (3–18) | <0.0001 |
30-day and 1-year clinical outcomes
CMS-linked data, used for estimation for 30-day and 1-year outcomes, were available for 68.9% of the overall population, with no differences between patients who required surgical bailout and those who did not. Baseline and procedural characteristics of those with and without CMS-linked data are available in Online Table 2. The median follow-up time was 288 days (interquartile range: 106 to 546 days). At 30 days, all-cause mortality (50.0% vs. 4.98%; p < 0.0001) and the incidence of all nonfatal complications, including MACE (54.64% vs. 7.36%; p < 0.0001) was significantly higher in the surgical bailout cohort (Table 4). Similarly, the 1-year incidence of all adverse clinical outcomes was significantly increased in the surgical bailout group, including MACE (63.92% vs. 20.29%; p < 0.0001) and all-cause mortality (59.79% vs. 17.06%; p < 0.0001). The cumulative incidence of MACE and all-cause mortality was significantly higher in patients who underwent surgical bailout compared with those who did not (Central Illustration). Furthermore, both the all-cause mortality and MACE were significantly increased for patients requiring surgical bailout between 30 days and 1 year in the landmark analysis (Online Figure 1).
| Overall (N = 32,758) | No Bailout (n = 32,370) | Bailout (n = 388) | p Value | |
|---|---|---|---|---|
| 30-day outcomes∗ | ||||
| All-cause mortality | 1,807 (5.52) | 1,613 (4.98) | 194 (50.0) | <0.0001 |
| Myocardial infarction | 215 (0.66) | 207 (0.64) | 8 (2.06) | <0.0001 |
| Stroke | 777 (2.37) | 751 (2.32) | 26 (6.70) | <0.0001 |
| MACE | 2,594 (7.92) | 2,382 (7.36) | 212 (54.64) | <0.0001 |
| Any bleeding | 4,104 (12.53) | 3,966 (12.25) | 138 (35.57) | <0.0001 |
| Access-site bleeding | 1,306 (3.99) | 1,280 (3.95) | 26 (6.70) | <0.0001 |
| Aortic valve reintervention | 152 (0.46) | 146 (0.45) | 6 (1.55) | <0.0001 |
| 1-yr outcomes∗ | ||||
| All-cause mortality | 5,755 (17.57) | 5,523 (17.06) | 232 (59.79) | <0.0001 |
| Myocardial infarction | 530 (1.62) | 520 (1.61) | 10 (2.58) | <0.0001 |
| Stroke | 1,140 (3.48) | 1,109 (3.43) | 31 (7.99) | <0.0001 |
| MACE | 6,817 (20.81) | 6,569 (20.29) | 248 (63.92) | <0.0001 |
| Any bleeding | 6,818 (20.81) | 6,667 (20.60) | 151 (38.92) | <0.0001 |
| Access-site bleeding | 1,309 (4.0) | 1,283 (3.96) | 26 (6.70) | <0.0001 |
| Aortic valve re-intervention | 360 (1.10) | 354 (1.09) | 6 (1.55) | <0.0001 |
Cumulative Incidence of All-Cause Mortality and Major Adverse Cardiovascular Events
(A) Cumulative incidence of 1-year all-cause mortality in patients requiring surgical bailout after transcatheter aortic valve replacement (TAVR) (blue line) and those without surgical bailout (orange line). (B) Cumulative incidence of 1-year major adverse cardiovascular event(s) (MACE) in patients requiring surgical bailout after TAVR (blue line) and those without surgical bailout (orange line). HR = hazard ratio.
Surgical bailout incidence and outcomes over time
The incidence of surgical bailout decreased significantly in the last time period (first tertile 1.25%, second tertile 1.43%, third tertile 1.04%; p = 0.0088) (Figure 2A). Pairwise comparisons among the time periods revealed a statistically significant difference in the incidence of surgical bailout only between the second and third time periods (1.43% vs. 1.04%; Bonferroni p = 0.0012). The use of nonfemoral access over time (first tertile 28.54%, second tertile 49.67%, third tertile 17.24%; p < 0.0001), but not the number of new TAVR sites (first tertile 77.46%, second tertile 28.93%, third tertile 18.67%; p < 0.001), paralleled the incidence of surgical bailout over time (Figure 2A).
Trends in the Incidence of Surgical Bailout and All-Cause Mortality by Time Period
(A) Trends in the incidence of surgical bailout, new transcatheter aortic valve replacement (TAVR) sites, and nonfemoral access by time period. The use of nonfemoral access over time, but not the number of new TAVR sites, paralleled the incidence of surgical bailout over time. (B) Cumulative incidence of all-cause mortality by time period. There was a trend toward a reduction in the cumulative incidence of all-cause mortality over time.
Reasons for surgical bailout and clinical outcomes by time period are presented in Table 5. There were no statistically significant differences among reasons for surgical bailout over time. In-hospital and 30-day clinical outcomes, including MACE and all-cause mortality, were similar among the 3 time periods, with the exception of MI, which was more frequent during the first time period. However, the 1-year incidence of MACE (p = 0.0044), stroke (p = 0.0265), MI (p < 0.0001), and all-cause mortality (p = 0.0025) significantly improved over time. There was a trend toward an improvement in the cumulative incidence of all-cause mortality over time (Figure 2B).
| Overall (N = 558) | First Time Period (n = 74) | Second Time Period (n = 191) | Third Time Period (n = 292) | p Value | |
|---|---|---|---|---|---|
| Surgical bailout | 558 (1.17) | 74 (1.25) | 191 (1.43) | 293 (1.04) | 0.0024 |
| Reason for surgical bailout | |||||
| Valve dislodged to aorta | 16 (2.87) | 2 (2.70) | 6 (3.14) | 8 (2.73) | 0.086 |
| Valve dislodged to left ventricle | 108 (19.35) | 13 (17.57) | 49 (25.65) | 46 (15.70) | |
| Ventricular rupture | 111 (19.89) | 11 (14.86) | 32 (16.75) | 68 (23.21) | |
| Annular rupture | 79 (14.16) | 9 (12.16) | 20 (10.47) | 50 (17.06) | |
| Aortic dissection | 46 (8.24) | 11 (14.86) | 9 (4.71) | 26 (8.87) | |
| Coronary occlusion | 34 (6.09) | 5 (6.76) | 15 (7.85) | 14 (4.78) | |
| Other | 164 (29.39) | 23 (31.08) | 60 (31.41) | 81 (27.65) | |
| In-hospital outcomes | |||||
| All-cause mortality | 277 (49.64) | 42 (56.76) | 97 (50.79) | 138 (47.10) | 0.1452 |
| Myocardial infarction | 9 (1.61) | 5 (6.76) | 3 (1.57) | 1 (0.34) | 0.0012 |
| Stroke | 32 (5.73) | 4 (5.41) | 11 (5.76) | 17 (5.80) | 0.9202 |
| MACE | 295 (52.87) | 45 (60.81) | 103 (53.93) | 147 (50.17) | 0.1164 |
| Any bleeding | 186 (33.33) | 25 (33.78) | 64 (33.51) | 97 (33.11) | 0.8990 |
| Access-site bleeding | 25 (4.48) | 3 (4.05) | 5 (2.62) | 17 (5.80) | 0.1681 |
| Aortic valve reintervention | 9 (1.61) | 0 (0.00) | 5 (2.62) | 4 (1.37) | 0.9677 |
| CMS-linked | 388 (69.53) | 52 (70.27) | 135 (70.68) | 201 (68.60) | 0.6423 |
| 30-day outcomes | |||||
| All-cause mortality | 194 (50.0) | 32 (61.54) | 67 (49.63) | 95 (47.26) | 0.1368 |
| Myocardial infarction | 8 (2.06) | 5 (9.62) | 3 (2.22) | 0 (0.00) | 0.0007 |
| Stroke | 26 (6.70) | 4 (7.69) | 9 (6.67) | 13 (6.47) | 0.4920 |
| MACE | 212 (54.64) | 34 (65.38) | 73 (54.07) | 105 (52.24) | 0.1798 |
| Any bleeding | 138 (35.57) | 19 (36.54) | 49 (36.30) | 70 (34.83) | 0.8160 |
| Access-site bleeding | 26 (6.70) | 4 (7.69) | 6 (4.44) | 16 (7.96) | 0.4570 |
| Aortic valve reintervention | 6 (1.55) | 0 (0.00) | 3 (2.22) | 3 (1.49) | 0.4485 |
| 1-yr outcomes | |||||
| All-cause mortality | 232 (59.79) | 39 (75.00) | 86 (63.70) | 107 (53.23) | 0.0025 |
| Myocardial infarction | 10 (2.58) | 5 (9.62) | 5 (3.70) | 0 (0.00) | <0.0001 |
| Stroke | 31 (7.99) | 6 (11.54) | 10 (7.41) | 15 (7.46) | 0.0265 |
| MACE | 248 (63.92) | 40 (76.92) | 92 (68.15) | 116 (57.71) | 0.0044 |
| Any bleeding | 151 (38.92) | 22 (42.31) | 54 (40.00) | 75 (37.31) | 0.3789 |
| Access-site bleeding | 26 (6.70) | 4 (7.69) | 6 (4.44) | 16 (7.96) | 0.0338 |
| Aortic valve reintervention | 6 (1.55) | 0 (0.00) | 3 (2.22) | 3 (1.49) | 0.0255 |
Predictors of the need for and survival after surgical bailout
A logistic regression model was built to identify covariates associated with need for surgical bailout using data from the second and third time periods and excluding patients for whom valve system type was missing. Baseline and procedural characteristics of those with and without valve system type information are included in Online Table 3. Female sex, hemoglobin, left ventricular ejection fraction, cardiogenic shock or use of a left ventricular assist device, salvage procedures, and nonfemoral access site were found to be independent predictors of the need for surgical bailout during TAVR (Figure 3A). When only patients undergoing transfemoral access were included in the logistic regression model, female sex, left ventricular ejection fraction, cardiogenic shock or use of a left ventricular assist device, emergent or salvage procedures, and balloon-expandable valve were found to be independent predictors of need for surgical bailout (Figure 3B).
Predictors of Need for Surgical Bailout
(A) Predictors of need for surgical bailout in the overall cohort. (B) Predictors of need for surgical bailout in the transfemoral cohort. BSA = body surface area; CABG = coronary artery bypass graft surgery; CI = confidence interval; GFR = glomerular filtration rate; ICD = implantable cardioverter-defibrillator; LAD = left anterior descending coronary artery; LVEF = left ventricular ejection fraction; MI = myocardial infarction; NYHA = New York Heart Association; OR = odds ratio; PCI = percutaneous coronary intervention; RVSP = right ventricular systolic pressure; TAVR = transcatheter aortic valve replacement.
A separate logistic regression model was constructed to identify covariates associated survival after surgical bailout during TAVR. Increasing body surface area was found to be the only predictor of survival after surgical bailout in the overall cohort and in the transfemoral access cohort (Figure 4). Predictors of the need for individual reasons for surgical bailout are included in Online Table 4.
Predictors of Survival After Surgical Bailout
(A) Predictors of survival after surgical bailout in the overall cohort. (B) Predictors of survival after surgical bailout in the transfemoral cohort. BSA = body surface area; CI = confidence interval; ICD = implantable cardioverter-defibrillator; NYHA = New York Heart Association; OR = odds ratio; PAD = peripheral artery disease.
Discussion
In the present analysis we evaluated the incidence, clinical outcomes, and predictors of surgical bailout during TAVR among all commercial cases in the United States. The main findings of the present study were as follows: 1) the need for emergent conversion to open heart surgery, or surgical bailout, during TAVR is about 1%; 2) the incidence of surgical bailout has significantly decreased over time, mirroring a decrease in the use of nonfemoral access during the study period; 3) the clinical outcomes of patients requiring surgical bailout during TAVR are poor, with 30-day and 1-year mortality rates of 50% and 59.8%, respectively, and >10-fold higher than in those patients not requiring emergent conversion to open heart surgery; and 4) independent predictors of need for surgical bailout included female sex, increasing hemoglobin, increasing left ventricular ejection fraction, nonelective cases, and nonfemoral access.
The overall incidence of surgical bailout from the TVT Registry during the study period was 1.17%. This was at the lower end of the reported incidence, which ranges from 1% to 6% (4,8–14). An important finding of the present analysis is that we were able to demonstrate a statistically significant, albeit numerically small, decrease in the incidence of surgical bailout over time. This is in line with the findings of other studies (10,14). The decreasing rate of surgical bailout is likely due to a combination of better patient selection, improved device technology, and increasing operator experience. Smaller delivery systems that allow more patients to be treated using transfemoral access, repositionable valves, pre-shaped working left ventricular guidewires, and pre-procedural computed tomographic annular sizing may have played a role by potentially decreasing transapical-related bleeding, valve embolization, left ventricular rupture, and annular rupture. This was in part suggested by the parallel decreases in the use of nonfemoral access and the incidence of surgical bailout during our study period.
In our cohort, patients who required surgical bailout were older, were more likely to be female, had a smaller median body surface area, and higher median left ventricular ejection fraction. Additionally, certain procedural characteristics were more frequent in those who underwent surgical bailout, including the use of nonfemoral access. Higher rates of surgical bailout in patients undergoing alternative-access TAVR have being previously reported (4,8,13). Although the actual difference was small, the total institutional TAVR volume was statistically significantly lower in the bailout group. Published data have already demonstrated the inverse relationship between TAVR volume and mortality and nonfatal adverse outcomes (15,16).
In-hospital and 30-day all-cause mortality of the surgical bailout group in the present study was about 50%, which is >10-fold higher than the mortality rates of patients who did not require bailout. Our findings confirmed the dismal clinical outcomes of patients requiring surgical bailout during TAVR reported by prior studies (4,8,10–13). Because of the large sample size of our study, we were able to evaluate the differences in in-hospital all-cause mortality among the reasons for bailout. Ventricular rupture, the most lethal reason, was associated with a 2- to 4-fold increase in in-hospital mortality compared with most of the other reasons for bailout. This finding is surprising, as annular rupture is typically regarded as the most fatal complication of TAVR, and may be due to the inability of this analysis to capture patients with annular rupture who did not survive to be able to undergo surgical bailout. Additionally, patients requiring the institution of cardiopulmonary bypass had a 2-fold increase in mortality compared with those not requiring bypass. Furthermore, the incidence of MACE and most of the other nonfatal adverse outcomes was significantly higher in the bailout group. Although most of the morbidity and mortality after surgical bailout occurs in the early post-operative period, the 30-day landmark analysis demonstrated that adverse outcomes continued to increase up to 1 year after TAVR. Despite the observed decrease in the incidence of surgical bailout over time, the in-hospital and 30-day outcomes of patients who required bailout were similar across the 3 time periods. Interestingly, the incidence of MACE, stroke, MI, and all-cause mortality 1 year after TAVR was better in those patients treated during the latest time period. This finding suggests better patient selection over time, reflected in improved 1-year outcomes, which are likely determined by baseline patient characteristics and their risk profile.
Prior studies have not provided any insights into possible patient and procedural characteristics associated with increased likelihood for the need for surgical bailout. This is likely due to limited sample sizes, as well as the small numbers of patients undergoing surgical bailout, in prior studies (8–13). In fact, the largest published dataset comes from a meta-analysis of 9,251 patients from 46 studies, of whom 102 required surgical bailout (4). Female sex, increasing hemoglobin, increasing left ventricular ejection fraction, emergent or salvage procedures, and nonfemoral access were found to be independent predictors of need for surgical bailout during TAVR in our study. Female sex has been previously shown to be associated with an increased risk for nonfatal complications after TAVR (17–19), which may be explained by anatomic differences, including smaller access vessels, smaller annular size, left ventricular outflow track, and left ventricular cavity size. Nonfemoral access strongly predicted the risk for surgical bailout in our study, and in fact, the decrease in the incidence of surgical bailout mirrored the decrease in the use of nonfemoral access over the study period. Data from observational studies have shown that alternative access is associated with increased short- and long-term morbidity and mortality compared with the transfemoral approach (20,21). Although not achieving statistical significance for the whole cohort, patients who underwent transfemoral TAVR with balloon-expandable valves had a significant 35% increase in the risk for needing surgical bailout. Balloon-expandable valves have been previously associated with an increased risk for annular rupture (11,22,23). Increasing body surface area was found to be the only independent predictor of survival after surgical bailout. Although its mechanism is not well understood, the protective effect of obesity in patients undergoing cardiac surgery has been previously reported (24). Several anatomic and procedural characteristics not included in the TVT Registry may have been associated with increased risk for surgical bailout. However, to our knowledge, the present study is the first to provide insights into possible predictors of need for and survival after surgical bailout during TAVR.
The need for on-site surgical backup for TAVR is currently a matter of debate. Data from a German TAVR registry have already shown similar outcomes, including the need for surgical bailout and mortality, in cases done at hospitals with or without on-site cardiac surgery (25). Although our findings cannot be used to change the current recommendation for on-site cardiac surgery backup, they may be used for refinement of the procedural planning and patient selection for TAVR. For example, on the basis of the novel predictors found in our study, elective transfemoral TAVR cases without an increased risk for additional complications (i.e., low coronary height, severe calcification, narrow left ventricular outflow tract, etc.), could be potentially done outside the operating room with low risk for surgical bailout. However, those patients at risk for surgical bailout should continue to be treated in the operating room given that surgical bailout during TAVR is not futile, as demonstrated in the present study.
In summary the key findings of the present analysis are as follows: 1) the incidence of surgical bailout during TAVR is low and has decreased over time; 2) the clinical outcomes of patients requiring surgical bailout are poor; and 3) there are multiple predictors of surgical bailout, and the most significant risks include nontransfemoral access and urgent or emergent procedures. Indeed, the existence of these independent predictors has implications for clinical practice and underscores that thoracic alternative access should be used as a last resource and that conscientious patient selection is important in nonelective cases. Although other predictors were also identified, including sex, higher hemoglobin, and high left ventricular ejection fraction, their relationships with the need for bailout are less understood, and their clinical implications are unclear at this time.
Study limitations
The endpoints captured in the TVT Registry are reported by each participating site and internally validated by electronic data checks but not adjudicated by a central committee. Some of the definitions in the TVT Registry are broad (i.e., ventricular rupture), not allowing more detailed analyses (i.e., right ventricular rupture from temporary pacing wire vs. left ventricular rupture). Data on the timing of surgical bailout with respect to TAVR and the specific surgical procedures (i.e., coronary artery bypass surgery vs. ventricular rupture repair vs. surgical aortic valve replacement, etc.) performed is not known. Furthermore, the inability to capture patients with severe intraoperative complications who did not survive to be able to undergo surgical bailout, or were deemed inoperable, also limited our analysis. Transcatheter technology and techniques have rapidly evolved since 2015, which may have further decreased the incidence of surgical bailout in current practice. The implanted valve type was not reported in about 13% of the patients in the surgical bailout group, either because that information was missing or no valve was ultimately implanted. CMS claims data were used for 30-day and 1-year outcomes, thereby limiting the granularity of outcomes at these time points.
Conclusions
The need for emergent conversion to open heart surgery, or surgical bailout, during TAVR among commercial cases in the United States is about 1% and is associated with an increased incidence of adverse clinical outcomes, including 30-day and 1-year mortality rates of about 50% and 60%, respectively, which is >10-fold higher than in patients not requiring emergent conversion to open heart surgery. The incidence of surgical bailout has statistically decreased over time, mirroring a decrease in the use of nonfemoral access during the study period. Several independent predictors of need for surgical bailout during TAVR were found, including female sex, nonelective cases, and nonfemoral access. The present study provides several novel findings that enhance our understanding of surgical bailout and may be used to refine patient selection and/or to identify potential candidates for TAVR in a nonsurgical environment, thereby improving the efficiency and safety of TAVR.
WHAT IS KNOWN? Surgical bailout during TAVR is associated with high morbidity and mortality.
WHAT IS NEW? The rate of surgical bailout among TAVR commercial cases in the United States has decreased over time and is about 1%, carrying a 30-day and 1-year all-cause mortality of >50%. Several predictors of need for surgical bailout were identified.
WHAT IS NEXT? Our findings may be used for refinement of procedural planning and patient selection for TAVR.
1. : "Transcatheter aortic-valve replacement with a self-expanding prosthesis". N Engl J Med 2014; 370: 1790.
2. : "The learning curve associated with transapical aortic valve implantation". Ann Cardiothorac Surg 2012; 1: 165.
3. : "Transcatheter versus surgical aortic-valve replacement in high-risk patients". N Engl J Med 2011; 364: 2187.
4. : "Emergent cardiac surgery during transcatheter aortic valve implantation (TAVI): a weighted meta-analysis of 9,251 patients from 46 studies". EuroIntervention 2013; 8: 1072.
5. : "Transcatheter aortic valve implantation: lessons from the learning curve of the first 270 high-risk patients". Catheter Cardiovasc Interv 2011; 78: 977.
6. : "Outcomes after transfemoral transcatheter aortic valve replacement a comparison of the randomized PARTNER (Placement of Aortic Transcatheter Valves) trial with the NCRA (Nonrandomized Continued Access) registry". J Am Coll Cardiol Interv 2014; 7: 1245.
7. : "Outcomes following transcatheter aortic valve replacement in the United States". JAMA 2013; 310: 2069.
8. : "The role of the heart team in complicated transcatheter aortic valve implantation: a 7-year single-centre experience". Eur J Cardiothorac Surg 2015; 47: 1090.
9. : "Single center TAVR experience with a focus on the prevention and management of catastrophic complications". Catheter Cardiovasc Interv 2014; 84: 834.
10. : "Severe intraprocedural complications after transcatheter aortic valve implantation: calling for a heart team approach". Eur J Cardiothorac Surg 2013; 44: 478.
11. : "Emergency cardiac surgery during transfemoral and transapical transcatheter aortic valve implantation: incidence, reasons, management, and outcomes of 411 patients from a single center". Catheter Cardiovasc Interv 2013; 82: E726.
12. : "Salvaging catastrophe in transcatheter aortic valve implantation: rehearsal, preassigned roles, and emergency preparedness". Can J Anesth 2015; 62: 918.
13. : "Outcome of patients after emergency conversion from transcatheter aortic valve implantation to surgery". EuroIntervention 2013; 9: 446.
14. : "The SOURCE registry: what is the learning curve in trans-apical aortic valve implantation?". Eur J Cardiothorac Surg 2011; 39: 853.
15. : "Association between hospital volume and 30-day readmissions following transcatheter aortic valve replacement". JAMA Cardiol 2017; 2: 732.
16. : "Procedural experience for transcatheter aortic valve replacement and relation to outcomes: the STS/ACC TVT Registry". J Am Coll Cardiol 2017; 70: 29.
17. : "Revisiting sex equality with transcatheter aortic valve replacement outcomes: a collaborative, patient-level meta-analysis of 11,310 patients". J Am Coll Cardiol 2015; 66: 221.
18. : "Sex-specific differences at presentation and outcomes among patients undergoing transcatheter aortic valve replacement: a cohort study". Ann Intern Med 2016; 164: 377.
19. : "Sex-based differences in outcomes with transcatheter aortic valve therapy: TVT Registry from 2011 to 2014". J Am Coll Cardiol 2016; 68: 2733.
20. : "Longterm results of transapical versus transfemoral TAVI in a real world population of 1000 patients with severe symptomatic aortic stenosis". Circ Cardiovasc Interv 2015; 8:
e000761 .21. : "A metaanalysis of mortality and major adverse cardiovascular and cerebrovascular events in patients undergoing transfemoral versus transapical transcatheter aortic valve implantation using Edwards valve for severe aortic stenosis". Am J Cardiol 2014; 114: 1882.
22. : "Annular rupture during transcatheter aortic valve replacement: classification, pathophysiology, diagnostics, treatment approaches, and prevention". J Am Coll Cardiol Intv 2015; 8: 1.
23. : "Anatomical and procedural features associated with aortic root rupture during balloon-expandable transcatheter aortic valve replacement". Circulation 2013; 128: 244.
24. : "The obesity paradox in elderly obese patients undergoing coronary artery bypass surgery". Interact Cardiovasc Thorac Surg 2011; 13: 124.
25. : "Outcomes of transfemoral transcatheter aortic valve implantation at hospitals with and without on-site cardiac surgery department: insights from the prospective German aortic valve replacement quality assurance registry (AQUA) in 17919 patients". Eur Heart J 2016; 37: 2240.
Abbreviations and Acronyms
| CI | confidence interval |
| CMS | Centers for Medicare and Medicaid Services |
| MACE | major adverse cardiovascular event(s) |
| MI | myocardial infarction |
| OR | odds ratio |
| TAVR | transcatheter aortic valve replacement |
Footnotes
Dr. Pineda has received consulting fees from Pfizer and TZ Medical. Dr. Harrison has received institutional grants from Abbott Vascular, Medtronic, Edwards Lifesciences, and Boston Scientific. Dr. Kleiman has received fees from Medtronic for providing educational services. Dr. Rihal has received institutional grants from Medtronic and Edwards Lifesciences. Dr. Kodali received research grants from Claret Medical, Edwards Lifesciences, Medtronic, Abbott Vascular, Boston Scientific, Admedus, and Meril Life Sciences; is on the scientific advisory board for Thubrikar Aortic Valve Inc., Dura Biotech, and Biotrace Medical; received honoraria from Claret Medical, Admedus, Meril Life Sciences, and Abbott Vascular; and received equity from Thubrikar Aortic Valve Inc., Dura Biotech, and Biotrace Medical. Dr. Kirtane has received institutional grants from Medtronic, Boston Scientific, Abbott Vascular, Abiomed, Cardiovascular Systems, Inc., CathWorks, Siemens, Philips, and ReCor Medical. Dr. Leon has received institutional research grants from Abbott Vascular, Boston Scientific, Edwards Lifesciences, and Medtronic. Dr. Sherwood has received consulting fees from Medtronic. Dr. Vemulapalli has received institutional grants from the American College of Cardiology and the Society of Thoracic Surgeons; has received personal grants from Abbott Vascular, the Patient-Centered Outcomes Research Institute, National Institutes of Health, and Boston Scientific; and has received consulting fees from Boston Scientific, Novella, Janssen, and Premiere. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
