Long-Term Outcomes of Veteran Patients After Transcatheter Aortic Valve Replacement

Background. Transcatheter aortic valve replacement (TAVR) has become a mainstay treatment for severe aortic stenosis and is increasingly used for veterans, producing excellent short-term outcomes. There is a paucity of long-term outcome data after TAVR in the veteran population. Methods. We examined consecutive patients who underwent TAVR at a single Veterans Affairs medical center through 2019. Baseline characteristics, echocardiographic and angiographic variables, and clinical outcomes were abstracted. All-cause mortality was the primary outcome of interest. Factors associated with all-cause mortality and cardiac-specific mortality, including the presence of significant non-revascularized coronary artery disease (CAD), were assessed with multivariable regression and competing-risk analyses. Results. The 189 consecutive patients enrolled (mean age, 76.6 ± 8.4 years) had a median Society of Thoracic Surgeons (STS) score of 6.0 (interquartile range [IQR], 4.0-8.5). After a maximum follow-up of 7.5 years, 71 (37.6%) deaths occurred, of which 76% had a cardiac cause. Median overall survival was 3.55 years (95% confidence interval [CI], 3.21-5.30); significant graded differences were observed across STS risk subgroups (P<.001). After multivariable adjustment, CAD was significantly associated with cardiac mortality (hazard ratio [HR], 2.6; 95% CI, 1.3-5.3) and all-cause mortality (HR, 2.2; 95% CI, 1.1-4.3). Other independent variables associated with all-cause mortality included age (P=.01), baseline creatinine (P<.01), and chronic obstructive pulmonary disease (P=.03). Baseline ejection fraction (P=.04), age (P<.01), creatinine (P=.02), and vascular disease (P=.04) were independently associated with cardiac-specific mortality. Conclusion. Long-term survival of veterans after TAVR is comparable to that of their non-veteran counterparts. Significant CAD, along with age and select comorbidities, was associated with poorer survival.

Key words: aortic valve replacement, coronary artery disease, outcomes

Transcatheter aortic valve replacement (TAVR) has become the most common therapeutic modality for severe aortic stenosis (AS). TAVR has been studied in several randomized clinical trials that have proven its benefits in high-, intermediate-, and low-risk patients.^1-9 The benefits observed in clinical TAVR trials were also corroborated at longer-term follow-up.^5,10-14 There is, however, a paucity of long-term, real-world outcome data after TAVR, especially in the veteran healthcare system.

Shortly after the United States Food and Drug Administration (FDA) approved TAVR, the Michael E. DeBakey Veterans Affairs Medical Center (MEDVAMC) performed the first TAVR in the Veterans Affairs (VA) healthcare system in 2011, and MEDVAMC continues to maintain one of the largest and most advanced TAVR programs in this unique and complex patient population.^15-17 Recently, Hall and colleagues reported excellent TAVR outcomes from 8 VA centers; these results compared favorably to benchmark outcome data outside the VA healthcare system.¹⁸ Their multicenter registry study of 959 consecutive veterans who underwent TAVR between 2012 and 2017 examined 30-day and 1-year survival only.¹⁸

With the established short-term benefits of TAVR in lower-risk patients,^4,6 it is necessary to assess the long-term valve durability and clinical outcomes of TAVR given its rapid expansion to younger patients. Moreover, given the unique cardiovascular profile of veteran patients,^19-21 it is imperative to assess predictors of long-term survival in this patient population. In particular, the impact of coronary artery disease (CAD) on outcomes among TAVR patients with severe AS remains controversial.²² Previous meta-analyses from observational retrospective studies have yielded conflicting results.^23,24 In the absence of data from randomized clinical trials, the decision of whether and when to perform percutaneous coronary intervention (PCI) in TAVR patients is typically made through shared decision-making by a multidisciplinary heart team.

We herein report the longest follow-up study of TAVR outcomes in a veteran population. We examined all-cause and cardiac-specific mortality, and the impact of concurrent significant CAD on mortality.

We included all consecutive patients who underwent TAVR at the MEDVAMC between December 2011 and May 2019. Most of the data input into the institutional registry were collected prospectively. Additional elements that were not part of the initial prospective input were added retrospectively, including follow-up outcomes and echocardiographic variables. None of the patients were part of prior trials. Our center’s patients are not included in the national Society of Thoracic Surgeons (STS)/American College of Cardiology (ACC) Transcatheter Valve Therapy (TVT) registry. The current observational study was approved by the local institutional review board.

Patient selection. All patients who underwent TAVR were evaluated by a multidisciplinary structural heart team comprised of a core group of cardiologists and cardiothoracic surgeons, 80% of whom have been part of our structural program since its inception. The decision on whether to proceed with surgical aortic valve replacement (SAVR) vs TAVR was based on the structural team’s clinical judgment after deliberations during our weekly heart team conference.

During the study period, 685 patients underwent either TAVR or SAVR. Patients with biventricular failure and marked cardiac cachexia were not offered SAVR/TAVR. Patients with severe chronic obstructive pulmonary disease (COPD), as defined by (1) a 1-second forced expiratory volume (FEV1) and diffusing capacity of carbon monoxide (DLCO) of <30% of predicted; and (2) an arterial partial pressure of oxygen (PaO₂) <50 mm Hg on room air or a partial pressure of carbon dioxide (PaCO₂) >55 mm Hg were usually not offered AVR. Patients with concomitant ascending, arch, and root aortic aneurysms were offered SAVR and the indicated operation for their aneurysm. Patients with bicuspid valves were offered SAVR unless they were deemed too high risk for open surgery. No patient was denied TAVR because of access issues.

Patients with significant CAD were considered for revascularization when deemed appropriate by the heart team. Patients with either significant left main coronary artery (LMCA) stenosis or significant proximal left anterior descending (LAD) artery stenosis were considered for SAVR + coronary artery bypass graft (CABG) surgery with left internal mammary arterial (LIMA) grafts to improve long-term outcomes. If the risk of open surgery was deemed prohibitive, TAVR with PCI was considered as an alternative option. The decision to decline PCI before TAVR was undertaken by consensus of the heart team when the patient’s coronary anatomy was not amenable to safe PCI.

Data extraction. Data abstracted included patients’ sociodemographic variables, baseline comorbidities, baseline angiographic and echocardiographic variables, and in-hospital and long-term clinical outcomes. The STS predicted risk of mortality (PROM) score was computed in all patients, who were divided into tertiles according to their risk scores: low (<3), intermediate (3-8), or high risk (>8).

Baseline coronary angiographic variables were prospectively adjudicated by interventional cardiologists. These include the anatomic location and percent stenosis of coronary lesions. Significant CAD was defined as unprotected LMCA or proximal-mid LAD diameter stenoses ≥50%. This definition was chosen a priori because long-term survival in patients with ischemic heart disease is likely to be influenced by the revascularization of significant stenoses of the LAD and/or LMCA, which place large segments of myocardium in jeopardy. A 50% cut-off was used because it is the threshold above which myocardial blood flow becomes compromised.²⁵ Patients who had previously undergone CABG with a patent LIMA graft to the LAD or who had PCI of the LMCA or LAD within 180 days before TAVR were considered “protected” and were not included in the significant-CAD group.

All patients underwent a transthoracic echocardiogram on postoperative day 1 or before discharge. Baseline and follow-up echocardiographic hemodynamic data were collected and compared. The definition of structural valve deterioration was adapted from the report by Dvir and colleagues²⁶ and included a mean aortic valve (AV) gradient of ≥ 20 mm Hg on the most recent follow-up echocardiogram or an increase in the mean AV gradient of ≥10 mm Hg from the immediate postoperative echocardiogram to the most recent one, or new onset of aortic insufficiency of at least moderate severity after TAVR.

The cause was determined by chart review according to Valve Academic Research Consortium (VARC)-2 definitions. If no clear non-cardiac cause was listed, then the death was presumed to be cardiac related. Postoperative complications were recorded according to VARC-2 criteria.²⁷

Statistical analyses. Between-group differences in the distribution of preoperative characteristics and postoperative outcomes were tested with Chi-squared analysis or the Fisher’s exact test for categorical variables, and with the Wilcoxon 2-sample test for continuous variables. Unadjusted overall survival was determined by Kaplan-Meier survival analysis. Intergroup comparisons were made by the log-rank test. Multivariable logistic regression was performed to identify independent risk factors for all-cause mortality among the following variables: CAD (unprotected LMCA or LAD ≥50% stenosis), history of myocardial infarction, previous CABG, year TAVR was performed, history of cancer, peripheral vascular disease, preoperative serum creatinine level, preoperative ejection fraction, age, congestive heart failure, COPD, cirrhosis, and diabetes. Fine-gray competing risk analysis was used to delineate cardiac-specific mortality. Two-sided P-values <.05 were considered significant. No adjustments were made for multiple comparisons. All statistical analyses were performed by using the SAS software, version 9.4 (SAS Institute).

Patient characteristics. A total of 189 consecutive patients (mean age, 76.6 ± 8.4 years; 98% men) with severe AS underwent TAVR from December 2011 through May 2019. Patients’ demographic information, clinical comorbidities, and operative details are summarized in Table 1. The patients’ median STS score was 6.0 (interquartile range [IQR], 4.0-8.5). One-third of patients had prior CABG, and one-quarter had previous PCI. The mean follow-up time was 3.28 ± 0.17 years, with a maximum follow-up of 7.5 years. A balloon-expandable valve (Edwards Lifesciences) was used in 92% of patients, and the femoral artery approach was used in 93% .

Mortality and its predictors. Overall, 71 deaths (37.6%) occurred, of which 54 (76.0%) were attributable to a cardiac cause. The median overall survival was 3.55 years (95% confidence interval [CI], 3.21-5.30 years). Thirty-day, 1-year, and long-term mortality rates were 2.6%, 10.6%, and 37.6%, respectively (Figure 1).

Five patients died within 30 days of surgery (30-day mortality, 2.6%). Three deaths were procedure related and 2 deaths were a result of postoperative complications. One patient died on postoperative day 0 because submitral chordal entanglement caused severe mitral regurgitation, for which extracorporeal membrane oxygenation and emergency open sternotomy were ineffective. One patient died of cardiogenic shock on postoperative day 4 after embolization of the prosthetic valve to the aortic arch requiring open sternotomy and SAVR. Another patient died because a failed Angio-Seal closure device (Terumo) caused a retroperitoneal bleed. Iliac-femoral artery stenting was performed for rescue. Nonetheless, the patient died of hemorrhagic shock on postoperative day 11. One patient discharged to a skilled nursing facility died of a gastrointestinal bleed associated with supratherapeutic anticoagulation. Lastly, 1 patient died of hospital-acquired pneumonia on postoperative day 25.

Twenty-one patients (11%) had an STS score <3, 106 patients (56%) had a score of 3-8, and 61 patients (32%) had a score >8. Survival in the STS <3 group did not reach a statistical median. Survival in the STS 3-8 group was 5.04 years (lower limit of 95% CI, 3.21) and survival in the STS >8 group was 2.71 years (95% CI, 2.07-3.37). There was a statistically significant survival difference among all 3 risk groups (P<.001) (Figure 1).

Four factors were independently associated with all-cause mortality: age (hazard ratio [HR], 1.04; 95% CI, 1.01-1.07), baseline creatinine (HR, 1.2; 95% CI, 1.07-1.36), COPD (HR, 1.81; 95% CI, 1.06-3.08), and significant CAD (HR, 2.19; 95% CI, 1.13-4.25) (Table 2). Fine-gray competing risk analysis revealed 5 independent predictors of cardiac-related mortality: age (HR, 1.05; 95% CI, 1.02-1.08), preoperative creatinine level (HR, 1.18; 95% CI, 1.03-1.34), peripheral vascular disease (HR, 1.74; 95% CI, 1.02-2.96), reduced baseline ejection fraction (HR, 1.02; 95% CI, 1.00-1.04), and significant CAD (HR, 2.62; 95% CI, 1.3-5.26) (Table 2 and Figure 2).The addition of severe stenosis of the left circumflex or the right coronary arteries to our definition of “significant CAD” had no additional effect on post-TAVR survival.

Procedural outcomes and postoperative complications. Postoperative complications for the entire cohort are listed in Table 3. Three patients had a cerebrovascular accident, 13 patients had a conduction disturbance requiring permanent pacemaker implantation, and 2 patients had a major vascular access-site related complication (1 iliac artery dissection and 1 retroperitoneal bleed). There were no coronary obstruction events. Unplanned use of cardiopulmonary bypass was necessary in 6 patients. Five patients were converted to open sternotomy for one of the following reasons: valve embolization into the left ventricle, valve embolization to the arch, contained aortic root rupture, mitral valve subchordal entanglement with severe mitral regurgitation, and ventricular septal defect.

Three patients required a reoperation within 30 days of their TAVR (debridement for a groin wound infection, placement of an inferior vena cava filter for deep vein thrombosis, and surgical explant of the prosthetic valve followed by SAVR and ventricular septal defect repair). Over the study period, 2 patients required SAVR for severe aortic insufficiency (1 for intravalvular aortic insufficiency and 1 for paravalvular leak).

Structural valve deterioration. Baseline and postoperative echocardiographic data are summarized in Table 1 and Table 4, respectively. The median time from the postoperative echocardiogram to the most recent surveillance echo was 1.55 years (95% CI, 0.30-2.53). A total of 167 patients (88%) had both immediate postoperative and subsequent follow-up echocardiograms available for comparison. Ten patients (6.0%) met criteria for at least moderate structural valve deterioration on follow-up echocardiogram. Median survival was not significantly different for those who developed moderate structural valve deterioration compared with all others (3.37 years vs 4.02 years, respectively; P=.96).

Our single-center study analyzed the longest existing follow-up data from TAVR patients in the VA healthcare system. We report favorable 30-day, 1-year, and long-term outcomes compared with those of non-veteran patients and show that significant CAD (unprotected LMCA or LAD ≥ 50% stenosis) is a significant independent predictor of long-term all-cause and cardiac mortality. Our VA clinical setting represents a unique opportunity to examine the outcomes of TAVR, given the availability of near-complete clinical follow-up data and prospective clinician-adjudicated angiographic data, and national quality initiatives to track and monitor care.^28,29

The favorability of our short- and long-term mortality findings observed in our cohort is self-evident, especially when these findings are compared with findings from clinical trials and real-world registries. We report 30-day and 1-year mortality rates of 2.6% and 10.6% in a cohort with a mean STS score of 6. The intermediate-risk trial of balloon-expandable TAVR in patients with an average STS of 5.8% (which is comparable to the STS score in our population) revealed all-cause mortality rates of 3.9% and 12.4% at 30 days and 1 year, respectively.^3,4 Even though the self-expanding TAVR platform was used in <10% of our cohort, our outcomes are also comparable with those of patients in the self-expanding TAVR intermediate-risk comparative trial (who had an STS score of 4.4 and 30-day and 1-year mortality rates of 2% and 7%, respectively).⁷

Our results also compare favorably with the most current real-world registry data. In their 2016 annual TVT registry report, Glover and colleagues reported reductions in in-hospital and 1-year mortality between 2012 and 2014 (from 5.7% to 2.9% and from 26% to 22%, respectively).³⁰ The most recent TVT registry annual report described 1-year all-cause mortality of 13.9% in a 2017 cohort with a median STS score of 5.1. Moreover, our survival rates are on par with those found in a large meta-analysis of 31 non-randomized TAVR studies with a total of 13,857 patients with at least 5 years of follow-up, in which survival rates at 1, 2, 3, and 5 years were 83%, 75%, 65%, and 48% vs 89%, 76%, 63%, and 43% in the present analysis, respectively.³¹

Of course, veterans have different demographics and risk profiles than their non-veteran counterparts,^1,3,4 which may not be fully reflected in their STS scores. Our 1-year survival rate of 89% is very similar to the 86% rate reported in 959 consecutive VA-TAVR patients from 8 VA-TAVR sites by Hall and colleagues.¹⁸ Our study extends their multicenter VA registry findings and adds long-term outcomes (maximum follow-up of 7.5 years).

We believe that our favorable outcomes are at least partially attributable to the early and consistent implementation of the heart team approach, the creation of a hybrid operating room from the early inception of the program, and most importantly, the careful selection process of eligible veterans after meticulous deliberation by the heart team. This process allowed us to appropriately select patients and achieve excellent outcomes comparable to those of higher-volume TAVR centers of excellence.

As with surgical bioprosthetic valves, the question of durability remains critically important in the nascent TAVR field, given especially that the earliest TAVR patients are nearing or have surpassed 5 years of post-TAVR follow-up. Overall, the structural valve deterioration rate observed in our cohort is similar to those in recently published reports.³² Long-term data from surgical bioprosthetic valves associate hemodynamic valve deterioration with greater mortality, but such data are lacking in TAVR patients.³² While we did not observe greater mortality in those with structural valve deterioration, longer follow-up may be necessary to reveal any meaningful clinical differences.

Finally, our report provides clinically relevant information regarding the prognostic implications of CAD in patients undergoing TAVR. Previous studies examining the effect of CAD on TAVR outcomes have yielded inconsistent and, in some cases, contradictory results.²² This is probably related to several factors including study design heterogeneity, variable definitions of CAD, inconsistent use of invasive physiological assessments (which is controversial anyway in patients with severe AS), and the possibility of implicit bias underlying treatment decisions by the heart team. We used a definition of CAD that included significant stenosis to the LMCA or LAD (that was not protected by a patent LIMA-LAD graft or recent PCI) to identify significant CAD that we believe is most clinically relevant. Our findings suggest that untreated CAD of the LMCA or LAD negatively affects survival after TAVR. While our data show the long-term prognostic implications of significant CAD, they do not provide information regarding the effects of revascularization.

Study limitations. There are several limitations to our analysis. As mentioned above, the veteran population is not representative of the national population in the United States. Although ours is one of the largest TAVR programs in the VA healthcare system, our single-institution experience is limited by its modest sample size. However, this was balanced out by the availability of complete and long-term clinical follow-up. Notably, the angiographic data were prospectively collected and adjudicated by the heart team, which adds robustness to our CAD findings. Finally, although we do not have 100% echocardiographic follow-up and adjudication, efforts are underway to have blinded operators examine and adjudicate serial echocardiograms from all patients, the results of which will be the subject of a separate follow-up report.

Long-term survival of veterans after TAVR is comparable to that of their non-veteran counterparts. Significant CAD, age, and specific medical comorbidities are independently associated with long-term all-cause and cardiac mortality. Continuous review of institutional TAVR outcomes in the VA healthcare system is needed to ascertain whether outcomes remain favorable and to allow better understanding of the effect of CAD in TAVR patients.

Acknowledgments. The authors wish to acknowledge Katherine Simpson for her statistical expertise, as well as Stephen N. Palmer, PhD, ELS and Jean Woodruff of the Department of Scientific Publications at the Texas Heart Institute for editorial assistance and graphical expertise, respectively.