Transcatheter aortic valve implantation (TAVI) has now become an established treatment for symptomatic severe aortic stenosis (AS). Multiple studies have demonstrated comparable, and in some instances, improved outcomes with TAVI than with surgical aortic valve replacement (SAVR), regardless of surgical risk.1 Coexisting coronary artery disease requiring revascularization is present in up to 20% of patients with severe AS with the prevalence of previous revascularization by either percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG) in this cohort being much higher.2–4 However, doubt remains regarding the optimal treatment for these patients, who have been excluded from most randomized trials. Recent data have suggested that PCI is safe prior to TAVI with low rates of target lesion and target vessel failure at 2 years.5 On the other hand, incomplete revascularization or high residual SYNTAX (Synergy Between PCI with Taxus and Cardiac Surgery) score confers an increased risk of major adverse cardiovascular and cerebrovascular events.5,6 Current guidelines suggest that both SAVR/CABG or TAVI/PCI may be appropriate in patients undergoing aortic valve replacement procedures but in the context of concomitant complex coronary artery disease, SAVR and CABG may be favored.7,8 However, head-to-head comparisons of TAVI/PCI vs SAVR/CABG are scarce. Recent meta-analyses of published studies to date have suggested comparable short-term outcomes between patients undergoing TAVI and PCI vs SAVR and CABG, but the included studies were small and with a high risk of bias.9,10 Furthermore, many studies on this important clinical question have included patients from the time when TAVI was first becoming recognized as a viable treatment for AS and may no longer reflect contemporary practice, current generation devices, and indeed, outcomes. As such, we aimed to assess contemporary outcomes in patients undergoing PCI and TAVI vs SAVR with concomitant CABG in a nationwide population.

From January 2016 to December 2019, 6194 patients were identified as meeting the inclusion criteria (977 TAVI and PCI vs 5217 SAVR and CABG) (figure 1). While all CABG procedures were performed during the same episode of care as SAVR procedures, the median time between PCI and TAVI was 61 [IQR 30-107] days. Several baseline characteristics differed between the groups. TAVI patients were older (81.1 vs 72 years, P <.001), were more commonly female (40.3 vs 26.4%, P <.001) and more commonly had a history of heart failure, prior CABG, chronic renal failure, atrial fibrillation, and chronic obstructive pulmonary disease (P <.001 for all comparisons). table 1 summarizes the baseline characteristics of the unmatched and matched cohorts.

Figure 1.

Flow chart depicting included and excluded patients from the original cohort. SAVR/CABG, surgical aortic valve replacement with coronary artery bypass grafting; TAVI/PCI, transcatheter aortic valve implantation with percutaneous coronary intervention.

(0.35MB).

Matched cohorts

Propensity score matching resulted in 774 pairs. The variables included in the matching were age, female sex, hypertension, disorders of lipid metabolism, diabetes, morbid obesity, other endocrine/metabolic/nutritional disorders, chronic renal failure, history of CABG or valve surgery, congestive heart failure, atrial fibrillation, presence of cardiac pacemaker, chronic obstructive pulmonary disease, stroke, liver, or biliary disease, and vascular or circulatory disease. Kernel density plots before and after matching are depicted in figure 2A, B. Although a close match was achieved, some differences remained. TAVI patients more commonly had a history of prior cardiac surgery (14.0 vs 3.9%; P <.001), hypertension (57.6 vs 50.5%; P=.005), and PPM (7.4 vs 4.9%; P=.044). However, the proportion of some variables (such as heart failure and chronic renal failure) were inverted between groups before and after the pairing, being more frequent in the SAVR cohort after the propensity match.

Figure 2.

Kernel Density plots representing pre- (A) and post- (B) matching. Patients were matched based on the following variables: age, sex, diabetes, obesity, chronic obstructive pulmonary disease, previous history of atrial fibrillation, previous stroke, renal failure, heart failure, liver disease and respiratory failure at presentation. SAVR/CABG, surgical aortic valve replacement with coronary artery bypass grafting; TAVI/PCI, transcatheter aortic valve implantation with percutaneous coronary intervention.

(0.11MB).

In-hospital outcomes

table 2 summarizes the in-hospital outcomes between the 2 cohorts. The primary endpoint (crude in-hospital all-cause mortality) was lower in the matched TAVI group vs SAVR (average treatment effect: 3.4% vs 9.4%; OR=0.34; 95%CI, 0.20-0.54; P <.001) (figure 3). After application of the sensitivity analysis, a crude all-cause mortality rate of 3.4% was obtained for patients with a diagnosis of AS undergoing PCI (AS/PCI) during in the study period who were not in either of the study groups. If this crude mortality rate were added to the average treatment effect of the TAVI group calculated in the original analysis (3.4%, best-case) a worst-case all-cause mortality of 6.8% would be obtained vs the SAVR/CABG average treatment effect: 9.4%, P <.005. Other in-hospital complications such as periprocedural stroke (0.9% vs 2.2%; P=.004), requirement for blood transfusion (9.6% vs 21.1%), acute kidney injury (4.3% vs 16.0%; P <.001), pericardial complications (1.4% vs 3.0%; P=.037), and hospital-acquired pneumonia (0.1% vs 1.7%; P=.001) were also more frequent in the SAVR group. Vascular complications (2.2% vs 0.5%; P=.004) and the need for in-hospital PPM implantation was higher in TAVI patients (12.0% vs 5.7%; P <.001). Length of stay, both in the intensive care unit and in-hospital overall, was significantly longer in those patients undergoing SAVR/CABG than in those undergoing TAVI/PCI. Data on 30-day cardiovascular readmission were available for all patients and did not differ between groups.

Table 2.

Postprocedure clinical endpoints for TAVI/PCI and SAVR/CABG cohorts

	Prematching			Postmatching
Clinical endpoints	TAVI & PCI(n=977)	SAVR & CABG(n=5217)	P	TAVI & PCI(n=774)	SAVR & CABG(n=774)	P
In-hospital all-cause mortality	3.0	7.0	<.001	3.4	9.4	<.001
Periprocedural stroke (CC100)	1.0	1.7	.103	0.9	2.2	.04
Acute kidney injury (CC135)	4.7	10.3	<.001	4.3	16.0	<.001
Blood transfusion	11.1	19.1	<.001	9.6	21.1	<.001
Pericardial complicationsa	1.5	3.5	.001	1.4	3.0	.037
Acute myocardial infarction (CC86)	0.4	1.4	.013	0.5	1.2	.164
Vascular complicationsb	3.0	0.6	<.001	2.2	0.5	.004
Aspiration and specified bacterial pneumonias (CC114)	0.2	1.2	.006	0.1	1.7	.001
New permanent pacemaker implantationc	13.0	3.5	<.001	12.0	5.7	<.001
New onset atrial fibrillation	1.3	2.8	.007	1.0	2.5	.033
Length of stay in intensive care	0 [0-1]	2 [0-4]	<.001	0 [0-1]	2 [0-4]	<.001
Length of hospital stay (days)	7 [6-10]	12 [9-20]	<.001	7 [5-10]	14 [9-23]	<.001
30-day cardiovascular readmission	7.55	7.8	0.63	6.4	6.8	.810

CABG, coronary artery bypass graft; PCI, percutaneous coronary intervention; SAVR, surgical aortic valve replacement; TAVI, transcatheter aortic valve implantation.

Values are expressed as No. (%) or median [interquartile range].

a

Pericardial complications including pericarditis, pericardial effusion, hemopericardium, cardiac tamponade, and the need for pericardiocentesis or pericardial window.

b

Vascular complications including accidental puncture or arteriovenous fistula.

c

Calculated only for patients without pre-exiting permanent pacemakers.

CC: Condition categories. Secondary diagnoses grouped on risk factors (Pope et al).11

Figure 3.

Central illustration. In-hospital complications for patients undergoing TAVI/PCI vs those undergoing SAVR/CABG. AKI, acute kidney injury; SAVR/CABG, surgical aortic valve replacement with coronary artery bypass grafting; TAVI/PCI, transcatheter aortic valve implantation with percutaneous coronary intervention.

(0.28MB).

Predictors of in-hospital mortality

The multivariable logistic regression model used for in-hospital all-cause mortality demonstrated acceptable discrimination with an AUROC of 0.70. When procedures were included in this model, SAVR had a higher risk of in-hospital all-cause mortality than TAVI (OR=3.1; 95%CI, 2.1-4.5; P <.001).

We also investigated whether procedure volume impacted on in-hospital all-cause mortality. Using a clustered algorithm with K-means, a cutoff of 130 procedures per year for both TAVI/PCI and SAVR/CABG was optimum for demonstrating a difference in the primary outcome. Those centers performing <130 procedures per year demonstrated an increased risk of in-hospital risk-adjusted all-cause mortality: TAVI/PCI 3.6% vs 2.9%; P <.001 and SAVR/CABG 8.3% vs 6.8%; P <.001 for low- and high-volume centers respectively (figure 4).

Figure 4.

Differences in in-hospital all-cause mortality for TAVI/PCI and SAVR/CABG by center volume (low <130, high ≥ 130 procedures per year). SAVR/CABG, surgical aortic valve replacement with coronary artery bypass grafting; TAVI/PCI, transcatheter aortic valve implantation with percutaneous coronary intervention.

(0.15MB).

Our study presents the results from a large propensity matched population of symptomatic patients with AS and severe coronary artery disease undergoing either entirely percutaneous (TAVI and PCI) or entirely surgical (concomitant SAVR and CABG) management. The main findings are as follows: a) in-hospital all-cause mortality was more frequent among SAVR/CABG patients than TAVI/PCI patients with an OR of 3.1; b) periprocedural complications including periprocedural stroke, requirement for blood transfusion, acute kidney injury, pericardial complications and hospital-acquired pneumonia were all more common in the SAVR/CABG group; c) rates of PPM implantation prior to discharge were twice as high in the TAVI/PCI cohort than the SAVR/CABG cohort; d) procedure volume significantly impacted in-hospital all-cause mortality in both (percutaneous and surgical) groups.

Coronary artery disease is highly prevalent among patients presenting with symptomatic severe AS and coronary revascularization continues to present a clinical dilemma in the choice between TAVI and SAVR. Indeed, even the indication for revascularization may be difficult to ascertain with certainty, due to considerable overlap between the symptoms of coronary artery disease and those of severe AS. Nonetheless, current guidelines recommend revascularization for stenosis ≥ 70% in proximal segments, and an almost 10-fold increase in percutaneous revascularization prior to TAVI has been reported.7,13 With repeatedly demonstrated equivalent outcomes between TAVI and SAVR, the concept of an entirely percutaneous approach in the presence of coronary artery disease presents an attractive, less invasive option. Registry data support a percutaneous approach with good outcomes for patients undergoing PCI prior to TAVI regardless of anatomical complexity.5 Direct comparisons of TAVI/PCI vs SAVR/CABG are, however, scarce. The only randomized data available to date on this topic is provided by the SURTAVI trial.14 In that study, a subgroup of patients with coronary artery disease (SYNTAX score ≤ 22 without left main disease) were randomized to revascularization by either PCI, in those undergoing TAVI, or CABG, in those undergoing SAVR, and no differences in all-cause mortality or stroke were seen at 30-days or 2 years. Similarly, a post hoc analysis of the PARTNER 3 study demonstrated no difference in the composite endpoint of death from any cause, stroke or rehospitalization (hazard ratio=0.52; 95%CI, 0.11-2.49) in a small subgroup of patients undergoing TAVI with concomitant PCI (n=32) vs SAVR with CABG (n=58)2. Many of the remaining data come from observational registry studies. In a propensity matched analysis of the OBSERVANT registry, Barbanti et al.15 found no difference in early mortality in those undergoing TAVI/PCI vs SAVR/CABG and, similarly, an analysis of the National Inpatient Sample database in the US between 2012 and 2017 again found no difference in in-hospital mortality between groups,13 although a more recent analysis of this database including only those treated between 2016 and 2017 demonstrated a lower adjusted OR for in-hospital mortality in the TAVI/PCI group (OR=0.32; 95%CI, 0.17-0.62; P=.001).16 More recently, in the context of complex coronary artery disease (left main stem or SYNTAX score> 22), no difference in major adverse cardiovascular or cerebrovascular events at 3 years was found between those undergoing TAVI/PCI vs SAVR/CABG, although a higher rate of new revascularization was seen in the TAVI/PCI group.17

Unlike these studies, our study found increased in-hospital all-cause mortality in patients undergoing SAVR/CABG vs TAVI/PCI. One explanation could be that, in our study, percutaneous revascularization was performed as a staged procedure prior to TAVI, allowing patients time to recover, while the previously cited studies included concomitant TAVI/PCI in some, but not all, patients. Additionally, in our study, patients who underwent PCI and died prior to undergoing TAVI were not included. However, in our sensitivity analysis, which attempted to account for this, TAVI/PCI continued to have lower in-hospital all-cause mortality vs SAVR/CABG when taking “worst-case scenario” mortality rates into account. Lastly, particularly in retrospective observational studies, PCI may have been performed as a complication of TAVI rather than as a planned procedure and may explain the absence of mortality difference between the groups compared with our study, which excluded patients who underwent PCI and TAVI in the same care episode to exclude those that had PCI as a result of a complication or as a ‘rescue’ treatment. The subanalysis of the National Inpatient Sample database by Patlolla et al.13 would support a treatment strategy of staged percutaneous procedures. In that study, patients who underwent staged PCI followed by TAVI had reduced in-hospital mortality compared with those undergoing same day TAVI/PCI. Conversely, concomitant SAVR with CABG is carried out simultaneously resulting in longer procedure times, and time on cardiopulmonary bypass, which may partly explain the increased in-hospital all-cause mortality seen in the SAVR/CABG group in our study. Although our study contributes important data on this issue, specific prospective randomized studies are clearly required to further compare TAVI/PCI with SAVR/CABG. The currently recruiting TCW trial (NCT03424941) aims to examine this issue using a more nuanced approach to percutaneous revascularization with prior fractional flow reserve assessment and the use of newer generation devices.

Acute kidney injury was higher in the SAVR/CABG group than in the TAVI/PCI group, which has been a consistent finding in all TAVI vs SAVR studies to date, with and without revascularization.9,18–20 The ability to stage revascularization prior to the TAVI procedure can result in reduced nephrotoxic agents (contrast) with less subsequent kidney injury and more time for renal function recovery between procedures. When given the choice, staging PCI and TAVI procedures appears to be the preferred option for many physicians, with both the SURTAVI study14 and that of Barbanti et al.15 having predominantly staged PCI procedures (59% and 92% respectively), despite no prespecification for staging (in fact in the SURTAVI study, concomitant TAVI and PCI procedures were encouraged). Subsequently, acute kidney injury rates in these studies were similar to those in our present study. Staging CABG and SAVR procedures, however, is not recommended and concomitant SAVR/CABG are longer, with longer cardiopulmonary bypass times being previously associated with a higher incidence of acute kidney injury.21 This may be a particularly important consideration in patients with chronic kidney disease, a prevalent comorbidity in patients with symptomatic AS (and present in ∼20% of our cohort). The importance of avoiding periprocedural acute kidney injury cannot be over emphasized, with acute kidney injury being consistently found as a predictor of both short- and long-term mortality in both TAVI and SAVR populations with and without pre-existing renal dysfunction.20–23 The lower incidence of this complication in the TAVI/PCI population is therefore an important consideration when choosing between purely percutaneous and surgical strategies in patients with severe AS and concomitant coronary artery disease, with TAVI/PCI offering some advantages in this regard.

Requirement of PPM continues to be a hurdle for TAVI, and our study again demonstrated an increase in PPM requirement among the TAVI population compared with SAVR. However, it is evident that the rate of PPM following TAVI has been steadily decreasing and this is reflected in the lower PPM requirement in our study than that of the SURTAVI trial, and similar to the more recent study by Barbanti.14,15 One likely explanation is that our study included all types of available transcatheter valves and more contemporary patients (inclusion from 2016-2019) compared with the SURTAVI trial, which included only self-expanding valves and recruited patients from 2012-2016. Recent advances in implantation techniques for TAVI particularly with self-expanding valves, has resulted in decreased pacemaker requirement in more contemporary studies, and adaption of these techniques across centers may explain the lower PPM rates seen in our study.

In this study, center experience reflected in procedure volume demonstrated an impact on in-hospital all-cause mortality. For both TAVI/PCI and SAVR/CABG, a procedure volume <130 per year was associated with increased in-hospital all-cause mortality compared with centers performing ≥ 130 procedures per year. The association between procedure volume and outcomes has been previously demonstrated in a meta-analysis by He et al., and analyses of the TVT registry.24–26 Similar to our study, reduced in-hospital and 30-day mortality, as well as other periprocedural complications such as bleeding and stroke have been found with increased procedural volume in these aforementioned studies.24–26 Patients requiring both coronary revascularization and treatment of severe AS are inherently complex, making these findings unsurprising. They do highlight, however, the potential beneficial impact on outcomes of high-volume, specialized heart centers with combined surgical and interventional skills acting as tertiary referral centers and this should be a point of discussion and further research.

In a large contemporary national database, patients with symptomatic severe AS and concomitant coronary artery disease requiring revascularization, percutaneous intervention (TAVI/PCI) resulted in decreased in-hospital all-cause mortality, periprocedural stroke, blood transfusion requirement, acute kidney injury, and hospital-acquired pneumonia than SAVR with concomitant CABG. Patients in the TAVI cohort had an increased requirement for PPM implantation. In both procedures, greater center experience was associated with reduced in-hospital all-cause mortality. These findings are important considerations for the cardiology community managing patients with AS and concomitant coronary artery disease and suggest the potential usefulness of an entirely percutaneous approach in these complex patients.

None.

A. McInerney, M. García-Marquez, J.L. Bernal, C. Fernandez-Perez, J. Elola, L. Nombela-Franco: substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work, drafting the work or revising it critically for important intellectual contente and final approval of the version to be published;

G. Tirado-Conte, P. Jimenez-Quevedo, N. Gonzalo, I. Nuñez-Gil, N. Prado, J. Escaned, A. Fernández-Ortiz: drafting the work or revising it critically for important intellectual contente and final approval of the version to be published.

L. Nombela-Franco has received consulting fees from Edwards Lifesciences and Abbott Vascular. The other authors have no relationships to disclose.