Introduction

Transcatheter aortic valve replacement (TAVR) has been shown to have similar or even better clinical outcomes compared with surgical aortic valve replacement (SAVR) in patients with severe symptomatic aortic stenosis (AS) and at least intermediate surgical risk (1–3). Prior cardiac surgery (PCS) is associated with increased morbidity and mortality in patients undergoing cardiac surgery, as it is technically challenging because of scarring of tissues resulting in loss of tissue planes, adhesions, and injury to adjacent anatomic structures (4). In patients with prior coronary artery bypass grafting (CABG), there is an increased risk for injury to patent grafts (arterial or venous) during subsequent cardiac surgery that has been associated with increased mortality and morbidity (5). Furthermore, there are technical issues related to myocardial protection during aortic cross-clamping (6–9). The increased risk associated with PCS is incorporated in the European System for Cardiac Operative Risk Evaluation (EuroSCORE I or II) (10,11), and Society of Thoracic Surgeons (STS) (12,13) scores. However, other observational studies have reported no additional risk in patients with PCS undergoing SAVR (14). Irrespective of the increase in surgical risk for patients with PCS, a single-center retrospective study reported comparable clinical outcomes of TAVR and SAVR in high-risk patients with PCS (15). Furthermore, an analysis of the PARTNER 1 (Placement of Aortic Transcatheter Valve) trial (2) reported that patients with prior CABG and high surgical risk had better 2-year clinical outcomes with SAVR than TAVR, because of higher rates of repeat hospitalizations and a trend toward a higher rate of all-cause death in the TAVR arm (16). In contrast, another subgroup analysis of patients with prior CABG in the CoreValve high-risk study found that TAVR had a significant morbidity advantage, with a trend toward improved survival over SAVR at 1 year (17). Thus, it is not yet clear whether the effect of TAVR versus SAVR is different for patients with and without PCS, especially in patients with intermediate surgical risk.

We sought to assess whether the relative 30-day and 2-year risk for adverse clinical outcomes after TAVR with SAPIEN XT compared with SAVR for patients with severe symptomatic AS and intermediate surgical risk was different for patients with versus without PCS in the PARTNER 2A trial (3).

Methods

Study design and population

The design and results of the PARTNER 2A trial ( NCT01314313) have been previously described (3). Briefly, Cohort A of the PARTNER 2 trial enrolled patients with severe, symptomatic AS at intermediate surgical risk at 57 sites in the United States and Canada. Severe AS was defined as: 1) aortic valve area ≤0.8 cm² or aortic valve area index ≤0.5 cm²/m²; and 2) mean aortic valve gradient >40 mm Hg or peak aortic jet velocity >4.0 m/s. Patients were considered to be at intermediate surgical risk if they had predicted 30-day surgical mortality of 4% to 8% as determined by the STS mortality risk model (possible range of risk: 0% to 100%; higher percentages indicate greater risk) (13) and a multidisciplinary heart team. Key exclusion criteria included patients with congenitally bicuspid aortic valves, severe renal disease, predominant aortic regurgitation, or left ventricular ejection fraction (LVEF) <20%. Patients were randomized to undergo TAVR with the SAPIEN XT valve or SAVR. The present analysis used the intention-to-treat population of patients randomized to TAVR or SAVR in PARTNER 2, Cohort A.

Definitions and event adjudication

The primary endpoint of the original study was a composite of death from any cause or disabling stroke at 2 years. The definitions of the various endpoints are provided in the supplementary appendix to the original publication (3). A clinical events committee adjudicated all adverse outcomes. All electrocardiograms and echocardiograms were interpreted by independent core laboratories using methodology previously described (18). The severity of bleeding, vascular complications, and acute kidney injury were graded according to the Valve Academic Research Consortium 2 (19). Notably, major vascular complications by this definition include: 1) any aortic dissection, aortic rupture, annular rupture, left ventricular perforation, or new apical aneurysm or pseudoaneurysm; 2) access site or access-related vascular injury (dissection, stenosis, perforation, rupture, arterio-venous fistula, pseudoaneurysm, hematoma, irreversible nerve injury, compartment syndrome, percutaneous closure device failure) leading to death, life-threatening or major bleeding, visceral ischemia, or neurological impairment; 3) distal embolization (noncerebral) from a vascular source requiring surgery or resulting in amputation or irreversible end-organ damage; 4) the use of unplanned endovascular or surgical intervention associated with death, major bleeding, visceral ischemia, or neurological impairment; 5) any new ipsilateral lower extremity ischemia documented by patient symptoms, physical examination, and/or decreased or absent blood flow on lower extremity angiography; 6) surgery for access site–related nerve injury; and 7) permanent access site–related nerve injury.

Statistical analysis

Continuous variables are presented as mean ± SD and were compared using the Student’s t-test. Categorical variables are reported as percentages and frequencies and were compared using the chi-square or Fisher exact test, as appropriate. Time-to-event variables are presented as Kaplan-Meier event rates and were compared using the log-rank test. Adjusted comparisons of clinical outcomes and echocardiographic parameters were conducted using multivariate Cox proportional hazards regression models. Covariates included in the adjusted models were age, sex, body mass index, diabetes mellitus, hypertension, dyslipidemia, chronic obstructive lung disease, oxygen-dependent chronic obstructive pulmonary disease, chronic kidney disease, LVEF, coronary artery disease, prior myocardial infarction, prior percutaneous coronary intervention, and peripheral vascular disease. Interaction terms were included in the covariate set to assess whether the effect of TAVR versus SAVR differed according to the presence or absence of PCS.

Results

Patient population and baseline characteristics

Of the 2,032 patients included in the present analysis, 509 (25.1%) had PCS, 245 (12.1%) in the TAVR group and 264 (13.0%) in the SAVR group. Baseline characteristics are presented in Table 1. Patients with PCS were significantly younger, more frequently male, and had higher rates of diabetes and higher body mass index compared with patients without PCS. Baseline STS score, logistic EuroSCORE, and SYNTAX (Synergy Between PCI With Taxus and Cardiac Surgery) score were significantly higher in patients with PCS. Expectedly, prior percutaneous intervention and prior myocardial infarction as well as peripheral vascular disease were more common among patients with prior PCS. In both groups (PCS and no PCS), there were no significant differences between patients randomized to TAVR or SAVR except for higher rates of hypertension and peripheral vascular disease among patients with PCS who were randomized to SAVR compared with those randomized to TAVR.

Baseline echocardiographic characteristics

Table 2 presents baseline echocardiographic characteristics stratified by PCS group (PCS vs. no PCS) and by the randomized treatment (TAVR vs. SAVR). Aortic valve area was significantly higher and aortic valve mean gradient and LVEF were both lower in patients with PCS. Comparing TAVR with SAVR patients within each PCS group revealed no significant differences, except for LVEF, which was lower in SAVR than TAVR patients in the no-PCS group.

Procedural characteristics

Procedural variables are displayed, stratified by PCS group, in Table 3. Among patients who underwent TAVR, in both the PCS and no-PCS groups, transfemoral access was the most commonly used approach. For patients who were not treated transfemorally, the transapical approach was more common in patients with PCS than without PCS, while the opposite was true for the transaortic approach. Prosthesis size in TAVR patients ranged from 23 to 29 mm; 18.1% of PCS patients and 40.5% of no-PCS patients (p < 0.0001) were implanted with prostheses 23 mm in size. Fluoroscopy duration and the time to discharge post-TAVR were both longer in no-PCS than in PCS patients. However, the volume of contrast media delivered during the procedure did not differ significantly between PCS groups. In contrast, prosthesis size in the SAVR group ranged from 17 to 29 mm, with 70.6% of PCS patients and 83.3% of no-PCS patients implanted with prostheses 23 mm in size or smaller (p < 0.0001). Valve size used with SAVR remained significantly smaller than with TAVR, after controlling for the PCS group (p < 0.0001). In patients treated with SAVR, procedure duration and cross-clamp time were significantly longer in patients with PCS.

Clinical outcomes

Overall, patients with PCS had similar 30-day outcomes as those with no-PCS, with the exception of life-threatening or disabling bleeding, which was more frequent in patients with PCS (Table 4, Figure 1). On the contrary, there was a trend toward higher 30-day rates of the primary endpoint and all-cause death among patients with no PCS. Myocardial infarction, including periprocedural myocardial infarction, was more common among patients with PCS; however, this did not reach statistical significance. A significant interaction was observed between PCS group and treatment arm for major vascular complication (p_interaction < 0.0001). Whereas in patients with no PCS, major vascular complications were significantly more common with TAVR than SAVR, the opposite was true for patients with PCS. Major vascular complications were mainly access site or access-related vascular injury. Bleeding (any) events were more common among patients who underwent SAVR than TAVR in both PCS groups, driven by life-threatening or disabling bleeding, but a significant interaction was observed between PCS group and treatment arm for any bleeding (p_interaction = 0.003), driven by life-threatening or disabling bleeding (p_interaction = 0.01). After multivariate adjustment, PCS remained a moderator of the effect of TAVR versus SAVR on the risk for major vascular complication, bleeding (any), and life-threatening or disabling bleeding, with disproportionally lower risk with TAVR versus SAVR among patients with PCS (Table 5).

At 2 years (median follow-up was 2.0 years; interquartile range: 1.7 to 2.0 years), the primary composite endpoint occurred more frequently in patients with no PCS than in patients with PCS (Table 6). This was driven by an increased all-cause death rate, resulting from a higher noncardiovascular death rate. Expectedly, patients with PCS had higher rates of 2-year myocardial infarction. Nonetheless, the relative risk associated with TAVR versus SAVR did not differ significantly between patients with versus without PCS (Table 6, Figure 2). No significant interaction was observed between the PCS group and the treatment arm at 2-year follow-up. These results remained similar after multivariate adjustment (Table 5).

When analyses were restricted to the as-treated population (n = 1,938; PCS, n = 485; no PCS, n = 1,453), or to only those patients who had undergone prior CABG (excluding the 12 patients who underwent sternotomy for other reasons besides CABG), these results remained consistent.

Discussion

The major finding of this subanalysis of the PARTNER 2A trial comparing the outcomes of TAVR and SAVR in intermediate-risk patients stratified by PCS is that patients with and without PCS undergoing TAVR with the SAPIEN XT valve had similar short- and long-term clinical outcomes compared with those treated with SAVR. However, major vascular complications and life-threatening bleeding were disproportionally more common among patients with PCS who underwent SAVR. The present study thus supports the long-term efficacy of both TAVR and SAVR in patients with PCS, with a possible early safety advantage for TAVR versus SAVR.

The similar mortality and repeat hospitalization rates observed between treatment arms in patients with PCS are in line with a previous small retrospective study (15) but in contradiction to a previous publication reporting a trend toward increased all-cause mortality rate and increased repeat hospitalization rate among patients with PCS who underwent TAVR compared with SAVR in the PARTNER 1 trial (16). This discrepancy might be explained by differences in the baseline risk of the study populations. The PARTNER 1 trial enrolled patients at high operative risk, while PARTNER 2 enrolled those at intermediate risk. In both trials, risk was defined using STS score, and PCS factors prominently in this calculation. Thus, in intermediate-risk patients, the relative weight of PCS in the STS score calculation is high, as these patients have relatively low rates of other comorbidities. This is likely the reason for the observed lower rate of 2-year all-cause death, specifically noncardiovascular death, in PCS patients compared to no-PCS patients in both treatment arms. On the contrary, in high-risk patients, the relative weight of PCS is low because the STS score may be driven by other comorbidities. Thus, the comparatively low comorbidity rates of patients with PCS in the present (intermediate-risk) study likely counteract the disadvantage of having PCS, resulting in similar long-term clinical outcomes. The last potential explanation is methodological, as the previous study did not examine the interaction between PCS group and treatment arm and only compared TAVR with SAVR in patients with PCS, hence rendering its results less statistically valid (20,21).

An improvement in procedural skills among TAVR operators due to accrued experience with the procedure, or iterative device improvements between the 2 studies (SAPIEN classic in PARTNER 1 vs. SAPIEN XT in PARTNER 2) (22) may also explain the current results. For example, 30-day moderate or severe paravalvular leak rates, which have been shown to affect prognosis (23–25), decreased from 12.2% in PARTNER 1A to <4% in PARTNER 2A (2,3). The clinical impact of this reduction may be especially pronounced in patients with PCS, as the presence of ischemic heart disease makes them particularly vulnerable to the negative influence of aortic regurgitation on coronary flow reserve (26).

In the present study, higher rates of major vascular complications observed in patients with PCS who underwent SAVR compared with TAVR were driven by access site–related complications. This is unlike the overall patient population, in which major vascular complications were significantly more common in TAVR compared with SAVR (3). There was a noticeably increased vascular complication rate in PCS patients who underwent SAVR, likely because the operating field was not naive: post-surgery adhesions might complicate access, and repeat surgery poses a risk for iatrogenic injury to the patent coronary graft (5). Compounding this interaction, we saw decreased vascular complication rates with TAVR in PCS patients compared with those without PCS. This is likely due to the comparatively low comorbidity burden of patients with PCS compared with those without PCS, as explained previously. The relatively low vascular complication rate in patients with PCS who underwent TAVR might also be attributed to significant differences in access route as transapical was relatively common and transaortic less common compared with no-PCS patients. In the subanalysis of patients who underwent prior CABG from the PARTNER 1 trial, higher rates of vascular complications were observed in TAVR compared with SAVR in patients with PCS (16). The discrepancy with our findings might be explained by the advances in device configuration, including the ability to mount the valve on the deployment balloon inside the abdominal aorta, allowing a decrease in the delivery system size that occurred between the 2 trials.

Finally, bleeding was significantly more common among patients with PCS who underwent SAVR than TAVR, driven by life-threatening or disabling bleeding. This finding is fairly expected, given that repeated sternotomy and lysis of adhesions are required during the SAVR procedure, and it is in line with the increase in access-related major vascular complications and with previous publications reporting higher rates of bleeding and blood transfusion among patients with PCS who underwent SAVR compared with TAVR (15,16,27). It is noteworthy that the increase in bleeding event rates had no effect on mortality or length of hospitalization in the present study.

Study limitations

The present study was a post hoc analysis of a randomized trial and therefore subject to the usual limitations for this type of analysis. The PARTNER 2A trial was not powered to examine outcomes according to the presence of PCS. Despite using an intention-to-treat population, and the fact that our findings regarding elevated major vascular complication and life-threatening bleeding rates among patients with PCS treated with SAVR remained statistically significant after multivariate adjustment, we cannot rule out the possibility that the analysis is confounded by other unmeasured factors that are correlated with PCS. Although the study identified differences in major vascular complications and bleeding events as defined by Valve Academic Research Consortium 2, the prognostic significance of these events may differ between patients treated with SAVR and TAVR.

Conclusions

In the PARTNER 2A trial, patients with severe AS at intermediate surgical risk and PCS had similar 2-year clinical outcomes when treated with SAPIEN XT TAVR or SAVR. However, 30-day major vascular complication and life-threatening bleeding events were disproportionally more common with SAVR than with TAVR in patients with versus without PCS. Thus, compared with SAVR, TAVR may be associated with a relatively lower risk for periprocedural complications for patients with PCS.

Footnotes