Original research•

Coronary artery disease and outcomes following transcatheter aortic valve implantation

Abstract

Background Aortic stenosis is a life-limiting condition for which transcatheter aortic valve implantation (TAVI) is an established therapy. Coronary artery disease (CAD) is frequently found in this patient group and optimal management in these patients remains uncertain.

Objectives We sought to examine the association of coexistent CAD on mortality and hospital readmission in patients undergoing TAVI.

Methods In this observational cohort study, we examined patients who underwent TAVI and segregated them by the presence of obstructive epicardial CAD. The primary outcome was 3-year mortality with secondary outcomes being readmission for (1) all-causes, (2) a MACE (Major Adverse Cardiovascular Event) composite endpoint and (3) acute coronary syndrome. Subsidiary outcomes included patient angina and breathlessness scores.

Results 898 patients underwent TAVI, of which 488 (54.3%) had unobstructed coronary arteries and 410 (45.7%) had obstructive CAD. Overall, n=298 (33.2%) patients experienced the primary mortality endpoint with no significant difference when stratified according to CAD (n=160 (32.9%) vs n=136 (33.2%), HR 0.98, CI 0.78 to 1.24). After multivariate analysis, the presence of CAD had no effect on the primary outcome (HR 0.98, CI 0.68 to 1.40). There was no significant difference in readmission for any cause (n=181, 37.1% (CAD) vs n=169, 41.2% (no CAD), p=0.23), including no significant difference on readmission for MACE (n=48, 9.8% (CAD) vs n=45, 11.0% (no CAD), p=0.11). CAD at the time of TAVI also did not alter breathlessness or angina scores before/after TAVI (p>0.05).

Conclusion Coexistent CAD had no significant association with mortality, any-cause readmission or symptoms for patients undergoing TAVI in our cohort.

What is already known on this topic

  • Concomitant coronary artery disease (CAD) is a frequent finding in patients undergoing transcatheter aortic valve implantation (TAVI).

  • There is limited and conflicting randomised controlled trial and observational data on the effect of percutaneous coronary intervention for stable CAD on outcomes following TAVI.

What this study adds

  • This is a large, single-centre retrospective study of patients undergoing TAVI in which we demonstrated that the presence of concomitant CAD had no association with mortality or all-cause readmissions at 3 years.

How this study might affect research, practice or policy

  • This study suggests that a default conservative management strategy of stable bystander CAD in patients undergoing TAVI could be safe and does not adversely affect clinical outcomes.

Introduction

Transcatheter aortic valve implantation (TAVI) is now a well-established therapy for severe aortic stenosis (AS) and has been shown to be superior to medical management and at least equivalent to surgical aortic valve replacement (sAVR) across different risk cohorts.1,2 With a progressively ageing population, the demand for TAVI is likely to rise significantly.3 Part of the morbidity and mortality associated with severe AS may be attributable to the presence of concomitant coronary artery disease (CAD). About 30% of patients undergoing sAVR have coexisting CAD,4 which is an independent predictor of outcomes in this patient population.5 It is an established surgical practice that bystander CAD >70% stenosis (in proximal vessels) be revascularised by bypass grafting at the time of sAVR.6

Compared with patients with severe AS undergoing sAVR, the prevalence of CAD is higher in those undergoing TAVI with estimates ranging from 40% to 75% depending on the definition of CAD used.7 Furthermore, the pattern of CAD tends to be multivessel often with vessel calcification.8 Despite this finding, the evidence for routine percutaneous revascularisation prior to TAVI is unclear. In a meta-analysis of over 8000 patients from 15 studies, pre-existing CAD had no impact on 30-day all-cause mortality (OR 1.07 (95% CI 0.82 to 1.40); p=0.62) but was associated with higher all-cause, 1-year mortality (OR 1.21 (95% CI 1.07 to 1.36); p=0.002).9 Another meta-analysis comprising seven studies with 2472 patients showed that CAD had no impact on mid-term outcome after a median follow-up of 452 days.10 Further, it has also been shown that those patients who underwent percutaneous coronary intervention (PCI) demonstrated higher rates of major vascular complications and mortality at 30 days.8 The only prospective randomised controlled trial (RCT) comparing elective preprocedure PCI versus no PCI in patients undergoing TAVI with significant (>70% stenosis) CAD is ACTIVATION.11 The study randomised 235 patients to the treatment of CAD by PCI versus medical management but failed to meet recruitment targets and was underpowered. Nonetheless, there was no demonstrated difference in the primary endpoints of death or rehospitalisation at 1 year of follow-up.

The impact of coexisting stable CAD in patients with TAVI, and its management, particularly in the long term, remains unclear with conflicting data from the literature. We therefore sought to examine this question by a large retrospective long-term analysis of patients undergoing TAVI at a high-volume centre in the UK.

Methods

Data for the analysis were extracted from the Glenfield TAVI registry. This is a database of all patients undergoing TAVI at the Glenfield Hospital, generated by extraction of information from patients’ medical records by clinically trained staff. These data are maintained on an electronic database which is curated by the audit team. Using this database, all patients who underwent TAVI between January 2013 and December 2017 were identified and included for analysis.

Assessment of CAD

As part of their diagnostic workup, all patients underwent invasive coronary angiography prior to their TAVI. The presence of CAD, for the purposes of the registry, was defined as >50% stenosis in two or more orthogonal views and was assessed by the performing operator. Patients were subsequently segregated into those with and without epicardial CAD. Further subgrouping was performed including stratifying patients into those with and without left main stem (LMS) disease and the number of epicardial coronary arteries with >50% luminal stenoses. In those patients who had previous surgical revascularisation, grafts were included in the assessment of CAD. Those with severe native CAD but unobstructed grafts to all myocardial territories were deemed not to have CAD for the purposes of this study. Conversely, those with >50% stenosis in one or more grafts were deemed as having CAD.

Clinical outcomes

The primary outcome assessed was mortality at 3 years of follow-up. This was obtained from the centralised electronic record system (Hospital Information Support System—HISS) which was managed by the University Hospitals of Leicester audit and data team.

Secondary outcomes included (1) readmission for any cause; (2) readmission for a non-hierarchical composite of myocardial infarction (MI), hospitalisation due to heart failure, stroke/transient ischaemic attack (TIA), peripheral vessel disease; and (3) readmissions for acute coronary syndrome (ACS) presentations including ST-Elevation Myocardial Infarction (STEMI), Non ST-Elevation Myocardial Infarction (NSTEMI) and unstable angina.

Readmission data were obtained from the medical records localised within a centralised database (HISS), and the reason for readmissions was captured from the clinical coding data. Similarly, medical records were analysed at other hospitals (from which patients were referred) to ensure all readmissions were captured.

Subsidiary outcomes included patient-reported symptoms of breathlessness (graded by New York Heart Association (NYHA) class) and angina (graded by Canadian Cardiovascular Society (CCS) class) before and after TAVI. This information was obtained from the TAVI registry.

Statistical analysis

Categorical variables are presented as numbers and percentages, whereas continuous variables are presented as means±SD, assuming normality unless otherwise specified. The statistical significance of baseline differences was determined by χ2 test or t-test as appropriate. Survival analysis was estimated by the Kaplan-Meier method, and differences in survival between groups was assessed by log-rank tests. For multivariate regressional modelling, initial non-hierarchical univariate analysis was performed (by either χ2 or unpaired t-test) using variables included within the database and deemed by the authors to be clinically important in patients with severe AS undergoing TAVI. This was followed by Cox proportional regression modelling. A p value <0.05 was deemed to be statistically significant.

Results

Patient characteristics

A total of 898 consecutive patients (54.2% male, mean age 82.4±6.5 years) underwent TAVI at the Glenfield Hospital between January 2013 and December 2017 and were included in the analysis. Approximately one-quarter (27.8%, n=250) of patients had a diagnosis of diabetes, with 60.0% (n=540) having current or previous exposure to smoking. Over 70% of patients had preserved Left Ventricular (LV) systolic function on transthoracic echo. Most patients (80.7%, n=725) had Clinical Frailty Scores (CFS) of 3 or less. A patient with a score of 4 or more was considered ‘frail’ (19.3%, n=173).

Procedural details

Most TAVI implants (n=889) were performed for degenerative severe AS (including 21 in restenosed bioprosthetic valves). Most procedures were performed via transfemoral access (n=807, 89.8%) under local anaesthesia and sedation, and the remaining 91 were either transapical, subclavian or axillary cutdowns performed under general anaesthetic (categorised as a ‘non-femoral’ approach). Of all patients, 86 had a pacemaker prior to their TAVI and 142 required a de novo permanent system (15.8%) post TAVI. Overall, 37.0% of valves were balloon-expandable and the remainder 63% were self-expanding valves.

Extent of CAD

Overall, 244 patients had a history of coronary revascularisation (n=155 had coronary artery bypass grafting; n=130 had PCI; n=41 patients had a hybrid of both modalities of revascularisation). Of all the patients within the registry, 18.4% (n=165) had documented evidence of MI, of which 68 were treated by PCI prior to their assessment for TAVI. Of those with a documented MI (n=165), a total of 18 patients (10.9%) had an event within 3 months prior to their TAVI. Overall, n=6 patients had PCI in the interval between their diagnostic coronary angiogram, performed as part of the TAVI assessment, and their TAVI procedure itself. The remaining 89.1% (n=147) of patients with a history of MI had their event more than 90 days prior to their TAVI.

When patients were stratified by the presence of residual CAD at the time of TAVI, 488 patients had unobstructed coronary arteries (54.3%). Of these patients, 66 had been previously revascularised by Coronary Artery Bypass Grafting (CABG) and 48 patients had previous PCI. The remaining 411 patients had at least one vessel with a 50% luminal stenosis prior to their TAVI, of which the majority (n=179) had single-vessel disease and n=116 had either LMS or three-vessel disease (Table able1). One patient with stenosis in the proximal LAD disease was electively revascularised by PCI as part of their TAVI procedure for intrusive angina (following their diagnostic coronary angiogram).

Table 1
|
Summary of data from all patients (n=898) describing extent of CAD at the time of diagnostic angiography

Patient characteristics of those with and without CAD

A total of 488 patients did not have obstructive CAD at the time of the pre-TAVI coronary angiogram compared with 410 who did and are summarised in table 2. The groups were of comparable age (82.1 vs 87.8 years, p=0.094), body mass index (BMI) (28.3 kg/m2 vs 27.6 kg/m2, p=0.052), prevalence of concomitant pulmonary disease (33.2% vs 29.8%, p=0.27) and were predominantly of Caucasian ethnicity (96.9% vs 95.4%, p=0.22). The cohort with residual CAD had a greater proportion of males (62.0% vs 47.5%, p<0.005) and a greater proportion of patients with a smoking history (63.7% vs 57.2%, p=0.041). In contrast, those patients without CAD had a greater prevalence of neurological disease including a previous stroke/TIA (23.4% vs 16.1%, p=0.007). Importantly, there was no statistically significant difference in LV ejection fraction or diabetic status between the two groups. The group with residual CAD had significantly greater proportion of patients with a Clinical Frailty Score of 4 or greater (11.2% vs 19.5%, p<0.05).

Table 2
|
Patient characteristics as stratified by the presence or absence of residual CAD

Mortality outcomes

Of all patients undergoing TAVI during the study period (n=898), 298 patients (33.2%) met the primary outcome of all-cause death over 36 months of follow-up. One-third (32.9%) of patients without residual CAD experienced the primary outcome at 3 years as compared with 33.2% in those with residual CAD, of which the survival curves were not significantly different (p=0.91, Mantel-Cox). Of those patients with CAD (n=410), mortality was quantified in those with one-vessel (34.9%), two-vessel (29.3%) and three-vessel (35%) CAD (figure 1, p=0.2, Mantel-Cox). Patients were also stratified according to the presence or absence of LMS disease and did not demonstrate a significant difference in survival (p=0.67, Mantel-Cox). Sequential non-hierarchical univariate analysis was performed using 16 variables (included within the TAVI database) to see if they were associated with the primary outcomes of death within 3 years(table 3). The presence of extracardiac arteriopathy (p=0.001), previous cardiac surgery (p=0.022) and non-femoral access (p=0.006) were all associated with worse outcomes with respect to mortality. Multivariate analysis was performed using Cox proportional regressional model using these 16 variables (table 4). With this analysis, CAD was not associated with increased mortality following TAVI (HR 0.95, 0.75–1.20).

Figure 1
Request permissionrequest permission icon
Figure 1

Kaplan-Meier survival curves for mortality following transcatheter aortic valve implantation (TAVI). Graphs depict all cause survival with months of follow-up (x-axis) and % of patients alive (y-axis). (A) Represents all patients within the cohort; overall 66.7% of patients survived at 3 years. (B) Stratifies survival in those with (red) and without (blue) coronary artery disease (CAD). There is no significant difference between the two groups (p=0.91, Mantel-Cox). (C) Stratifies survival according to the extent of CAD including no CAD (blue), one-vessel (green), two-vessel (pink) and three-vessel disease (red). There is no significant difference between curves (p=0.727, Mantel-Cox). (D) Depicts mortality stratified by the presence (red) or absence (green) of left main stem (LMS) disease. There is no significant difference between curves (p=0.673, Mantel-Cox).

Table 3
|
Sequential, non-hierarchal univariate analysis of mortality outcomes and different patient factors
Table 4
|
Multivariate analysis using Cox proportional regressional modelling

Effect of CAD on readmission

Of all the patients that had a TAVI during the period of study, 355 patients (39.5%) had an unplanned re-presentation to hospital, for any reason, within 3 years of their index procedure. One hundred and eighty-one patients (37.1%) without CAD re-presented as compared with 169 patients (41.2%) who had CAD at the time of TAVI ((p=0.32, Mantel-Cox).

For cardiovascular readmissions (a composite of MI, stroke, admissions for heart failure (HF) or peripheral vascular disease), a total of 93 patients (10.4%) re-presented within 3 years. When stratified by the presence of residual CAD, 48 patients (9.8%) without residual CAD as compared with 45 (11.0%) in the arm with CAD had an unplanned admission with this composite endpoint (p=0.39 Mantel-Cox). When examining admissions for ACS only, n=10 patients (2.0%) re-presented in the no CAD arm as compared with n=13 (3.3%) in the CAD arm (p=0.32, Mantel-Cox)(figure 2).

Figure 2
Request permissionrequest permission icon
Figure 2

Kaplan-Meier curves of re-admissions following transcatheter aortic valve implantation (TAVI). Graph depicts readmission events over 3 years of follow-up (x-axis) and the % of patients that were readmitted (y-axis). (A) Depicts readmission for any cause stratified according to those with (red) and without (blue) coronary artery disease (CAD). There was no significant difference between the curves (p=0.23, Mantel-Cox). (B) Shows readmission for a composite of myocardial infarction, heart failure, stroke or peripheral vascular disease. The inset is the same graph but magnified to demonstrate the separation of the curves. There was no overall significant difference between the curves (p=0.11, Mantel-Cox). (C) Depicts readmission for acute coronary syndrome in those with (red) and without (blue) prior CAD. The inset is magnified to demonstrate better the separation of curves. Overall, readmissions for acute coronary syndrome were greater in the group with CAD than without but this was not statistically significant (p=0.32, Mantel-Cox).

Effect of CAD on symptom control

Both angina and breathlessness were assessed in patients undergoing TAVI. Most patients did not report angina (median=0); however, in those patients that did have angina, there was a significant reduction following TAVI in both patients with and without CAD (p<0.001, Mann-Whitney) (figure 3). The median NYHA breathlessness was 3 before TAVI which reduced to 1 following the procedure. This reduction was statistically significant in both groups of patients (p<0.01, Mann-Whitney).

Figure 3
Request permissionrequest permission icon
Figure 3

Angina and breathlessness scores before and after transcatheter aortic valve implantation (TAVI). Graph on the left depicts mean CCS anginal scores, whereas the graph on the right depicts mean NYHA breathlessness scores. Error bars indicate SD. The graphs are further divided into those patients who had no residual CAD (blue bars) and those with residual coronary artery disease (CAD) (red bars).

Discussion

This large observational study was designed to examine the association of concomitant CAD on mortality and readmission outcomes in patients who underwent a TAVI procedure. This all-comer study included patients over a 7-year period with a 3-year follow-up and therefore is representative of daily practice within a tertiary cardiology centre. From the analysis, there is no correlation between concomitant CAD at the time of TAVI with either mortality or readmission of patients following their procedure in a centre in which it is common practice not to routinely revascularise concomitant CAD. Furthermore, the presence of left main CAD and multivessel disease were also not correlated with these outcomes.

This finding is consistent with the evidence for stable CAD revascularisation outside of the context of severe AS, where there has not been a demonstrable mortality benefit.12 13 However, most contemporary guidance advocate revascularisation for LMS or multivessel disease in chronic stable angina on both symptomatic and prognostic grounds.14 The finding that this pattern of CAD does not have an effect on mortality over a 3-year follow-up period in our study might reflect frailty and multimorbidity in this patient group; something which may not necessarily be reflected in the trials of intervention in chronic stable angina. Furthermore, it is important to note that even for treated severe AS, mortality at 1 year was approximately 20% in this cohort. This may ‘drown out’ any differences in mortality related to CAD. This study did not examine the cause of death which might have demonstrated differential cardiovascular outcomes, but we postulate that absolute differences would be small.

In addition to the mortality benefits in severe AS, TAVI also serves as an effective treatment of symptoms with significant reductions in NYHA class of breathlessness.15 In this cohort, there were significant reductions in both CCS angina and NYHA breathlessness scores after TAVI which were independent of the presence of concomitant CAD. This suggests that the resolution of aortic valve pathology is the main driver of symptom improvement in this cohort rather than coexisting CAD.

Coexisting CAD was also demonstrated not to be correlated significantly with all-cause readmissions following TAVI. Readmission is an important clinical outcome in this cohort that tends to be multimorbid and frail. Arguably, readmissions are of as much significance to patients as any mortality benefits from TAVI. The Kaplan-Meier curves for readmission diverge at 18 months of follow-up with a numerically higher number of patients being readmitted who had concomitant CAD. The Kaplan-Meier survival curves diverge earlier when restricted to those readmissions for a composite cardiovascular endpoint including MI, but still not statistically significant. Of those patients readmitted with ACS, only limited numbers were treated by PCI and the remainder were managed medically. This result should be interpreted in the context of bias as it requires accurate coding and the knowledge of pre-existing coronary disease may bias a physician’s decision to admit, treat and ultimately code a readmission as an ACS.

Some argue that angioplasty after the fact is rendered more difficult with a TAVI prosthesis and therefore would support revascularisation for severe lesions on a prophylactic basis. However, our data suggest otherwise. Firstly, only a very small proportion of patients re-present with an ACS following a TAVI. Secondly, our data have shown that of those who had PCI for ACS after TAVI, only one patient had the culprit lesion identified on the pre-TAVI angiogram. This suggests that the presence of CAD on angiography is a general marker of risk in this cohort rather than a robust mechanism by which to identify future culprit coronary lesions.

The most important limitation of this study is that it is observational and is prone to the pitfalls associated with non-randomised data, such as inclusion, interpretation bias and unmeasured confounding. In addition, this registry employed a 50% angiographic stenosis, to be deemed a significant coronary artery lesion. This is an arbitrary cut-off based on a visual angiographic assessment in two or more orthogonal views. It does not account for functional impact of these lesions with physiological indices such as FFR/iFR. It also does not account for description of plaque morphology by intracoronary imaging both of which are potential predictors of future cardiovascular events.16 Having said this, physiological indices are difficult to interpret in the context of severe AS, with different thresholds suggested.17 Our findings are contrasted with other registry data in which concomitant CAD was associated with worse outcomes at 5 years of follow-up.18 This was particularly true of those with more complex CAD (Syntax score >22). However, even though up to 30% of patients underwent pre-TAVI PCI, there was no effect on outcomes. Furthermore, at 3 years of follow-up, there were no significant differences in all-cause mortality in patients with and without CAD.

While the pitfalls of observational data are evident, RCTs for this cohort of patients have proved difficult to recruit to. The largest RCT, ACTIVATION,11 failed to recruit a sufficient number of patients. The NOTION 3 study is currently running and will randomise patients with severe AS selected for TAVI and at least one coronary lesion of >90% stenosis or with an FFR <0.8.19 Again, questions will remain about the validity of physiology in this cohort, in addition to the difficulty in recruiting patients, as the authors of ACTIVATION found.

This large observational study demonstrates no association between concomitant CAD on either mortality or readmissions following TAVI. It suggests that having a strategy whereby coronary lesions are not routinely revascularised around the time of TAVI is generally safe. With the difficulty in performing RCTs in this area, it may be that findings from large cohorts, such as this, are sufficient to inform our practice of how to manage CAD in patients undergoing TAVI.

Article metrics
Altmetricopen-url
Dimensionsopen-url