Article Text
Abstract
Objective Patients with moderate aortic stenosis (AS) exhibit high morbidity and mortality. Limited evidence exists on the role of aortic valve replacement (AVR) in this patient population. To investigate the benefit of AVR in moderate AS on survival and left ventricular function.
Methods In a retrospective cohort study, patients with moderate AS between 2008 and 2016 were selected from the Cleveland Clinic echocardiography database and followed until 2018. Patients were classified as receiving AVR or managed medically (clinical surveillance). All-cause and cardiovascular mortality were assessed by survival analyses. Temporal haemodynamic and structural changes were assessed with longitudinal analyses using linear mixed effects models.
Results We included 1421 patients (mean age, 75.3±5.4 years and 39.9% women) followed over a median duration of 6 years. Patients in the AVR group had lower risk of all-cause (adjusted HR (aHR)=0.51, 95% CI: 0.34 to 0.77; p=0.001) and cardiovascular mortality (aHR=0.50, 95% CI: 0.31 to 0.80; p=0.004) compared with those in the clinical surveillance group irrespective of sex, receipt of other open-heart surgeries and underlying malignancy. These findings were seen only in those with preserved left ventricular ejection fraction (LVEF) ≥50%. Further, patients in the AVR group had a significant trend towards an increase in LVEF and a decrease in right ventricular systolic pressure compared with those in the clinical surveillance group.
Conclusions In patients with moderate AS, AVR was associated with favourable clinical outcomes and left ventricular remodelling.
- aortic valve stenosis
- transcatheter aortic valve replacement
- aortic diseases
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Moderate aortic stenosis (AS) is associated with high morbidity and mortality, yet the role of aortic valve replacement (AVR) in managing this condition remains unclear.
WHAT THIS STUDY ADDS
Patients with moderate AS who underwent AVR had improved long-term survival and left ventricular function compared with clinical surveillance alone. AVR was associated with lower risks of all-cause and cardiovascular mortality, especially in those with preserved left ventricular ejection fraction and smaller aortic valve area.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
It would highlight the need for prospective studies and randomised controlled trials to study the benefits of AVR in patients with moderate AS.
Introduction
Aortic stenosis (AS) is the most common valvular heart disease, with an increasing prevalence due to the ageing of the population.1 2 Aortic valve replacement (AVR) is recommended for patients with severe, symptomatic AS or when there is evidence of left ventricular systolic dysfunction.3 4 However, for patients with moderate AS, these indications are not well-established, with current guidelines recommending surveillance every 1–2 years with deferring AVR until AS becomes severe.3 4 Surgical AVR (SAVR) may also be considered in patients undergoing concomitant open-heart surgeries for other indications. Transcatheter aortic valve replacement (TAVR) is currently not indicated for moderate AS.3–5 However, recent advancements in surgical and catheter-based aortic valve interventions have resulted in improved procedural outcomes, with low mortality rates for both TAVR and SAVR.6–8
Recent data consistently show that moderate AS is associated with a high risk of cardiovascular and all-cause mortality.9–12 These findings highlight the need to investigate different strategies for the management of patients with moderate AS and the potential for earlier intervention. However, the benefits of AVR in patients with moderate AS are not yet fully understood mainly due to a paucity of published studies on the benefit of intervention in moderate AS, many of which are limited by small sample sizes and insufficient follow-up times.13–15 We aimed to compare the long-term clinical outcomes and echocardiographic changes in patients with moderate AS who underwent AVR versus clinical surveillance using real-world data from a large and contemporary cohort of patients.
Methods
Study population
Patients aged 65 years or greater with native moderate native AS, defined as aortic valve area (AVA) between 1.0 and 1.5 cm2 at index echocardiogram, were identified from the Cleveland Clinic Echocardiography Database from January 2008 to June 2018. Using AVA for determining the severity of AS was preferred over transvalvular gradients to avoid misclassifying low-flow, low-gradient severe AS as moderate AS. All included patients underwent at least two echocardiograms. We included data from all available echocardiograms for a given patient over the study duration. Patients with bicuspid aortic valve (n=112), left ventricular ejection fraction (LVEF) <20% (n=20) or without an echocardiogram at least 2 years from the index imaging were excluded. Figure 1 highlights the flowchart for patient selection.
Study groups
We had two distinct comparisons for the purposes of this study: patients with moderate AS who did not undergo AVR during the follow-up period, the clinical surveillance group, and patients with moderate AS who underwent SAVR or TAVR during the follow-up period, the AVR group. We compared patients who did versus did not have AVR at a time when they had moderate severity of their AS.
Study variables and assessment of aortic valve disease
Baseline patient characteristics including demographics, comorbidities, medications and laboratory data were obtained from the electronic medical records (EMRs) based on a comprehensive chart review of data at or nearest to the initial echocardiogram date. Further, a manual review of EMRs by practicing physicians was done to ascertain receipt of SAVR or TAVR, indications for surgery and functional status through NYHA classification. Baseline characteristics and echocardiographic data for patients in the clinical surveillance group were recorded at the time of the first echocardiogram and those in the AVR group were recorded at the time of surgery. Details on the measurements of all echocardiographic parameters obtained are available in the online supplemental material. Echocardiographic and Doppler measurements were obtained by an experienced sonographer and adjudicated by an expert, board-certified echocardiogram reader according to established guidelines.3
Supplemental material
Study outcomes
Our study outcomes were all-cause mortality and cardiovascular mortality, which were ascertained by practicing physicians through a comprehensive manual review of the EMR for all included patients. Cardiovascular mortality was defined as death attributable to myocardial ischemia and infarction, heart failure, cardiac arrest because of other or unknown cause, or cerebrovascular accident. We also analysed the relative longitudinal changes over time in echocardiographic parameters of valvular and left ventricular dimensions and hemodynamics between the two study groups.
Statistical analysis
Baseline characteristics were compared between study groups using a two-sided Student’s t-test or Mann-Whitney U test for continuous variables and analysis of variance for categorical variables. Continuous variables are represented as mean (SD) or median (IQR), and categorical variables are reported as proportions.
The associations with the primary outcomes were assessed through survival analyses using the Kaplan-Meier non-parametric method. To account for differences in baseline characteristics and calculate survival estimates, we used a multivariable Cox proportional-hazard model for all-cause mortality and competing risks regression using Fine and Gray’s proportional subhazards model for cardiovascular mortality with all-cause mortality as a competing endpoint; both models were adjusted for demographics and baseline clinical and echocardiographic features known to be correlated with clinical outcomes in AS.16 Stratified survival analyses were done by stratifying the population according to sex and baseline or preoperative LVEF (for AVR group) less than 50% or ≥50%, and p values for interaction were calculated. Sensitivity analyses of the primary outcomes were performed by (1) excluding patients with underlying active malignancy, defined as having a diagnosis of any malignancy with active chemotherapy or radiation therapy, (2) excluding surgical AVR from the AVR group (ie, patients who underwent TAVR only), (3) excluding patients who underwent concomitant open-heart surgeries in the AVR group and (4) by stratifying outcomes by presence or absence of symptoms as an indication for AVR.
To address potential biases, notably immortal time bias, our analysis approach accounted for the specific follow-up periods for both groups. For the clinical surveillance group, follow-up started from the time of the first echocardiogram to the last date of follow-up or until an event occurred. In the AVR group, the follow-up was initiated from the time of AVR surgery and continued until the last date of follow-up or an event, thereby excluding the period before AVR from the survival analysis. The duration of follow-up, which is reported in table 1, was adjusted for in the survival analysis.
Additionally, we repeated primary outcome analyses using a 2:1 propensity-matched group of patients who underwent AVR or clinical surveillance using the greedy matching strategy. A propensity score for a patient in the AVR group was considered matched to the closest patient in the clinical surveillance group having a propensity score within a difference of 0.1. The method was repeated until all patients in the AVR group were matched or all propensity scores deviated by more than 0.1 between the groups. Propensity matching was assessed by determining covariate balance as measured by standardised mean difference in the selected variables between groups before and after propensity matching. The results of these comparisons demonstrated that adequate propensity matching was achieved between both groups (online supplemental figure 1).
Supplemental material
Echocardiographic data at varying time intervals after the index echocardiogram were included in the longitudinal analyses of the relative echocardiographic changes over time between both groups. For patients in the AVR group, we only included those who had follow-up echocardiograms available post-AVR (n=86). Linear regression plots were done using means of echocardiographic measures on follow-up echocardiograms. We used linear mixed effect models with random intercept and interaction term with time to estimate differences in the temporal changes in individual echocardiographic parameters, including AVA, mean gradient (MG), transvalvular maximum velocity (Vmax), LVEF, right ventricular systolic pressure (RVSP), left ventricular mass index, left ventricular end-diastolic diameter (LVEDD) and E/e′ velocity across the follow-up period. Beta coefficients were calculated to estimate the between-subjects effect of AVR on echocardiographic parameters in the linear mixed effects model. All longitudinal analyses were stratified by AVR status and performed after adjusting for potential confounders including demographics, comorbidities, medications, as well as baseline echocardiographic data. Statistical significance was defined by p value <0.05. All analyses were conducted using STATA V.13.0 (StataCorp) and R studio (V.1.3.1073).
Results
Baseline characteristics
Table 1 demonstrates the baseline demographic, clinical and echocardiographic characteristics of the study population. Among 1421 included patients, the mean age was 75.3±5.4 years, 568 (39.9%) were women and 1268 (89.2%) were white. The median (IQR) duration of follow-up was 6.1 (4.0–8.2) years. Among patients in the AVR group, the majority had surgical AVR (73.6%), and the most common indication for surgery was symptomatic disease (64%), defined as having symptoms of heart failure, followed by undergoing cardiac surgery for other indications (26%), and rapid disease progression on echocardiogram, which included a rapid increase in transvalvular gradients and left ventricular systolic dysfunction (10%). Those who had other open-heart surgeries with AVR most commonly had included coronary bypass graft (19.1%), mitral valve surgery (6.7%), tricuspid valve surgery (4%) and aortic root surgery (9.6%). Further, patients in the AVR group had higher NYHA class preoperatively compared with those in the clinical surveillance group at baseline. Additionally, patients in the clinical surveillance group had significantly higher prevalence of hypertension (83.4 vs 72.2%, p<0.001), hyperlipidemia (77.1 vs 66.9%, p<0.0001), chronic kidney disease (15.2 vs 9.0%, p=0.006) but lower prevalence of coronary artery disease (50.4 vs 60.2%, p=0.005) and malignancy (7.7 vs 11.4%, p=0.041) and lower median serum N-terminal pro b-type natriuretic peptide (NTproBNP) (716 vs 2492 pg/mL, p=0.007). No difference was noted in the prevalence of diabetes, atrial fibrillation at baseline between the two groups. In terms of baseline echocardiographic data, patients in the clinical surveillance group had significantly higher median indexed AVA (0.72 vs 0.66 mm2/m2, p=0.004) and mean RVSP (36.5±12.7 vs 33.7±10.7 mm Hg, p=0.035), and mean tricuspid regurgitant peak velocity (2.74±5.12 vs 2.61±5.51, p=0.02), but lower means of MG (25.1±7.8 vs 29.8±9.2 mm Hg, p<0.001), Vmax (3.1±0.6 vs 3.5±0.7 cm/s, p<0.001) and LVEF (50.6±10.8 vs 57.2±11.1, p=0.04).
Primary outcomes
Overall, there were 363 (25.5%) deaths, among which 266 (18.7%) were of cardiovascular reasons. The death rates were 6.8 versus 2.4 per 1000 person-years among patients in the clinical surveillance and AVR groups, respectively (log rank p<0.001). In multivariable-adjusted regression analyses, patients in the AVR group had a lower risk of all-cause mortality (adjusted HR (aHR)=0.51, 95% CI: 0.34 to 0.77; p=0.001) and cardiovascular mortality (aHR=0.50, 95% CI: 0.31 to 0.80; p=0.004). In sensitivity analyses including a 2:1 propensity-matched group of patients who had clinical surveillance or AVR (figure 2), patients in the AVR group had lower risk of all-cause mortality (Log rank p=0.0003). When looking at the group of patients who underwent TAVR without open-heart surgery (n=79), they still had lower risk of all-cause mortality at 10 years (aHR=0.45, 95% CI: 0.22 to 0.91; p=0.026) (online supplemental figure 2). These findings remained significant when stratifying by sex (p for interaction=0.23) (figure 3). However, when stratifying by LVEF (figure 4), only patients in the AVR group with LVEF ≥50% (n=1198, p for interaction=0.011) had mortality benefits. The benefit of AVR on all-cause mortality was apparent when excluding patients who underwent concomitant open-heart surgery (online supplemental figure 3A) and those with underlying malignancy (online supplemental figure 3B). When stratifying by presence or absence of symptoms as an indication for AVR, the benefit of AVR on all-cause mortality was apparent in both groups (aHR1=0.57, 95% CI: 0.36 to 0.89; p=0.015 and aHR2=0.36, 95% CI: 0.19 to 0.69; p=0.002, respectively) (online supplemental figure 4).
Supplemental material
Supplemental material
Supplemental material
Longitudinal echocardiographic changes
Figure 5 reveals the temporal changes in select AS parameters across the study period in both study groups. Notably, patients in the AVR group who had follow-up echocardiograms postoperatively exhibited improvements in several echocardiographic parameters, including an increase in LVEF, decrease in LVEDD, RVSP and LV mass index and no reduction in AVA. In multivariable-adjusted linear mixed effects regression, those who underwent AVR had a significant trend toward higher LVEF (β=2.86, p=0.014) and had lower RVSP (β=−4.73, p=0.011) over time compared with those who were managed with clinical surveillance, who had continued decrease in LVEF and increase in RVSP over the follow-up period.
Discussion
In this large retrospective cohort study which included patients with moderate AS followed for a median period of 6 years, all-cause mortality and cardiovascular mortality were found to be significantly lower following AVR when compared with clinical surveillance. To our knowledge, this is one of the largest studies to provide real-world data comparing outcomes between intervention for moderate AS and clinical surveillance.
The interest in investigating whether earlier intervention in patients with AS stems from multiple bodies of evidence indicating that moderate AS is associated with a poorer prognosis than previously thought compared with those with milder forms of AS or not AS, irrespective of LVEF.9 10 In a recent study, Khan et al assessed clinical outcomes in 9133 patients with varying degrees of AS and heart failure with reduced ejection fraction (HFrEF) (n=374 patients with moderate AS). Over a median follow-up of 3.1 years, moderate AS was independently and significantly associated with HF hospitalisation and all-cause mortality in a 1:1 propensity score-matched cohort of patients with moderate AS and no AS. Notably, those with moderate who had SAVR/TAVR (n=46) had improved survival (aHR: 0.60, 95% CI: 0.36 to 0.99; p<0.05).11 In a study looking at a population of 25 827 patients with mild to severe AS (3315 with moderate AS), these had an increased risk of long-term morality when compared with an age-adjusted and sex-adjusted control group without AS (aHR: 1.44 to 2.09; p<0.001 for all comparisons). The 5-year mortality for patients with moderate AS was 56%, indicating poor survival rates in this group of patients.12 The magnitude of benefit to AVR in moderate AS in prior observational data is similar to our findings, where those in the AVR group had ≤50% lower risk of all-cause and cardiovascular mortality.13 14 Additionally, Jean et al showed that in a matched group of 262 patients with moderate AS compared with 262 matched controls with no AS, all-cause mortality and HF hospitalisation were significantly higher in those with moderate AS (HR: 2.34, 95% CI: 1. 72 to 3.21; p<0.0001). However, those who underwent AVR (16.7%) benefited from significantly improved survival after a median follow-up of 10.9 months (HR: 0.59, 95% CI: 0.35 to 0.98; p=0.04).15 Our study adds to the above published data in having a large and more contemporary cohort with long duration and follow-up and including multiple sensitivity analyses to ascertain our findings of relative survival difference between the two groups. The benefit of AVR was notable only in those with LVEF ≥50%; though there was signal for benefit in those with LVEF <50%, statistical significance could have been limited by small sample size of those with low LVEF in the AVR group. Another explanation is that those with low LVEF already have advanced cardiac structural and haemodynamic damage so that the benefit of AVR on cardiac remodelling and function could be limited, and this was seen in reports on patients undergoing AVR with severe AS.17
Poor prognosis in moderate AS can be explained by adverse LV remodelling including diastolic dysfunction, fibrosis and hypertrophy with possible left atrial enlargement and pulmonary hypertension.18–21 As illustrated in figure 5, when comparing follow-up echocardiogram data post-AVR versus clinical surveillance, we found that AVR was associated with a significant trend toward improvement of LVEF. Moreover, AVR was associated with a significant lower RVSP indicating improved right-sided pressures. The low number of patients with ≥2 follow-up echocardiograms in the AVR group could have impacted some of the trends seen in echocardiographic parameters such as peak velocities and mean gradients. Furthermore, keeping in mind the possibility of having low-gradient AS in the AVR group, changes in gradients in this group might not be as apparent as changes in valve area of LVEF.
When comparing TAVR to clinical surveillance in matched controls, we also reported a significant lower incidence of all-cause mortality (aHR=0.45, 95% CI: 0.22 to 0.91; p=0.026) (online supplemental figure 2). Many randomised controlled trials (RCTs) and meta-analyses compared outcomes between TAVR and SAVR in severe AS high-risk, intermediate-risk and low-risk patients.22–24 Importantly, when comparing outcomes in patients with moderate AS and HFrEF to a 1:1 matched group with heart failure without AS, Jean et al found that patients who underwent TAVR (n=15) had significantly improved survival (HR: 0.43, 95% CI: 0.18 to 1.00; p=0.05) whereas patients who had SAVR (n=29) did not (p=0.92).15 We found evidence for possible benefit in both TAVR and SAVR approaches in the AVR group, and the difference compared with the above study is likely due to differences in sample size and surgical outcomes at our institution. The clinical benefit of TAVR in moderate AS is currently investigated in the RCTs TAVR-UNLOAD (NCT02661451),25 PROGRESS (NCT04889872)26 and Evolut EXPAND TAVR II (NCT05149755).27
In light of the potential need for future redo AVR when treating AS earlier in the disease course, it is crucial for operators to carefully evaluate root anatomy, type and size of surgical aortic valve replaced and the feasibility of future needed valve-in-valve TAVR prior to intervention.
Study limitations
Our analysis may be influenced by biases that are inherent in a retrospective study design including the inability to adjust for unmeasured confounders and completely adjust for the baseline difference in clinical characteristics between the two groups at baseline. Since the selection of the treatment regimen can depend on the severity of the patient’s condition, susceptibility bias (confounding by indication) and ascertainment bias can result and the differences in baseline characteristics between the two groups may confound the results. It is also difficult to capture the granular data on the indications for AVR as it frequently involves complex clinical decision making between operators and patients. The majority of AVR in our cohort were surgical, which reflects the study period during which the transcatheter approach was not as widely used as present. Finally, patients are included from a single centre, and though generalisability can be limited, our institution is a quaternary referral centre for patients from many different parts of the country.
In conclusion, AVR appears to be a safe and promising option in patients with moderate AS when compared with conservative management through clinical surveillance. AVR was associated with a reduction in all-cause mortality and cardiovascular mortality notably in patients with preserved LVEF. Results from ongoing trials in that field are necessary to further inform us on the clinical utility and ideal timing and candidates for intervention versus medical management in this vulnerable patient population.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by the Institutional Review Board of the Cleveland Clinic, with waived informed consent.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
X @GrantReedMD
EHH and OB contributed equally.
Contributors EHH and OB equally contributed with concept, design, data acquisition, analysis and interpretation, writing manuscript drafts and revisions. JK, HL and WS contributed with writing manuscript, data acquisition and data interpretation.RB contributed with data acquisition and analysis. SCH, RP, GWR, AK, LGS contributed with data interpretation and critical revisions to manuscript draft and final revisions. SK contributed with concept, design, data analysis and interpretation, writing manuscript drafts, critical revisions and final approval.
Funding This work was supported by unrestricted philanthropic support to the Cleveland Clinic Heart, Vascular and Thoracic Institute.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.