Author + information
- Received April 13, 2015
- Revision received September 7, 2015
- Accepted September 10, 2015
- Published online July 1, 2016.
- Maxime Berthelot-Richer, MD, MSc,
- Philippe Pibarot, DVM, PhD,
- Romain Capoulade, PhD,
- Jean G. Dumesnil, MD,
- Abdellaziz Dahou, MD,
- Christophe Thebault, MD,
- Florent Le Ven, MD and
- Marie-Annick Clavel, DVM, PhD∗ ()
- Institut Universitaire de Cardiologie et de Pneumologie de Québec/Québec Heart & Lung Institute, Laval University, Québec, Canada
- ↵∗Reprint requests and correspondence:
Dr. Marie-Annick Clavel, Institut Universitaire de Cardiologie et de Pneumologie de Québec, 2725 Chemin Sainte-Foy, #A-2047, Québec (QC) G1V-4G5, Canada.
Objectives This study sought to assess the survival benefit associated with aortic valve replacement (AVR) according to different strata of echocardiographic parameters of aortic stenosis (AS) severity, and especially in patients with an aortic valve area (AVA) comprised between 0.8 cm2 and 1 cm2.
Background Discordant findings between AVA (≤1.0 cm2) and mean gradient (MG) (<40 mm Hg) raise uncertainty regarding the actual severity of AS. Some studies suggested that the AVA threshold value to define severe AS should be decreased to 0.8 cm2 to reconcile these discordances.
Methods A total of 1,710 patients with documented moderate to severe AS by Doppler echocardiography were separated into 4 strata of AS severity based alternatively on AVA, indexed AVA, MG, or peak aortic jet velocity (Vpeak). We compared the survival rates of medically versus surgically treated patients. To eliminate covariate differences that may lead to biased estimates of treatment effect, a propensity matching with a greedy 5-to-1 digit-matching algorithm was used.
Results Mean AVA was 0.9 ± 0.3 cm2, mean MG 33 ± 18 mm Hg, and mean Vpeak 3.6 ± 0.9 m/s. A total of 1,030 (60%) patients underwent AVR within 3 months following echocardiographic evaluation. During a mean follow-up of 4.4 ± 3.0 years there were 469 deaths. Patients with an AVA between 0.8 cm2 and 1.0 cm2 had a significant observed survival benefit with AVR (hazard ratio: 0.37 [95% confidence interval: 0.21 to 0.63]; p = 0.0002). AVR was also associated with improved survival in patients with MG between 25 mm Hg and 40 mm Hg or Vpeak between 3 m/s and 4 m/s, but only in patients with concomitant AVA ≤1 cm2 (p = 0.001 vs. p = 0.46 in patients with AVA >1 cm2).
Conclusions These results do not support decreasing the AVA threshold value for severity to 0.8 cm2 and they confirm that AVR is associated with improved survival in a substantial number of patients with discordant aortic grading.
According to American and European guidelines (1–3), severe aortic stenosis (AS) is defined by several echocardiographic criteria including aortic valve area (AVA) ≤1 cm2, indexed AVA (AVAi) ≤0.6 cm2/m2, mean gradient (MG) ≥40 mm Hg, and peak aortic jet velocity (Vpeak) ≥4 m/s. However, up to 30% of patients with AS present with discordant echocardiographic parameters of AS severity. The most frequent discordant grading pattern is an AVA ≤1 cm2 and/or AVAi ≤0.6 cm2/m2 (indicating a severe disease) with an MG <40 mm Hg and/or Vpeak <4 m/s (rather consistent with a moderate disease) (4–7). In that situation, uncertainty remains regarding the actual AS severity and whether or not to refer the symptomatic patient to aortic valve replacement (AVR).
Although discordance (i.e., small AVA/AVAi but low MG/Vpeak) in severe low-flow AS, with or without preserved ejection fraction, is accepted, the possibility of discordance in severe AS with normal flow is highly controverted. Nevertheless, discordant echocardiographic grading of AS severity in a normal flow patient may be related to multiple factors including measurement errors (3,8), small body size, reduced arterial compliance (7,9), or inherent discordance between AVA/AVAi and MG/Vpeak cutpoints used for the definition of severe AS (5). With regards to the latter factor, it has been demonstrated that, from a hemodynamic standpoint, an MG of 40 mm Hg does not correspond to an AVA of 1.0 cm2 but rather to an AVA of 0.8 cm2 (5). Hence, to reconcile the AVA/AVAi and MG/Vpeak criteria to define severe AS, some authors have proposed to lower the AVA cutpoint from 1.0 to 0.8 cm2 (5,10,11). However, survival studies have reported that an AVA ≤1.0 cm2 is the most powerful and sensitive predictor of death (12). Furthermore, AVR is associated with better survival in symptomatic patients with AVA ≤1.0 cm2 and low MG (<40 mm Hg) (6,13,14). However, no previous study has assessed the benefit of AVR according to the different echocardiographic criteria proposed in the literature. The primary objective of this study was thus to assess the survival benefit associated with surgical AVR according to different strata of echocardiographic parameters of AS severity. The secondary objective was to assess the survival benefit associated with AVR in patients with discordant aortic grading and normal flow.
All adult patients who underwent a comprehensive Doppler echocardiographic examination for moderate to severe aortic valve stenosis (AVA ≤1.5 cm2) at our center between 2000 and 2012 were eligible for this study. Patients with a history of rheumatic valve disease or endocarditis, life-threatening comorbid conditions at diagnosis, more than mild aortic regurgitation (i.e., vena contracta ≥3 mm, regurgitant volume ≥30 cc, regurgitant fraction ≥30%, and/or regurgitant effective orifice area ≥0.1 cm2) (15), mild mitral stenosis (mitral valve area ≤1.5 cm2 or MG ≥5 mm Hg) (3), and/or mild mitral regurgitation (vena contracta ≥3 mm, regurgitant volume ≥30 cc, regurgitant fraction ≥30% or regurgitant effective orifice area ≥0.2 cm2) (16), or any other valve disease or prior valve replacement were excluded.
We also excluded patients who underwent any concomitant intervention except coronary artery bypass graft (CABG) at the time of AVR (including aortic root replacement, mitral annuloplasty, and so forth); and patients who underwent transcatheter AVR because the aim of this study was to assess the benefit of surgical AVR. Clinical, Doppler echocardiographic, and operative data were prospectively collected in consecutive patients and were retrospectively analyzed.
Doppler echocardiographic measurements
Left ventricular (LV) dimensions, mass, and left ventricular ejection fraction (LVEF) were measured according to recommendations of the American Society of Echocardiography (17). Measurement of the LV outflow tract, Vpeak, and time velocity integrals allowed calculation of stroke volume, MG by modified Bernoulli formula, and AVA by the continuity equation. AVA and stroke volume were indexed to body surface area. Normal LV function and flow was defined as LVEF ≥50% and stroke volume index >35 ml/m2. To minimize measurement error as a potential cause of discordance, we excluded suboptimal echocardiograms.
Therapeutic management and outcomes
All patients who underwent AVR within 3 months after baseline echocardiographic evaluation were classified in the AVR treatment group. In all patients of this group, the decision to refer the patient to AVR was decided by the treating physician before or at the time of echocardiographic evaluation. Patients who did not undergo AVR (n = 627) or who underwent AVR >3 months (n = 53 with medical mean follow-up of 4.4 ± 3.0 years) after echocardiographic evaluation were classified in the medical treatment group. In patients with delayed AVR (i.e., >3 months), only medical follow-up was taken into account (i.e., follow-up was censored at that time of AVR). Mortality data were obtained from the Institut National de la Statistique du Québec. The follow-up data were complete for all patients.
Results are expressed as mean ± SD or percentages. Differences between groups were analyzed with the use of the 2-sided Student t test for continuous variables, with the Wilcoxon rank sum test for ordinal variables and the chi-square test of Fisher exact test for categorical variables as appropriate.
We grouped patients into strata according to their levels of echocardiographic parameters. Numbers of strata were decided to have sufficient statistical power for the multivariate survival analysis. Each echocardiographic parameter (AVA, AVAi, MG, and Vpeak) was divided into 4 strata in the whole cohort analysis and 3 strata in the normal LV function/flow patients’ subgroup analysis (Online Table SM1). The impact of AVR was assessed in each stratum with the use of 2 methods to adjust for selection bias: Cox hazard multivariate model and propensity greedy matching.
Survival analyses were performed with the use of Cox proportional hazards and were presented as hazard ratio (HR) and 95% confidence interval (CI). The clinically relevant variables and/or the variables associated with death in univariate analysis were included in multivariable analysis. All multivariable models were constructed with age, male sex, presence of symptoms, atrial fibrillation, coronary artery disease, diabetes, chronic kidney disease, chronic pulmonary disease, hypertension, LV mass index, and LVEF.
Then, propensity score match 1:1 was used to eliminate covariate differences that may lead to biased estimates of treatment effect. The following relevant variables were included in the propensity score: age, male sex, presence of symptoms, atrial fibrillation, coronary artery disease, diabetes, chronic obstructive pulmonary disease, chronic kidney disease, dyslipidemia, hypertension, body surface area, body mass index, AVA, LV mass index, mean aortic gradient, aortic regurgitation (none vs. mild), mitral stenosis (none vs. mild), mitral regurgitation (none vs. mild), and LVEF. C statistic for the propensity was 0.93 for predicting adjudication to the surgical or medical treatment group. Then patients with similar probability of AVR (i.e., propensity score) were matched using a greedy 5-to-1 digit-matching algorithm. Before analyzing survival, we assessed the success of the propensity score match by comparing the distribution of patient characteristics in the matched sample.
A total of 1,710 patients were included in this study. The mean AVA was 0.9 ± 0.3 cm2, AVAi 0.50 ± 0.17 cm2/m2, MG 33 ± 18 mm Hg, and Vpeak 3.6 ± 0.9 m/s (Table 1). There were several differences in the baseline characteristics between the surgical (n = 1,030) and medical (n = 680) groups (Table 1). More patients in the surgical group had AS-related symptoms (95% vs. 66%; p < 0.0001) and these patients had more severe stenosis (Table 1). Overall patients in the medical group were older, more likely to be female, had more chronic kidney disease, and a trend for more chronic obstructive pulmonary disease (Table 1). However, all differences in baseline characteristics were abolished by the propensity score.
Among the 1,710 patients, 940 presented with normal LV function and flow defined as LVEF ≥50% and stroke volume index >35 ml/m2. In this subcohort, the differences between the surgical and the medical group were comparable or lighter than in the whole cohort, and the propensity score match also eliminated these differences (Table 2).
Among patients with AVA ≤1 cm2, patients with low gradient and/or low flow were more prevalent in the medical treatment group (Online Table SM2). Among patients who underwent AVR, 503 (48%) had a concomitant CABG in the whole cohort and 258 (47%) in the normal LV function and flow group.
Relationship between AVA and MG/Vpeak
AVA/AVAi-MG/Vpeak correlations were good in all patients (all r2 ≥0.59; all p < 0.0001) (Online Figures SM1 to SM4). A total of 280 (29.8%) patients with normal ejection fraction and flow had discordant parameters of AS severity with an AVA ≤1.0 cm2 and a MG <40 mm Hg. A total of 247 (26.3%) had an AVA ≤1.0 cm2 and a Vpeak <4.0 m/s. Correspondence between AVA, AVAi, MG, and Vpeak thresholds to define patients with moderate and severe AS are presented in the online material (Online Results, Online Figures SM1 to SM4). To note, there was an excellent correlation between MG and Vpeak (r2 = 0.98; p < 0.0001) and the cutpoints for severity between the 2 parameters were concordant (Online Results, Online Figure SM5).
Observed survival benefit associated with AVR, separate analysis for each echocardiographic parameter
During a mean follow-up of 4.4 ± 3.0 years, there were 469 deaths. There was a median of 118 (range 56 to 192) deaths in each studied stratum in the whole cohort and 80 (range 44 to 99) deaths in the normal ejection fraction and flow cohort.
AVA and AVAi strata
Baseline characteristics according to AVA strata are presented in the Online Materials in Online Table SM3 for the whole cohort and Online Table SM4 for the normal LVEF and normal flow cohort. As expected, patients with more severe AS were older and had higher prevalence of symptoms and comorbidities (Online Tables SM3 and SM4). However, these differences did not influence the results of survival analyses that were performed within each stratum. The year of inclusion had no significant effect on treatment referral or on survival (p = 0.31, p = 0.56, respectively).
In the whole cohort, multivariable and matched analysis showed that AVR was associated with improved survival in patients with an AVA ≤1.0 cm2 (i.e., in strata with AVA ≤0.6 cm2, AVA between 0.6 cm2 and 0.8 cm2, and AVA between 0.8 cm2 and 1.0 cm2) (Figure 1A). When the analysis was restricted to patients with normal LVEF and flow, the observed benefit of AVR on survival remained statistically significant in all strata with an AVA <1.0 cm2 (all p ≤ 0.01) (Figure 1B).
As shown in Figure 2, AVR was associated with improved survival early in the follow-up in patients with AVA between 0.8 cm2 and 1 cm2 of the whole cohort and of the normal LVEF and flow cohort. To further ascertain observed survival benefit associated with AVR in patients with an AVA between 0.8 cm2 and 1.0 cm2, patients who underwent a concomitant CABG were excluded and benefit of AVR was confirmed (matched HR: 0.31 [95% CI: 0.11 to 0.80]; p = 0.02 and adjusted HR: 0.36 [95% CI: 0.12 to 0.97]; p = 0.04).
Accordingly, AVR was associated with improved survival in patients with an AVAi ≤0.6 cm2/m2 in the whole cohort, in the subset of patients with normal LV function/flow and in the subset of patients who underwent an isolated AVR (all p ≤ 0.04) (Figure 3, Online Figure SM6).
MG and Vpeak strata
When the severity of the stenosis was analyzed by strata of MG, the observed benefit of AVR on survival was present in all strata where MG ≥25 mm Hg in the whole cohort and in the subset of patients with normal LVEF/flow (all p ≤ 0.02) (Figure 4). Of the 275 patients with MG between 25 mm Hg and 40 mm Hg, 54 (19.6%) had concordant moderate AS (AVA >1 cm2) and 221 (80.4%) discordant grading AS (AVA ≤1 cm2). The association between AVR and survival was significant only in the discordant patients (i.e., MG <40 mm Hg but AVA ≤1 cm2) (HR: 0.35 [95% CI: 0.19 to 0.64]; p = 0.001), whereas not in patients with concordant moderate AS (i.e., MG <40 mm Hg but AVA >1 cm2) (p = 0.46).
After exclusion of patients who underwent a concomitant CABG, patients with normal LVEF and flow and MG between 25 mm Hg and 40 mm Hg who underwent an isolated AVR (n = 83) had better survival than medically (n = 85) treated patients (matched HR: 0.34 [95% CI: 0.13 to 0.80]; p = 0.01 and adjusted HR: 0.39 [95% CI: 0.17 to 0.86]; p = 0.02).
The analysis of Vpeak was highly similar to MG, because of very good concordance between MG and Vpeak (Online Appendix, Online Figure SM7).
The main findings of this study were that patients with an AVA comprised between 0.8 cm2 and 1.0 cm2 exhibit a significant improvement in survival with isolated AVR; and a significant proportion of patients with normal flow and MG <40 mm Hg or Vpeak <4 m/s, that is, patients with discordant aortic grading (AVA ≤1 cm2 but MG <40 mm Hg or Vpeak <4 m/s), present an observed survival benefit associated with isolated AVR.
Our findings also corroborate previously published data showing that discordance between echocardiographic markers of AS severity is frequent (5,7,11,18). Indeed, more than 25% of patients with normal LV ejection fraction/flow included in the present study presented with a small AVA but with nonsevere range of MG/Vpeak. We also confirm that there is an intrinsic discordance (i.e., in normal LV function/flow patients) in grading criteria of AS severity, with an AVA of 1.0 cm2 or an AVAi of 0.6 cm2/m2 corresponding to a MG of around 25 mm Hg and a Vpeak around 3.0 m/s, whereas an MG of 40 mm Hg and Vpeak of 4.0 m/s correspond to an AVA of around 0.8 cm2 and an AVAi of around 0.45 cm2/m2.
To reconcile these discordant echocardiographic parameters, some authors have proposed to lower the AVA cutoff to 0.8 cm2 (5,10). Although such an approach would indeed help to reduce the proportion of patients with discordant AS grading, the validity of this lower AVA cutpoint to predict prognosis has not been validated. Furthermore, most studies that have established the prognostic value of the Doppler echocardiographic parameters of stenosis severity used AVR or a composite of death or AVR (which was mainly driven by AVR) as the primary endpoint (19–22). A bias referral could have occurred in these previous studies given that the treating physicians were not blinded to Doppler echocardiography findings and could thus have been more prone to refer patients with high gradients or velocities to surgery. Moreover, recent studies have shown that although high gradients/velocities are predictive of AVR, they are not necessarily predictive of survival (6,13,23,24). However, numerous studies analyzed impact of AVR in patients with a small AVA and low gradient with or without regards to flow pattern and all but 2 showed a beneficial impact of AVR on survival (6,12–14,18,24–28). Among these 2 studies, at least 1 had very limited statistical power (28,29).
In our large cohort of all-comer patients with AS, and in patients with normal LV function/flow, we found that AVR, with or without concomitant CABG, was associated with a lower mortality compared with conservative treatment when AVA is ≤1.0 cm2 or AVAi is ≤0.6 cm2/m2. In particular, patients with an AVA between 0.8 cm2 and 1.0 cm2 had a survival benefit associated with AVR. These findings do not support the suggestion of lowering the AVA cutpoint to 0.8 cm2. Moreover, AVR was associated with improved survival in patients with discordant aortic grading (AVA ≤1.0 cm2 but GM <40 mm Hg or Vmax <4 m/s).
Nevertheless, patients with discordant echocardiographic grading of AS severity represent a highly heterogeneous group including: 1) patients with nonsevere AS caused by measurement error or small body size; 2) patients with pseudo-severe AS caused by low flow; 3) patients with severe AS and underestimation of stenosis severity caused by low flow (i.e., true severe AS) or systemic arterial compliance; and 4) patients with discordant grading caused by the inherent “misalignment” of the AVA-gradient cutpoints proposed in the guidelines to define severe stenosis (7,30). Recent studies evaluating the actual severity of AS by stress echocardiography or computed tomography revealed that 50% to 70% of these patients with small AVA but a low gradient/velocity have severe stenosis (7,30). This may explain, at least in part, the strong observed survival benefit associated with AVR in the present study even for patients with gradients/velocities in the nonsevere range.
Discordant echocardiographic grading of AS stenosis severity becomes a highly challenging situation for the treating physician when the patient is symptomatic. Indeed, in such a situation, a severe stenosis would constitute a class I indication for AVR, whereas a nonsevere stenosis would rather be a contraindication for AVR unless another cardiac procedure is required (1,2). One of the strengths of this study is that it has been conducted in a large population of patients with AS, with 80% of them being symptomatic at the time of the echocardiographic evaluation, thus allowing enlightenment of this complex situation. The results of this study support the status quo with regard to cutpoints of AVA, AVAi MG, and Vpeak for the identification of severe AS. MG and Vpeak are specific markers of AS severity and thus when reached, AS could be considered as severe, whereas AVA and AVAi are sensitive markers and should not be neglected. Indeed, in case of discordant grading with small AVA/AVAi and low MG/Vpeak (with or without normal flow), additional diagnostic tests including more comprehensive resting echocardiography, stress echocardiography, and/or computed tomography may be required to determine stenosis severity and need for AVR.
Because this is not a randomized controlled trial, we can only describe an association between AVR and survival and not a causal relationship. Also, because of the retrospective analysis, unmeasured factors may have influenced patients’ therapeutic management. We minimized these biases with the use of propensity score, which was excellent to predict referral to surgery versus medical treatment (C statistic of 0.93). Yet, it is possible that unmeasured clinical variables may have introduced biases into the data analyses.
The statistical analyses are based on observational comparisons of a therapeutic benefit and this study has therefore inherent limitations related to its exploratory and nonrandomized design. Nonetheless, we believe this analysis would be the best doable analysis, because a randomized controlled trial would be nonethical.
In this large series of patients with moderate to severe AS, discordant echocardiographic grading of stenosis severity was encountered in 30% of patients with moderate to severe AS and normal ejection fraction/flow. Patients with an AVA comprised between 0.8 cm2 and 1.0 cm2 displayed a significant observed survival benefit associated with isolated AVR. Moreover, patients with discordant aortic grading, with an AVA ≤1.0 cm2 but MG or Vpeak in the nonsevere range (i.e., MG 25 mm Hg to 40 mm Hg or Vpeak 3 m/s to 4 m/s, respectively, exhibited a significant improved survival associated with AVR. Hence, our results do not support the previously proposed suggestion of lowering the AVA threshold for AS severity to 0.8 cm2. They rather support the status quo with the use of AVA ≤1.0 cm2 (and AVAi ≤0.6 cm2/m2) and MG ≥40 mm Hg (and Vpeak ≥4 m/s) as sensitive and specific markers of stenosis severity, respectively. Facing discordant echocardiographic parameters of AS severity, one should confirm the stenosis severity by a multimodality and multiparametric evaluation to assess the potential need of AVR.
COMPETENCY IN MEDICAL KNOWLEDGE: The findings of this study emphasize the importance of AVA calculation to assess AS severity. Moreover the threshold of 1.0 cm2 of AVA to define severe AS and potentially refer a patient to AVR is accurate and should not be lowered to 0.8 cm2 for AVA and gradient/velocity thresholds reconciliation.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Patients with low gradient/velocity and with normal LVEF and normal flow may benefit from AVR when AVA ≤1 cm2. Thus, AVR should not automatically be denied to a symptomatic patient with small AVA, low gradient/velocity, normal LVEF, and normal flow. In the presence of discordant echocardiographic markers of AS severity (i.e., low gradient/velocity and tight AVA) with normal LVEF and normal flow, individualized evaluation of AS severity should be performed with other diagnostic modalities, such as exercise stress echocardiography or computed tomography, to ensure that the patient is truly asymptomatic and the stenosis is truly moderate.
TRANSLATIONAL OUTLOOK: Additional studies are needed to validate the benefit of AVR according to aortic valve calcification as measured by computed tomography in this specific population of patients with small AVA, low gradient/velocity, normal LVEF, and normal flow.
For supplemental materials, results, tables, and figures, please see the online version of this article.
All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic stenosis
- aortic valve area
- indexed AVA
- aortic valve replacement
- coronary arteries bypass graft
- confidence interval
- left ventricular
- mean gradient
- peak aortic jet velocity
- Received April 13, 2015.
- Revision received September 7, 2015.
- Accepted September 10, 2015.
- 2016 American College of Cardiology Foundation
- Nishimura R.A.,
- Otto C.M.,
- Bonow R.O.,
- et al.
- Minners J.,
- Allgeier M.,
- Gohlke-Baerwolf C.,
- Kienzle R.P.,
- Neumann F.J.,
- Jander N.
- Belkin R.N.,
- Khalique O.,
- Aronow W.S.,
- Ahn C.,
- Sharma M.
- Clavel M.A.,
- Messika-Zeitoun D.,
- Pibarot P.,
- et al.
- Quinones M.A.,
- Otto C.M.,
- Stoddard M.,
- Waggoner A.,
- Zoghbi W.A.
- Kadem L.,
- Dumesnil J.G.,
- Rieu R.,
- Durand L.G.,
- Garcia D.,
- Pibarot P.
- Zoghbi W.A.
- Ozkan A.,
- Hachamovitch R.,
- Kapadia S.R.,
- Tuzcu E.M.,
- Marwick T.H.
- Otto C.M.,
- Burwash I.G.,
- Legget M.E.,
- et al.
- Pellikka P.A.,
- Sarano M.E.,
- Nishimura R.A.,
- et al.
- Hachicha Z.,
- Dumesnil J.G.,
- Bogaty P.,
- Pibarot P.
- Clavel M.A.,
- Dumesnil J.G.,
- Capoulade R.,
- Mathieu P.,
- Sénéchal M.,
- Pibarot P.
- Eleid M.F.,
- Sorajja P.,
- Michelena H.I.,
- Malouf J.F.,
- Scott C.G.,
- Pellikka P.A.
- Maes F.,
- Boulif J.,
- Pierard S.,
- et al.
- Tribouilloy C.,
- Rusinaru D.,
- Marechaux S.,
- et al.
- Pibarot P.,
- Clavel M.A.