Article Text

Original article
Right ventricular longitudinal strain for risk stratification in low-flow, low-gradient aortic stenosis with low ejection fraction
  1. Abdellaziz Dahou1,
  2. Marie-Annick Clavel1,
  3. Romain Capoulade1,
  4. Philipp Emanuel Bartko2,
  5. Julien Magne3,4,
  6. Gerald Mundigler2,
  7. Jutta Bergler-Klein2,
  8. Ian Burwash5,
  9. Julia Mascherbauer2,
  10. Henrique B Ribeiro1,
  11. Kim O'Connor1,
  12. Helmut Baumgartner6,
  13. Mario Sénéchal1,
  14. Jean G Dumesnil1,
  15. Raphael Rosenhek2,
  16. Patrick Mathieu1,
  17. Eric Larose1,
  18. Josep Rodés-Cabau1,
  19. Philippe Pibarot1
  1. 1Institut Universitaire de Cardiologie et de Pneumologie de Québec/Québec Heart & Lung Institute, Laval University, Québec City, Québec, Canada
  2. 2Department of Internal Medicine II, Division of Cardiology, Medical University of Vienna, Vienna, Austria
  3. 3Service Cardiologie, CHU Limoges, Hôpital Dupuytren, Limoges, France
  4. 4Faculté de médecine de Limoges, INSERM 1094, Limoges, France
  5. 5University of Ottawa Heart Institute, Ottawa, Ontario, Canada
  6. 6Division of Adult Congenital and Valvular Heart Disease, Department of Cardiovascular Medicine, University Hospital Muenster, Muenster, Germany
  1. Correspondence to Dr Philippe Pibarot, Institut Universitaire de Cardiologie et de Pneumologie de Québec, 2725 Chemin Sainte-Foy, Québec city, Québec, Canada G1V-4G5; philippe.pibarot{at}med.ulaval.ca

Abstract

Background Left ventricular global longitudinal strain (LVLS) is a powerful predictor of outcome in patients with low-flow, low-gradient aortic stenosis (LF-LG AS) and low LV ejection fraction (LVEF). However, the impact of right ventricular (RV) function on the outcome of these patients remains unknown.

Objectives The aim of this study was to examine the impact of RV function as evaluated by RV free wall longitudinal strain (RVLS) on mortality in patients with LF-LG AS and low LVEF.

Methods 211 patients with LF-LG AS (mean gradient <40 mm Hg and indexed aortic valve area (AVA) ≤0.6 cm2/m2) and low LVEF (≤40%)) were prospectively recruited in the True or Pseudo-severe Aortic Stenosis study. AS severity was assessed using the projected AVA (AVAproj) at normal flow rate. Among the 211 patients, 128 had RVLS measurement available at rest and were included in this analysis. RVLS measurement at dobutamine stress echocardiography (DSE) was available in 58 of the 128 patients.

Results Two-year survival was lower in patients with RVLS<|13|% (53%±9%) compared with those with RVLS>|13|% (69%±5%) (p=0.04). In multivariable Cox analysis stratified for the type of treatment (aortic valve replacement vs conservative) and adjusted for age, AS severity, previous myocardial infarction and LVLS, rest RVLS<|13|% (HR=2.70; 95% CI 1.19 to 6.11; p=0.018) was independently associated with all-cause mortality. RVLS had incremental prognostic value over baseline risk factors and LVLS (χ2=20.13 vs 13.56; p=0.01). Reduced stress RVLS was also associated with increased risk of mortality (stress RVLS<|14|%: HR=2.98; 95% CI 1.30 to 6.52; p=0.01). In multivariable Cox analysis, stress RVLS<|14|% remained independently associated with mortality (HR=2.94; 95% CI 1.23 to 7.02; p=0.015). After further adjustment for rest RVLS, stress RVLS<|14|% remained independently associated with mortality (HR=3.29; 95% CI 1.17 to 9.25; p=0.024), whereas rest RVLS was not (p>0.05).

Conclusions In this series of patients with LF-LG AS and low LVEF, reduced RVLS was independently associated with increased risk of mortality. Furthermore, stress RVLS provided incremental prognostic value beyond that obtained from rest RVLS. Thus, RVLS measurement at rest and at DSE may be helpful to enhance risk stratification in this high-risk population.

Trial registration number NCT01835028; Results.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Low-flow, low-gradient aortic stenosis (LF-LG AS) with reduced left ventricular ejection fraction (LVEF) is observed in approximately 5%–10% of patients with severe AS and it is classified as a D2 stage in the 2014 American College of Cardiology/American Heart Association (ACC/AHA) guidelines if severity is confirmed by dobutamine stress echocardiography (DSE).1 ,2 These patients generally have a poor prognosis when treated medically, but also a higher operative risk when undergoing surgical aortic valve replacement (AVR).3–10 Accurate assessment of the underlying severity of the valve stenosis and the degree of myocardial impairment are crucial for optimal risk stratification and therapeutic management in this challenging subset of patients.11

We previously reported that left ventricular global longitudinal strain (LVLS) is a powerful predictor of outcome in patients with LF-LG AS and low LVEF.12 However, the impact of right ventricular (RV) function on the outcome of these patients is unknown. RV longitudinal strain (RVLS) measured by speckle tracking has been shown to be useful for the assessment of RV function.13–16 The purpose of this study was to examine the impact of RV function as evaluated by RVLS on all-cause mortality in patients with LF-LG AS and low LVEF.

Methods

Study protocol

The protocol of the True or Pseudo-severe Aortic Stenosis (TOPAS) multicentre prospective observational study (https://clinicaltrials.gov; NCT 01835028) has been previously described in detail.17 Briefly, between July 2002 and March 2012, we recruited 211 patients with LF-LG AS (mean transvalvular gradient<40 mm Hg and aortic valve area (AVA) ≤1.2 cm2 or ≤0.6 cm2/m2) with reduced LVEF (≤40%). Patients were excluded if they had >mild concomitant aortic regurgitation, organic mitral regurgitation or mitral stenosis. The study was approved by the Institutional Review Board Committee of the participating centres and the subjects gave informed consent.

Clinical data

Clinical data included age, gender, height, weight, body surface area and systolic and diastolic blood pressure, New York Heart Association functional class, documented diagnosis of traditional cardiovascular risk factors including hypertension, diabetes, dyslipidaemia and smoking, chronic obstructive pulmonary disease, coronary artery disease (history of myocardial infarction (MI) or ≥50% coronary artery stenosis on coronary angiography) and logistic EuroSCORE. Medication was recorded at the time of echocardiography.

Doppler echocardiography

At entry in the study, all patients underwent a comprehensive echocardiography using commercially available equipment (IE33, Philips or Vivid 7, GE Healthcare) at rest and at DSE as previously described.17 ,18 The dobutamine infusion protocol consisted of 8 min increments of 2.5–5 µg/kg/min up to a maximum dosage of 20 µg/kg/min. Continuous-wave Doppler of the aortic valve velocity spectrum, as well as pulsed-wave Doppler of the LV outflow tract velocity spectrum, was recorded at rest and at each stage of the dobutamine protocol. LVEF was measured by the biplane Simpson method. AS severity was assessed using the projected AVA at a normal transvalvular flow rate (AVAproj) and was calculated by the simplified method.17 ,19: AVAProj= AVArest+(ΔAVA/ΔQ)×(250−Qrest), where Q is mean transvalvular flow rate, AVARest and QRest are the AVA and Q at rest, and ΔAVA and ΔQ are the absolute increases in AVA and Q during DSE. Other parameters of RV function including tricuspid annular plane systolic excursion (TAPSE) and tricuspid annulus peak systolic velocity were measured in a subset of 37 patients using the methods recommended in the American Society of Echocardiography guidelines.20 The severity of tricuspid regurgitation was assessed using an integrated approach and graded as none (0), mild (grade 1), mild-to-moderate (grade 2), moderate-to-severe (grade 3) and severe (grade 4) according to the current guidelines.21–23

Longitudinal strain measurements

LV and RV strain parameters were retrospectively measured in an echocardiography core laboratory using dedicated commercial software for speckle tracking imaging (2D Cardiac Performance Analysis, TomTec Imaging Systems, Munich, Germany). LVLS was measured from the four-chamber, two-chamber and three-chamber apical views as previously described.12

Longitudinal strain of the RV free wall (RVLS) was measured from the four-chamber view at rest and at peak DSE. RVLS was the average value of that of the three segments of the RV free wall. Rest RVLS was available in 128 (61%) of the 211 patients in the TOPAS study and stress RVLS in 58 (45%) of the 128 patients with rest RVLS available. The major reasons for missing RVLS were inadequate frame rate (<50 fps) and poor-quality images. RVLS data were expressed using absolute values (|%|). Reproducibility of RV strain measurements was evaluated with the intraclass correlation coefficient in 10 patients. Intraclass correlation coefficient was 0.95 for intraobserver and 0.84 for interobserver measurements, respectively.

Study endpoints

The primary endpoint for this study was all-cause mortality. The outcomes were updated by annual TOPAS visit follow-up, up to 5 years or telephone interviews after 5 years.

Statistical analysis

Results are expressed as mean ±SD or percentages. Differences between patients with RVLS>|13|% versus RVLS<|13|% or stress RVLS>|14|% versus stress RVLS<|14|% were compared with the use of the t test for continuous variables and the χ2 test or Fisher's exact test for categorical variables as appropriate. The |13|% and |14|% cut-point values corresponded to the lower quartiles of rest RVLS and stress RVLS, respectively. Correlation between rest RVLS and stress RVLS was assessed using the Pearson's product–moment correlation. Kaplan–Meier curves and log-rank test of the time-to-event data were used to assess the effect of RVLS>|13|% versus RVLS<|13|% and stress RVLS>|14|% versus stress RVLS<|14|% on the primary endpoint (ie, all-cause mortality). The effect of the clinical and Doppler-echocardiographic variables on survival was assessed with the use of Cox proportional hazard models. The proportional hazards assumption was assessed with the use of Schoenfeld residuals and all variables met the assumption. Variables with a p value <0.10 in univariable analysis and those with clinical relevance regardless of their level of significance (ie, age, AS severity) were incorporated in multivariable model. Further adjustment for echocardiographic parameters of LV systolic function (ie, LVEF, stroke volume index (SVi) or LVLS) was performed. All the multivariable models were stratified for the type of treatment (AVR vs conservative). The incremental predictive value of rest and stress RVLS was assessed using χ2 values from the models and by calculating the category-free Net Reclassification Improvement for 1-year mortality. A p value <0.05 was considered statistically significant. The p values were from two-sided test. Statistical analysis was performed using JMP V.12 (SAS, Cary, North Carolina, USA) and SPSS V.20 (V.20.0, SPSS, Chicago, Illinois, USA) software.

Results

Patient characteristics

There was not statistical difference in the baseline clinical and echocardiographic characteristics between patients with RVLS measure available (study subset; n=128) versus those without RVLS (excluded subset), except for the resting mean transvalvular gradient, which was higher (p=0.02) in the former subset (see online supplementary table S1). The other parameters of AS severity as well as the 2-year survival rate (65%±4% vs 62%±6%, respectively; p=0.66) were however similar in the two subsets.

Table 1 shows the baseline characteristics of the 128 patients included in the study. Eighty of these patients (63%) underwent AVR within 3 months following enrolment and the remaining 48 patients (37%) were treated conservatively.

Table 1

Baseline clinical and Doppler-echocardiographic characteristics of the study population

Average LVLS was |9.5|%±3.0%, and average RVLS was |17.4|%±5.5%. In the subset of patients with stress RVLS available (n=58), rest and stress values of RVLS were not statically different (|18.3|%±5.5% vs |18.1|%±6.0%; p=0.87). The prevalence of any RV dysfunction and severe RV dysfunction, defined as a RVLS<|20|% and <|13|%,24 ,25 respectively, was higher in the subset of patients with severe impairment of LV longitudinal function (LVLS<|9|%)12 than in those with mild/moderate impairment of LV function (any RV dysfunction: 76% vs 60% and severe RV dysfunction 38% vs 10%, p<0.05). Figure 1 shows the distribution of rest RVLS values according to the severity of LV dysfunction.

Figure 1

Distribution of RVLS according to the severity of left ventricular dysfunction. This figure shows the distribution of RVLS according to the absence/presence of severe LV longitudinal dysfunction (LVLS ≥ vs <|9|%). LV, left ventricular; LVLS, left ventricular global longitudinal strain; RVLS, right ventricular longitudinal strain.

Patients with rest RVLS <|13|% had similar baseline clinical characteristics compared with those with RVLS >|13|%, except for diabetes (50% vs 26%; p=0.01) and previous MI (59% vs 40%; p=0.04) (table 1). With regard to echocardiographic data, patients with RVLS<|13|% had lower SVi, LVEF and LVLS compared with those with RVLS>|13|% (table 1).

Rest RVLS correlated with TAPSE (r=0.66; p=0.0001), tricuspid annulus systolic velocity (r=0.59; p=0.0001), tricuspid annulus diameter (r=−0.31, p=0.0006), LVLS (r=0.37; p<0.0001) and LVEF (r=0.32; p=0.0003).

Survival according to rest RVLS

During a mean follow-up of 2.4±2.1 years, 59 (46%) patients died, including 31 deaths in the subset of patients treated conservatively and 28 deaths in the surgically treated patients. AVR was associated with reduced risk of mortality (HR=0.48; 95% CI 0.24 to 0.94; p=0.03). The 2-year survival was 69%±5% in patients with RVLS<|13|% compared with 53%±9% in those with RVLS<|13|% (p=0.04; figure 2A). Figure 2B shows the results after further dichotomisation for the type of treatment. RVLS<|13|% was associated with increased risk of mortality in the group managed conservatively (p=0.02) and there was trend towards increased risk of mortality in the group treated by AVR (p=0.14). There was, however, no significant interaction (p=0.47) between RVLS<|13|% and type of treatment with respect to impact on mortality.

Figure 2

Impact of rest RVLS on all-cause mortality. Panel (A) Shows the survival in patients with RVLS>|13|% versus those with RVLS<|13|%. The numbers at the bottom of the graph represent the number of patients at risk at each follow-up year. The p value is that of the log-rank test. Panel (B) shows the survival in patients with RVLS>|13|% versus those with RVLS<|13|% after further dichotomisation for the type of treatment (ie, AVR vs conservative: ie, no AVR); *p<0.05 versus RVLS>|13|%+AVR; §p<0.05 versus RVLS>|13|%+AVR, RVLS>|13|%+No AVR and RVLS<|13|%+AVR. The numbers at the bottom of the graph represent the number of patients at risk at each follow-up year. The p value is that of the log-rank test. AVR, aortic valve replacement; RVLS, right ventricular longitudinal strain.

In a Cox analysis stratified for the type of treatment (AVR vs conservative), rest RVLS<|13|% was significantly associated with an increased risk of all-cause mortality (HR=1.92; 95% CI 1.06 to 3.47; p=0.031; table 2). In multivariable Cox analysis stratified for type of treatment and adjusted for age, previous MI, AVAproj and LVEF, RVLS<|13|% remained independently associated with all-cause mortality (HR=3.06; 95% CI 1.41 to 6.63; p=0.005; table 2, Model #1). Results were similar when LVEF was replaced by SVi (HR=3.32; 95% CI 1.47 to 7.28; p=0.004) or LVLS (HR=2.70; 95% CI 1.19 to 6.11; p=0.018) in the model (table 2, Model#2) or after further adjustment for diabetes (HR=3.12; 95% CI 1.42 to 6.72; p=0.005). The addition of RVLS had incremental prognostic value over baseline risk factors and LVLS (χ2=20.13 vs 13.56; p=0.01; figure 3). The Net Reclassification Improvement for rest RVLS was 0.43 (p=0.04).

Table 2

Multivariable analysis of the effect of rest and stress RVLS on all-cause mortality

Figure 3

Incremental prognostic value of rest RVLS over baseline risk factors and LVLS. AVA, aortic valve area; AVR, aortic valve replacement; BSL, baseline risk factors (age, AVR vs conservative treatment, AVA proj, previous myocardial infarction); LVLS, left ventricular global longitudinal strain; RVLS, right ventricular longitudinal strain.

Tricuspid regurgitation >mild (HR=2.18; 95% CI 1.30 to 3.65; p=0.004) was associated with mortality in univariable analysis (see online supplementary table S2). After adjustment for the same variables included in Model #2 and tricuspid regurgitation, RVLS <13% remained associated with all-cause mortality (HR=2.43; 95% CI 1.04 to 5.68; p=0.04). Tricuspid regurgitation >mild also was associated with mortality in multivariable analysis (HR=2.51; 95% CI 1.32 to 4.80; p=0.005).

In sequential Cox analyses, the addition of rest RVLS also provided significant incremental prognostic value beyond the other following risk factors previously identified in patients with LF-LG AS: that is, TR >mild (p=0.025), Duke activity status index (p=0.023), 6 min walking distance (p=0.05) or flow reserve on DSE (stroke volume increase >20% during DSE) (p=0.016) (see online supplementary table S3).

Survival according to stress RVLS

In the subset of patients with stress RVLS (n=58), 28 (48%) patients died. Two-year survival was 73%±7% in patients with stress RVLS >|14|% compared with 36%±12% in patients with stress RVLS <|14|% (p=0.005; figure 4). There was a modest correlation between stress RVLS and rest RVLS (r=0.49; p=0.0001). After stratification for type of treatment, stress RVLS <|14|% remained significantly associated with an increased risk of all-cause mortality (HR=3.00; 95% CI 1.33 to 6.79; p=0.008). In multivariable Cox analysis adjusted for age, AVAproj and LVEF, stress RVLS <|14|% remained independently associated with all-cause mortality (HR=2.94; 95% CI 1.23 to 7.02; p=0.015; table 2, Model #1). After further adjustment for rest RVLS, stress RVLS <|14|% remained independently associated with mortality (HR=3.29; 95% CI 1.17 to 9.25; p=0.024; table 2, Model #2), whereas rest RVLS was not (p>0.05). The addition of stress RVLS to the baseline model including LVLS and rest RVLS improved the ability of predicting outcomes in sequential Cox model (χ2=26.2 vs 18.4; p=0.005). The Net Reclassification Improvement for stress RVLS was 0.52 (p=0.009).

Figure 4

Impact of stress RVLS on all-cause mortality. Figure shows the survival in patients with stress RVLS>|14|% versus those with stress RVLS<|14|%. The numbers at the bottom of the graph represent the number of patients at risk at each follow-up year. The p value is that of the log-rank test. RVLS, right ventricular longitudinal strain.

Discussion

The main findings of the present study are as follows: (1) severe RV longitudinal dysfunction defined as rest RVLS <|13|% is independently associated with all-cause mortality in patients with LF-LG AS and low LVEF and provides incremental prognostic value beyond clinical risk factors and echocardiographic parameters of AS severity and LV systolic function. (2) RVLS measured during DSE provides additional prognostic value beyond that obtained with rest RVLS.

Rest RVLS and outcomes in low LVEF, LF-LG AS

The presence of severely depressed LV systolic longitudinal function, defined as LVLS <|9|%, has been reported to be associated with worse outcomes in patients with LF-LG AS and low LVEF.12 However, the impact of RV dysfunction on prognosis is unknown. Several studies have shown that echocardiographic parameters of RV function are useful to predict outcomes in patients with pulmonary hypertension, patients with ischaemic heart disease and patients undergoing cardiac surgery.13 ,25–29 Recent studies suggested that RVLS assessed by speckle tracking better correlates with RVEF measured by cardiac magnetic resonance as compared with other echocardiographic parameters of RV function such as TAPSE, tricuspid annulus systolic velocity or myocardial performance index 16 ,30 and better predicts outcome in patients with ischaemic cardiomyopathy.16 ,31 Furthermore, in a recent study including patients with end stage of heart failure, RV free wall strain was, among all echocardiographic parameters of RV function, the one that best correlated with the extent of RV myocardial fibrosis and patient's exercise capacity.32 In a recent study, we reported that the presence of >mild tricuspid regurgitation is associated with increased risk of mortality in patients with LF-LG AS.33 In the present study, we found that reduced RVLS is associated with increased risk of mortality. Indeed, the presence of severe RV dysfunction defined as an RVLS <|13|% is associated with a threefold increase in mortality in patients with LF-LG AS and low LVEF, after comprehensive adjustment including LVLS and tricuspid regurgitation severity. It is important to note that in this subset of patients, the presence of severe LV dysfunction does not necessarily imply the presence of severe RV dysfunction or vice versa (figure 1). Hence, rest RVLS provides incremental prognostic value beyond the parameters of LV global or longitudinal function.

Stress RVLS and outcomes in low LVEF, LF-LG AS

DSE is useful to assess the degree of LV myocardial impairment and improves the risk stratification of patients with LF-LG AS and low LVEF.5 ,17–19 ,34–36 No study has examined the impact of DSE on RV function in these patients. In the present study, there was no significant difference between the average values of rest and peak stress RVLS; however, some patients displayed a marked improvement in RVLS with DSE, whereas others had an important deterioration in RV function. Furthermore, stress RVLS provided significant incremental prognostic value beyond that obtained from rest RVLS. We previously reported similar findings with LV longitudinal function, where the stress LVLS provided additional prognostic information beyond that of rest LVLS. Hence, there appears to be an advantage to assess both LVLS and RVLS at rest and during DSE. DSE is routinely performed in patients with LF-LG AS and low LVEF and the strain analyses could be performed on the acquired images.

Clinical implications

The results of the present study show that RV dysfunction as assessed by RVLS is associated with increased risk of mortality, independent of the type of treatment: conservative or AVR. RV dysfunction is not included in the current surgical risk scores. Recent studies report a significant impairment of RV function after surgical AVR, but an improvement or no change in RV function after transcatheter AVR.15 ,37 These studies suggest that transcatheter AVR may be superior to surgical AVR in patients with pre-existing RV dysfunction. Absence of flow reserve on DSE and severely reduced LV longitudinal function at rest and/or DSE may also shift the balance in favour of transcatheter AVR.12 Further studies are needed to compare the outcomes of transcatheter versus surgical AVR in patients with LF-LG AS having RV dysfunction, severely impaired LV longitudinal function and/or no LV flow reserve.

Study limitations

RVLS at rest and at peak DSE were not available in all patients due to inadequate frame rate or poor-quality images. However, there were no baseline or outcome differences between patients with and without rest and/or peak RVLS measurements. We did not test the effect of other parameters of RV function (ie, TAPSE, tricuspid annulus systolic velocity, and so on) because they were available in only a small proportion of patients. The population size may have limited our ability to detect significant associations with other factors in the multivariable analysis and did not allow us to perform an analysis of perioperative mortality or the impact in each treatment arm independently. This may also have limited our ability to detect a significant interaction between RVLS and type of treatment. However, a larger study in this high-risk population would be difficult to achieve due to the low prevalence of low LVEF, LF-LG AS. Some of our multivariable models could be overfitted due to the number of variables included in our analysis, especially in the subset of patients with stress RVLS.

Conclusion

In this series of patients with LF-LG AS and low LVEF, severe RV dysfunction as defined by RVLS <|13|% was independently associated with an increased risk of mortality. Furthermore, stress RVLS provided incremental prognostic value beyond that obtained from rest RVLS. Measurement of RVLS both at rest and during DSE may be helpful to improve risk stratification in this high-risk population.

Key messages

What is already known on this subject?

  • Left ventricular (LV) longitudinal strain is independently associated with mortality in patients with low-flow, low-gradient aortic stenosis (LF-LG AS) and reduced LV ejection fraction. However, the prognostic value of right ventricular (RV) dysfunction as measured by strain is unknown in this high-risk population.

What might this study add?

  • This study shows that reduced RV free wall longitudinal strain (RVLS<|13|%) is independently associated with increased risk of mortality in patients with LF-LG AS even after adjustment for other risk factors including LV longitudinal strain and tricuspid regurgitation severity. Furthermore, RVLS measured during dobutamine stress echocardiography provides incremental prognostic value beyond that obtained from rest RVLS.

How might this impact on clinical practice?

  • Measurement of RVLS both at rest and during dobutamine stress echocardiography may be helpful to enhance risk stratification in this high-risk population with LF-LG AS.

Acknowledgments

We thank Jocelyn Beauchemin, Dominique Labrèche, Isabelle Laforest, Isabelle Fortin and Martine Poulin for the data collection and their technical assistance.

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

  • Contributors AD: coordination of the study, acquisition and analysis of the data, preparation of the first draft of the manuscript. M-AC: participation to acquisition and analysis of data; critical revisions on the manuscript. RC: critical revisions on the manuscript. PEB: participation to acquisition of data in one of the participating site. JM: participation to analysis of data; critical revisions on the manuscript. GM: local principal investigator of one of the participating sites. JB-K, IB, HB: local principal investigators of one of the participating sites; critical revisions on the manuscript. JM, RR: critical revisions on the manuscript. HBR: participation in data acquisition; critical revisions on the manuscript. KO'C: central analysis of echos; critical revisions on the manuscript. MS: participation in data acquisition; central analysis of echos. JGD, EL, JR-C: participated in the design of the study; critical revisions on the manuscript. PM: supervision of the surgical aspects of the study; critical revisions on the manuscript. PP: principal investigator of the TOPAS study; supervision of the acquisition, analysis and interpretation of the data; critical revisions and validation of the final version of the paper.

  • Funding This work was supported by a grant (# MOP-57445 for TOPAS-II and # MOP-126072 and FDN-143225 for TOPAS-III) from the Canadian Institutes of Health Research, Ottawa, Canada. AD was supported by a fellowship grant from “L'Agence de la santé et des services sociaux de la Capitale nationale-ADLSSS”, Québec, Québec, Canada. M-AC and RC were supported by a post-doctoral fellowship from the Canadian Institutes of Health Research, Ottawa. HBR was supported by a research PhD grant from ‘Conselho Nacional de Desenvolvimento Científico e Tecnológico—Brasil’. PM is a research scholar from the ‘Fonds de Recherche en Santé du Québec, Montreal, Québec, Canada’. PP holds the Canada Research Chair in Valvular Heart Diseases, Canadian Institutes of Health Research.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Quebec Heart and Lung Institute Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.