Author + information
- Received November 2, 2017
- Revision received January 16, 2018
- Accepted January 17, 2018
- Published online March 14, 2018.
- Anastasia Vamvakidou, MBBSa,b,c,
- Wenying Jin, MD, PhDa,b,
- Oleksandr Danylenko, MD, PhDa,b,
- Navtej Chahal, MBBS, PhDa,b,c,
- Rajdeep Khattar, MBBS, DMb,c and
- Roxy Senior, MD, DMa,b,c,∗ ()
- aNorthwick Park Hospital, Harrow, United Kingdom
- bRoyal Brompton Hospital, London, United Kingdom
- cBiomedical Research Unit, National Heart and Lung Institute, Imperial College, London, United Kingdom
- ↵∗Address for correspondence:
Dr. Roxy Senior, Department of Cardiology, Royal Brompton Hospital, Sydney Street, London SW3 6NP, United Kingdom.
Objectives This study aimed to assess the value of low transvalvular flow rate (FR) for the prediction of mortality compared with low stroke volume index (SVi) in patients with low-gradient (mean gradient: <40 mm Hg), low aortic valve area (<1 cm2) aortic stenosis (AS) following aortic valve intervention.
Background Transaortic FR defined as stroke volume/left ventricular ejection time is also a marker of flow; however, no data exist comparing the relative prognostic value of these 2 transvalvular flow markers in patients with low-gradient AS who had undergone valve intervention.
Methods We retrospectively followed prospectively assessed consecutive patients with low-gradient, low aortic valve area AS who underwent aortic valve intervention between 2010 and 2014 for all-cause mortality.
Results Of the 218 patients with mean age 75 ± 12 years, 102 (46.8%) had low stroke volume index (SVi) (<35 ml/m2), 95 (43.6%) had low FR (<200 ml/s), and 58 (26.6%) had low left ventricular ejection fraction <50%. The concordance between FR and SVi was 78.8% (p < 0.005). Over a median follow-up of 46.8 ± 21 months, 52 (23.9%) deaths occurred. Patients with low FR had significantly worse outcome compared with those with normal FR (p < 0.005). In patients with low SVi, a low FR conferred a worse outcome than a normal FR (p = 0.005), but FR status did not discriminate outcome in patients with normal SVi. By contrast, SVi did not discriminate survival either in patients with normal or low FR. Low FR was an independent predictor of mortality (p = 0.013) after adjusting for age, clinical prognostic factors, European System for Cardiac Operative Risk Evaluation II, dimensionless velocity index, left ventricular mass index, left ventricular ejection fraction, heart rate, time, type of aortic valve intervention, and SVi (p = 0.59).
Conclusions In patients with low-gradient, low valve area aortic stenosis undergoing aortic valve intervention, low FR, not low SVi, was an independent predictor of medium-term mortality.
Severe aortic stenosis (AS) may be present in patients with discordant aortic valve area (AVA) and transvalvular mean gradient (i.e., low AVA [<1 cm2] but low mean gradient [<40 mm Hg]). This mainly occurs from a low flow state, provided measurement errors have been addressed. This phenomenon may be present in patients both with normal and low left ventricular ejection fraction (LVEF) (≤50%) (1). Conventionally, left ventricular outflow tract (LVOT) Doppler-derived stroke volume indexed to body surface area (BSA-SVi) has been used as a marker of transaortic valve flow. Several studies have shown that low SVi (<35 ml/m2) is a strong predictor of mortality in patients with symptomatic severe AS; however, its prognostic value, especially in the medium term, following aortic valve intervention has shown inconsistent results (2–4).
It was shown in an experimental model of low-flow low-gradient AS that when transvalvular flow rate (FR), defined as stroke volume (SV)/left ventricular ejection time (LVET), was increased beyond 200 ml/s, considered to be normal FR, valve area in severe AS did not change further compared with less severe AS. FR as a marker of transaortic flow was also proposed by Blais et al. (5) in such patients; however, the group considered 250 ml/s as normal flow. Preliminary small cross-sectional studies have suggested that the true severity of AS may be better determined when FR is ≥200 ml/s rather than when SVi is normal both at rest and following “flow correction” during dobutamine stress echocardiography (6,7). Recent data from the SEAS (Simvastatin and Ezetimibe in Aortic Stenosis) population, which consisted of patients with asymptomatic mild-moderate AS, showed that low transaortic FR <200 ml/s predicted mortality even after adjusting for low SVi (8); however, no data exist assessing the relative prognostic value of transaortic FR and SVi in patients with symptomatic discordant AS (low-gradient, low-AVA AS) who have undergone aortic valve intervention and who, therefore, would require careful risk stratification.
This project was granted approval by the Research & Development departments of Northwick Park (London North West Healthcare NHS Trust) and Royal Brompton (Royal Brompton and Harefield NHS Foundation Trust) Hospitals.
Between January 2010 and December 2014, 218 consecutive patients with AVA <1 cm2 and mean gradient <40 mm Hg who had aortic valve intervention including surgical aortic valve replacement (AVR), transcatheter aortic valve replacement (TAVR), or balloon aortic valvuloplasty (BAV) were evaluated. Patients with subaortic stenosis such as subvalvular membrane, supravalvular stenosis, and previous aortic valve intervention were excluded from the patient cohort. Patients who had concurrent coronary artery bypass surgery were also excluded from the analysis because in these cases aortic valve intervention may have been performed secondary to ischemic heart disease (9).
For the aortic valve assessment 2-dimensional (2D) left ventricular apical and parasternal views, as well as outflow tract and transaortic Doppler measurements were captured using a broadband frequency transducer (iE33, Philips Medical Systems, Eindhoven, the Netherlands). A Medcon reporting workstation was used for the off-line analysis.
Transthoracic echocardiography parameters
All echocardiographic data were prospectively collected but reevaluated retrospectively by 3 experienced investigators in echocardiography. LVOT was measured in mid-systole at the aortic annulus level. The LVOT velocity time integral (VTI), aortic valve VTI, mean, and peak gradients were also measured. The dimensionless velocity index (DVI) was calculated as LVOT VTI/aortic valve VTI and the AVA from the continuity equation. SV was derived as LVOT VTI × LVOTarea and was indexed to BSA (SVi). FR was calculated as SV divided by LVET. The latter was measured using the Doppler outflow signal as the time interval between the beginning and the endpoint of the previously traced LVOT VTI. Left ventricular mass was calculated using the following formula and was indexed for BSA (LV mass index):in which LVEDD = left ventricular end-diastolic diameter, IVSd = intraventricular septum diameter in end-diastole, and PWd = posterior wall diameter in end-diastole.
Finally, LVEF was derived from the biplane Simpson method. In cases in which image quality was poor and endocardial border definition uncertain, eyeball assessment of LVEF was used. For patients in sinus rhythm, both 2D and Doppler measurements were performed in the beat with the clearest image or Doppler signal, whereas for patients in atrial fibrillation, an average of 5 measurements was taken.
Interobserver variability was assessed by the 3 main investigators who were involved in the analysis of the transthoracic echocardiography images. Fifty random cases that included both patients in sinus rhythm and patients in atrial fibrillation were assigned to each of the 3 individuals for the evaluation of LVOT diameter, LVOT VTI, aortic valve VTI, LVET at the LVOT level, and LVEF. The 3 investigators performed the measurements independently.
The outcome assessed was all-cause mortality. The hospital electronic database was searched to check the date of aortic valve intervention and the NHS Care Record Service was used to identify any deaths and document the date of death. Survivors were also contacted by phone, when required, to collect information regarding valve intervention performed in other centers.
Categorical and continuous variables were expressed as percentages and mean ± SD respectively, apart from follow-up days, which are reported as median values with interquartile range. Categorical variables were compared with the chi-square test. Correlation between continuous variables was assessed by Pearson test. The correlations between flow (SVi and FR) and parameters of AS severity were assessed with Pearson correlation and the AS parameter with the weakest correlation was selected for the Cox regression analysis.
Univariable and multivariable Cox regression analysis were performed to identify predictors of mortality. Variables with p < 0.1 in the univariable analysis were subsequently entered into the multivariable model. Because SVi and FR have the same numerator, interaction between the cutoffs of FR and SVi was assessed by computing centered variables (FR and SVi) and centered product FR·SVi and evaluating them with linear regression analysis. Collinearity between the 2 flow parameters was further examined with linear regression collinearity statistics. Both FR and SVi were not only examined in the same model, but also in separate multivariable models to avoid any issue of collinearity. For this purpose, the same number of subjects was included for both analyses. Hierarchical Cox-regression analysis was used to identify the incremental prognostic value of the different flow parameters. Kaplan-Meier survival curves and log-rank p values were used to evaluate the impact of specific variables on mortality. Intraclass correlation coefficient for single measures was assessed to test the variability of the echocardiographic measurements among the 3 investigators. Finally, receiver-operating characteristic (ROC) curve was used to identify the best cutoff of continuous variables for the prediction of mortality.
For all tests, p < 0.05 was considered statistically significant. Hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated. Statistical analysis was performed with SPSS, version 22.0 (SPSS Inc., Chicago, Illinois).
Baseline clinical and echocardiographic characteristics
A total of 218 patients were included in the study, of which 126 (57.8%) underwent TAVR, 83 (38.1%) had surgical AVR, and 9 (4.1%) underwent BAV (Table 1). One hundred and two patients (46.8%) had low SVi, 95 (43.6%) had low FR. and 58 (26.6%) had low LVEF (≤50%).
Patients with low flow (FR <200 ml/s) had similar demography and EuroSCORE II (European System for Cardiac Operative Risk Evaluation) compared with those with normal FR, except that the prevalence of atrial fibrillation was higher in the low FR group. The time to aortic valve intervention was also shorter in the low-flow group. The low-flow group had significantly smaller AVA and lower DVI, mean gradient, and LVEF versus the normal-flow group (Table 1). LVOT diameter and SVi were also lower in patients with low FR versus normal FR. Among the 213 patients in whom SVi was calculated (availability of BSA data), the concordance between low and normal flow classified by FR and SVi was 78.8%. Of the 91 patients with low FR, 74 (81.3%) demonstrated low SVi; of the 122 patients with normal FR, 28 (23%) had low SVi (p < 0.005).
The intraclass correlation coefficients among the 3 observers were 0.68 (95% CI: 0.54 to 0.79), 0.80 (95% CI: 0.69 to 0.87), and 0.87 (95% CI: 0.79 to 0.92) for the measurements of LVOT diameter, LVEF, and LVET, respectively.
Correlation among SVi, FR, and echocardiographic parameters of AS severity
The 2 flow parameters (i.e., SVi and FR) were moderately correlated (r = 0.71; p < 0.005). The correlation between FR and AVA (r = 0.71; p < 0.005) was stronger compared with SVi and AVA (r = 0.59; p < 0.005); however, both SVi and FR had weaker but significant and similar correlations with the transvalvular mean gradient (r = 0.46, p < 0.005 for SVi; r = 0.42, p < 0.005 for FR) and DVI (r = 0.46, p < 0.005 for SVi; r = 0.34, p < 0.005 for FR).
Predictors of low FR
Amongst clinical and echocardiographic parameters, significant univariable predictors of low flow (FR <200 ml/s) were atrial fibrillation, AVA, mean gradient, LVOT diameter, and LVEF. Only AVA, mean gradient, and LVOT diameter were independent predictors, however (Table 2).
Prediction of mortality
During a median follow-up of 46.8 ± 21 months from the time of echocardiography, 52 deaths occurred (23.9%). The mortality was higher in patients with low FR (n = 35, 36.8%) compared with those with normal FR (n = 17, 13.8%) (log-rank p < 0.005), as shown in Figure 1. Mortality was also higher in those with low SVi (n = 31, 30.4%) compared with normal SVi (n = 20, 18%) (log-rank p = 0.02); however, as shown in Figure 2A, in patients with low SVi (n = 102), low FR predicted higher mortality (n = 29 of 74, 39.2%) versus normal FR (n = 2 of 28, 7.1%) (log-rank p = 0.005), but FR status did not discriminate outcome in patients with normal SVi (log-rank p = 0.30) (Figure 2B). In the group with low SVi, patients with normal FR were younger with a lower prevalence of peripheral vascular disease and a trend toward lower prevalence of coronary artery disease. They also had higher AVA, mean gradient, and SV, but lower LVET (Table 3). In these patients, the prevalence of other factors that may influence LVET (i.e., bradycardia, right ventricular pacing, and left bundle branch block was only 15.7% [n = 16]) and there was no difference in outcome between patients with and without these abnormalities (log-rank p = 0.47). Furthermore, SVi did not discriminate survival either in patients with normal or low FR (log-rank p = 0.33 and p = 0.28, respectively).
FR status discriminated mortality in patients with normal LVEF (log-rank p < 0.005) but not in patients with low LVEF (log-rank p = 0.28) (Figures 3A and 3B, respectively). Similar to SVi, in patients with either normal or low FR, LVEF status did not have an impact on outcome (log-rank p = 0.13 and p = 0.92, respectively).
We assessed interaction between FR and SVi with linear regression analysis and have found p values for FR (p = 0.033), SVi (p = 0.619), and FR·SVi (p = 0.068), which were statistically nonsignificant. Collinearity between the 2 flow parameters was examined with linear regression analysis and the SVi variance inflation factor was 1.73 and tolerance 0.58, whereas the FR variance inflation factor was 1.58 and tolerance 0.63, which suggested collinearity was unlikely. Table 4 lists the univariable predictors of mortality. When both flow parameters were entered in the same multivariable model, low transaortic FR remained an independent predictor of mortality (HR: 2.89; 95% CI: 1.25 to 6.69; p = 0.013), but not low SVi (HR: 0.79; 95% CI: 0.33 to 1.90; p = 0.59) (Table 5). To address the possible clinical issue of collinearity between FR and SVi, separate multivariable models were used containing the same prognostic parameters as previously. Low FR was an independent predictor of mortality (HR: 2.5; 95% CI: 1.31 to 4.77; p = 0.006), whereas low SVi was not (HR: 1.63; 95% CI: 0.85 to 3.11; p = 0.14). To minimize the effect of overfitting, the variables in the multivariable model were reduced by creating clinical risk and aortic valve intervention scores based on significant univariable factors. The clinical risk score contained age, previous ischemic heart disease, peripheral vascular disease, and chronic kidney disease. Based on the ROC curve for the prediction of mortality, optimum age cutoff was 75 years. Clinical risk was scored from 0 to 4 (0 = absent risk, 4 = all 4 risk factors). The aortic valve intervention risk score contained the time to aortic valve intervention (converted to categorical variable with cutoff 175 days based on ROC curve analysis) and the type of aortic valve intervention. The multivariable model comprising clinical risk score, aortic valve intervention risk score, LVEF, low FR, and low SVi showed that low FR remained an independent predictor of mortality (HR: 3.19; 95% CI: 1.43 to 7.14; p = 0.005), whereas low SVi did not (HR: 0.87; 95% CI: 0.39 to 1.96; p = 0.74).
Amongst the different echocardiographic parameters of AS severity, DVI was the one introduced in the univariable analysis because it displayed lesser collinearity with FR; however, even when AVA (HR: 0.05; 95% CI: 0.01 to 0.28; p = 0.001) or transvalvular mean gradient (HR: 0.97; 95% CI: 0.93 to 1.01: p = 0.09) were entered in the multivariable model, low FR remained an independent predictor of mortality (p = 0.02 after adjusting for either AVA or the mean transvalvular gradient). SVi remained a nonsignificant predictor (p = 0.42 and p = 0.35, respectively) after adjusting for AVA and mean transvalvular gradient.
LVEF was not an independent predictor of mortality. When low LVEF ≤50% was entered into the model, low FR remained an independent predictor of mortality (HR: 2.71; 95% CI: 1.42 to 5.18; p = 0.002).
When prognostic information was assessed in a hierarchal manner as done clinically (Figure 4) (i.e., prognostic clinical variables, followed by LVEF followed subsequently by the time/type of aortic valve intervention), then by SVi and finally by FR, LVEF, and the aortic valve intervention parameters significantly improved the predictive value of the model containing clinical variables (p = 0.04 and p < 0.005, respectively). SVi did not provide any incremental prognostic information beyond the model containing clinical variables, LVEF, and the aortic valve intervention parameters (p = 0.15). By contrast, when FR was added to the model containing the previous factors, it significantly improved (p = 0.01) the prediction of mortality.
Figure 5 shows the association between mortality and grades of FR. There is a stepwise increase in mortality with decreasing FR. ROC curve analysis showed that the optimal cutoff value of FR for the prediction of mortality following aortic valve intervention was close to 200 ml/s (208 ml/s) (area under the curve: 0.67; p < 0.005) with sensitivity, specificity, positive predictive value, and negative predictive value of 75.0%, 53.0%, 33.0%, and 86.4%, respectively.
This is the first study to show that transaortic FR is an independent predictor of mortality during medium-term follow-up of patients with low-gradient low-AVA AS who underwent aortic valve intervention. Low FR (<200 ml/s) predicted almost 3-fold increase in mortality compared with normal FR; furthermore, FR status demonstrated stepwise increase in mortality with worsening FR. FR was superior to both SVi and LVEF for the prediction of outcome, and the prognostic information provided was also independent of clinical data. When FR was assessed in a clinically oriented hierarchical manner (i.e., clinical prognostic variables followed by LVEF, the type/timing of aortic valve intervention, and SVi), FR provided further incremental prognostic information.
A recent study from SEAS database also showed that low transaortic FR <200 ml/s predicted mortality (8); however, the population consisted of asymptomatic patients with mild-moderate AS who were free of cardiovascular disease, diabetes mellitus, and renal disease that did not warrant valve intervention. By contrast, the present population had symptomatic discordant AS, was a decade older, had significant cardiovascular morbidity, and underwent aortic valve intervention. Unlike SEAS, this population was more representative of those seen in our daily clinical practice, requiring careful risk stratification, and in whom assessment of risk such as flow status was of vital importance prior to valve intervention.
Importance of determining flow in low-gradient AS
Low-gradient AS has been shown to confer worse prognosis after aortic valve intervention than high gradient AS (11,12). In this study, mortality was approximately 24% over nearly 4 years following aortic valve intervention. The cause of higher mortality in low-gradient AS is multifactorial; the primary determinant of transvalvular gradient is flow. A small reduction in flow results in exponential reduction in gradient because the latter is the squared function of transvalvular flow. The flow may in turn be reduced because of low LVEF (classical low-flow low-gradient AS) (1,13); however, flow may also be reduced in patients with normal LVEF (paradoxical low-flow low-gradient AS) (14). This is mainly because of restrictive diastolic filling (myocardial disease with and without ischemic heart disease), small LV size, atrial fibrillation, significant mitral valve disease, tricuspid regurgitation, or constrictive pericardial disease (15). Low flow may also occur in the presence of reduced vascular compliance and systemic hypertension (16,17); thus, a low-flow status may be the culmination of multiple significant underlying cardiovascular pathological processes. It is not surprising therefore that low-flow state reflects a high-risk patient group; this has been shown using SVi as a marker of flow (11–13).
This study demonstrated, however, that the prediction of outcome in patients with a low-flow state as determined by SVi may be further improved by assessing FR. It has been shown in an experimental model that AVA at a FR of ≥200 ml/s may reflect the true severity of AS (18). In another small cross-sectional preliminary study in patients with low-gradient discordant AS, FR performed better than SVi for the detection of true severe AS (7). Beyond a FR of 200 ml/s, AVA in true severe AS showed very little change, whereas those patients with nonsevere AS showed continued increase in valve area (7,18); therefore, if patients with low-gradient and AVA <1 cm2 demonstrated a FR ≥200 ml/s, this most likely represents true severe AS. This entity, known as normal-flow low-gradient AS, was associated with better outcome compared with low-flow state following valve intervention. Normal FR in this group was a marker of better LV function (higher SV) and less severe AS (shorter LVET). Nevertheless, these patients had true severe AS because, despite normal FR, AVA remained low (mean 0.81 cm2) with a mean gradient of 31 mm Hg, which corresponds to AVA <1 cm2 (19,20). These factors explain improved survival in patients with normal FR compared with low FR despite low SVi.
Physiological basis of difference between FR and SVi
Transaortic FR is defined as SV/LVET (ml/s). According to the formula, the transaortic FR may be normal or low at the same SV with a shorter or longer LVET, respectively. Because in severe AS, the valve resistance is high, LVET is prolonged; thus, with the same SV, the FR may be low in very severe AS or normal in less severe AS. In the present study, AVA correlated more strongly with FR than with SVi. The numerator of SV being the same, the difference between FR and SVi ultimately is likely due to the difference in the denominators used to derive these parameters (i.e., LVET and BSA, respectively) (Figure 6). Accordingly, low FR may also occur in other conditions that prolong LVET (i.e., bradycardia, left bundle branch block, and right ventricular pacing), all of which may be prevalent in this population and may adversely affect outcome. The prevalence of this cohort in the population with low SVi where FR predicted outcome was only 16%, however, and it did not influence outcome. In the patients with normal FR, despite similar SV, SVi may be low because of obesity (large BSA). Obesity was thought to play a role in predicting higher mortality in patients with low SVi in AS (21); however, in the present population, the prevalence of obesity was low (mean BSA 1.8 m2) and may explain in part the attenuated prognostic power of SVi.
The study suggests that when FR is normal in patients with low-gradient low-AVA AS, this most likely represents true severe, though not critical, AS. A normal FR also tends to reflect better preserved intrinsic myocardial function. Both of these factors contributed to improved medium-term survival in patients with normal FR following aortic valve intervention. In patients with low FR, even if LVEF was normal, the outcome was significantly compromised. Finally, a normal SVi predicted improved outcome, but a low SVi did not necessarily portend an adverse outcome, especially when FR was normal. Thus, this study suggests that in patients with low-gradient low-AVA AS, FR may be useful in predicting medium-term risk before aortic valve intervention, although larger population studies are needed for further confirmation.
The strength of the study is that it is a real-world study comprising a population of patients with low-gradient low-AVA AS. This is a challenging population to manage because an assessment of the true severity of AS is often difficult; this may have a significant impact on outcome following valve intervention. To the best of our knowledge, the cohort studied is fairly large relative to what is published and the outcome assessed was all-cause mortality, which is a robust marker of outcome (4,22,23). The event rate was also sufficient to not compromise the overall conclusion of the study.
Limitations of the study are that this is a prospective study with retrospective assessment of data in which patients underwent either surgical AVR, TAVR, or BAV based on their comorbidities, symptoms, concomitant ischemic heart disease, and their own preferences; therefore, the treatment provided was not predetermined. As a result, the prevalence of low LVEF in this low-gradient AS population was low and likely related to the exclusion of patients with severe LV systolic dysfunction from aortic valve intervention. Discordant results occurred in 45 patients, and thus represent the population responsible for the finding that low FR is superior to low SVi. The findings need to be confirmed, ideally in large randomized trials with core laboratory-measured data. Despite the reasonably good agreement between the observers who performed the echocardiographic measurements, the observations may still be subject to errors relating to the underestimation of the LVOT diameter by 2D transthoracic echocardiography, which is common, and compounded with difficulties in obtaining reliable measurements in the presence of atrial fibrillation. Nevertheless, in this study, low AVA was independent of LVOT diameter as a predictor of low FR.
FR, a measure of transvalvular flow, provided incremental survival information in patients with low-gradient AS undergoing aortic valve intervention beyond that provided by routine clinical, LVEF, and SVi data.
COMPETENCY IN MEDICAL KNOWLEDGE: In patients with discordant AS who had undergone aortic valve intervention, low transvalvular FR (determined echocardiographically by SV/LVET) predicts almost 3-fold increase in mortality compared with normal FR. Furthermore, FR is superior to both SVi and LVEF for the prediction of outcome independently of other clinical and echocardiographic prognostic factors.
TRANSLATIONAL OUTLOOK: Further large-scale studies are needed to establish the prognostic value of FR compared with SVi in patients with discordant AS for routine inclusion of FR in the risk stratification of patients undergoing aortic valve intervention.
The authors thank Paul Bassett for his valuable assistance with the statistical methodology of the manuscript.
Dr. Senior has received speaker fees from Bracco Imaging, Lantheus Medical Imaging, and Philips Healthcare. All other 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
- aortic valve replacement
- balloon aortic valvuloplasty
- body surface area
- confidence interval
- dimensionless velocity index
- flow rate
- hazard ratio
- left ventricular ejection fraction
- left ventricular ejection time
- left ventricular outflow tract
- receiver- operating characteristic
- stroke volume
- stroke volume index
- transcatheter aortic valve replacement
- velocity-time integral
- Received November 2, 2017.
- Revision received January 16, 2018.
- Accepted January 17, 2018.
- 2018 American College of Cardiology Foundation
- Pibarot P.,
- Dumesnil J.
- Le Ven F.,
- Freeman M.,
- Webb J.,
- et al.
- Herrmann H.C.,
- Pibarot P.,
- Hueter I.,
- et al.
- Clavel M.A.,
- Berthelot-Richer M.,
- Le Ven F.,
- et al.
- Blais C.,
- Burwash I.,
- Mundigler G.,
- et al.
- Vamvakidou A.,
- Chahal N.,
- Senior R.
- Chahal N.,
- Drakopoulou M.,
- Gonzalez-Gonzalez A.,
- Manivarmane R.,
- Khattar R.,
- Senior R.
- Saeed S.,
- Senior R.,
- Chahal N.S.,
- et al.
- Vahanian A.,
- Alfieri O.,
- Andreotti F.,
- et al.
- Levy F.,
- Laurent M.,
- Monin J.,
- et al.
- Hachicha Z.,
- Dumesnil J.,
- Bogaty P.,
- Pibarot P.
- Leong D.P.,
- Pizzale S.,
- Haroun M.J.,
- et al.
- Briand M.,
- Dumesnil J.G.,
- Kadem L.,
- et al.
- Eleid M.F.,
- Nishimura R.A.,
- Sorajja P.,
- Borlaug B.A.
- Voelker W.,
- Reul H.,
- Nienhaus G.,
- et al.
- Michelena H.I.,
- Margaryan E.,
- Miller F.A.,
- et al.
- Grayburn P.A.
- Mohty D.,
- Magne J.,
- Deltreuil M.,
- et al.
- Eleid M.,
- Sorajja P.,
- Michelena H.,
- Malouf J.,
- Scott C.,
- Pellikka P.