Author + information
- Received September 2, 2015
- Revision received March 7, 2016
- Accepted March 8, 2016
- Published online May 24, 2016.
- Suzanne J. Baron, MD, MSca,
- Suzanne V. Arnold, MD, MHAa,
- Howard C. Herrmann, MDb,
- David R. Holmes Jr., MDc,
- Wilson Y. Szeto, MDb,
- Keith B. Allen, MDa,
- Adnan K. Chhatriwalla, MDa,
- Sreekaanth Vemulapali, MDd,
- Sean O’Brien, PhDd,
- Dadi Dai, PhDd and
- David J. Cohen, MD, MSca,∗ ()
- aSaint Luke’s Mid America Heart Institute, University of Missouri-Kansas City, Kansas City, Missouri
- bHospital of the University of Pennsylvania, Philadelphia, Philadelphia
- cMayo Clinic, Rochester, Minnesota
- dDuke Clinical Research Institute, Durham, North Carolina
- ↵∗Reprint requests and correspondence:
Dr. David J. Cohen, Saint Luke’s Mid America Heart Institute, University of Missouri-Kansas City School of Medicine, 4401 Wornall Road, Kansas City, Missouri 64111.
Background In patients with aortic stenosis undergoing transcatheter aortic valve replacement (TAVR), studies have suggested that reduced left ventricular (LV) ejection fraction (LVEF) and low aortic valve gradient (AVG) are associated with worse long-term outcomes. Because these conditions commonly coexist, the extent to which they are independently associated with outcomes after TAVR is unknown.
Objectives The purpose of this study was to evaluate the impact of LVEF and AVG on clinical outcomes after TAVR and to determine whether the effect of AVG on outcomes is modified by LVEF.
Methods Using data from 11,292 patients who underwent TAVR as part of the Transcatheter Valve Therapies Registry, we examined rates of 1-year mortality and recurrent heart failure in patients with varying levels of LV dysfunction (LVEF <30% vs. 30% to 50% vs. >50%) and AVG (<40 mm Hg vs. ≥40 mm Hg). Multivariable models were used to estimate the independent effect of AVG and LVEF on outcomes.
Results During the first year of follow-up after TAVR, patients with LV dysfunction and low AVG had higher rates of death and recurrent heart failure. After adjustment for other clinical factors, only low AVG was associated with higher mortality (hazard ratio: 1.21; 95% confidence interval: 1.11 to 1.32; p < 0.001) and higher rates of heart failure (hazard ratio: 1.52; 95% confidence interval: 1.36 to 1.69; p <0.001), whereas the effect of LVEF was no longer significant. There was no evidence of effect modification between AVG and LVEF with respect to either endpoint.
Conclusions In this series of real-world patients undergoing TAVR, low AVG, but not LV dysfunction, was associated with higher rates of mortality and recurrent heart failure. Although these findings suggest that AVG should be considered when evaluating the risks and benefits of TAVR for individual patients, neither severe LV dysfunction nor low AVG alone or in combination provide sufficient prognostic discrimination to preclude treatment with TAVR.
Left ventricular (LV) dysfunction is associated with an increased risk of poor periprocedural outcome in patients with severe aortic stenosis (AS) undergoing surgical aortic valve replacement (SAVR) (1). Despite this early hazard, SAVR has been shown to improve symptoms and survival when compared with medical therapy alone in such patients (2). Recently, transcatheter aortic valve replacement (TAVR) has emerged as an alternative treatment for those patients who are considered either inoperable or at high risk for complications of SAVR. For such patients, TAVR has been shown to provide substantial improvements in both survival and quality of life when compared with medical therapy alone, and similar intermediate term survival when compared with SAVR (3–6). Although several studies have demonstrated that the benefits of TAVR are preserved among patients with moderate LV dysfunction (7–9), little is known about the outcomes of TAVR among patients with severe LV dysfunction—in part because such patients have generally been excluded from pivotal clinical trials.
Moreover, recent studies have suggested that other hemodynamic parameters may affect clinical outcomes after aortic valve replacement. In particular, AS patients with low aortic valve gradient (AVG) (resulting from either LV dysfunction or reduced transvalvular flow in the setting of preserved ejection fraction [EF]) generally have poorer survival rates with medical management, SAVR, or TAVR than those with high AVG (10–15). Furthermore, little is known about the interaction between AVG and LV dysfunction and how these 2 variables affect outcomes after TAVR in a real-world population. To address these questions, we used data from the Society of Thoracic Surgeons/American College of Cardiology’s TVT (Transcatheter Valve Therapies) Registry to investigate the association between baseline LVEF and AVG and clinical and health status outcomes of TAVR.
The population for this study was derived from the TVT Registry, a national registry (16) designed to track outcomes of TAVR and other therapies for valvular heart disease. Data are collected using standardized definitions for clinical and procedural details and outcomes as previously described (16). Registry activities have been approved by a central institutional review board, and the Duke University School of Medicine institutional review board granted a waiver of informed consent and authorization for this study.
TVT Registry data for procedures performed between November 9, 2011, and June 27, 2014, were linked to Medicare administrative claims using direct patient identifiers. Patients were excluded from the study if records from the index procedure were unable or ineligible to be linked to a Medicare inpatient claim, if data were missing on baseline AVG or LVEF, or if the procedure was aborted. Only first admissions for each patient were included in this analysis.
The analytic cohort was stratified according to LVEF and AVG. First, the cohort was divided into 3 groups according to LVEF using clinically relevant cutpoints: severe LV dysfunction (LVEF <30%); mild/moderate LV dysfunction (LVEF 30% to 50%); and preserved LV function (LVEF >50%). Next, the cohort was divided into 2 groups according to mean AVG as assessed by preprocedure echocardiography: low AVG (mean AVG <40 mm Hg) and high AVG (mean AVG ≥40 mm Hg). Of note, the TVT Registry does not currently distinguish whether the AVG is derived from a resting or stress echocardiogram. Furthermore, the assessment of stroke volume index (SVI) was not feasible because of the inability to reliably calculate flow based on available data elements in the TVT Registry.
For clinical endpoints, outcomes were evaluated at hospital discharge and at 1 year using the TVT Registry data and Medicare claims data, respectively. In-hospital outcomes included death, myocardial infarction (MI), stroke, new requirement for dialysis, and length of hospital stay. Clinical outcomes at 1 year included death, MI, stroke, and hospitalization for recurrent heart failure. Medicare claims files were used for detection of rehospitalization using the following International Classification of Diseases-Ninth Revision-Clinical Modification codes: for MI, 410.x1; for stroke, 433.x1, 434.x1, 997.02, 437.1, 437.9, 430, 431, and 432.x; for heart failure, 398.x, 402.x1, 404.x1, 404.x3, 428.x). For outcomes associated with rehospitalizations, follow-up was censored at the time of death, at the time of loss of Medicare Part A or B coverage or FFS eligibility, or at the end of 1-year follow-up.
Patient-reported outcomes were evaluated using the Kansas City Cardiomyopathy Questionnaire (KCCQ) at baseline and 30-day follow-up. The KCCQ is a disease-specific instrument that has proven to be a reliable measure of health status in patients with severe AS (17). As previously described, substantial health status improvement was defined as a >20-point increase in the KCCQ Overall Summary Score (KCCQ-OS) compared with baseline (18). One-year KCCQ data were not examined for this study because of high rates of missing data and because previous studies have demonstrated that the majority of functional recovery after TAVR occurs within the first 30 days—particularly with transfemoral access (19,20).
Baseline characteristics are summarized as medians for continuous variables and proportions for categorical variables and were compared across the strata of LVEF and AVG using the chi-square test, the Kruskal-Wallis test, and the Wilcoxon test as appropriate. Unadjusted comparisons for in-hospital clinical outcomes and 30-day health status outcomes were performed using the chi-square test for categorical outcomes and the Kruskal-Wallis test for continuous outcomes. Mortality rates are summarized using Kaplan-Meier estimates and were compared using Cox proportional hazards models. For stroke, MI, and recurrent heart failure, the cumulative incidence function was used to estimate the probability of each event occurring over the first year of follow-up, with death as a competing event. For each endpoint, the cumulative incidence at 1 year was estimated nonparametrically using the Fine and Gray method (21).
Adjusted analyses were also performed to examine the independent association between both AVG and LVEF and each outcome. For these analyses, LVEF was modeled as a continuous variable. This approach was prespecified based on examination of the univariate association between LVEF and mortality (which failed to identify meaningful cutpoints) and to allow for inclusion of interpretable interaction terms between the LVEF and AVG (which was retained as a dichotomous variable). Additional covariates included in the models were age, sex, current dialysis, glomerular filtration rate, diabetes, peripheral arterial disease, cerebrovascular disease, home oxygen use, chronic lung disease, prior MI, prior percutaneous coronary intervention, extent of coronary artery disease, prior coronary artery bypass graft surgery, New York Heart Association functional class IV symptoms, current tobacco use, pacemaker, implantable cardiac defibrillator, access site, moderate/severe mitral regurgitation, and moderate/severe tricuspid regurgitation. For adjusted analyses involving health status outcomes, baseline KCCQ-12 OS score was also included as a covariate.
We used logistic regression models to evaluate the association between LVEF and AVG and in-hospital outcomes and binary health status outcomes and linear regression models to identify the association between LVEF and AVG and continuous outcomes. The Generalized Estimating Equation method was used to account for within-hospital clustering to control for the possibility that patients at the same hospital would be likely to have similar response relative to patients in other hospitals (22). For mortality, we used a Cox proportional hazards model to assess the independent association between LVEF and AVG and mortality. For stroke, MI, and recurrent heart failure, differences in the adjusted incidence of these events were assessed based on hazard ratios from a Fine and Gray proportional subdistributions hazards model, with death as a competing event (21). For each model, in addition to the main effects and the covariates listed previously, interaction terms for AVG and LVEF were also evaluated. All analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, North Carolina).
Between November 2011 and June 2014, 15,938 patients underwent TAVR over 16,054 admissions and were included in the TVT Registry (Figure 1). Of these 16,054 admissions, 12,182 were identified as index admissions and were linked successfully to Centers for Medicare and Medicaid Services records. After patients who had aborted procedures or missing data for LVEF or AVG were excluded, 11,292 patients were included in the final analytic population (Figure 1).
Baseline characteristics stratified according to LVEF are shown in Table 1. The median LVEF was 23%, 42%, and 60% across the LVEF strata. Median AVG was significantly lower in patients with severe LV dysfunction and mild/moderate LV dysfunction as compared with those with preserved LV function (37 vs. 41 vs. 45 mm Hg, p < 0.001). When baseline characteristics were compared across LVEF strata, there were significant differences with respect to most comorbidities, and patients with severe LV dysfunction and mild/moderate LV dysfunction had significantly higher Society of Thoracic Surgeons mortality risk scores when compared with patients with preserved LVEF (9.7% vs. 8.0% vs. 6.6%; p < 0.001).
Baseline characteristics stratified according to mean AVG are shown in Table 2. The median AVG was 32 mm Hg in the low AVG group and 49 mm Hg in the high AVG group (p < 0.001). Median LVEF was significantly lower in patients with low AVG (55% vs. 60%; p < 0.001). Similar to the comparisons across LVEF strata, patients with low versus high AVG differed with respect to most baseline characteristics, and Society of Thoracic Surgeons mortality risk scores were significantly higher in patients with low AVG (7.6% vs. 6.9%; p < 0.001).
In-hospital outcomes stratified by LVEF and AVG are summarized in Tables 3 and 4⇓⇓, respectively. In unadjusted analyses, LV dysfunction was associated with increased length of stay (median: 7 vs. 7 vs. 6 days; p < 0.001) and a trend toward higher mortality (6.4% vs. 5.4% vs. 4.7%; p = 0.069). Patients with low AVG also tended to have worse in-hospital outcomes including higher rates of mortality (5.6% vs. 4.7%, p = 0.035) and longer length of stay (7 vs. 6 days, p < 0.001).
1-year clinical outcomes
Severe LV dysfunction was associated with higher rates of mortality (29.3% vs. 25.5% vs. 21.9%, p < 0.001) and of recurrent heart failure (19.3% vs. 17.2% vs. 12.8%, p < 0.001) at 1 year (Figures 2A and 2B, respectively). Similarly, patients with low AVG had higher rates of 1-year mortality (27.1% vs. 21.5%, p < 0.001) and of hospitalization for heart failure (19.2% vs. 11.9%, p < 0.001) when compared with patients with high AVG (Figures 3A and 3B, respectively). There was no association between baseline LVEF or AVG and 1-year rates of either stroke or MI.
When the study cohort was stratified simultaneously by both LV function and AVG, rates of 1-year mortality and heart failure were higher for patients with low AVG within each LVEF stratum (Figures 4A and 4B, Online Figure SA-1). Patients with preserved LV function and high AVG had the most favorable clinical outcomes, with 1-year rates of mortality and heart failure of 23.6% and 11.2%, respectively. Conversely, patients with severe LV dysfunction and low AVG fared the worst with rates of 1-year mortality and heart failure of 33.1% and 23.6%, respectively. After adjustment for baseline clinical factors, only low AVG was independently associated with 1-year mortality (adjusted hazard ratio [HR]: 1.21; 95% confidence interval [CI]: 1.11 to 1.32; p < 0.001) and recurrent heart failure (adjusted HR: 1.52; 95% CI: 1.36 to 1.69; p < 0.001) (Central Illustration). On the other hand, after adjustment for clinical factors and AVG, baseline LV dysfunction was no longer significantly associated with mortality (adjusted HR: 1.03 per 10 percentage point reduction in EF; 95% CI: 0.99 to 1.06; p = 0.116) or recurrent heart failure (adjusted HR: 1.03 per 10 percentage point reduction in EF; 95% CI: 0.99 to 1.07; p = 0.199). There was no evidence of effect modification with respect to LVEF or AVG on rates of mortality or heart failure at 1 year (p values for interaction terms were nonsignificant).
Further stratification of AVG (<20 mm Hg, 20 to 30 mm Hg, 30 to 40 mm Hg) demonstrated a graded relationship between reduced AVG and 1-year mortality, whereas the relationship between AVG and recurrent HF was similar at all levels of reduced AVG (Online Figure SA-2). When LVEF was analyzed as a categorical variable (<30%, 30% to 50%, >50%), the association between low AVG and 1-year mortality tended to increase with decreasing LVEF (Online Figure SA-3); however, similar to the results of our primary analysis, this effect was not statistically significant (p value for interaction = 0.222).
Health status outcomes
At 30-day follow-up, health status, as measured by the KCCQ-OS, improved across all levels of LVEF (Table 5); however, the absolute change in KCCQ-OS scores was greatest for patients with severe LV dysfunction and least for patients with normal LV function at baseline (32.8 vs. 28.1 vs. 24.0 points, respectively; p < 0.001). Patients with severe LV dysfunction were most likely to experience a substantial health status improvement (56.9% vs. 52.3% vs. 48.2%; p < 0.001), and these differences persisted in risk-adjusted analyses.
In contrast, there were no significant differences between the low and high AVG group for any of the 30-day health status outcomes. After controlling for baseline factors (Table 6), low AVG was associated with a marginally smaller improvement in the KCCQ-OS score (mean adjusted difference: -1.7 points; p = 0.018). There was no evidence of effect modification with respect to LV dysfunction or AVG on any health status outcomes (all p values for interaction were nonsignificant).
In this large study of a real-world population undergoing commercial TAVR implantation, we found that patients with either LV dysfunction or low AVG had higher 1-year rates of death and heart failure hospitalization. After adjustment for other clinical factors, however, only low AVG was associated with worse clinical outcomes, whereas the effect of LV dysfunction was no longer significant. The association between low AVG and clinical outcomes was similar regardless of baseline EF, suggesting that LVEF is not a significant modifier of the relationship between AVG and 1-year outcomes of TAVR. In contrast, LVEF had a more substantial effect on health status outcomes at 30 days; patients with lower baseline EFs were more likely to experience substantial improvement in their follow-up health status, whereas there were no significant differences in the extent of health status improvement across AVG strata.
This study both confirms and extends the results of previous research regarding the benefits of both SAVR and TAVR in patients with severe AS. Numerous studies have demonstrated that LV dysfunction as well as low AVG are associated with increased early and late mortality following SAVR (1,11,23). Among high-risk AS patients treated with TAVR, several small studies (<1,000 patients) have also suggested that low AVG is associated with reduced long-term survival after TAVR even in the presence of preserved LVEF (24,25).
Our study adds to the existing published reports in several ways. First, it is the largest study to date to examine the impact of LV dysfunction and low AVG on long-term outcomes after TAVR. As such, it is the first study with sufficient power to examine the effect of each of these measures independently, to test for interactions between EF and AVG, and to adjust for a broad range of additional comorbidities. Given the large sample size afforded by the TVT Registry, our finding that LV dysfunction was not independently associated with long-term mortality after adjusting for AVG and other factors provides important reassurance regarding the benefits of TAVR, even in patients with severe LV dysfunction. Second, as a database of real-world patients, the TVT Registry allows for the evaluation of patients who have generally been excluded from the pivotal trials (e.g., patients with severe LV dysfunction or low AVG). Because these patients are frequently encountered in contemporary practice, this study offers insight into the clinical outcomes and prognosis for such patients. Finally, our study is the first to examine the association between measures of LV dysfunction and health status outcomes. The finding that patients with severe LV dysfunction derive greater health status benefits from TAVR than patients with preserved LV function may reflect several factors including their greater health status impairment at baseline and the fact that LVEF often improves substantially among surviving patients after TAVR (7).
Our finding that low AVG, but not reduced LVEF, was associated with increased long-term mortality after TAVR likely reflects that low AVG may be an indication of reduced flow, which is often related to intrinsic myocyte dysfunction. In fact, in some studies, low SVI has been shown to be a more powerful independent predictor of post-TAVR mortality than either EF or AVG (14). Previous studies have demonstrated that patients with low-flow, low-gradient AS have evidence of myocardial fibrosis (26)—a finding that has been linked to abnormal LV remodeling and reduced compliance and filling of the LV (27,28) as well as to poorer clinical outcomes in patients with severe AS (29). On the other hand, reduced LVEF in patients with severe AS could reflect either irreversible myocardial dysfunction or afterload mismatch resulting from valvular obstruction. In the latter case, LV function may improve substantially after either surgical (2,30) or transcatheter (7) AVR, particularly when the resting AVG is high. In such cases, it is likely that the relief of valvular obstruction by AVR results in a significant decrease in afterload and allows for beneficial LV remodeling and recovery of LV function.
From a practical perspective, our findings suggest that the presence of low AVG (<40 mm Hg) may identify a cohort of AS patients, who derive less long-term benefit from TAVR. Nevertheless, it is important to recognize that neither LV dysfunction nor low AVG identifies a group of patients with sufficiently poor outcomes to preclude consideration for TAVR in the absence of other indicators of poor prognosis. Indeed, even among those patients with severely reduced LVEF and low AVG, 1-year mortality in this all-comers population was 33%. In contrast, 1-year mortality among extreme risk AS patients who were managed medically in the PARTNER B (Placement of AoRTic TraNscathetER valve) trial was nearly 50% (3). Although cross-trial comparisons should be undertaken with caution, these results, in combination with other smaller studies, suggest that TAVR can provide some degree of benefit among certain patients with both reduced EF and low AVG (13,14).
Our study should be interpreted in light of several limitations. Despite collection of extensive echocardiographic and hemodynamic data in the TVT Registry, certain important variables were not collected and hence were not included in our analyses. For example, although prior studies have demonstrated the importance of contractile reserve as a predictor of prognosis after both TAVR and surgical AVR (31,32), the presence or absence of contractile reserve was not collected in the TVT Registry and therefore, the effect of this variable could not be evaluated. Furthermore, it was not possible to calculate transvalvular flow in an accurate fashion. These calculations would have required combining data from several diagnostic tests (e.g., echocardiogram, catheterization) that were performed at different time points by different observers, thereby leading to inaccurate conclusions. Because SVI is increasingly recognized as an important measurement in the evaluation of aortic stenosis (33) as well as an important prognostic factor (14,15,34), the inability to assess the multiway interaction among transvalvular flow, AVG, and EF is a significant limitation of this analysis. Second, approximately 50% of patients had incomplete KCCQ data at 30-day follow-up. Although we assumed that these data were missing at random, it is certainly possible that there was response bias because patients with better health status may be more likely to complete the KCCQ questionnaire. Consequently, the generalizability of our findings on health status outcomes to the overall TAVR population is uncertain. Third, because follow-up outcomes were derived from administrative claims data, it is possible that some hospitalizations for heart failure were missed because of miscoding. Fourth, echocardiographic and hemodynamic data from the TVT registry are site-reported and not adjudicated via a core laboratory, the use of which has been shown to provide more reliable data (35). As such, there is likely some interpretation variability in the reported echocardiographic and hemodynamic parameters. Finally, it is likely that there were some confounding factors (e.g., frailty, pulmonary hypertension), that were not accounted for by our regression models.
Among patients undergoing TAVR, both reduced LVEF and low AVG are common and are associated with higher rates of 1-year mortality and recurrent heart failure. After adjustment for other baseline factors, only AVG was found to be a significant predictor of clinical outcomes regardless of LVEF. Although these data suggest that low AVG may identify a group of patients less likely to benefit from TAVR and should be considered when evaluating the risks and benefits of TAVR for individual patients, neither severe LV dysfunction nor low AVG alone or in combination provide sufficient prognostic discrimination to preclude treatment with TAVR in the absence of other adverse prognostic factors.
COMPETENCY IN MEDICAL KNOWLEDGE: In patients with severe aortic stenosis undergoing TAVR, an AVG <40 mm Hg is associated with worse clinical outcomes, but reduced LVEF is not.
TRANSLATIONAL OUTLOOK: Further evaluation of the factors responsible for the interactions among transvalvular flow, AVG, LVEF, and clinical outcomes could provide helpful prognostic guidance for patients undergoing TAVR.
For supplemental figures, please see the online version of this article.
This research was supported by the American College of Cardiology’s National Cardiovascular Data Registry (NCDR). The views expressed in this manuscript represent those of the author(s), and do not necessarily represent the official views of the NCDR or its associated professional societies identified at CVQuality.ACC.org/NCDR. The Society of Thoracic Surgeons/American College of Cardiology’s Transcatheter Valve Therapies Registry is an initiative of The Society of Thoracic Surgeons and the American College of Cardiology. Dr. Baron has received speaker honoraria and consulting income from Edwards Lifesciences; and consulting income from St. Jude Medical. Dr. Herrmann has received research support from Abbott Vascular, Boston Scientific, Medtronic, Siemens, Edwards Lifesciences, and St. Jude Medical; and consulting income from Siemens and Edwards Lifesciences. Dr. Szeto is a principal investigator with Edwards Lifesciences and Medtronic. Dr. Chhatriwalla has received travel reimbursement from Edwards Lifesciences, Medtronic, and St. Jude Medical. Dr. Vemulapali has received research support from Boston Scientific and the American College of Cardiology; consulting income from Premiere Inc.; and travel reimbursement from Abbott Vascular. Dr. Cohen has received research grant support from Edwards Lifesciences, Medtronic, and Boston Scientific; and consulting income from Edwards Lifesciences and Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper.
- Abbreviations and Acronyms
- aortic stenosis
- aortic valve gradient
- confidence interval
- hazard ratio
- Kansas City Cardiomyopathy Questionnaire Overall Summary Score
- left ventricular
- left ventricular ejection fraction
- myocardial infarction
- surgical aortic valve replacement
- stroke volume index
- transcatheter aortic valve replacement
- Received September 2, 2015.
- Revision received March 7, 2016.
- Accepted March 8, 2016.
- 2016 American College of Cardiology Foundation
- Elmariah S.,
- Palacios I.F.,
- McAndrew T.,
- et al.
- Passeri J.J.,
- Elmariah S.,
- Xu K.,
- et al.
- Pibarot P.,
- Dumesnil J.G.
- Hachicha Z.,
- Dumesnil J.G.,
- Bogaty P.,
- Pibarot P.
- Herrmann H.C.,
- Pibarot P.,
- Hueter I.,
- et al.
- Le Ven F.,
- Freeman M.,
- Webb J.,
- et al.
- Carroll J.D.,
- Edwards F.H.,
- Marinac-Dabic D.,
- et al.
- Arnold S.V.,
- Spertus J.A.,
- Lei Y.,
- et al.
- Arnold S.V.,
- Spertus J.A.,
- Lei Y.,
- et al.
- Reynolds M.R.,
- Magnuson E.A.,
- Lei Y.,
- et al.
- Reynolds M.R.,
- Magnuson E.A.,
- Wang K.,
- et al.
- Herrmann S.,
- Stork S.,
- Niemann M.,
- et al.
- Adda J.,
- Mielot C.,
- Giorgi R.,
- et al.
- Azevedo C.F.,
- Nigri M.,
- Higuchi M.L.,
- et al.
- Barbash I.M.,
- Minha S.,
- Ben-Dor I.,
- et al.
- Levy F.,
- Laurent M.,
- Monin J.L.,
- et al.
- Nishimura R.A.,
- Otto C.M.,
- Bonow R.O.,
- et al.
- Dayan V.,
- Vignolo G.,
- Magne J.,
- et al.