OUP user menu

Long-term analysis of left ventricular ejection fraction in patients with stable multivessel coronary disease undergoing medicine, angioplasty or surgery: 10-year follow-up of the MASS II trial

Cibele Larrosa Garzillo, Whady Hueb, Bernard J. Gersh, Eduardo Gomes Lima, Paulo Cury Rezende, Alexandre Ciappina Hueb, Ricardo D'Oliveira Vieira, Desiderio Favarato, Alexandre Costa Pereira, Paulo Rogério Soares, Carlos Vicente Serrano Jr., José Antônio Franchini Ramires, Roberto Kalil Filho
DOI: http://dx.doi.org/10.1093/eurheartj/eht201 3370-3377 First published online: 4 July 2013


Background Assuming that coronary interventions, both coronary bypass surgery (CABG) and percutaneous coronary intervention (PCI), are directed to preserve left ventricular function, it is not known whether medical therapy alone (MT) can achieve this protection. Thus, we evaluated the evolution of LV ejection fraction (LVEF) in patients with stable coronary artery disease (CAD) treated by CABG, PCI, or MT as a post hoc analysis of a randomized controlled trial with a follow-up of 10 years.

Methods Left ventricle ejection fraction was assessed with transthoracic echocardiography in patients with multivessel CAD, participants of the MASS II trial before randomization to CABG, PCI, or MT, and re-evaluated after 10 years of follow-up.

Results Of the 611 patients, 422 were alive after 10.32 ± 1.43 years. Three hundred and fifty had LVEF reassessed: 108 patients from MT, 111 from CABG, and 131 from PCI. There was no difference in LVEF at the beginning (0.61 ± 0.07, 0.61 ± 0.08, 0.61 ± 0.09, respectively, for PCI, CABG, and MT, P = 0.675) or at the end of follow-up (0.56 ± 0.11, 0.55 ± 0.11, 0.55 ± 0.12, P = 0.675), or in the decline of LVEF (reduction delta of −7.2 ± 17.13, −9.08 ± 18.77, and −7.54 ± 22.74). Acute myocardial infarction (AMI) during the follow-up was associated with greater reduction in LVEF. The presence of previous AMI (OR: 2.50, 95% CI: 1.40–4.45; P = 0.0007) and during the follow-up (OR: 2.73, 95% CI: 1.25–5.92; P = 0.005) was associated with development of LVEF <45%.

Conclusion Regardless of the therapeutic option applied, LVEF remains preserved in the absence of a major adverse cardiac event after 10 years of follow-up.

Clinical Trial Registration URL: http://www.controlled-trials.com. Registration number ISRCTN66068876.

  • Left ventricular ejection fraction
  • Coronary artery disease
  • Coronary bypass surgery
  • Percutaneous coronary intervention
  • Medical therapy

See page 3339 for the editorial comment on this article (doi:10.1093/eurheartj/eht297)


Studies of ageing of the cardiovascular system in healthy subjects, with focus on left ventricular ejection fraction (LVEF), are controversial. Some authors demonstrated that systolic function usually remains preserved, with worsening of diastolic function13 due mainly to adaptative mechanisms, such as myocyte hypertrophy and changes in diastolic dynamics.

However, Cheng et al.,4 studying left ventricular function changes in patients without known cardiovascular disease, found a paradoxical increase in LVEF with ageing. Another study5 demonstrated a reduction in LVEF with ageing in patients with and without diabetes, after ruling out the presence of myocardial ischaemia.

On the other hand, in the population with coronary artery disease (CAD), many of whom have associated comorbidities, the evolution of LVEF is unclear. It is assumed that ventricular function deterioration could occur, if associated with the presence of continued but unalleviated chronic myocardial ischaemia.

Since ventricular function is the most important predictor of long-term survival in patients with CAD, regardless of the number of arteries affected or the degree of arterial stenosis,610 all therapeutic strategies that confer additional protection for the ischaemic myocardium are relevant in an attempt to improve long-term prognosis. In this setting, revascularization procedures, percutaneous coronary intervention (PCI) and coronary artery bypass graft (CABG), have the potential to contribute to the reduction of myocardial ischaemia and, in this way, possibly prevent progressive impairment of LVEF.

A post hoc substudy from the (Optimal Medical Therapy With and Without Percutaneous Coronary Intervention) COURAGE trial11 compared the occurrence of the primary endpoint [death or acute myocardial infarction (AMI)] in patients with CAD randomized to optimal medical therapy (OMT) with or without PCI, according to the presence and extent of ischaemia at baseline. Surprisingly, regardless of the ischaemic extent, even in the subgroup of patients with moderate-to-severe ischaemia, the addition of PCI to OMT failed to demonstrate any additional benefit in long-term prognosis. Further study on the effectiveness of available treatments for CAD in patients with documented myocardium ischaemia of greater extension and severity, the ongoing ISCHEMIA trial,12 will compare the efficacy of different therapeutic strategies is patients with moderate-to-severe ischaemia. Additionally, the role of revascularization procedures in regard to their potential to protect the function of the ischaemic myocardium is not well established. In fact, there are very few studies that aimed to investigate, in depth, the role of such interventions on ventricular function during a long period of follow-up.13,14 Other trials comparing CABG and PCI strategies,1517 or conservative vs. interventional therapies,1821 focused on primary endpoints, such as mortality, occurrence of acute myocardial infarction, or the need of additional revascularization.

The aim of this study was to evaluate the changes in LVEF, by sequential echocardiography, in patients with chronic multivessel coronary heart disease and initially preserved ventricular function, who underwent three therapeutic options (CABG, PCI, or MT alone) participants of the MASS II trial, during 10 years of follow-up. The second Medical, Angioplasty, or Surgery Study (MASS II) is a single-centre study designed to compare the long-term effects of MT, stent angioplasty, or surgical strategies among patients with multivessel CAD with stable angina and preserved ventricular function who are appropriate candidates for all three therapies.20,21


MASS II patient selection

Briefly, patients with angiographically documented proximal multivessel coronary stenosis of >70% by visual assessment and documented ischaemia were considered for inclusion. Patients were enrolled and randomized if the surgeons, attending physicians, and interventional cardiologists agreed that revascularization could be attained by either strategy.20,21

Patients gave written informed consent and were randomly assigned to a treatment group. The Ethics Committee of the Heart Institute (InCor) of the University of São Paulo Medical School in São Paulo, Brazil, approved the trial, and all the procedures were performed in accordance with the Helsinki Declaration.

Clinical criteria for exclusion included mandatory revascularization, ventricular aneurysm that required surgical repair, LVEF <40%, a history of PCI, CABG, congenital, or valvular heart disease; left main coronary artery stenosis of 50% or more, or another co-existing condition that was a contraindication to CABG or PCI.

Treatment intervention

In MASS II, all the patients were placed on an optimal medical regimen after randomization until the end of follow-up. Patients were randomized to continue with aggressive MT alone or to undergo PCI or CABG in addition to MT.

Investigators were required to perform optimum coronary revascularization. For patients assigned to PCI, the procedure was available within 3 weeks after randomization and was performed according to a standard protocol.22 Devices used for catheter-based therapeutic strategies included stents, lasers, directional atherectomy, and balloon angioplasty. For patients assigned to CABG, the procedures were available within 12 weeks after randomization, and revascularization performed with standard surgical techniques,23 using saphenous vein grafts, internal mammary arteries, and other arterial conduits. No off-pump CABG was performed.


Adverse and other clinical events were tracked from randomization. Patients were evaluated and angina symptoms were graduated24 with follow-up visits every 6 months for at least 10 years at Heart Institute. Patients underwent the treadmill exercise test according to a modified Bruce protocol, electrocardiography and routine blood tests on regular basis.

Myocardial infarction was defined as the presence of significant new Q waves in at least two ECG leads or symptoms compatible with MI associated with creatine kinase-MB fraction concentrations that were more than three times the upper limit of the reference range.

The vital status of each patient was ascertained on 28 February 2011. The predefined primary endpoints were the incidence of total mortality, Q-wave MI, or refractory angina that required revascularization. Secondary endpoints included angina status, death due to a cardiac cause, and cerebrovascular accident.

Left ventricular ejection fraction assessment

Most participants of the MASS II trial underwent transtoracic echocardiography with colour Doppler before randomization. Of those, left ventricular function was evaluated at two different time periods: the first, at the beginning of the study, and the second, after 10 years of follow-up. All echocardiographic parameters assessed were predefined and images were analysed in a core lab, by expert physicians.

Left ventricular ejection fraction was measured by the biplane method (Simpson) when the endocardial border of the left ventricle was well-defined and whenever regional wall-motion abnormalities were present, or alternatively by the Teichholz method.25

Statistical analysis

Baseline characteristics were summarized for all patients as percentages for categorical variables and as means with standard deviations for continuous variables. Comparisons between means of the groups used Student's t test for parametric26 and the Mann–Whitney test for non-parametric variables.

The means of three or more groups were compared by one-way ANOVA, followed by the Bonferroni multiple comparison test for parametric variables. For non-parametric variables, we used the Kruskal–Wallis test, followed by multiple comparisons based on Dunn's test. Tests were two-sided.

The evaluation of homogeneity between proportions was performed using the χ2 or the Fisher exact test.26 To evaluate the behaviour of groups along the evaluations (initial and 10 years) analysis of variance with repeated measures was used.27

An additional analysis was performed using logistic regression to evaluate the association of several variables to determine LVEF dysfunction, defined as LVEF ≤45% in 10 year survivors.

All data were analysed according to the intention-to-treat principle and values of P < 0.05 were considered statistically significant. Statistical analysis was performed with SPSS 15.0 for Windows.


Patients characteristics

Between May 1995 and May 2000, 611 eligible patients who met all entry criteria were randomly assigned to one of three therapeutic strategies: PCI, MT, or CABG. After a mean follow-up of 10.32 (±1.43) years, 422 patients were alive and 350 patients had their LVEF reassessed by echocardiography (108 patients from the MT group, 111 from CABG, and 131 from PCI). For several reasons, 72 patients could not have their LVEF reassessed, most commonly due to patient refusal, geographic inaccessibility and severe non-cardiac illness.

Randomization created balanced treatment groups with respect to important prognostic characteristics (Table 1), except for the angina status (less frequent in the MT group).

View this table:
Table 1

Baseline characteristics of the 350 patients with left ventricular ejection fraction reassessed according to treatment in MASS II

PCI (n = 131)CABG (n = 111)MT (n = 108)P-value
Demographic variables
 Age, years58.42 ± 9.0458.10 ± 7.6857.87 ± 8.760.886
 Age ≥65 years (%)30.5322.52250.346
 Male (%)63.3668.4770.370.485
 Current or past smoker (%)37.6135.4532.820.738
Clinical history and status
 History of MI (%)50.3843.2437.960.152
 History of hypertension (%)58.7857.6649.070.275
 Treated diabetes mellitus (%)24.4336.0434.260.106
 History of angina (%)92.3789.1980.560.018
Laboratory values
 Total cholesterol (mmol/L)5.92 ± 1.275.58 ± 1.145.74 ± 1.060.485
 LDL cholesterol (mmol/L)3.95 ± 1.143.77 ± 1.063.80 ± 0.880.521
 HDL cholesterol (mmol/L)0.95 ± 0.260.93 ± 0.230.95 ± 0.210.213
 Triglycerides (mmol/L)5.04 ± 2.534.50 ± 2.074.91 ± 2.640.082
Positive treadmill test (%)55.8353.4649.490.59
Angiographic profile
 Double-vessel disease (%)41.9843.2445.370.870
 Triple-vessel disease (%)58.0256.7654.630.870
 Double-vessel with proximal LAD disease (%)78.1877.0885.710.579
 Coronary collateral circulation (%)38.5842.5743.300.736
  • LDL, low-density lipoprotein; HDL, high-density lipoprotein; LAD, left anterior descending coronary artery. Some patients had both angina and positive treadmill tests. Unless otherwise indicated, data are mean ± SD.

Among patients eligible at 10.32 years to undergo LVEF evaluation, baseline characteristics of those who underwent LVEF assessment (n = 350) were compared with those who did not (n = 72). All variables analysed were similar in prevalence, except for diabetes, which was more frequent in patients without LVEF reassessment (Table 2). The need of additional revascularization and new AMI was similar between the groups.

View this table:
Table 2

Comparison of baseline characteristics and events between patients with and without left ventricular ejection fraction measurements

VariableLVEF available (n = 350)LVEF missing (n = 72)P-value
Male sex (%)67.4%73.6%0.374
Age (years)57.96 ± 8.760 ± 90.073
History of hypertension (%)55.14%65.3%0.146
Treated diabetes (%)31.4%47.2%0.013
Previous AMI (%)44.2837.50.353
Current or past smoker (%)35.1431.90.701
Double-vessel disease (%)42.8547.20.532
Triple-vessel disease57.1452.8
Proximal LAD disease9086.10.445
Total cholesterol (mmol/L)5.77 ± 1.165.84 ± 1.420.649
LDL cholesterol (mmol/L)3.83 ± 1.033.90 ± 1.160.630
HDL cholesterol (mmol/L)0.96 ± 0.230.96 ± 0.260.828
Triglycerides (mmol/L)4.86 ± 2.565.09 ± 3.360.554
PCI treatment (%)37.4229.20.278
CABG treatment (%)31.4240.3
MT treatment (%)31.1430.6
New AMI (%)12.286.940.193
Additional PCI (%)15.718.330.104
Additional CABG (%)13.718.330.213
  • LDL, low-density lipoprotein; HDL, high-density lipoprotein; LAD, left anterior descending coronary artery; AMI, acute myocardial infarction. Unless otherwise indicated, data are mean ± SD.

Treatment outcomes

The overall major adverse cardiac and cerebrovascular events at 10-year follow-up of the patients who had their LVEF reassessed, according to one of three therapeutic strategies are shown in Table 3. The groups did not differ with respect to the occurrence of acute myocardial infarction. However, they differ regarding the need of additional revascularization (less frequent in the CABG group).

View this table:
Table 3

Major adverse cardiac events at 10 years

PCI (n = 131) (%)CABG (n = 111) (%)MT (n = 108) (%)P-value
AMI13 (9.92)9 (8.11)16 (14.81)0.255
Additional PCI36 (27.48)10 (9.01)15 (13.89)<0.001
Additional CABG15 (11.45)4 (3.60)31 (28.70)<0.001
Combined events64 (48.45)23 (20.72)62 (57.41)<0.001
CVA3 (2.29)7 (6.31)6 (5.56)0.277
  • AMI, acute myocardial infarction; CVA, cerebrovascular accident; combined events are the sum of AMI, additional PCI or CABG.

Left ventricular ejection fraction assessment

The echocardiographic assessment identified similarity of LVEF in the three groups, both in the beginning and at the end of the 10-year follow-up (P = 0.675, Table 4). Regardless of the treatment strategy, the three groups experienced the same pattern and magnitude of ventricular function variation (P= 0.641), with a decline of LVEF (P< 0.001), represented by the reduction delta (−7.2 ± 17.13, −9.08 ± 18.77, and −7,54 ± 22.74 for PCI, CABG, and MT, respectively, Table 4 and Figure 1).

View this table:
Table 4

Left ventricular ejection fraction evolution according the treatment assigned

LVEF baseline0.61 ± 0.070.61 ± 0.080.61 ± 0.090.675
LVEF 10 years0.56 ± 0.110.55 ± 0.110.55 ± 0.120.675
Reduction delta (%)−7.2 ± 17.13−9.08 ± 18.77−7.54 ± 22.740.631
  • Reduction delta = 100 × (LVEF 10 year – LVEF baseline)/LVEF baseline.

Figure 1

Left ventricular ejection fraction at baseline and the 10th year, according to treatment assigned: P = 0.675: mean left ventricular ejection fraction similar among groups in the beginning and end of the follow-up; P = 0.641: similar evolution among the groups; P < 0.001: left ventricular ejection fraction decrease over time

Analysis of other variables identified a modest decline in LVEF overall and consistent across subgroups, except for AMI during the follow-up. Among those patients the extent of the reduction in LVEF was statistically significantly greater (Table 5).

View this table:
Table 5

Reduction of left ventricular ejection fraction according to the presence of several variables

Variable (n)Reduction delta (%)P-value
Male (235)−8.71 ± 20.000.278
Female (115)−6.23 ± 18.36
<65 years (258)−7.51 ± 19.480.336
≥65 years (92)−9.00 ± 19.56
Diabetes (109)−7.27 ± 18.960.779
Non-diabetes (241)−8.18 ± 19.75
Smoking−4.30 ± 18.130.089
Non-smoking−8.77 ± 18.15
Previous AMI−9.25 ± 14.520.248
No previous AMI−6.82 ± 16.13
Double-vessel without proximal LAD (30)−9.29 ± 17.630.727
Double vessel with proximal LAD disease (122)−6.78 ± 20.05
Triple-vessel disease (198)−8.37 ± 19.46
Presence of coronary collateral circulation (144)−8.10 ± 18.550.825
Absence of coronary collateral circulation (206)−7.75 ± 20.15
Additional PCI (61)−8.30 ± 18.970.884
No Additional PCI (289)−7.81 ± 19.62
Additional CABG (50)−11.15 ± 23.080.152
No additional CABG (300)−7.36 ± 18.81
New AMI (38)−18.29 ± 21.220.001
No AMI during follow-up (312)−6.63 ± 18.91
  • Reduction delta = 100 × (LVEF 10 year– LVEF baseline)/ LVEF baseline.

When those variables were correlated with the development of ventricular dysfunction, defined as LVEF ≤45%, the presence of previous AMI and AMI during the follow-up were associated with a greater chance of worsening of ventricular function (Figure 2). Other demographic and anatomic factors were not different between those with and without a decline in LVEF.

Figure 2

Association of several variables and development of left ventricular ejection fraction dysfunction (LVEF ≤45%). Double/triple indicates double vs. triple-vessel disease; Col. Circ., presence of coronary collateral circulation; MT/Revasc., medical treatment vs. revascularization (either PCI or CABG), according to treatment assigned from randomization; Add. PCI, additional PCI; Add CABG, additional CABG.

Diastolic function was re-evaluated after 10 years of follow-up in 300 patients who had this measure obtained at baseline. One hundred and ten patients had diastolic dysfunction at baseline and 90% of those maintained this dysfunction at re-evaluation. On the other hand, 190 patients had normal diastolic function at baseline: in 34 patients (17.89%) this measure remained preserved, and 155 patients (82.11%) presented diastolic dysfunction in the end of the follow-up (47, 59, and 49 patients respectively, in the MT, PCI, and CABG group, with no difference in evolution according to treatment applied, P = 0.331).

Of the 350 patients with LVEF reassessed after 10 years of follow-up, 252 had ischaemia evaluated with the treadmill exercise test according to a modified Bruce protocol. There was no difference in the prevalence of positive tests, irrespective of treatment applied (Table 6). Besides, patients with ischaemia demonstrated by treadmill tests presented the same evolution of ventricular function, compared with the patients with no ischaemia at the end of the follow-up (Table 6).

View this table:
Table 6

Ischaemia assessment with treadmill test after 10 years follow-up

Ischaemic changes presentIschaemic changes absentP-value
PCI (n)17790.728
MT (n)1756
CABG (n)1667
Reduction delta−5.18 ± 17.83−7.27 ± 18.310.468
  • Reduction delta = 100 × (LVEF 10 year– LVEF baseline)/ LVEF baseline.


This single institution study evaluated long-term evolution of left ventricular function, which was initially preserved, in patients with multivessel CAD who underwent one of three therapeutic strategies (PCI, CABG, or MT alone), participants of the MASS II randomized trial. Final results have shown that, regardless of the treatment option, LVEF remained preserved at the end of follow-up, albeit with an overall significant modest decline. This was observed irrespective of the coronary angiographic anatomy at the time of baseline and the need of additional revascularization.

However, patients affected by AMI had a worse outcome, with a greater reduction of LVEF in comparison with all other subgroups. Moreover, patients with previous AMI or with AMI during the follow-up had a greater chance of developing ventricular dysfunction, characterized as LVEF ≤45%. Also, the need of CABG during the follow-up was marginally associated with the development of LVEF ≤45% in the multivariated model (P = 0.05).

Considering that our population is composed by patients with severe CAD and other comorbidities, and that abnormalities in endothelium-dependent vasodilation have been described in epicardial vessels and microcirculation in patients with known CAD, as well as in those with only risk factors for atherosclerosis,2830 it can be inferred that both revascularization strategies (PCI and CABG) were of little relevance in the protection of the ischaemic myocardium, since ventricular function remained preserved in the absence of AMI. It is possible that the endothelial dysfunction provoked by the epicardial vessel atherosclerosis in our patients could be responsible for the slight decrease in LVEF observed in all subgroups, irrespective of the treatment option and the presence of other risk profiles.

In fact, the results shown by LVEF analysis demonstrate that, regardless of the therapeutic option, ventricular function was significantly impaired only when there was a history of AMI (prior to randomization or during the follow-up). One possible explanation for this result is that the subgroup with previous AMI included patients with LVEF slightly >45%, so that any decline in ventricular function, as a continuous variable, would led to LVEF value <45%. In this subgroup, as well as in all other patients in the MASS II trial, a slight decrease in LVEF was observed independently of the occurrence of AMI during the follow-up, but in the presence of this event the magnitude was greater.

In this same way, investigators of the Bypass Angioplasty Revascularization Investigation (BARI) trial evaluated the LVEF (initially preserved) in 1220 subjects with multivessel coronary heart disease, 5 years after randomization for balloon PCI or CABG. They found a similarity in LVEF at that time period between the groups. However, attention was not drawn to the slight reduction in ventricular function observed in the comparison of baseline values of LVEF and the 5th year evaluation.13 Despite the similarity of those results with the present study, the BARI trial did not analyse a group with optimal medical treatment alone. In MASS II, LVEF of patients from the MT group demonstrated the same pattern of changes as those who underwent revascularization procedures.

Another subset analysis of 100 of the 988 patients from the SoS trial (Stent or Surgery trial) evaluated the LVEF 30 days, 4 and 10 years after CABG or PCI with bare metal stents. They found an initial improvement in ventricular function after 30 days in both groups that was sustained at 4 and 10 year's revaluation. It is important to note, however, that those results represent only one of the 53 sites of the trial and were confined to only 68 of the original 100 patients from that specific site.14

The COURAGE nuclear substudy31 also assessed ventricular function measurements at baseline and 6–18 months after interventions. Results identified a similarity in rest and post-stress LVEF between patients in OMT, with or without PCI. Despite the short follow-up period, those results indicated that, similar to the findings of the present study, PCI did not confer any additional protection to the ventricular function. However, the nuclear substudy of COURAGE included only 314 of the 2287 participants of the original trial and additionally, measurements of ventricular function were missing in 141 patients at baseline and in 26 at follow-up. It is possible that the population selected for this substudy does not represent the entire population of the COURAGE trial.

In the present study, of the 422 patients alive with a mean follow-up of 10 years and eligible for LVEF reassessment, echocardiography was not performed in 17% (72 patients). Although the prevalence of diabetes was greater in the group without revaluation of the ventricular function, this variable did not impact on poorer evolution of ventricular function in those with LVEF reassessed, and the variables responsible for worst outcome (AMI prior to randomization and during the follow-up) were similar between patients with and without LVEF revaluated. This would suggest that results presented here, i.e. the absence of a difference according to initial treatment assignment can probably be extrapolated to those patients who did not undergo LVEF reassessment.

Noteworthy, this analysis of measurements performed at 10 years did not include any patient who died before that time. These results therefore apply only to patients alive at 10 years. It is possible that patients with lower LVEF died before our evaluation.

Recent randomized clinical trials confirm equivalent long-term clinical outcomes for stable CAD patients treated with intensive medical management with or without coronary revascularization.18,19 In the MASS II trial, patients in CABG and PCI groups were under medication therapy as aggressive as patients of the MT group. The current findings indicate that an initial conservative strategy is effective for the prevention of MACE, as well as for the preservation of ventricular function in patients with stable CAD. Besides, it is possible that the intensive treatment for all other comorbidities could protect ventricular function of those patients. Additionally, the potential benefit of revascularization procedures over ventricular function could be lost after 10 years follow-up, considering graft patency and stent stenosis, as well as progression of atherosclerosis in native arteries.

Clinical implications and conclusions

Since LVEF is a major predictor of prognosis in patients with stable CAD, the impact of strategies on ventricular function is of interest and relevance.6,7,9 In this way, revascularization techniques were believed to be of greater importance.

The MASS II trial has already shown similar survival of MT patients, when compared with PCI and CABG after a follow-up of 10 years.21 The current study demonstrated that initial medical treatment was equivalent to revascularization procedures in respect to long-term evolution of LVEF. Also, we have found that the occurrence of AMI was the only variable associated with deterioration of LVEF.

Thus, aggressive MT and lifestyle prescriptions with comprehensive risk factor control are valuable and should not be underestimate in the treatment of patients with stable multivessel CAD.


This work was supported by a research grant from the Zerbini Foundation, São Paulo, Brazil. There is no relationship with industry. Funding was provided by the Zerbini Foundation.

Conflict of interest: none declared.


Medical writing support was provided by Ann Conti Morcos of MorcosMedia during the preparation of this paper, supported by Zerbini Foundation.


View Abstract