Author + information
- Received October 17, 2014
- Revision received November 26, 2014
- Accepted December 1, 2014
- Published online April 1, 2015.
- Magnus Thorsten Jensen, MD∗,†∗ (, )
- Peter Sogaard, MD∗,
- Henrik Ullits Andersen, MD†,
- Jan Bech, MD∗,
- Thomas Fritz Hansen, MD∗,
- Tor Biering-Sørensen, MD∗,‡,
- Peter Godsk Jørgensen, MD∗,
- Søren Galatius, MD∗,
- Jan Kyst Madsen, MD∗,
- Peter Rossing, MD†,§,‖ and
- Jan Skov Jensen, MD∗,‡
- ∗Department of Cardiology, Copenhagen University Hospital Gentofte, Hellerup, Denmark
- †Steno Diabetes Center, Gentofte, Denmark
- ‡Institute of Clinical Medicine, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark
- §Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark
- ‖Faculty of Health, University of Aarhus, Aarhus, Denmark
- ↵∗Reprint requests and correspondence:
Dr. Magnus Thorsten Jensen, Department of Cardiology, Copenhagen University Hospital Gentofte, Kildegårdsvej 28, 2900 Hellerup, Denmark.
Objectives The purpose of this study was to investigate if systolic myocardial function is reduced in all patients with type 1 diabetes (T1DM) or only in patients with albuminuria.
Background Heart failure is a common cause of mortality in T1DM, and a specific diabetic cardiomyopathy has been suggested. It is not known whether myocardial dysfunction is a feature of T1DM per se or primarily associated with diabetes with albuminuria.
Methods This cross-sectional study compared 1,065 T1DM patients without known heart disease from the outpatient clinic at the Steno Diabetes Center with 198 healthy control subjects. Conventional echocardiography and global longitudinal strain (GLS) by 2-dimensional speckle-tracking echocardiography was performed and analyzed in relation to normoalbuminuria (n = 739), microalbuminuria (n = 223), and macroalbuminuria (n = 103). Data were analyzed in univariable and multivariable linear regression models adjusted for confounding factors including conventional risk factors, medication, and systolic and diastolic dysfunction. Investigators were blinded to degree of albuminuria.
Results Mean age was 49.5 years, 52% men, mean glycated hemoglobin 8.2% (66 mmol/mol), mean body mass index 25.5 kg/m2, and mean diabetes duration 26.1 years. In unadjusted analyses, GLS differed significantly between T1DM patients and control subjects (p = 0.02). When stratified by degrees of albuminuria, the difference in GLS compared with control subjects was −18.8 ± 2.5% versus −18.5 ± 2.5% for normoalbuminuria (p = 0.28), versus −17.9 ± 2.7% for microalbuminuria (p = 0.001), and versus −17.4 ± 2.9% for macroalbuminuria (p < 0.001). Multivariable analyses, including clinical characteristics, diastolic and systolic dysfunction, and use of medication, did not change this relationship.
Conclusions Systolic function assessed by GLS was reduced in T1DM compared with control subjects. This difference, however, was driven solely by decreased GLS in T1DM patients with albuminuria. T1DM patients with normoalbuminuria have systolic myocardial function similar to healthy control subjects. These findings do not support the presence of specific diabetic cardiomyopathy without albuminuria.
Cardiovascular disease (CVD) is one of the most common comorbidities and causes of death in diabetes (1). Early identification of patients at particular risk is, therefore, of major importance in the day-to-day clinical setting.
Currently, the most well-known heart disease in diabetes is the premature development of coronary atherosclerosis, which leads to ischemic heart disease; however, a special subset of heart failure in diabetes has been proposed: the diabetic cardiomyopathy, which is the impairment of myocardial function without major coronary vessel disease (2,3). Diabetic cardiomyopathy is believed to be associated with nephropathy, which is a common complication in diabetes.
Type 1 diabetes mellitus (T1DM) is associated with incidence rates of heart failure comparable to a 10-year-older background population, and the risk of death due to CVD is increased 6- to 12-fold (4,5). So far, studies of a possible diabetic cardiomyopathy have been small and primarily in patients with type 2 diabetes (T2DM) (3). T1DM, however, may be a better patient category to study because the myocardial changes associated with T1DM are the result of a lack of insulin and are, to a lesser degree than in T2DM, confounded by lifestyle factors.
Speckle-tracking echocardiography is an imaging technique developed as a method to objectively quantify myocardial function and provides information about myocardial strain (6). In the clinical setting, left ventricular systolic function is usually assessed visually or by Simpson’s biplane method. Assessment of global ventricular longitudinal function using speckle-tracking imaging may, however, be able to identify discrete changes that are not detectable with conventional echocardiography. Studies in patients with heart failure and ischemic heart disease have shown that impaired global longitudinal strain (GLS) is associated with adverse events and provides incremental prognostic information beyond conventional echocardiography (7–9).
In the present study, we investigated the relationship between GLS and degrees of albuminuria in T1DM patients without known heart disease randomly selected from an outpatient clinic compared with healthy control subjects. The aim of the present study was to examine if T1DM per se is associated with reduced myocardial systolic function or if reduced myocardial function is a feature of concomitant albuminuria, that is, to examine if diabetic cardiomyopathy exists without the presence of albuminuria in T1DM.
Steno Diabetes Center and Department of Cardiology, Copenhagen University Hospital Gentofte
The Steno Diabetes Center is an integrated part of the Danish public health care system, and 90% of the patients are referred from the Copenhagen Capital Region. The total number of patients followed at Steno Diabetes Center is around 6,000, of whom 3,500 patients have T1DM and the remaining patients have T2DM. T1DM patients are followed lifelong, with outpatient visits approximately every third month.
The Department of Cardiology, Copenhagen University Hospital Gentofte, is an invasive center with a core echocardiography laboratory that performs more than 6,400 echocardiographic examinations annually.
The Thousand & 1 study is a cross-sectional cohort study of T1DM patients without known heart disease.
Invitation, screening, and inclusion of patients started April 1, 2010, and inclusion was completed April 1, 2012. Patients were eligible if they were 18 years or older, attending the outpatient clinic at Steno Diabetes Center, diagnosed with T1DM, without known heart disease, and willing to participate. Known heart disease was defined as heart failure; coronary artery disease (CAD), including previous myocardial infarction, stable angina, previous percutaneous coronary intervention, or coronary artery bypass surgery; atrial fibrillation or atrial flutter; left bundle branch block; congenital heart disease; and pacemaker or implantable cardioverter-defibrillator insertion, all of which were exclusion criteria. No financial compensation was offered to patients for their participation. The study population and study visit has been described in detail elsewhere (10). As previously shown, the included study population was representative of T1DM patients without known heart disease followed at Steno Diabetes Center without any major selection bias.
The study was performed in accordance with the second Helsinki declaration and was approved by the regional ethics committee (H-3-2009-139) and the Danish Data Protection Agency (00934-Geh-2010-003). All subjects gave written informed consent.
Prior to the echocardiographic examination, all patients received study information, signed the consent form, and filled out a questionnaire with information about lifestyle factors, including smoking, exercise, alcohol consumption, cardiorespiratory symptoms, and use of medication. Blood pressure and electrocardiogram at rest were recorded in the supine position.
A healthy, nondiabetic control population was sampled from the fifth cross-sectional survey of the CCHS (Copenhagen City Heart Study), which is currently in the inclusion phase. The CCHS is a prospective cardiovascular study of the general population (11). In the present fifth survey, subjects are invited if they have previously been invited to earlier cross-sectional surveys of the CCHS; subjects are also randomly selected through the Civil Registration System and invited to participate to include up to 5,000 subjects for the present survey. So far, around 2,500 patients have been included since 2011.
After relevant approval from the steering committee of the CCHS was obtained, a total of 200 subjects with similar age and sex distribution were included as control subjects if they had no history of CVD in the National Patient Registry; answered “No” to previous CVD on the questionnaire; did not take any cardiovascular medication; did not have diabetes; did not have known kidney disease; did not have known hypertension; and had an echocardiography performed with the examination stored in the EchoPac BT11 (GE Healthcare, Little Chalfont, United Kingdom) format, a format which was replaced by EchoPac BT12 in 2012. Subjects who met these criteria were stratified consecutively by inclusion date in 5-year age and sex strata and sampled to reflect age and sex distributions, similar to the T1DM patients. Characteristics are shown in Table 1. Echocardiograms were analyzed off-line by 1 investigator (M.T.J.) using the same methodology as for the Thousand & 1 population described in the following text.
Echocardiography was performed with a General Electric, Vivid 7 Dimension imaging system device (GE Vingmed Ultrasound AS, Horten, Norway) with a 3.5-MHz transducer in accordance with the recommendations from the European Association of Echocardiography/American Society of Echocardiography (12). Echocardiographic examinations were read and analyzed using General Electric EchoPac software (BT11). Three consecutive heart cycles were recorded. Left ventricular ejection fraction was determined by Simpson’s biplane method. Left atrial volume was determined by the recommended biplane area-length method and indexed for body surface area. Left ventricular mass was determined by the linear method and indexed for body surface area. Pulsed-wave Doppler was performed in the apical 4-chamber view, with the sample volume placed between the mitral leaflet tips to obtain diastolic mitral early (E) and atrial (A) inflow velocities and deceleration time of the E velocity wave. Pulsed-wave early diastolic tissue Doppler velocities (e′) were determined from the apical 4-chamber view at the lateral region of the mitral annulus (13).
Global left ventricular longitudinal strain
Global left ventricular longitudinal strain (GLS) was measured using 2-dimensional speckle-tracking echocardiography, where deformation of the left ventricle is determined by tracking speckles from frame to frame (Figure 1). First, the timing of aortic valve closure is defined. Thereafter, 3 points are anchored inside the myocardial tissue at the mitral valve plane as well as apically to allow for semiautomated tracking. The tracking algorithm follows the endocardium from a single frame throughout the cardiac cycle and allows for a further manual adjustment of the region of interest to ensure that all myocardial regions are included and tracked accurately throughout the cardiac cycle. Accurate tracking is ensured by visual evaluation of tracking and segmental strain curves. Apical images of the 2-, 3-, and 4-chamber views of the left ventricle are divided into 6 segments (basal, mid, and apical segments in opposing walls), and the tracking quality of each of the 18 segments are manually approved or rejected. The algorithm subsequently calculates a strain score for each apical view. The method has been described in detail elsewhere (6,14,15). For the present analysis, GLS was determined as the average of the 3 apical views, thereby providing a GLS measure for the entire left ventricle. Consequently, only patients who had information from all 3 apical views available were included in the analysis, thus excluding 28 patients (2.6%). For the included 1,065 patients, overall feasibility was 96.6%. Images were obtained at a mean frame rate of 76 frames/s.
All echocardiographic examinations were performed, read, and analyzed by Dr. Jensen, and were validated with a second opinion by another experienced imaging cardiologist. Interobserver and intraobserver variability was assessed with 25 randomly selected patients and showed good agreement with no systematic bias. For Simpson’s biplane, left ventricular ejection fraction limits of agreement were (mean of difference ±2 SD) −0.6% (−11.0 ± 10.0%) for interobserver variability and 0.6% (−9.0 ± 7.0%) for intraobserver variability (Bland-Altman plots not shown) (16), and the GLS interobserver and intraobserver variability was −0.18% (−1.9 ± 1.5%) and −0.12% (−1.5 ± 1.3%), respectively.
Information about biochemistry, such as glycated hemoglobin, p-creatinine, and albuminuric status, was collected from electronic patient files at Steno Diabetes Center from the ambulatory visit closest to study inclusion, which was maximally ±4 months from inclusion. This information was collected after analyzing the echocardiogram, ensuring that the investigators were blinded.
Urinary albumin excretion rate (UAER) was measured in 24-h sterile urine collections by enzyme immunoassay. Patients were categorized as normoalbuminuric if UAER, in 2 of 3 consecutive measurements, was <30 mg/24 h, microalbuminuric if UAER was between 30 and 300 mg/24 h, and macroalbuminuric if UAER was >300 mg/24 h. Glycated hemoglobin was measured by high-performance liquid chromatography (Variant, Bio-Rad Laboratories, Munich, Germany) (normal range 21 to 46 mmol/mol [4.1% to 6.4%]) and serum creatinine concentration by an enzymatic method (Hitachi 912, Roche Diagnostics, Mannheim, Germany). Estimated glomerular filtration rate was calculated by the MDRD (Modification of Diet in Renal Disease) method (17).
All analyses were performed with STATA version 12.1 (STATACorp LP, College Station, Texas). Categorical variables were analyzed with the chi-square test and continuous variables with analysis of variance or the Student t test.
GLS was studied in T1DM patients versus control subjects with similar age and sex distribution, and in T1DM patients by degree of albuminuria versus control subjects. Four different models were performed: an unadjusted model; a multivariable model (model 1) including baseline characteristics and adjustments for systolic and diastolic dysfunction as well as left ventricular mass index (LVMI); a multivariable model (model 2) including the same variables as model 1 but excluding patients with systolic dysfunction; and a multivariable model (model 3) including the same variables as in model 2 but also including use of medication. Baseline characteristics included age, sex, body mass index, smoking status, systolic blood pressure, diastolic blood pressure, LVMI, and systolic and diastolic dysfunction. Age and sex were included in the analyses stratified by degree of albuminuria but were not included when analyzing the T1DM patients compared with the control subjects because the control subjects were matched by age and sex. Smoking status and blood pressure are well-known risk factors of CVD, and LVMI was included because it is an overall indicator of cardiac remodeling. Systolic dysfunction was defined as left ventricular ejection fraction <45%, and diastolic dysfunction was defined as E/e′ lateral >12 because this measure has been shown to be the best indicator of diastolic dysfunction (18). To account for multiple comparisons in the baseline characteristics of between-albuminuria-group differences, a p value <0.017 (Bonferroni) was considered statistically significant in these analyses (Table 2). A p value <0.05 was considered statistically significant unless otherwise stated.
A total of 1,065 patients with complete information on GLS and albuminuria were included; 739 patients were classified as having normoalbuminuria, 223 patients as microalbuminuria, and 103 patients as macroalbuminuria. In addition, 198 sex- and age-matched healthy control subjects with GLS were included. Demographics are shown in Tables 1 and 2.
Echocardiographic baseline characteristics are presented in Tables 1 and 2. LV dimensions were significantly different between groups of albuminuria. Transmitral Doppler E/A ratio decreased with increasing degree of albuminuria. Also, tissue Doppler early diastolic velocity, e′, decreased, and an E/e′ >12 was more prevalent with increasing degree of albuminuria. In the T1DM population, 1.6% of the population had systolic dysfunction compared with none of the control subjects.
GLS in T1DM versus control subjects
Women had greater strain values than men; in the Thousand & 1 study: women −19.1 ± 2.6% versus men −17.6 ± 2.4%, p < 0.001; in control subjects: women −19.7 ± 2.3% versus men −17.9 ± 1.9%, p < 0.001.
The population mean values of GLS were −18.3% for the T1DM patients and −18.8% for the control subjects, and this difference was significant (p = 0.02) (Table 3). In multivariable model 1, the estimate was essentially unchanged. The differences persisted when excluding patients with systolic dysfunction, as shown in model 2. When including use of medication (model 3), none of the medications reached statistical significance in relation to GLS. The inclusion of the medication variables, however, attenuated the observed difference in GLS between T1DM patients and control subjects (Table 3).
GLS in T1DM by degree of albuminuria versus control subjects
The patient population was stratified by degrees of albuminuria. Values were −18.5 ± 2.5% for patients with normoalbuminuria, −17.9 ± 2.7% for patients with microalbuminuria, and −17.4 ± 2.9% for patients with macroalbuminuria. As shown in Table 4, GLS did not differ between control subjects and T1DM patients with normoalbuminuria, but differences were highly significant between control subjects and patients with microalbuminuria (p = 0.001) and macroalbuminuria (p < 0.001). In the multivariable models (models 1, 2, and 3) the estimates changed only slightly when including baseline characteristics, systolic and diastolic function, and use of medication. Use of any of the medications was not independently associated with GLS. In a model where use of medication was included as a dichotomous variable, group differences were similar to the multivariable model 3 where medications were included individually (control subjects vs. normoalbuminuria, p = 0.39; control subjects vs. microalbuminuria, p = 0.031; and control subjects vs. macroalbuminuria, p = 0.021).
Diastolic dysfunction and GLS
As shown in the multivariate models 1, 2, and 3 (Tables 3 and 4), diastolic dysfunction was independently associated with decreased GLS. Inclusion of diastolic dysfunction in the multivariable models did not change the relationship between albuminuria and reduced GLS in patients with T1DM.
In the present study, 1,065 T1DM patients without known heart disease were included from the outpatient clinic at Steno Diabetes Center and compared with 198 nondiabetic control subjects from the currently ongoing fifth cross-sectional survey of the Copenhagen City Heart Study.
The concept of a specific diabetic cardiomyopathy has received much attention (3,19,20). An important question to answer in this respect is whether diabetes per se is associated with myocardial impairment or if impaired myocardial function is present only in relation to concomitant complications, such as albuminuria and kidney disease. The findings of the present study suggest that T1DM patients without albuminuria have a systolic function that is similar to a healthy background population. In the present study, the observed difference in systolic myocardial function measured by GLS between patients with T1DM and healthy control subjects was driven alone by reduced GLS in T1DM patients with albuminuria. Thus, these findings do not support the concept of a specific diabetic cardiomyopathy in patients without albuminuria.
In the day-to-day clinical setting, the finding that myocardial systolic function is reduced only in T1DM patients with albuminuria underscores the importance of preventing, diagnosing, and treating T1DM patients with kidney disease. It is well known that the presence of albuminuria is a major risk factor for the development of heart disease in T1DM, and albuminuria in T1DM is associated with increased mortality. Furthermore, it appears that patients with T1DM but without albuminuria have a prognosis that is similar to the background population (21–23). This is in line with the findings from the present study. As shown in this study, 30% of the patients with microalbuminuria and 15% of the patients with macroalbuminuria did not receive an angiotensin-converting enzyme inhibitor or angiotensin II antagonist. The prevention of diabetic kidney disease by ensuring proper renoprotective and cardioprotective medication may be a feasible way of preventing the reduction in myocardial function in patients with T1DM.
In the present study, an association between albuminuria and reduced myocardial function was demonstrated. Here, a clear dose–response relationship was found, as patients with macroalbuminuria had lower GLS compared with patients with microalbuminuria. Although diastolic dysfunction was associated with degree of albuminuria, the reduction in GLS was independent of diastolic dysfunction. This finding is interesting, as diastolic dysfunction is often described as being the first sign of diabetic cardiomyopathy (3). The association between albuminuria and risk of CVD and mortality has been shown in several population studies, and albuminuria has been shown to be 1 of the most important prognostic factors in diabetes (21,23,24). The association between albuminuria, myocardial impairment, and adverse events seems particularly strong for heart failure (25), a disease entity which, in terms of complications with diabetes, is receiving increasing attention (26,27). Albuminuria is believed to be a marker of general vascular damage (28), and microalbuminuria has been shown to also be predictive of ischemic heart disease in the general population (29). In line with the increased incidence of diabetes and its complications, there has been increased focus on a possible diabetic cardiomyopathy, which is myocardial dysfunction associated with diabetic nephropathy in the absence of CAD (2,19,20,30). Several studies support the findings of altered myocardial function in patients with diabetes and in patients with kidney disease (31). The underlying factors relate to changes in the renin-angiotensin-aldosterone system (RAAS) and arterial blood pressure, and treatment is, therefore, primarily focused on RAAS inhibition (32,33). Connelly et al. (34) have shown that hyperglycemia and the presence of kidney disease/changes in the RAAS seem instrumental in the development of diabetic cardiomyopathy. Here, we found that these results can be translated from studies in animals to clinical studies in ambulatory patients.
GLS is a relatively new measure of systolic function and measures deformation, that is, systolic contraction. Deformation imaging, or strain, has been developed to detect subtle changes in systolic function. Speckle-tracking echocardiography was, therefore, used to detect possible reduction in systolic function not detectable by conventional echocardiography in patients with different degrees of albuminuria compared with a control population. Other speckle-tracking measures, such as strain rate or radial and circumferential strain, could possibly expand our knowledge about cardiac mechanics in T1DM (35); GLS, however, appears to show the best reproducibility across different vendors, and may, therefore, be the most robust marker of changes in systolic function in the day-to-day clinical setting (36). GLS has previously been shown to be able to predict prognosis in patients with myocardial infarction beyond conventional echocardiography (7) and may be used as an addition to conventional echocardiography in an extended cardiac evaluation to detect discrete changes in myocardial function.
First, the current population is the largest cohort of T1DM patients studied with speckle-tracking echocardiography to date; it is, therefore, possible to detect changes and study the influence of several factors on myocardial function not previously reported.
A limitation in the present study may be that the presence of subclinical CAD was unknown, and the few patients with systolic dysfunction may represent undiagnosed CAD. The Thousand & 1 population, however, represents real-life ambulatory patients without known heart disease. It is also worth noting that these patients are followed every third month at the Steno Diabetes Center throughout their lives, and clinical signs or a history of cardiac symptoms will prompt further investigation. If these patients develop heart disease, there is a good chance that it will be diagnosed. Also, indicating against significant CAD is the finding that patients with normoalbuminuria had a systolic function comparable to the control subjects. We performed multivariable analyses both adjusting for and excluding patients with systolic dysfunction, and this did not change the overall results. A functional test or further studies of possible CAD would, however, have strengthened the findings further.
In the present study of 1,065 T1DM patients without known heart disease compared with 198 nondiabetic healthy control subjects, we found that systolic function assessed by GLS is reduced in T1DM compared with control subjects. The difference between groups, however, is driven solely by decreased GLS in patients with albuminuria in that there is no difference between control subjects and T1DM patients with normoalbuminuria.
These findings suggest that reduced systolic myocardial function in T1DM is associated primarily with the presence of albuminuria. T1DM patients with normoalbuminuria have systolic myocardial function similar to healthy control subjects.
COMPETENCY IN MEDICAL KNOWLEDGE: Type 1 diabetes is not associated with systolic myocardial impairment per se. Systolic myocardial impairment in type 1 diabetes occurs in association with the presence of albuminuria. Type 1 diabetes patients without albuminuria have a systolic myocardial function similar to that of healthy controls.
TRANSLATIONAL OUTLOOK: Further studies of the pathogenesis and prevention of heart disease in diabetes are needed with special emphasis on patients with albuminuria.
The authors are indebted to the staff and patients of the Steno Diabetes Center for their participation and contribution to the Thousand & 1 study. The controls were examined with a cardiac ultrasound machine made available by GE Healthcare.
This work was supported by The European Foundation for the Study of Diabetes/Pfizer European Programme 2010 for Research into Cardiovascular Risk Reduction in Patients with Diabetes and The Danish Heart Foundation (12-04-R90-A3840-22725). Additional funding has been received from The Torben & Alice Frimodts Foundation, Carl & Ellen Hertz’ Legat til Dansk Læge-og Naturvidenskab, and The Beckett Foundation. Dr. Sogaard has received research grants from, and has given talks for GE Health Care, Biotronik, and Bayer. Dr. Andersen has been an advisory board member for Abbott; has stock in Novo Nordisk; and the Steno Diabetes Center is an affiliate of Novo Nordisk. Dr. Rossing has board memberships in AstraZeneca/Bristol-Myers Squibb, Eli Lilly, Janssen, Novo Nordisk, and Astellas; has received grants and/or has grants pending from Novo Nordisk, Novartis, and Abbott; has received payment for lectures by AstraZeneca/Bristol-Myers Squibb, Novartis, and Sanofi; and has stock in Novo Nordisk. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery disease
- cardiovascular disease
- global longitudinal strain
- left ventricular mass index
- type 1 diabetes mellitus
- type 2 diabetes mellitus
- Received October 17, 2014.
- Revision received November 26, 2014.
- Accepted December 1, 2014.
- 2015 American College of Cardiology Foundation
- Soedamah-Muthu S.S.,
- Fuller J.H.,
- Mulnier H.E.,
- Raleigh V.S.,
- Lawrenson R.A.,
- Colhoun H.M.
- Gorcsan J. 3rd.,
- Tanaka H.
- Ersbøll M.,
- Valeur N.,
- Mogensen U.M.,
- et al.
- Cho G.-Y.,
- Marwick T.H.,
- Kim H.-S.,
- Kim M.-K.,
- Hong K.-S.,
- Oh D.-J.
- Stanton T.,
- Leano R.,
- Marwick T.H.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.
- Delgado V.,
- Mollema S.A.,
- Ypenburg C.,
- et al.
- Kasner M.,
- Westermann D.,
- Steendijk P.,
- et al.
- Orchard T.J.,
- Secrest A.M.,
- Miller R.G.,
- Costacou T.
- Groop P.-H.,
- Thomas M.C.,
- Moran J.L.,
- et al.
- Soedamah-Muthu S.S.,
- Fuller J.H.,
- Mulnier H.E.,
- Raleigh V.S.,
- Lawrenson R.A.,
- Colhoun H.M.
- Braunwald E.
- Borch-Johnsen K.,
- Feldt-Rasmussen B.,
- Strandgaard S.,
- Schroll M.,
- Jensen J.S.
- Forbes J.M.,
- Cooper M.E.
- Andrésdóttir G.,
- Jensen M.L.,
- Carstensen B.,
- et al.