Author + information
- Received July 31, 2014
- Revision received October 17, 2014
- Accepted December 5, 2014
- Published online August 1, 2015.
- Julian W. Sacre, PhD∗,†,
- Christine L. Jellis, MD, PhD‡,
- Brian A. Haluska, PhD‡,
- Carly Jenkins, PhD‡,
- Jeff S. Coombes, PhD∗,
- Thomas H. Marwick, MD, PhD‡,§∗ ( and )
- Michelle A. Keske, PhD§
- ∗School of Human Movement and Nutrition Sciences, University of Queensland, Brisbane, Australia
- †Baker IDI Heart and Diabetes Institute, Melbourne, Australia
- ‡School of Medicine, University of Queensland, Brisbane, Australia
- §Menzies Research Institute Tasmania, University of Tasmania, Hobart, Australia
- ↵∗Reprint requests and correspondence:
Prof. Thomas H. Marwick, Menzies Research Institute Tasmania, Private Bag 23, Hobart TAS 7000, Australia.
Objectives This study sought to investigate the association of exercise intolerance in type 2 diabetes (T2DM) with skeletal muscle capillary blood flow (CBF) reserve.
Background Exercise intolerance in T2DM strongly predicts adverse prognosis, but associations with muscle blood flow independent of cardiac dysfunction are undefined.
Methods In 134 T2DM patients without cardiovascular disease, left ventricular function and contrast-enhanced ultrasound of the quadriceps (for CBF; i.e., product of capillary blood volume and velocity) were assessed at rest and immediately following treadmill exercise for peak oxygen uptake (Vo2peak). Left ventricular systolic and diastolic functional reserve indexes were derived from changes in systolic and early diastolic color tissue Doppler velocities. Cardiac index reserve and its constituents (stroke volume and chronotropic indexes) and left ventricular filling pressure (ratio of early diastolic mitral inflow and annular velocities) were also measured.
Results Vo2peak correlated with muscle CBF reserve (β = 0.16, p = 0.005) independent of cardiac index reserve and clinical covariates. This was explained by higher muscle capillary blood velocity reserve (β = 0.18, p = 0.002), rather than blood volume reserve (p > 0.10) in patients with higher Vo2peak. A concurrent association of Vo2peak with cardiac index reserve (β = 0.20, p < 0.001) appeared to reflect chronotropic index (β = 0.15, p = 0.012) rather than stroke volume index reserve (p > 0.10), although the systolic functional reserve index was also identified as an independent correlate (β = 0.16, p = 0.028). No associations of Vo2peak with diastolic functional reserve were identified (p > 0.10).
Conclusions Vo2peak is associated with muscle CBF reserve in T2DM, independent of parallel associations with cardiac functional reserve. This is consistent with a multifactorial basis for exercise intolerance in T2DM.
Exercise intolerance is now widely recognized in type 2 diabetes mellitus (T2DM) per se—that is, independent of cardiovascular disease or other comorbidities such as obesity (1–4). However, despite strong predictive power for cardiovascular and all-cause mortality (5,6), the determinants of exercise capacity in T2DM remain incompletely understood. Subclinical cardiovascular dysfunction secondary to the abnormal metabolic milieu may be central to a multifactorial etiology (1,3,4,7); however, the relative contributions of cardiac versus peripheral vascular abnormalities remain difficult to discern. Exercise intolerance as a consequence of peripheral vascular dysfunction in T2DM may reflect compromised arterial blood flow during exercise secondary to impaired endothelium-dependent vasodilation (8). However, blood flow reserve at the muscle capillary level (i.e., site of oxygen/substrate exchange) relies on microvascular function as well as on upstream hyperemia. Indeed, blunting of muscle capillary blood flow (CBF) during forearm contractions is described in T2DM patients with microvascular complications (9) and unites with slowed or reduced flow reserve during submaximal exercise (7,10). These observations certainly argue for important roles for peripheral vascular sequelae; however, associations with maximal exercise capacity in T2DM—particularly relative to left ventricular (LV) dysfunction—are unknown.
In the current study of patients with T2DM without concurrent cardiovascular disease, we sought the association of skeletal muscle CBF reserve with exercise capacity (peak oxygen uptake [Vo2peak]) independent of LV functional reserve and other potential covariates. Skeletal muscle CBF (contrast-enhanced ultrasound) (11) and LV function (echocardiography) were assessed at rest and immediately following maximal treadmill exercise.
Patients with T2DM (n = 189), aged ≥40 years, and with no history of cardiovascular, psychiatric, or other severe illness, and with no symptomatic macrovascular or microvascular complications of T2DM, were recruited from the community. Patients were required to have an ejection fraction of ≥50%, no valvular disease, and no resting or inducible wall motion abnormalities indicative of ischemia, as previously described (12,13). The study was approved by the local human research ethics committee and all patients provided written informed consent.
Fasting blood glucose, glycosylated hemoglobin (HbA1c), lipids, hemoglobin, creatinine, and a random urinary albumin-to-creatinine ratio (for microalbuminuria; i.e., ≥3.5 mg/mmol [female subjects] or ≥2.5 mg/mmol [male subjects]), were measured by standard hospital pathology laboratory protocols. Resting heart rate and blood pressure (BP) were recorded during supine rest.
Echocardiography for LV function and contrast-enhanced ultrasound for skeletal muscle CBF were performed sequentially at rest and then repeated immediately following maximal treadmill exercise testing according to the Bruce protocol. The imaging protocols were not performed simultaneously due to microbubble influence on Doppler-based echocardiographic measurements. In accordance with usual image acquisition times (14), post-exercise echocardiography was completed within 1 to 2 min after exercise (views for wall motion analysis were prioritized and completed first). Contrast-enhanced ultrasound was performed during the subsequent ∼2.5 to 3.5 min (i.e., all imaging protocols completed by ∼3.5 to 5.5 min post-exercise).
During treadmill exercise, an electrocardiogram (ECG) was recorded continuously and BP was measured during the final minute of each stage. Exercise capacity was measured by expired breath-by-breath gas analysis for Vo2peak following 20-s interval data averaging (Vmax, SensorMedics, Yorba Linda, California). Patients with exercise-induced arrhythmias or abnormal exercise BP necessitating test termination (based on American College of Cardiology/American Heart Association exercise testing guidelines) (15), or with a peak respiratory quotient <1.0 (indicating submaximal effort), were excluded. The heart rate response to exercise (chronotropic index) was quantified by heart rate reserve (peak − resting heart rate) as a percentage of age-predicted heart rate reserve (12). Antihypertensive therapy was withheld for the preceding 24 h.
Contrast-enhanced ultrasound for skeletal muscle CBF was performed according to a protocol modified from Vincent et al. (11). Ultraharmonic images (Sonos 7500, Philips Medical Systems, Andover, Massachusetts) of the quadriceps (approximately one-third of the distance from the patella to inguinal fold) were acquired in cross section at centerline transmission and received frequencies of 1.3 and 3.6 MHz, respectively. Relevant settings were mechanical index = 1.0 (sufficient to destroy microbubbles in the ultrasound beam), gain = 1, and compression = 80%. Depth and focus were optimized for each study. Images were acquired intermittently during intravenous infusion of microbubbles (Definity, Lantheus Medical Imaging, North Billerica, Massachusetts) at ∼0.15 ml/min (operator-controlled), using an internal ECG trigger timer to progressively increase the time between successive pulses (pulsing interval) from 1 to 40 RR intervals (∼1 s to ≥25 s). Video intensity was derived from images at each pulsing interval in a region of interest encompassing the deep quadriceps (MCE software, University of Virginia, Charlottesville, Virginia). Because mean video intensity of images acquired at the lowest pulsing interval (i.e., 1 RR interval, which corresponded to 910 ± 133 ms at rest and 636 ± 108 ms post-exercise) is approximately representative of contributions of larger noncapillary vessels and background signal intensity, this was digitally subtracted from the mean video intensity recorded at subsequent pulsing intervals. Mean subtracted video intensity (y) was plotted as a function of the pulsing interval (t) to generate a replenishment curve according to the formula: y = A (1 – e−βt). Here, A represents muscle capillary blood volume and β represents the muscle capillary blood velocity (11). Calculation of muscle CBF (A × β) by this technique has been validated in cardiac muscle (16). CBF reserve variables were defined by the change from rest to post-exercise.
Rest and post-exercise echocardiographic assessment of LV function was performed using a commercial ultrasound machine (Vivid 7, GE Medical, Horten, Norway). Apical views in grayscale and with tissue Doppler imaging were acquired by standard methods (17). LV end-diastolic and end-systolic volumes (modified Simpson biplane) were indexed to body surface area and used to derive stroke volume index (SVI), cardiac index (SVI × heart rate) and ejection fraction. SVI and cardiac index reserve were each determined from the rest to post-exercise change (Δ) in their respective values. Color tissue Doppler images were analyzed for peak systolic (Sm) and early diastolic (Em) tissue velocities (mean of 6 basal LV segments) using commercial software (Echopac PC, GE Medical). LV systolic and diastolic longitudinal functional reserves were assessed by the Sm and Em reserve indexes, respectively (i.e., rest to post-exercise change in tissue velocity, indexed to the resting tissue velocity according to the formula proposed by Ha et al. ).
Mitral inflow was recorded by pulsed wave Doppler, with LV filling pressure estimated from the ratio of early diastolic mitral inflow (E) and septal annular velocities (E/Em). Change in LV filling pressure was estimated by ΔE/Em, though the Eexercise/Emrest and (Eexercise − Erest)/Emrest ratios were also calculated because they have been reported to be potentially closer correlates of the absolute, or change in, pulmonary capillary wedge pressure during exercise (19).
LV mass (indexed to body surface area) was derived from 2-dimensional targeted M-mode echocardiography and calculated using the Devereux formula (20).
Data were analyzed using SPSS Statistics version 19.0 (SPSS, Inc., Chicago, Illinois) and expressed as mean ± SD or median (interquartile range), depending on normality of distribution (assessed by the Kolmogorov-Smirnov test). Categorical data were expressed as counts (percentages). Linear associations were assessed by Pearson or Spearman rank correlation coefficient, where appropriate. Multiple linear regression models were performed using the enter method. Statistical significance was defined by p < 0.05.
Of the original 189 patients, 55 required exclusion from the final analyses. Reasons included premature exercise test termination due to ECG or BP criteria (n = 3), submaximal effort indicated by respiratory quotient <1.0 (n = 17), and technical inadequacies in gas exchange data (n = 10) or ultrasound image quality (n = 25). Clinical characteristics of the final study population (n = 134) are displayed in Table 1. On the basis of established normative data for Vo2peak, mean exercise tolerance was <30th percentile for age and sex (21). The patients were mostly overweight or obese, but demonstrated relatively favorable cardiometabolic risk factor control (i.e., HbA1c, BP, and lipids) and were predominantly treated by oral hypoglycemic medications rather than insulin. Echocardiographic data (Table 2) were consistent with no overt systolic or diastolic dysfunction (based on ejection fraction, Doppler, and tissue Doppler). No patients had evidence of ischemia on the resting or exercise echocardiogram. The battery of post-exercise LV filling pressure markers could be feasibly measured in 117 patients (87%); thus, analyses involving these variables were limited to this subgroup.
Univariate correlations of Vo2peak with clinical factors and LV functional reserve
In all patients, clinical correlates of a higher Vo2peak included the following: younger age (r = −0.35, p < 0.001); male sex (r = 0.52, p < 0.001); lower body mass index (r = −0.47, p < 0.001); lower waist circumference (r = −0.31, p < 0.001); shorter duration of diabetes (r = −0.18, p = 0.041); lower HbA1c (r = −0.19, p = 0.028), lower systolic BP (r = −0.29, p = 0.001); lower resting heart rate (r = −0.24, p = 0.006); previous/current smoking (r = 0.18, p = 0.041) and higher hemoglobin (r = 0.49, p < 0.001). Of patients’ medical therapies, only β-blockade demonstrated an association with Vo2peak (r = −0.18, p = 0.038). There were borderline significant relations with resting mean arterial pressure (MAP) (r = −0.15, p = 0.085) and triglycerides (r = −0.17, p = 0.067). No correlations were observed between Vo2peak and total, low-density, or high-density cholesterol; fasting glucose; creatinine; microalbuminuria; or hypertension history.
With respect to LV functional reserve, Vo2peak correlated positively with cardiac index reserve (r = 0.26, p = 0.003), Sm reserve index (r = 0.55, p < 0.001), and the chronotropic index (r = 0.27, p = 0.002), but not with SVI reserve or Em reserve index (p > 0.10). Of the 4 markers used to estimate LV filling pressure, Vo2peak was inversely associated with post-exercise E/Em (r = −0.27, p = 0.003), Eexercise/Emrest (r = −0.22, p = 0.020), and (Eexercise – Erest)/Emrest (r = −0.21, p = 0.023), but was unrelated to ΔE/Em (p > 0.10).
Univariate correlations of Vo2peak with muscle CBF
Correlations of Vo2peak with muscle CBF parameters (rest, post-exercise, and reserve) are displayed in Figure 1. Vo2peak was positively associated with muscle CBF at rest and post-exercise (Figure 1E). A rightward shift of the linear regression line for post-exercise versus resting flow was consistent with an overall positive CBF reserve, which in turn, correlated positively with Vo2peak (Figure 1F). Examination of constituents of muscle CBF revealed that Vo2peak was related to capillary blood volume at rest and post-exercise (Figure 1A). However, no correlation was observed with capillary blood volume reserve (Figure 1B). In contrast, higher capillary blood velocity post-exercise, but not at rest, was associated with higher Vo2peak (Figure 1C). In addition, capillary blood velocity reserve was positively related to Vo2peak (Figure 1D).
Independent associations of Vo2peak with muscle CBF versus LV functional reserve
To determine whether associations of Vo2peak with skeletal muscle CBF reserve parameters were independent of LV functional reserve and clinical factors, a series of multiple linear regression models were created (Table 3). Age, sex, body mass index, systolic BP, HbA1c, diabetes duration, previous/current smoking, hemoglobin, β-blockade, and ΔMAP were included as covariates in all models. These were selected on the basis of a current relation with Vo2peak or—in the case of ΔMAP—potential covariance with LV functional or muscle CBF reserve indexes.
Model 1 featured muscle CBF reserve and cardiac index reserve, and both parameters were identified as positive independent correlates of Vo2peak. The underlying origins of these associations were investigated in model 2 via replacement of each variable by its constituents (i.e., cardiac index reserve replaced by SVI reserve and chronotropic index; muscle CBF reserve replaced by muscle capillary blood volume and velocity reserves). Consistent with the univariate analyses, this model revealed an independent association of Vo2peak with muscle capillary blood velocity reserve, but not with capillary blood volume reserve. Furthermore, the chronotropic index, but not SVI reserve, was identified as an independent correlate. In model 3, independence of LV systolic and diastolic functional reserve was investigated specifically. For this purpose, SVI reserve was replaced by Sm and Em reserve indexes, with chronotropic index and muscle capillary blood velocity reserve retained from model 2. This revealed that Vo2peak was positively and independently related to Sm, but not Em reserve index. Associations with muscle capillary blood velocity reserve and chronotropic index were still observed, although the latter was of borderline significance (p = 0.057). Because potential associations of Vo2peak with diastolic functional reserve may reflect the exercise-induced change in LV filling pressure, rather than the change in longitudinal tissue velocity, we also analyzed 4 different variations of model 3 in which Em reserve index was replaced by each of the following, in turn: post-exercise E/Em, ΔE/Em, Eexercise/Emrest, and (Eexercise – Erest)/Emrest. However, no independent associations were observed between Vo2peak and any of these 4 variables (p > 0.10) in the subgroup in whom they were measured (n = 117).
Within all models in Table 3, younger age (β range −0.13 to −0.16, p < 0.05), male sex (β range 0.33 to 0.42, p < 0.001), lower body mass index (β range −0.34 to −0.36, p < 0.001), shorter duration of diabetes (β range −0.12 to −0.13, p < 0.05), and higher hemoglobin (β range 0.20 to 0.27, p < 0.01) were all consistently identified as significant independent correlates of higher Vo2peak.
In the current study of patients with T2DM without concurrent cardiovascular disease, Vo2peak correlated positively with muscle CBF reserve independently of cardiac index reserve and other covariates—an association that appeared to be driven by muscle capillary blood velocity reserve. Although Vo2peak correlated with absolute resting and post-exercise muscle capillary blood volume, it was unrelated to capillary blood volume reserve (i.e., capillary recruitment).
Exercise capacity and skeletal muscle perfusion
Subclinical cardiac dysfunction is associated with (3), but incompletely explains, the exercise intolerance of T2DM per se (1). In keeping with the contention that this may at least partly reflect an incremental contribution from abnormalities in skeletal muscle perfusion, we identified an association of Vo2peak with muscle CBF reserve that was independent of the cardiac output response, changes in perfusion pressure (MAP), and other potential covariates. Functional impairment in the muscle microcirculation in T2DM has previously been established from acute exercise experiments applying near-infrared spectroscopy and laser Doppler techniques. Muscle blood flow augmentation has been demonstrated to be slowed (10) and—in the case of patients with poor glycemic control—of diminished magnitude (7) in response to submaximal cycling exercise. These findings unite with blunted CBF responses to submaximal forearm contractile exercise in T2DM patients with microvascular complications (9). The current study extends these observations by highlighting the potential independent role of abnormal skeletal muscle perfusion in T2DM-related exercise intolerance. Notably, a similar association has been described in a T2DM cohort complicated by peripheral artery disease (i.e., claudication onset time during treadmill walking was independently predicted by calf muscle CBF reserve) (22).
It was notable that the relation of Vo2peak with CBF reserve was driven primarily by muscle capillary blood velocity reserve, rather than blood volume reserve. This may implicate pre-capillary factors—particularly endothelium-dependent vasodilation—which is abnormal in T2DM and tightly coupled to local vascular conductance and exercise hyperemia (8). Our results may indicate that the association of endothelial dysfunction with exercise intolerance in T2DM (23) is mediated by diminished muscle capillary blood velocity reserve—a more specific determinant of patients’ capacity to increase oxygen and substrate delivery to the site of blood-myocyte exchange.
Correlations of Vo2peak with absolute values of muscle capillary blood volume both at rest and post-exercise may be explained by structural factors; specifically, differences in capillary density (which may be abnormal in T2DM) (24) between patients with low versus high Vo2peak. The apparent dissociation of Vo2peak with capillary blood volume reserve is consistent with normal exercise-mediated capillary recruitment in animal models of diabetes (25) and in uncomplicated T2DM patients (9). However, this result should also be considered in the context of ∼25% of the study cohort demonstrating a paradoxical reduction in capillary blood volume from rest to post-exercise. Given the atypical nature of this phenomenon, exploratory analyses were performed to identify potential explanations. On regression modeling, independent correlates of capillary blood volume reserve (adjusted for age, sex, systolic BP, HbA1c, diabetes duration, ΔMAP, medications, and univariate correlates; data not shown) included body mass index, total cholesterol, cardiac index reserve, capillary blood velocity reserve, and resting capillary blood volume. Of these variables, patients with negative volume reserve were distinguished from those with positive volume reserve only by higher values of resting capillary blood volume and cardiac index reserve (p < 0.001 for both; values not shown). Collectively, these findings may be interpreted to reflect confounding of some patients’ capillary blood volume measurements by technical factors—specifically, the combination of high opacification at rest and increased exercising limb inflow (secondary to high cardiac index reserve) leading to post-exercise signal saturation. This has the effect of reducing the replenishment curve’s A value, with the consequence that capillary blood volume reserve becomes artificially blunted. On the other hand, negative capillary blood volume reserve may reflect a genuine physiologic phenomenon. One possibility lies with regional heterogeneity of blood flow distribution during exercise, the relevance of which in the current study is underscored by CBF being characterized from a single cross section of 1 of numerous muscle groups activated during treadmill exercise. That these and other muscle groups would be activated to varying degrees during exercise (e.g., due to differences in gait) may also heighten regional variation and thereby confound CBF reserve measures.
Muscle CBF parameters are known to be affected by peripheral artery disease (22), which we did not screen for objectively. However, the absence of lower extremity symptoms on maximal exercise testing combined with the absence of apparent coronary artery disease on stress echocardiography, collectively point to a low probability of this condition in this cohort.
In evaluating correlates of exercise capacity, we have not considered training status, pulmonary factors or the mitochondrial dysfunction of skeletal muscle in T2DM (26). Nonetheless, our multivariate models explained up to 70% of the variance in Vo2peak. The standardized coefficients for LV functional reserve and muscle CBF reserve were relatively low in these models, indicating that a large proportion of explained variance reflected contributions from model covariates; however, because T2DM-related exercise intolerance is multifactorial, it may be expected that associations with any 1 single factor would be modest. Moreover, Vo2peak will always depend to a certain extent on nonmodifiable factors such as age and sex. Of course, the cross-sectional study design precludes assumptions of causation and the extent to which independent correlates are abnormal in T2DM will require comparisons with a nondiabetic control group.
Previous work in this field has investigated LV function and muscle blood flow during submaximal exercise (7,8,10,18). This contrasts with the current study, where cardiovascular reserve was based on measurements taken post-maximal exercise. Although each approach is associated with its own advantages and limitations, we focused on maximal exercise testing because of the robust evidence supporting prognostic significance of peak exercise capacity (5) and because of previous success in the application of post-exercise echocardiography to determine factors predictive of exercise intolerance in heart disease (27). Maximal exercise testing may also reveal deficits in cardiovascular function that remain hidden in the setting of less potent hemodynamic stresses imposed by submaximal exercise. Nonetheless, it must be acknowledged that treadmill exercise echocardiography introduces an unavoidable delay between exercise cessation and imaging that probably causes underestimation of cardiovascular reserve indexes. Although this may be particularly relevant to muscle CBF (because echocardiography preceded contrast-enhanced ultrasound in the post-exercise imaging protocol), the mean CBF reserve in this study (2.3-fold) was similar to the CBF reserve reported in T2DM patients in whom contrast-enhanced ultrasound was performed during high-intensity contractile exercise (80% of maximal strength) (9).
Some additional technical limitations require acknowledgment. Two-dimensional echocardiography may have led to underestimation of the absolute values of SVI and cardiac index due to image foreshortening. Contrast-enhanced ultrasound was performed during an operator-controlled continuous infusion, which is less robust than could have been achieved with a syringe pump. Furthermore, reliance on an ECG trigger predisposed to a heart rate–dependent pulsing interval for background subtraction, which afforded variation in the minimum velocity of excluded larger vessels. Although this velocity cutpoint would have been raised in the post-exercise condition due to RR interval shortening, this would probably have been offset by higher blood velocities in the same vessels that were excluded at rest. Adjustment for ΔMAP relied on the peak BP recorded during exercise, which is not necessarily reflective of BP at the time of post-exercise echocardiography or contrast-enhanced ultrasound.
Finally, the number of potential correlates investigated in this study—relative to the final sample size—may have led to type 1 error.
This study has demonstrated novel associations of exercise capacity with skeletal muscle CBF in patients with T2DM. In this context, exercise intolerance appears to have a multifactorial cardiovascular basis characterized by LV and peripheral vascular dysfunction. Investigation of the responsiveness of these factors to intervention trials, particularly exercise training, is warranted to confirm the mechanistic origins of these associations.
COMPETENCY IN MEDICAL KNOWLEDGE: Exercise intolerance is an important prognostic parameter in T2DM. When a history of impaired exercise tolerance is elicited, the potential causes of this symptom should be considered. Chronotropic incompetence, reduced systolic contractile reserve, and peripheral effects on muscle blood flow are all potential explanations.
TRANSLATIONAL OUTLOOK: Muscle capillary blood flow reserve is a measurable parameter that might improve the understanding of exercise intolerance. Muscle reserve may provide a new target for therapies directed at the exercise intolerance of T2DM.
This work was supported in part by a Centre of Clinical Research Excellence award from the National Health and Medical Research Council, Canberra, Australia. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Sherif Nagueh, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- blood pressure
- capillary blood flow
- ratio of early diastolic mitral inflow and septal annular velocities
- early diastolic tissue velocity
- glycosylated hemoglobin
- left ventricular
- mean arterial pressure
- systolic tissue velocity
- stroke volume index
- type 2 diabetes mellitus
- peak exercise oxygen uptake
- Received July 31, 2014.
- Revision received October 17, 2014.
- Accepted December 5, 2014.
- American College of Cardiology Foundation
- Fang Z.Y.,
- Sharman J.,
- Prins J.B.,
- Marwick T.H.
- Joshi D.,
- Shiwalkar A.,
- Cross M.R.,
- Sharma S.K.,
- Vachhani A.,
- Dutt C.
- Kingwell B.A.,
- Formosa M.,
- Muhlmann M.,
- Bradley S.J.,
- McConell G.K.
- Womack L.,
- Peters D.,
- Barrett E.J.,
- Kaul S.,
- Price W.,
- Lindner J.R.
- Bauer T.A.,
- Reusch J.E.,
- Levi M.,
- Regensteiner J.G.
- Vincent M.A.,
- Clerk L.H.,
- Lindner J.R.,
- et al.
- Sacre J.W.,
- Franjic B.,
- Jellis C.L.,
- Jenkins C.,
- Coombes J.S.,
- Marwick T.H.
- Kusunose K.,
- Motoki H.,
- Popovic Z.B.,
- Thomas J.D.,
- Klein A.L.,
- Marwick T.H.
- Gibbons R.J.,
- Balady G.J.,
- Bricker J.T.,
- et al.
- Wei K.,
- Jayaweera A.R.,
- Firoozan S.,
- Linka A.,
- Skyba D.M.,
- Kaul S.
- Jellis C.,
- Wright J.,
- Kennedy D.,
- et al.
- Ha J.W.,
- Lee H.C.,
- Kang E.S.,
- et al.
- Devereux R.B.,
- Reichek N.
- American College of Sports Medicine
- Lindner J.R.,
- Womack L.,
- Barrett E.J.,
- et al.
- Regensteiner J.G.,
- Bauer T.A.,
- Reusch J.E.
- Marin P.,
- Andersson B.,
- Krotkiewski M.,
- Björntorp P.
- Wheatley C.M.,
- Rattigan S.,
- Richards S.M.,
- Barrett E.J.,
- Clark M.G.
- Mogensen M.,
- Sahlin K.,
- Fernström M.,
- et al.