- Email: [email protected]
Influence of Heart Rate Reduction on Doppler Myocardial Imaging Parameters in a Small Animal Model
Ultrasound in Med. & Biol., Vol. 35, No. 1, pp. 30 –35, 2009 Copyright © 2008 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/09/$–see front matter
doi:10.1016/j.ultrasmedbio.2008.08.006
● Original Contribution INFLUENCE OF HEART RATE REDUCTION ON DOPPLER MYOCARDIAL IMAGING PARAMETERS IN A SMALL ANIMAL MODEL C. WEYTJENS,*† J. D’HOOGE,‡ S. DROOGMANS,*† A. VAN DEN BERGH,‡ B. COSYNS,*† T. LAHOUTTE,† P. HERIJGERS‡ and G. VAN CAMP*† *Department of Cardiology, UZ Brussel, Brussels; †In Vivo Cellular and Molecular Imaging (ICMI) Laboratory, Faculty of Medicine and Pharmacy, Vrije Universiteit Brussel (VUB), Brussels; and ‡Department of Cardiovascular Diseases Research, Katholieke Universiteit Leuven (KULeuven), Leuven, Belgium (Received 21 May 2008; revised 22 July 2008; in final form 6 August 2008)
Abstract—In small animals studies, sick animals often have a significant reduction in heart rate while under anesthesia. The influence of heart rate reduction on Doppler myocardial imaging (DMI) parameters is not known. The aim of the present study was to assess the effect of heart rate reduction on DMI parameters in a small animal model. Twenty-four rats underwent transthoracic echocardiography at baseline and during the administration of ivabradine IV. In all rats, left ventricular (LV) systolic velocity, strain and strain rate were measured in the anteroseptal and inferolateral wall segments from short axis views. In 12 rats (group A), M-mode analysis was also performed for assessment of global LV function. In the other 12 rats (groups B), contractility was quantified invasively using the end-systolic pressure-volume relation (ESPVR) and the preload recruitable stroke work (PRSW). During ivabradine, administration heart rate decreased by 18% in group A (p < 0.001) and 36% in group B (p < 0.001). There was a slight increase in LVEDD and LVESD, with no change in cardiac output or LV ejection fraction. During ivabradine administration, DMI parameters did not change significantly in any group. No significant correlation between DMI parameters and heart rate (r2 ⴝ 0.05) or ejection time (r2 ⴝ 0.14) could be found. The absence of changes in contractility was confirmed by the absence of change in PRSW and end-systolic elastance (Ees). In conclusion, moderate heart rate reduction did not influence DMI measurements in this specific rat model. Therefore, in the interpretation of DMI data when performing small animal studies, moderate heart rate reduction does not need to be taken into account. (E-mail: [email protected]) © 2008 World Federation for Ultrasound in Medicine & Biology. Key Words: Echocardiography, Doppler myocardial imaging, Small animals, Ivabradine, Heart rate.
In small animals with developing cardiac diseases, we often see a more significant decrease in heart rate during anesthesia compared with healthy controls. The difference in heart rate between normal and sick rats can be as high as 30 –35% (Mihm et al. 2001; Yoon et al. 2005; Weytjens et al. 2008). This makes the interpretation of noninvasive measurements for LV function, especially tissue Doppler parameters, difficult. Indeed velocity and strain rate are time-dependent parameters. They might in theory be influenced by changes in heart rate. Furthermore, changes in heart rate can influence LV contractility by changing preload and by the staircase (treppe) phenomenon (Katz 2006). The influence of heart rate on LV filling and left ventricular contractility has already been studied extensively but little is known about the effect of heart rate on DMI parameters (Bombardini 2005). Therefore, the aim
INTRODUCTION Doppler myocardial imaging (DMI) has become a feasible and accurate tool in the quantitative assessment of regional LV function. Recent studies have demonstrated that, even in small animals, DMI can be used to assess LV function (Hirano et al. 2005; Sebag et al. 2005; Weytjens et al. 2006). Using DMI in small animals is very challenging because they not only have small hearts but also very fast heart rates. In the conscious adult rat, the heart rate is approximately 400 bpm. Depending on the anesthetic drug, heart rate usually decreases during anesthesia (Weytjens et al. 2006; Stein et al. 2007; Droogmans et al. 2008). Address correspondence to: Caroline Weytjens, Department of Cardiology, UZ Brussel, Laarbeeklaan 101, 1090 Brussels, Belgium. E-mail: [email protected] 30
Doppler imaging and HR reduction ● C. WEYTJENS et al.
of our study was to assess the effect of a pharmacological decrease in heart rate using an If channel blocker in a healthy small animal model. To ensure that LV contractility was not affected, invasive measurements of contractility were performed in a subgroup of animals. METHODS Study design A total of 24 adult male Wistar rats (10 weeks old) were studied before and during IV administration of ivabradine. Twelve animals (391 ⫾ 24 g) underwent echocardiography with DMI analyses (group A) 10 minutes after administration of pentobarbital 50 mg/kg IP. The 12 remaining animals (397 ⫾ 29 g) were assigned to invasive measurements (group B) under anesthesia with urethane 600 mg/kg and alpha-chloralose 160 mg/kg IP. Echocardiography with DMI analysis was performed in group B after stabilization of hemodynamic parameters. The study was approved by the local Institutional Care and Animal Research Committee. All animals were treated in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Ivabradine Ivabradine was dissolved in physiological saline to obtain a solution of 1 mg/mL. Ivabradine was administered IV by continuous drip of 0.1 mL/min via a tail vein in group A (noninvasive protocol) and via a jugular vein in group B (invasive protocol) until a heart rate reduction of at least 15% was obtained. Ivabradine infusion was then stopped and echocardiographic images where acquired within 30 min. Echocardiographic evaluation After administration of the anesthetic drug, chest hair was shaved and animals were positioned in left lateral decubitus for the echocardiographic examination. Transthoracic echocardiography was performed using a Vivid 7 (GE VingMed, Horten, Norway) with a linear 13-MHz probe (i13L) in all animals before and during administration of ivabradine. The ECG was monitored continuously by fixing the electrodes to the limbs. Two-dimensional guided M-mode tracings were recorded from parasternal short-axis views at the midpapillary level. M-mode measurements were performed offline using EchoPAC PC (GE Vingmed, version 3.1.6). The thickness of the interventricular septum, posterior wall and left ventricular (LV) dimensions were measured according to the edge-to-edge method. Measurements of three consecutive cycles during expiration were averaged. To calculate LV volumes and EF, the ellipsoid formula Pi*D3/6a (D ⫽ diameter of the LV; a ⫽ ellip-
31
ticity factor) was used. We used an ellipticity factor of 1/3 as D ⫽ L/3 (L ⫽ length of the ellipsoid). Color Doppler myocardial images were recorded in the parasternal short-axis view at the midventricular level. High frame rate (⬎300 fps) was obtained by reducing view width. The Nyquist limit was set as low as possible (while avoiding aliasing) to include the low myocardial velocities (between 9 and 15 cm/s). Velocity, strain and strain rate measurements were performed offline using dedicated software (SPEQLE, Catholic University Leuven, Belgium). The sample volume was placed in the middle of the anteroseptal and inferolateral wall. A strain estimation length of 0.8 mm was used. Timings of the beginning and ending of the ejection phase were obtained using the electrocardiogram and the velocity trace as previously demonstrated (Urheim et al. 2000). At least six consecutive cycles were used for the offline analysis. The reproducibility and repeatability of the deformation parameters has recently been assessed and published by our group (Weytjens et al. 2006). Invasive measurements Twelve animals (group B) underwent invasive assessment of hemodynamics before and during administration of ivabradine. The animals were placed on a controlled heating pad to maintain body temperature. A tracheotomy was performed to insert a 25-g cannula, and ventilation was started using a pressure-controlled respirator (end-inspiratory pressure, 12 mm Hg; end-expiratory pressure, 2 mm Hg; Hugo Sachs Elektronik). The animals were placed under a microscope (Zeiss, Götingen, Germany) and the jugular vein was cannulated for infusion of hypertonic saline or ivabradine. Via the carotid artery, a 2-F, micromanometer conductance catheter (Millar Instruments, Houston, TX, USA) was then inserted in the LV cavity guided by online pressure and volume signals. After hemodynamic stabilization heart rate, end-diastolic and end-systolic pressures were assessed. The abdomen was opened via a small incision just below the diaphragm to enable temporary preload reduction by directly compressing the inferior vena cava using a cotton-tipped stick. To quantify contractility we used the end-systolic pressure-volume relation (ESPVR) and the preload recruitable stroke work (PRSW). Endsystolic elastance (Ees) was derived from the linear regression of several ESPVR loops obtained during pressure manipulation (Katz AM 2006). The parallel conductance attributed to the tissues surrounding the LV cavity was estimated by injection of a bolus of 30% saline into the jugular vein. Volume calibrations of the measured conductance were performed at the end of each experiment with known blood volume. Analysis was performed with PVAN 2.9 (Millar Instruments).
32
Ultrasound in Medicine and Biology
Volume 35, Number 1, 2009
Table 1. Influence of heart rate on M-mode parameters Group A (n ⫽ 12)
Heart rate (bpm) LVEDD (mm) LVESD (mm) SV (mL) CO (mL/min) EF (%)
Rest
Ivabradine
p-value
350 ⫾ 24 7.2 ⫾ 0.7 4.4 ⫾ 0.5 0.31 ⫾ 0.08 106 ⫾ 28 76 ⫾ 4
273 ⫾ 31 7.9 ⫾ 0.6 4.8 ⫾ 0.5 0.40 ⫾ 0.09 108 ⫾ 24 78 ⫾ 4
⬍0.001 0.003 0.05 0.002 0.79 0.46
LVED(S)D ⫽ Left ventricular end diastolic (systolic) parameter; SV ⫽ stroke volume; CO ⫽ cardiac output; EF ⫽ ejection fraction.
Statistical analysis All data are expressed as the mean ⫾ SD. Comparison between groups was performed using an unpaired Student’s t-test. Comparison between repeated measures was performed using a paired Student’s t-test test. Statistical significance was defined as p ⬍ 0.05. For pvalues ⬍ 0.000 a cut-off value of ⬍ 0.001 was chosen. RESULTS In group A, all DMI datasets were used for offline analysis. In group B, one rat died during the protocol, and in two animals DMI trace quality was too low for adequate quantification. Heart rate reduction There was a significant reduction in heart rate during ivabradine administration in both groups (Tables 1 and 2). Heart rate was reduced by 18% in group A and by 36% in group B (p ⬍ 0.02). The ejection time only increased significantly in group B (Table 2). Figure 1 represents the heart rate and ejection time of each rat before and during ivabradine administration. Influence of heart rate on conventional echo measurements Left ventricular M-mode analysis was only performed in the noninvasive protocol (group A) because
Fig. 1. Heart rate and ejection time before and during administration of ivabradine in group A (grey lines; open circles) and group B (black lines; solid circles).
the conductance catheter interfered with the posterior endocardial delineation. There was a significant increase in LV end-diastolic diameter (EDD), but no change in LVEF and CO (Table 1). Influence of heart rate on DMI parameters The mean values for systolic velocity, strain and strain rate in the anteroseptal and inferolateral walls before and during administration of ivabradine are given in Table 2. Absolute values for systolic velocity, strain and strain rate were significantly lower in ventilated animals compared with nonventilated animals. During induced bradycardia, these parameters re-
Table 2. Influence of heart rate on DMI parameters Group A (n ⫽ 12)
Heart rate (bpm) Ejection time (ms) Anteroseptal Velocity (cm/s) Strain (%) Strain rate (1/s) Inferolateral Velocity (cm/s) Strain (%) Strain rate (1/s)
Group B (n ⫽ 9)
Rest
Ivabradine
p-value
Rest
Ivabradine
p-value
317 ⫾ 23 78 ⫾ 10
262 ⫾ 25 82 ⫾ 13
⬍0.001 0.18
408 ⫾ 55 59 ⫾ 8
263 ⫾ 62 83 ⫾ 24
⬍0.001 ⬍0.02
⫺0.9 ⫾ 0.3 39 ⫾ 10 11.0 ⫾ 3.0
⫺1.1 ⫾ 0.4 42 ⫾ 11 12.0 ⫾ 2.9
0.26 0.50 0.47
⫺0.9 ⫾ 0.3 21 ⫾ 9 9.9 ⫾ 3.9
⫺1.0 ⫾ 0.5 28 ⫾ 10 8.8 ⫾ 3.7
0.83 0.09 0.49
3.6 ⫾ 0.4 53 ⫾ 14 16.5 ⫾ 1.2
3.6 ⫾ 0.5 46 ⫾ 8 14.8 ⫾ 2.6
0.46 0.10 0.2
2.8 ⫾ 0.8 25 ⫾ 10 13.8 ⫾ 4.7
3.0 ⫾ 1.1 35 ⫾ 12 14.2 ⫾ 5.8
0.30 0.12 0.61
Doppler imaging and HR reduction ● C. WEYTJENS et al.
33
Fig. 3. Correlation between change in heart rate and ejection time and change in strain rate during ivabradine administration (group A ⫹ group B).
ative dP/dt and ED and ES pressures changed significantly during the administration of ivabradine (Table 3). DISCUSSION
Fig. 2. Velocity, strain and strain rate of the inferolateral wall before and during administration of ivabradine in group A (grey lines, open circles) and group B (black lines, solid circles).
mained unchanged in both groups. Figure 2 shows the individual responses of velocity, strain and strain rate during ivabradine infusion for the inferolateral wall. The wide range in absolute values is because groups A and B are represented in the same figure. No significant correlation between velocity, strain or strain rate and heart rate or ejection time was found. Figure 3 shows the relation between the change in strain rate and the change in heart rate and ejection time (r2 ⫽ 0.05 and 0.14, respectively). Influence of heart rate on invasive hemodynamics There was no significant change in PRSW and Ees during heart rate reduction. However, positive and neg-
This study demonstrates that, although contractility of the LV myocardium remains constant, a decrease in heart rate, induced by ivabradine, has no significant effect on velocity, strain or strain rate parameters in this small animal model. We can therefore conclude that neither velocity nor strain rate are influenced by moderate changes in heart rate.
Table 3. Influence of heart rate on invasive measurements Group B (n ⫽ 11)
Heart rate (bpm) PRSW EES ⫹dP/dt (mm Hg/s) ⫺dP/dt (mm Hg/s) LVEDP (mm Hg) LVESP (mm Hg)
Rest
Ivabradine
p-value
383 ⫾ 35 68.7 ⫾ 18.1 1.6 ⫾ 0.7 6534 ⫾ 2647 ⫺7864 ⫾ 2482 5.2 ⫾ 3.3 109.9 ⫾ 17.1
271 ⫾ 33 63.7 ⫾ 15.4 1.6 ⫾ 1.1 5233 ⫾ 2825 ⫺6315 ⫾ 2758 6.7 ⫾ 4.0 95.4 ⫾ 19
⬍0.001 0.12 0.67 0.001 0.001 0.006 0.001
PRSW ⫽ Preload recruitable stroke work; EES ⫽ end systolic elastance.
34
Ultrasound in Medicine and Biology
When performing small animal studies, heart rate often significantly decreases during anesthesia (Rottman et al. 2003; Stein et al. 2007). Ideally, echocardiography would be performed in conscious rats. However, when performing echocardiographic studies in conscious rodents, human manipulation in lab environment can markedly increase heart rate (Rottman et al. 2003; Stein et al. 2007). Furthermore, the use of an anesthetic drug is necessary to keep the animal immobilized during the echocardiographic examination. Our group recently analyzed LV function in an experimental rat model for type I diabetes (Weytjens et al. 2008). In this study, during anesthesia, a significant difference in heart rate was noted between control and diabetic rats (315 ⫾ 29 bpm vs. 244 ⫾ 21 bpm in diabetes), making the interpretation of the data difficult. Several authors have encountered the same problem in experimental cardiomyopathy, but until now little attention has been given to this issue (Mihm et al. 2001; Yoon et al. 2005). To date, only a few reports have studied the effect of heart rate on DMI parameters (Weidemann et al. 2002; Boettler et al. 2005). Both studies evaluated the effect of an increase in heart rate on specific strain rate measurements. Weideman et al. (2002) demonstrated that, in a closed-chest pig model, pacing the hearts up to 180 bpm did not influence strain rate significantly. Strain on the other hand significantly decreased during atrial pacing. In a stepwise multiple regression analysis, strain rate correlated best with contractility (measured by ⫹dP/dt) and strain with stroke volume (measured by M-mode derived volumes). The authors concluded that strain can be related to increased heart rate and altered loading conditions, whereas strain rate was relatively independent of heart rate, reflecting better LV contractility. Because atrial pacing can induce significant changes in LV filling and LV contractility, we believe that a pacing model is not the ideal method to assess the effect of heart rate changes on DMI parameters. Furthermore, dP/dt is a rate-dependent parameter and should increase with increasing heart rate (Appleton 1991). A better way to assess LV contractility would be to look at the pressure–volume relation and to measure PRSW (Lee et al. 2003; Bombardini 2005). Boettler et al. (2005) measured velocity and strain rate in a large number of infants with a wide range of resting heart rates and could not find any significant correlation between velocity or strain rate and heart rate. This suggests that, even with a physiological increase in heart rate, tissue Doppler parameters are not affected. This study unfortunately was unable to correlate their results with invasive measurements, and no information about preload, afterload or contractility was given.
Volume 35, Number 1, 2009
To assess the effect of heart rate reduction on parameters for LV function, beta-blocking agents cannot be used because they are well known to influence LV contractility. Ivabradine is a potent If channel blocker that is able to induce a reduction in heart rate without modifying myocardial contractility or mean arterial pressure (Colin et al. 2003; Borer 2004). This drug was therefore chosen for evaluating the effect of heart rate reduction on deformation parameters in the present study. Our experimental study confirms the absence of dependence on heart rate of velocity and strain rate when relative bradycardia of 18 –36% is induced. To ensure that the decrease in heart rate did not influence LV function, we measured preload recruitable stroke work (PRSW), the gold standard technique for evaluation of LV contractility (Katz AM 2006). In contrast with previous authors, we demonstrated a significant decrease in dP/dt in response to ivabradine. This is not surprising because the change in pressure is expressed over the ejection time period, which was significantly prolonged during ivabradine administration. In contrast to the study by Weideman et al. (2002), we did not observe a significant change in strain. Weideman et al (2002) measured a significant decrease in stroke volume during pacemaker-induced tachycardia at 160 and 180 bpm. The lower strains, therefore, reflected the decrease in total deformation caused mainly by the decrease in ejection time. During bradycardia, one would expect an increase in systolic strain. In our study, although we could measure a slight increase in strain, the difference was not statistically significant. This can be explained by the fact that the change in heart rate that was obtained during our study protocol was less significant (18 –36 % decrease vs. 30 – 40 % increase). Limitations Anesthetic drugs can have an influence on heart rate and myocardial function (Stein et al. 2007). The results of our study, therefore, need to be interpreted with regard to the anaesthetic drug. Moreover, a different type of anaesthesia was used for analysis of DMI parameters in both groups. This was done because in the invasive protocol the animals required analgesic drugs in addition to narcotic drugs. This explains why heart rate was more reduced in group B than in group A and why lower absolute values for velocity, strain and strain rate were found in group B. Because we were especially interested in the relative changes in the measurements in response to ivabradine, the type of anaesthesia used is of minor importance. We were only able to measure deformation in short axis images and thus only analyzed radial function. The use of DMI in the analysis of longitudinal function is
Doppler imaging and HR reduction ● C. WEYTJENS et al.
much more difficult because artefacts and alignment with the beam are often a problem. We cannot exclude the possibility that longitudinal function is not influenced by heart rate reduction in rats and can contribute to significant changes in LV function. However, this is unlikely because invasive measurements remained unchanged. No inotropic stimulation was performed, which would have given the opportunity to show that, with restoration of heart rate but at higher contractility, deformation parameters change significantly. This has, however, already been demonstrated (Weidemann et al. 2002; Sebag et al. 2005). CONCLUSION This study shows that a moderate heart rate reduction does not significantly influence DMI measurements in this specific rat model. Therefore, in the interpretation of DMI data in small animal studies, heart rate reduction does not need to be taken into account, given limited heart rate and loading changes. Acknowledgments—Steven Droogmans is an Aspirant of the Research Foundation–Flanders (Belgium) (FWO). Tony Lahoutte is a Senior Clinical Investigator of the Research Foundation–Flanders (Belgium) (FWO). The research at ICMI is funded by the Inter-University Attraction Poles Programme–Belgian State–Belgian Science Policy.
REFERENCES Appleton CP. Influence of incremental changes in heart rate on mitral flow velocity: Assessment in lightly sedated, conscious dogs. J Am Coll Cardiol 1991;17:227–236. Boettler P, Hartmann M, Watzl K, Maroula E, Schulte-Moenting J, Knirsch W, Dittrich S, Kececioglu D. Heart rate effects on strain and strain rate in healthy children. J Am Soc Echocardiogr 2005; 18:1121–1130. Bombardini T. Myocardial contractility in the echo lab: Molecular, cellular and pathophysiological basis. Cardiovasc Ultrasound 2005; 3:000 – 000. Borer JS. Drug insight: If inhibitors as specific heart-rate-reducing agents. Nat Clin Pract Cardiovasc Med 2004;00:103–109. Colin P, Ghaleh B, Monnet X, Su J, Hittinger L, Giudicelli JF, Berdeaux A. Contributions of heart rate and contractility to myocardial oxygen balance during exercise. Am J Physiol Heart Circ Physiol 2003;284:H676 –H682. Droogmans S, Lauwers R, Cosyns B, Roosens B, Franken PR, Weytjens C, Bossuyt A, Lahoutte T, Schoors D, Van Camp G.
35
Impact of anesthesia on valvular function in normal rats during echocardiography. Ultrasound Med Biol 2008;00:(ahead of print). Hirano T, Asanuma T, Azakami R, Okuda K, Ishikura F, Beppu S. Noninvasive quantification of regional ventricular function in rats: Assessment of serial change and spatial distribution using ultrasound strain analysis. J Am Soc Echocardiogr 2005;18:907–912. Katz AM. Physiology of the heart. C.H.I.P.S., 2006. Lee W, Huang W, Yu W, Chiou K, Ding PY, Chen C. Estimation of preload recruitable stroke work relationship by a single-beat technique in humans. Am J Physiol Heart Circ Physiol 2003;284: H744 –H750. Mihm MJ, Seifert JL, Coyle CM, Bauer JA. Diabetes related cardiomyopathy time dependent echocardiographic evaluation in an experimental rat model. Life Sci 2001;69:527–542. Rottman JN, Ni G, Khoo M, Wang Z, Zhang W, Anderson ME, M. EC. Temporal changes in ventricular function assessed echocardiographically in conscious and anesthetized mice. J Am Soc Echocardiogr 2003;16:1150 –1157. Sebag I, Handschumacher M, Ichinose F, Morgan J, Hataishi R, Rodrigues A, Guerrero J, Steudel W, Raher M, Halpern E, Derumeaux G, Bloch K, Picard M, Scherrer-Crosbie M. Quantitative assessment of regional myocardial function in mice by tissue Doppler imaging: Comparison with hemodynamics and sonomicrometry. Circulation 2005;111:2611–2616. Stein AB, Tiwari S, Thomas P, Hunt G, Levent C, Stoddard MF, Tang XL, Bolli R, Dawn B. Effects of anesthesia on echocardiographic assessment of left ventricular structure and function in rats. J Basic Res Cardiol 2007;102:28 – 41. Urheim S, Edvardsen T, Torp H, Angelsen B, Smiseth OA. Myocardial strain by Doppler echocardiography validation of a new method to quantify regional myocardial function. Circulation 2000;102:1158 – 1164. Weidemann F, Jamal F, Kowalski M, Kukulski T, D’hooge J, Bijnens B, Hatle L, De Scheerder I, Sutherland GR. Can strain rate and strain quantify changes in regional systolic function during dobutamine infusion, B-blockade, and atrial pacing. Implications for quantitative stress echocardiography. J Am Soc Echocardiogr 2002; 15:416 – 424 Weytjens C, Cosyns B, D’hooge J, Gallez C, Droogmans S, Lahoute T, Franken PR, Van Camp G. Doppler myocardial imaging in adult male rats: Reference values and reproducibility of velocity and deformation parameters. Eur J Echocardiogr 2006;7:411– 417. Weytjens C, Franken PR, D’hooge J, Droogmans S, Cosyns B, Lahoutte T, Van Camp G. Doppler myocardial imaging in the diagnosis of early systolic left ventricular dysfunction in diabetic rats. Eur J Echocardiogr 2008;9:326 –333. Yoon Y-S, Uchida S, Masuo O, Cejna M, Park J-S, Gwon H-C, Kirchmair R, Bahlman F, Walter D, Curry C, Hanley A, Isner JM, Losordo DW. Progressive attenuation of myocardial vascular endothelial growth factor expression is a seminal event in diabetic cardiomyopathy: Restoration of microvascular homeostasis and recovery of cardiac function in diabetic cardiomyopathy after replenishment of local vascular endothelial growth factor. Circulation 2005;111:2073–2085.
doi:10.1016/j.ultrasmedbio.2008.08.006
● Original Contribution INFLUENCE OF HEART RATE REDUCTION ON DOPPLER MYOCARDIAL IMAGING PARAMETERS IN A SMALL ANIMAL MODEL C. WEYTJENS,*† J. D’HOOGE,‡ S. DROOGMANS,*† A. VAN DEN BERGH,‡ B. COSYNS,*† T. LAHOUTTE,† P. HERIJGERS‡ and G. VAN CAMP*† *Department of Cardiology, UZ Brussel, Brussels; †In Vivo Cellular and Molecular Imaging (ICMI) Laboratory, Faculty of Medicine and Pharmacy, Vrije Universiteit Brussel (VUB), Brussels; and ‡Department of Cardiovascular Diseases Research, Katholieke Universiteit Leuven (KULeuven), Leuven, Belgium (Received 21 May 2008; revised 22 July 2008; in final form 6 August 2008)
Abstract—In small animals studies, sick animals often have a significant reduction in heart rate while under anesthesia. The influence of heart rate reduction on Doppler myocardial imaging (DMI) parameters is not known. The aim of the present study was to assess the effect of heart rate reduction on DMI parameters in a small animal model. Twenty-four rats underwent transthoracic echocardiography at baseline and during the administration of ivabradine IV. In all rats, left ventricular (LV) systolic velocity, strain and strain rate were measured in the anteroseptal and inferolateral wall segments from short axis views. In 12 rats (group A), M-mode analysis was also performed for assessment of global LV function. In the other 12 rats (groups B), contractility was quantified invasively using the end-systolic pressure-volume relation (ESPVR) and the preload recruitable stroke work (PRSW). During ivabradine, administration heart rate decreased by 18% in group A (p < 0.001) and 36% in group B (p < 0.001). There was a slight increase in LVEDD and LVESD, with no change in cardiac output or LV ejection fraction. During ivabradine administration, DMI parameters did not change significantly in any group. No significant correlation between DMI parameters and heart rate (r2 ⴝ 0.05) or ejection time (r2 ⴝ 0.14) could be found. The absence of changes in contractility was confirmed by the absence of change in PRSW and end-systolic elastance (Ees). In conclusion, moderate heart rate reduction did not influence DMI measurements in this specific rat model. Therefore, in the interpretation of DMI data when performing small animal studies, moderate heart rate reduction does not need to be taken into account. (E-mail: [email protected]) © 2008 World Federation for Ultrasound in Medicine & Biology. Key Words: Echocardiography, Doppler myocardial imaging, Small animals, Ivabradine, Heart rate.
In small animals with developing cardiac diseases, we often see a more significant decrease in heart rate during anesthesia compared with healthy controls. The difference in heart rate between normal and sick rats can be as high as 30 –35% (Mihm et al. 2001; Yoon et al. 2005; Weytjens et al. 2008). This makes the interpretation of noninvasive measurements for LV function, especially tissue Doppler parameters, difficult. Indeed velocity and strain rate are time-dependent parameters. They might in theory be influenced by changes in heart rate. Furthermore, changes in heart rate can influence LV contractility by changing preload and by the staircase (treppe) phenomenon (Katz 2006). The influence of heart rate on LV filling and left ventricular contractility has already been studied extensively but little is known about the effect of heart rate on DMI parameters (Bombardini 2005). Therefore, the aim
INTRODUCTION Doppler myocardial imaging (DMI) has become a feasible and accurate tool in the quantitative assessment of regional LV function. Recent studies have demonstrated that, even in small animals, DMI can be used to assess LV function (Hirano et al. 2005; Sebag et al. 2005; Weytjens et al. 2006). Using DMI in small animals is very challenging because they not only have small hearts but also very fast heart rates. In the conscious adult rat, the heart rate is approximately 400 bpm. Depending on the anesthetic drug, heart rate usually decreases during anesthesia (Weytjens et al. 2006; Stein et al. 2007; Droogmans et al. 2008). Address correspondence to: Caroline Weytjens, Department of Cardiology, UZ Brussel, Laarbeeklaan 101, 1090 Brussels, Belgium. E-mail: [email protected] 30
Doppler imaging and HR reduction ● C. WEYTJENS et al.
of our study was to assess the effect of a pharmacological decrease in heart rate using an If channel blocker in a healthy small animal model. To ensure that LV contractility was not affected, invasive measurements of contractility were performed in a subgroup of animals. METHODS Study design A total of 24 adult male Wistar rats (10 weeks old) were studied before and during IV administration of ivabradine. Twelve animals (391 ⫾ 24 g) underwent echocardiography with DMI analyses (group A) 10 minutes after administration of pentobarbital 50 mg/kg IP. The 12 remaining animals (397 ⫾ 29 g) were assigned to invasive measurements (group B) under anesthesia with urethane 600 mg/kg and alpha-chloralose 160 mg/kg IP. Echocardiography with DMI analysis was performed in group B after stabilization of hemodynamic parameters. The study was approved by the local Institutional Care and Animal Research Committee. All animals were treated in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Ivabradine Ivabradine was dissolved in physiological saline to obtain a solution of 1 mg/mL. Ivabradine was administered IV by continuous drip of 0.1 mL/min via a tail vein in group A (noninvasive protocol) and via a jugular vein in group B (invasive protocol) until a heart rate reduction of at least 15% was obtained. Ivabradine infusion was then stopped and echocardiographic images where acquired within 30 min. Echocardiographic evaluation After administration of the anesthetic drug, chest hair was shaved and animals were positioned in left lateral decubitus for the echocardiographic examination. Transthoracic echocardiography was performed using a Vivid 7 (GE VingMed, Horten, Norway) with a linear 13-MHz probe (i13L) in all animals before and during administration of ivabradine. The ECG was monitored continuously by fixing the electrodes to the limbs. Two-dimensional guided M-mode tracings were recorded from parasternal short-axis views at the midpapillary level. M-mode measurements were performed offline using EchoPAC PC (GE Vingmed, version 3.1.6). The thickness of the interventricular septum, posterior wall and left ventricular (LV) dimensions were measured according to the edge-to-edge method. Measurements of three consecutive cycles during expiration were averaged. To calculate LV volumes and EF, the ellipsoid formula Pi*D3/6a (D ⫽ diameter of the LV; a ⫽ ellip-
31
ticity factor) was used. We used an ellipticity factor of 1/3 as D ⫽ L/3 (L ⫽ length of the ellipsoid). Color Doppler myocardial images were recorded in the parasternal short-axis view at the midventricular level. High frame rate (⬎300 fps) was obtained by reducing view width. The Nyquist limit was set as low as possible (while avoiding aliasing) to include the low myocardial velocities (between 9 and 15 cm/s). Velocity, strain and strain rate measurements were performed offline using dedicated software (SPEQLE, Catholic University Leuven, Belgium). The sample volume was placed in the middle of the anteroseptal and inferolateral wall. A strain estimation length of 0.8 mm was used. Timings of the beginning and ending of the ejection phase were obtained using the electrocardiogram and the velocity trace as previously demonstrated (Urheim et al. 2000). At least six consecutive cycles were used for the offline analysis. The reproducibility and repeatability of the deformation parameters has recently been assessed and published by our group (Weytjens et al. 2006). Invasive measurements Twelve animals (group B) underwent invasive assessment of hemodynamics before and during administration of ivabradine. The animals were placed on a controlled heating pad to maintain body temperature. A tracheotomy was performed to insert a 25-g cannula, and ventilation was started using a pressure-controlled respirator (end-inspiratory pressure, 12 mm Hg; end-expiratory pressure, 2 mm Hg; Hugo Sachs Elektronik). The animals were placed under a microscope (Zeiss, Götingen, Germany) and the jugular vein was cannulated for infusion of hypertonic saline or ivabradine. Via the carotid artery, a 2-F, micromanometer conductance catheter (Millar Instruments, Houston, TX, USA) was then inserted in the LV cavity guided by online pressure and volume signals. After hemodynamic stabilization heart rate, end-diastolic and end-systolic pressures were assessed. The abdomen was opened via a small incision just below the diaphragm to enable temporary preload reduction by directly compressing the inferior vena cava using a cotton-tipped stick. To quantify contractility we used the end-systolic pressure-volume relation (ESPVR) and the preload recruitable stroke work (PRSW). Endsystolic elastance (Ees) was derived from the linear regression of several ESPVR loops obtained during pressure manipulation (Katz AM 2006). The parallel conductance attributed to the tissues surrounding the LV cavity was estimated by injection of a bolus of 30% saline into the jugular vein. Volume calibrations of the measured conductance were performed at the end of each experiment with known blood volume. Analysis was performed with PVAN 2.9 (Millar Instruments).
32
Ultrasound in Medicine and Biology
Volume 35, Number 1, 2009
Table 1. Influence of heart rate on M-mode parameters Group A (n ⫽ 12)
Heart rate (bpm) LVEDD (mm) LVESD (mm) SV (mL) CO (mL/min) EF (%)
Rest
Ivabradine
p-value
350 ⫾ 24 7.2 ⫾ 0.7 4.4 ⫾ 0.5 0.31 ⫾ 0.08 106 ⫾ 28 76 ⫾ 4
273 ⫾ 31 7.9 ⫾ 0.6 4.8 ⫾ 0.5 0.40 ⫾ 0.09 108 ⫾ 24 78 ⫾ 4
⬍0.001 0.003 0.05 0.002 0.79 0.46
LVED(S)D ⫽ Left ventricular end diastolic (systolic) parameter; SV ⫽ stroke volume; CO ⫽ cardiac output; EF ⫽ ejection fraction.
Statistical analysis All data are expressed as the mean ⫾ SD. Comparison between groups was performed using an unpaired Student’s t-test. Comparison between repeated measures was performed using a paired Student’s t-test test. Statistical significance was defined as p ⬍ 0.05. For pvalues ⬍ 0.000 a cut-off value of ⬍ 0.001 was chosen. RESULTS In group A, all DMI datasets were used for offline analysis. In group B, one rat died during the protocol, and in two animals DMI trace quality was too low for adequate quantification. Heart rate reduction There was a significant reduction in heart rate during ivabradine administration in both groups (Tables 1 and 2). Heart rate was reduced by 18% in group A and by 36% in group B (p ⬍ 0.02). The ejection time only increased significantly in group B (Table 2). Figure 1 represents the heart rate and ejection time of each rat before and during ivabradine administration. Influence of heart rate on conventional echo measurements Left ventricular M-mode analysis was only performed in the noninvasive protocol (group A) because
Fig. 1. Heart rate and ejection time before and during administration of ivabradine in group A (grey lines; open circles) and group B (black lines; solid circles).
the conductance catheter interfered with the posterior endocardial delineation. There was a significant increase in LV end-diastolic diameter (EDD), but no change in LVEF and CO (Table 1). Influence of heart rate on DMI parameters The mean values for systolic velocity, strain and strain rate in the anteroseptal and inferolateral walls before and during administration of ivabradine are given in Table 2. Absolute values for systolic velocity, strain and strain rate were significantly lower in ventilated animals compared with nonventilated animals. During induced bradycardia, these parameters re-
Table 2. Influence of heart rate on DMI parameters Group A (n ⫽ 12)
Heart rate (bpm) Ejection time (ms) Anteroseptal Velocity (cm/s) Strain (%) Strain rate (1/s) Inferolateral Velocity (cm/s) Strain (%) Strain rate (1/s)
Group B (n ⫽ 9)
Rest
Ivabradine
p-value
Rest
Ivabradine
p-value
317 ⫾ 23 78 ⫾ 10
262 ⫾ 25 82 ⫾ 13
⬍0.001 0.18
408 ⫾ 55 59 ⫾ 8
263 ⫾ 62 83 ⫾ 24
⬍0.001 ⬍0.02
⫺0.9 ⫾ 0.3 39 ⫾ 10 11.0 ⫾ 3.0
⫺1.1 ⫾ 0.4 42 ⫾ 11 12.0 ⫾ 2.9
0.26 0.50 0.47
⫺0.9 ⫾ 0.3 21 ⫾ 9 9.9 ⫾ 3.9
⫺1.0 ⫾ 0.5 28 ⫾ 10 8.8 ⫾ 3.7
0.83 0.09 0.49
3.6 ⫾ 0.4 53 ⫾ 14 16.5 ⫾ 1.2
3.6 ⫾ 0.5 46 ⫾ 8 14.8 ⫾ 2.6
0.46 0.10 0.2
2.8 ⫾ 0.8 25 ⫾ 10 13.8 ⫾ 4.7
3.0 ⫾ 1.1 35 ⫾ 12 14.2 ⫾ 5.8
0.30 0.12 0.61
Doppler imaging and HR reduction ● C. WEYTJENS et al.
33
Fig. 3. Correlation between change in heart rate and ejection time and change in strain rate during ivabradine administration (group A ⫹ group B).
ative dP/dt and ED and ES pressures changed significantly during the administration of ivabradine (Table 3). DISCUSSION
Fig. 2. Velocity, strain and strain rate of the inferolateral wall before and during administration of ivabradine in group A (grey lines, open circles) and group B (black lines, solid circles).
mained unchanged in both groups. Figure 2 shows the individual responses of velocity, strain and strain rate during ivabradine infusion for the inferolateral wall. The wide range in absolute values is because groups A and B are represented in the same figure. No significant correlation between velocity, strain or strain rate and heart rate or ejection time was found. Figure 3 shows the relation between the change in strain rate and the change in heart rate and ejection time (r2 ⫽ 0.05 and 0.14, respectively). Influence of heart rate on invasive hemodynamics There was no significant change in PRSW and Ees during heart rate reduction. However, positive and neg-
This study demonstrates that, although contractility of the LV myocardium remains constant, a decrease in heart rate, induced by ivabradine, has no significant effect on velocity, strain or strain rate parameters in this small animal model. We can therefore conclude that neither velocity nor strain rate are influenced by moderate changes in heart rate.
Table 3. Influence of heart rate on invasive measurements Group B (n ⫽ 11)
Heart rate (bpm) PRSW EES ⫹dP/dt (mm Hg/s) ⫺dP/dt (mm Hg/s) LVEDP (mm Hg) LVESP (mm Hg)
Rest
Ivabradine
p-value
383 ⫾ 35 68.7 ⫾ 18.1 1.6 ⫾ 0.7 6534 ⫾ 2647 ⫺7864 ⫾ 2482 5.2 ⫾ 3.3 109.9 ⫾ 17.1
271 ⫾ 33 63.7 ⫾ 15.4 1.6 ⫾ 1.1 5233 ⫾ 2825 ⫺6315 ⫾ 2758 6.7 ⫾ 4.0 95.4 ⫾ 19
⬍0.001 0.12 0.67 0.001 0.001 0.006 0.001
PRSW ⫽ Preload recruitable stroke work; EES ⫽ end systolic elastance.
34
Ultrasound in Medicine and Biology
When performing small animal studies, heart rate often significantly decreases during anesthesia (Rottman et al. 2003; Stein et al. 2007). Ideally, echocardiography would be performed in conscious rats. However, when performing echocardiographic studies in conscious rodents, human manipulation in lab environment can markedly increase heart rate (Rottman et al. 2003; Stein et al. 2007). Furthermore, the use of an anesthetic drug is necessary to keep the animal immobilized during the echocardiographic examination. Our group recently analyzed LV function in an experimental rat model for type I diabetes (Weytjens et al. 2008). In this study, during anesthesia, a significant difference in heart rate was noted between control and diabetic rats (315 ⫾ 29 bpm vs. 244 ⫾ 21 bpm in diabetes), making the interpretation of the data difficult. Several authors have encountered the same problem in experimental cardiomyopathy, but until now little attention has been given to this issue (Mihm et al. 2001; Yoon et al. 2005). To date, only a few reports have studied the effect of heart rate on DMI parameters (Weidemann et al. 2002; Boettler et al. 2005). Both studies evaluated the effect of an increase in heart rate on specific strain rate measurements. Weideman et al. (2002) demonstrated that, in a closed-chest pig model, pacing the hearts up to 180 bpm did not influence strain rate significantly. Strain on the other hand significantly decreased during atrial pacing. In a stepwise multiple regression analysis, strain rate correlated best with contractility (measured by ⫹dP/dt) and strain with stroke volume (measured by M-mode derived volumes). The authors concluded that strain can be related to increased heart rate and altered loading conditions, whereas strain rate was relatively independent of heart rate, reflecting better LV contractility. Because atrial pacing can induce significant changes in LV filling and LV contractility, we believe that a pacing model is not the ideal method to assess the effect of heart rate changes on DMI parameters. Furthermore, dP/dt is a rate-dependent parameter and should increase with increasing heart rate (Appleton 1991). A better way to assess LV contractility would be to look at the pressure–volume relation and to measure PRSW (Lee et al. 2003; Bombardini 2005). Boettler et al. (2005) measured velocity and strain rate in a large number of infants with a wide range of resting heart rates and could not find any significant correlation between velocity or strain rate and heart rate. This suggests that, even with a physiological increase in heart rate, tissue Doppler parameters are not affected. This study unfortunately was unable to correlate their results with invasive measurements, and no information about preload, afterload or contractility was given.
Volume 35, Number 1, 2009
To assess the effect of heart rate reduction on parameters for LV function, beta-blocking agents cannot be used because they are well known to influence LV contractility. Ivabradine is a potent If channel blocker that is able to induce a reduction in heart rate without modifying myocardial contractility or mean arterial pressure (Colin et al. 2003; Borer 2004). This drug was therefore chosen for evaluating the effect of heart rate reduction on deformation parameters in the present study. Our experimental study confirms the absence of dependence on heart rate of velocity and strain rate when relative bradycardia of 18 –36% is induced. To ensure that the decrease in heart rate did not influence LV function, we measured preload recruitable stroke work (PRSW), the gold standard technique for evaluation of LV contractility (Katz AM 2006). In contrast with previous authors, we demonstrated a significant decrease in dP/dt in response to ivabradine. This is not surprising because the change in pressure is expressed over the ejection time period, which was significantly prolonged during ivabradine administration. In contrast to the study by Weideman et al. (2002), we did not observe a significant change in strain. Weideman et al (2002) measured a significant decrease in stroke volume during pacemaker-induced tachycardia at 160 and 180 bpm. The lower strains, therefore, reflected the decrease in total deformation caused mainly by the decrease in ejection time. During bradycardia, one would expect an increase in systolic strain. In our study, although we could measure a slight increase in strain, the difference was not statistically significant. This can be explained by the fact that the change in heart rate that was obtained during our study protocol was less significant (18 –36 % decrease vs. 30 – 40 % increase). Limitations Anesthetic drugs can have an influence on heart rate and myocardial function (Stein et al. 2007). The results of our study, therefore, need to be interpreted with regard to the anaesthetic drug. Moreover, a different type of anaesthesia was used for analysis of DMI parameters in both groups. This was done because in the invasive protocol the animals required analgesic drugs in addition to narcotic drugs. This explains why heart rate was more reduced in group B than in group A and why lower absolute values for velocity, strain and strain rate were found in group B. Because we were especially interested in the relative changes in the measurements in response to ivabradine, the type of anaesthesia used is of minor importance. We were only able to measure deformation in short axis images and thus only analyzed radial function. The use of DMI in the analysis of longitudinal function is
Doppler imaging and HR reduction ● C. WEYTJENS et al.
much more difficult because artefacts and alignment with the beam are often a problem. We cannot exclude the possibility that longitudinal function is not influenced by heart rate reduction in rats and can contribute to significant changes in LV function. However, this is unlikely because invasive measurements remained unchanged. No inotropic stimulation was performed, which would have given the opportunity to show that, with restoration of heart rate but at higher contractility, deformation parameters change significantly. This has, however, already been demonstrated (Weidemann et al. 2002; Sebag et al. 2005). CONCLUSION This study shows that a moderate heart rate reduction does not significantly influence DMI measurements in this specific rat model. Therefore, in the interpretation of DMI data in small animal studies, heart rate reduction does not need to be taken into account, given limited heart rate and loading changes. Acknowledgments—Steven Droogmans is an Aspirant of the Research Foundation–Flanders (Belgium) (FWO). Tony Lahoutte is a Senior Clinical Investigator of the Research Foundation–Flanders (Belgium) (FWO). The research at ICMI is funded by the Inter-University Attraction Poles Programme–Belgian State–Belgian Science Policy.
REFERENCES Appleton CP. Influence of incremental changes in heart rate on mitral flow velocity: Assessment in lightly sedated, conscious dogs. J Am Coll Cardiol 1991;17:227–236. Boettler P, Hartmann M, Watzl K, Maroula E, Schulte-Moenting J, Knirsch W, Dittrich S, Kececioglu D. Heart rate effects on strain and strain rate in healthy children. J Am Soc Echocardiogr 2005; 18:1121–1130. Bombardini T. Myocardial contractility in the echo lab: Molecular, cellular and pathophysiological basis. Cardiovasc Ultrasound 2005; 3:000 – 000. Borer JS. Drug insight: If inhibitors as specific heart-rate-reducing agents. Nat Clin Pract Cardiovasc Med 2004;00:103–109. Colin P, Ghaleh B, Monnet X, Su J, Hittinger L, Giudicelli JF, Berdeaux A. Contributions of heart rate and contractility to myocardial oxygen balance during exercise. Am J Physiol Heart Circ Physiol 2003;284:H676 –H682. Droogmans S, Lauwers R, Cosyns B, Roosens B, Franken PR, Weytjens C, Bossuyt A, Lahoutte T, Schoors D, Van Camp G.
35
Impact of anesthesia on valvular function in normal rats during echocardiography. Ultrasound Med Biol 2008;00:(ahead of print). Hirano T, Asanuma T, Azakami R, Okuda K, Ishikura F, Beppu S. Noninvasive quantification of regional ventricular function in rats: Assessment of serial change and spatial distribution using ultrasound strain analysis. J Am Soc Echocardiogr 2005;18:907–912. Katz AM. Physiology of the heart. C.H.I.P.S., 2006. Lee W, Huang W, Yu W, Chiou K, Ding PY, Chen C. Estimation of preload recruitable stroke work relationship by a single-beat technique in humans. Am J Physiol Heart Circ Physiol 2003;284: H744 –H750. Mihm MJ, Seifert JL, Coyle CM, Bauer JA. Diabetes related cardiomyopathy time dependent echocardiographic evaluation in an experimental rat model. Life Sci 2001;69:527–542. Rottman JN, Ni G, Khoo M, Wang Z, Zhang W, Anderson ME, M. EC. Temporal changes in ventricular function assessed echocardiographically in conscious and anesthetized mice. J Am Soc Echocardiogr 2003;16:1150 –1157. Sebag I, Handschumacher M, Ichinose F, Morgan J, Hataishi R, Rodrigues A, Guerrero J, Steudel W, Raher M, Halpern E, Derumeaux G, Bloch K, Picard M, Scherrer-Crosbie M. Quantitative assessment of regional myocardial function in mice by tissue Doppler imaging: Comparison with hemodynamics and sonomicrometry. Circulation 2005;111:2611–2616. Stein AB, Tiwari S, Thomas P, Hunt G, Levent C, Stoddard MF, Tang XL, Bolli R, Dawn B. Effects of anesthesia on echocardiographic assessment of left ventricular structure and function in rats. J Basic Res Cardiol 2007;102:28 – 41. Urheim S, Edvardsen T, Torp H, Angelsen B, Smiseth OA. Myocardial strain by Doppler echocardiography validation of a new method to quantify regional myocardial function. Circulation 2000;102:1158 – 1164. Weidemann F, Jamal F, Kowalski M, Kukulski T, D’hooge J, Bijnens B, Hatle L, De Scheerder I, Sutherland GR. Can strain rate and strain quantify changes in regional systolic function during dobutamine infusion, B-blockade, and atrial pacing. Implications for quantitative stress echocardiography. J Am Soc Echocardiogr 2002; 15:416 – 424 Weytjens C, Cosyns B, D’hooge J, Gallez C, Droogmans S, Lahoute T, Franken PR, Van Camp G. Doppler myocardial imaging in adult male rats: Reference values and reproducibility of velocity and deformation parameters. Eur J Echocardiogr 2006;7:411– 417. Weytjens C, Franken PR, D’hooge J, Droogmans S, Cosyns B, Lahoutte T, Van Camp G. Doppler myocardial imaging in the diagnosis of early systolic left ventricular dysfunction in diabetic rats. Eur J Echocardiogr 2008;9:326 –333. Yoon Y-S, Uchida S, Masuo O, Cejna M, Park J-S, Gwon H-C, Kirchmair R, Bahlman F, Walter D, Curry C, Hanley A, Isner JM, Losordo DW. Progressive attenuation of myocardial vascular endothelial growth factor expression is a seminal event in diabetic cardiomyopathy: Restoration of microvascular homeostasis and recovery of cardiac function in diabetic cardiomyopathy after replenishment of local vascular endothelial growth factor. Circulation 2005;111:2073–2085.