Functional and Bioenergetic Consequences of Postinfarction Left Ventricular Remodeling in a New Porcine Model
MRI and 31P-MRS Study
Background The underlying mechanisms by which left ventricular remodeling (LVR) leads to congestive heart failure (CHF) are unclear. This study examined the functional and bioenergetic abnormalities associated with postinfarction ventricular remodeling in a new, large animal model.
Methods and Results Remodeling was induced by circumflex coronary artery ligation in young pigs. LV mass, volume, ejection fraction (EF), the ratio of scar surface area to LV surface area, and LV wall stresses were calculated from magnetic resonance imaging anatomic data and simultaneously measured LV pressure. Hemodynamics, transmural blood flow, and high-energy phosphates (spatially localized 31P–nuclear magnetic resonance) were measured under basal conditions, during hyperperfusion induced by pharmacological vasodilation with adenosine, and during pyruvate infusion (11 mg/kg per minute IV). Six of 18 animals with coronary ligation developed clinical CHF while the remaining 12 animals had LV dilation (LVR) without CHF. The results were compared with 16 normal animals. EF decreased from 55.9±5.6% in normals to 34.6±2.3% in the LVR group (P<.05) and 24.2±2.8% in the CHF group (P<.05 versus LVR). The infarct scar was larger in CHF hearts than in LVR hearts (P<.05). In normals, LV myocardial creatine phosphate (CP)/ATP ratios were 2.10±0.10, 2.06±0.16, and 1.92±0.12 in subepicardium (EPI), mid myocardium (MID), and subendocardium (ENDO), respectively. In LVR hearts, the corresponding ratios were decreased to 1.99±0.13, 1.80±0.14, and 1.57±0.15 (ENDO P<.05 versus normal). In CHF hearts, CP/ATP ratios were 1.41±0.14, 1.33±0.15, and 1.25±0.15; (P<.05 versus LVR in EPI and MID). The calculated myocardial free ADP levels were significantly increased only in CHF hearts.
Conclusions Bioenergetic abnormalities in remodeled myocardium are related to the severity of LV dysfunction, which, in turn, is dependent on the severity of the initiating myocardial infarction.
Acute myocardial infarction results in remodeling of the noninfarcted region of the LV, which can ultimately lead to the development of CHF.1 2 3 4 5 Although HEP abnormalities have been observed in remodeled myocardium,1 4 it is unclear whether they are caused by perfusion abnormalities or altered intermediary metabolism. Furthermore, it is uncertain whether these bioenergetic abnormalities contribute to the development of CHF. The primary goals of this study were (1) to develop a reproducible large animal model of postinfarction ventricular remodeling that would make possible detailed in vivo physiological and biochemical studies and (2) to examine the relationships between the severity of LV dysfunction and abnormalities in morphological, physiological, and bioenergetic characteristics of this model. To achieve these goals, proximal LCx occlusion was used to generate a myocardial infarction in young pigs. The animals then were followed over several months while remodeling of the left ventricle developed. LV wall thickness, volume, EF, and wall stress were measured by MRI.1 4 6 Myocardial ATP, CP, and inorganic phosphate (Pi) levels were measured with spatially localized 31P-NMR spectroscopy, and regional myocardial blood flow was measured with radioactive microspheres.1 4 7
On the basis of observations in previous models of remodeling1 4 and pressure-overload7 and volume-overload8 hypertrophy, we hypothesized that (1) the severity of remodeling would be related to the size of the initiating myocardial infarction, (2) myocardial HEP levels and CP/ATP ratios would be decreased in remodeled myocardium and that the severity of these reductions would be correlated with the severity of LV dysfunction, (3) the bioenergetic abnormalities might result from myocardial underperfusion, and (4) defects in the metabolism of carbohydrates or lipids also might contribute to these abnormalities. To examine these hypotheses, measurements were obtained under basal conditions, during coronary hyperperfusion induced with adenosine, and during the infusion of pyruvate, a substrate that bypasses glycolysis and activates pyruvate dehydrogenase. The results support the concept that in remodeled myocardium, metabolic abnormalities are related to the severity of LV dysfunction, which, in turn, is dependent on the severity of the initiating myocardial infarction, and that the bioenergetic abnormalities can be explained partially by abnormalities of carbohydrate metabolism. However, the data do not support the concept that persistant underperfusion of the myocardium is the cause of the aforementioned metabolic abnormalities.
Studies were performed in 18 swine with LCx occlusion and 16 normal animals. All experimental procedures performed were approved by the University of Minnesota Animal Resources Committee. The investigation conformed with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1985).
Infarct Production by Coronary Ligation
Young Yorkshire swine (age, 45 days; weight, ≈10 kg) were anesthetized with sodium pentobarbital (30 mg/kg IV), intubated, and ventilated with a respirator with supplemental oxygen. Arterial blood gases were maintained within the physiological range by adjustments of the respiratory settings and oxygen flow. A left thoracotomy was performed, and 0.5 cm of the proximal LCx was dissected free and completely occluded with a ligature. After the ligation, the animals were observed in the open chest state for 30 minutes. When ventricular fibrillation occurred, electrical defibrillation was performed immediately. This procedure usually was successful. The chest then was closed; if the heart was dilated, the pericardium was left open. The animals were given standard postoperative care including analgesia until they were active and eating normally.
All MRI studies were performed on the standard Siemens Medical System VISION operating at 1.5 T. The animals were anesthetized with sodium pentobarbital. A catheter was placed into the femoral artery and advanced into the LV chamber for LV pressure recording. Animals then were placed on their left side in a Helmholtz coil with a diameter of 18 cm, which was used to improve signal to noise. To compute LV wall stress, the image acquisition was triggered by the LV pressure through the fluid-filled LV catheter. All of the imaging sequences were synchronized to the LV pressure trace. The electronic LV pressure signal was recorded and fed to a comparator set to a threshold level of 10% of the upslope of the LV pressure curve at the beginning of systole. The signal from the comparator was sent to a pulse former and then fed to the ECG port of the magnetic resonance system, where it was treated like the standard electrographic input to run the pulse sequences. Scout images were taken in the axial plane with a single-shot, ultrafast gradient echo sequence.9 10 11 12 From the axial image, both horizontal and vertical long-axis images were obtained. By alternating back and forth several times, a true vertical long axis of the left ventricle was obtained. From the long-axis scout image, short-axis segmented cine turboflash9 10 11 slices were prescribed to cover the myocardium from apex to base. The double-oblique, short-axis turboflash images cover the heart from apex to base with a slice thickness of 10 mm, with no interslice gap.
MRI Cine Technique
The parameters of the segmented cine sequence were TR/TE/flip angle=33 ms/6.1 ms/25 degrees with an FOV=17.5 cm and a matrix of 87×128 (pixel size, 2 mm×1.4 mm) and slice thickness of 7 to 10 mm.9 11 The sequence used segmented k-space acquisition such that three phase-encoded lines were gathered per cardiac phase per heartbeat. Total image acquisition required approximately 52 heartbeats for each slice location. The temporal image resolution (data acquisition window) of this sequence was 33 ms per cardiac image. Each myocardial level took <1.5 minutes to acquire, since two acquisitions were used and the average heart rate of the animals was 120 bpm. The average number of short-axis slices needed to image the entire myocardium from apex to base was 6 to 8. This 10-minute protocol provided high signal-to-noise cine sequences covering the entire heart.
To obtain high-resolution anatomic heart images, multislice, single-phase spin-echo images trigged in the systolic phase were acquired to cover the entire heart. These images permitted the precise delineation of the extent of the scar region of the heart.9 11 Images were taken with a slice thickness of 5 mm and an FOV of 17.5 cm, resulting in a true spatial resolution of 2 mm×1.4 mm pixel size. The TR for this sequence equals the RR interval (500 ms) and the echo time TE was set to 30 ms. Total measurement time for an average of 10 to 14 slices was 5 minutes.
Image Analysis of the MRI Cine Studies
The imaging data were archived to optical disk and copied to a SUN SPARC 10 workstation for evaluation with the use of an automatic segmentation program (ImageView, Siemens Cooperate Research, Inc). As previously described,12 the program is based on robust deformable models of endocardial and epicardial border segmentation of ventricular boundaries in cardiac magnetic resonance images. This segmentation technique has been combined with a user interface that allows one to load, sort, visualize, and analyze a cardiac study in <20 minutes.11 12 The segmentation algorithm is based on the steepest descent as well as dynamic programming strategies integrated via multiscale analysis for minimizing the energy function of the resulting contour. The ventricular boundaries are used to construct a three-dimensional model for visualization and to compute hemodynamic parameters. Automatic segmentation of endocardial and epicardial boundaries was performed for calculation of ventricular volumes, EF, LV diastolic and systolic volumes, and absolute myocardial mass from multislice, multiphase magnetic resonance cine images. Starting with a user-specified approximate boundary or an interior point of the ventricle for one starting image in one slice, the algorithm generated automatic contours corresponding to the epicardium and the endocardium and automatically propagated them to other slices in the cardiac phase (spatial propagation) and to other phases for a given slice location (temporal propagation) of the cardiac study. The observer then could make some manual corrections to the six or seven pairs of contours in the first column of the temporal-spatial matrix. Manual modifications generally were made on the apex and base levels. Mean LV wall thickness for each short-axis ring was averaged from three measurements of the remote zone (anterior wall and septum wall). Thickness of the scar was averaged from three measurements of the scar area. LVSA measurement in each slice was computed by subtracting the total area enclosed by the endocardium from that enclosed by the subepicardium; the resultant area was multiplied by the slice thickness to obtain the volume of each slice; the total LV mass volume was calculated by adding up the volumes of all the short-axis slices. The total LVSA was obtained by dividing the total LV wall mass volume by the mean of LV wall thickness of each slice. Similarly, the LVSSA was obtained by dividing the total scar volume, which was the sum of the scar volume of each short axis, by the mean of the scar thickness of each short axis. LV mass was computed by the total LV wall mass volume multiplied by 1.05 (specific gravity of myocardium) to calculate the LV mass. The LV end-diastolic volume (Vd) and end-systolic volume (Vs) of each slice were represented by the area enclosed by the endocardium. The total LV volume was computed by adding the volumes of all slices. LVEF was calculated by 100%×(V−Vs)/Vd. Interobserver and intraobserver errors for the calculations of LV mass and LV volumes have been shown previously to be <3 mg and 3 mL, respectively.4 Meridianal wall stress was computed from the LV pressure and simultaneously obtained short-axis view of LV MRI (LV cavity diameter and average thickness the remote LV wall) as described by Grossman et al.6
Acute Experimental Preparation for MRS Study
Eighteen animals with MRI-documented LVR and 16 normal animals were anesthetized with sodium α-chloralose (100 mg/kg and 20 mg/kg per hour IV), intubated, and ventilated with a respirator with supplemental oxygen. Arterial blood gases were maintained within the physiological range by adjustments of the respiratory settings and oxygen flow. A heparin-filled polyvinyl chloride catheter (outer diameter, 3.0 mm) was introduced into the right femoral artery and advanced into the ascending aorta. A sternotomy was performed and the heart suspended in a pericardial cradle. A second heparin-filled catheter was introduced into the left ventricle through the apical dimple and secured with a purse-string suture. A similar catheter was placed into the left atrium through the atrial appendage. A bipolar epicardial pacing electrode was sutured to the right atrial appendage. A 25-mm-diameter NMR surface coil was sutured onto the LV anterior wall; care was taken to avoid the infarct region. The pericardial cradle then was released, and the heart was allowed to assume its normal position. The surface coil leads were connected to a balanced-tuned circuit external and perpendicular to the thoracotomy incision. The animals then were placed in a Lucite cradle and positioned within the magnet.
Spatially Localized 31P-NMR Spectroscopic Technique
Measurements were performed in a 40-cm-bore, 4.7-T magnet interfaced with a SISCO (Spectroscopy Imaging Systems Corporation) console. The LV pressure signal was used to gate NMR data acquisition to the cardiac cycle, while respiratory gating was achieved by triggering the ventilator to the cardiac cycle between data acquisitions.13 14 31P- and 1H-NMR frequencies were 81 and 200.1 MHz, respectively. Spectra were recorded in late diastole with a pulse repetition time of 6 to 7 seconds. This repetition time allowed full relaxation for ATP and Pi resonances and ≈90% relaxation for the CP resonance.1 7 13 14 CP resonance intensities were corrected for this minor saturation; the correction factor was determined for each heart from two spectra recorded consecutively without transmural differentiation, one with 15-second repetition time to allow full relaxation and the other with the 6- to 7-second repetition time used in all the other measurements.
Radiofrequency transmission and signal detection were performed with a 25-mm-diameter surface coil. The coil was cemented to a sheet of silicone rubber 0.7 mm in thickness and ≈50% larger in diameter than the coil itself. A capillary containing 15 μL of 3 mol/L phosphonoacetic acid was placed at the coil center to serve as a reference. The proton signal from water detected with the surface coil was used to homogenize the magnetic field and to adjust the position of the animal in the magnet so that the coil was at or near the magnet and gradient isocenters. This was accomplished with a spin-echo experiment and a readout gradient. The information gathered in this step also was used to determine the spatial coordinates for spectroscopic localization.13 14 Chemical shifts were measured relative to CP, which was assigned a chemical shift of −2.55 ppm relative to 85% phosphoric acid at 0 ppm.
Spatial localization across the LV wall was performed with the RAPP-ISIS/FSW method.15 Detailed data regarding voxel profiles, voxel volume, and extensive documentation of the accuracy of the spatial localization obtained in phantom studies and in vivo have been published elsewhere.13 14 15 Briefly, signal origin was restricted with the use of B0 gradients and adiabatic inversion pulses to a column coaxial with the surface coil perpendicular to the LV wall. The column dimensions were 17×17 mm. Within this column, the signal was further localized using the B1 gradient to five voxels centered about 45°, 60°, 90°, 120°, and 135° spin rotation increments.13 14 15 FSW localization used a nine-term Fourier series expansion. The Fourier coefficients, number of free induction decays acquired for each term in the Fourier expansion, and the multiplication factors used to construct the voxels have been reported previously.13 14 15 The position of the voxels relative to the coil was set using the B1 magnitude at the coil center, which was experimentally determined in each case by measurement of the 90° pulse length for the phosphonoacetic acid reference located in the coil center. Each set of spatially localized transmural spectra were acquired in 10 minutes. A total of 96 scans was accumulated within each 10-minute block.
Resonance intensities were quantified with the use of integration routines provided by the SISCO software. ATPγ resonance was used for ATP determination. Since data were acquired with the transmitter frequency being positioned between the ATPγ and CP resonance, the off resonance effects on these peaks were negligible. The numeric values for CP and ATP in each voxel were expressed as ratios of CP/ATP. Pi levels were measured as changes from baseline values (ΔPi) with the use of integrals obtained in the region covering the Pi resonance.
Measurement of Intracellular Mg2+
Cytosolic Mg2+ was estimated from the chemical shift differences of the ATPα and ATPβ (δαβ) resonances and cytosolic pH measurements.16 Whole wall spectra were used for this determination.
Calculation of Myocardial Free ADP Level
The myocardial free ADP level was calculated from the CK equilibrium expression17 18 |<|[|>|ADP|<|]|>||<|=|>|(|<|[|>|ATP|<|]|>| |<|[|>|CR_|<|free|>||<|]|>|)/(|<|[|>|CP|<|]|>||<|\times|>|K_|<|obs|>| |<|[|>|Mg^|<|2|<|+|>||>|, pH|<|]|>|)where Kobs (Mg2+, pH) was calculated according to the equations of Lawson and Veech.18 CP and ATP values were obtained from the NMR spectra and calibrated with the biopsy-measured ATP values. Free creatine was calculated by subtracting the CP values from the biopsy-obtained measurement of total creatine.
Myocardial Blood Flow
Myocardial blood flow was measured with radioactive microspheres 15 μm in diameter, labeled with 141Ce, 51Cr, 95Nb, 85Sr, or 46Sc (NEN Corp) as described previously.7 19 Blood flow was expressed as mL/min per gram of myocardium.
After completion of the study, a drill myocardial biopsy and LV wall specimen were taken and frozen in liquid nitrogen for subsequent analysis of ATP and total creatine contents with the use of a high-performance liquid chromatography technique.20 The heart then was fixed in 10% buffered formalin. The atria, right ventricle, aorta, and large epicardial vessels were dissected from the left ventricle. The left ventricle then was sectioned into four transverse rings of approximately equal thickness parallel to the mitral valve ring so that a myocardial ring ≈2.0 cm in thickness contained the region of myocardium located directly beneath the surface coil. The region of myocardium directly beneath the surface coil was removed and sectioned into three transmural layers from epicardium to endocardium, weighed on an analytical balance, and placed into vials for counting. Similar myocardial specimens were obtained from the lateral and posterior LV walls to ensure that the measurements from region of myocardium corresponding to the surface coil were typical for the entire left ventricle.
After the circumflex ligation, the LVR process was evaluated by MRI once every month for the following 2 months. If CHF developed as evidenced by coexistence of two of the following signs: cyanosis, ascites, decreased activity, or LVEDP >20 mm Hg, the final MRI and 31P-NMR spectroscopy study was performed immediately.
In animals not manifesting failure at the last MRI study, a 31P-MRS study was performed 3 to 5 days later. Aortic and LV pressures were measured with the aforementioned catheters and Spectramed pressure transducers positioned at mid chest level and recorded on an eight-channel direct writing recorder (Coulbourne Instrument Co). LV pressure was recorded at normal and high gain for measurement of end-diastolic pressure. Hemodynamic measurements and 31P-MRS spectra first were obtained under basal conditions. Midway through the 10-minute MRS acquisition period, a microsphere injection was performed for determination of myocardial blood flow.
After completion of baseline measurements, an infusion of adenosine was started at a rate of 125 μg/kg per minute IV and increased by 125 μg/kg per minute every 3 minutes until one of the following two end points was reached: (1) LV peak systolic pressure decreased by 10 to 25 mm Hg or (2) the adenosine dose reached 500 μg/kg per minute. This dose was continued for the following 10 to 12 minutes while all measurements were repeated. The adenosine infusion then was stopped and hemodynamic measurements were allowed to return over the ensuing 10 minutes. Pyruvate then was infused in a dose of 11 mg/kg per minute IV, and all measurements were repeated. This dose of pyruvate was chosen because it has been reported previously to result in blood levels that cause a significant decrease in myocardial ADP and to significantly increase myocardial phosphorylation potential in normal dogs.21
Hemodynamic data were measured directly from the chart recordings. The integral numeric values for CP, ATP, and Pi during each experimental condition were expressed as CP/ATP and ΔPi/CP ratios. All the integrals were performed by an observer unaware of MRI results and were obtained as described above. 31P-NMR spectra from the first, third, and fifth voxels were taken to represent subepicardium, midmyocardium, and subendocardium, respectively.
The unpaired t test was used for the comparison of intergroup data, and one-way ANOVA with replications was used for the data during different experimental conditions within the same group. A value of P<.05 was required for significance. When significance was found, individual comparisons were made with use of the method of Scheffe´. All values are expressed as mean±SE.
The anatomic data and MRI measurements are summarized in Table 1⇓. Circumflex ligation was performed on 22 pigs. One pig died immediately, and 3 animals died within the first 10 days after infarction. In the first 30 days after infarction, 6 animals developed clinical CHF characterized by cyanosis (observed in 2 of the 6), weight loss, ascites, decreased activity, or LVEDP >20 mm Hg. When two of these characteristics appeared, the final MRI and 31P-MRS studies were performed immediately. These 6 pigs formed the LVR+CHF group (Table 1⇓). In the remaining 12 infarct animals, which formed the compensated LVR group (Table 1⇓), the mean period from infarction to 31P-NMR study was 52.5±3.2 days. In 5 sham-operated littermates, the 31P-NMR studies were performed 48.8±5.9 days after operation. There was no significant difference in examined variables between sham-operated animals and 5 similarly sized normal pigs. Therefore, the data from the unoperated and sham-operated animals were combined to form control group data for LVR animals (Normal A; Table 1⇓). Because the LVR+CHF group animals weighed less than those in the LVR group, another group (n=6) of similarly sized normal animals was also studied, and these data served as the control group for the LVR+CHF group (Normal B; Table 1⇓).
LVW/BW (g/kg) was significantly higher in both groups with LVR compared with their respective controls, and this ratio was the highest in animals with CHF (Table 1⇑). A similar trend was seen in RVW/BW, with the highest values observed in animals with CHF (Table 1⇑). There was no evidence of RV infarction in any of the animals studied. The scar surface area to total surface area ratio (SSA/LVSA) was significantly higher in animals with CHF compared with animals with LV remodeling only (Table 1⇑). LV EF decreased significantly in animals with LVR and was lowest in animals with LVR+CHF (Table 1⇑). LV diastolic volume increased significantly only in animals with LVR+CHF. LV end-systolic volume increased significantly in both groups of animals with remodeling (Table 1⇑) and was significantly higher in LVR+CHF hearts than in LVR hearts (Table 1⇑). Both LV diastolic wall stresses and mean LV systolic wall stresses were significantly increased in hearts with LVR and were further increased in LVR+CHF hearts.
Fig 1⇓ illustrates LV apical views of hearts after formalin fixation. A heart (LVR+CHF group) with severe LV dilation (right) is compared with a heart from a sham-operated littermate (left). The animal with LCx ligation did not survive the terminal study on the 10th day after LCx ligation surgery. This animal had ≈400 mL of ascites by postmortem examination. The littermate was studied at 63 days of age. The pig with LCx ligation had severe LV dilation, as illustrated in the figure. Fig 2⇓ compares the MRI cine studies of a normal heart and a heart from the LVR group. Hearts with LVR had a significant decrease in LV EF. Fig 3⇓ compares multislice spin-echo MRI images of a normal heart and a heart 24 days after LCx ligation. Thinning of the infarcted zone is clearly shown. Arrows point to the scarred area.
Hemodynamic measurements obtained during the experimental periods are shown in Table 2⇓. Heart rates were not significantly different between groups throughout the study protocol. Under the basal conditions, mean aortic pressure and LV systolic pressure tended to be lower in animals with ventricular remodeling, but this difference was not significant. Adenosine infusion (234.4±40.6, 230.3±31.8, 250.0±64.5, 229.1±59.7 mg/kg per minute IV for groups Normal A, LVR, Normal B, and LVR+CHF, respectively) tended to reduce aortic and LV systolic pressures. This was significant only in group Normal A (Table 2⇓). During pyruvate infusion, aortic and LV pressure remained at basal levels. During basal conditions, LVEDP was significantly elevated in the LVR+CHF group. None of the groups showed significant changes in LVEDP in response to experimental interventions (Table 2⇓). The product of heart rate times LV systolic pressure was not different between the groups of animals through the study protocol (Table 2⇓).
Myocardial Blood Flow
Myocardial blood flow data are summarized in Table 3⇓. Myocardial blood flow was not significantly different among the different groups of animals under the basal conditions. Adenosine increased myocardial blood flow in each layer of the LV wall in each group (Table 3⇓). However, the subendocardial blood flow and Endo/Epi blood flow ratio were significantly lower in animals in the LVR+CHF group compared with its control group during adenosine infusion (Table 3⇓). Pyruvate infusion did not change myocardial blood flow significantly (Table 3⇓).
31P-NMR Spectroscopic Measurements
Fig 4⇓ illustrates transmural sets of spectra acquired from a normal heart under basal conditions and during pyruvate infusion. Corresponding spectra from an LVR heart are shown in Fig 5⇓. Spectra obtained during adenosine infusion are not shown because no significant changes were observed in either group during this intervention. Fig 6⇓ illustrates a set of spectra from an LVR+CHF heart under basal conditions. The LVEF was 28% two days before this spectra set was obtained. This animal had cyanosis and 200 mL of ascites by postmortem examination. Spectra obtained during adenosine or pyruvate infusion are not shown because no significant changes were observed in these hearts. In these transmurally differentiated spectra sets, the voxels were positioned such that the voxel labeled EPI was over the outer edge of the LV wall while the voxel most distant from the coil, labeled as ENDO, was positioned over the subendocardium with little penetration into the LV cavity. The voxel labeled MID encompasses the midwall. We have shown previously that these three voxels have virtually no overlap.7 13 14 15 Spectra in all layers of LV wall are characterized by high CP/ATP ratios (Figs 4 through 6⇓⇓⇓). It is evident that CP/ATP ratios are markedly reduced in all myocardial layers of the LVR+CHF heart compared with normal heart (Figs 4 and 6⇓⇓ and Table 4⇓). In hearts with compensated LVR, the CP/ATP ratios were reduced in the inner layers of the LV wall (Figs 4A and 5A⇓⇓ and Table 4⇓). In response to pyruvate infusion, CP/ATP ratios increased in the subendocardial layers of LVR heart (Fig 5⇓ and Table 4⇓); this did not occur in the normal hearts or CHF hearts (Fig 4⇓ and Table 4⇓). Pi was not visible at the signal-to-noise ratio of the spectra obtained from the normal hearts or hearts with compensated LVR (Figs 4 and 5⇓⇓). Pi was increased in LVR+CHF hearts under basal conditions (Fig 6⇓) in 2 of the 6 hearts. Adenosine or pyruvate infusion had no effect on Pi levels of these hearts.
The measured CP/ATP ratios from the spectra for each group during the experimental periods are shown in Table 4⇑. As noted, the Pi resonance was too low to be detected in most of the hearts studied and was unchanged in response to experimental interventions; ΔPi/CP ratio data are therefore not presented. Adenosine did not affect myocardial HEP levels in any layer of any group (Table 4⇑). Pyruvate infusion increased the subendocardial CP/ATP ratio significantly in the LVR group only (Table 4⇑).
Biopsy ATP data are summarized in Table 5⇓. ATP levels obtained from subepicardial biopsies were significantly lower in both groups of remodeled ventricles compared with normals (Table 5⇓). The ATP level was not significantly lower in the LVR+CHF group than in the LVR group. Myocardial total creatine levels were significantly less than control in both groups of hearts with LVR, but the difference between the LVR and LVR+CHF groups was not significant (Table 5⇓).
Myocardial pH and Free Mg2+ and ADP Levels
Calculated whole wall cytosolic pH, free Mg2+, and ADP levels are shown in Table 6⇓. In most studies, the Pi level was too low to be detected under basal conditions (see Figs 4 and 5⇑⇑). The pH data for the two LVR+CHF hearts with detectable Pi are included in Table 6⇓. In the adequately perfused remodeled myocardium, in the presence of normal arterial blood pH, it is likely that intracellular pH was normal at a value of ≈7.1.17 This was the pH measured in the two remodeled hearts with visible Pi (Table 6⇓), and this value was used in ADP calculations. Myocardial free Mg2+ levels in each group under each experimental condition were ≈0.5 mmol/L, and this value is within the range of reported intracellular pH values. With the use of the equations of Lawson and Veech18 with these intracellular pH and Mg2+ values, the calculated Kobs of CK was 120. This value was used for myocardial free ADP calculations (Table 6⇓). The calculated ADP tended to be increased in both remodeled groups, but this was statistically significant only in the LVR+CHF group.
Relationships Between the Size of the Scar and Alterations in LV Function, Remodeling, and Energetics
It is important to characterize the relationships between the severity of initial myocardial damage and subsequent alterations in LV function, energetics, and the severity of LVR. In the current study, the SSA/LVSA was used as an indirect measurement of the severity of the initial myocardial damage. Fig 7⇓ presents EF and LVW/BW plotted against SSA/LVSA as well as the CP/ATP ratio plotted against EF% and SSA/LVSA. These plots include all 18 animals with circumflex ligation. All plots show a fair amount of scatter. There is a borderline-significant correlation between severity of LV dysfunction (EF%) and the index of scar size (P=.06, Fig 7A⇓) and a significant correlation between the severity of LVR and the severity of myocardial damage (Fig 7B⇓), between CP/ATP ratio and EF (Fig 7C⇓), and SSA/LVSA (Fig 7D⇓).
The present study was designed to develop a new porcine model of postinfarction LVR and to define its characteristics of bioenergetic, morphological, and functional abnormalities. The main findings are that (1) the severity of LV dysfunction is related to infarct size and that clinical CHF developed in each animal with an SSA/LVSA ratio >0.24, (2) in the remodeled zone, both LV systolic and diastolic wall stresses are significantly increased and these values are greatest in animals with clinical CHF, (3) ATP, CP, and creatine levels and CP/ATP ratios are significantly decreased in remodeled myocardium, with the most severe changes being present in animals with CHF, (4) increasing myocardial blood flow with adenosine did not affect HEP levels in any group, although in remodeled hearts from CHF animals, subendocardial blood flow reserve is impaired, (5) in remodeled hearts without CHF, the reduced CP/ATP ratio is alleviated by pyruvate infusion, and (6) in remodeled hearts from CHF animals, the remarkable reduction of CP/ATP across the LV wall is not alleviated by pyruvate infusion. The subsequent discussion will focus first on the morphological and functional abnormalities in the remodeled hearts and then will consider the bioenergetic abnormalities and their implications.
General Characteristics of the Model
In the present study the LCx was occluded to avoid the high mortality rate that we observed after acute left anterior descending coronary artery occlusion in pigs. Unlike the dog, which has an abundant native intercoronary collateral circulation so that infarcts are commonly nontransmural, pigs have sparse native collateral vessels and very low collateral flow rates after acute coronary occlusion. As a result, acute coronary occlusion in these pigs resulted in a dense transmural infarct, which evolved to a thin-walled scar that was associated with prominent remodeling of the uninvolved myocardium.
After recovery from acute coronary ligation, the incidence of clinical CHF was between 30% and 40% (if animals with sudden death are included) within the first 4 weeks. Once an animal developed signs of CHF, survival was truncated and physiological studies were performed promptly. The infarct animals with CHF were ≈30 days younger when studied than those without CHF so that the body weights of the pigs in the CHF group were lower than those without CHF. Hence, separate groups of weight-matched controls were required for the LVR and LVR+CHF subgroups of infarct animals (Table 1⇑). LV mass was increased 12% in the LVR group and 52% in the LVR+CHF animals, while the respective LVW/BW ratios were increased 20% and 62% compared with their respective control groups (Table 1⇑). The degree of hypertrophy of the remodeled myocardium may be underestimated by simple LVWs because of the presence of thin scar, which comprised a significant fraction of the left ventricle (Figs 1 through 3⇑⇑⇑). The LVW/BW ratios may overestimate the degree of hypertrophy in the infarct groups if they had growth retardation or weight loss compared with the controls. Conversely, the LVW/BW ratios could represent an underestimate in the CHF group because of water retention. A more accurate way to assess the degree of hypertrophy would be to measure myocyte lengths and volumes in littermates with and without remodeling.
We believe that the in vivo MRI measurement of SSA/LVSA provides a reasonable indication of infarct size relative to heart size, although underestimation caused by remote zone dilation and overestimation resulting from scar area aneurysm formation probably decreased the validity of this assessment (Figs 2 and 3⇑⇑). With these methodological limitations in mind, the results suggest that early development of CHF is related to infarct size (Table 1⇑). This conclusion is in agreement with previous studies.5 22 In the current study, clinical signs of CHF developed in each individual in which SSA/LVSA was >24% (Table 1⇑). It is possible that animals with smaller infarcts and “compensated” LVR would develop CHF if followed for a longer time, since in the clinical situation the development of CHF in patients with LVR may not occur for years.5 In a canine LVR model in which myocardial damage was produced by DC shock, animals followed for 13 months did not develop clinical signs of CHF although evidence of significant LVR was present.1 Last, coronary occlusion was produced during a period of rapid body growth. It is possible that the rapid body growth represents an additional stress for the heart that contributed to the remodeling process.
As shown in Table 1⇑ and Figs 1 through 3⇑⇑⇑, wall stress is significantly elevated in the remodeled hearts, with the highest values being present in the LVR+CHF group. The calculation of the LV wall stress was based on a formula derived from studies of the normal heart,6 and application of this calculation is complicated by the fact that thin scar has a higher average wall stress, whereas the remote area is thicker and would have decreased average wall stress (Figs 2 and 3⇑⇑). Despite these limitations, the calculated end-diastolic and mean systolic wall stresses probably are useful approximations of the remodeled hearts. The relative contributions of the marked increase in afterload imposed by the altered chamber geometry and the primary decrease in myocyte contractile capacity to the markedly reduced EFs observed in the remodeled hearts remain to be determined.
LV Contractile Function
LV systolic function was evaluated by means of MRI measurements of EF and LV systolic shortening fraction. As shown in Table 1⇑ and Fig 2⇑, LV systolic function (represented by EF%) was markedly reduced in both remodeled groups; longitudinal systolic shortening fraction (data not shown) also was reduced. As pointed out above, it is not clear to what extent the marked decrease in the “pump” function of the remodeled hearts reflects alterations in loading conditions and to what extent primary (ie, non–load-dependent) contractile dysfunction of the myocyte is contributory. This issue is an important area for future experimentation. The uptake of calcium by sarcoplasmic reticulum (SR) is dependent on the Ca2+-activated ATPase of SR. The activity of this enzyme is decreased in myocardium of failing hearts.23 24 25 This alteration is thought to cause myocyte diastolic Ca2+ overload and ultimately diastolic heart failure.23 24 The results of the study of myocardium from patients transplanted for CHF indicate that the performance of mechanical work is essential for inducing abnormalities of excitation contraction coupling and intracellular calcium homeostasis.26 It is quite possible that primary abnormalities of myocyte function will be found in the current model.
Another issue of interest is whether myocardial ischemia plays a role in the evolution of the remodeling process or the transition from the compensated to the decompensated state. In a pacing CHF model, Wilson et al27 found that the cardiac dilation and LV dysfunction observed under basal conditions in these hearts were not caused by myocardial ischemia, as indicated by the presence of normal lactate extraction. Lactate extraction, however, does not provide information regarding the transmural differences of metabolic consequence. It is possible that the mix of ischemic subendocardial venous blood and nonischemic outer layer venous blood might have masked abnormal subendocardial lactate metabolism. However, as pointed out above, the absence of any HEP change in response to adenosine argues against underperfusion under basal conditions. The reduced subendocardial flow reserve in the decompensated (but not compensated) remodeled hearts indicates that at some point blood flow inadequacies may play a role in the later stages of the remodeling process. In remodeled LV, during pacing-induced tachycardia, a “vicious circle” of subendocardial underperfusion, contractile dysfunction, and loss of myocardial HEPs was observed that may be initiated by subendocardial ischemia.1
Last, the transmural distribution of energy expenditure in remodeled myocardium is of interest. There is evidence that in normal myocardium, the rate of oxygen consumption may be greatest in the subendocardium.28 We have presented evidence suggesting that at very high work states in normal hearts, this gradient may be altered and that subepicardial ATP expenditure is equal to or greater than that of the subendocardium29 and that the same phenomenon may be present in hypertrophied myocardium. Hence, it will be of interest to address this issue in both compensated and decompensated remodeled hearts. It may be that recruitment of subepicardial contractile function is a compensatory mechanism in remodeled myocardium, and this is another issue of future experimental interest.
Myocardial ATP Levels
The biopsy data in the present study indicate that myocardial ATP is significantly decreased in remodeled myocardium (Table 6⇑). The ATP values tended to be lowest in LVR+CHF hearts, but differences between the two groups of remodeled hearts were not significant. The reasons for the depressed level of ATP in remodeled hearts are not known. Repetitive episodes of ischemia can cause persistent depression of ATP levels, and this could apply to the subendocardium of the LVR+CHF hearts in which coronary reserve was impaired (Table 3⇑). However, no such abnormality was present in other layers of the LVR+CHF group or in any layer of the LVR group despite reduced ATP levels, so this explanation seems unlikely. An alternative possibility is that the higher ADP levels in the remodeled hearts enhanced ATP catabolism via the myokinase30 and subsequent degradative reactions. In the myocyte, the mechanism that controls the set point of steady-state ATP levels is unclear, but it must reflect a balance between adenine nucleotide synthesis and degradation, and if degradative reaction rates are even slightly enhanced in the absence of a compensatory increase in nucleotide synthetic rates, the net result could be a lower set point for steady-state ATP levels.
Myocardial CP, ADP, and Creatine Levels and the CP/ATP Ratio
In normoxic myocardium, steady-state CP levels are related to ADP levels because CK is a near equilibrium enzyme31 and the concentrations of other substrates of the reaction exceed their respective Km values.32 Hence, in the current study, the reductions in CP and, derivatively, the CP/ATP ratio, reflect elevation of ADP levels (Table⇑s 4 and 6). The mechanism of the fall in creatine levels is not clear. As is the case with ATP, the mechanisms that determine “normal” levels of free creatine in the myocyte are unknown. Creatine is not synthesized in the myocyte; rather, it is transported into the cell and a fraction of intracellular creatine is trapped in the form of CP, a hydrophilic compound that does not traverse the normal cell membrane. If the myocyte attempts to set free creatine at some “optimum” value, then a fall of CP will obligatorily result in the reduction of total cellular creatine until the desired level of free creatine is reached.
Whether the reductions of CP, ATP, or the CP/ATP ratio reflect ATP synthetic abnormalities that constrain contractile function is unknown. In a mitral regurgitation model, we found that a significantly decreased CP/ATP ratio across the LV wall was associated with normal LV function under baseline conditions and during inotropic stimulation.8 These observations suggest that reduction of the CP/ATP ratio under conditions when myocardial blood flow is not restricted does not necessarily reflect limitation of either ATP synthetic capacity or myocardial function. Hence, reduction of the CP/ATP ratio under some circumstances may result from alterations in intermediary metabolism and/or oxidative phosphorylation that do not obligatorily limit ATP synthetic capacity sufficiently to restrain contractile performance.
The basis of the ADP abnormality observed in remodeled myocardium is of great interest, even if this abnormality does not prove to reflect a functionally significant limitation of oxidative phosphorylation. A number of variables may affect ADP levels in the absence of ischemia. Isolated hearts perfused with glucose are known to have higher ADP levels than those perfused with pyruvate, yet contractile function of the glucose-perfused hearts is comparable to that of pyruvate-perfused hearts over a wide range of MVo2 values.17 The higher ADP levels in the glucose-perfused hearts may relate to lower levels of intramitochondrial NADH.17 Because the level of CP is approximately inversely related to that of ADP (via the CK equilibrium relationship) under steady-state conditions, any increase in ADP levels such as that induced by predominant metabolism of glucose would be expected decrease CP levels. There is evidence that both pressure-overload33 34 and volume-overload35 36 hypertrophy are associated with decreased long-chain fatty acid utilization and increased glucose metabolism. A comparable change in substrate utilization patterns occurring in remodeled myocardium could result in the higher ADP and lower CP levels observed in the present study.
Alternatively, a change in some rate-limiting step in the reactions of oxidative phosphorylation could result in an apparent increase in the Km of ADP with respect to ATP synthase; this alteration would be reflected in higher ADP levels at any level of O2 consumption and would not have to be associated with a change of the Vmax for O2. Again, this issue can only be resolved with more experimentation. With regard to the latter possibility, there is considerable dispute regarding whether or not mitochondrial oxidative phosphorylation is normal in CHF hearts. In failing hamster hearts, the oxidative phosphorylation is abnormal.37 In contrast, data obtained from failing human hearts38 and experimentally produced CHF of cat hearts39 suggest that mitochondrial function is normal.
Last, it is possible that increased ADP levels may reflect decreased CK activity in mitochondria. Mitochondrial CK is an obligatory link in the “creatine phosphate shuttle,” which has been considered by some40 to be an important mechanism for the facilitation of HEP delivery to sites of consumption in the myocyte and the ultimate return of ADP to ATP synthase for rephosphorylation.41 A derivative prediction of this hypothesis is that a rate-limiting decrease in mitochondrial CK activity would increase the Km value for ADP by making ADP return to the ATP synthase more dependent on transmitochondrial transport. Myocardial CK activity has been shown to be significantly reduced in abnormal myocardium including the postinfarction remodeling rat model,42 43 and isolated, permeabolized myocytes with reduced or inhibited mitochondrial CK activity have been shown to have higher Km (with regard to MVo2) values for ADP.44 Hence, it is possible that higher ADP values in remodeled porcine myocardium reflect physiologically significant reductions of mitochondrial CK activity.
Relationships Between the Size of the Scar and Alterations in LV Function, Remodeling, and Energetics
Fig 7⇑ characterizes the relationships between the size of the scar (as an indication of the severity of initial myocardial damage) and subsequent alterations in LV function, energetics, and the severity of LVR. These linear relationships suggest a “vicious circle”1 in which LV dysfunction triggered by myocardial damage results in LVR (chamber dilation), which adds an extra workload on the LV and consequently causes further worsening of LV function and myocardial energetics and remodeling.
A significant decrease of the CP/ATP ratio was observed in the subendocardial layer of the LVR group that was corrected by pyruvate infusion. This response to pyruvate infusion in LVR myocardium suggests that either glycolytic generation of pyruvate was limiting and/or that under physiological conditions, pyruvate dehydrogenase (PDH) activity was suboptimal. Because high concentrations of pyruvate are known to activate PDH,17 the response to pyruvate infusion cannot differentiate between these possibilities. The positive pyruvate response also suggests that there may be abnormalities of fatty acid metabolism in remodeled myocardium. In normal porcine myocardium, fatty acid is the dominant substrate, and unless limitations of pyruvate generation decrease anapleurotic generation of TCA cycle intermediates to the point where TCA cycle turnover is affected, fatty acid oxidation would be expected to maintain adequate intramitochondrial NADH levels and, in consequence, the CP/ATP ratio, despite a reduction in glucose metabolism. Long-chain fatty acid utilization has been shown previously to be abnormal in both pressure-overload32 and volume-overload35 hypertrophy. The present preliminary results indicate that detailed analyses of glucose and fatty acid metabolism and their relationships to myocardial function and remodeling are warranted in remodeled myocardium. Interestingly, this positive pyruvate response was not present in the LVR+CHF group despite the presence of substantially lower CP/ATP ratios in all myocardial layers (Table 4⇑). The reason for this lack of response to pyruvate in the LVR+CHF hearts is unknown. It does suggest that other factors (see above discussion) determine the steady-state ADP values in end-stage CHF hearts.
The results of the current study suggest that (1) metabolic abnormalities in remodeled myocardium are related to the severity of LV dysfunction which, in turn, is dependent on the severity of the initiating myocardial infarction, (2) the bioenergetic abnormalities may be explained partially by abnormalities of carbohydrate metabolism in compensated remodeled left ventricle, and (3) in hearts with remodeling and CHF, the mechanism of more severe alterations in bioenergetics remains to be elucidated.
Selected Abbreviations and Acronyms
|CHF||=||congestive heart failure|
|LCx||=||left circumflex coronary artery|
|LVEDP||=||LV end-diastolic pressure|
|LVEDV||=||LV end-diastolic volume|
|LVESV||=||LV end-systolic volume|
|MRI||=||magnetic resonance imaging|
|MRS||=||magnetic resonance spectroscopy|
|NMR||=||nuclear magnetic resonance|
|SSA||=||scar surface area|
This work was supported by US Public Health Service Grants HL-21872, HL-33600, HL-32427, and HL-50470 (J.Z.), a Grant-in-Aid from the American Heart Association National Center (J.Z.), a Grant-in-Aid from the American Heart Association Minnesota Affiliate, and the Department of Veterans Affairs Medical Research Funds. M.H.J. Eijgelshoven was a recipient of a research Fellowship of the Netherlands Foundation for Scientific Research (NWO).
- Received September 28, 1995.
- Revision received March 6, 1996.
- Accepted March 13, 1996.
- Copyright © 1996 by American Heart Association
Zhang J, McDonald K. Bioenergetic consequence of left ventricular remodeling secondary to discrete myocardial infarction. Circulation. 1995;92:1011-1019.
Sharpe N, Smith H, Murphy J, Hannon S. Treatment of patients with symptomless left ventricular dysfunction after myocardial infarction. Eur Heart J. 1990;11(suppl B):147-150.
Anversa P, Beghi C, Kikkawa Y, Olivetti G. Myocardial infarction in rats: infarct size, myocyte hypertrophy, and capillary growth. Circ Res. 1986;58:26-37.
Pfeffer MA, Braunwald E. Ventricular remodeling after myocardial infarction: experimental observations and clinical implications. Circulation. 1990;81:1161-1172.
Grossman W, Jones D, McLaurin LP. Wall stress and patterns of hypertrophy in human left ventricle. J Clin Invest. 1975;56:56-64.
Zhang J, Merkle H, From AHL, Ugurbil K, Bache RJ. Bioenergetic abnormalities associated with severe left ventricular hypertrophy. J Clin Invest. 1993;92:993-1003.
Toher C, Zhang J, Zhang Y, Ugurbil K, Bache RJ, Homans D. Bioenergetic consequences of volume-overloaded hypertrophy. Circulation. 1994;90(suppl I):I-160. Abstract.
Robitaille PM, Lew B, Merkle H, Path G, Sublett E, Hendrich K, Lindstrom P, From AHL, Garwood M, Bache RJ, Ugurbil K. Transmural high energy phosphate distribution and response to alterations in workload in the normal canine myocardium as studied with spatially localized 31P NMR spectroscopy. Magn Reson Med. 1990;16:91-116.
Hendrich K, Merkle H, Weisdorf S, Vine W, Garwood M, Ugurbil K. Phase modulated rotating frame spectroscopic localization using an adiabatic plane rotation pulse and a single surface coil. J Magn Reson. 1991:92:258-275.
Lawson JWR, Veech RL. Effects of pH and free Mg2+ on the Keq of the creatine kinase reactioned other phosphate hydrolyses and phosphate transfer reactions. J Biol Chem. 1979;254:6528-6537.
Bache RJ, Zhang J, Path G, Merkle H, Hendrich K, From AHL, Urgubil K. High-energy phosphate responses to tachycardia and inotropic stimulation in left ventricular hypertrophy. Am J Physiol. 1994;266(Heart Circ Physiol 35):H1959-H1970.
Laughlin MR, Taylor J, Chesnick AS, DeGroot M, Balaban RS. Pyruvate and lactate metabolism in the in vivo dog heart. Am J Physiol. 1993;264:H2068-H2079.
Kostuk WJ, Kazamias TM, Gander MP, Simon AL, Ross J Jr. Left ventricular size after acute myocardial infarction: serial changes and their prognostic significance. Circulation. 1973;47:1174-1179.
Mercadier JJ, Lompree AM, Duc P, Boheler KR, Fraysse JB, Wisnewsky C, Allen PD, Komajda M, Schwartz K. Altered sarcoplasmic reticulum Ca2+-ATPase gene expression in the human ventricle during end-stage heart failure. J Clin Invest. 1990;85:305-309.
de la Bastie D, Levitsky D, Rappaport L, Mercadier JJ, Marotte F, Wisnewsky C, Brovkovisch V, Schwartz K, Lompre AM. Function of the sarcoplasmic reticulum and expression of its Ca2+-ATPase gene in pressure overload-induced cardiac hypertrophy in the rat. Circ Res. 1990;66:554-564.
Gwathmey JK, Copelas L, MacKinnon R, Schoen FJ, Feldman MD, Grossman W, Morgan JP. Abnormal intracellular calcium handing in myocardium from patients with end-stage heart failure. Circ Res. 1987;61:70-76.
Vahl CF, Bonz A, Hagl TS. Intracellular calcium transient of working human myocardium of seven patients transplanted for congestive heart failure. Circ Res. 1994;74:952-958.
Wilson JR, Douglas P, Hickey WF, Lanoce V, Ferraro N, Muhammad A, Reichek N. Experimental congestive heart failure produced by rapid ventricular pacing in the dog: cardiac effects. Circulation. 1987;75:857-867.
Weiss HR, Lipp JA, Sinha AK. Regional oxygen saturation of small arteries and veins in the canine myocardium. Circ Res. 1978;42:119-126.
Zhang J, Duncker DJ, Xu Y, Zhang Y, Path G, Merkle H, Hendrich K, From AHL, Bache RJ, Ugurbil K. Bioenergetic responses of normal myocardium at very high workstates: an in vivo transmural 31P NMR study. Am J Physiol. 1995;268:H1891-H1905.
Zhang J, Duncker DJ, Ya X, Zhang Y, Pavek T, Wei H, Merkle H, Ugurbil K, From AHL, Bache RJ. Effect of left ventricular hypertrophy secondary to chronic pressure overload on transmural myocardial glucose uptake: a 31P NMR spectroscopic study. Circulation. 1995;92:1274-1283.
Bishop SP, Altschuld RA. Increased glycolytic metabolism in cardiac hypertrophy and congestive failure. Am J Physiol. 1970;218:153-159.
Wittels B, Spann JF Jr. Defective lipid metabolism in the failing heart. J Clin Invest. 1968;47:1787-1794.
Sievers R, Parmley WW, James T, Coffelt-Wilman J. Energy levels at systole versus diastole in normal hamster hearts versus myopathic hamster hearts. Circ Res. 1983;53:759.
Chidsey CA, Weinbach EC, Pool PE, Morrow AG. Biochemical studies of energy production in the failing human heart. J Clin Invest. 1966;45:40-50.
Sobel BE, Spann JF Jr, Pool PE, Sonnenblick EH, Braunwald E. Normal oxidative phosphorylation in mitochondria from the failing heart. Circ Res. 1967;21:355-363.
Meyer RA, Sweeney HL, Kushmerick MJ. A simple analysis of the ‘phosphocreatine shuttle.’ Am J Physiol. 1984;246:C365-C377.
Neubauer S, Horn M, Schindler R, Mader H, Lubke D, Kaiser WA, Neubauer TS, Horn M, Schindler R, Mader H, Lubke D, Kaiser WA, Krahe T, Hahn D, Ertl G. Impaired cardiac high-energy phosphate metabolism in patients with aortic valve disease: clinical and hemodynamic correlates. Circulation. 1993;88(suppl I):I-531. Abstract.
Neubauer S, Horn M, Naumann A, Tian R, Hu K, Laser M, Friedrich J, Gaudron P, Schnackerz K, Ingwall JS, Ertl G. Impairment of energy metabolism in intact residual myocardium of rat hearts with chronic myocardial infarction. J Clin Invest. 1995;95:1092-1100.