31P Nuclear Magnetic Resonance Spectroscopic Imaging of Regions of Remodeled Myocardium in the Infarcted Rat Heart
Background The clinical course of a patient with a myocardial infarction (MI) depends largely on the ability of the noninfarcted region to remodel and compensate for the loss of the infarcted region. Previous studies have shown that the remaining viable myocardium remodels morphologically, functionally, and biochemically. The purpose of this study was to define the regional distribution of the biochemical remodeling that occurs after MI in rat hearts by use of a technique that could be applied noninvasively to human subjects.
Methods and Results Infarcts of the left ventricular apex and anterolateral wall were induced by occluding a coronary artery. Eight to 10 weeks after infarction, one-dimensional chemical shift imaging (CSI) was used to obtain 31P nuclear magnetic resonance (NMR) spectra of eight 2.5-mm-thick cross-sectional slices along the long axis (from base to apex) of isolated buffer-perfused rat hearts. Regional ATP and phosphocreatine (PCr) contents were compared in remodeled versus normal (sham) myocardium. Spin-echo 1H MR images identified the mass of each slice, allowing calculations of metabolite amount per unit myocardium in each slice. 1H MR images identify the hypertrophy of remodeled myocardium but do not discriminate between scar and viable tissue. In contrast, 31P CSI does distinguish viable tissue. Compared with shams, there was less 31P signal in the slices distal to the occlusion containing mainly scar tissue and increased signal intensity in slices proximal to the occlusion because of myocyte hypertrophy. The ATP signal intensity changed in direct proportion to the viable tissue mass in the slice, suggesting that the amount of ATP per unit mass in viable remodeled myocardium is the same as that of the shams. In contrast, the amount of PCr per unit mass in remodeled myocardium decreased. This decrease is uniform across the slices, correlates with infarct size, and parallels a similar decrease in tissue creatine content.
Conclusions 31P CSI of post-MI hearts shows that (1) PCr decreases uniformly (ie, independent of the distance from the scar) in the noninfarcted remodeled myocardium, and its amount inversely correlates with infarct size; and (2) the ATP signal provides a profile of viable myocardium and is a biochemical marker of morphological remodeling and hypertrophy that has occurred in noninfarcted regions. Thus, 31P CSI provides both a marker that tissue injury has occurred (decreased PCr) and a marker of the extent of remodeling in response to injury (ATP distribution) in a single set of noninvasive measurements.
Myocardial infarction with loss of function in the infarcted region imposes an increased mechanical load on the noninfarcted region. The subsequent hemodynamic course and life span of a subject with a myocardial infarction depends in large part on the ability of the noninfarcted region to remodel, adapt, and compensate for the loss of the infarcted region.
Studies using rat hearts with permanent coronary artery ligation have been useful for understanding this remodeling process. Chronic hemodynamic alterations result in functional and morphological changes, including LV dilatation leading to decreased cardiac performance1 2 3 4 and myocyte hypertrophy with inadequate adaptation of the capillary vasculature.5 In addition, biochemical assay of homogenates of the intact rat heart,6 7 and more recently of the isolated chamber walls,8 9 have shown decreased PCr and creatine levels and slightly decreased or unchanged ATP levels in each of the chambers studied. These changes in animal models of chronic coronary artery disease faithfully reproduce earlier results obtained from human biopsy samples. Samples obtained from patients undergoing heart surgery10 11 show decreased creatine and PCr and essentially unchanged ATP. These results show that the remaining viable myocardium in chronically infarcted hearts has remodeled not only morphologically and functionally but also biochemically.
The purpose of this study was to define the regional distribution of the biochemical remodeling that occurs in chronically infarcted rat hearts by a noninvasive method that can potentially be applied to human subjects with MI. Assessment of the time course, location, and extent of biochemical remodeling in the post-MI noninfarcted region is critical for achieving a better understanding of the progression of this region to subsequent hemodynamic failure. Moreover, identifying and quantifying regional biochemical changes may provide a sensitive noninvasive measure of the state of progression to failure and lead to more aggressive therapeutic strategies to slow such progression. It could also serve as a tool to assess the efficacy of interventional strategies such as revascularization.
31P NMR spectroscopy has proved to be a useful technique for measuring ATP and PCr contents in intact hearts of small animals, including the intact post-MI heart.12 31P spectroscopic imaging and spatial localization techniques to identify the heart have been used to measure overall cardiac PCr and ATP levels in patients with cardiomyopathy,13 14 15 16 aortic valve disease,17 acute MI,18 and chronic MI.19 31P spectroscopic imaging techniques provide measurement of localized metabolite contents in different regions of the heart. Regional analyses have been used to determine transmural gradients in phosphate content across the LV wall of dog hearts,20 focusing on one portion of the free wall. 31P chemical shift imaging has also been used to measure regional pH differences in skeletal muscle.21
In this study, we use 31P spectroscopic imaging techniques to define the changes in regional distribution of ATP and PCr throughout the remodeled myocardium. Using the rat heart model of chronic MI, we compared the remodeled myocardium subsequent to both small and large infarcts. We determined whether these changes are uniform in the remaining noninfarcted tissue or whether there is a gradient dependent on the distance from the infarct. To do this, we combined 31P and 1H NMR to obtain 31P spectra and 1H images from eight different regions in isolated perfused rat hearts that had been subjected to coronary artery ligation 8 to 10 weeks earlier.
Ligation of the large marginal branch of the circumflex coronary artery was performed22 23 on 250- to 360-g male Wistar rats (Charles River Breeding Laboratories, Inc, Kingston, Conn) (in the rat, the territory served by the left anterior descending artery in humans, ie, the anterolateral wall of the left ventricle, is usually served by a large marginal branch of the circumflex artery). Briefly, under artificial ventilation with air, a left thoracotomy was performed under pentobarbital sodium anesthesia. The heart was accessed through the fifth intercostal space. The pericardial sac was cut and the heart exteriorized through the space. The vessel was ligated with a silk suture approximately midway between the left atrium and the apex of the heart. The heart was repositioned in the thorax, the lungs were reinflated by a brief increase in positive end-expiratory pressure, and the wound was sutured closed immediately. Rats remained on the ventilator for ≈30 minutes until they were able to breathe on their own. The mortality rate for rats undergoing this operation was ≈50%; death usually occurred in the first 24 hours after the operation. Sham operations were performed according to an identical procedure, except that the heart was not exteriorized and the coronary artery not ligated. Surviving rats were housed 1 per cage and given free access to laboratory chow and water. Hearts from 10 infarcted and 10 sham rats were harvested 8 to 10 weeks after surgery. Three additional hearts (2 infarcted and 1 sham) were harvested 17 weeks after surgery. Since their results were indistinguishable, all data were pooled. All procedures conformed to the guiding principles of the American Physiological Society.
Isolated Perfused Hearts
Rats were anesthetized with pentobarbital sodium (20 mg/kg body wt IP). The heart was removed through a median sternotomy and rinsed in ice-cold buffer, and the cut aorta was securely tied to a polyethylene cannula. Retrograde aortic perfusion was carried out in the Langendorff mode at constant pressure of 100 mm Hg with 37°C modified Krebs-Henseleit buffer saturated with 95% O2/5% CO2. The buffer contained (in mmol/L): NaCl 118, KCl 4.7, CaCl2 1.75, MgSO4 1.2, Na4EDTA 0.5, NaHCO3 25, and glucose 11. To arrest the heart, 25.3 mmol/L NaCl was replaced by KCl to give a total KCl concentration of 30.0 mmol/L. An incision in the pulmonary artery and a short polyethylene vent pierced through the apex of the left ventricle allowed drainage of flow from the coronary sinus and thebesian veins. A water-filled latex balloon was inserted into the left ventricle through the mitral valve after incision of the left auricle. The balloon was filled with water so that the LV end-diastolic pressure was 5 to 10 mm Hg. The water-filled balloon was connected to a pressure transducer (Statham P23Db, Gould Instruments) for continuous monitoring of heart rate and LV pressure. Each heart was centered in a 20-mm NMR sample tube such that the midpoints of the ventricular tissue were at the same vertical height in the NMR-sensitive volume. The site of ligation for all the infarcted hearts was close to this midpoint. The heart was immersed in its effluent in the NMR tube and inserted into a vertical-bore 9.4-T magnet (Oxford Instruments). Coronary flow was estimated periodically by timed coronary effluent samples collected in a calibrated cylinder.
31P NMR Spectroscopic Imaging and 1H MR Imaging
Spectra were obtained with a General Electric GN400 spectrometer equipped with a Microstar imaging accessory operating at 161.94 MHz for 31P and 400.05 MHz for 1H. In beating hearts, one-dimensional CSI was used to obtain 31P NMR spectra from different regions of the heart, while conventional NMR spectroscopy was used to obtain 31P spectra from the entire heart. For 6 of the infarcted and 7 of the sham hearts, 1H MR images from different regions of the heart were also obtained after completion of the CSI experiment.
One-dimensional CSI was used to obtain 31P NMR spectra from eight 2.5-mm-thick slices along the long axis of the heart (ie, from base to apex). The total field of view along the spatial dimension was 20 mm, encompassing the length of the isolated perfused hearts within the length of the NMR-sensitive volume of the probe. CSI spectra were obtained by signal-averaging 400 scans for each spatial phase-encoding increment. Each scan used a 60° (127-ms) broad-band excitation pulse followed by a 400-μs phase-encoding time period during which a successively incremented linear magnetic field gradient in the direction of the long axis of the heart was applied. This was followed by signal acquisition in the absence of any gradients. The repetition time was 2.3 seconds, resulting in a total acquisition time of 2 hours. Individual time domain signals in the chemical shift dimension were weighted with an exponential 20-Hz line-broadening function before two-dimensional Fourier transformation. The spectral width of 6000 Hz was digitized by 1024 acquisition points for each spectrum in the chemical shift dimension. CSI spectra are displayed in absolute value mode.
Conventional 31P NMR spectra of the entire heart, acquired immediately before and after each CSI experiment, were obtained over 8 minutes by signal-averaging 200 scans of 60° broad-band excitation pulses separated by an interscan delay of 2.3 seconds. Spectra were processed similarly to the chemical shift dimension of the CSI spectra except that individual free induction decays were zero-filled from 1024 to 2048 data points24 before Fourier transformation. PCr peak areas of spectra taken before and after the 2-hour CSI experiment decreased by 20±3% in shams and 12±4% in MI hearts (P=.11); [γ-P]ATP peak areas decreased by 12±4% and 3±4% (P=.12), respectively. System gain was held constant for all 31P spectra and spectroscopic images.
1H MR images were obtained on nonbeating hearts after the 31P CSI spectra were acquired. For these experiments, the probe was retuned for 1H, and the perfusion solution was changed to Krebs-Henseleit buffer containing 30 mmol/L KCl. In addition, suction of the coronary effluent was begun from below (instead of above) the heart so that during image acquisition, the heart was no longer bathed in buffer. With the hearts in an arrested state, eight interleaved, 2.5-mm-thick, 256×128 spin-echo 1H MR images were obtained (TR=2 seconds, TE=18 ms) by signal-averaging four scans over 4 minutes. These eight images correspond to the tissue from which each 31P CSI spectrum was obtained. The relative mass of tissue in each slice was estimated by multiplication of the area of the signal of each image determined by planimetry by 2.5 mm. This calculation assumes that the width of the tissue in the 2.5-mm slice was uniform.
Infarct Size Measurement
After acquisition of the NMR data, all hearts were removed from the perfusion apparatus and blotted. The right ventricle was separated from the left ventricle and interventricular septum, which was considered to be part of the left ventricle. Grossly, the 8- to 10-week-old infarct regions were uniformly white and fibrotic, and these scars were easily distinguished from the adjacent noninfarcted reddish-brown myocardium. These fibrotic infarct regions were carefully cut along their edges and separated from the noninfarct regions. The two ventricles and infarct regions were then individually weighed and frozen. Infarct sizes in all hearts were determined as the ratio of infarct region weight divided by the combined weight of the left ventricle plus infarct region weight.
To have another measure of infarct size based on surface area rather than tissue mass, a subset of the infarcted left ventricles was opened with an incision along the long axis of the left ventricle adjacent to the septum before the infarct region was dissected away. The entire left ventricle (including infarct region) was placed between glass plates marked by a millimeter grid, and photographs of the epicardial and endocardial surfaces were taken. The photographs were used to estimate the areas of the fibrous infarct region on the endocardial and epicardial surfaces as a proportion of the total LV area and to compare these measurements with the proportion of infarcted ventricle as determined by the weight of the fibrous infarct region as described above.
For each heart, a 5- to 15-mg myocardial tissue sample obtained from the noninfarcted LV region was homogenized in 0.6N perchloric acid at 0°C. The biopsies were taken from random sites in the noninfarcted region. Aliquots were removed for measurement of protein according to the method of Lowry et al25 with bovine serum albumin as the standard. Measuring protein by this method minimizes the contribution from extracellular proteins because they contain relatively low amounts of aromatic amino acids, the target for this assay. The homogenate was neutralized and centrifuged for 5 minutes. Measurement of total creatine content was performed on other aliquots by use of the fluorometric assay of Kammermeier.26
Integrals of the resonance peaks in 31P NMR spectra corresponding to PCr and ATP contents were measured by the NMR1 curve-fitting routine (New Methods Research Inc). The [γ-P]ATP peak was used to estimate ATP content, since the relatively weak B1 field (≈1300 Hz at 161 MHz) resulted in unreliable [β-P]ATP peak integrals (its offset from the transmitter, midway between the PCr and [γ-P]ATP resonances, is almost two times that of the B1 field). The integrated signal intensities were corrected for differential T1 relaxation based on PCr and ATP relaxation time constant measurements.27 T1 relaxation time constants of 31P metabolites in the remaining viable myocardium were assumed to remain unchanged in the infarcted hearts. Signal intensities were also corrected for small differences in T2* relaxation during the phase-encoding time period (T2*=28 ms for PCr and 5 ms for [γ-P]ATP, based on non–exponentially weighted line-width measurements). The concentrations of ATP and PCr in each heart were estimated by division of integrated 31P signal intensities in each slice by the corresponding tissue mass estimated from the 1H images.
Absolute metabolite amounts and concentrations were calculated by use of a baseline ATP concentration of 31.2 nmol/mg protein obtained for control rat hearts.28 This was indistinguishable from those obtained for comparable shams in a separate study.12 We also used measured values for Lowry protein of 0.139±0.005 (sham) and 0.137±0.004 (infarcted) mg protein/g tissue wet wt (P=.76), 0.5 mL intracellular water/g tissue wet wt obtained for isolated perfused rat hearts,29 and ratios of integrated signal intensities from NMR spectra relative to ATP in shams to calculate the amounts and concentrations of the other high-energy phosphate metabolites in sham and infarcted hearts. Intracellular pH was measured by comparing the chemical shifts between the inorganic phosphate and PCr resonances with values obtained from a standard curve.
Unpaired two-tailed t tests were used to compare ATP and PCr amounts and concentrations and areas in 1H images for comparable slices from sham and infarcted hearts. The relationship of these parameters to infarct size was determined by linear regression for separate slices. This slice-to-slice comparison is valid because the midpoint of each heart was fixed at the same vertical height, and the total length of the infarcted hearts over the range of infarcts studied is <1 mm (<0.13 mm/slice) longer than the sham hearts (R. Tian, MD, PhD, personal communication, December 1994). Unpaired two-tailed t tests for all sham compared with all infarcted hearts were used to compare physiological parameters (biventricular heart weight, body weight, cardiac performance, and coronary flow).
In addition, for the ATP and PCr concentrations and PCr/ATP ratio measurements, two-factor (group and slice) ANOVA was used. This analysis showed a significant group effect between sham and infarcted hearts for all three parameters (P<.0001, P<.0001, and P<.004, respectively). Thus, the data for sham and infarcted hearts were separated, and one-factor ANOVA was used to test for statistically significant differences (P<.05) among slices in each group for the three parameters.
All calculations were performed with statview 512+ (BrainPower Inc). All data are presented as mean±SEM.
Morphological Properties of Infarcted Hearts
The surgical procedure produced variable-size infarcts of the LV apex and anterolateral free wall. The mean value of the weight of infarct region divided by the combined weight of left ventricle and infarct region was 19% and ranged from 5% to 40%. Infarct size by weight correlated well with an estimation of infarct size by surface area calculated as a percentage of total LV endocardial area obtained from photographs of the hearts taken immediately after NMR data collection (r=.92, P=.003). Data displayed in Table 1⇓ show the average body weights, biventricular heart weights with and without infarct scars, and ratios of biventricular heart weight to body weight for the sham and infarcted animals. The ratio of biventricular heart to body weight of the infarcted compared with sham animals was not different, in agreement with others studying small or moderate-sized infarcts.3 For larger infarcts with more scar, this ratio has been found to be elevated.3 12 This is because the remaining viable tissue hypertrophies until the viable tissue weight is similar to preinfarct weight.3 In agreement with these observations, the three animals with the largest infarcts (28%, 35%, and 40% of the left ventricle by weight) in this study had a significant increase in the ratio of total biventricular heart weight (including scar) to body weight: 3.2±0.2 versus 2.6±0.1 for shams (P=.004).
Fig 1⇓ shows a photograph of representative sham and infarcted Langendorff-perfused hearts illustrating the site of the surgical occlusion, location of the infarcted tissue, and the hypertrophy proximal to the occlusion. The hypertrophy proximal to the site of occlusion changes the heart shape from “ellipsoid” to more “boxlike.” Fig 2⇓ is a schematic of the rat heart ventricles shown in the photograph in Fig 1⇓ illustrating the positions of the eight 2.5-mm-thick cross-sectional slices relative to the surgical occlusion. The site of occlusion is between slices 4 and 5. These eight slices contain the tissue from which both the 1H MR images and 31P NMR spectra described below were obtained.
Cardiac Performance and Coronary Flow
Because of the length of the NMR data acquisition protocol, cardiac performance, estimated as the RPP, was set at a value near the low end of the physiological range by adjusting the LV balloon volume as discussed in the “Methods” section. RPP was measured throughout the time period during which the 31P spectroscopic images were collected so that the RPP and metabolic data were obtained concomitantly. Average RPP in the infarcted hearts was 42% lower than in shams (9.5±1.2×103 versus 16.5±1.4×103 mm Hg/min [P=.0008]). This decrease was similar to that measured at matched preload by Vleeming et al.30 The decrease in RPP was due to a decrease in developed pressure (68±7 versus 40±4 mm Hg [P=.003]) rather than a decrease in heart rate (236±8 versus 250±10 beats per minute [P=.29]). RPP correlated negatively with infarct size (r=.67, P=.0003).
Average coronary flow of viable myocardium (excluding scar weight) was not significantly different (P=.32) between infarcted hearts (13.8±0.6 mL·min−1·g−1) and shams (14.2±0.3 mL·min−1·g−1).
Fig 3⇓ shows typical 1H MR spin-echo images for both a sham and an infarcted KCl-arrested heart. Each image was obtained from a 25×25-mm cross-sectional slice 2.5 mm thick that corresponded to the tissue from which each 31P NMR spectrum was obtained. These images allow the identification of the left ventricle, interventricular septum, and right ventricle. They illustrate the increased signal intensity in the infarcted hearts, most evident in slices 2 and 3, resulting from the increase in heart mass due to hypertrophy proximal to the site of occlusion. It is important to note that in these 1H images, scar tissue (parts of slices 5 through 8 in the infarcted heart) cannot be distinguished from viable myocardium because both contain primarily water.
Fig 4⇓ compares the average slice areas from the 1H images of infarcted with those of sham hearts. This figure shows that the infarcted hearts are significantly larger in slices 1 through 6 because of hypertrophic remodeling of the noninfarcted tissue. This degree of hypertrophy correlates positively with infarct size in these slices (see figure legend for quantitative analyses).
31P NMR Spectroscopy
Fig 5⇓ shows typical conventional 31P NMR spectra obtained from the entire heart for one sham and three infarcted hearts, each with a different infarct size. The spectra show that the PCr resonance area and thus the total amount of PCr in each infarcted heart was lower than for the sham heart. Biochemical analysis of LV tissue homogenates from these hearts showed that total creatine content was also lower. The decreased amounts of PCr and creatine per heart each correlated with infarct size (r=.58, P=.003 for PCr and r=.42, P<.05 for creatine). Neither the amount of ATP nor inorganic phosphate was different in the infarcted hearts compared with shams. No monophosphate ester peak was detectable in any of the hearts. Intracellular pH was also not different.
Table 2⇓ shows average metabolite concentrations for all sham and infarcted hearts. Because both the mass of viable tissue and protein content are indistinguishable between sham and infarcted hearts, tissue concentrations show the same proportional differences as total (whole-heart) metabolite amounts.
31P NMR Regional Spectroscopic Imaging
31P NMR spectra obtained by one-dimensional CSI for each of the eight slices for a typical sham and infarcted heart are shown in Fig 6⇓. The sum of the signal intensities of these eight spectra correlates well with the signal intensity obtained with the conventional spectra of the entire heart for all hearts (r=.95, P<.0001), with an intercept not statistically different from zero (P=.06). For example, like the 31P NMR spectroscopy results for the whole heart, the total amount of PCr measured by summing all CSI slices in each of the infarcted hearts decreased by 22% on average (P=.005) compared with shams. Thus, the results of all eight CSI spectra added together are in accord with those observed by conventional NMR spectroscopy. This analysis also indicates that the signal attenuation effects of the additional phase-encoding time period before data acquisition and the switched magnetic field gradients resulted in a negligible loss in signal intensity in the CSI spectra.
Because each CSI spectrum was obtained from a specific region of the heart, we can determine whether there are any regional changes in the amounts of PCr and ATP in chronically infarcted hearts by comparing integrated signal intensities in slices of the CSI spectra. Because of the presence of the scar tissue, which contains few or no phosphorus-containing metabolites,8 12 there is less ATP and PCr in the spectra from slices 5 through 8 distal to the occlusion in the infarcted heart compared with the sham (Fig 6⇑). Conversely, PCr and ATP signal areas in the slices proximal to the ligation (slices 2 through 4) are greater in the infarcted heart than in the sham because of the remodeling that has occurred in the viable myocardium in response to the infarct. Average amounts of ATP and PCr per slice are shown in Figs 7⇓ and 8⇓, respectively. To illustrate that the changes in the amount of ATP and PCr in separate slices depend on infarct size, the data for the infarcted hearts in these two figures have been divided into two subgroups, those with small (<15% of LV mass) and large (>15% of LV mass) infarcts, each compared with shams. Correlation coefficients relating ATP and PCr amounts per slice and infarct size for all hearts are given in these figure legends.
Fig 7⇑ shows the average amount of ATP in each of the eight slices of hearts with large and small infarcts compared with shams. The total amount of ATP per heart is equal to the area under each curve. For hearts with either small or large infarcts, ATP content of the entire heart is not significantly decreased from that of shams. However, ATP distribution across the slices differs in large versus small infarcts. For hearts with large infarcts, there is more ATP in slices 1 through 4 proximal to the site of occlusion and less ATP in slices 6 and 7 distal to the site of occlusion than in corresponding slices of shams. Hearts with small infarcts show a comparable but attenuated pattern. This pattern corresponds to the presence of scar tissue distally and increased girth in the heart due to hypertrophy of existing myocytes in regions proximal to the site of occlusion. The changes in ATP amounts per slice correlate with infarct size, inversely for slices that contain the infarct and directly for slices proximal to the site of occlusion. These data show that the extent of change in the regional distribution of ATP depends on the size of the infarct.
In contrast to conservation of total ATP, the total amount of PCr in the infarcted hearts (area under each curve shown in Fig 8⇑) is decreased compared with that of shams. The total (ie, nonregional) amount of PCr was lower for hearts with both small and large infarcts, but the decrease is greater in hearts with large infarcts. Regional analysis shows that the amount of PCr per slice is lower in all the slices distal to the site of occlusion. This decrease correlates with infarct size. In the slices proximal to the occlusion, the amount of PCr per slice is similar to that for the shams for both large and small infarcts. For hearts with the larger infarcts, the pattern of PCr shows a distribution similar to that observed for ATP: the slice with maximum signal intensity shifts proximally, from slice 4 or 5 in the shams to slice 3 or 4 in the hearts with the larger infarcts.
Estimating Metabolite Concentrations by Combining 1H MRI and 31P CSI
CSI defines the amount of each metabolite in the slice. To calculate metabolite concentrations, the amount of metabolite needs to be related to the amount of tissue in each slice. This was obtained from the 1H MR images. Dividing the metabolite amounts in each slice obtained from the 31P spectroscopic images by slice volume obtained from the 1H images provides an estimate of the metabolite concentrations (ie, amount per unit volume) in each slice. Results from this calculation are shown in Fig 9⇓ for the signal-containing slices distal to the site of occlusion and the first slice proximal to the site of occlusion (which has the highest 31P signal intensities).
In the sham hearts, ATP and PCr concentrations are not different among slices. This is expected, because all the slices in the sham hearts contain only viable myocytes, and biochemical analyses have shown essentially uniform distribution of both ATP and PCr concentrations in left and right ventricular tissue.8 9 In the infarcted hearts, in contrast, ATP and PCr concentrations in slices that include both viable myocardium and scar (slices distal to the site of occlusion) are lower. This is because slices 5 through 7 consist partially of scar tissue containing essentially no phosphorus-containing metabolites. Moving proximally from slice 7, the tissue concentrations increase toward those measured in the sham hearts. This is because the slices contain progressively less scar tissue. Average ATP concentration of the infarcted hearts in slice 4, which is proximal to the site of ligation and contains only remodeled myocardium (ie, no scar), is indistinguishable from that in the sham hearts (P=.71). This analysis shows that the ATP concentration in remodeled viable myocardium is the same as that for sham myocardium. This is supported by biopsy sample measurements, which also show indistinguishable ATP concentrations.12
Tissue concentrations of PCr in the infarcted hearts also increase as one moves proximally from slice 7. However, unlike ATP, PCr concentration for the infarcted hearts in the slice containing only remodeled myocardium (eg, slice 4) is always less than for sham hearts. PCr concentrations in all signal-containing slices proximal to the infarct (slices 2 through 4) are not statistically different from each other (P=.16). These results show that PCr concentrations are uniformly decreased in slices proximal to the site of occlusion.
The unchanged ATP concentration and decreased PCr concentration in remodeled myocardium can also be deduced by comparing tissue masses shown in Fig 4⇑ and metabolite amounts in Figs 7⇑ and 8⇑. The increased tissue masses in slices 1 through 4 in Fig 4⇑ are comparable to the increased amounts of ATP in Fig 7⇑ (hence the unchanged ATP tissue concentration), but the increased masses are greater than the increased amounts of PCr in Fig 8⇑ (hence the decreased PCr tissue concentration).
Since the amount of ATP in each slice is directly proportional to the amount of viable myocardium in that slice, we may use the ATP signal as an index of viable myocardial mass. This is particularly important in slices 5 through 7, in which we do not know what proportion is made up of scar. Normalizing PCr content by ATP content instead of total tissue mass eliminates the contribution from nonviable scar tissue. Within each group, PCr/ATP ratios, shown in Fig 9⇑, were indistinguishable among the slices for sham hearts (P=.26) and among the slices for infarcted hearts (P=.96). However, the absolute value of the mean PCr/ATP ratio for infarcted hearts (1.2±0.1) was significantly lower than for sham hearts (1.7±0.1, P=.004). The decrease in PCr/ATP also shows that the decreased PCr tissue concentration is a result not only of the presence of scar tissue in the slices distal to the infarct but also of decreased PCr in remodeled myocardium. Moreover, the constancy of the PCr/ATP ratio across all slices in the infarcted hearts confirms that the decrease in PCr is uniformly distributed in remodeled myocardium in the scar-containing slices distal to the infarct.
We used high-field (9.4-T) 31P NMR spectroscopy and spectroscopic imaging coupled with 1H MR imaging of the chronically infarcted isolated perfused rat heart to define regional changes in high-energy phosphate contents in remodeled viable myocardium. Combining these techniques allowed the determination of both amount and concentration of ATP and PCr in eight separate 2.5-mm-thick slices in these 1.5-g hearts. Because scar contains little or no ATP or PCr,8 12 31P signals report metabolite distribution only in viable remodeled myocardium. Compared with shams, the whole-heart ATP amount was unchanged; however, the regional distribution of ATP was altered both by the presence of the scar tissue and by hypertrophy proximal to the site of occlusion. Regional PCr content also differed. Unlike ATP, PCr amount and concentration were uniformly decreased in the noninfarcted viable region.
Whole-Heart PCr and ATP Contents
31P NMR spectroscopy of whole hearts and sums of chemical shift images of separate regions both showed that the total amount of PCr in intact hearts containing infarcts decreased by 22%. Biochemical assay of tissue homogenates prepared from these hearts showed a similar (27%) decrease in total creatine content in the infarcted hearts. This suggests that the proportion of creatine that is phosphorylated is similar in remodeled myocardium and in sham hearts. The decrease in the total creatine pool is similar to that observed by others 8 weeks after infarction.8 12 A major result from the present study shows that the magnitude of the decrease depends on infarct size. Use of only the three largest infarcts (28%, 35%, and 40% of LV by mass) from this study gives average decreases of 33% for PCr by NMR spectroscopy, 39% for PCr by CSI, and 32% for total creatine. Consistent with this finding, Neubauer et al12 observed decreases of 31% for PCr and 35% for total creatine in hearts with infarcts of comparable sizes. Biopsy samples of hearts of patients with coronary artery disease also showed 40% decreases in PCr10 and total creatine.11 31P localized spectroscopy in patients with coronary artery disease31 and chronic MI19 showed comparable decreases in PCr. Thus, our results using 31P spectroscopic imaging are consistent with these observed decreases in the guanidino pool.
In contrast to these changes in the amount of PCr, the amount of ATP measured by 31P NMR was not decreased in viable myocardium of chronically infarcted rat hearts compared with the shams. This is in agreement with the results of Neubauer et al,12 who used 31P NMR spectroscopy of whole heart and chemical assay, but in disagreement with Sanbe et al,8 who reported a 20% reduction in ATP in residual myocardial tissue samples. This apparent discrepancy may be explained by the fact that Sanbe et al8 related high-energy phosphate amounts to dry tissue weight. ATP levels calculated in this way are most likely lower because the nonmyocytic extracellular matrix increases in residual intact LV tissue (hydroxyproline content doubled in their study). Swain et al10 showed essentially unchanged ATP amounts in biopsy samples of patients with coronary artery disease. 31P localized spectroscopic studies in patients reported either decreased (chronic MI19 ) or unchanged (coronary artery disease31 ) ATP contents. Our results are consistent with the reports showing unchanged ATP amounts in remodeled myocardium.
Magnetic Resonance Imaging
Morphological remodeling in response to the presence of an infarct was detectable in the 1H MR images as hypertrophy localized to the slices proximal to the site of occlusion. The 1H MR images showed uniformly more signal intensity in the infarcted hearts because of their larger size, but importantly, the images did not distinguish scar tissue from viable myocardium. In some of the hearts, we attempted to identify the scar tissue by incrementally increasing TE up to 120 ms; however, as the overall image intensity progressively decreased, we were still unable to distinguish scar tissue in these T2-weighted images. The inability to distinguish chronically infarcted scar tissue from viable myocardium on the basis of differences in relaxation time constants is consistent with the results of other groups. In acute MI, the infarcted region exhibits increased signal intensity compared with noninfarcted myocardium in T2-weighted images.32 33 34 35 This is because of the edema that develops in the injured tissue. However, as the scar matures, the increase in signal intensity is no longer observed, since the T2 of the infarcted region becomes indistinguishable from that of viable myocardium.36 37 In contrast, 31P NMR spectroscopic images identified the location of the infarct along the dimension of the phase-encoding gradient.
Regional ATP and PCr Concentrations
Combining data from corresponding 1H images and 31P CSI spectra allows determination of tissue concentrations of ATP and PCr in the eight slices encompassing the entire rat heart. Our results show that ATP concentration in remodeled myocardium is indistinguishable from that in normal myocardium. This conclusion is supported by three lines of evidence. First, the total amount of ATP determined both by conventional NMR and by CSI was the same in infarcted and sham hearts. Second, tissue concentrations of ATP in all slices in the shams and in the slices of infarcted hearts containing only viable myocardium proximal to the site of ligation were not different. Finally, ATP concentrations measured in biopsy specimens obtained from remodeled viable myocardium (containing no scar) are the same as for normal myocardium.12 Given equal ATP tissue concentrations, the increased amount of ATP in the slices proximal to the site of ligation must represent increased myocardial mass. Thus, the results presented here show that the amount of ATP in each slice provides a profile of the hypertrophic remodeling that has occurred in viable tissue in response to the presence of an infarct.
PCr concentrations in all slices containing only remodeled myocardium are indistinguishable. This shows that the decrease in PCr concentration is uniform proximal to the site of ligation. Furthermore, since ATP is unchanged in remodeled myocardium, it can be used to normalize PCr concentration in viable myocardium of the scar-containing slices distal to the site of ligation. The constancy of the PCr/ATP ratio observed across the slices of the infarcted hearts indicates that, within the resolution of this method (2.5 mm), the PCr concentration is also uniformly decreased in remodeled myocardium of chronically infarcted hearts distal to the site of ligation. Thus, a major result of the work reported here is that the decrease in the amount of PCr per heart appears to be equally distributed across the heart even though the infarcted region is confined to the LV apex and anterolateral free wall. Constant PCr/ATP ratios among the different slices within the infarcted heart group and within the sham heart group are observed despite relatively large differences in the amounts of ATP and PCr in individual slices. This illustrates that measuring only PCr/ATP ratios can mask regional differences in the amounts of these two metabolites. Because the presence of an infarct decreases the PCr concentration in the entire remaining viable myocardium, these results underscore the importance of intervening to prevent ischemic myocardium from progressing to infarction.
Decreased PCr levels have been reported for several pathophysiological states in the heart, including the transition from compensated hypertrophy to failure and frank failure.11 12 38 39 Sanbe et al8 presented evidence that rat hearts with infarcts similar in size to those reported here are failing. Thus, this decrease in the PCr and creatine pools may be a common characteristic of failing myocardium. The correlation between the decrease in PCr and infarct size reported here suggests that loss of the primary energy reserve compound may be one factor explaining why hearts with large infarcts are more likely to progress to heart failure than hearts with small infarcts.
In summary, our results using in vivo 31P spectroscopic imaging techniques in chronically infarcted hearts support two conclusions: (1) PCr decreases uniformly in the noninfarcted remodeled myocardium, and this decrease correlates with infarct size; and (2) distribution of ATP provides a profile of viable myocardium and is a biochemical marker of the morphological remodeling (regional hypertrophy) that has occurred in the different regions of the noninfarcted viable myocardium. Thus, the technique provides both a marker that tissue injury has occurred (decreased PCr) and a marker of remodeling in response to the injury (ATP distribution) in a single set of noninvasive measurements.
First, the CSI technique used here delineates the infarct and remodeled hypertrophic region along only one dimension, and 2.5-mm resolution provides only eight slices. Although higher resolution is always desirable, increasing the resolution by decreasing the width of each slice or increasing the number of spatial dimensions (while keeping the data acquisition time constant) would decrease the sensitivity of the experiment. This is because each region would be smaller and contain less signal. 31P NMR, which detects phosphorus metabolite concentrations in the millimolar range, is much less sensitive than 1H NMR, which detects water with a hydrogen concentration around 80 mol/L in myocardium. Although sensitivity limits resolution for the small hearts studied here, this approach should be useful clinically.
Second, there are several factors that could cause imprecise delineations of the slices used to compare sham and infarcted hearts and also between 1H images and 31P CSI. First, 31P CSI experiments were not gated to the cardiac cycle, so heart movement during data acquisition could alter registration. We expect this overlap to be negligible because during contraction, heart movement in the Langendorff perfusion mode is primarily horizontal (ie, perpendicular to the spatial dimension in our study). Second, the positioning of each heart is another potential source of error. As discussed in the “Methods” section, each heart was centered at the same vertical height. We estimate that this positioning error was <1 mm. Third, valid comparison of 1H images and 31P CSI requires that the position of the heart did not change when the heart was arrested and the bath removed. We observed the position of Langendorff-perfused hearts outside the magnet while arresting them with KCl-containing buffer and beginning effluent suction from below, and we could not observe any changes in position. This is probably because in our preparation, the left ventricle is occupied by an isovolumic fluid-filled balloon imparting a constant, normal level of wall stress. This minimizes LV sagging or distortion. Finally, we did not take into account the small overlap of adjacent slices inherent in the method of applying phase-encoding gradients in the CSI experiment and imperfect selective pulses in the imaging experiment.
Several studies of diseased myocardium in humans have demonstrated the feasibility of applying 31P NMR spectroscopic techniques to a clinical setting.13 14 15 16 17 18 19 31 40 41 Our results from the intact functioning isolated heart demonstrate the feasibility of obtaining measurements of regional high-energy phosphate metabolism. The same methodology could be applied to human subjects to yield regional metabolic data that may be highly pertinent to clinical treatment.
Decreases in PCr/ATP ratio or PCr have been observed in acute18 and chronic19 MI and in patients with coronary artery disease.31 In addition, 31P NMR spatial localization and imaging studies have shown decreases in PCr/ATP of 15% to 22%15 and 19%16 in patients with dilated cardiomyopathy and 27% in aortic valve disease17 accompanied by heart failure. Furthermore, some of these studies were able to demonstrate either unchanged13 16 17 or smaller15 decreases in overall PCr/ATP in patients with less severe or no signs of congestive heart failure. Although no significant correlations between PCr/ATP ratio and LV ejection fractions were observed,15 16 the magnitude of the decrease in PCr/ATP did correlate with New York Heart Association class.16 These studies suggest that overall changes in PCr/ATP may be dependent on the degree of heart failure. The results presented here from an experimental animal model of chronic infarction showed a comparable mean overall decrease in PCr/ATP and a significant correlation with infarct size. Thus, our results in this well-controlled setting provide a foundation for the observations made in the human studies.
Clinical cardiac decisions are now guided largely by hemodynamic observations. Poor prognostic signs, such as a decreasing ejection fraction or increasing LV cavity diameter, provide part of the basis for either medical or surgical interventions. Moreover, the evaluation of the results of such interventions uses the same hemodynamic parameters to judge success or failure. Our results suggest the possibility of using noninvasive measures such as the PCr content as a biochemical marker of myocardial metabolic failure. The values and time course of the PCr content of the noninfarcted region or the magnitude of the LV dysfunction with a decreased PCr content (or PCr/ATP ratio) may be better indexes of prognosis and better guides to intervention than currently used hemodynamic or angiographic indexes. The functional significance of a “borderline” coronary narrowing on angiography may be clarified by high-energy phosphorus–containing metabolite measurements of the region served by the vessel in question. These speculations will require clinical investigations correlating regional metabolite content abnormalities with clinical course.
Selected Abbreviations and Acronyms
|CSI||=||chemical shift imaging|
|NMR||=||nuclear magnetic resonance|
|RPP||=||product of heart rate and left ventricular developed pressure|
This research was supported in part by National Institutes of Health grants HL-43170 and HL-52320 (Dr Ingwall) and HL-48175 (Drs Apstein and Ingwall). Dr Friedrich was supported by Harvard-MIT Division of Health Sciences and Technology Johnson & Johnson and Research Fund fellowships and an AHA Medical Student Research Fellowship. We thank Drs Luigi Nascimben, Stefan Neubauer, and Franz Eberli for helpful discussions. We are grateful for the technical assistance of Souen Ngoy (Cardiac Muscle Research Laboratory, Whitaker Cardiovascular Institute, Boston [Mass] University School of Medicine), Jonathan Rose, BA, and Ilana Reis, MSc.
- Received December 27, 1994.
- Revision received June 5, 1995.
- Accepted August 1, 1995.
- Copyright © 1995 by American Heart Association
Pfeffer MA, Pfeffer JM, Fishbein MC, Fletcher PJ, Spadaro J, Kloner RA, Braunwald E. Myocardial infarct size and ventricular function in rats. Circ Res. 1979;44:503-512.
Fletcher PJ, Pfeffer JM, Pfeffer MA, Braunwald E. Left ventricular diastolic pressure-volume relations in rats with healed myocardial infarction. Circ Res. 1981;49:618-626.
Pfeffer JM, Pfeffer MA, Fletcher PJ, Braunwald E. Progressive ventricular remodeling in rat with myocardial infarction. Am J Physiol. 1991;260(Heart Circ Physiol 29):H1406-H1414.
Litwin SE, Litwin CM, Raya TE, Warner AL, Goldman S. Contractility and stiffness of noninfarcted myocardium after coronary ligation in rats: effects of chronic angiotensin converting enzyme inhibition. Circulation. 1991;83:1028-1037.
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.
Fellenius E, Hansen CA, Mjøs O, Neely JR. Chronic infarction decreases maximum cardiac work and sensitivity of heart to extracellular calcium. Am J Physiol. 1985;249(Heart Circ Physiol 18):H80-H87.
Neubauer S, Reis I, Naumann A, Tian R, Schwarz K, Ingwall JS. Regional analysis of the creatine kinase system in noninfarcted tissue of rat heart following myocardial infarction. Circulation. 1993;88(suppl I):I-246. Abstract.
Swain JL, Sabina RL, Peyton RB, Jones RN, Wechsler AS, Holmes EW. Derangements in myocardial purine and pyrimidine nucleotide metabolism in patients with coronary artery disease and left ventricular hypertrophy. Proc Natl Acad Sci U S A. 1982;79:655-659.
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.
Neubauer S, Krahe T, Schindler R, Horn M, Hillenbrand H, Entzeroth C, Mader H, Kromer EP, Riegger GAJ, Lackner K, Ertl G. 31P magnetic resonance spectroscopy in dilated cardiomyopathy and coronary artery disease: altered high-energy phosphate metabolism in heart failure. Circulation. 1992;86:1810-1818.
Robitaille P-ML, Merkle H, Lew B, Path G, 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.
Selye H, Bajusz E, Grasso S, Mendell P. Simple techniques for the surgical occlusion of coronary vessels in the rat. Angiology. 1960;11:398-407.
Maclean D, Fishbein MC, Braunwald E, Maroko PR. Long-term preservation of ischemic myocardium after experimental coronary artery occlusion. J Clin Invest. 1978;61:541-551.
McLeod K, Comisarow MB. Systematic errors in the discrete integration of Fourier transform NMR spectra. J Magn Reson. 1989;84:490-500.
Lowry OH, Rosebrough MJ, Farr AL, Randall RJ. Protein measurement with the Folin reagent. J Biol Chem. 1951;193:265-275.
Bak MI, Ingwall JS. NMR-invisible ATP in heart: fact or fiction? Am J Physiol. 1992;262(Endocrinol Metab 25):E943-E947.
Clarke K, Anderson RE, Nedelec J-F, Foster DO, Ally A. Intracellular and extracellular spaces and the direct quantification of molar intracellular concentrations of phosphorus metabolites in the isolated rat heart using 31P NMR spectroscopy and phosphonate markers. Magn Reson Med. 1994;32:181-188.
Yabe T, Mitsunami K, Okada M, Kinoshita M, Morikawa S, Inubushi T. Quantitative measurements of phosphorus metabolites in coronary artery diseases by 31P slice-selected one-dimensional chemical shift imaging. Proc SMR, Second Meeting, San Francisco, 1994. p 1220.
Pflugfelder PW, Wisenberg G, Prato FS, Carroll SE, Turner KL. Early detection of canine myocardial infarction by magnetic resonance imaging in vivo. Circulation. 1985;71:587-594.
Johnston DL, Brady TJ, Ratner AV, Rosen BR, Newell JB, Pohost GM, Okada RD. Assessment of myocardial ischemia with proton magnetic resonance: effects of a three hour coronary occlusion with and without reperfusion. Circulation. 1985;71:595-601.
McNamara MT, Higgins CB, Schechtmann N, Botvinick E, Lipton MJ, Chatterjee K, Amparo EG. Detection and characterization of acute myocardial infarction in man with use of gated magnetic resonance. Circulation. 1985;71:717-724.
Johnston DL, Homma S, Liu P, Weilbaecher DG, Rokey R, Brady TJ, Okada RD. Serial changes in nuclear magnetic resonance relaxation times after myocardial infarction in the rabbit: relationship to water content, severity of ischemia, and histopathology over a six-month period. Magn Reson Med. 1988;8:363-379.
Ingwall JS. Is cardiac failure a consequence of decreased energy reserve? Circulation. 1993;87(suppl VII):VII-58-VII-62.
Nascimben L, Friedrich J, Liao R, Pauletto P, Pessina AC, Ingwall JS. Enalapril treatment increases cardiac performance and energy reserve via the creatine kinase reaction in myocardium of Syrian myopathic hamsters with advanced heart failure. Circulation. 1995;91:1824-1833.
Heatherington HP, Luney DJE, Vaughan JT, Pan JW, Ponder SL, Tschendel O, Twieg DB, Pohost GM. 3D 31P spectroscopic imaging of the human heart at 4.1T. Proc SMR, Second Meeting, San Francisco, 1994. p 86.