Myocardial Gd-DTPA Kinetics Determine MRI Contrast Enhancement and Reflect the Extent and Severity of Myocardial Injury After Acute Reperfused Infarction
Background Contrast medium–enhanced magnetic resonance images of acute, reperfused infarcts have shown hypoenhanced and hyperenhanced regions in areas of injured myocardium. The precise mechanisms that lead to these altered enhancement patterns are unknown. This study was designed to evaluate possible mechanisms and to relate altered enhancement patterns to myocardial perfusion and viability.
Methods and Results Thirteen rabbits underwent in situ coronary artery occlusion and reperfusion followed by isolated perfusion with cardioplegic solution. T1-weighted spin-echo images were acquired continuously during step changes in perfusate Gd-DTPA concentration. Regional blood flow was also measured by use of radioactive microspheres in all rabbits. There were marked differences in Gd-DTPA wash-in and washout time constants (wash-in, 0.8±0.1, 2.1±0.2, and 16.3±2.4 minutes, P<.001; washout, 1.6±0.1, 4.8±0.5, and 31.1±3.3 minutes, P<.001) in normal, infarct rim, and infarct core regions, respectively, resulting in differential enhancement of these regions. Microsphere flows in the infarct rim and core were 42.9±4.0% and 12.0±1.6% of normal myocardium and correlated well with washout time constants (r=.86, y=0.77x−0.002, P<.001), suggesting that these time constants index the severity of microvascular damage. In addition, spatial maps of washout time constants were produced. The extent of regions with abnormal time constants correlated well with triphenyltetrazolium chloride–determined infarct size (r=.94, y=0.95x+4.17, P<.001).
Conclusions In contrast-enhanced magnetic resonance images of acute, reperfused rabbit infarcts, differential image intensity is primarily due to regional differences in contrast agent wash-in and washout time constants. These regional differences in time constants also indicate the extent and severity of myocardial injury.
Several recent studies have focused on the potential of MRI with paramagnetic contrast agents to assess the extent and type of tissue injury after myocardial infarction.1 2 3 4 5 6 7 For example, our group recently reported that in humans with acute myocardial infarction, altered MR contrast enhancement patterns were found to correlate with fixed thallium defects, regional dysfunction, prolonged occlusion of the infarct-related artery,1 and scar size at 6 months after infarct.8 In addition, a study in closed-chest dogs with 2-day-old reperfused myocardial infarcts showed that central hypoenhanced zones were related to the “no-reflow” phenomenon, whereas hyperenhanced regions correlated well with nonviable myocardium.2 These observations of altered contrast medium enhancement in injured myocardium point to the potential of using contrast-enhanced MRI as a clinical tool to portray complex pathophysiological changes in myocardial tissue noninvasively.
However, these empirical findings have contributed little to our understanding of the underlying mechanisms that cause these patterns to arise. The physical principles by which paramagnetic contrast agents change MR signal intensity have long been studied and are relatively well understood in simple aqueous solutions.9 What is not well understood is the role of the myocardial tissue environment in determining the patterns of contrast enhancement. For example, it has been hypothesized that hyperenhancement after ischemic injury is caused by increased contrast medium volume of distribution.10 Others suggest that injured tissue contrast enhancement is caused by changes in contrast medium wash-in and washout compared with normal tissue.2 5 11
To explore these issues, we studied myocardial contrast enhancement in a rabbit model of acute, reperfused myocardial infarction. Similar rabbit models have been studied in the past by our group,12 13 and this model was specifically chosen to reproduce the infarct enhancement patterns observed in vivo, only under a controlled setting. In situ infarction and reperfusion were followed by crystalloid perfusion of the isolated, intact heart so that perfusate contrast agent concentration could be directly manipulated and contrast recirculation eliminated. This approach allowed us to perform sequential imaging after step changes in perfusate contrast concentration, thereby allowing us to determine the importance of regional differences in contrast agent volumes of distribution and wash-in/washout kinetics in determining myocardial enhancement patterns. To investigate the physiological basis of these altered enhancement patterns, we compared regional flow measured by radioactive microspheres with regional contrast agent kinetics. This allowed us to test whether MR enhancement patterns are related to the severity of perfusion impairments caused by infarction and/or reperfusion. In addition, the spatial extent of altered enhancement, processed by a variety of different methods, was compared with infarct size determined histochemically to evaluate the potential utility of contrast-enhanced MRI in quantifying acute myocardial infarction.
Twenty New Zealand White rabbits (3.5 to 4.0 kg) were anesthetized with intravenous sodium pentobarbital (≈27 mg/kg), intubated, and mechanically ventilated. The care and treatment of all animals involved in this study were in accordance with the Position of the American Heart Association on Research Animal Use, adopted November 15, 1984. A median sternotomy was performed, and a reversible snare ligature was placed around an anterior branch of the left coronary artery. After 35 minutes of in situ occlusion followed by 60 minutes of reperfusion, the hearts (n=13) were rapidly excised and retrogradely perfused with cardioplegic solution at room temperature. Five additional rabbits died before completion of the ischemic occlusion period. The hearts were perfused at constant pressure with flow set to 10 mL/min (≈1.0 to 1.5 mL·min−1·g−1) as measured with an in-line electromagnetic flowmeter (model 1401, Skalar Medical). The perfusate was not recirculated. Perfusate composition (in mmol/L) was Na+ 120, K+ 16, Mg2+ 16, Cl− 160, and HCO3− 10.14 Adenosine was added to ensure maximal coronary vasodilation (0.5 mmol/L), and the perfusate was equilibrated with a 95% O2/5% CO2 gas mixture to maintain pH at 7.4 to 7.5.15 We have previously shown that hearts isolated in this manner remain viable for several hours.14 16 After a small (1-mm-diameter) Gd-DTPA–filled polyethylene tube was attached to the right ventricle to locate the infarct territory, the hearts were hung vertically in a 30-mm-diameter radiofrequency volume coil and placed in the magnet.
Imaging was performed on a GE/Bruker 4.7-T Omega system. Spin-echo imaging parameters were TR, 300 ms; TE, 16 ms; field of view, 50 mm; average number, 1; matrix size, 256×128; voxel size, 0.2×0.4×1.5 mm; and image acquisition time, 38 seconds. Slice selection over the infarcted territory was achieved by acquisition of T1-weighted LV short-axis scout images until the epicardial marker was visualized.
Sequential images were obtained before, during, and after contrast (Magnevist Gd-DTPA; Berlex/Schering, gadopentetate dimeglumine) infusion. Step changes in contrast concentration (from 0 to 1.0 mmol/L for 31 minutes, then back to 0 mmol/L for 30 minutes) were performed by rapid turning of a stopcock to switch between perfusate reservoirs. A total of 40 images (20 each for wash-in and washout) were acquired to evaluate the myocardial wash-in and washout kinetics of Gd-DTPA.
Myocardial Perfusion Measurements/Histology
After MR imaging, ≈2×105 sonicated radioactive microspheres (DuPont; 153Gd, 15±1-μm diameter) were injected into the perfusate line. Each heart was then cut into short-axis slices. The imaged slice, which was easily identified by the epicardial marker (see “MR Methodology”), was photographed to determine the extent of gross hemorrhage and then incubated in a 1% TTC solution at 37°C to 40°C for 15 minutes. Since TTC forms a red precipitate in the presence of intact dehydrogenase enzyme systems and reducing coenzymes, viable myocardium stains brick red, whereas necrotic areas fail to stain.17 The TTC-stained myocardial slice was also photographed, and the resultant 35-mm slides were digitally scanned for subsequent analysis.
Tissue samples (20 to 100 mg) were then obtained for microsphere counting from the following areas of myocardium: (1) the central core of the infarct region (determined by TTC staining), (2) the subepicardial (1.5 mm) rim of the infarct region, and (3) a remote, normal region of myocardium on the opposite side of the LV cavity from the infarcted region. The location of each of the three regions and outlines of the myocardial slice and the infarct zone were traced by hand on clear acetate paper placed over the myocardial slice. Each of the samples was weighed and then counted in a well gamma emission spectrometer (Hewlett-Packard 5986), and myocardial perfusion was calculated by standard methods.18
Two additional rabbits underwent the same ischemic protocol, then their hearts were cut into five LV short-axis slices, with alternate slices of heart (slices 1, 3, and 5) stained with TTC and the remaining slices (slices 2 and 4) immersed in cold fixative (4% formaldehyde/1% glutaraldehyde). After infarct and noninfarct zones were identified in the TTC-stained slices, two tissue specimens (2 to 3 mm3) from the infarcted region and one specimen from a normal region were taken from the corresponding surfaces of the fixed slices and processed for electron microscopy.19 No attempt was made to sample tissue from infarct rim regions, because the infarct border was indirectly ascertained from the TTC slices. The frequency of myocyte membrane rupture was determined by counting of all cells within five contiguous grid squares (each 90×90 μm2) in each sample and determination of the percentage of cells with sarcolemmal tears. Intracapillary plugging was defined as the occurrence of more than five contiguous erythrocytes per capillary in longitudinal section. The results were expressed as the average number of plugged capillaries seen in each 90×90-μm2 grid.
Since the epicardial marker had guided selection of both the TTC and MR slices, the previously traced outlines of the LV, the infarct zone, and the three regions sent for microsphere counting from the TTC-stained slice were superimposed over the MR images to guide selection of MR image ROIs. First, the region of altered contrast enhancement (hypoenhanced or hyperenhanced) was determined visually from the entire set of contrast wash-in and washout images. Next, the MR images were scaled and rotated (by use of NIH Image) to match the region of altered contrast enhancement with the TTC infarct outline. The traced outlines for the three tissue samples sent for microsphere sampling were then used to draw comparable regions on the MR images. In all hearts, the region of altered enhancement on the MR images was similar to the region of abnormal TTC staining. Myocardial image time-intensity curves were then generated within the three ROIs (core of altered enhancement, subepicardial rim of altered enhancement, and remote/normal region) for the contrast wash-in and washout time periods. Time-intensity curves were normalized by expression of all signal intensities as a percentage of the precontrast normal myocardium. Signal intensity ratios were statistically compared for the three ROIs at five different time points: before contrast arrival, after 2.5 and 28.5 minutes of contrast wash-in, and after 3 and 25 minutes of contrast washout. Regional signal intensity wash-in and washout time constants were determined by fitting the time-intensity curves to monoexponential curves.
Monoexponential curve-fitting for contrast agent washout was also performed on a pixel-by-pixel basis. This was done to form an image of washout time constants (where pixel intensity represents the myocardial τwo for that pixel location) independent of operator biases in ROI selection.
Spatial Extent of Altered Enhancement
MR regions were measured by two independent observers in three different images: (1) an early (3 minutes) contrast washout image, (2) a late (25 minutes) contrast washout image, and (3) the τwo image. For each image, observers were instructed to measure myocardial regions with pixels >2 SD (of normal tissue) higher than the mean value. The area of increased signal intensity was expressed as a percentage of the LV myocardium, and the results for the two independent observers were averaged. Infarct size was measured by a third independent observer using the digitized TTC images and compared with the MR measurements. The extent of gross hemorrhage seen on digitized pre–TTC staining images was also measured and compared with TTC-negative regions.
All results were expressed as mean±SEM. Repeated-measures ANOVA20 was used to test the hypotheses that signal intensities varied with time in different regions of the same heart (repeated measures on both factors), that wash-in/washout time constants varied in different regions, and that microsphere flows varied in different regions. Differences between specific regions were isolated by Bonferroni t tests.20 The correlation between extent of infarction determined histologically and by MRI was determined by least-squares linear regression analysis, as was the correlation between extent of infarction and gross hemorrhage and the correlation between τwos and microsphere flow. The spatial extent of τwo abnormalities was compared with TTC-determined infarct size by a paired t test. Values of P<.05 were deemed significant.
The upper two rows (A through F) of Fig 1⇓ depict typical MR images acquired before, during wash-in, and during washout of contrast. Before contrast administration (Fig 1A⇓), image intensity in all myocardial regions was homogeneous and low, consistent with our in vivo observations in dogs and humans.1 2 After contrast administration, normal myocardium showed a rapid homogeneous increase in signal intensity, with little further change after the first 2 to 3 minutes. However, in the region visually corresponding to infarcted myocardium (Fig 1H⇓), a hypoenhanced region appeared 2.5 minutes after contrast infusion (Fig 1B⇓). After 9 minutes of contrast wash-in (Fig 1C⇓), the periphery of the formerly hypoenhanced region had increased substantially in signal intensity, but central hypoenhanced regions still remained. By 28.5 minutes (Fig 1D⇓), almost the entire region had filled in and in fact was modestly hyperenhanced compared with normal tissue. During contrast washout, the findings were nearly symmetrical to those of wash-in but with reversed direction. Normal tissue quickly decreased in signal intensity while a hyperenhanced region appeared (Fig 1E⇓), again corresponding in location to infarcted tissue. By 25 minutes (Fig 1F⇓), the periphery of the hyperenhanced region had decreased in signal intensity back to the precontrast state, but central patches with increased signal intensity still remained.
Fig 2⇓ shows selected images from two different rabbit hearts. Fig 2A1⇓⇑ demonstrates (in contrast to Fig 1D⇑) that central regions of the infarct were occasionally still hypoenhanced at 28.5 minutes of contrast wash-in. Fig 2B1⇓⇑ demonstrates minimal differential enhancement compared with normal tissue in either central or peripheral regions of the infarct and shows that MR images after long contrast infusion frequently do not delineate infarcted regions.
Normalized Time-Intensity Curves
Fig 3⇓ shows normalized time-intensity curves (n=13) obtained from the remote (normal) region and the core and rim of altered enhancement. In normal regions, contrast wash-in and washout were rapid, reaching steady state within 2 to 3 minutes (0.8±0.1 and 1.6±0.1 minutes for wash-in/washout time constants, respectively). Rim regions had significantly delayed wash-in and washout compared with normal regions (2.1±0.2 and 4.8±0.5 minutes, respectively; P<.001 for both). Core regions had significantly delayed wash-in and washout compared with both normal and rim regions (16.3±2.4 and 31.1±3.3 minutes; P<.001). Contrast kinetics in core regions were so slow that signal intensity did not reach steady state after 30 minutes of either contrast wash-in or washout.
Signal Intensity Ratios
Fig 4⇓ plots the signal intensity ratios of altered enhancement regions compared with remote (normal) regions for five time points. Before contrast infusion, there were virtually no differences in signal intensity. The core and rim were 97.1±4.1% and 102.6±1.7% of normal tissue (both P=NS). Early during contrast wash-in, both the core (40.6±2.1%; P<.001) and rim (84.9±3.6%; P<.05) were hypoenhanced compared with normal tissue, but by the end of wash-in, the core was, on average, as intense as normal (95.9±6.6%; P=NS), whereas the rim showed only mild hyperenhancement (114.8±2.7%; P<.05). Early during contrast washout, both core and rim were significantly hyperenhanced (240.1±17.2% and 196.9±11.9%, respectively; both P<.001). Late during contrast washout, the core was still significantly hyperenhanced (209.4±10.7%, P<.001), but the rim had nearly returned to baseline (120.6±5.8%, P=NS).
Figs 1I, 2A3, and 2B3⇑⇑⇑⇑⇑ are examples of images constructed from washout time constants calculated on a pixel-by-pixel basis. Comparison with the TTC-stained images shows that regions with abnormal washout kinetics clearly delineate infarcted regions. Compared with the MR images, τwo images show better spatial contrast between normal and abnormal tissue, because regional differences in τwos (≈20- to 40-fold higher in core than in normal) were much larger than differences in MR signal intensity (≈2- to 3-fold higher in core than in normal).
Relationship of Image Enhancement Patterns to Infarct Regions
Fig 5⇓ shows the spatial extent of hyperenhancement (signal intensity at least 2 SD above normal tissue) seen in early contrast washout images, late contrast washout images, and τwo images plotted against the spatial extent of TTC-negative regions. MR measurements correlated well with TTC measurements, although early washout images frequently overestimated TTC-negative regions (r=.88, y=0.85x+14.0, SEE=8.00, P<.001) and late washout images underestimated TTC-negative regions (r=.95, y=0.81x−4.57, SEE=4.75, P<.001). Regions defined by prolonged washout time constants on τwo images best approximated infarct size (r=.94, y=0.95x+4.17, SEE=5.65, P<.001). No significant differences in the sizes of τwo hyperenhanced versus TTC-negative regions were found (P=NS; mean, 25.9% versus 22.9%; range, 2.6% to 69.9% versus 3.2% to 64.5%). Repeat analysis of early washout images and τwo images using an intensity threshold of 4 SD above normal tissue showed a large decrease in the extent of hyperenhancement in early washout images (y=0.80x−0.76) but only a small decrease in extent of hyperenhancement in τwo images (y=0.89x+1.48). Again, no significant differences in the sizes of τwo hyperenhanced versus TTC-negative regions were found (P=NS).
Relationship of Microsphere Flow to τwo
Microsphere flow in the infarct rim and core were 42.9±4.0% and 12.0±1.6% of remote-region flow, respectively. These regional differences in flow were compared with regional differences in τwos from corresponding locations. The inverse of normalized washout decay time (remote-region τwo/τwo) correlated well with normalized flow (flow/remote-region flow) in a pooled analysis of all infarcted regions (Fig 6⇓; r=.86, y=0.77x−0.002, SEE=0.08, P<.001).
Relationship of Gross Hemorrhage to Infarct Regions
Fig 1G⇑ shows gross hemorrhage within the infarct region. Areas of hemorrhage were predominantly subendocardial and were always within areas of necrosis. Fig 7⇓ shows that there was excellent correlation between the extent of hemorrhage and the extent of infarction (r=.99, y=0.80x−1.24, SEE=2.30, P<.001). On average, hemorrhagic areas were 69±6% of necrotic regions.
In contrast to specimens from normal regions displaying intact myocyte ultrastructure (Fig 8A⇓), tissue taken from infarcted regions displayed marked cellular edema with subsarcolemmal blebs and frequent myocyte membrane rupture (Fig 8B⇓). In infarcted regions, 98 of 111 myocytes (88%) had ruptured membranes, whereas in normal regions, 0 of 89 myocytes (0%) had ruptured membranes. Other diagnostic features of myocardial necrosis were seen, such as hypercontracted and overstretched sarcomeres producing contraction bands and swollen, disrupted mitochondria with amorphous densities. Extensive microvascular damage was seen, with endothelial thinning and disruption. There were also frequent examples of capillary plugging with red blood cell stasis, suggesting mechanical obstruction to flow (Fig 8C⇓). Normal regions had, on average, 0.1±0.1 capillaries with red cell stasis per 90×90-μm2 grid, whereas infarcted regions had 3.7±0.4 capillaries with red cell stasis over the same tissue area (P<.001).
Although the infarction-reperfusion protocol used in this study was performed in situ, all MR data reported here were acquired in isolated hearts. This experimental design was chosen to precisely control perfusate [Gd-DTPA], thereby allowing us to distinguish between contrast kinetics and volumes of distribution as potential mechanisms of hyperenhancement. Nevertheless, some physiological differences between our data and in vivo changes may be expected.
Mechanisms of Myocardial Enhancement
Gross Hemorrhage and Other Direct Tissue Effects
Although tissue changes such as edema or hemorrhage could lead to hyperenhanced4 or hypoenhanced21 regions, respectively, on T2-weighted images, we1 2 and others3 4 have found that standard T1-weighted imaging without contrast medium often does not delineate acute reperfused infarction. Fig 4⇑ shows no differences in precontrast signal intensities between normal and infarcted myocardium. This result suggests that hemorrhage and other tissue changes associated with infarction have little direct effect on signal intensity, given our imaging protocol. Although it is possible that tissue changes could affect contrast medium relaxivity and thus image intensity after contrast injection, this seems unlikely, because several studies have shown that tissue Gd-DTPA is present uncomplexed with protein or blood cells.22 23 Wendland et al24 showed that there is increased relaxivity of Gd-BOPTA in reperfused infarction consistent with weak binding between Gd-BOPTA and serum albumin, but this was not found to occur for Gd-DTPA, the agent used in the present study. Hemorrhage, however, could indirectly affect signal intensity, because myocardial hemorrhage is known to be associated with worse vascular injury.25 Vascular injury could then influence contrast wash-in/washout kinetics and thus regional signal intensity (see “Regional Differences in [Gd-DTPA]”).
Since MR signal is derived from protons rather than from the contrast agent itself, signal intensity may depend on the number of water protons that are exposed to the agent. Gd-DTPA has been found to equilibrate rapidly from the vasculature to the interstitium.13 26 It is excluded from the intracellular space by the cell membrane.22 23 Cellular disruption in infarcted regions could lead to increased free water access and therefore hyperenhancement without increased contrast agent concentration in infarcted regions. Although Donahue et al27 showed that there is fast intracellular-extracellular exchange of water protons without cellular disruption for Gd-DTPA concentrations, similar to the present study, we cannot exclude the possibility that this mechanism may play a role in differential contrast enhancement of injured myocardium.
Nonetheless, if signal intensity before Gd-DTPA infusion is nearly the same in all regions of myocardium, new differences after Gd-DTPA arrival are probably a result of regional differences in tissue [Gd-DTPA]. This conclusion is consistent with the results of several studies that have directly compared myocardial [Gd-DTPA], measured with either radiolabeled gadolinium6 10 28 or inductively coupled plasma–atomic emission spectrometry,29 with MR signal intensity on T1-weighted spin-echo images.
Regional Differences in [Gd-DTPA]
To date, two basic mechanisms have been proposed to account for the [Gd-DTPA] differences and hence the enhancement patterns seen in reperfused, infarcted myocardium. First, many investigators have speculated that there is an increase in the volume of distribution of Gd-DTPA in reperfused, infarcted tissue.1 4 10 Increased extracellular volume in regions of acute infarction due to interstitial edema and/or loss of cell membrane integrity would increase the volume of distribution of Gd-DTPA, with a corresponding increase in the effective voxel concentration of Gd-DTPA. Fig 8B⇑ shows an example of myocyte membrane rupture in a tissue specimen taken from infarcted myocardium. Our data show that 88% of myocytes in infarcted regions have sarcolemmal tears, compared with 0% in normal regions. This suggests that Gd-DTPA molecules in infarcted regions can passively diffuse into what used to be intracellular space and that loss of membrane integrity may indeed play a role in increased voxel concentrations of Gd-DTPA. Second, other investigators have argued that injured myocardium may also have abnormal Gd-DTPA wash-in and washout kinetics.1 2 5 11 Slow wash-in in reperfused, infarcted regions would lead to early, low tissue contrast concentrations, whereas slow washout would eventually lead to higher contrast concentrations compared with normal tissue. These changes in kinetics could be due to many factors, including changes in coronary flow rates, capillary permeability, or functional capillary density (see “Physiological Basis of Prolonged Wash-in/Washout Time Constants” for further discussion).
Our results in Figs 3 and 4⇑⇑ clearly show, at least for this ischemic model, that patterns of hypoenhancement and hyperenhancement are caused primarily by regional differences in wash-in/washout time constants. After nearly 30 minutes of continuous contrast infusion, only minor differences in signal intensity were seen. Rim regions of altered enhancement reached steady-state signal intensities, on average, 14.8±2.7% higher than normal regions. This leads to the conclusion that rim regions have only mildly increased gadolinium volumes of distribution. Core regions after 30 minutes were, on average, isointense with normal regions, although intensity was still increasing and had not reached steady state. Thus, core regions probably also have increased gadolinium volumes of distribution, but unless contrast loading is prolonged, the hyperenhancement seen in these regions is probably due to impaired contrast clearance. This conclusion is emphasized by the τwo images (Figs 1I, 2A3, and 2B3⇑⇑⇑⇑⇑), which eliminate steady-state influences on signal intensity and yet show excellent delineation of altered enhancement regions. One should be reminded, however, that these conclusions may not be valid for reperfused infarcts imaged long after the ischemic insult, in which physiological changes that occur during infarct healing may affect contrast kinetics and volumes of distribution. Tong et al30 measured tissue:blood partition coefficients in a canine model of reperfused infarction. The volume of distribution of Gd-DTPA was only slightly increased after 2 hours of reperfusion but was markedly increased after 6 days of reperfusion. In the latter situation, hyperenhancement may in large part be due to increased Gd-DTPA volumes of distribution.
Applicability to In Vivo Imaging
The pharmacokinetics of in vivo injection of Gd-DTPA have been described as a two-compartment open model with mean distribution and elimination half-lives of 0.20±0.13 and 1.58±0.13 hours, respectively.26 Since normal tissue time constants are short (1 to 2 minutes) compared with the fast component of blood Gd-DTPA clearance (12 minutes), normal tissue [Gd-DTPA] will closely follow changes in blood [Gd-DTPA].1 2 Infarct tissue time constants, however, are long (up to 30 minutes) compared with the fast component of Gd-DTPA clearance. Therefore, Gd-DTPA clearance from infarcted tissue will no longer follow blood or normal tissue clearance, and differences in myocardial signal intensity due to differences in tissue [Gd-DTPA] will be observed. Extrapolation of these results to enhancement patterns observed in vivo in humans1 31 and in animals2 5 6 would lead to the conclusion that hypoenhancement is due to delayed contrast penetration, and hyperenhancement is primarily due to slow contrast washout. The specific enhancement pattern seen will depend on regional differences in tissue wash-in/washout kinetics, as well as the time after injection of contrast when the image is acquired.
Physiological Basis of Prolonged Wash-in/Washout Time Constants
The results of Burstein et al32 and Johnston et al33 suggest that coronary flow rate probably has minor effects on either the kinetics of contrast enhancement or the steady-state T1 unless coronary flow is markedly reduced. Fig 6⇑, however, shows a good correlation between microsphere-determined flow in infarct regions and washout time constants. This apparent discrepancy emphasizes the importance of distinguishing between studies that evaluate MR contrast enhancement under conditions of acute ischemia from studies of reperfused infarction. In the latter case, patency of the infarct-related artery leads to early hyperemic flow followed by progressive impairment of flow (“no-reflow” phenomenon).19 34 Tissue hypoperfusion is due to obstruction of flow at the capillary level rather than in large epicardial vessels and is associated with intravascular neutrophil accumulation as well as marked intracapillary red blood cell stasis.19 35 Fig 8C⇑ shows an example of capillary plugging in a tissue specimen taken from infarcted myocardium. Our data show that infarct regions compared with normal regions had a 37-fold higher number of capillaries with red cell stasis. Thus, regional perfusion measurements may reflect functional capillary density rather than microvascular flow rates. A decrease in functional capillary density would then lead to prolonged wash-in/washout time constants, for two reasons. First, there would be a decrease in capillary surface area available for solute transport, and second, once in the interstitium there would be increased diffusion distances to fill the extracellular space. Since diffusion time increases with the square of distance,36 moderate changes in functional capillary density could lead to greatly prolonged wash-in/washout time constants.
Because our data were acquired during maximal vasodilation with adenosine, regional differences in microsphere flow should indicate regional differences in the extent of microvascular damage occurring during acute infarction with reperfusion. This interpretation of our data suggests that washout time constants and microsphere flow are both markers for microvascular injury rather than Gd-DTPA kinetics being directly related to myocardial blood flow in mL·min−1·g−1. Likewise, the correlation between washout time constants and microsphere flow should not imply that washout time constants may be used to measure regional blood flow, since washout time constants are probably determined by several additional factors besides blood flow (eg, functional capillary density, rate of diffusion of contrast out of the vascular space, water exchange rates). However, the correlation between washout time constants and microsphere flow does imply that core regions of the infarct have more extensive microvascular damage than subepicardial rim regions of the infarct. Our finding that areas of gross hemorrhage were predominantly subendocardial and central within infarct regions is also consistent with this interpretation.
Comparison of MR-Altered Enhancement Zones With Infarct Size
Previous studies by us1 2 and others4 7 37 have shown a good correlation between the extent of hyperenhanced regions on MR images and the extent of infarction measured by other techniques. This study, however, demonstrates several new findings. (1) The spatial extent of MR signal abnormalities can change, depending on the signal intensity used to define injured regions. Although an intensity threshold of 2 SD above remote regions on early washout images led to overestimation of TTC-negative regions (Fig 5⇑), a higher cutoff (4 SD) led to underestimation of TTC-negative regions. One possible explanation for this observation is that partial volume effects due to thick slice widths can lead to infarct edge blurring. This will be true particularly if the infarct border migrates significantly across the imaged volume. (2) The spatial extent of MR signal abnormalities can change, depending on time. The time-intensity curves of Fig 3⇑ show that a region that is hypoenhanced or hyperenhanced compared with normal tissue at one time point may become isoenhanced at another time point and no longer be distinguishable from normal. The actual time when this occurs depends greatly on the wash-in/washout time constant for a given region, but this observation is a likely explanation for why hyperenhanced regions on late washout images (see Figs 1F and 5⇑⇑) underestimate TTC-negative regions and why any single image at one time point may not show the entire region of altered enhancement. (3) Regions defined by prolonged washout time constants in τwo images (Fig 5⇑) correlate best with true infarct size. Since τwo images are derived from the entire set of washout images, rim regions that appear to be indistinguishable from normal tissue on particular images will still be delineated by their abnormal washout kinetics. Also, since the dynamic range of tissue washout time constants is much greater than the dynamic range of tissue signal intensities, τwo images show improved spatial contrast between normal and abnormal tissue and less sensitivity to arbitrary threshold levels.
These results show that injured myocardium is well characterized by Gd-DTPA washout kinetics and that although there are difficulties in measuring MR signal abnormalities on any single image, the full region of altered contrast enhancement does in fact appear to be an accurate representation of the infarcted territory.
We conclude that altered contrast enhancement patterns seen in MR images of myocardial tissue subject to acute infarction with reperfusion are caused primarily by regional differences in wash-in/washout time constants and that the spatial extent of regions with abnormal time constants measures true infarct size. The physiological basis of these regional differences in time constants appears to be related to differences in functional capillary density, and therefore, evaluation of myocardial Gd-DTPA kinetics may provide useful information indicating the severity of microvascular damage after this type of myocardial injury.
Selected Abbreviations and Acronyms
|BOPTA||=||benzyloxypropionic tetraacetic acid|
|ROI||=||region of interest|
|τwo||=||washout time constant|
This work was supported in part by a Biomedical Engineering Research Grant from the Whitaker Foundation (Dr Judd), the Frank T. McClure Fellowship in Cardiovascular Research (Dr Judd), and NIH-NHLBI grants R29-HL-53411 (Dr Judd) and T32-HL-07227 (Dr Kim). The authors thank Frank C.P. Yin, MD, PhD, for his helpful suggestions.
- Received April 30, 1996.
- Revision received July 15, 1996.
- Accepted July 18, 1996.
- Copyright © 1996 by American Heart Association
Lima JAC, Judd RM, Bazille A, Schulman SP, Atalar E, Zerhouni EA. Regional heterogeneity of human myocardial infarcts demonstrated by contrast-enhanced MRI: potential mechanisms. Circulation. 1995;92:1117-1125.
Judd RM, Lugo-Olivieri CH, Arai M, Kondo T, Croisille P, Lima JAC, Mohan V, Becker LC, Zerhouni EA. Physiological basis of myocardial contrast enhancement in fast magnetic resonance images of 2-day-old reperfused canine infarcts. Circulation. 1995;92:1902-1910.
Wolfe CL, Moseley ME, Wikstrom MG, Sievers RE, Wendland MR, Dupon JW, Finkbeiner WE, Lipton MJ, Parmley WW, Brasch RC. Assessment of myocardial salvage after ischemia and reperfusion using magnetic resonance imaging and spectroscopy. Circulation. 1989;80:969-982.
Peshock RM, Malloy CR, Buja LM, Nunnally RL, Parkey RW, Willerson JT. Magnetic resonance imaging of acute myocardial infarction: gadolinium diethylenetriamine pentaacetic acid as a marker of reperfusion. Circulation. 1986;74:1434-1440.
Goldman MR, Brady TJ, Pykett IL, Burt CT, Buonanno FS, Kistler P, Newhouse JH, Hinshaw WS, Pohost GM. Quantification of experimental myocardial infarction using nuclear magnetic resonance imaging and paramagnetic ion contrast enhancement in excised canine hearts. Circulation. 1982;66:1012-1016.
Lima JAC, Judd RM, Lugo-Olivieri CH, Schulman SP, Atalar E, Gerstenblith G, Zerhouni EA. Myocardial perfusion pattern by contrast enhanced MRI in the acute post-infarct period can predict myocardial necrosis and eventual extent of scar formation. Circulation. 1994;90(pt 2):I-410. Abstract.
May-Newman K, Judd RM, Zerhouni EA. Myocardial T1 following step changes in perfusate contrast concentration. Proc Soc Magn Reson Imaging. 1994;1:1017. Abstract.
Resar JR, Livingston JZ, Yin FCP. In-plane myocardial wall stress is not the primary determinant of coronary systolic flow impediment. Circ Res. 1992;70:583-592.
Resar JR, Judd RM, Halperin HR, Chacko VP, Weiss RG, Yin FCP. Direct evidence that coronary perfusion affects diastolic myocardial mechanical properties. Cardiovasc Res. 1993;27:403-410.
Domenech RJ, Hoffman JIE, Noble MIM, Saunders KB, Henson JR, Subijanto S. Total and regional coronary blood flow measured by radioactive microspheres in conscious and anesthetized dogs. Circ Res. 1969;25:581-596.
Ambrosio G, Weisman HF, Mannisi JA, Becker LC. Progressive impairment of regional myocardial perfusion after initial restoration of postischemic blood flow. Circulation. 1989;80:1846-1861.
Winer BJ. Statistical Principles in Experimental Design. New York, NY: McGraw Hill; 1971:351-359.
Wendland MF, Saeed M, Lauerma K, Derugin N, Cavagna F, Higgins CB. Alterations in T1 in normal and reperfused infarcted myocardium following GdBOPTA or GdDTPA on inversion recovery EPI. Proc Soc Magn Reson Imaging. 1993;2:788. Abstract.
Fishbein MC, Y-Rit J, Lando U, Kanmatsuse K, Mercier JC, Ganz W. The relationship of vascular injury and myocardial hemorrhage to necrosis after reperfusion. Circulation. 1980;62:1274-1279.
Weinmann HJ, Laniado M, Mutzel W. Pharmacokinetics of GdDTPA-dimeglumine after intravenous injection into healthy volunteers. Physiol Chem Phys. 1984;16:167-172.
Tong CY, Prato FS, Wisenberg G, Sandler D, Wills JM. Measurement of the extraction efficiency and distribution volume for Gd-DTPA in canine reperfused myocardium. Proc Soc Magn Reson Imaging. 1992;1:1114. Abstract.
Johnston DL, Liu P, Lauffer RB, Newell JB, Wedeen VJ, Rosen BR, Brady TJ, Okada RD. Use of gadolinium-DTPA as a myocardial perfusion agent: potential applications and limitations for magnetic resonance imaging. J Nucl Med. 1987;28:871-877.
Engler RL, Dahlgren MD, Peterson MA, Dobbs A, Schmid-Schoenbein G. Accumulation of polymorphonuclear leukocytes during 3 hour experimental myocardial ischemia. Am J Physiol. 1986;251:H93-H100.
Bassingthwaighte JB, Goresky CA. Modeling in the analysis of solute and water exchange in the microvasculature. In: Renkin EM, Michel CC, Geiger SR, eds. Handbook of Physiology. Bethesda, Md: American Physiological Society; 1984:549-626.