In Vivo MRI Visualization of Acute Myocardial Ischemia and Reperfusion in Ferrets by the Persistent Action of the Contrast Agent Gd(BME-DTTA)
Background Contrast agent–enhanced magnetic resonance imaging (MRI) has the potential to visualize myocardial ischemia. To date, however, no agent has been found that has a sustained effect that allows MRI detection for the entire duration of ischemia and reperfusion and thus is useful in conjunction with stress test MRI. In this article, we introduce the gadolinium complex of N3,N6-bis(2′-myristoyloxyethyl)-1,8-dioxo-triethylene-tetraamine-N,N,N′,N′-tetraacetic acid [Gd(BME-DTTA)], an agent potentially useful for such a purpose.
Methods and Results Four protocols were carried out. ECG-triggered, partially T1-weighted, spin-echo MRI was used in protocols A through C. In protocol A, in nonischemic ferrets, 50 μmol/kg Gd(BME-DTTA) induced a 70±5% intensity enhancement lasting 3 hours. In protocol B, the left anterior descending coronary artery was occluded, and a 99mTc-sestamibi–induced autoradiographic contrast verified (r=.87, P<.01) a Gd(BME-DTTA)-induced (n=5) or Gd(DTPA)- induced (n=4) MRI contrast. In the Gd(BME-DTTA) group a sustained contrast and in the Gd(DTPA) group a short-lived contrast were observed. In protocol C (n=11), during ischemia, a 31±3.3% (P<.02) contrast was evident between the ischemic and nonischemic myocardial regions. Upon reperfusion, a contrast of 19±3% (P<.05) and 13±4.5% (P<.05) persisted for 5 and 15 minutes, respectively. Beyond 15 minutes, the contrast continued to diminish gradually. Nonradioactive microspheres verified (r=.87, P<.05) ischemia and reperfusion in this model. In protocol D (n=4), blood ΔR1 data showed that the blood pool retained Gd(BME-DTTA) for the entire time frame of the experiment at high enough concentration to provide an appropriate wash-in effect during the initial contrast enhancement and during reperfusion.
Conclusions This study demonstrates that Gd(BME-DTTA) induces a sustained MRI contrast between regions of normal versus ischemic myocardium, showing the potential of this agent for the diagnosis of ischemic heart disease in conjunction with stress tests.
Coronary artery disease causes more deaths, disability, and economic loss in industrialized nations than any other group of diseases.1 Therefore, the screening and noninvasive diagnosis of ischemic heart disease is highly important. MRI is a potentially powerful tool for this purpose2 3 because of its ability to show excellent soft-tissue contrast. This is best achieved when tissue differences in relaxation times, T1 or T2, are present. Under such conditions, myocardial MRI contrasts can be obtained. Beyond 3 hours of coronary occlusion, T1 and T2 increase in the ischemic region.4 Within the first hour of occlusion, however, the inherent relaxation rate difference between well-perfused and ischemic myocardial tissue is not sufficient for generating reasonable MRI contrast for the detection of myocardial underperfusion.5 6 Thus, the use of contrast agents is necessary for assessing acute myocardial underperfusion.7 8
Several studies had shown that myocardial ischemic areas could be detected with radiolabeled fatty acids.9 10 11 On the basis of these data, the bifunctional ligand BME-DTTA was designed and synthesized in our laboratory by the addition of a covalent attachment of a chelator moiety to myristoyl, a long fatty acyl chain.12 Chu et al13 reported that at a dose of 50 μmol/kg, liposomal Gd(BME-DTTA) did not cause significant cardiovascular side effects in ferrets, as indeed was expected from our previous LD50 measurements.12 Also, when this contrast agent was administered to ferrets, a specific MRI IE was detected in the heart muscle.13
The primary purpose of the present study was to determine whether Gd(BME-DTTA) would induce a contrast between ischemic and nonischemic myocardium in the acutely ischemic ferret heart and whether this myocardial contrast would persist for the entire duration of the ischemic period as well as during early reperfusion. Verification of the extent and location of the myocardial ischemic region as detected by agent-enhanced MRI was carried out by a corresponding non-MRI imaging method, autoradiography with 99mTc-sestamibi. Correlation of contrast with MP was also carried out by use of nonradioactive microspheres. The cardiospecificity of the agent was demonstrated by myocardial tissue kinetics quite distinct from its blood-pool kinetics.
Gd(BME-DTTA) was synthesized and incorporated into liposomes as described previously.12 Liposomal Gd(BME-DTTA) was dialyzed at 4°C for 24 hours against a saline solution (pH 7.4) that contained 0.9 wt% NaCl, 20 mmol/L HEPES, and 0.5 wt% Chelex-100 to remove any weakly bound gadolinium ions. In vitro NMR relaxation rate (1/T1) was measured on an IBM PC-20 (0.47 T) Multispec NMR instrument at 37°C. An in vitro longitudinal relaxivity of 27.1±0.3 s−1·mmol/L−1 was routinely found.
Thirty-three male ferrets weighing 0.9 to 1.2 kg (Marshall Farms) were anesthetized with sodium pentobarbital (25 mg/kg). A tracheal tube was inserted and connected to a respirator (intermediate animal ventilator, Harvard Apparatus, Inc) with a setting for 45-mL tidal volume at the rate of 25 cycles per minute. The left jugular vein was isolated, and an intravenous line was inserted to allow the administration of infusion and contrast agent. The left carotid artery was instrumented for blood pressure measurements, and an ECG was recorded. In 20 ferrets, transverse thoracotomy was performed. Coronary arteries in ferrets submerge into the myocardium immediately after originating from the aorta. Therefore, direct isolation of a coronary artery could not be carried out. A surgical suture was placed superficially through the myocardium under the most visible proximal segment of the LAD. An already inflated, manometer-controlled balloon was positioned close to the surgical suture. Following the closure of the surgical suture around the middle portion of the balloon, a livid discoloration developed at the interventricular groove and the anterior myocardial area. After the deflation of the balloon, this livid discoloration disappeared. An epicardial ECG was used to detect the presence of significant myocardial ischemia. During a 5-minute temporary occlusion, an immediate, significant ST-segment elevation developed in the epicardial ECG. The ST-segment elevation as well as the livid discoloration in the ischemic area disappeared after the release of this occlusion. This visible discoloration, associated with a >5-mV increase in ST-segment elevation, was subsequently used as an indicator of significant myocardial ischemia. The ECG leads were removed after the surgical procedure. During the MRI experiment, a Hewlett-Packard telemetric ECG system was used to trigger the MRI acquisions and to monitor surface ECG changes during the ischemic period.
A 1.5-T Philips Gyroscan MR imager with a head coil was used for ferret heart imaging. An ECG-gated, relatively T1-weighted (TR=600 ms, TE=30 ms), spin-echo pulse sequence was used in a multiple-slice, multiple-phase, dynamic study with seven dynamic intervals (10 minutes each), 256×256 image matrix, field of view of 200 mm, and slice thickness of 3 mm. During the 10 minutes of each dynamic interval, nine images with three tomographic slices in three cardiac phases (two diastolic and one systolic phase) were obtained. Three in vivo imaging protocols and one ex vivo protocol were carried out.
In protocol A, in 6 nonischemic ferrets, 50 μmol/kg Gd(BME-DTTA) was injected, and signal IE was monitored over a 3-hour interval.
In protocol B, in 9 ferrets with the acutely ischemic heart model, after the acquisition of control images the LAD was permanently ligated by a surgical snare, and either 50 μmol/kg Gd(BME-DTTA) (n=5) or 200 μmol/kg Gd(DTPA) (n=4) was injected. Three sets of MR images were collected during a 30-minute period. At 20 minutes of ischemia, 99mTc-sestamibi (99TC-Sestamibi, DuPont; 20 to 30 mCi) was injected intravenously. At 30 minutes of ischemia, the hearts were arrested and quickly removed. The right and left ventricular chambers were filled with embedding medium (Tissue-Tek O.C.T. Compound) and cooled on dry ice. The heart was then bread-sliced into 5 to 7 slices (each 3 to 4 mm thick). Care was taken to preserve close correspondence in slice thickness and position between these physical slices and the MRI tomographic slices previously acquired. The common starting point for both the physical slices and the MR image slices was the site of the ligation, which is clearly identifiable on the MR image. The physical slices were placed on a precooled, plastic surface and positioned against a Kodak MRM1 film for an exposure time of 6 hours. The developed films were scanned into a Macintosh computer, and the autoradiographic images were compared with the corresponding transverse MR images by use of NIH image 1.54 software.
In protocol C, in 11 ferrets with the acutely ischemic heart model, control images were obtained, and the prepositioned balloon was inflated to occlude the LAD. Simultaneously, 50 μmol/kg Gd(BME-DTTA) (n=9) or isotonic saline solution (n=2) was injected. In all 11 ferrets, images were subsequently acquired during 30 minutes of ischemia. Upon balloon deflation, 40 minutes of reperfusion was allowed. A 30-minute occlusion period was selected because it had been shown previously that edema did not occur until after 40 minutes of ischemia6 and the subsequent biochemical changes during 30 minutes of ischemia were reversible.8 In protocol C, in 4 ferrets of the 9 above that were injected with 50 μmol/kg Gd(BME-DTTA) and MR imaged, microspheres 15±0.1 mm in diameter and dyed with a given color (DyeTrak, Triton Technology, Inc)14 were used to measure MP at the different stages of the experiment. One million microspheres, of a different color each time, were injected during the control, ischemia, and reperfusion time periods into the left atrium through a Tygon catheter and flushed with 4 mL of saline at room temperature. Before injection, the microspheres were ultrasonicated and vortex agitated to obtain optimal dispersion. Upon termination of the MRI experiment, the LAD was reoccluded and 5 mL MB was injected into the left atrium to delineate the ischemic and nonischemic myocardial tissue areas, since MB is known to yield a sharp border at the regions in which MP falls below 50% of baseline.15 After the injection of MB, the heart was arrested with KCl and quickly removed. The heart was then bread-sliced into 5 to 7 slices (each 3 to 4 mm thick). Ischemic areas were recognized by postmortem analysis of the MB dye staining; blue tissue as nonischemic regions (2.46±0.13 g) and nonstained tissue as ischemic regions (0.43±0.1 g) were differentiated. For further analysis, a procedure described by Kowalik at al14 was followed. The latter procedure included tissue digestion by 4 mol/L KOH, 2% Tween 80 mixture at 72°C (4 hours in a water bath shaker), filtering (microspheres remain on the surface of the 8-mm pore size, polyester filter disks), and eluting of the three different dyes by dimethyl formamide. The latter suspension was centrifuged (5 minutes, 2000g), and the supernatant, which contained the dyes, was retained. The samples thus obtained were then analyzed on a Beckman DU-65 spectrophotometer. The three peaks in the spectrum (320 to 820 nm), each corresponding to one of the three dyes, were digitized and the peak areas integrated by a computer-interfaced digitizing board. The peak areas were normalized to the corresponding individual peaks in reference spectra. From these normalized areas, the microsphere content per gram tissue for each of the three dyes was calculated for each tissue sample. The MP values were normalized to the preischemic control.
To demonstrate specific agent accumulation in the myocardium by the end of 30 minutes of LAD occlusion, myocardial gadolinium content and tissue T1 were measured in a total of 3 ferrets given a 50 μmol/kg (n=2) or a 100 μmol/kg (n=1) dose of Gd(BME-DTTA) upon occlusion. Nonischemic myocardial tissue samples (≈1.0 g) were cut out on the basis of the distribution of MB. First, R1 relaxation rates (1/T1) were determined at 1.5 T, and subsequently the same samples were sent to Galbraith Laboratories for direct measurement of tissue concentration of gadolinium by plasma emission spectroscopy.
In protocol D, in 4 ferrets, whole-blood T1 was determined ex vivo. Arterial blood samples, 0.5 mL each, were taken from the carotid artery. After the drawing of a control blood sample, 50 μmol/kg Gd(BME-DTTA) was injected intravenously, and samples were drawn at 2, 4, 6, 8, 10, 15, 20, 25, 30, 40, 50, 60, 90, 120, 150, and 180 minutes after injection. The T1 value of each blood sample was determined at 1.5 T at room temperature.
A phantom consisting of a plastic beaker filled with agarose gel was used as an external intensity reference. In transverse myocardial slices (short-axis, dual-chamber view), septal, anterior, lateral, and posterior segments were selected as ROIs (Fig 1⇓). All the MR and autoradiographic images presented in this article are oriented in the manner indicated in Fig 1⇓. Average intensity in each ROI was measured and normalized to the intensity of the external reference. In protocol A, the IE in each ROI was expressed by
where Ipre=Iorgan/Ireference (before injection of agent), Ipost=Iorgan/Ireference (after injection of agent), and I denotes signal intensity.
After LAD occlusion, the lateral and posterior myocardial regions are not expected to be ischemic. In protocol B, the locations of the nonischemic myocardial area and of the ischemic myocardial region (ROI inside the anteroseptal myocardial region) were verified by 99mTc-sestamibi autoradiography. Thus, on the basis of this verification, in protocols B and C the average MR image signal intensities of these lateral and posterior ROIs were used as nonischemic controls, and the average MR image signal intensity of the ROI inside the anteroseptal myocardial region was followed in consecutive MR images as ischemic intensity values. The size of a given ROI was kept constant in any specific experiment in a specific myocardial region, in a specific slice and cardiac phase. The software used, NIH image, allowed maintaining the same size (same outline) and position of any given ROI within a sequential image set acquired over time. In the analysis of one set of our MR images (example: slice 1, phase 1, dynamic 1 to 7), an image stack was created in which all the MR images were zoomed exactly to the same extent. Next, multiple ROIs were drawn, the ROIs were kept in the same size and position inside all the MR images in the stack, and the average intensity in each ROI in each MR image was measured. Thus, the size and location for both nonischemic and ischemic ROIs were kept unchanged during the experiment, and myocardial contrasts were calculated and expressed by
To compare the defect sizes between 99mTc-sestamibi– and Gd(BME-DTTA)–induced myocardial contrasts, we defined defect boundaries as follows: (1) The boundary of the ischemic region in the 99mTc-sestamibi image was drawn manually around an area with a signal intensity threshold of 1%. In this case, however, the endocardial and epicardial boundaries of the ischemic region are imaginary because of the nature of the method. (2) In the MR image, the boundary of the ischemic spot was drawn on the basis of a signal intensity reduction of at least 20% (real boundaries on each side of the ischemic region).
The data represented as mean±SEM. Statistical analysis was performed with Number Cruncher Statistical Systems software. General linear-models ANOVA and repeated-measures ANOVA were performed between multiple serial measurements and control measurement. The Duncan test was performed when control measurements were compared with the multiple serial measurements (protocols A through D). Duncan’s multiple-range test was performed to compare multiple serial measurements (protocols B and C). In protocol B, the perfusion defect size was determined either by 99mTc-sestamibi–or by Gd(BME-DTTA)–induced myocardial contrast.The corresponding perfusion defect sizes determined by these two techniques were correlated by linear regression analysis. In protocol D, the relaxation rate in blood was analyzed as a function of time by curve fitting. Data with P<.05 were considered significant.
In nonischemic ferrets (protocol A; n=6), the intensity in heart muscle increased by 65±5% (P<.001) within 15 minutes after administration of Gd(BME-DTTA). In subsequent images obtained during the first hour, a 70±5% enhancement in MRI signal intensity in the myocardium was observed. The average IE remained at a plateau level for 3 hours, indicating a conveniently long lifetime of this contrast agent in the heart tissue (Fig 2⇓). In contradistinction, during the same experimental time window, blood R1 enhancement, induced by the contrast agent present in the blood compartment, decreased monotonically (Fig 2⇓).
In LAD-ligated ferrets (protocol B), autoradiography with 99mTc-sestamibi was used to verify the extent and location of the ischemic region highlighted by Gd(BME-DTTA). 99mTc-sestamibi activity accumulates into the nonischemic myocardial tissue but not into the ischemic myocardial tissue. In Fig 3⇓, the 99mTc-sestamibi– and the Gd(BME-DTTA)–induced myocardial contrasts are shown in corresponding slices. The anterior ischemic region delineated by the Gd(BME-DTTA)–enhanced MR image is detected as early as 5 minutes after injection (Fig 3⇓, II), and it corresponds in location and extent to the 99mTc-sestamibi defect (Fig 3⇓, I). This delineation remains in effect even after 25 minutes (Fig 3⇓, III). As a comparison, in Fig 4⇓, the 99mTc-sestamibi– and the Gd(DTPA)-induced myocardial contrasts are shown in the autoradiographic (Fig 4⇓, I) and in the MR (Fig 4⇓, II and III) images. After administration of Gd(DTPA), the anteroseptal ischemic region became delineated at 5 minutes and corresponded in location and extent to the 99mTc-sestamibi defect. Contrary to the Gd(BME-DTTA)–enhanced image, however, this contrast quickly disappeared, as demonstrated in the late MR image (Fig 4⇓, MR image III).
A quantitative depiction of the time dependence of the above myocardial IEs and contrasts are shown in Fig 5⇓. After LAD occlusion and administration of Gd(BME-DTTA), a sustained myocardial IE developed in the nonischemic region, and a nonsignificant IE took place in the ischemic spot (Fig 5A⇓). The time dependence of the corresponding contrast, obtained in accordance with Equation 2, is shown in Fig 5C⇓. This contrast reached a plateau at 5 minutes, and at the end of 25 minutes of occlusion it was still at its plateau level. With Gd(DTPA), however, the contrast was short-lived (Fig 5C⇓); the MR IE in the nonischemic myocardial regions decreased rapidly and the IE in the ischemic myocardial region increased significantly beyond 5 minutes after the administration of this agent (Fig 5B⇓). The ischemic defects as detected by both Gd(BME-DTTA) and 99mTc-sestamibi were localized in the anteroseptal myocardial wall, and the same location was consistently identified by both the radionuclide perfusion tracer and the MRI agent. A positive correlation (r=.87, P<.01) was found between the defect sizes determined by contrast agent–enhanced MRI versus autoradiography (Fig 6⇓).
The effect of Gd(BME-DTTA) in a ferret model (protocol C) in which reperfusion followed the LAD occlusion is shown in typical transverse MR image tomographic slices (Fig 7⇓). After the administration of Gd(BME-DTTA) and LAD occlusion, a significant increase in signal intensity is observed in the nonischemic heart muscle, ie, in the lateral and posterior regions. In the ischemic, septal region, the intensity does not increase during ischemia (Fig 7B⇓ and 7C⇓). Thus, during focal ischemia induced by LAD occlusion, a contrast develops between ischemic (anteroseptal) and nonischemic (lateral and posterior) myocardial tissue areas within a relatively short time, and this contrast persists while ischemia is present. In this particular experiment, the myocardial signal intensity is higher in the endocardial surface compared with the epicardial region of the ischemic spot. This observation, however, is not present consistently in all of our animals. It is likely that because of ischemia-induced dyskinesia, the blood movement in the left ventricular cavity adjacent to the endocardial surface in question is slowed. In slice-selective spin-echo MRI, this would increase the signal intensity in the affected image region. Upon reperfusion (Fig 7D⇓), the image intensity in the anteroseptal (ischemic) region begins to increase, and the contrast between previously ischemic and nonischemic tissue areas disappears with time.
Fig 8⇓ depicts the detailed time dependence of the IEs and contrast for the entire group (n=9) in the occlusion-plus-reperfusion model. The MRI IE in the nonischemic, ie, posterior and lateral, regions of the myocardial tissue increased with time (Fig 8A⇓). The IE in the ischemic, anteroseptal region gradually increased during both the occlusion and the reperfusion periods (Fig 8A⇓), although this increase became steep enough to achieve statistical significance only during reperfusion. A myocardial contrast between the septal and the lateral and posterior myocardial slices was not present during preocclusion control (Fig 8B⇓). During ischemia, a significant 31±3.3% (P<.05) contrast was evident between the ischemic (anteroseptal) and nonischemic myocardial tissue areas (Fig 8B⇓). Upon reperfusion, the Gd(BME-DTTA)–induced contrast gradually decreased but did not immediately disappear. Rather, a contrast of 19±3% (P<.05) and 13±4.5% (P<.05) persisted for 5 and 15 minutes, respectively. Beyond the first 15 minutes of reperfusion, the signal intensity increased further in the ischemic myocardium (Fig 8A⇓), and thus, the contrast between the ischemic and normal myocardial regions continued to diminish gradually (Fig 8B⇓).
In another subgroup of ferrets subjected to 30 minutes of occlusion followed by reperfusion (protocol C, n=4), measurement of MP, along with acquisition of MR images, was carried out during the three different stages of the experiment (control, ischemia, and reperfusion). MP and Gd(BME-DTTA)–induced IE data during ischemia show a linear correlation with a correlation coefficient of .82 (P<.01), demonstrating a close correspondence between MP and Gd(BME-DTTA)–induced myocardial MRI signal intensity (Fig 9⇓). A 66±5% reduction in MP in the ischemic region was evident during ischemia compared with the MP of the same tissue during the control period (Fig 10B⇓). A similar reduction in MP (63±6%) was observed when the MP in the ischemic region was compared with the MP in the nonischemic region. In the corresponding Gd(BME-DTTA)–enhanced MRI data, a 47±3% lower IE occurred in the ischemic spot compared with the IE in the nonischemic myocardial region during ischemia (Fig 10A⇓). During reperfusion, the average MP rises both in the nonischemic and previously ischemic areas (129±21% versus 187±34% of control, respectively). In the latter, however, the increase in MP is significantly larger (P<.05), possibly due to the effect of reactive hyperemia. The increase in MP in the nonischemic region from 95±5% to 129±21% (see Table 1⇓) is reflected in only a small, nonsignificant increase in MRI IE (from 59±6% to 63±6%). The increase in MP in the ischemic region (from 34±5% to 187±34%), however, is reflected in a 40±5% increase in MRI IE (from 12±6% to 52±6%) (Fig 10A⇓). In this latter case, the increase in both MP and MRI IE is highly significant (P<.01).
Myocardial R1 and gadolinium content, determined ex vivo (n=3) at the end of 30 minutes of occlusion, indicated an enhanced R1 of 1.67±0.09 and 2.89 s−1 at the low and high agent doses, respectively, and a 1.35- or 1.9-fold myocardial accumulation of Gd in the nonischemic region after administration of 50 or 100 μmol/kg doses of Gd(BME-DTTA), respectively, given at the onset of occlusion (Table 2⇓).
To follow agent kinetics in the blood pool, T1 values of ferret arterial blood samples were determined at a 1.5-T magnetic field before and after administration of Gd(BME-DTTA) (protocol D, n=4). A blood T1 value of 1.32±0.04 seconds was found during control. After the administration of Gd(BME-DTTA), a fast initial bolus effect in blood is detected. This effect reaches a minimum T1 value of 0.14±0.02 seconds within less than 2 minutes, followed by a slow increase in T1, and reaching a blood T1 of 0.25±0.05 seconds at the end of 1 hour after administration of the contrast agent. On the basis of the T1 values, the R1 rates were calculated, the Gd(BME-DTTA)–induced ΔR1, which was linearly proportional to the blood concentration of the contrast agent, was plotted versus time, and the data were curve fitted (Fig 11⇓). The results indicate a three-phase kinetic behavior. First, a fast bolus effect brings ΔR1 to an initial maximum within the first 2 minutes. Next, an intermediate rate of decay is evident with a time constant of 6.8 minutes. This decay brings ΔR1 down to about 82% of the maximum effect within a 15-minute time period (Fig 11A⇓). The third phase is a slow decay whose time constant is 197 minutes (Fig 11B⇓). At the 30-minute time point, 74% of the maximum effect is still observed.
The present study demonstrates that the MRI contrast agent Gd(BME-DTTA) displays a sufficiently long half-life in the bloodstream and a specific accumulation in the myocardial tissue inducing a sustained myocardial signal IE and also retains a reasonably efficient in vivo relaxivity at relatively low dosage. We have found that Gd(BME-DTTA) induces a well-defined contrast between ischemic and nonischemic myocardial tissue areas in the acutely ischemic ferret heart. We provide evidence that the myocardial contrasts induced by both 99mTc-sestamibi and Gd(BME-DTTA) show similar sizes and locations of the myocardial ischemic regions in corresponding slices. The agent-induced myocardial MRI contrast exists for the entire duration of the ischemic and early reperfusion time periods. Thus, the data indicate the potential usefulness of Gd(BME-DTTA) for the early diagnosis of ischemic heart disease.
Our experimental design deliberately consisted of 30 minutes of coronary occlusion, since the absence of myocardial edema6 and the presence of the reversibility of the biochemical changes due to the ischemic myocardial event8 defined a model of acute myocardial ischemia. The possibility that we observed an infarct rather than an ischemic spot was precluded by (1) short duration of occlusion and (2) the fact that without the agent, no contrast could be detected in the same model, since in two ferrets no myocardial MR image contrast was observed after LAD ligation and administration of isotonic saline. The presence of myocardial ischemia due to the decrease in MP was tested by epicardial ECG before MRI and also monitored during the MRI experiment. During MRI, both diastolic and systolic MR images were collected at each experimental phase, and the impact of the decrease in regional MP on regional ventricular function could be tested. The observed dyskinesia, which developed during occlusion and partially improved during reperfusion, indicated that a myocardial ischemia was associated with the decrease in MP. Thus, we use both terms, ie, underperfusion and ischemia, in this work.
The size and location of the myocardial ischemic region are verified by a non-MRI method, autoradiography with 99mTc-sestamibi. 99mTc-sestamibi is a single photon emitting radionuclide tracer that is taken up by viable myocardium in proportion to the distribution of MP.16 17 The mode of tissue uptake of Gd(BME-DTTA) may be similar to that of 99mTc-sestamibi,18 since both agents are lipophilic compounds. Therefore, autoradiography with 99mTc-sestamibi is used as a comparative standard to verify the ability of our MRI agent to detect the location and size of underperfused myocardium by comparing the defects detected by the two techniques. In 5 ferrets, the ischemic defects as detected by both Gd(BME-DTTA) and 99mTc-sestamibi were localized in the anteroseptal myocardial wall, and the same location was consistently identified by both the radionuclide perfusion tracer and the MRI agent. A significant correlation (r=.87, P<.01) was found between the defect sizes determined by contrast agent–enhanced MRI versus autoradiography (Figs 3⇑ and 6⇑).
In animal models, microspheres are well known to be useful in the measurement of MP.15 Therefore, we monitored the MP changes during ischemia and reperfusion using color microspheres.14 A positive correlation (r=.82, P<.05) between MP, measured by microspheres, and Gd(BME-DTTA)–induced MRI signal IE indicated a correspondence between these two parameters during the ischemic time period (Fig 11⇑). Furthermore, the significant increase in MP in the ischemic region during the reperfusion time period was reflected in a significant increase in IE in MR images (Fig 10A⇑). Thus, the microsphere data provided direct verification for the occurrence of both ischemia and reperfusion. The reasons for the different values of the ΔIE/ΔMP ratio in the nonischemic versus the ischemic regions during the reperfusion period are explained in the “Appendix.”
The potential of MRI to show a contrast between normal and infarcted canine myocardium arising from the prolongation of the T1 and T2 relaxation times in the infarcted region was demonstrated early.19 20 21 22 23 This prolongation was correlated with increased water content in the infarct compared with normal myocardium.4 20 24 Significant myocardial edema, however, does not occur until at least 30 minutes after coronary occlusion,6 and thus, acute ischemia would not be visualized by MRI in the absence of contrast agent. In excised hearts, no difference in signal intensity between ischemic and normal myocardium was observed in the absence of agent in spin-echo images after 1 minute of LAD ligation in dogs.25 The lack of MRI contrast in early ischemia is also demonstrated by the in vivo sham experiment in ferrets in the present work.
Because of its low toxicity, Gd(DTPA) has been extensively studied and used as a contrast agent for experimental26 27 and clinical MRI28 29 of the heart and other organs. It has been found, however, that Gd(DTPA) has a short half-life (20 minutes) in blood and is quickly excreted through the kidney.30 Therefore, in this work, we decided to compare the effect of this agent with that of Gd(BME-DTTA) in the same animal model under a similar set of experimental conditions. Gd(DTPA) enhances the visualization of acute myocardial infarction in relatively T1-weighted MR images.31 Contrary to its effect in an infarct several days old, Gd(DTPA) is not able to induce a sustained myocardial contrast between acutely ischemic and nonischemic myocardial tissue areas using T1-weighted spin-echo MRI,2 19 as is also demonstrated in our study. Our study has demonstrated, however, that Gd(DTPA) does induce a short-lived myocardial contrast between acutely ischemic and nonischemic myocardial tissue areas (Figs 4⇑ and 5⇑), which would allow detection only during the first several minutes after agent administration. Indeed, similar effects were demonstrated by ultrafast MRI techniques.2 32
The lack of a sustained effect of Gd(DTPA) in spin-echo MRI in acute ischemia in the absence of a well-developed infarct is probably due to differences in wash-in, washout effects in infarcted versus in acutely ischemic myocardial tissue. In our experiments during acute myocardial ischemia, the short persistence of Gd(DTPA) in the bloodstream and the lack of a sustained differential signal enhancement in the normally perfused versus the underperfused myocardial tissue areas resulted in a short-lived myocardial contrast. To allow a sustained monitoring of an entire stress-induced duration of ischemia and reperfusion, a new myocardial contrast agent (1) should have a sufficiently long half-life in the blood stream to maintain the appropriate wash-in, washout effects; (2) must show sufficient organ specificity; and (3) must retain reasonably efficient in vivo relaxivity at relatively low dosage. Such an agent is likely to induce contrast between ischemic and nonischemic myocardial tissue areas for a period of time sufficient for MRI in conjunction with exercise testing. Gd(BME-DTTA) fulfills the above requirements, since it (1) displays a sustained effect; (2) shows characteristic, specific myocardial accumulation (see Figs 2⇑ and 9⇑ and Table 2⇑); and (3) retains efficient in vivo relaxivity at a low dose (Table 2⇑).
Until recently, cardiac MRI could be performed only at rest or with pharmacological stress. Treadmill- or bicycle-stress cardiac MRI to detect ischemia has not been performed. Immediate postexercise cardiac MRI, however, is possible in both cases. In the immediate postexercise period, however, no contrast will be seen with previous or current contrast agents8 33 34 35 36 37 because of their inability to induce a persistent myocardial contrast during both acute ischemia and early reperfusion. Gd(BME-DTTA) induced a sustained myocardial contrast during ischemia (Figs 5⇑, 7⇑, and 8⇑). Furthermore, upon reperfusion, a contrast of 19±3% (P<.05) and 13±4.5% (P<.05) persisted for 5 and 15 minutes, respectively. These were different from the contrast during ischemia (31±3.3%) with a confidence of P<.05. The contrasts at 5 and 15 minutes of reperfusion, however, were not statistically different from each other but were different from the contrast during control and from the contrast at 35 minutes of reperfusion (P<.05). Thus, the myocardial contrast values at 5 and 15 minutes of reperfusion still highlight the ischemic spot in a significant manner (Fig 8⇑). It is our conclusion that after completion of symptom-limited treadmill- or bicycle-stress testing, MRI contrast between ischemic and normal myocardium is likely to be detected when Gd(BME-DTTA) is injected at the peak of exercise.
The present work has demonstrated that liposomal Gd(BME-DTTA) yields a well-defined contrast between regions of normal and ischemic myocardium by differentiating the ischemic region at the relatively low dosage of 50 μmol/kg. The mechanism of this effect is most likely the creation of a difference in T1 in the normal versus the ischemic region due to the presence of unequal concentrations of contrast agent, which is distributed in proportion to blood flow and specific myocardial uptake. The contrast between the ischemic and nonischemic regions increases with time after agent administration, probably resulting from a wash-in effect of the agent and the time-dependent accumulation of contrast agent in the nonischemic region. After the onset of reperfusion, the contrast between the ischemic and nonischemic regions decreases gradually and disappears, most likely by the reestablishment of normal wash-in to the previously ischemic regions.
Our results demonstrate that liposomal Gd(BME-DTTA), at a relatively low dose, clearly differentiates the tissue areas associated with the ischemic insult. In this study, 99mTc-sestamibi autoradiography used as a comparative standard has verified the ability of our MRI agent to detect the location and size of underperfused myocardium. The Gd(BME-DTTA)–induced myocardial contrast persists for the entire duration of the coronary occlusion and the early postischemic period, thus showing potential usefulness for the diagnosis of ischemic heart disease either by pharmacological stress combined with cardiac MRI (MRI before, during, and after stress) or by physical stress test (MRI before stress and during the postexercise period).
Selected Abbreviations and Acronyms
|LAD||=||left anterior descending coronary artery|
|MRI||=||magnetic resonance imaging|
|R1||=||relaxation rate (1/T1)|
|ROI||=||region of interest|
In spin-echo MRI, the signal intensity (I) is determined by
where Mo is a constant proportional to the spin density, TE is the echo time, TR is the repetition time, R1 is the longitudinal relaxation rate, and R2 is the transverse relaxation rate.
In general, the administration of an MRI contrast agent changes both R1 and R2. Because of the large initial value of R2 compared with R1, ΔR2/R2<<ΔR1/R1, and therefore, the paramagnetic change in R2 can be neglected. Thus, taking the derivative of I with respect to R1 in Equation 3 yields
where ΔI is, in fact, the IE, and ΔR1 is the paramagnetic change in the longitudinal relaxation rate. This ΔR1 is linearly proportional to the contrast agent concentration ([CA]) in the tissue, and since it is reasonable to assume that [CA] in any tissue region is proportional to the MP into that tissue, Equation 3 becomes
Then the change in MRI IE that results from a change in perfusion is given by
It is clear from Equation 8 that for a tissue region with a shorter T1, ie, larger R1, the ΔIE/ΔMP ratio becomes smaller. This indeed is the case in the nonischemic region compared with the previously ischemic region. In the former versus the latter, T1 is already shorter, because of a higher [CA], at the onset of reperfusion. This fact accounts for the smaller ΔIE/ΔMP in the nonischemic versus ischemic regions (see Table 1⇑). Indeed, the more than twofold ratio in this index as observed between these two tissue regions can be simulated by calculation using Equation 8 with T1 values of 0.5 and 1.1 seconds for the nonischemic and ischemic regions, respectively. Note that such T1 values are similar, within experimental error, to those measured ex vivo, ie, 0.6 and 1.0 second, respectively.
We thank Dr Tamas Jilling for his help with the process of autoradiographic image digitization and for his helpful suggestions in image processing.
Presented in part at the 11th Annual Meeting of the Society of Magnetic Resonance in Medicine, Berlin, Germany, August 8-14, 1992, and at the second Annual Meeting of the Society of Magnetic Resonance, San Francisco, Calif, August 8-12, 1994.
- Received August 17, 1993.
- Revision received July 17, 1995.
- Accepted August 8, 1995.
- Copyright © 1995 by American Heart Association
Gotto AM, Farmer JA. Risk factors for coronary artery disease. In: Braunwald E, ed. Heart Disease: A Textbook of Cardiovascular Medicine. Philadelphia, Pa: WB Saunders; 1988:1153.
de Roos A, van der Wall EE, Bruschke AV, van Voorthuisen AE. Magnetic resonance imaging in the diagnosis and evaluation of myocardial infarction. Magn Reson Q. 1992;7:191-207.
Johnston JL, 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.
Ratner AV, Okada RD, Newell JB, Pohost GM. The relationship between proton nuclear magnetic resonance relaxation parameters and myocardial perfusion with acute coronary arterial occlusion and reperfusion. Circulation. 1985;71:823-828.
Williams ES, Kaplan JI, Thatcher F, Zimmerman G, Knoebel SB. Prolongation of proton spin lattice relaxation times in regionally ischemic tissue from dog heart. J Nucl Med. 1980;21:449-453.
Wolfe CL, Moseley ME, Wikstrom MG, Sievers RE, Wendland MF, Dupon JW, Finkbeiner WE, Lipton LJ, Parmley WW, Brasch RC. Assessment of myocardial salvage after ischemia and reperfusion using magnetic resonance imaging and spectroscopy. Circulation. 1989;80:969-982.
Evans JR, Gunton RW, Baker RG. Use of radioiodinated fatty acid for photoscans of the heart. Circ Res. 1965;16:1-10.
Chu W-J, Simor T, Pohost GM, Elgavish GA. Liposomal Gd(BME-DTTA) as a NMR myocardial contrast agent: studies of its biodistribution and its effect on heart function. Presented at the Society of Magnetic Resonance in Medicine 11th Annual Meeting, Berlin, Germany. 1992. Abstract.
Kowalik P, Shultz R, Guth B, Schade A, Paffhausen W, Gross R, Heush G. Measurement of regional myocardial blood flow with multiple colored microspheres. Circ Res. 1991;83:974-982.
Gibbons RJ, Verani MS, Behrenbeck T, Pellikka PA, O’Connor MK, Mahmarian JJ, Chesebro JH, Wackers FJ. Feasibility of tomographic 99mTc-hexakis-2-methoxy-2-methylpropyl-isonitrile imaging for the assessment of area at risk and the effect of treatment in acute myocardial infarction. Circulation. 1989;80:1277-1286.
Sinusas AJ, Shi Q, Vitols PJ, Fetterman RC, Maniawski P, Zaret BL, Wackers FJ. Impact of regional ventricular function, geometry, and dobutamine stress on quantitative 99mTc-sestamibi defect size. Circulation. 1993;88:2224-2234.
Dilsizian V, Bonow RO. Current diagnostic techniques of assessing myocardial viability in patients with hibernating and stunned myocardium. Circulation. 1993;87:1-16.
Wesbey GE, Higgins CB, Lanzer P, Botvinick E, Lipton MJ, Sievers R. In vivo imaging and characterization of acute myocardial infarction using gated nuclear magnetic resonance. Circulation. 1984;69:125-130.
McNamara MT, Higgins CB, Schechtmann N. Detection and characterization of acute myocardial infarction in man with use of gated magnetic resonance. Circulation. 1985;71:717-724.
Van Rossum AC, Visser FC, Van Eenige MJ, Sprenger M, Valk J, Verheugt FW, Roos JP. Value of gadolinium-diethylene-triamine pentaacetic acid dynamics in magnetic resonance imaging of acute myocardial infarction with occluded and reperfused coronary arteries after thrombolysis. Am J Cardiol. 1990;65:845-851.