Assessment of myocardial infarct size by means of T2-weighted 1H nuclear magnetic resonance imaging

Volume 117 Number

24.

25.

26. 27.

28.

Heparinlbatroxobin-enhanced

2

flow variations and causes thromholysis following acute coronary thrombosis. AM HEART J 1987;113:898-996. Cercek B, Lew AS, Satoh Y, Isojima K, Laramee P, Yano J, Reddv KNN. Maddahi J. Ganz W. Henarin enhances exnerimental thrombolysis by preventing new fibrin deposition [Abstract]. Circulation 1985;72(Suppl 3):194. Gertz SD, Uretaky G, Wajnberg RS, Navot N, Gotsman MS. Endothelial cell damage and thrombus formation after partial constriction: relevant to the role of coronary artery spasms in the pathogenesis of myocardial infarction. Circulation 1981;63:476-86. Baumgartner HR. The role of blood flow in platelet adhesion, fibrin deposition, and formation of mural thrombi. Microvasc Res 1973;50:167-79. Horie T, Sekiguchi M, Hirosawa K. Coronary thrombosis in pathogenesis of acute myocardial infarction: histopathological study of coronary arteries in 108 necronsied cases using serial section. Br Heart J 1978;40:153-61. Dormandy JA, Reid HC. Controlled defibrination in the treatment of peripheral vascular disease. Angiology 1978; 29:80-6.

Assessment of myocardial of T,-weighted lH nuclear imaging

TPA thrombo~ys~s

29. Olsson P, Blonback M, Egberg N, Ekestrom S. Experience of extensive vascular surgery of defibrase-defibrigenated patients, Thromb Res 1976;9:277-87. 30. Belch JJC, Meed DR, Lowe G, Meed DR, Campbell AF, Young AB, Forbes CD, Prentice CRM. Subcutaneous ancrod in prevention of deep vein thrombosis after hip surgery. Thromb Res 1982;25:23-31. 31. Leon MB, Cannon RO, Watson RM, Rosing DR, Epstein SE. Long-term clot-specific thrombolytic therapy in patients with refractory chronic stable angina: angiographic coronary blood flow and metabolic changes [Abstract]. Circulation 1984;7O(Suppl‘2):323. 32. Prentic CRM, Edgar W, McNicol GP. Characterization of fibrin degradation products in patients on scrod therapy: comparison with fibrinogen derivatives produced by plasmin. Br J Haematol 1974;27:77-8. 33. Marder JV, Sherry S. Thrombolytic therapy: current status. N Engl J Med 1988;318:1585-95.

infarct size by means magnetic resonance

Proton (‘H) nuclear magnetic resonance (NM) imaging is thought to depict zones of recent myocardial infarction in contrast to noninfarcted myocardium. This is related to T2 increases in infarct zones that have been verified previously by relaxometry measurements In excised myocardlel samples. Accordingly the present study was undertaken to evaluate a ‘H NMR imaging method to optimize Tz contrast and measure infarct size in a high-field imaging system (1.5 T). To accomplish this, NMR images were acquired every other R wave with echo times of 30 and 100 msec. The flrst echo image was used for myocardlal border definition and the second echo image, which hlghllghted the myocardial infarction, for infarct border definition. This T,-weighted approach yielded a significant correlation between infarct size by NMR and pathologic methods. However, NMR imaging tended to overestimate infarct size, and the NMR image depicted abnormal signal well beyond the extent of the pathology&determined infarct. There was a significant relationship between NMR-imaged infarct size and myocardial mass with microsphere-determined reduction in blood flow of 25% or more. These data suggest that T,-weighted NMR imaging depicts not only infarct but also some reversibly Injured myocardium. (AM HEART J 1989;117:281.)

Alain Bouchard, MD, Russell C. Reeves, MD, Gregory Cranney, MD, Sanford P. Bishop, DVM, PhD, and Gerald M. Pohost, MD, with the technical assistance of Patricia Bischoff, RT, CNMT. Birmingham, Ala.

From the Division of Cardiovascular Disease, Department of Medicine, University of Alabama at Birmingham. Received for publication April 27, 1988; accepted Sept. 6, 1988. Reprint requests: Alain Bouchard, MD, University of Alabama at Birmingham, Department of Medicine, 321 Tinsley Harrison Tower,

Division Birmingham,

of Cardiovascular AL 35294.

Disease,

Room

There is considerable interest in determining the size of myocardial infarction because of its prognostic value and its potential use for guiding therapy.1-3 A method such as nuclear magnetic resonance (NMR) imaging is well suited for this purpose 281

282

Bouchard

et al.

American

Fabrusry ISBS Heart Journal

Fig. 1A. Representative late echo image (TE, = 100 msec) of ex vivo study (left) and corresponding anatomic slice (right) showinganterior myocardial infarction. Area of increasedsignal intensity on NMR imagesexceedsarea of necrosisasdelineated on anatomic specimen.RV, right ventricle; LV, left ventricle; ANT, anterior; INF, inferior.

because of its noninvasive properties. The proton relaxation parameters T, and T, are related to the biophysical and biochemical properties of the tissue.4*5 Variation within the relaxation parameters can be used to characterize myocardial tissue. The changes in relaxation times, which have been attributed mostly to alterations in myocardial water,4-s will cause an increase in signal intensity of a spinecho NMR image. 9 In this study we evaluated a method to optimize the contrast between normal and infarcted myocardium by using a T,-weighted approach. We used the technique to quantitate infarct size 1 week after myocardial infarction in a canine model and compared the results to infarct mass determined by quantitative morphology and by area of reduced blood flow. METHODS Experimental preparation. We produced myocardial infarction in the dog by meansof a closed-chesttechnique and measured infarct size 1 week later. Seven adult mongrel dogswere anesthetized with intravenous sodium pentobarbital (20 mg/kg), and the lungs were ventilated with 50% nitrous oxide, 1% halothane, and oxygen by meansof a Harvard respirator. Pellets used for creating the infarcts were constructed by using epoxy to attach a 7 mm segmentof Delrin plastic onto a 3 to 4 mm length of 7F or SF radiopaque vesseldilator. This produced a 2.2 to 2.5 mm diameter pellet with a 3 mm length of Delrin tail that fit snugly to the end of a Sonescatheter. The Sones catheter, with an appropriately sized pellet on its tip, was advanced into the ascendingaorta from the right carotid artery and then into either the left anterior descending coronary artery or the left circumflex coronary artery. Just before occlusion all animals received lidocaine, 2 mg/kg, and potassiumchloride, 20 mEq/L, intravenously in 0.9% saline solution. The pellet was dislodged from the tip of the catheter to occlude the coronary artery by a 0.035inch diameter guide wire. The catheter was then removed and

the cervical incision repaired. The animal wasthen placed in the recovery room until fully recovered from the anesthesia.Prophylactic antibiotics were given after the surgical procedure. One week after pellet embolization (7 days, n = 6; 5 days, n = l), animalswere anesthetized for NMR studies. The dogs were fasted overnight and sedated with 20 mg droperidol and 0.4 mg fentanyl (1 ml Innovar-vet). Anesthesia was induced and maintained with halothane, nitrous oxide, and oxygen by meansof positive-pressure respiration created with a Harvard respirator. Anesthesia initially wasinduced with 26% oxygen, 70% nitrous oxide, and 4% halothane by means of a volume ventilator

(Harvard apparatus) and a recirculation

anesthetic sys-

tem with 2 L/min gasflow into the system. The halothane concentration in the gasmixture inlet to the system was decreasedto 1% in 2 to 5 minutes. Throughout the NMR imaging study the animals’ lungs were ventilated by means of a Harvard respirator with 29% oxygen, 70% nitrous oxide, and 1% or lesshalothane depending on the eyelid reflex. After completion of the imagingprotocol, the animals were taken to the surgery suite where left-sided thoracotomy and pericardiotomy were performed. A 3F vinyl catheter was inserted into the left atrial appendage for injection of radiolabeled microspheres (15 Mm, 3M Corp., St. Paul, Minn., and New England Nuclear, North Billerica, Mass.). A 5F woven Dacron catheter was placed in the right femoral artery, to obtain reference blood samples,via a Bucher peristaltic pump, for 3 minutes during injection of microspheres. Radioactive microspheres(Ce-141, SC-46,or Sr-85) were injected into the left atrium for measurement of myocardial blood flow. The dogswere killed by intravenous injection of potassium chloride, and hearts were excisedand rinsed with cold saline solution to remove the blood. The left and right ventricles were stuffed with rubber gloves to maintain their shape, and each heart was placed inside a rubber glove for ex vivo NMR imaging. Wft imaging. NMR imager.

Images were obtained with a Philips

Volume Number

117 2

Proton NMR imaging

of

AMI

283

Fig. 1 B. Gated spin-echo images showing anterior wall infarction by means of in vivo lH NMR imaging. First echo image (TE,) was used to delineate myocardial borders and second echo image (TE,) for definition of myocardial infarct. Epicardial sparing is not seen on these images.

Gyroscan (Philips Medical Systems, Inc., Shelton, Conn.) NMR imaging system operating at a field of 1.5 Tesla (64 MHz). This system can perform a spin-echo pulse sequence with first echo times of 20,30 and 50. By means of a single-slice technique, multiple-echo (up to eight ethos) images can be obtained while using a multiple-slice technique that allows acquisition of two spin-echo images at multiple anatomic levels and/or at multiple phases. Spatial encoding is achieved by means of a two-dimensional Fourier transform technique with selective irradiation for plane selection. Acquisition of data can be performed in virtually any plane (coronal, sagittal, or transverse) by controlling a set of gradient angles to define the desired vector through the animal. The ECG gating is done by using three standard nonferromagnetic electrodes placed on the animal’s chest. ECG signals are transmitted by telemetry to a remote receiver. The peak of the R wave on the ECG is used to estimate the time of end diastole and to trigger the image acquisition. Imaging technique IN vrvo NMR IMAGING. Under anesthesia the dogs were placed on their right side inside a 50 cm diameter body coil. Imaging in the transverse orientation is first performed by centering at 2 cm above the apical impulse. This permits localization of the heart chambers and the ability to plan for the next acquisition in the coronal plane. The coronal plane allows visualization of the left ventricle in the long axis. The transverse plane is then oriented so that the imaging plane is perpendicular to the long axis and the number of slices selected so as to encompass the entire left ventricle from baae to apex. In the first three dogs gated spin-echo imaging was performed by means of a multiecho single-slice technique with echo times (TE) of 50,100,150, and 200 msec triggered 5 msec after the R wave. The repetition time (TR) varied between 1000 and 1100 msec related to the dog’s heart rate. However, because of motion-induced loss of signal at the late echo times (150 and 200 msec), and because of slightly improved signal-

to-noise and image quality with the use of the multislice technique, the remaining four dogs were studied by means of a multiple-slice, cyclic-gated, double spin-echo technique. The first echo image was obtained at a TE varying between 20 and 50 msec (mean 45 msec), and the second echo image was obtained at 100 msec. Six to seven tomographic slices along the short axis of the left ventricle were acquired every other R wave (TR varied between 1200 and 1400 msec) with a 1 mm gap. This sequence allowed acquisition of two echo images at six or seven phases of the cardiac cycle for each slice. By using two measurements, 10 mm thick slices were acquired on a 128 x 256 matrix with a field of view of 45 cm. Ex vrvo NMR IMAGING. The excised heart was positioned in the center of a 32 cm head coil. Tomographic images were obtained perpendicular to the long axis of the left ventricle. Five to eight 10 mm thick slices of the left ventricle were obtained by means of a single-slice, multiecho technique (four echo times) with TE = 50,100,150, and 200 msec and TR = 1500 msec. Images were acquired by means of a 128 X 256 matrix with only one measurement and a field of view of 25 cm. Image analysis and NMR quantitation of infarct size. In both in vivo and ex vivo studies the second echo image showed excellent contrast of the myocardial infarction but poor definition of the myocardial border. By means of a computer-generated region of interest, the first echo image was selected for tracing the left ventricular myocardial borders. The width and level of the window were chosen to optimize the visualization of the left ventricular myocardium. A track ball was used to control the computer-generated outline of the endocardial and epicardial interfaces. When necessary the late echo image was used to discriminate between the myocardial wall and the intracavitary signal. In the apical slice the endocardium could not usually be identified. On the second echo image the myocardial infarction appeared as an area of increased signal intensity. The width and level of the window were

February

284

Bouchard et al.

American

INFARCT r= 86 30 - SEE= 3.8 (gm.) n= 7

Infarct

Heart

lS8S Journal

SIZE r= .96 3o - SEE= 2 3 (%) n= 7

Mass (g)

% LV infarcted Pathology

I

INFARCT

SIZE

30

0’

r= .88

om Infarct

Mass (g)

1 1 10 20 % LV Infarcted

Pathology

Fig. 2. Top, Correlation betweeninfarct size determined by proton NMR imaging of excisedhearts and pathologic methods. Percentage of left ventricle (LV) infarcted was also determined (infarct mass/LV massx 100). Bottom, Correlation between in vivo NMR imaging and pathologic methods.

Table

I. Infarct size (gm)

NMR Dog No. 4036 4084 4128 4201 4231 4232 4278

Mean SD *Percentage

Pathology 7.8 17.4 21.7 8.1 9.8 14.5 19.5 14.1 5.6 of ilow reduction

imaging

Myocardial blood flow (microsphere)1 Pow reduction*

In vivo

Ex vivo

25%

40%

9.54 19.48 16.68 8.64 12.10 16.26 22.08 14.97 5.05

10.78 26.19 22.45 12.37 7.55 15.96 20.34 16.52 6.76

7.03 18.5 21.7 -

4.97 15.5 14.8 -

14.2 17.3 27.06 17.6 6.2

12.3 13.8 23.08 14.2 5.5

relative

to respective

control

zone tissue.

adjusted so that the signal from the normal myocardium (opposite the area with increasedsignal intensity) would be null. The outer borders of the zone with increased intensity were then planimetered with the track ball.

When necessarythe myocardial borders traced on the first echo image were recalled, so that only increased signal from the myocardium was included. The endocardial area wassubtracted from the epicardial area to derive the myocardial wall area. The wall area wasmultiplied by the slice thickness (10 mm) and by the gap thickness (when the multislice technique wasused)to obtain the myocardial volume for each slice and the total myocardial volume. After summation of slice volume, the resultant total left ventricular myocardial plus interslice gap was multiplied by the specific gravity of the myocardium (1.05 g/cc) to derive the left ventricular massand similarly the myocardial infarct mass:1.05g/cc multiplied by the sum of (infarct area X [lo mm t 1 mm interslice gap [with the multiple slicetechnique]) of each slice. The infarction was expressed as infarct mass and infarct percentage of the total myocardial mass. To determine intra- and interobserver variability, each image was planimetered on two separate occasions, 3 months apart, by the sameobserver and independently by a secondobserver without knowledge of the other observer’s results. OuantiWive afdemk evW*n. Hearts were immersion fixed in 10% phosphate-buffered formalin after the NMR study was completed. The right ventricular free wall, atria, and valves were dissectedfrom the left ventri-

Volume

117

Number

2

Proton NMR imaging of AMI

285

Fig. 3. Proton NMR imagesshowingextensive area of increasedsignal intensity on late echo (TE2 = 100 msec) involving inferoseptal and inferolateral regions.LAT, lateral.

cle and septum, which were then weighedand sliced at 10 mm intervals. Each slicewasweighedand an outline of the endocardial and epicardial surfaces,as well as the grossly identifiable infarct region, was made on a transparent overlay to record the location of samplesfor histology and microsphereblood flow analysis.Selectedtissuesfrom the infarct regionswere embeddedin paraffin, sectioned,and stained with hematoxylin and eosin and with Gomori aldehyde fuchsin trichrome to verify the grossidentification of infarct borders. Photographs of the basilar aspect of each slice were enlargedto approximately 20 X 25 cm, and a sonic digitizer (G&/pen, Science Accessories Corp., Southport, Conn.) interfaced to a Hewlett-Packard 9825A computer was used to determine the total ventricular and infarct area in eachslice. Infarct massand percentageof total left ventricle plus septum infarcted werecalculated from these areasand the slice weights. Microsphere

technique

Injection. Approximately 2 x lo6 polystyrene nonbiodegradable15 f 1.0 micron diameter tracer microspheres (3M or New England Nuclear) were injected into the left atrium over a period of 20 to 30 seconds.Microspheres were labeledwith either Ce-141,Cr-51, Sr-85, or SC-46and suspended in 10% dextran with 0.05% Tween-80 to prevent aggregation.Absenceof clumping wasdetermined microscopically. Microspheres were injected from a 2.0 ml glassinjector vial with the use of 20 to 30 ml saline wash. Five to 10 seconds before injection of microspheres, withdrawal of the reference blood flow samplewasstarted

at a rate of 7 to 10 ml/min into preweighedsampletubes by meansof a Buchler polystaltic pump. Weight of blood in the tubes/blood specific gravity (1.06) was used to determine the actual withdrawal flow rate. Blood flow analysis. The entire left ventricle and interventricular septum were slicedat 1 cm intervals from apex to base. Each slice of the left ventricle and septum was divided into nine or more transmural sections,and these sectionswere cut into epicardial, midwall, and endocardial pieces weighing 0.5 to 1.5 gm each.‘OThis resulted in a minimum of 27 samplesper 1.0 cm slice. The location of each samplewasrecorded on the slicemap. Sampleswere placed in tubes and gammaradioactivity determined in a LKB 1282CompuGammacounter with a 3-inch well-type sodium iodide crystal detector. Tissue weight was determined from dry weight obtained by oven drying the tissue after counting and converted to wet weight assuming78% water content. Standard gamma spectrometry with correction for spillover of radioactivity into adjacent regions was used to calculate activity for each isotope, and blood flow to each samplewas determined from these corrected counts by meansof the following equation: Qt = (C!,x Q,)/ C,, where Qt = tissue blood flow, C, is corrected tissue radioactivity, Qr is withdrawal rate of the reference blood flow sample,and C, is radioactivity in the reference flow sample. The total amount of tissue with reduced blood flow was determined as the sum of the weights of all samples from the region at risk with flow relative to respective control tissue zone (epicardium, midwall, and endocardium) below a predetermined value.

286

Bouchard

et al.

American

February 1989 Heart Journal

NMR IMAGING r= 0.80 30 - SEE= 4 5 grn n= 7

0

10, Infarct

r= 093 30-SEE=26% “= I

30I

20 Mass (gl

0

10

20 % LV Infarcted

30

Ex Vwa NMR Imaging

Fig. 4. Correlation between quantitation of infarct size determined by in vivo and ex vivo NMR

imaging. II. Regional myocardial blood flow (ml/gm/min) 1 week after coronary occlusion Table

Normalzone surrounding infarct

Infarct zone

Epicardium 0.74 z!z 0.29* 1.06 f 0.25 Midwall 0.44 f 0.18* 1.08 iz 0.30 Endocardium 0.32+ 0.12* 0.99-t 0.29* *Different

from control

zone at a statistical

Control zone 1.25 k 0.36 1.35 zk 0.42 1.45 * 0.49

level of p < 0.05

Statistical analysis. The infarct size was expressed as mean + standard deviation. Differences in siie of myocardial infarction determined by different techniques were compared by means of analysis of variance. The intra- and interobserver variance components were estimated from a

random-effect (type II)

analysis of variance. Linear

regression analysis was used to compare infarct size as determined by NMR imaging and determination by microsphere and anatomic methods. Tests were made for a zero intercept and a slope of 1, as well as for the significance of the regressionline. Relative strength of

correlations between NMFt measurements and pathology and flow measurements were assessedby means of partial sums of squares from multiple regression. RESULTS Relation

between

anatomic

and

NMR infarct

size.

The coronary artery occlusion created a myocardial infarction in all seven dogs. The pellet was located in the left anterior descending artery in five animals and in the circumflex artery in two. Anatomically the infarct mass varied from 8 to 22 gm or 7% to 27% of the left ventricular myocardium. NMR imaging visualized the myocardial infarct in alI seven dogs. Fig. 1A shows an NMR image of an excised heart and the corresponding fixed anatomic slices of the heart. The increased signal intensity on the second echo image (TE = 100 msec) can be seen in the region of the anterior walI. However, the lack

of involvement of the subepicardium on the anatomic section is not seen on the NMR image. The same phenomenon was seen on the in vivo images (Fig. 1B). The results of quantitation of infarct size for each dog are shown in Table I. When compared to anatomic data, the analaysis of ex vivo images overestimated the infarct mass. The largest discrepancy between the two measurements was 8.8 gm and was seen in a dog (No. 4084) with a moderately large infarct. There was good agreement between left ventricular mass calculated by ex vivo NMR and the actual myocardial weight (r = 0.92, standard error of estimate [SEE] = 9.8 gm). The equation describing the relationship was: NMR left ventricular mass = 0.95 x left ventricular weight - 1.3. The correlation coefficients between infarct mass and percentage of left ventricle infarcted measured by ex vivo NMR imaging and anatomy were significant at 0.86 (p = 0.013) and 0.96 (p = 0.0019), respectively (Fig. 2). The equations describing the relationship were: NMR = 1.0 X PATH + 2.05 for infarct mass and NMR = 1.1 X PATH + 2.6 for percentage of left ventricle infarcted. Despite significant correlations, the SEE was rather high indicating a limitation of the technique. In the in vivo imaging the use of plane angulation permitted visualization of the left ventricular myocardium in its short axis and accurate localization of infarcts in all animals, including the two dogs with involvement of the inferior wall (Fig. 3). The mean infarct mass as assessed by in viva NMR imaging was 15.0 + 5 gm and approached the value in the ex vivo NMR evaluation (16.5 & 7 gm). Quantitation of infarct size by means of in vivo NMR imaging correlated significantly with results of ex vivo NMR imaging (p = 0.03) (Fig. 4). The equations describing the relationship were: NMR in vivo = 0.6 X 2.0 for percentage of left ventricle infarcted. Again the SEE was rather high. A significant relationship was also observed between in vivo NMR imaging and

Volume Number

117 2

Proton NMR imaging

(F =

0.88,

between infarct size by NMR imaging analysis. The myocardial blood

and

pathology-determined p = 0.009; Fig. 2). Relation microsphere

infarct

size

INFARCT

3or flow All dogs had

data are summarized in Table II. substantial reduction of blood flow in the infarct zone compared to the control regions. For the normal-appearing myocardium immediately surrounding the infarct, the “border zone,” the reduction in blood flow to the epicardium, midwall, and endocardium averaged 17 % , 16%) and 33% respective to their control zones. The infarct sizing by ex vivo NMR imaging approximated the amount of myocardium with a reduction in blood flow of 25% or more relative to the control tissue zone (17.2 + 7.1 gm vs 17.6 + 6.8 gm, p = NS). The correlation coefficient between hypoperfused myocardial mass and NMR infarct size was 0.66 (p = 0.15). The equation describing the relationship was: NMR = 0.70 x flow + 4.9 (Fig. 5). Here the SEE is even worse than the correlation with anatomy. This poor correlation is largely related to the overestimation of infarct size in dog No. 4084. This dog had a mural thrombus that appeared bright on the ex vivo NMR image and was included in the calculation of infarct size. The exclusion of this dog brings the correlation between hypoperfused myocardium and ex vivo NMR imaging to 0.82. However, mural thrombus did not affect the in vivo NMR imaging, since the signal intensity of the thrombus decreases with the second echo. The mean infarct size of in vivo NMR imaging was 16.02 f 4.6 gm and approximated the mean hypoperfused myocardial mass with a reduction of blood flow of 25% or more, (17.6 f 6.8 gms). The correlation coefficient between hypoperfused myocardial mass and in vivo NMR infarct size was 0.92 (p = 0.008; Fig. 5). The multiple regression analysis showed a better correlation between in vivo NMR imaging and microsphere-determined hypoperfused myocardium with reductions in blood flow of 25% and 40% compared to respective control zones than pathology-determined infarct size (p = 0.001). Observer variability. The interobserver variabilities for infarct size measured by means of in vivo NMR imaging, ex vivo NMR imaging, and anatomy were 5.0%) 5.9%) and 3.6% respectively. The intraobserver variabilities for infarct size measured by means of in vivo and ex vivo NMR imaging were similar at 5.6% and 5.2%) respectively. DISCUSSION

The aim of this study was to evaluate T,-weighted proton NMR imaging for quantitation of the extent of the myocardial ischemic insult 1 week after

. Ex VIVO 0 In VIVO

of

AMI

287

SIZE

.

20-

NMR Imaging km)

0

10 Hypoperfused

t 30

20 Myocardium

(pm)

Fig. 5. Correlation between qua&it&ion

of ischemic insult by ex vivo (0) and in oivo (0) NMFI imaging and hypoperfused myocardium (>25 % reduction respective to control zone) as determined by microsphereblood flow. T Values were 0.66 and 0.92, respectively.

coronary occlusion in a canine model. Considering that the relaxation parameter T, increases substantially in zones of myocardial infarct as assessed by relaxometry of excised myocardial samples,6-8 we evaluated an approach for applying a double spinecho technique with echo times of 30 msec and 100 msec and a relatively long repetition time to reduce the effects of Tl. The rationale for choosing these two echo times was that the short TE would provide anatomic definition of the left ventricular myocardium and the long TE would increase the contrast between infarct (with prolonged TJ and normal tissues.” The long TR parameter allows more complete relaxation and minimizes the T, effects, which tend to decrease the signal intensity on spin-echo imaging. In instances where T, values in the normal and infarct zone are 700 msec and 900 msec, respectively, doubling the TR from 800 msec to 1600 msec will increase the signal intensity overall by 13% and reduce the T, effect by 50% according to the modified Bloch equation6’12: SI = H f(v) x l/eTErr2) X [l - l/e-*) 1, where SI = signal intensity, H = local hydrogen density, and f(v) = fraction and velocity of the moving nuclei through the imaged region. Results of the present study show that a T,weighted NMR imaging approach is useful for highlighting myocardium subjected to 1 week of coronary occlusion. The imaging approach used yielded a significant correlation between infarct size determined by NMR and that determined by pathologic methods (F = 0.86 and 0.96, respectively, for ex vivo

288

Eouchard

et al.

and 0.88 and 0.97 for in vivo NMR imaging). The differences in infarct size measurements between ex vivo and in vivo NMR imaging could be attributed in part to myocardial wall motion between pulses. However, NMR imaging generally tended to overestimate the actual size of the infarct. Although it is likely that partial volume effects contributed to this overestimation (particularly in hearts where the infarct involved the apex), measurements of myocardial blood flow distribution demonstrated a significant relationship between NMR estimation of infarct size and ischemic or hypoperfused zones. Although we did not measure blood flow immediately after occlusion, collateral vessel development would have increased flow to the infarcted region over the l-week period13,14 resulting in our relatively high blood flow measurements. The l-week period would also have reduced the area of hypoperfusion as measured by microspheres. Fig. 1 shows that the NMR image depicts abnormal signal well beyond the extent of the pathology-determined infarct. Thus it is evident from these studies that an abnormal increase in NMR signal intensity is present in regions that are both reversibly and irreversibly damaged. One explanation for this is that myocardial edema in reversibly damaged myocardium may persist despite only slightly reduced or even normalized regional blood flow. The presence of myocardial edema has been noted in both infarcted and noninfarcted but ischemic tissue.15 Accordingly the region of ischemic injury would have increased signal intensity on proton NMR images and would appear larger than that predicted from pathologic conditions alone. Moreover, the multiple regression analysis showed a better relationship between in vivo NMR imaging and microsphere-determined “hypoperfused” myocardium with reductions in flow of 25% or 40% than the relationship between in vivo NMR results and pathology-determined infarct size (p < 0.0001). This was not the case in the ex vivo NMR imaging study. The lesser correlation was attributed to the overestimation of infarct size caused by the presence of a mural thrombus in dog No. 4084. The thrombus appeared as a bright signal on the late echo image of the ex vivo NMR study and was incorporated into the infarct size measurement, as compared to the in vivo NMR imaging study where the mural thrombus appeared as a less intense signal on the late echo image. Therefore in this canine model of l-week coronary occlusion, ‘H NMR estimation of ischemic injury is sensitive to regions with reduced myocardial blood flow possibly because edema extends beyond the confines of the infarcted myocardium. This overestimation of

American

February 1989 Heart Journal

infarct size was also noted in a study by Buda et al.,‘” who used a canine model with 2 to 4 hours of coronary occlusion followed by reperfusion. In their study imaging was performed on the excised heart only, and the mean NMR infarct size compared better with hypoperfused myocardium as determined by autoradiography of technetium-99m microsphere distribution. However, their correlation between the different techniques of infarct size evaluation is difficult to interpret because the regression line obtained was substantially influenced by the three dogs without infarction. Results of several previous studies have shown that accurate quantitation of experimental infarct size is feasible by proton NMR imaging.17, I8 Rokey et a1.17used a calculated T, image for the quantitation of infarct size in an ex vivo NMR study in a canine model of 6-hour occlusion. There was excellent correlation between NMR and infarct size by histochemical staining (r = 0.98, SEE = 1.86%). Caputo et al.18 also used the calculated T, NMR image to quantitate infarct size 3 and 21 days after occlusion of the left anterior descending artery. The in vivo NMR infarct mass correlated significantly with pathologic estimation (r = 0.94, SEE = 1.54 gm). The infarct mass at 3 and 21 days was not significantly different. The calculated T, image has the advantage of being able to discriminate between increased signal caused by stasis of blood adjacent to the akinetic region of myocardium and increased signal caused by the infarcted myocardium itself. The even-echo rephasing of slowly moving blood results in a negative T, value.lg However, errors in calculated T, images can be introduced when only two echo times are used and when the two echo times do not correspond to the same phase of the cardiac cycle (30 msec apart). Caputo et al.lR also used a transverse imaging plane, which can create partial-volume problems when the inferior or the lateral wall of the left ventricle is involved. In summary, we have reported that a T,-weighted proton NMR imaging approach can be used to show infarcted myocardium. Accurate measurements of infarct size can be achieved in both ex vivo and in vivo imaging studies with a high-field (1.5 Tesla) NMR imaging system. The increased signal intensity seen on the late echo images (TE = 100 msec) correlated significantly with the necrotic zone as assessed by pathologic conditions. NMR imaging, however, seemed to overestimate the infarct size. Our data suggest that this NMR method reflects a zone comprised of both infarct and reversibly injured myocardium. In fact, the correlation was

Volume Number

117 2

better with “hypoperfused myocardium” 025 % reduction respective to control zone) as evaluated by microsphere blood flow determination. These observations suggest a potential role for the application of this imaging method in evaluating patients after a recent myocardial infarction. We acknowledge the statistical assistance of Katherine Kirk, PhD, and the secretarial assistance of Margaret Burchfield. REFERENCES

1. Sobel BE, Bresnahan GF, Shell WE, Yoder RD. Estimation of infarct size in man and its relation to prognosis. Circulation 1972;46:640-8. 2. Page DL, Caulfield JB, Kastor JA, et al. Myocardial changes associated with cardiogenic shock. N Engl J Med 1971; 285:133-7. 3. Cox JR, Roberta R, Ambus HD, et al. Relations between enzymatically estimated myocardial infarct size and early ventricular dysrhythmia. Circulation 1976;53(suppl 1):1505. 4. Bottomley PA, Hardy CJ, Argersinger RE, Allen-Moore G. A review of ‘H nuclear magnetic resonance relaxation in pathology: are T, and T, diagnostic? Med Phys 1987;14:1-37. 5. Herfkens R, Davis PL, Crooks LE, Kaufman L, Price D, Miller T, Margulis AR, Watts J, Hoenninger J, Arakawa M, McRee R. NMR imaging of the abnormal live rat and correlation with tissue characteristics. Radiology 1981; 141:211-18. 6. Ratner AV, Okada R, Newell J, Pohost GM. The relationship between proton NMR relaxation parameters and myocardial perfusion with acute coronary artery occlusion and reperfusion. Circulation 1985,71:823-8. 7. Higgins CB, Herfkens R, Lipton MJ, Sievers R, Sheldon P, Kaufman L, Crooks LE. Nuclear magnetic resonance imaging of acute myocardial infarction in dogs: alterations in magnetic relaxation times. Am J Cardiol 1983;52:184-8.

Proton NMR imaging of AMI

289

8. Canby RC, Reeves RC, Evanochko WT, Elgavish GA, Pohost GM. Proton nuclear magnetic resonance relaxation times in severe myocardial ischemia. J Am Co11 Cardiol 1987;10:41220. 9. Crooks LE. Overview of NMR imaging techniques. In: Kaufman L, Crooks LE, Margulis AR, eds. Nuclear magnetic resonance imaging in medicine. Tokyo: Igaku-Shoin, 1981:3052. 10. Bishop SP. How well can we measure coronary flow, risk zones, and infarct size? In: Hearse DJ, Yellon DM, eds. Therapeutic approaches to myocardial infarct size limitation. New York: Raven Press, 1984134. 11. Crooks LE, Ortendahl DA, Kaufman L, et al. Clinical efficiency of nuclear magnetic resonance imaging. Radiology 1983;146:123-8. 12. Bloch F. Nuclear induction. Physiol Rev 1946;70:460-73. 13. Bishop SP, White FC, Bloor CM. Regional myocardial blood flow during acute myocardial infarction in the conscious dog. Circ Res 1976;38:429-38. 14. Rivas F, Cobb FR, Bathe RJ, Greenfield JC. Relationship between blood flow to ischemic regions and extent of myocardial infarction. Serial measurement of blood flow to ischemic regions in dogs. Circ Res 1976;38:439-47. 15. Powell Jr WJ, DiBona DR, Flores J, Leaf A. The protective effect of hyperosmotic mannitol in myocardial ischemia and necrosis. Circulation 1976;54:603. 16. Buda AJ, Aisen AM, Juni JE, Gallagher KP, Zotz RJ. Detection and sizing of myocardial ischemia and infarction by nuclear magnetic resonance imaging in the canine heart. AM HEART J 1985;110:1284-90. 17. Rokey R, Verani MS, Bolli R, et al. Myocardial infarct size quantification by NMR imaging early after coronary artery occlusion in dogs. Radiology 1986;158:771-4. 18. Caputo GR, Sechtem V, Tscholakoff D, Higgins CB. Measurement of myocardial infarct size at early and late time intervals using NMR imaging. An experimental study in dogs. Radiology 1987;149:237-43. 19. Walach V, Ka-Siu L, Fernandez EJ, Spalter C. The appearance of rapidly flowing blood on magnetic resonance imaging. Am Med J 1984;143:1167-74.