Elias A. Zerhouni

Prognostic Significance of Microvascular Obstruction by Magnetic Resonance Imaging in Patients With Acute Myocardial Infarction

BACKGROUNDnThe extent of microvascular obstruction during acute coronary occlusion may determine the eventual magnitude of myocardial damage and thus, patient prognosis after infarction. By contrast-enhanced MRI, regions of profound microvascular obstruction at the infarct core are hypoenhanced and correspond to greater myocardial damage acutely. We investigated whether profound microvascular obstruction after infarction predicts 2-year cardiovascular morbidity and mortality.nnnMETHODS AND RESULTSnForty-four patients underwent MRI 10 +/- 6 days after infarction. Microvascular obstruction was defined as hypoenhancement seen 1 to 2 minutes after contrast injection. Infarct size was assessed as percent left ventricular mass hyperenhanced 5 to 10 minutes after contrast. Patients were followed clinically for 16 +/- 5 months. Seventeen patients returned 6 months after infarction for repeat MRI. Patients with microvascular obstruction (n = 11) had more cardiovascular events than those without (45% versus 9%; P=.016). In fact, microvascular status predicted occurrence of cardiovascular complications (chi2 = 6.46, P<.01). The risk of adverse events increased with infarct extent (30%, 43%, and 71% for small [n = 10], midsized [n = 14], and large [n = 14] infarcts, P<.05). Even after infarct size was controlled for, the presence of microvascular obstruction remained a prognostic marker of postinfarction complications (chi2 = 5.17, P<.05). Among those returning for follow-up imaging, the presence of microvascular obstruction was associated with fibrous scar formation (chi2 = 10.0, P<.01) and left ventricular remodeling (P<.05).nnnCONCLUSIONSnAfter infarction, MRI-determined microvascular obstruction predicts more frequent cardiovascular complications. In addition, infarct size determined by MRI also relates directly to long-term prognosis in patients with acute myocardial infarction. Moreover, microvascular status remains a strong prognostic marker even after control for infarct size.

Physiological Basis of Myocardial Contrast Enhancement in Fast Magnetic Resonance Images of 2-Day-Old Reperfused Canine Infarcts

BACKGROUNDnContrast-enhanced fast magnetic resonance (MR) images of acute, reperfused human infarcts demonstrate regions of hypoenhancement and hyperenhancement. The relations between the spatial extent and time course of these enhancement patterns to myocardial risk, infarct, and no-reflow regions have not been well characterized.nnnMETHODS AND RESULTSnThe proximal left anterior descending coronary artery was occluded in 11 closed-chest dogs for 90 minutes followed by 2 days of reperfusion. Regional blood flow was determined by use of radioactive microspheres. The animals were studied at the 2-day time point with contrast-enhanced fast MRI (Signa 1.5 T, General Electric). Thioflavin-S was administered to demarcate no-reflow regions. The hearts were then excised, sectioned into five base-to-apex slices, stained with 2,3,5-triphenyltetrazolium chloride (TTC), and photographed under room light (for TTC) and ultraviolet light (for thioflavin). The spatial extents of thioflavin-negative, TTC-negative, and risk regions were compared planimetrically with MRI hypoenhanced and hyperenhanced regions. The spatial locations of subendocardial hypoenhancement in MR images correlated closely with those of thioflavin-negative regions. Microsphere blood flow in these regions was significantly reduced compared with remote regions (0.37 +/- 0.09 versus 0.88 +/- 0.10 mL/min per gram, respectively, P < .001) and with baseline (0.37 +/- 0.09 versus 0.87 +/- 0.15 mL/min per gram, P < .01). The spatial extent of hyperenhancement was smaller than the risk region (r = .64, slope = 0.48, P < .001) but highly correlated with TTC-negative regions and were, on average, 12% larger (r = .93, slope = 1.12, P = .035).nnnCONCLUSIONSnIn contrast-enhanced MR images of 2-day-old reperfused canine infarcts, myocardial regions of hypoenhancement are related to the no-reflow phenomenon. Approximately 90% of the myocardium within hyperenhanced regions is nonviable.

Regional Heterogeneity of Human Myocardial Infarcts Demonstrated by Contrast-Enhanced MRI Potential Mechanisms

BACKGROUNDnMyocardial reperfusion is pivotal to the prognosis of patients with acute myocardial infarction. In these patients, coronary flow is generally assessed by angiography and tissue perfusion by tracer scintigraphy. This study was designed to examine whether magnetic resonance imaging (MRI) provides information on myocardial perfusion and damage beyond that supplied by angiography and thallium scintigraphy after acute myocardial infarction.nnnMETHODS AND RESULTSnTwenty-two patients with recent myocardial infarction had ECG, echocardiography, coronary angiography, and fast contrast-enhanced MRI. Twelve patients also had exercise thallium scintigraphy. Time-intensity curves obtained from infarcted and noninfarcted regions were correlated with coronary anatomy and left ventricular function. Two perfusion patterns were observed in infarcted regions by comparison with the normal myocardial pattern. All patients but 1 had persistent myocardial hyperenhancement within the infarcted region up to 10 minutes after contrast. In 10 patients, this hyperenhanced region surrounded a subendocardial area of decreased signal at the center of the infarcted region associated with coronary occlusion at angiography, Q waves on ECG, and greater regional dysfunction by echocardiography. Moreover, the extent and location of the MRI abnormalities correlated well with the extent and location of the fixed single-photon emission computed tomography thallium defects.nnnCONCLUSIONSnLarge human infarcts, associated with prolonged obstruction of the infarct-related artery, are characterized by central dark zones surrounded by hyperenhanced regions on MRI. Conversely, reperfused infarcts with less regional dysfunction have uniform signal hyperenhancement. The MRI hyperenhanced segment correlates well with the fixed scintigraphic defect in patients with acute myocardial infarction.

Noninvasive quantification of left ventricular rotational deformation in normal humans using magnetic resonance imaging myocardial tagging.

It has been postulated that rotation of the left ventricular apex with respect to the base is a component of normal systolic function in humans, but it has been difficult to measure it noninvasively. Tagging is a new magnetic resonance imaging technique that labels specific areas of myocardium by selective radio-frequency excitation of narrow planes orthogonal to the imaging plane before acquiring an image. Tags appear as black lines and persist in myocardium for 400-500 msec and, if applied at end diastole, will move with the myocardium through systole. Tagging was used to noninvasively quantify left ventricular torsion and circumferential-longitudinal shear (shearCL) in humans. Eight normal volunteers, aged 24-38 years, were imaged in a 0.38-T iron-core resistive magnet. Five short-axis left ventricular images, positioned to encompass the entire left ventricle (LV), were obtained separately at end systole. Four equiangular radial tags had been applied at end diastole, intersecting the myocardium at eight locations. We calculated the difference in angular displacement of each epicardial and endocardial tag point (a tag point being where the tag crossed the epicardium or endocardium) at end systole from the systolic position of the corresponding tag point on the basal plane. This value was called the torsion angle. From this, shearCL, the angle inscribed on the epicardial or endocardial surface between the systolic tag position, the corresponding basal tag position, and its projection onto the slice of interest could be calculated at 32 points in the left ventricular wall.(ABSTRACT TRUNCATED AT 250 WORDS)

Dissociation between left ventricular untwisting and filling. Accentuation by catecholamines.

Background Efficient early diastolic filling is essential for normal cardiac function. Diastolic suction, as evidenced by a decreasing left ventricular pressure during early filling, could result from restoring forces (the release of potential energy stored during systolic deformation) dependent on myofilament relaxation. Although these restoring forces have been envisioned within individual myofibers, recent studies suggest that gross fiber rearrangement involving the connective tissue network occurs early in diastole. This may lead to the release of potential energy stored during systole and suction-aided filling. Methods and Results To establish precisely the timing and extent of restoration of the systolic torsional deformation of the left ventricle with respect to early filling at baseline and with enhanced relaxation, we studied untwisting during control conditions and with catecholamine stimulation. Using noninvasive and nondestructive magnetic resonance tagging, torsional deformation of the left ventricle was measured at 20-msec intervals in 10 open-chest, atrially paced dogs, starting at aortic valve closure. Eight equiangular tags intersected the epicardium and endocardium in three short-axis imaging planes (base, mid, and apex). From the intersection points, epicardial and endocardial circumferential chord and arc lengths were measured and angular twist of mid and apical levels with respect to the base (maximal torsion and its reversal, untwisting) was calculated. Echo-Doppler provided timing of aortic valve closure and of mitral valve opening. Zero torsion was defined at end diastole. Torsion at the apical level reversed rapidly between its maximum and the time immediately after mitral valve opening: from 7.0±5.80 to 3.2±5.40 and 12.0±8.50 to 6.9±7.80 (mean±SD, both p < 0.01) at the epicardium and endocardium, respectively. During the same period, no significant circumferential segment length changes occurred. As expected, after mitral valve opening, filling resulted in significant circumferential segment lengthening, whereas further reversal of torsion was small and nonsignificant. During dobutamine infusion, torsion at end systole was greater and reversal during isovolumic relaxation was much more rapid and greater in extent (p < 0.01). Torsion reversed from 11.5±4.3° to 5.7±4.8° and 17.4±6.4° to 6.9±7.7° at epicardium and endocardium. Conclusions Untwisting occurs principally during isovolumic relaxation before filling and is markedly enhanced in speed and magnitude by catecholamines. This partial return of the left ventricle to its preejection configuration before mitral valve opening could represent an important mechanism for the release of potential energy stored in elastic elements during the systolic deformation. These myocardial restoring forces would be markedly enhanced by physiological changes consequent to catecholamines such as during exercise, offsetting the concomitant shortening of the filling period.

Quantification and Time Course of Microvascular Obstruction by Contrast-Enhanced Echocardiography and Magnetic Resonance Imaging Following Acute Myocardial Infarction and Reperfusion

OBJECTIVESnWe aimed to validate contrast-enhanced echocardiography (CE) in the quantification of microvascular obstruction (MO) against magnetic resonance imaging (MRI) and the histopathologic standards of radioactive microspheres and thioflavin-S staining. We also determined the time course of MO at days 2 and 9 after infarction and reperfusion.nnnBACKGROUNDnPostinfarction MO occurs because prolonged ischemia produces microvessel occlusion at the infarct core, preventing adequate reperfusion. Microvascular obstruction expands up to 48 h after reperfusion; the time course beyond 2 days is unknown. Though used to study MO, CE has not been compared with MRI and thioflavin-S, which yield precise visual maps of MO.nnnMETHODSnTen closed-chest dogs underwent 90-min coronary artery occlusion and reperfusion. Both CE and MRI were performed at 2 and 9 days after reperfusion. The MO regions by both methods were quantified as percent left ventricular (% LV) mass. Radioactive microspheres were injected for blood flow determination. Postmortem, the myocardium was stained with thioflavin-S and 2,3,5-triphenyltetrazolium chloride.nnnRESULTSnExpressed as % total LV, MO by MRI matched in size MO by microspheres using a flow threshold of <40% remote (4.96+/-3.52% vs. 5.32+/-3.98%, p=NS). For matched LV cross sections, MO by CE matched in size MO by microspheres using a flow threshold of <60% remote (13.27+/-4.31% vs. 13.5+/-4.94%, p=NS). Both noninvasive techniques correlated well with microspheres (MRI vs. CE, r=0.87 vs. 0.74; p=NS). Microvascular obstruction by CE corresponded spatially to MRI-hypoenhanced regions and thioflavin-negative regions. For matched LV slices at 9 days after reperfusion, MO measured 12.94+/-4.51% by CE, 7.11+/-3.68% by MRI and 9.18+/-4.32% by thioflavin-S. Compared to thioflavin-S, both noninvasive techniques correlated well (CE vs. MRI, r=0.79 vs. 0.91; p=NS). Microvascular obstruction size was unchanged at 2 and 9 days (CE: 13.23+/-4.11% vs. 12.69+/-4.97%; MRI: 5.53+/-4.94% vs. 4.68+/-3.44%; p=NS for both).nnnCONCLUSIONSnBoth CE and MRI can quantify MO. Both correlate well with the histopathologic standards. While MRI can detect regions of MO with blood flow <40% of remote, the threshold for MO by CE is <60% remote. The extent of MO is unchanged at 2 and 9 days after reperfusion.

Fast determination of regional myocardial strain fields from tagged cardiac images using harmonic phase MRI

BACKGROUNDnTagged MRI of the heart is difficult to implement clinically because of the lack of fast analytical techniques. We investigated the accuracy of harmonic phase (HARP) imaging for rapid quantification of myocardial strains and for detailed analysis of left ventricular (LV) function during dobutamine stimulation.nnnMETHODS AND RESULTSnTagged MRI was performed in 10 volunteers at rest and during 5 to 20 microg(-1). kg(-1). min(-1) dobutamine and in 9 postinfarct patients at rest. We compared 2D myocardial strains (circumferential shortening, Ecc; maximal shortening, E(2); and E(2), direction) as assessed by a conventional technique and by HARP. Full quantitative analysis of the data was 10 times faster with HARP. For pooled data, the regression coefficient was r=0.93 for each strain (P<0.001). In volunteers, Ecc and E(2) were greater in the free wall than in the septum (P<0.01), but recruitable myocardial strain at peak dobutamine was greater in the LV septum (P<0.01). E(2) orientation shifted away from the circumferential direction at peak dobutamine (P<0.01). HARP accurately detected subtle changes in myocardial strain fields under increasing doses of dobutamine. In patients, HARP-determined Ecc and E(2) values were dramatically reduced in the asynergic segments as compared with remote (P<0.001), and E(2) direction shifted away from the circumferential direction (P<0.001).nnnCONCLUSIONSnHARP MRI provides fast, accurate assessment of myocardial strains from tagged MR images in normal subjects and in patients with coronary artery disease with wall motion abnormalities. HARP correctly indexes dobutamine-induced changes in strains and has the potential for on-line quantitative monitoring of LV function during stress testing.

Determination of left ventricular mass by magnetic resonance imaging in hearts deformed by acute infarction.

Measurement of left ventricular (LV) mass by magnetic resonance imaging (MRI) is accurate in normal hearts. Because determination of mass by MRI does not require assumptions about ventricular shape, this method may be well suited for evaluating hearts distorted by infarction. To test this hypothesis, gated MRI was performed in 15 dogs before and after acute myocardial infarction. The LV mass of each dog was calculated from five short-axis images acquired at end systole, when shape distortion is greatest, at end diastole, and also from slices at varying phases of the cycle with a multiphase mode that required only one acquisition. Correlation was excellent between actual mass and end-systolic mass before infarction (p less than 0.001, r = 0.98, and SEE = 5.1 g) and after infarction (p less than 0.001, r = 0.97, and SEE = 6.6 g). Likewise, values correlated closely at end diastole before (p less than 0.001, r = 0.96, and SEE = 6.7 g) and after infarction (p less than 0.001, r = 0.94, and SEE = 8.7 g). Surprisingly, measurements of mass by a multiphase mode were also very accurate before (p less than 0.001, r = 0.98, and SEE = 5.1 g) and after (p less than 0.001, r = 0.95, and SEE = 6.49 g) infarction. Therefore, at the same phase and at multiphases of the cardiac cycle, MRI permits accurate determination of LV mass in distorted hearts.

Quantification of and correction for left ventricular systolic long-axis shortening by magnetic resonance tissue tagging and slice isolation.

BackgroundMeasurement of regional left ventricular (LV) function is predicated on the ability to compare equivalent LV segments at different time points during the cardiac cycle. Standard techniques of short-axis acquisition in two-dimensional echocardiography, cine computed tomography, and standard magnetic resonance imaging (MRI) acquire images from a fixed plane and fail to compensate for through-plane motion. The shortening of the left ventricle along its long axis during systole results in planar images of two different levels of the ventricle, leading to error in any derived functional measurements. LV systolic long-axis motion was measured in 19 normal volunteers using MRI. Methods and ResultsWith a selective radio frequency (RF) tissue-tagging technique, three short-axis planes were labeled at end diastole and standard spin-echo images were acquired at end systole in the two- and four-chamber orientations. Persistence of the tags through systole allowed visualization of the intersecting short-axis tags in the long-axis images and allowed precise quantification of long-axis motion of the septum, lateral, anterior, and inferior walls at the base, mid, and apical LV levels. The total change in position along the long axis between end diastole and end systole was greatest at the base, which moved toward the apex 12.8 ± 3.8 mm. The mid left ventricle moved 6.9 ± 2.6 mm, and the apex was nearly stationary, moving only 1.6 ± 2.2 mm (p < 0.001). Having quantified the normal range of long-axis shortening, we developed a technique that isolates a slice of tissue between selective RF saturation planes at end diastole. Combining this with a wide end-systolic image slice, end-systolic images were acquired without contamination of signal from adjacent tissue moving into the imaging plane. This technique was validated in a moving phantom and in normal volunteers. ConclusionsSignificant LV systolic long-axis shortening exists, and this effect is seen the most at the base and the least at the apex. At a given ventricular level, shortening varied significantly according to location. A method using selective saturation pulses and gated spin-echo MRI automatically corrects for this motion and thus eliminates misregistration artifact from regional function analysis.

Accurate systolic wall thickening by nuclear magnetic resonance imaging with tissue tagging: Correlation with sonomicrometers in normal and ischemic myocardium☆

OBJECTIVESnThis study examined whether the correlation of systolic wall thickening (%WT) by nuclear magnetic resonance (NMR) imaging with wall thickening by sonomicrometry (SM) is improved by using a three-dimensional volume element model of the left ventricular wall.nnnBACKGROUNDnLeft ventricular wall obliquity with respect to the imaging plane causes overestimation of wall thickness by planar imaging techniques. Wall thickness perpendicular to the endocardial surface can be accurately calculated by three-dimensional reconstruction of left ventricular wall segments.nnnMETHODSnSonomicrometers were placed transmurally in 11 dogs (left anterior descending artery territory) with an imaging marker, visible on NMR imaging, sewn to the epicardial crystal. Two adjacent NMR short-axis image planes were radially segmented by four perpendicular spin-saturated planes (tags), dividing the myocardium into eight volume elements, one of which contained the sonomicrometer crystal pair. Left ventricular thickness and thickening were calculated by four methods: 1) linear = distance between epicardium and endocardium at midpoint in the segment with the sonomicrometer; 2) planar = area of that segment divided by the mean of the endocardial and epicardial arc lengths; 3) biplanar = average of wall thicknesses calculated by the planar method from the segment with sonomicrometers and the corresponding segment located in the adjacent short-axis imaging plane; and 4) three-dimensional = volume of the element with the sonomicrometers divided by the mean of the endocardial and epicardial surface areas.nnnRESULTSnRegressions for all methods using pooled data from control periods and during ischemia: Linear %WT = 0.59 + 1.31 SM%WT (r = 0.71, SEE = 0.28, p < 0.0002) Planar %WT = 1.43 + 1.62 SM%WT (r = 0.87, SEE = 0.19, p < 0.0001) Biplanar %WT = 2.09 + 1.46 SM%WT (r = 0.90, SEE = 0.15, p < 0.0001) Three-dimensional %WT = 0.19 + 1.49 SM%WT (r = 0.95, SEE = 0.10, p < 0.0001)nnnCONCLUSIONSnNuclear magnetic resonance imaging with tissue tagging allows accurate noninvasive assessment of systolic wall thickening. The three-dimensional volume element approach, by accounting for obliquity between the image plane and the left ventricular wall, provides the strongest correlation between NMR imaging and percent systolic wall thickening by sonomicrometer crystals.