Author + information
- Received April 17, 2000
- Revision received September 7, 2000
- Accepted September 14, 2000
- Published online January 1, 2001.
- Hisashi Masugata, MDa,
- Barry Peters, MDa,
- Stephane Lafitte, MDa,
- G.Monet Strachan, RDCSa,
- Koji Ohmori, MDa and
- Anthony N DeMaria, MD, FACCa,* ()
- ↵*Reprint requests and correspondence: Dr. Anthony N. DeMaria, Cardiovascular Division, UCSD Medical Center, 200 West Arbor Street, San Diego, California 92103
The present study examined the ability of real-time myocardial contrast echocardiography (MCE) to delineate abnormalities produced by graded coronary stenoses and to correlate signal intensity (SI) parameters derived from destruction/refilling curves with regional myocardial blood flow (MBF) and contractile function.
Recent technological advances have enabled myocardial opacification by MCE to be achieved during real-time imaging.
In eight open-chest dogs, we created LAD occlusion and graded stenoses that were either flow-limiting at rest (FLS) or reduced adenosine hyperemia (non-flow-limiting at rest = NFLS). Myocardial contrast echo used Optison infusion and low-energy real-time power pulse inversion imaging. High-energy FLASH frames destroyed bubbles every 15 cardiac cycles. Myocardial SI-versus-time plots were fitted to a one-exponential function to obtain the rate of SI rise (b) and peak SI in the last frame.
Dyssynergy was not observed during any NFLS, but perfusion abnormalities were. Visual detection of decreased opacification was possible with severe NFLS and FLS. bdemonstrated a significant reduction with severe NFLS and near significant with moderate NFLS; peak SI did not. All exponential parameters were significantly decreased with FL stenosis and occlusion. The MBF ratio in LAD/LCx beds (fluorescent microspheres) correlated with b(r = 0.79) and the product of the peak SI and b(r = 0.80).
In an open-chest dog model, parameters derived from microbubble refilling of the imaging field by real-time MCE correlate well with myocardial blood flow and can identify coronary stenosis.
The ability of ultrasound to destroy contrast microbubbles was initially demonstrated by Porter et al. (1). Shortly thereafter, it was recognized that the myocardial contrast echocardiography (MCE) intensity observed at any level of ultrasound exposure was determined by the degree of bubble destruction produced and the velocity at which blood containing undestroyed microbubbles refilled the imaging field between transmitted pulses (2). Wei et al. (3)applied progressively prolonged intervals between electrocardiogram (ECG)-gated imaging pulses and demonstrated that signal intensity increased over time until a peak plateau value was reached, at which refilling of the imaging field between pulses was complete. They fitted the time-intensity data to an exponential equation as Y = A(1 − e−bt) (where A was peak plateau intensity and b was the rate of signal increase), and demonstrated that both b and the product of A and b could be correlated with myocardial blood flow. However, clinical application of this approach has been hampered by the limited number of interpulse intervals available with ECG gating, the potential for the myocardium within the beam to be displaced during the long intervals between imaging pulses, and the inability to evaluate contractile function.
Recent technological advances have enabled myocardial opacification by MCE to be achieved during real-time imaging (4,5). Using real-time MCE, high-energy ultrasound bursts (referred to as FLASH frames) can be transmitted every 10 to 15 cardiac cycles to produce bubble destruction, after which subsequent consecutive frames delineate the restoration of contrast intensity. Time-intensity plots from the resulting data can be subjected to curve fitting and provide measures of the rate of increase and the peak level of signal intensity.
At present, no data exist regarding the ability of real-time MCE time-intensity curves to identify and quantify perfusion abnormalities produced by coronary stenosis. Nor are data available regarding the relationship of intensity measurements derived from real-time MCE refilling curves with either myocardial blood flow or contractile performance in the setting of graded coronary stenoses. Therefore, we examined the ability of real-time intravenous MCE to delineate myocardial perfusion abnormalities produced by graded coronary stenoses, and to correlate signal intensity (SI) parameters derived from real-time refilling curves with measures of regional myocardial blood flow (MBF) and contractile function.
The study was approved by the University of California at San Diego Animal Research Committee. In eight anesthetized mongrel dogs femoral artery catheters (7F) measured hemodynamics and provided blood samples. The heart was exposed through a left lateral thoracotomy and suspended in a pericardial cradle. Fluorescent microspheres were injected through a left atrial catheter. The proximal portion of the left anterior descending coronary artery (LAD) was dissected free from the surrounding tissue and a transit-time flow probe (Transonics series 2RB) connected to a digital flowmeter (model T 201 Transonics System, Ithaca, New York) was placed snugly around the vessel. A custom-designed screw occluder distal to the flow probe produced graded stenoses.
Myocardial contrast echocardiography by PPI
Contrast was produced by the continuous infusion of 0.2 ml/min FS-069 (Optison; Molecular Biosystems, San Diego, California) with a gently agitated volumetric pump. The dose was selected on the basis of pilot experiments and was the lowest dose that provided definite visible myocardial opacification in real-time imaging. Recordings were obtained 2 min after initiating infusion to ensure that myocardial opacification had reached a plateau intensity. A latex bag filled with degassed saline functioned as an acoustic interface between the heart and the transducer, which was positioned to image the LAD perfusion territory.
Imaging was performed with a commercial instrument (HDI 5000; Advanced Technology Laboratories, Inc., Seattle, Washington) using a broadband 4-2 MHz transducer. Color-coded harmonic power pulse inversion (PPI) imaging was performed with ultrasound transmitted at 2 MHz and received at 4 MHz in short-axis view at papillary muscle level using low-energy (mechanical index 0.12) real-time imaging at 11 to 13 frames/s (4,5). Pulse inversion imaging is a pulse cancellation technique that transmits two pulses per image line, each of which is the inverse phase of the preceding pulse. Adding the returned signals cancels the fundamental frequencies from tissue but not the nonlinear signals from bubbles. A low dynamic range (dB) was used and the pulse repetition frequency was fixed at 2,500 Hz. Instrument settings were held constant for each experiment. Profound microbubble destruction was produced by two or three consecutive high-energy (mechanical index of 0.8) FLASH frames transmitted every 15 cardiac cycles to yield unopacified myocardium. The low-energy real-time scanning allowed microbubble replenishment into the ultrasound field. Because tissue harmonic signals are produced primarily at high mechanical index energy, operating at very low mechanical index not only reduces microbubble destruction but also reduces the tissue harmonic component in the image.
Frames preceding and following the high-energy FLASH were digitally captured and analyzed off-line. Myocardial signal intensity (SI) was measured for each 15 cardiac cycle digital cineloop with HDI Lab software (ATL Ultrasound, Bothell, Washington). In this software, LAD bed SI at every frame (every phase of cardiac cycle) versus time after FLASH was fitted to an exponential function: y = A(1 − exp−bt) + C, where y is SI at any given time, A is the peak plateau amplitude that reflects microvascular cross-sectional area or myocardial blood volume, b is the rate of SI rise (slope of the curve) that reflects myocardial microbubble velocity, t is time after FLASH, and C is offset of intensity. The product of A and b provides a measure of myocardial blood flow (3). The rate of intensity increase was substantially diminished in the presence of stenoses of severe grade so that a true plateau was not apparent. Therefore, A was taken as the maximal intensity reached at the end of the imaging sequence. The LAD perfusion territory was identified as the transmural area of absent opacification following contrast injection during coronary occlusion. Transmural regions of interest (ROI) were selected that encompassed the LAD perfusion territory and adjacent nonattenuated area of the left circumflex (LCx) coronary artery but excluded high-intensity signals from the epicardium and endocardium. The ratio of LAD/LCx values was then used to normalize intensity values and was subjected to linear regression analysis compared with MBF.
To assess regional left ventricular (LV) contraction, regional wall thickness was measured from gray scale with digital calipers in systole and diastole at the center of risk area. Wall thickening (WT) was calculated as WT (%) = [(end-systolic wall thickness) − (end-diastolic wall thickness)]/end-diastolic wall thickness × 100.
Myocardial blood flow measurement
Myocardial blood flow (MBF) was measured by injecting fluorescent microspheres (Molecular Probes, Eugene, Oregon) into the left atrium during blood withdrawal from the femoral artery. After the animal was euthanized, the heart was sliced into cross-sectional segments and the slice corresponding to the short-axis image was cut into 12 wedge-shaped transmural tissue pieces, each of which was divided into endocardial and epicardial segments. The tissue and the arterial reference sample were processed to count the microspheres. MBF to each endocardial and epicardial segment was calculated from the equation Qm = (Cm × Qr)/Cr, where Qm is the blood flow to the myocardial segment (ml/min), Cm is tissue count, Qr is rate of arterial sample withdrawal (ml/min), and Cr is arterial reference sample count (6). Transmural MBF (ml/min/g) to 12 wedge-shaped pieces was calculated as the quotient of the summed flows to the individual segments within that piece and their combined weight. MBF to the LAD and LCx beds defined by monastral blue dye injection was then calculated by averaging the transmural MBF in the pieces from each bed respectively. The LAD/LCx ratio of transmural MBF was also calculated.
After the acquisition of baseline MCE and MBF data, adenosine was infused intravenously at 140 μg/kg/min. When stable hyperemia was reached, MCE and MBF data were repeated. Thereafter, guided by flow probe, the LAD area was progressively compressed with a screw occluder to produce three levels of non-flow-limiting stenosis (NFLS), which reduced the increase in LAD flow produced by hyperemia (i.e., coronary flow reserve) by 30% (mild), 60% (moderate), or 90% (severe). MCE and MBF measurement were performed at each NFLS. After the three levels of NFLS, the adenosine infusion was stopped and LAD flow was allowed to decrease to baseline value in all dogs. Thereafter, a flow-limiting-stenosis (FLS) was created by reducing LAD flow by 50% of the baseline value. Adenosine was infused again and MCE and MBF data were reacquired. Finally, total LAD occlusion was produced and MCE and MBF measurements were performed. At the end of the experiment, the LAD was occluded at the occluder site, and monastral blue dye was injected into the left atrium to delineate the LAD and LCx beds. The dog was then euthanized and the slice corresponding to the MCE imaging plane was processed for fluorescent microsphere analysis.
Data were expressed as mean ± SD. Comparisons of hemodynamics, MBF, and MCE data among all stages were performed using repeated-measures ANOVA, and Bonferroni t-test was used to assess the statistical difference between multiple comparisons. Correlation between MBF and MCE data was performed by linear regression analysis. For differences, a value of p < 0.05 was considered significant.
Hemodynamic data are presented in Table 1. Systolic and diastolic blood pressure decreased during adenosine infusion, but was significant only with complete LAD occlusion. LAD blood flow increased during vasodilation both in the absence of stenosis and in the presence of NFLS, but decreased with FLS.
Severity of coronary stenosis
The ratio of LAD/LCx MBF by fluorescent microspheres progressively decreased with greater levels of stenosis (Fig. 1). The reduction in LAD/LCx ratio was significant at all levels except mild NFLS. The endocardial/epicardial ratio of MBF measured fluorescent microspheres in LAD bed was 1.35 ± 0.33 without stenosis, 1.13 ± 0.30 with mild NFLS, 1.01 ± 0.24 with moderate NFLS, 0.87 ± 0.21 with severe NFLS, 0.78 ± 0.13 with FLS, and 0.81 ± 0.16 with total occlusion, respectively. The endocardial/epicardial ratio of MBF was < 1.0 and the reduction from without stenosis was statistically significant (p < 0.05) in lesions of severe grade or greater.
Regional LV contraction
Figure 1shows the regional LV wall thickening of the risk area at each level of stenosis during vasodilation. Regional wall thickening of the risk area did not decrease at the three levels of NFLS compared with no stenosis. However, a significant reduction in wall thickening was observed with FLS and occlusion.
Visual analysis of stenosis
Real-time images were visually analyzed by two independent reviewers. The reviewers agreed on the absence of perfusion defects during mild NFLS and the presence of perfusion defects during severe NFLS, FLS, and total occlusion in all eight animals. During moderate NFLS, the reviewers agreed on the absence of perfusion defect in six animals and the presence of perfusion defect in one animal. However, the two reviewers disagreed on the presence or absence of perfusion defect in one animal. Of the 40 real-time images that were analyzed in the eight animals, the agreement between the reviewers was 98%.
Figure 2shows changes in end-systolic images after FLASH without stenosis during adenosine-induced vasodilation. Because microbubbles in the myocardium were destroyed by the high-energy frames, both the risk myocardium (LAD) and control (LCx) regions were unopacified just after the FLASH. From the first to fourth cardiac cycles after FLASH, the SI of both regions increased quickly because of prompt replenishment of microbubbles into the myocardium. After five cardiac cycles, the SI of both areas reached a maximum of equal magnitude. The speed of SI rise after FLASH looked to be similar by visual analysis in risk and control areas.
Visual analysis of real-time imaging did not detect a myocardial perfusion abnormality with mild NFLS in any animal. With moderate NFLS, visual analysis of real-time imaging did not detect a perfusion abnormality in six animals (75%). Despite the difficulty in detecting a perfusion abnormality during real-time MCE with moderate NFLS, visual analysis of consecutive end-systolic frames did reveal opacification defects with moderate NFLS lesions (Fig. 3). Figure 3depicts consecutive end-systolic frames obtained from an animal that showed no perfusion defect in real-time imaging. The visual presentation of such abnormalities changed rapidly over time, with subendocardial deficits (arrow heads) disappearing quickly and the size of transmural defects (arrows) becoming progressively smaller with sequential frames after FLASH.
Visual examination was able to detect perfusion abnormalities during real-time imaging in the presence of severe NFLS (Fig. 4), FLS, and total occlusion. Again, analysis of serial end-systolic frames revealed that the perfusion defects were largest in the first post-FLASH frame and progressively decreased thereafter, often being limited to the subendocardial region (arrow heads in Fig. 4). The shrinkage of the perfusion defect with time after FLASH was less for more severe lesions.
Relation of perfusion abnormalities to contractile abnormalities
Myocardial perfusion abnormalities were observed in end-systolic frames of stenoses of moderate grade or greater and during real-time imaging in lesions of severe grade or greater. However, contractile dysfunction (reduced endocardial excursion, wall thickening, and LV dilation) was observed only in the presence of FLS and total occlusion. Thus, perfusion defects were observed under lesser grades of coronary stenosis than were contractile abnormalities.
Myocardial intensity in the presence of graded coronary stenosis
Figure 5shows the myocardial SI-versus-time plots at each level of stenosis from the LAD risk area of the same animal shown in Figures 2–4, and the results of fitting the data to the one-exponential function described in the Methods section. Although the value for bobtained with mild NFLS was not different from normal, it was reduced with moderate NFLS and progressively decreased with greater levels of stenosis. Peak SI obtained at the very end of the curve was not different from baseline with mild or moderate NFLS. However, the value for peak SI with severe NFLS was less than without stenosis and also progressively decreased with greater levels of stenosis.
Figure 6shows averaged values for the parameters of myocardial SI in the risk area derived from curve fitting at each stage in eight dogs. These curves demonstrated that the values both of band of peak SI at the tail end of the curve progressively decreased with greater levels of stenosis.
Comparison of parameters from microbubble refilling curves during graded coronary stenoses
Mean values for the maximal SI reached, the rate of SI rise, and the product of these two parameters at each level of stenosis are presented in Figure 7. Although the peak SI with the three levels of NFLS was decreased compared to normal, the reduction was not significant. The peak SI did significantly decrease with FLS and occlusion. The rate of SI rise (b) demonstrated a significant reduction even with severe NFLS lesions. The reduction in the rate of SI rise with FLS and occlusion was greater than NFLS. Furthermore, the product of peak SI and the rate of SI demonstrated a greater reduction than peak SI and the rate of SI rise with all levels of graded stenosis, and showed a significant reduction even with moderate NFLS.
Correlation between MBF and indexes derived from PPI
The results of linear regression analysis showed that the ratio of MBF in the LAD versus LCx bed correlated with the ratio of peak SI in these regions (r = 0.69, p < 0.0001). The correlation with MBF ratios was closer for b(r = 0.79, p < 0.0001), as was the product of peak SI and b(r = 0.80, p < 0.0001).
Previous studies have reported that MBF can be quantitatively assessed by MCE using ECG gated intermittent triggered imaging (3). This study presents the first data demonstrating the ability of real-time MCE to detect stenosis-induced myocardial perfusion abnormalities without the need for and limitations of ECG-gated triggered imaging, and provides the initial quantitative assessment of the severity of stenosis from the microbubble refilling rate. The data in this study enable the following conclusions in an open-chest dog model. 1) In conjunction with high-energy bubble destruction, real-time MCE enables definition of the time course of restoration of myocardial intensity through refilling with microbubbles. 2) Real-time MCE measurements of the rate of intensity rise (b) and peak intensity are altered in the presence of coronary stenoses. 3) Both band peak intensity decrease progressively as coronary blood flow is incrementally diminished by more severe stenoses, and is significantly reduced with severe NFLS and FLS stenoses. 4) Frame-by-frame analysis of real-time MCE can identify opacification abnormalities associated with moderate NFLS lesions that are not visible when microbubble refilling is complete. 5) Real-time MCE can identify perfusion abnormalities associated with severe NFLS lesions that do not produce contractile dysfunction. 6) Measurements of band the product of btimes peak intensity, normalized as the ratio of values in ischemic (LAD) to normal (LCx) coronary perfusion beds, show a good correlation with a similar ratio of MBF derived from fluorescent microspheres. These data indicate that quantitative analysis of real-time intravenous MCE should be of clinical value in the assessment of coronary artery disease.
Quantitative analysis of intensity-versus-time curves derived using a one-exponential function demonstrates that the peak SI at the termination of the curve, which reflects microvascular blood volume and cross-sectional area, is significantly reduced with FLS and occlusion and insignificantly reduced with NFLS. The rate of SI rise, which is related to microbubble velocity, decreases significantly with severe NFLS, FLS, and occlusion, and shows a trend (p = 0.15) toward a significant reduction with moderate NFLS. The product of peak SI and rate of SI rise that represents volumetric MBF demonstrates the greatest reduction in values at each level of graded stenosis and exhibits a statistically significant change even with moderate NFLS. The product of peak SI and rate of SI rise, as well as the rate of SI rise alone, yielded a closer correlation with fluorescent microsphere-derived MBF measures than the peak plateau SI and will likely be the most valuable criteria for clinical abnormality.
Relationship between myocardial perfusion and LV wall motion abnormalities in real-time MCE
No data exist regarding the relationship between myocardial perfusion assessed by MCE and contractile function. In the present study contractile dysfunction could not be detected during adenosine infusion at any of the three levels of NFLS, which exhibited a reduced ratio of LAD/LCx MBF ranging from 0.88 for mild to 0.63 for severe grades. Dyssynergy could be detected with FLS and coronary occlusion (the ratio of LAD/LCx MBF; 0.50 ± 0.13 and 0.46 ± 0.17, respectively) (Fig. 1). Conversely, real-time MCE was able to record a perfusion defect with severe NFLS. In addition, although subtle, careful observation at every end-systolic cardiac cycle demonstrated small, short-lasting, and often subendocardial opacification defects immediately after FLASH transmission even with moderate NFLS. This difference is likely due to the fact that mild myocardial perfusion abnormalities may not exhibit LV wall dysfunction (7), and suggests that myocardial perfusion abnormalities assessed by MCE may be more sensitive than wall motion abnormalities in detecting stenoses, especially in vasodilator stress echocardiography.
Temporal and spatial changes in opacification defects after bubble destruction
We reviewed every end-systolic frame after FLASH transmission using cineloop processing for moderate NFLS stenoses that did not show an opacification defect in real-time imaging (six of eight animals). In four of the six animals that did not show opacification defect during moderate NFLS, either a subendocardial defect was observed early after FLASH and disappeared later, or a defect that was initially transmural became smaller as the sequence continued. Thus, temporal and spatial changes in the opacification defect were often observed after microbubble destruction. Therefore, careful frame-by-frame observation during the early phase after microbubble destruction may be needed to identify milder noncritical stenosis by visual analysis during real-time imaging.
These visual findings are in agreement with the quantitative results obtained from analysis of MCE refilling curves. Although peak SI values were significantly reduced only with FLS, the microbubble replenishment rate (b) was reduced with moderate and severe NFLS as well (Figs. 6 and 7). Thus, the percent reduction from control for moderate and severe NFLS was 14 ± 25% and 21 ± 13% for peak SI and 31 ± 18% and 63 ± 20% for b.
In the current study, the SI reached at the end of an imaging sequence between two FLASH transmissions was taken as the peak intensity value. Although maximal plateau intensity values were also derived from fitting the time-intensity data to the exponential function, these calculated values often differed from those measured directly. We believe this disparity was due to the profound reduction in rate of rise that occurred with severe reductions of coronary flow and the resultant failure to reach or inability to identify a true plateau intensity. Nevertheless, as is readily apparent from Figure 6, peak intensity was clearly and progressively decreased with stenoses of increasing severity. Because maximal intensity values are a reflection of myocardial blood volume, and therefore microcirculatory cross-sectional area, the decrease observed with incremental reductions of MBF supports the concept that the mechanism by which stenoses produce perfusion defects during coronary vasodilation is a reduced microcirculatory bed, likely due to “derecruitment” or closure of capillaries (8).
The current study was performed in open-chest dogs. Although this has implications for extrapolation to the clinical setting, we believe the principles established in this study remain valid. We also utilized a single contrast agent and dose. The dose of agent was adequate to opacify the imaging field and allow determination of replenishment characteristics. The values obtained also depend on machine settings, such as mechanical index, that determine the degree of microbubble destruction. Although the precise results obtained with other machine settings might differ, these data have established the basis for the relationship of microbubble replenishment rate into myocardium and lesion severity in real-time imaging. Finally, in this protocol we examined only stenoses in an anterior location and did not apply other vasodilator stimuli except adenosine. Because attenuation or lateral dropout were sometimes seen with real-time imaging, similar to that seen with ECG-gated intermittent-triggered imaging, assessment of myocardial perfusion of lateral and posterior wall may be problematic. Other pharmacological stress agents such as dobutamine have a different mechanism of inducing ischemia than does adenosine, so the relationship between myocardial perfusion and LV wall motion abnormalities may differ from that obtained with adenosine infusion.
Consensus is still lacking in regard to the optimal criteria to be applied to MCE images in the evaluation of coronary artery disease. Our experimental data demonstrate that microbubble replenishment rate and peak intensity after bubble destruction, as assessed by real-time MCE imaging, provide excellent parameters of regional microcirculatory flow. These parameters are progressively diminished to an extent similar to MBF in the presence of coronary stenoses. Recent studies have developed methods to assess myocardial perfusion from filling and clearance of contrast on angiogram and have related these measures to mortality risk in acute myocardial infarction (9). The MCE measurements in this study are similarly based upon assessment of blood-flow velocity and may provide microcirculatory data of comparable clinical significance. Although LV dyssynergy may not occur during adenosine infusion even with moderate or severe stenoses, perfusion defects are more readily identified both visually and quantitatively, and disturbed transmural distribution of myocardial perfusion can be identified by real-time MCE. We believe that our findings, although obtained in open-chest canines, support the usefulness of quantitative measures of contrast intensity and careful frame-by-frame observation during the early phase following microbubble destruction for the accurate assessment of myocardial perfusion. Real-time MCE readily conveyed this information in this experimental study and should provide comparable utility in the clinical setting.
☆ Dr. DeMaria received financial support from ATL Ultrasound, Inc., Bothell, Washington; Mallinckrodt, Inc., St. Louis, Missouri; and Molecular Biosystems, Inc., San Diego, California.
- flow-limiting stenosis
- left anterior descending coronary artery
- left circumflex coronary artery
- left ventricular
- myocardial blood flow
- myocardial contrast echocardiography
- non-flow-limiting stenosis
- power pulse inversion
- regions of interest
- signal intensity
- wall thickening
- Received April 17, 2000.
- Revision received September 7, 2000.
- Accepted September 14, 2000.
- American College of Cardiology
- Wei K,
- Jayaweera A.R,
- Firoozan S,
- Linka A,
- Skyba D.M,
- Kaul S
- Skyba D,
- Averkiou M,
- Loflin C,
- et al.
- Glenny R.W,
- Bernard S,
- Brinkley M
- Vatner S.F
- Jayaweera A.R,
- Wei K,
- Coggins M,
- Bin J.P,
- Goodman C,
- Kaul S
- Gibson C.M,
- Cannon C.P,
- Murphy S.A,
- et al.