Myocardial Contrast Echocardiography

HomeCirculationVol. 96, No. 10Myocardial Contrast Echocardiography Free AccessResearch ArticleDownload EPUBAboutView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissions ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toFree AccessResearch ArticleDownload EPUBMyocardial Contrast Echocardiography 15 Years of Research and Development Sanjiv Kaul Sanjiv KaulSanjiv Kaul From the Cardiovascular Division, University of Virginia School of Medicine, Charlottesville, Va. Originally published18 Nov 1997https://doi.org/10.1161/01.CIR.96.10.3745Circulation. 1997;96:3745–3760"An untroubled mind, no longer seeking to consider what is right and what is wrong; A mind beyond judgements, watches and understands." The Buddha (translated from the Dhammapada)The purpose of this article is to describe our personal experience in translating observations made in the experimental laboratory using MCE into the clinical setting. It is not intended to be an exhaustive review of MCE, for which readers are referred elsewhere.123 The work of others in MCE and related subjects will be mentioned only when it has influenced our own work. Our bench-to-bedside experience with MCE over the past 15 years will be discussed under these six broad categories: (a) technical issues; (b) AMI, (c) detection of CAD, (d) applications in the operating room, (e) quantification of myocardial perfusion, and (f) work in progress.Technical IssuesHistorically, it has not been possible to directly assess myocardial perfusion with echocardiography. Its clinical focus has involved the evaluation of cardiac chamber size and function, valve morphology and kinetics, pericardial space and great vessels, and intracavitary blood flow velocities. Yet, echocardiography is highly suited for the evaluation of myocardial perfusion for the following reasons: (a) It has very good spatial resolution (<1 mm in the axial direction), which is far superior to that offered by SPECT and positron emission tomography, although not as good as magnetic resonance imaging and ultrafast cine computed tomography; (b) its temporal resolution is excellent (30 to 120 Hz) and exceeds that of other commonly used imaging technologies; (c) for an imaging modality, it is inexpensive and has low overhead costs; (d) it is an integral tool in the day-to-day activities of clinical cardiologists, who can obtain advanced training in its use without needing to learn an entirely new technology.The study of myocardial perfusion with echocardiography involves the intravascular injection of tracers that can scatter ultrasound.4 Because the compressibility of a particle is the most important determinant of its scattering cross section,5 microbubbles are the most ideal tracers. We have used these bubbles to assess myocardial perfusion in both the spatial and the temporal domain. We have shown that their relative concentrations in different regions of the myocardium reflect the relative MBV in those regions, which is the volume of blood within the myocardial microvasculature.6 Since a large portion of the myocardial microvasculature consists of capillaries,7 MCE actually offers the potential to evaluate tissue perfusion at the level where oxygen transfer to the myocytes occurs.8 MCE can, therefore, provide insights into the functional status of the myocardial microvasculature. The spatial resolution of echocardiography also allows the evaluation of the transmural distribution of abnormal perfusion.Our initial observations with MCE were made using direct coronary injections of a mixture of hand-agitated Renografin-76 and saline4 in dogs.91011 These microbubbles provided adequate information on the spatial distribution of myocardial perfusion (Fig 1), but their large size (>10 μm) precluded their clinical use. Subsequently, with the development of sonication,12 small bubbles were formed when air in liquid media was exposed to high-energy ultrasound. We found that microbubble solutions produced by the sonication of Renografin-76 were safe in humans when injected directly into the coronary arteries.13 Because bubbles made by sonication of most liquid media have a very short half-life,12 they are unable to opacify the LV cavity after a venous injection. The sonication of 5% human serum albumin solutions, however, resulted in microbubbles with a thin (15 nm) shell consisting of denatured albumin formed by the heat from the sonication process.14 These bubbles were relatively stable and had a long shelf life. They cleared the pulmonary microcirculation after venous injection because of their small size (mean of 4.3 μm) and opacified the LV cavity.15 This approach was successfully applied to the commercial production of Albunex, the first echocardiographic contrast agent approved by the US Food and Drug Administration for LV cavity opacification from a venous injection.15Because of our interest in myocardial perfusion, we tested the effect of direct intracoronary injections of sonicated albumin microbubbles on left heart and systemic hemodynamics, as well as MBF. Unlike iodinated contrast agents,16 albumin is iso-osmolar and possesses no calcium chelating properties. Unlike these agents, therefore, it does not produce hemodynamic effects after direct intracoronary injections. Importantly, the introduction of microbubbles in albumin does not cause any additional hemodynamic effects.1718 This effect of albumin microbubbles is secondary to their intravascular rheology, which we found to be similar to that of red blood cells.19 As expected, no obstruction of the microvasculature was observed.Like other first-generation contrast agents, Albunex contains air, which is highly diffusible and leaks out of the shell after the bubbles are exposed to blood.20 Because the scattering cross section of a bubble is related to the 6th power of its radius,2 even a small reduction in its size results in a large decrement in its scattering cross section. Thus, this agent does not always result in LV cavity opacification after a venous injection,15 particularly when the duration of its contact with blood is longer than usual, such as in low-output states. Second-generation bubbles contain high-molecular-weight gases that are not easily diffusible.20 These agents invariably produce LV cavity opacification. Most of these agents consist of preformed bubbles, and we have found no adverse effects on systemic hemodynamics, resting MBF, or pulmonary gas exchange with their use.212223 The venous agents tested in our laboratory are listed in the Table.For many years, we used regular echocardiographic equipment with high-frequency transducers for our experimental work. We obtained adequate myocardial opacification with intracoronary, aortic root, and left atrial injections of microbubbles. We could also opacify the myocardium from right heart injections of Albunex using highly concentrated solutions not suitable for clinical use.24 When the second-generation contrast agents were injected intravenously at the recommended doses, myocardial opacification was not visually striking, although changes in myocardial video intensity could be quantified. Since <5% of the stroke volume enters the coronary circulation at rest, we postulated that the primary reason for suboptimal myocardial opacification was the poor signal-to-noise ratio.We therefore developed image-processing algorithms to display and measure video intensity from images produced by venous injections of ultrasound contrast agents. These algorithms are particularly helpful because the human eye is very poor at differentiating levels of gray but very adept at discriminating between hues of color. The following steps are involved in our approach to image processing: separate alignment and averaging of a few (3 to 6) pre-contrast and contrast enhanced images; realignment of the averaged pre-contrast and contrast enhanced images before the former are subtracted from the latter; and rescaling of the digitally subtracted image over a dynamic range of 256 gray-level values before colors are assigned to these values. The highest gray level within the myocardium is assigned the color white with progressively lower gray levels assigned colors of yellow, orange, and red. Gray-level values of ≤10 are considered to represent noise and are not assigned a color.25 Relative video intensity measurements can then be made directly from these color-coded images.About the same time that the new ultrasound contrast agents were being developed, nonlinear scattering properties of microbubbles were reported.26 Because ultrasound creates bands of compression and rarefaction of the medium through which it travels, bubbles exposed to ultrasound also alternately contract and expand equally (or linearly). If ultrasound of a specific frequency (resonant frequency, which depends on bubble size and shell thickness, among other variables) is used, the oscillations of the bubbles could become nonlinear, and result in the production of signals that contain not only the frequency to which the bubbles were originally exposed (the fundamental frequency) but also harmonics of those frequencies.2627 Since harmonic signals emanate primarily from bubbles, the signal-to-noise ratio is much higher.28By serendipity, the resonant frequencies for bubbles small enough to transit the pulmonary capillaries (<6 to 7 μm) are in the range of frequencies used clinically in adults (<3 MHz). To take advantage of the improved signal-to-noise ratio during harmonic imaging, transducers have been designed to transmit ultrasound at a particular frequency (say, 1.67 MHz) and receive at twice this frequency (in this case, 3.3 MHz). Even with harmonic imaging, however, the increase in myocardial opacification during venous injections of microbubbles was minimal. The value of harmonic imaging was truly recognized, however, when an increase in myocardial opacification from venous injection of microbubbles was reported from a chance observation when ultrasound was resumed after it had been accidentally suspended for a short period.29We demonstrated that the lack of contrast effect during real-time (continuous) imaging was related to bubble destruction caused by ultrasound.30 Fig 2 illustrates video intensity produced by a second-generation microbubble (FS-069, which is similar to Albunex, except that instead of air alone it also contains perfluoropropane) suspended in saline, constantly mixed, and imaged continuously. The video intensity gradually declined (filled squares) and Coulter counter measurements demonstrated a marked decrease in both microbubble size and concentration. When ultrasound was paused for 30 seconds (arrow) in a similar experiment (open squares), no change in video intensity was noted during that pause immediately after resumption of ultrasound. These data indicate that no bubbles were destroyed during the cessation of ultrasound.The implications of these findings in the in vivo setting are as follows: during a pause of ultrasound transmission, microbubbles in the pulmonary circulation and LV cavity, which have not been destroyed by ultrasound, enter the coronary microcirculation. Pausing ultrasound transmission results in greater opacification by permitting the myocardial sample being insonified by the ultrasound beam (whose thickness is approximately 0.5 cm) to become replenished with these new bubbles. If flow is high (in centimeters times seconds−1), then replenishment will occur even when imaging is performed in real time. This is the reason why larger myocardial vessels, such as septal perforators, are seen with continuous harmonic imaging while the rest of the myocardium does not show any opacification. If flow is slow (in millimeters times seconds−1), as occurs in capillaries (average resting flow <0.1 cm · s−1), then beam replenishment will take several cycles to complete. Imaging in real time will not produce much opacification because bubbles are destroyed very soon after they reenter the beam. At normal resting flows, therefore, the best myocardial opacification is seen when imaging is performed once every few (5 to 8) cardiac cycles, when the entire beam is replenished by microbubbles.The exact mechanism of bubble destruction is not known. It could occur from nonlinear oscillations produced by a resonant frequency or simply by the acoustic power of ultrasound itself. Since acoustic emissions resulting from bubble destruction contain many frequencies, a signal will be recorded when imaging is performed at any of these frequencies, including the harmonic frequency. Irrespective of the mechanism behind the production of harmonic signals from microbubbles, since normal myocardium has little harmonic properties, the signal-to-noise ratio during MCE is significantly greater with harmonic compared to fundamental imaging (where the signal includes backscatter from both bubbles and tissue).28 Furthermore, because ultrasound causes microbubble destruction, decreasing their exposure to ultrasound with intermittent imaging results in better myocardial opacification as discussed above.29 Consequently, intermittent harmonic imaging is currently the best approach for obtaining optimal myocardial opacification after venous injection of microbubbles.2930 It is also likely that continuous infusion of microbubbles during intermittent harmonic imaging will prove to be the ideal method for assessing myocardial perfusion with MCE in the clinical setting.Acute Myocardial InfarctionOur first applications of MCE were in the setting of AMI where a substantial portion of the microvasculature is hypoperfused. We showed that MCE could be used in vivo to accurately define risk area (Fig 1) in real time, something that was heretofore not possible. Our interest at that time was to assess serial changes in myocardial perfusion during coronary occlusion and to determine the effect of the duration of coronary occlusion on the risk area/infarct size ratios.31 We also wanted to understand the influence of the size of the risk area on clinical measurements such as regional and global LV function, as well as systemic hemodynamics in the setting of AMI. We found that because of compensatory mechanisms, these variables did not become abnormal until risk area involved a very sizeable portion of the LV.31It was previously demonstrated in dogs that within 30 to 45 minutes after coronary occlusion, necrosis is initiated in the subendocardium and progresses transmurally over time.3233 It was shown that the ultimate extent of necrosis was related to both the duration of occlusion and the residual MBF within the risk area.3233 The former message was heard clearly in the clinical world. Accordingly, the duration between the onset of symptoms and the restoration of infarct-related artery patency became of utmost importance, forming the basis for emergent reperfusion during AMI. Unfortunately, the second message was lost—that is, necrosis would not be severe if adequate residual MBF was present within the risk area.3233We demonstrated the ability of MCE to measure spatial changes in collateral perfusion caused by altering the collateral driving pressure.34 We also noted myocardial opacification in regions remote from those receiving grafts in patients undergoing CABG surgery when microbubbles were injected into the cross-clamped aortic root.35 This finding implied the presence of collateral perfusion. Additionally, we found extensive collateral perfusion on MCE in patients with both recent36 and old37 infarction in the presence of an occluded infarct-related artery. Interestingly, a poor correlation was noted between the extent of myocardial opacification from collateral vessels and the angiographic collateral score.353637On the basis of these observations, we hypothesized that, although collateral-derived MBF may not be adequate for normal myocardial systolic function distal to an occluded vessel, it may be enough to maintain viability for prolonged periods after coronary occlusion. We argued that providing anterograde coronary flow to such myocardium would result in improvement of regional function over time. We studied patients with recent (2 days to 5 weeks) AMI and an occluded infarct-related artery.38 These patients were referred for definition of their coronary anatomy and not because of postinfarction angina. We injected microbubbles directly into the left main and the right coronary arteries. We then attempted to open the infarct-related artery by angioplasty with success in about 80% of cases. Fig 3 illustrates a parasternal LV short-axis view in a patient with a recent inferior AMI who exhibited moderate hypokinesia of that region. Fig 3A shows opacification of the entire LV myocardium from a left main injection prior to angioplasty. After successful angioplasty of the occluded RCA, microbubbles were injected directly into it to define its vascular bed (B). It is obvious that this region was supplied by left-to-right collaterals during RCA occlusion (A). Coronary angiography showed poor collaterals in this patient. Assessment of collateral perfusion is more accurate with MCE than with coronary angiography because the latter can define only vessels >100 μm in size, while most collateral channels are much smaller.39In this study, nearly 80% of the patients exhibiting adequate collateral perfusion within the infarct bed by MCE demonstrated an improvement in regional systolic function 1 month after anterograde flow was restored.38 Importantly, as long as there was adequate collateral perfusion within the infarct zone (involving >50% of the bed), function improved even when anterograde flow was established days to weeks after AMI.38 These data indicate that, in addition to the duration of coronary occlusion, the spatial extent of residual myocardial perfusion is also a major determinant of infarct size. Thus, myocardial regions that suffer an infarction but are not immediately reperfused are still likely to be viable if they have adequate collateral flow. Conversely, if not revascularized, these regions are likely to undergo repeated episodes of ischemia with its multiple sequelae. We believe that our findings provide the physiological basis for the "open artery hypothesis" and explain why patients with open infarct-related arteries post-AMI have a better prognosis than those with occluded arteries.It was also known for many years that despite reflow, adequate amounts of tissue perfusion may not be achieved (the "low reflow" or the "no reflow" phenomenon) because of microvascular disruption, plugging by debris, or myocardial edema.4041 This phenomenon is localized only within the infarct and indicates the presence of severe necrosis, which was successfully demonstrated in a canine model of coronary occlusion and reperfusion using MCE.42 A great deal of interest was, however, generated when this same finding was first described in patients with AMI who underwent MCE in the cardiac catheterization laboratory immediately after patency of the infarct-related artery was restored.43 In one fourth of the patients, tissue perfusion was not seen despite good angiographic results. It was shown that patients exhibiting the "no reflow" phenomenon had worse regional and global function 1 month later compared to those who did not show this phenomenon.43We were invited to write an editorial in response to the article. We emphasized the importance of assessing microvascular rather than epicardial coronary flow in patients after reperfusion, since the presence of "flow" in the epicardial coronary artery was not indicative of actual nutrient perfusion.44 We also raised the concern that infarct size may be underestimated by MCE immediately after reflow because of the possibility of reactive hyperemia. We suggested that the best time to assess microvascular perfusion may be after reactive hyperemia had abated (12 to 24 hours later).44 It was previously shown that despite reactive hyperemia, microvascular reserve within the infarct bed was diminished consequent to functional if not anatomic damage to the microvasculature.4145 We therefore indicated that if it were necessary to determine infarct size immediately after reperfusion, MCE should be performed in the presence of a coronary vasodilator.44 We postulated that since the microvasculature in noninfarcted tissue is normal, the response to hyperemia in the normal bed would be greater than that within the infarct bed. Consequently, despite an increase in the absolute MBF in the infarct bed, MBF in that bed would appear to be less compared with the normal bed during exogenous hyperemia.To prove these hypotheses, we studied dogs undergoing 2 to 6 hours of LAD occlusion followed by 3 hours of reperfusion.46 Fig 4 illustrates radiolabeled microsphere–derived MBF data obtained from the infarct bed during reflow after the values were normalized to those of the remote noninfarcted bed. Wide variations and fluctuations are seen in MBF to the infarct bed in individual dogs during reflow (A). Because of reactive hyperemia, the average MBF within the infarct bed is similar to that in the normal bed despite considerable microvascular damage in the infarct bed (B). Importantly, abnormal MBF reserve within the infarct bed (compared with the normal bed) is unmasked by infusion of dipyridamole. Fig 5 illustrates color-coded MCE images at different points in time after establishment of reflow. Figs 5A & 5B (45 minutes and 3 hours after reperfusion) shows small areas of "no reflow," while in reality the infarct size is much larger (D). In the presence of dipyridamole, however, since all gray-level values are scaled to the highest value (in the normal bed in this case), the perfusion defect within the infarct bed appears more extensive (C) and accurately reflects infarct size.46Abnormal MBF reserve in the infarct bed in the above study was measured 3 hours after reflow. We have demonstrated that similar information can be obtained immediately (15 minutes in this case) after reflow.47 We also showed that although basal MBF to the infarct bed varies widely between 15 minutes and 3 hours after reflow, the abnormality in microvascular MBF reserve is remarkably consistent at both periods.47 Thus, the assessment of microvascular reserve any time after reflow should provide an accurate assessment of infarct size irrespective of the degree of reactive hyperemia and the unpredictable fluctuations in basal MBF.In the last two studies discussed above, reflow was established in the absence of any residual stenosis, which is usually not the case in patients with AMI unless angioplasty is also performed in conjunction with or instead of thrombolysis. There are two confounding issues when a stenosis is present. First, if it is critical and does not permit reactive hyperemia to occur, then the infarct size could be accurately measured immediately after reflow is established. Although such a situation may occasionally exist, the residual stenosis is usually not critical after the thrombus has resolved and, therefore, various degrees of reactive hyperemia may still be present. Second, if exogenous hyperemia is induced in the presence of a stenosis, then the entire infarct bed should demonstrate relative hypoperfusion compared with beds not supplied by stenotic vessels or by vessels with less stenosis. MCE during exogenous hyperemia could, therefore, conceivably overestimate infarct size.We have demonstrated that although perfusion is reduced to the entire bed distal to a stenosis, there is still marked heterogeneity in the microvascular reserve within the infarct bed.48 The area with the most diminished reserve corresponds to the necrotic area and can be easily discriminated from other areas within the infarct bed. Fig 6A depicts an MCE perfusion defect in the LCx bed immediately after reflow, prior to an infusion of dobutamine, where infarct size (D) is underestimated. After infusion of dobutamine, infarct size is accurately estimated (B). Dobutamine causes coronary hyperemia by increasing myocardial oxygen consumption. Fig 6C illustrates an MCE perfusion defect in the presence of dobutamine, after the placement of a stenosis. Although the intensity of colors is less in the entire LCx bed because of reduced reserve in that bed caused by the stenosis, infarcted areas can still be separated from noninfarcted ones because of the marked disparity in MBF reserve within these regions in the same bed.We have shown in humans that if MCE is performed after reactive hyperemia has abated (at least 1 day after establishment of reflow), the extent of adequate microvascular perfusion within the infarct zone indicates the extent of myocellular integrity.4950 Fig 7 shows MCE images from a patient with an anteroapical AMI who had previously received thrombolytic therapy. Angiography of the large wraparound LAD showed "excellent flow." Microbubbles were then injected into the left main coronary artery. No opacification is seen in the apex, and patchy opacification is noted in the middle and apical portions of the anterior interventricular septum. Only the basal interventricular septum demonstrates adequate perfusion in this apical 4-chamber view. In this study of patients who had patent infarct-related arteries without flow-limiting stenoses at rest (<80% luminal diameter narrowing), regional function 1 month later correlated inversely with the spatial extent of microvascular perfusion at the time of the original examination.4950 Dysfunctional regions with extensive myocardial opacification showed near-normal function 1 month later, while those with no opacification or that limited to very small regions showed the most dysfunction. Areas where the spatial extent of microvascular perfusion was intermediate showed intermediate function 1 month later. Thus, MCE provides optimal assessment of viability in patients with AMI undergoing reperfusion therapy after hyperemia has abated.As described above, our early experience with MCE in both the experimental and clinical settings involved direct coronary arterial, aortic, or left atrial51 injections of microbubbles. With the advent of second-generation contrast agents and intermittent harmonic imaging, it has become possible to obtain similar results in animals and humans using venous injections of microbubbles. Fig 8 illustrates MCE images obtained in a dog undergoing LCx occlusion and reperfusion where 1 mL of FS-069 was injected intravenously. Data were acquired using intermittent (once every systole) harmonic imaging. Fig 8A depicts the baseline image prior to coronary occlusion. Other than some posterior wall attenuation caused by the presence of contrast in the LV cavity, myocardial opacification is homogeneous. Fig 8B shows an image after proximal LCx occlusion where a large transmural defect is seen, indicating the location and spatial extent of the risk area. Following reflow and during dipyridamole infusion, injection of microbubbles reveals opacification of the previously occluded bed (C). The perfusion pattern, however, is not homogeneous, with very little opacification seen in the central portion of the bed. The location and topography of this defect closely mimic that of the actual infarct (D).The clinical implications of these findings are obvious, particularly in patients presenting to the emergency department. Only one third of those with ongoing AMI have diagnostic ECGs at the time of presentation to the emergency department.5253 The majority of patients admitted to a hospital bed do not have AMI and may not even have CAD. Defining the presence or absence of a risk area could be very useful in such patients. The size of the risk area could also be important in determining management strategies. Documentation of normal myocardial perfusion and normal microvascular reserve could rule out not only acute ischemia but also CAD. The success of attempted reperfusion could be documented as could the residual infarct size, which could be used for risk stratification and long-term management.Although we have not yet received US Food and Drug Administration approval to study patients with AMI, we have been able to study those with chronic CAD, including some with old infarction, using the same techniques as those in the animal studies discussed above. One such study was performed in the United Kingdom in collaboration with colleagues at the Northwick Park Hospital in Harrow.54 Fig 9 illustrates examples of resting MCE perfusion defects from 2 patients, which correspond to those on resting 99mTc-sestamibi-SPECT. Whereas these data are preliminary, they are very encouraging and indicate that at least post-AMI, MCE has the same potential to assess myocardial viability as other commonly used imaging modalities.54Detection of CADDespite the presence of coronary stenoses, resting MBF is normal in the majority of stable patients with CAD who have not had a previous myocardial infarction. MBF is maintained distal to a stenosis by autoregulation. As the stenosis severity increases, <300-μm arterioles distal to it dilate in order to maintain resting MBF as close to normal as possible.55 When autoregulation is exhausted (usually at ≥85% luminal diameter narrowing), resting MBF begins to decline, particularly if collateral flow is low or absent. A sudden decrease in resting MBF can manifest as an acute ischemic syndrome, while a more gradual decrease can result in congestive heart failure with or without chest pain. In either case, resting perfusion and function are abnormal.For the majority of patients with CAD (those with <85% luminal diameter narrowing of the coronary arteries), resting perfusion is normal and the detection of CAD depends on unmasking abnormal MBF reserve within regions supplied by stenotic vessels. Since microvessels in these regions have already