|Home | About | Journals | Submit | Contact Us | Français|
Pathology in the heart can be examined at several scales, ranging from the molecular to the macroscopic. Traditionally, fluorescence-based techniques such as flow cytometry have been used to study the myocardium at the molecular, cellular, and microscopic levels. Recent advances in magnetic resonance imaging (MRI), however, have made it possible to image certain cellular and molecular events in the myocardium noninvasively in vivo. In addition, diffusion MRI has been used to image myocardial fiber architecture and microstructure in the intact heart. Diffusion MRI tractography, in particular, is providing novel insights into myocardial microsctructure in both health and disease. Recent developments have also been made in fluorescence imaging, making it possible to image fluorescent probes in the heart of small animals non-invasively in vivo. Moreover, techniques have been developed to perform in vivo fluorescence tomography of the mouse heart. These advances in MRI and fluorescence imaging allow events in the myocardium to be imaged at several scales linking molecular changes to alterations in microstructure and microstructural changes to gross function. A complete and integrated picture of pathophysiology in the myocardium is thus obtained. This multiscale approach has the potential to be of significant value not only in preclinical research but, ultimately, in the clinical arena as well.
Characterizing changes in the myocardium often requires evaluation across multiple spatial scales. Recently developed molecular imaging techniques have and are continuing to provide insights into cellular and subcellular aspects of cardiac disease [1, 2]. New techniques are also being developed to measure the microstructural organization of the myocardium and elucidate the biological mechanisms that link cellular and whole-organ pathology . At the other end of the scale, macroscopic imaging methods to acquire global information concerning overall organ health and function continue to be refined . In the current article, we review recent advances in molecular and microstructural imaging of the myocardium and focus particularly on the use of fluorescence molecular tomography (FMT) and magnetic resonance imaging (MRI).
MRI has been traditionally used to image macroscopic phenomena such as contraction, flow, and infarction in the myocardium . During the last decade, however, several molecular MRI contrast agents have been developed to image apoptosis, necrosis, inflammation, peroxidase activity, and angiogenesis in the myocardium [6–8]. While the principal challenge in molecular MRI has been the development of novel imaging agents , the principal breakthroughs in fluorescence imaging have been in probe chemistry as well as hardware and image reconstruction algorithms . The development of near-infrared fluorochromes has allowed structures 3 to 4 cm deep to be imaged [11, 12]. Tomographic systems with source-detector configurations similar to those of X-ray computed tomography have been developed , and reconstruction algorithms which account for the scattering of light by tissue have been introduced [11, 14]. Thus, molecular and cellular events in the myocardium of small animals can now be imaged noninvasively with FMT.
Microstructural imaging of the myocardium is most frequently performed with diffusion-encoded MRI . This allows the orientation of myofiber tracts and sheets to be resolved non-destructively in three dimensions. Several diffusion-encoding techniques have been described, and the reader interested in a technical description of these is referred to several recent reviews [15–17]. In this article, we will focus on microstructural imaging with the recently developed diffusion spectrum MRI tractography (DSI) technique. This technique allows myofiber tracts to be resolved and provides a readout of myocardial organization between that of the cell and the whole organ . Both MRI and fluorescence imaging have thus evolved from their traditional niches and now provide the opportunity for multiscale imaging of cells, fibers, and physiology in the myocardium.
Molecular MRI of the myocardium can be performed with magnetic nanoparticles (MNPs) or gadolinium-based (Gd) agents. MNPs, such as cross-linked iron oxide (CLIO), are small (less than 50 nm in size), consist of a superparamagnetic iron oxide core, have high magnetic relaxivities, are highly stable, and are coated with a biologically inert material such as dextran . Several formulations of MNPs circulate in the blood stream for an extended period, making them ideal for imaging of the cardiovascular system. Unmodified MNPs have a circulatory half-life of approximately 12 h in mice, but conjugation of a ligand to the particles can decrease the half-life to less than 3 h [7, 19], an advantage as acute processes within the myocardium can be imaged with less background signal. In acute injury, the capillary membrane becomes hyperpermeable, allowing MNPs to reach the interstitial space rapidly in large amounts. MNPs are able to cross normal capillary membranes, although this process can take several hours. Once in the interstitial space, the inert nature of certain MNPs allows them to avoid non-specific binding and thus bind specifically to the target of interest on the cardiomyocyte surface. Once bound to the cell surface, these targeted MNPs are internalized and aggregate in lysosomes, forming a nano-assembly of particles with a higher transverse relaxivity (r2*) and hence higher detectability .
Gd-based molecular MRI probes for myocardial imaging can be either small chelates or large nanoparticulate constructs (i.e., micelles, liposomes, quantum dots) . Large constructs can be used to image endothelial targets in the myocardium but cannot easily penetrate into the interstitial space. In addition, these large constructs are frequently taken up non-specifically in the complex environment of the interstitial space and are thus not suited to targeted molecular MRI of the interstitium or the cardiomyocyte surface. Small Gd chelates do not suffer from these issues but have far lower sensitivities (micro-molar vs. nanomolar) and can thus only be used to image highly expressed targets in the myocardium such as collagen .
The largest experience with molecular MRI in the myocardium has been in the imaging of apoptosis. An apoptosis sensing MNP called annexin (Anx)CLIO-Cy5.5 has been developed which provides the ability to visualize the probe with both MRI and fluorescence. The Anx portion of the probe binds to phosphatidylserine, a phospholipid component that is usually kept on the inside of the cell membrane but becomes exposed on the cell surface once apoptosis has begun. The CLIO portion of the MNP provides the MR readout while the Cy5.5 fluorochrome allows for the localization of probe at a microscopic level with near-infrared fluorescence (NIRF). AnxCLIO-Cy5.5 has been able to image cardiomyocyte apoptosis in both ischemic injury and heart failure [7–9]. The extent of apoptosis and hence the uptake of the agent in acute ischemia was influenced significantly by the severity of injury (Fig. 1). In mild–moderate injury, AnxCLIO-Cy5.5 uptake was confined to the midmyocardium. In mice with severe injury, the uptake of the agent was more transmural. NIRF microscopy confirmed that the probe was primarily bound to the cell surface of morphologically intact cardiomyocytes (Fig. 1). This suggests that CM apoptosis begins in the midmyocardium and spreads transmurally as the severity of injury increases. Of note, the imaging in this study was performed 4 h after ischemic injury, during which apoptosis is maximal and well before the myocardium becomes infiltrated with inflammatory cells (see below).
The myocardium becomes heavily infiltrated with proteolytic macrophages 24 to 72 h after an infarction . MNPs are actively engulfed by macrophages and can be used to image cardiac inflammation associated with ischemic heart disease. Both conventional gradient-echo techniques and off-resonance MRI techniques have been used to image macrophage infiltration 72 to 96 h after myocardial infarction [10, 24]. MNPs have high transverse relaxivity, significantly reducing the intensity of the MRI signal in T2*-weighted gradient-echo sequences (Fig. 2). Thus, macrophages that have taken up MNPs produce signal hypointensity. A positive-contrast off-resonance technique has been tested in the identical infarct model (Fig. 2), but this technique was less sensitive and did not display the same linear relationship seen with the conventional gradient-echo approach . Regardless, off-resonance and other positive-contrast approaches demonstrate the range and flexibility of contrast mechanisms available to molecular MRI.
Macrophages infiltrating infarcted myocardium secrete numerous degradative and cytotoxic enzymes including myeloperoxidase (MPO). In the presence of MPO, serotonin is oxidized, causing a Gd-serotonin chelate to form dimers and oligomers with higher relaxivity . This change in magnetic relaxivity can be exploited to image MPO activity in infarcted myocardium . Activation of the probe causes an increase in r1 and, hence, signal hyperintensity of T1-weighted images. Knockout mice that were either heterozygous or homozygous for the MPO gene, however, did not show the same signal changes (Fig. 2d–f). This result confirmed the specificity of the contrast agent for MPO and demonstrated the utility of MRI for imaging enzymatic activity in vivo.
Delayed Gd enhancement is frequently used to assess myocardial viability. The cellular visualization of Gd distribution in infarcted myocardium, however, has proven difficult. The conjugation of conventional fluorochromes to small Gd chelates usually changes the pharmacokinetics of the probe significantly. However, the recent development of a novel fluorescent small Gd chelate (gadolinium diethylenetriamine pentaacetic acid nitrobenzoxadiazole, Gd-DTPA-NBD) has produced a Gd chelate that remains suitable for delayed enhancement imaging of the myocardium (Fig. 3a–c) . NBD is small, carries no charge, and has only a small protein binding potential. In addition, Gd-DTPA-NBD can be detected by MRI, fluorescence imaging, and immunohistochemistry. Initial testing of the probe was performed in a mouse model of myocardial infarction. Gd-DTPA-NBD showed the typical pattern of delayed enhancement by MRI, which corresponded well to fluorescence reflectance imaging of the agent. Immunohistochemistry confirmed that the chelate accumulated only in regions of necrosis and cell rupture, where the volume fraction of the extracellular space had increased .
Angiogenesis is part of the early healing response of the myocardium to infarction. Recently, the development of a cyclic Asn-Gly-Arg (cNGR)-labeled paramagnetic quantum dot (pQDs) was reported and shown to bind to CD13, a zinc-dependent aminopeptidase that is upregulated during angiogenesis . Infarcted mice injected with this agent 7 days after coronary artery ligation showed strong negative contrast in the injured myocardium (Fig. 3d). Ex vivo validation with two-photon laser scanning microscopy revealed a strong colocalization of the cNGR-pQDs probe with vascular endothelial cells. Current research is being done to couple this ligand to MNPs in order to develop a probe that may be more clinically applicable.
In addition to angiogenesis, the healing infarct is characterized by collagen deposition to form a scar. Healed infarcts are thus characterized by collagen-rich scars, which oppose expansion and rupture of the infarct . To quantify the development of fibrosis, a Gd-based MR contrast agent with a peptide specific for type I collagen has recently been developed . After injection of the probe, serial imaging of chronically infarcted mice was performed using a T1-weighted inversion recovery sequence. This revealed that the control probe was characterized by delayed washout and enhancement but, nevertheless, had fully washed out of the infarct within 40 min of injection. The active collagen binding probe, however, was retained in the infarct and allowed collagen content of the infarct scar to be clearly visualized (Fig. 3e) .
Several approaches are being developed to image more than one molecular imaging agent simultaneously. These include the use of fluorine MRI and novel endogenous contrast mechanisms [30–33]. It was also recently shown that cardiomyocyte necrosis and apoptosis can be imaged simultaneously in vivo using a dual-contrast approach . The probe AnxCLIO-Cy5.5 is taken up in both apoptotic cells (which express phosphatidylserine on the outer membrane surface) and necrotic cells, making it a marker for composite cell death (Fig. 4). The dual use of delayed enhancement imaging with Gd-DTPA-NBD, however, allowed the uptake of AnxCLIO-Cy5.5 due to apoptosis and necrosis to be differentiated . Areas of myocardium showing uptake of AnxCLIO-Cy5.5 were classified as apoptotic, while those showing the uptake of both probes were labeled as necrotic. Using this approach in a mouse model, it was shown that large areas of apoptotic (but potentially viable) myocardium are present in the mid-myocardium within the first few hours of ischemia–reperfusion injury .
The attributes of fluorescence imaging are highly complementary to those of MRI. The sensitivity and multispectral nature of fluorescence imaging have made it an indispensible tool for in vitro imaging. Current schemes to perform fluorescence imaging in vivo, however, are usually invasive and limited to surface imaging. For fluorescence imaging of organs deep within the body to be performed noninvasively, light at both the excitation and emission wavelengths must be able to penetrate tissue without being excessively attenuated. Thus, organic fluorochromes in the near-infrared spectrum must be used for noninvasive fluorescence imaging of deep structures . Unlike X-ray radiation that is only absorbed, the propagation of light in tissue is diffuse in nature and is a function of both scattering and absorption [12, 13].
Some of the first noninvasive in vivo fluorescence imaging of the myocardium focused on the detection of macrophage infiltration in mice with healing myocardial infarctions . Mice injected with CLIO-Cy5.5, a cross-linked iron oxide probe with an attached Cy5.5 fluorochrome, showed robust nanoparticle accumulation 96 h after infarction (Fig. 5). Fluorescence microscopy and immunohistochemistry confirmed that the probe was endocytosed by macrophages in the injured tissue. While planar fluorescence imaging is of major value, it is not depth-resolved. This creates challenges in accurately quantifying the fluorescence signal and ensuring that it is arising solely from the organ of interest . FMT of near-infrared fluorochromes in the heart overcomes some of these limitations and has been actively pursued.
Several generations of FMT systems have been developed [10, 13]. FMT of the myocardium was first performed on second-generation systems that required the mice to be immersed in an intralipid solution. Current third-generation systems employ 360° ring detector configurations and no longer require the use of optical matching media. FMT of the myocardium was also first performed in infarcted mice, injected with the CLIO-Cy5.5 nanoparticle, and imaged 96 h after infarction (Fig. 6) . Analogous to the signal seen in T2*-weighted MRI, the FMT signal increased linearly with the dose of CLIO-Cy5.5 injected. In a subsequent study, FMT was used to image a near-infrared probe that can be activated by proteases, principally cathepsins, in healing infarcts . Healing infarcts attract a large number of activated macrophages, which in turn produce copious amounts of active cathepsins. These enzymes then cleave a recognition site in the linker portion of the probe between discrete Cy5.5 molecules, allowing for the separation of tightly spaced fluorochromes that were previously quenched in the inactive state . FMT of the myocardium has become an extremely valuable research tool in the preclinical setting. However, even in the near-infrared portion of the spectrum, significant absorption and scattering of light limit penetration to a few centimeters. This limits fluorescence imaging in humans to superficial structures or invasive techniques. Thus, fluorescence imaging of the human heart will likely require surgical or catheter-based approaches for the foreseeable future.
Cardiac mechanics is greatly affected by the microstructure of the myocardium. Specifically, the orientation of myofiber tracts in the myocardium plays a central role in the contraction of the ventricle. Seminal studies using histology showed that the heart is composed of an array of crossing helical myofibers [36, 37]. MRI has the ability to image these muscle tracts by measuring the diffusion of water molecules within the tissue. MR diffusion tensor imaging (DTI) has been implemented to approximate the three-dimensional fiber structure of the heart  and has been successfully used to quantify infarct healing after ischemic injury in animal models [39, 40]. Differences were seen in mean diffusivity (MD), fractional anisotropy (FA), and fiber orientation between healthy and infarcted regions. MD is a measure of how freely water diffuses and is reduced by the organized structure of the myofibers and connective tissue network in normal myocardium. FA is a measure of whether water diffuses preferentially in one direction. Patients and animals with acute infarcts have shown an increase in MD and decrease in FA. These scalar measurements of myocardial microstructure have the potential to play an important role in the ongoing development and application of microstructural imaging in the myocardium.
DTI can be used to derive several important vector measurements of myocardial microstructure. Perhaps the most important of these is the primary eigenvector, which indicates the principal direction of water diffusion. The primary eigenvectors in a 3D field can be integrated into streamlines, which indicate the course of myofiber tracts. While this has been extensively performed in the brain , the application of this technique to the myocardium has been relatively recent. A further advance in the field of microstructural imaging has been the development of DSI . While complex and time-consuming, DSI is considered to be the gold standard of all the diffusion imaging techniques that sample diffusion space (or q-space) with high angular resolution [42–44]. Every voxel in the spatial (x, y, z) domain of a DSI acquisition has its own 3D diffusion or q-space associated with it (Fig. 7). Since water diffuses primarily along the long axis of myofibers, the Fourier transformation of q-space produces a probability density function (PDF), where the local maxima represent the directions of fiber orientation within each voxel (Fig. 7). The reader interested in more information on the physical basis of DTI, DSI, and other diffusion imaging techniques is referred to a recent review .
Myofibers form a series of crossing spiral structures in healthy myocardium where the helix angle varies from a right-handed helix in the subendocardium to a left-handed helix in the subepicardium [36, 37]. In a recent DSI tractography study of rat hearts, a smooth transition in fiber orientation from epicardium to endocardium was seen (Figs. 8 and 9a–c) [3, 16]. In infarcted hearts, however, myofiber architecture was severely disrupted (Fig. 9d–f). Numerous residual myofibers were present within the infarcts, forming a mesh-like network of orthogonal fibers in the basal and septal portions of the infarcted region. Perpendicular fibers within this area often lay in direct contact with each other, creating nodes of myofiber contact . The implications of such a pattern remain to be determined, but it is possible these fibers provide greater structural support while increasing the risk of the formation of lethal reentrant arrhythmias.
Efforts to perform DTI in the heart in vivo have been made for over a decade . While in vivo imaging of 3D fiber architecture remains a challenge, 2D DTI in humans has been described [45, 46]. In vivo 2D DTI in patients with recent infarcts showed an increase in mean diffusivity, a decrease in FA, and a loss of right-handed (subendocardial) fibers in the infarct zone . A follow-up study in these patients demonstrated the potential of DTI to resolve serial changes in myocardial microstructure in both the infarct and the remote zone: An increase in right-handed fibers in the remote zone correlated with better function . These studies demonstrate the potential of microstructural imaging in humans and the value of correlating microstructural changes with functional remodeling. Tractography of the human heart in vivo remains experimental. Nonetheless, a method for the 3D interpolation and reconstruction of fiber architecture in the left ventricle from sparse DTI datasets in normal volunteers has recently been described . Further work, however, will be needed to bring diffusion tractography into the clinical mainstream.
The role of both MRI and fluorescence imaging in studies of myocardial injury and remodeling continues to grow. While fluorescence imaging is extremely sensitive and multispectral in nature, molecular MRI provides high-resolution images that can correlate myocardial structure and function. The ability to combine molecular and microstructural imaging is particularly appealing since it allows events at the subcellular level to be linked to gross changes of the heart. Further efforts to refine both MRI and optical instrumentation, as well as probes for multi-modality molecular imaging, are likely to further improve this multiscale approach. The characterization of the myocardium with noninvasive imaging techniques has thus moved well beyond traditional macroscopic indices and into the era of multiscale measurement. Molecular and microstuctural imaging are already playing an important role in preclinical investigation and have the potential to play an equally important role in the clinical arena as well.
Dr. Sosnovik has been funded in part by the following National Institutes of Health grant: R01 HL093038.
Craig J. Goergen, Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.
David E. Sosnovik, Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. Cardiology Division, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Building 149, Room 2301, 13th Street, Charlestown, MA 02129, USA.