|Home | About | Journals | Submit | Contact Us | Français|
The past few decades have seen an explosion in the knowledge of the molecular basis of cardiovascular diseases, owing to rapid advances in molecular biology research. An improved understanding of disease pathogenesis at the genomic, transcriptional, and proteomic levels has led to the discovery of promising experimental strategies for the prevention, diagnosis, and treatment of cardiovascular disease. Unfortunately, only a minute number of these strategies have survived the rigors of pre-clinical and clinical trials to become therapeutically useful. Furthermore, even these successful strategies must endure a prolonged process of translation from bench to bedside, partially due to the lack of tools to directly interrogate the molecular events in patients. The strong impetus to develop noninvasive imaging techniques to visualize the molecular changes in patients has given birth to the field of molecular imaging.1, 2
Molecular imaging has its roots in nuclear medicine, in which radiolabeled imaging probes are injected into living subjects to assess the functionality of different organ systems. Unlike conventional diagnostic imaging techniques (e.g., X-ray and CT) that delineate the anatomy of the cardiovascular system (e.g., coronary luminal diameter), molecular imaging techniques have been designed and validated to study much smaller scale molecular events (e.g., gene expression) which may underlie disease processes. The complexity of molecular imaging lies in the requirement for molecular targeting, the design of which requires a solid understanding of the pharmacokinetics of the imaging probe and how it interacts with the molecular target. When the target is proven to be a biomarker, molecular imaging becomes a valuable tool for 1) detecting disease before its clinical manifestation, 2) stratifying disease severity, 3) predicting disease progression, 4) monitoring treatment efficacy, and 5) prognosticating disease. These challenges must be met before the true potential of personalized medicine can be fully realized.3
Significant advances have been made in molecular imaging to make it a clinical reality in many areas. Instrumental in its development is the advancement of small animal imaging technologies over the past 2 decades, including fluorescence imaging (FI), bioluminescence imaging (BLI), ultrasound (US), micro-positron emission tomography (microPET), micro-single photon emission computed tomography (microSPECT), high-field small animal magnetic resonance imaging (MRI), and micro-computed tomography (microCT) (Figure 1A). Parallel in development is the construction of sophisticated imaging probes with high specificity for various molecular targets. The use of these imaging systems and probes has facilitated the validation of molecular imaging techniques in animal models. Ongoing translation of these techniques into the clinical arena is being achieved by using either clinical versions of small animal scanners or special imaging platforms specifically developed for clinical translation. A wide range of exciting cardiovascular molecular imaging applications has now been developed and reviewed previously.1, 4, 5 After a brief overview on the fundamentals of molecular imaging approaches, this review will focus on the latest advances in the areas of atherosclerosis, heart failure, and stem cell therapy. The article will conclude with discussions on future prospects of molecular imaging in the clinical arena, as well as future directions that will shape molecular imaging in the post-genomic era.
Imaging of molecular events requires the selection of a molecular target, an imaging probe, and an imaging system. The molecular target can be DNA, RNA, or protein in the form of an intracellular enzyme, a cell surface receptor, a membrane transporter, or an extracellular enzyme. Proteins are of most practical interest due to their abundance within a cell (0.01-1 million, compared to 1-2 DNA and 10-1,000 RNA), which contributes to the imaging sensitivity. The cellular compartment in which the target resides needs special consideration because a transport mechanism (e.g., transporter) may be required to bring the probe across the cell or organelle membrane for intracellular target-probe interaction. In contrast, imaging of cell surface or extracellular targets can circumvent cellular transport barriers and can be performed using a generalized platform as shown in Figure 1B. The commonly used molecular imaging modalities are discussed below to highlight their relative strengths and weaknesses (see also Table 1).
FI is typically performed by exogenously delivering a fluorescent probe (e.g., an organic dye) that interacts with the target or by directly imaging an endogenously expressed fluorescent protein. Signal generation is achieved by exciting the fluorescent probe or protein at a given light wavelength (λ) and detecting light emission at another λ using a charge-coupled device (CCD) camera built into either a planar (fluorescence reflectance imaging; FRI) or a tomographic imaging system (fluorescence molecular tomography; FMT).6 Because light attenuation by tissue is wavelength-dependent, fluorescent probes or proteins with more red-shifted emission λ [near-infrared (NIR) probes or fluorescent proteins with λ=700-900 nm] have been developed to maximize imaging sensitivity and specificity. More recently, fluorescent semiconductor nanoparticles (2-8 nm), also known as quantum dots (QDs), have also been customized to allow simultaneous imaging of multiple targets using a single excitation wavelength and QDs of varying sizes.7 The advent of QD technology alleviates the problems associated with organic fluorophores (e.g., photobleaching, low quantum yield, low absorbance, and broad emission band). The cytotoxicity of QDs, however, needs to be addressed before they can be used clinically.8 Magnetofluorescent nanoparticles for both MRI and FI represent yet another class of fluorescent probes, which have found wide pre-clinical applications in studying inflammatory atherosclerosis9, post-infarct healing10, transplant rejection11, and early aortic valve disease12. These agents enable the attainment of both high imaging sensitivity from FI and high spatial resolution from MRI, which helps to also compensate for the limited imaging depths of FI. More recently, an intravascular near-infrared fluorescence-sensing catheter has been developed to allow imaging close to the vascular target.13 This novel imaging technique should help open the door for many promising human applications, including imaging of micro-thrombi associated with vulnerable coronary plaques.
BLI uses light generated from an enzyme-substrate pair as an imaging signal and an ultrasensitive cooled CCD camera for signal detection. Exogenous expression of a luciferase enzyme, followed by systemic delivery of its substrate, forms the basis of in vivo BLI. To date, luciferases cloned from beetles (firefly luciferase, FLuc; click beetle red; CBR) and sea pansy (Renilla luciferase; RLuc) have been used together with their respective substrates (D-Luciferin and coelenterazine) for in vivo BLI.14 The beetle luciferases are generally preferred over the sea pansy luciferase because their greater red-shifted emission spectrum produces better tissue penetration. Their light generation, however, requires O2, magnesium, and ATP, whereas the sea pansy luciferase can function well in extracellular environments lacking ATP. For optimal in vivo application, a mutant sea pansy luciferase with red-shifted emission wavelength (λ=535 nm) has been developed for improved imaging sensitivity and extracellular targeting.15 BLI has excellent imaging sensitivity (10-15-10-17 mole/L) due to its negligible background signal in living animals.1 Its low-cost and high-throughput capability make BLI the preferred modality for tracking stem cell fate16 and gene therapy17 in small animal models. However, due to the limited light penetration, the requirement for exogenous reporter genes, and the need for the injection of mass amounts (μg to mg) of potentially immunogenic substrates, it is unlikely that BLI will be clinically useful.
Molecular US is performed using microbubbles (~1-10 μm) that are composed of gas (e.g., perfluorocarbon) enclosed in spherical shells of lipid, protein, or biocompatible polymer. Targeting of microbubbles to diseased vasculature can be achieved by either directly conjugating ligands to the microbubble shell, or indirectly via an interjacent molecule (e.g., streptavidin) using biotinylated ligands.18 The size of the microbubbles, however, precludes their vascular extravasation, thereby limiting the selection of molecular targets to the endothelium.19 Signal enhancement from the microbubbles depends on their size, compressibility, and the frequency and power of the incident ultrasound wave.20 Imaging of microbubbles at their resonance frequency with medium power can stimulate non-linear oscillation around their equilibrium dimension, producing harmonic signals that can be separated from the surrounding tissue signal.21 Microbubble-specific signal can otherwise be produced by destroying the microbubbles using high acoustic powers, thereby, releasing highly echogenic free gas.22 Molecular US retains the advantages of conventional US (e.g., real-time imaging, portability, and low-cost), in addition to being highly sensitive. The disadvantages include its high degree of operator dependency, limited access to extravascular targets, low microbubble adhesion efficiency, and poor ability to image past bony structures.
Single photon emission computed tomography (SPECT) and positron emission tomography (PET) are radionuclide-based techniques commonly used for molecular imaging that differ mainly in the radionuclide used and the mechanism for signal generation and detection. SPECT imaging probes (tracers) are labeled with gamma-emitting radionuclides (e.g., 99mTc, 111In, 123I, and 131I), some of which can be milked from a generator, whereas PET tracers are labeled with positron-emitting radionuclides (e.g., 15O, 13N, 11C, and 18F) mainly produced in an on-site cyclotron. The former set of radionuclides produces high-energy gamma rays of different energy levels, which can be detected with a SPECT camera equipped with lead collimators. The latter set of radionuclides produces signal via annihilation events, in which each positron emitted collides with a nearby electron to produce two 511 keV gamma rays at 180° apart, which can then be coincidentally detected and localized by a PET scanner. Because both SPECT and PET use high energy gamma rays as signal, these techniques can be performed in large animals and humans as well as small animals. Co-registration of SPECT/PET and CT images acquired sequentially allows anatomical localization of probe activity.23 Combining PET imaging with MRI can further enhance anatomical localization by taking advantage of MRI's superb ability to differentiate soft tissue boundaries.24 Currently, PET and SPECT are the most translatable noninvasive molecular imaging platforms due to 1) the availability of existing clinical scanners, 2) the availability of versatile radiochemistry options, and 3) the ongoing development of new imaging agents, which only need to be administered in a small, non-pharmacological dose (nanograms) for highly sensitive imaging (SPECT: 10-10-10-11 mole/L; PET: 10-11-10-12 mole/L).1 The last merit should greatly facilitate the translation of new radionuclide-based imaging probes into clinical trials, where toxicity from the trace amount of probe injected is least expected. The disadvantages of radionuclide imaging include low spatial resolution (clinical SPECT: 7-15 mm; clinical PET: 6-10 mm; microSPECT: 0.5-2 mm; microPET: 1-2 mm), radiation exposure, lack of anatomical information (without CT), and the need for a generator/cyclotron for radionuclide production.
MRI detects the net magnetic moment of a collection of nuclei in a strong magnetic field (B0) following a radio-frequency (RF) pulse. The rates at which the net magnetic moment recovers or decays in the axes parallel and perpendicular to the direction of B0 are characterized by the longitudinal and transverse relaxation times, T1 and T2/T2*. 1H is the most commonly studied nucleus, but 13C, 19F, 23Na, 31P, and others have also been used for imaging.25 Molecular MRI is performed by using targeted contrast agents that can alter either the T1 or T2/T2* relaxation of water protons near the target, thereby creating signal contrast. Gadolinium (Gd) chelates and superparamagnetic iron oxide nanoparticles (SPIO; 60-150 nm), including the ultra-small (USPIO; 10-40 nm) and micron-sized versions (MPIO; 0.9-8 μm), are the two most popular classes of imaging agents that can generate contrast by shortening T1 and T2/T2*, respectively. The lower intrinsic moment of Gd, however, generally requires multiple Gd chelates to be linked together via a carrier (e.g., nanoparticle or peptide) to improve its sensitivity.26 Conjugation of Gd-chelates (or nanoparticles) to antibodies, peptides, or peptidomimetics are needed for molecular targeting. Note that more signal amplification can be accomplished by incorporating multiple Gd-chelates into antibody-coated nanoparticles27, or more sophisticated protein/liposome assemblies (e.g., recombinant high-density lipoprotein-like nanoparticles28 or immunomicelles29). Alternatively, the MRI signal can be further enhanced and made more specific by using novel Gd-based contrast agents that oligomerize in response to target enzyme activation.30 Other exciting contrast agents that are under active development include exchange saturation transfer (CEST) agents31 and 19F perfluorocarbon (PFC) nanoparticles.32 The latter agents, which confer both positive contrast and high specificity, have been successfully used to label circulating monocytes or macrophages in vivo.33 The main advantages of MRI include its superb spatial resolution (1.5T: 0.5-1.5 mm; ≥3T: 10-100 μm), great soft-tissue discrimination, lack of radiation exposure, good clinical translatability, and the ability to combine functional, anatomical, and molecular information. Its disadvantages include its low imaging sensitivity (10-3–10-5 mole/L)1, long scan or post-processing time, and high maintenance cost.
CT measures the relative ability of different tissues to attenuate incident X-ray by rotating an X-ray source and a detector around the 3-dimensional volume of a subject. Traditionally, the role of CT in molecular imaging has been limited to providing anatomical roadmaps for functional SPECT or PET scans (e.g., SPECT/CT or PET/CT applications), largely because of its low imaging sensitivity (10-2-10-3 mole/L).1 However, recent development of iodinated nanoparticulate contrast agent, N1177, has made it feasible to image cellular activity in vivo (i.e., macrophage accumulation in atherosclerotic plaques), beyond just outlining the adjacent anatomical structures.34 The ability to manufacture polymeric nanoparticulate contrast agents containing high concentrations of organometallics or radio-opaque organically soluble elements has further pushed the limit of CT's imaging sensitivity (10-9-10-10 mole/L).35 The main advantages of CT include its high spatial resolution (microCT: 20-300 μm; CT: 0.5-2 mm), excellent hard-tissue imaging, and short scan time. Its disadvantages include its low imaging sensitivity, limited soft tissue discrimination, and radiation exposure.
Atherosclerosis, a leading cause of mortality in the developed world, is a systemic inflammatory process characterized by fatty-streak development, plaque (atheroma) formation, and the potential of plaque rupture or erosion, leading to myocardial infarction (MI) or cerebrovascular accidents (CVA).36 X-ray angiography can be used to assess luminal narrowing but cannot evaluate atherosclerosis that does not encroach on the lumen. Intravascular ultrasound (IVUS), MRI, and 3D carotid ultrasound are capable of detecting plaques based on their morphology or tissue composition, but cannot distinguish the molecular signatures of rupture-prone (vulnerable) plaques from those of stable plaques. An active area of research in cardiovascular molecular imaging is, therefore, to develop techniques that can detect the molecular events underpinning plaque vulnerability. Such techniques should help identify patients at a high risk for cardiovascular events, guide their therapy, and monitor their treatment accordingly.
The predominant form of vulnerable plaque, termed thin-cap fibroatheroma, has been shown by ex vivo histology of human cadavars to have a thin-fibrous cap (<65 μm), a large necrotic lipid core (>10% plaque area), paucity of smooth muscle cells, lack of thrombus, and heavy infiltration of fibrous cap by inflammatory cells (i.e., macrophages).37 The initiation of plaque rupture has been linked to macrophages which contribute to the digestion of the fibrous cap by up-regulating metalloproteinases.38 PET imaging of plaque macrophages using 18F-fluorodeoxygluocse (18F-FDG) (radiolabelled glucose analogue) represents an attractive strategy to study plaque progression because: 1) macrophage density is greater in ruptured plaques than stable plaques, 2) macrophages in the anaerobic plaque interior use glucose as substrate, and 3) activated macrophages in cell culture express high levels of glucose transporters and hexokinase needed for 18F-FDG uptake and trapping.39
The potential of using 18F-FDG PET to image atherosclerosis was first suggested more than a decade ago based on the observation that deoxyglucose can be significantly trapped by macrophages residing within a tumor.40 Subsequently, the feasibility of 18F-FDG plaque imaging was established in animal models of aortic atherosclerosis (e.g., rabbits with heritable hyperlipidemia41 or exposed to balloon denudation plus atherogenic diet42), with results showing up to 3-5-fold greater 18F-FDG uptake in atherosclerotic aortas compared to normal aortas. The clinical feasibility of 18F-FDG plaque imaging has since been demonstrated in patients with carotid, aortic, vertebral, femoral, and iliac atherosclerosis.43-45 Although it has not been fully validated that 18F-FDG reports on plaque macrophage activity, a positive correlation has been observed between 18F-FDG PET activity and macrophage content in human plaque samples.44 The short-term (1-2 weeks) reproducibility of 18F-FDG PET for imaging systemic atherosclerosis is excellent, reportedly better than those of MRI, IVUS, and CT angiography for aortic and carotid lesions, with negligible intra- and inter-observer variabilities (<5%).46 Plaque inflammation assessed by 18F-FDG imaging following statin therapy (Figure 2) has been noted to precede changes in plaque morphology by ~9 months, demonstrating the importance of imaging molecular or cellular changes that often precede morphological change.47
18F-FDG PET/CT imaging of coronary plaque inflammation has been attempted with preliminary success, despite suboptimal image contrast resulting from elevated myocardial 18F-FDG uptake, motion artifact, and partial volume effect due to small vessel size (typically 2-6 mm).48 The first challenge has been addressed by using an overnight low carbohydrate, high fat diet regimen which can suppress myocardial 18F-FDG uptake by 3.5-fold greater than fasting.49 Ongoing efforts are aimed at reducing motion artifacts by instituting cardiac gating and breath hold techniques, and minimizing partial volume errors by performing post-reconstruction corrections using anatomic priors obtained from MRI.45 Efforts to standardize imaging protocols and data analysis are underway to help improve image quality and data accuracy.50 Ultimately, large prospective trials to test the power of 18F-FDG at predicting cardiovascular event risks are needed.
The translatability of 18F-FDG PET as a technique to study atherosclerosis in humans is excellent, owing to increased utilization of PET scanners worldwide for routine oncological investigations. Furthermore, 18F-FDG can now be made available to sites without cyclotrons at reasonable costs. With pharmaceutical companies eager to develop treatment strategies to either prevent or stabilize vulnerable plaques, 18F-FDG PET is poised to take on an essential role in the area of clinical atherosclerosis imaging.
Imaging of human plaque inflammation has also been achieved using USPIO as an MR contrast agent. USPIO is preferred over SPIO due to its longer circulation time in humans (>24 hours),51 which maximizes probe-target interaction. Following intravenous delivery, USPIO is predominantly engulfed by macrophages, and to a limited degree, by endothelial or smooth muscle cells within the inflamed plaques. The propensity of USPIO to accumulate in ruptured or rupture-prone plaques (75%), compared to stable plaques (7%), makes it a promising probe for imaging plaque vulnerability.52
Early studies involving symptomatic patients undergoing carotid endarterectomy found that up to 24% change in MRI signal could be observed for plaques 24 hours post-USPIO delivery with histological proof of USPIO uptake.52 A subsequent optimization study found an imaging time between 24 and 36 hours post-USPIO injection yielded the greatest detectability.53 No target image contrast could be observed after 72 hours due to either USPIO “washout” from the plaques or physiological recycling of endocytosed iron. The fast iron turnover renders repetitive imaging with this technique feasible and potentially useful for monitoring treatment efficacy. In fact, a recent prospective human study (ATHEROMA) showed the feasibility of using USPIO-MRI to serially monitor the effect of atorvastatin on plaque inflammation at 6 week intervals (Figure 3).54 Reduction of USPIO uptake was detected as early as 6 weeks after high dose (80 mg) atorvastatin therapy, months to years earlier than changes in plaque morphology. These findings reaffirmed the advantage of molecular imaging over morphometric imaging for assessing the effectiveness of plaque-altering therapy.
The major challenge facing USPIO imaging concerns the routine use of gradient recalled echo (GRE) pulse sequences, which can lead to signal loss in areas of USPIO uptake. This hypointensity is sometimes masked by signal loss arising from motion, tissue voids, or heavy calcification, leading to erroneous interpretation of plaque USPIO uptake. Newer MR imaging techniques that allow detection of USPIO as a positive contrast signal are under development and should help improve the diagnostic accuracy of the USPIO-enhanced MRI technique.55
Plaque erosion presents another form (besides plaque rupture) by which luminal thrombosis can occur. More commonly seen in premenopausal women and the coronary circulation, it accounts for ~40% of coronary thrombi in patients who died suddenly from coronary artery atherosclerosis.56 Histologically, plaque erosion is characterized by a luminal thrombus in an area of endothelial denudation, with the underlying intima enriched with smooth muscle cells and proteoglycan matrix, but almost devoid of inflammation or calcification.37 Recent studies have shown that the luminal thrombus typically undergoes days to weeks of organization and healing prior to causing an acute event.57 Therefore, a time window exists by which early detection of thrombosis with imaging can help guide prompt intervention and prevent coronary accidents.
MRI of fibrin-rich thrombus has been demonstrated in swine models of coronary thrombosis using EP-2104R (a novel Gd-based fibrin-binding peptide manufactured by EPIX Pharmaceuticals), a clinical 1.5T MR scanner, and a T1-weighted cardiac-triggered inversion recovery black-blood gradient echo sequence.58, 59 Early validation of this agent was performed by imaging an ex vivo engineered human clot that had been previously delivered into the swine left main coronary artery.59 The fibrin-rich clot used was larger than the size of a typical subclinical microthrombus overlying a vulnerable plaque. Nevertheless, as a proof of principle, high contrast-to-noise ratios (CNRs) (>17) comparing clot to blood pool were demonstrated after systemic delivery of EP-2104R. In a subsequent study, the MRI signal was further proven to correlate with ex vivo thrombus size.26 Afterwards, EP-2104R was validated in a phase II clinical trial involving a subset of patients with thrombi in different vessel territories potentially responsible for stroke (left ventricle, left atrium, thoracic aorta, carotid artery), with results showing comparable post-contrast CNRs (>20).60 More recent data from the phase II trial further showed a sensitivity of 84% for detecting known thrombi in the aforementioned vascular territories using this agent.61 As the application of this versatile imaging agent continues to expand (e.g., imaging of deep vein thrombosis62), ongoing efforts are aimed at further increasing its imaging sensitivity by optimizing the contrast dose and various imaging parameters. For now, this agent is more suitable for imaging advanced large plaque thrombosis rather than subclinical microthombi that overlie eroded plaques.
MRI of thrombus has been otherwise achieved at high fields (9.4T) in mouse models of carotid thrombosis using single-chain antibody-conjugated MPIOs (1 μm) targeted to the glycoprotein IIb/IIIa integrin of activated platelets.63 Compared to EP-2104R, this agent reportedly has the ability to detect smaller thrombi because of the high imaging sensitivity associated with MPIOs. The MRI signal achieved with this targeted agent has been found to correlate well with ex vivo thrombus size and respond to thrombolytic treatment. The lack of toxicity observed with the use of these MPIOs in animal models is reassuring and should help propel their further testing in humans.
More recently, a novel imaging agent (P975) composed of a single Gd moiety attached to a cyclic arginine-glycine-aspartic acid (RGD) peptide has been developed to target activated platelets within thrombi via binding to their αIIbβ3 integrin. In mouse models of arachidonic acid-induced carotid thrombosis, this agent was capable of leading to a 4-fold greater ΔCNR (the change in CNR after contrast injection) than sham surgery controls.64 Importantly, P975 was able to bind specifically to αIIbβ3 integrin, even though other integrin subtypes (e.g., αvβ3 on smooth muscle cells) could potentially compete with its binding. Thus far, only nearly occlusive thrombi have been imaged with this agent. It remains to be seen whether this single-Gd-containing agent would allow imaging of even smaller thrombi.
To date, MRI of coronary thrombosis has been difficult due to the small-size of coronary vessels and the cardiac/respiratory motion complicating high-resolution imaging. Nevertheless, recent successful imaging of human coronary plaques using Gd-DTPA, a non-targeted MR contrast agent, is encouraging and should pave the way for further testing of targeted coronary thrombosis imaging agents.65
Recent advances in intravascular ultrasound (IVUS)-Virtual Histology (IVUS-VH) with radiofrequency analysis have made it possible to accurately (~92%) characterize plaque composition with the potential to predict risk of plaque rupture.66 The detection of molecular events that precede changes in plaque composition should further accelerate the workup for vulnerable plaques. For example, NIRF imaging of aortic atheroma has been extensively validated in mouse models of atherosclerosis using activatable fluorescent probes.67 Theses probes have been custom-designed to fluoresce in living subjects only upon encountering cysteine proteases or metalloproteinases (MMPs), whose up-regulation has been closely linked to plaque expansion and rupture.38 The clinical translation of these imaging approaches has been propelled by the latest development of a NIRF sensing catheter, which has allowed imaging of macrophage-associated cathepsin B protease activity in rabbit aortic atheroma using a protease-activatable probe.68 Images with target-to-back ratios >6 have been achieved, despite a 30% light attenuation by blood. Current efforts are aimed at improving the specificity of the probe for cathepsin B and the sampling capability of the imaging device. Further refinement of this novel intravascular NIRF imaging platform may open the door for early intracoronary detection of vulnerable plaques.
The apoptosis of smooth muscle cells or macrophages within a plaque has been causally linked to plaque rupture and represents yet another promising imaging target for the detection of vulnerable plaques. Among many cellular or molecular changes that accompany apoptosis, the translocation of phosphatidylserine to the external surface of the cell membrane represents an integral event that precedes many morphologic or nuclear changes. For this reason, 99mTc-labelled annexin A5, a plasma protein with high affinity for phosphatidylserine, has been tested in a small pilot study of 4 patients for its ability to detect cellular apoptosis associated with unstable human carotid plaques.69 Encouragingly, following probed delivery, significant 99mTc-annexin A5 uptake was found in the culprit lesions of only patients with a recent, but not remote, history of transient ischemic attack. Ex vivo histological analysis of the culprit lesions following endarterectomy further demonstrated annexin A5 binding to the macrophage membranes only in plaques with unstable characteristics. Therefore, although tested in only a limited number of patients, this probe appears to be capable of distinguishing vulnerable from stable plaques. Future prospective outcome studies in a larger group of patients will help determine its true discriminatory potential.
Other promising plaque imaging techniques include, but are not limited to: 1) 99mTc-interleukin-2 for imaging activated T-cells in human carotid plaques70, 2) αvβ3 integrin-targeting nanoparticles for MRI of plaque-associated angiogenesis,71 3) micelles containing gadolinium (Gd) and oxidation-specific antibodies for MRI of oxidation-rich plaques,72 and 4) 18F-labeled vascular cell adhesion molecule-1 (VCAM-1)-targeting peptide73 or VCAM-1-targeted magnetic nanoparticles (MNPs)74 for PET/CT or MR imaging of activated endothelium in inflammatory atherosclerosis. The readers are referred to other excellent reviews for techniques not covered in this review.75-77
Congestive heart failure (CHF) is a devastating condition responsible for 1 in 8.6 deaths in the US as of 2006, and is expected to incur a cost of ~$39.2 billion in 2010.78 The astounding cost reflects the difficulty of managing CHF, a disease with complicated pathogenesis. Our current knowledge of CHF points to inflammation, oxidative stress, extracellular matrix remodeling, neurohormonal activation, and myocyte injury/stress as the main contributors. Serum-based biomarkers associated with these processes are currently being investigated for their ability to stratify disease severity, predict mortality, guide therapy, and assess treatment efficacy.79 However, not all valuable biomarkers are secretory nor can they be specifically attributed to myocardial processes. The development of molecular imaging techniques to detect and localize myocardial biomarkers with high sensitivity should complement the use of serum-based biomarkers for better management of CHF.
CHF has been linked to a hyperadrenergic state in which an increased norepinephrine (NE) release from the cardiac sympathetic post-ganglionic nerve terminals causes desensitization of the post-synaptic β-adrenergic receptors, leading to worsening of left ventricular (LV) systolic function.80 A decreased efficiency of NE re-uptake through the pre-synaptic NE transporter-1 (NET-1) further contributes to NE spillover into the bloodstream in CHF patients.81 The clinical significance of such neuropathic derangement is evidenced by the known benefit of β-blockade for reducing mortality. Noninvasive investigation of the myocardial adrenergic transmission in CHF patients should, therefore, be able to reveal their risk of cardiac death.
123I-metaiodo-benzylguanidine (123I-MIBG), a radiolabeled NE analogue, has been used for years to assess cardiac sympathetic innervation in CHF patients (Figure 4).82, 83 Although similar to NE in terms of its storage, release, and re-uptake in the pre-synaptic nerve terminal, it neither binds to the post-synaptic receptors nor undergoes pre-synaptic or post-synaptic metabolism.84, 85 123I-MIBG's myocardial retention after systemic delivery depends mainly on: 1) the integrity of pre-synaptic nerve terminal for 123I-MIBG storage and 2) the capability of pre-synaptic 123I-MIBG re-uptake through NET-1 relative to its release (sympathetic drive). The first factor has been semi-quantitatively assessed using early heart-to-mediastinum ratio (H/M; 10-20 minutes), and the latter using either late H/M (3-4 hours) or washout rate (WR). Compared to healthy subjects, CHF patients have a lower H/M (early or late), which decreases with disease severity. Conversely, they also have a higher WR, which increases with CHF severity. The prognostic value of 123I-MIBG planar imaging has been extensively studied. Earlier studies have suggested that 123I-MIBG uptake (cutoff late H/M=1.2) is a better predictor of 2-year event-free survival than plasma NE, echocardiography, or radionuclide ventriculography for CHF patients (NYHA II-IV; LVEF<45%) with either non-ischemic or ischemic cardiomyopathy.86 A recent meta-analysis of 18 123I-MIBG studies (N=52-205 each; N=1755 total) has found that either low late H/M or high WR in patients with CHF predicts high mortality.87 However, the largest study to date (ADMIRE-HF; phase 3; N=961; NYHA II-III; LVEF<35%) showed that only late H/M is consistently associated with risks for cardiac events in a multivariate analysis.83
A growing area of interest in 123I-MIBG imaging research is to investigate whether such a tool can be used to predict sudden death due to severe arrhythmia, which has been linked to sympathetic hyperactivity in CHF.88 If proven to be prognostic, 123I-MIBG can be used to identify patients who would benefit from ICD placement. Thus far, multiple single center studies (N<100) have shown that 123I-MIBG planar imaging, using either abnormal WR (≥27%)89 or late H/M (1.95) in combination with plasma BNP (>187 pg/ml), can independently predict sudden cardiac death with positive and negative predictive values reaching 82% and 94%, respectively.90 Although larger studies are needed to confirm these results, these studies highlight the feasibility of combining 123I-MIBG with other serum markers to augment predictability.
A failing heart has been likened to “an engine out of fuel” such that increasing its efficiency rather than its workload helps to maintain its lifetime.91 In CHF, the heart has an innate ability to increase its metabolic efficiency by switching from predominantly fatty acid metabolism at rest to glucose metabolism, resulting in 11% more ATP produced per molecule of oxygen. Such metabolic transformation is also seen in myocardial ischemia, in which the myocardium exposed to limited oxygenation increases anaerobic glycolysis to preserve viability.92 In fact, the clinical use of 18F-FDG-PET together with myocardial perfusion imaging to identify hibernating myocardium in patients undergoing consideration for coronary revascularization is based on this concept.93 The utility of 18F-FDG-PET has recently been expanded to include studying therapies that affect myocardial metabolism, including cardiac resynchronization therapy (CRT). CRT has been shown in numerous trials to improve long-term LVEF and ventricular remodeling by minimizing contractile dyssynchrony.94 In addition, 18F-FDG PET studies have shown that CRT helps to homogenize myocardial glucose utilization in patients with dilated cardiomyopathy and left bundle branch block, restoring the reverse mismatch pattern (low septal 18F-FDG uptake-to-blood flow ratio) commonly seen in these patients.95 As ~1/3 of CHF patients do not respond to CRT, this pattern has been employed to predict response to CRT. With the Receiver Operating Characteristics area under curve (AUC) reaching 0.93, this technique has been shown to be more predictive of positive CRT response than either LVEF (0.66) or QRS duration (0.75).96 In light of the recent multi-center PROSPECT trial showing poor predictability for 12 echocardiographic prognostic parameters (all AUC≤0.62),97 18F-FDG-based analysis presents an attractive alternative strategy for selecting patients who will likely respond to CRT.
The nature of tissue repair following myocardial ischemia or infarction can significantly influence the chance that further ventricular remodeling will lead to CHF. Understanding the kinetics of molecular or cellular events that accompany infarct repair (e.g., cardiomyocyte apoptosis, leukocyte trafficking, protease activation) through the use of molecular imaging can help facilitate the design of therapeutic strategies to both minimize ventricular remodeling and prevent CHF progression.
The apoptotic pathway plays a major role by which cardiomyocytes die after myocardial ischemia reperfusion and represents a promising target for therapy. An early attempt at noninvasively studying cardiomyocyte apoptosis in patients presenting with acute coronary syndromes utilized SPECT and 99mTc-labeled annexin.98 The dynamic range of this technique, however, was found to be limited by the high background signal within 12 hours of probe administration, thus making early detection of apoptosis after ischemia difficult. Furthermore, this probe does not effectively differentiate apoptosis from necrosis, both of which exposes the imaging target (phosphatidylserine) to annexin. Recent development of annexin-labeled magnetofluorecent nanparticles (AnxCLIO-cy5.5) for high-resolution MRI made it more feasible to image cardiomyocyte apoptosis within 4-6 hours of ischemia reperfusion, when apoptosis is most prevalent.99 When used in conjunction with Gd-DTPA-NBD (a Gd-based probe for necrosis) in a cleverly designed mouse study, this agent allowed the apoptotic cells to be differentiated from necrotic cells, further helping to identify a group of apoptotic yet viable cells in the mid-myocardium (4-6 hours after ischemia reperfusion), which can benefit from anti-apoptotic therapy. As recently demonstrated in a transgenic model of chronic heart failure, AnxCLIO-cy5.5 has an impressive ability to detect very low levels (2%) of cardiomyocyte apoptosis.100 The dual contrast imaging technique using both AnxCLIO-cy5.5 and Gd-DTPA-NBD has a great potential for clinical translation and will improve further with ongoing optimization of various imaging parameters (e.g., echo time) and contrast agent dose.
Post-infarct myocardial remodeling represents another imaging target of significant interest due to its central role in the development of ischemic cardiomyopathy. The initial phase of remodeling is triggered by the recruitment of phagocytes, followed by their release of proteases to degrade extracellular matrix in an attempt to repair the infracted myocardium. The feasibility of imaging macrophage infiltration in the infarcted myocardium with a combination of FMT and MRI was initially established using a magnetofluorescent nanoparticle (CLIO-Cy5.5).101 The temporal kinetics of macrophage/neutrophil infiltration and their secretion of proteases was later noninvasively studied in a transgenic mouse model of impaired wound healing (FXIII-/-) using CLIO-VT750 (a magnetofluorescent nanoparticle for phagocyte imaging), Prosense-680 (a fluorescence reporter on cathepsin B activity), and a combination of MRI and multispectral FMT.10 The lower CLIO-VT750 and Prosense-680 heart signals in these animals, compared to wide-type mice, following coronary ligation suggested both impaired phagocyte recruitment and protease secretion, both of which could account for the previously observed inability of these mice to repair infarction. This dual modality imaging technique thus offered new insights into the potential causes of pathological post-infarct remodeling. By closely monitoring myocardial repair and healing, such a powerful imaging tool may one day allow physicians or scientists to devise strategies to ensure more optimized scar formation so as to prevent heart failure.
An impressive repertoire of molecular imaging techniques has been developed for studying heart failure. Some of these novel techniques include: 1) 99mTc-labeled Cy5.5-RGD imaging peptide (CRIP) for imaging fibrinogenesis by myofibroblasts during myocardial remodeling102, 103, 2) 99mTc-labeled losartan for imaging angiotensin receptor II upregulation following myocardial infarction104, 3) 99mTc-labeled collagelin, a collagen-targeted peptidomimetic of the platelet collagen receptor glycoprotein VI, for imaging myocardial fibrosis105, 4) 111In-labeled affinity peptide (111In-DOTA-FXIII) for imaging transglutaminase factor XIII activity in the healing infarct106, and 5) 111In- or 99mTc labeled radiotracers for imaging metalloproteinase (MMP) activities in post-infarction remodeling.107 The readers are referred to other outstanding reviews on imaging techniques not being covered here.108-110
Myocardial infarction, an undesirable consequence of atherothrombosis, is associated with a considerable rate of progression to CHF.78 Currently, the only cure for end-stage CHF is heart transplantation, which is limited by organ shortage and high cost. Left ventricular assist devices can help prolong life by several months, but is only a temporizing measure before heart transplantation. For these reasons, stem cell transplantation has been extensively studied for the past decade as a novel therapy to reverse or minimize myocardial injury, with the goals of improving cardiac function and halting progression from MI to CHF.111, 112
The majority of pre-clinical studies thus far have shown that implantation of various cell types (e.g., skeletal myoblasts, bone-marrow derived mononuclear stem cells, mesenchymal stem cells, circulating progenitor cells, embryonic stem cell derived cardiac or endothelial cells, and cardiac resident stem cells) into ischemic or infarcted myocardium can lead to varying degrees of benefits, including reduced infarct size, reversed ventricular remodeling, and improved left ventricular systolic function.111, 112 The exact mechanism by which cells exert their benefits is still being unraveled, but has generally been linked to 1) cells secreting growth factors locally (paracrine function) to either stimulate angiogenesis or recruit endogenous stem cells for enhanced cardiac repair, 2) cells acting as mechanical scaffolds to strengthen the mechanically weak myocardium, or 3) stem cells differentiating into cardiomyocytes capable of contracting in synchrony with the host myocardium.111, 112
Despite encouraging findings from early animal and initial small-scale human studies, the more recent large-scale, randomized, phase II trials have shown inconsistent results, mainly with the use of bone marrow-derived stem cells (BMCs). Two recent meta-analyses of 18 and 10 trials, respectively, on intracoronary infusion of BMCs for the treatment of acute MI have shown only marginal benefits (~3-4% increase in LVEF).113, 114 The reasons for these perplexing, if not disappointing, results are not entirely clear, but are gradually being unraveled by the increasing use of imaging tools in pre-clinical studies and in clinical trials to monitor stem cell behavior.115
Directly imaging the behavior of cells following implantation into living subjects can offer great insights into their mechanisms of action, as well as their therapeutic efficacy. Cell imaging requires using a cell marker that ideally 1) generates signal only when the marker is associated with the cell while it is viable (imaging specificity), 2) emits adequate signal for detection (imaging sensitivity), 3) minimally perturbs cellular function (cytotoxicity), and 4) causes minimal toxicity to the subject when it (or its metabolite) is released into the circulation during excretion (systemic toxicity).
Cell labeling has been performed using 2 general approaches, namely direct cell labeling and reporter gene/probe labeling (Figure 5).116 The former approach is accomplished by incubating cells with contrast agents that either bind to the cell surface or are imported into the cells via diffusion, endocytosis, or active transport (e.g., 18F-FDG, SPIO). The agents may be trapped intracellularly or may leak out of the cell over time. The latter approach is performed by transfecting/transducing cells with a reporter gene, whose protein can be 1) a membrane transporter that actively imports exogenously delivered probes (e.g., sodium iodide symporter, NIS),117 2) an intracellular enzyme that actively accumulates/interacts with exogenously delivered probes [e.g., herpes simplex virus type 1 thymidine kinase (HSV1-TK) or its mutant (HSV1-sr39TK)],118 3) an intracellular storage protein that actively concentrates endogenous contrast elements [e.g., ferritin (FT) for concentrating iron],119 or 4) a cell surface receptor that binds to an exogenously delivered probe (e.g., dopamine type 2 receptor, D2R).120
The main advantage of the reporter gene/probe labeling approach is that the imaging signal becomes 1) specific to only viable implanted cells capable of mediating reporter protein-probe interaction and 2) reflective of cell number if the reporter gene is integrated into the genome and replicates with cell division. In contrast, with direct cell labeling, the contrast agent is diluted over time with cell division and may be engulfed by other cell types (e.g., macrophages) upon cell death, leading to the signal either not reflective of cell number or not specific to the implanted cells.121, 122 The main concerns over genetic-based stem cell labeling for human use are potential immunogenicity of some of the reporter genes (e.g., HSV1-sr39tk) and random reporter gene integration into cellular chromosomes (e.g., lentiviral transfection), which can lead to tumorigenesis. The former issue can be either circumvented by using endogenous reporters (e.g., human ferritin heavy chain) or addressed by using a human version of the exogenous reporter gene (e.g., human mitochondrial thymidine kinase type 2 for HSV1-sr39tk).123 The latter issue necessitates consideration of safe, site-specific transgene integration strategies under development.124 Other barriers to translation of reporter gene technology are related to the necessity to achieve robust and persistent reporter gene expression for utmost imaging sensitivity without significantly perturbing the functionality (differentiation capacity) of the implanted stem cells. In this regard, the use of a strong endogenous promoter (e.g., human ubiquitin promoter) to regulate reporter gene expression can minimize gene silencing due to promoter methylation.125 Enhanced reporter gene expression can be otherwise achieved by using strategies to enhance promoter activity (e.g., two-step transcriptional amplification126), reporter protein stability (site-directed mutagenesis127), or potentially transcript stability, which needs further exploration. It should be noted that due to the finite imaging sensitivity of any imaging device, it may be the case that even a lower number of cells than can be imaged needs to be assayed for clinical management. The development and concomitant use of secretory reporters that can be sensitively assayed by ex vivo nanotechnology-powered diagnostic devices should help ensure long-term monitoring of stem cells after implantation. Lastly, the potential toxicity from high levels of reporter gene expression (or repeated accumulation of the reporter probe) needs to be carefully assessed using more sophisticated proteomic analyses as shown previously.128 Adherence to the regulations implemented by the U.S. Food and Drug Administration (FDA) regarding reporter gene transfection of stem cells, in addition to the safe use of stem cell products and contrast agents, will also be key to quickly translating reporter gene technology to the clinical setting. As demonstrated in a recent clinical study involving the use of PET to track the homing of cytolytic T cells stably expressing HSV1-tk in a glioblastoma patient undergoing adoptive cellular immunotherapy129, reporter gene imaging can already be accomplished in the clinical setting; it should only be made easier and safer after all of the above issues are addressed.
A variety of cell imaging techniques have been validated in pre-clinical models, with each having its unique strengths and weaknesses, as previously reviewed elsewhere.116 However, only the radionuclide-based imaging techniques [PET, SPECT, and gamma camera imaging (GCI)] with direct cell labeling have been used in human studies (Table 2).130-137 These approaches can be quickly translated into the clinical setting because: 1) radionuclide imaging techniques are routinely performed in nuclear medicine, 2) similar cell labeling/imaging techniques have been used clinically for years to monitor the trafficking of other cell populations (e.g., 111In-oxine-labeled leukocytes for imaging infection),138 such that both the toxicity profile and the technical limitations are already known, and 3) their high imaging sensitivity allows tracking of low number of cells.
To date, all cardiac-related human stem cell imaging studies have been performed using either bone marrow-derived stem cells (BMCs)130, 131, 133, 135 or circulating progenitor cells (CPCs).132, 134, 136, 137 Direct labeling of stem cells has been achieved with 18F-FDG for PET imaging130, 132, 134, 137 and 99mTc-HMPAO131, 133, 135 or 111In-oxine134, 136 for SPECT/planar GCI. The cytotoxicity associated with these techniques is generally low (cell viability > 90% post-labeling).130, 131, 133, 135-137 The labeling efficiency, however, can vary greatly from study to study, even with the same cell marker (<10% to >70% for 18F-FDG).132, 137 Following intracoronary infusion, all stem cells have been found to engraft poorly (<10% and <5% at 2h and 24h, respectively), regardless of the cell type and the number of cells implanted (15-4,000 million) (Figure 6A).130-137 Cell retention has also been shown to correlate inversely with infarct age, presumably because more chemotactic cytokine is released during acute MI.136 However, the limited number of subjects in all of these imaging studies has precluded an adequate analysis of therapeutic efficacy. Nevertheless, the systemic toxicity profile of these techniques appears to be excellent, with no reported morbidity/mortality or breach of current radiation safety standards.132 In all cases, intravenous injection of stem cells, regardless of cell type, leads to undetectable myocardial homing.130, 132 It should be noted that this does not mean BMCs do not home to the myocardium following intravenous injection, but rather the number of BMCs homing to the heart is below the detection threshold of PET, since the more sensitive BLI approach has been able to illustrate the spatiotemporal kinetics of BMC homing following MI in small animal models.139
A major limitation of the aforementioned studies concerns the inability to monitor cell viability beyond 4-5 days. This duration is technically limited by the half-lives of the common isotopes used [18F (~110 min), 111In (~2.8 d), and 99mTc (~6 hr)] and biologically limited by the poor cell engraftment and survival. The technical limitation can be circumvented by using PET reporter gene/probe cell labeling, which has allowed longitudinal PET imaging of stem cells for at least 10 days in porcine animal models140 and greater than 2 weeks in murine models (Figure 6B).118, 125 The biological limitation has been addressed in preclinical studies using molecular imaging tools to study factors that influence stem cell survival (e.g., stem cell type16, 141, 142, timing of cell delivery after myocardial infarction143, route of cell delivery139, stem cell immunogenicity144), in addition to stem cell function (e.g., differentiation145) and biology (e.g., tumorigenicity146). Effective genetic or pharmacological strategies to promote transplant engraftment and survival have also been developed and validated in small animals.147 With further refinement of these strategies, long-term monitoring of stem cell therapy in humans should be feasible and could help optimize its therapeutic efficacy.
The clinical assessment of stem cell therapy has relied on the use of echocardiography, MRI, and nuclear perfusion imaging to assess the physiological consequences (e.g., changes in LVEF or myocardial blood flow) of stem cell implantation. However, none of these measurements reflects the impact of stem cells at the molecular level which may occur long before any noticeable physiological changes. Because a major benefit of stem cell implantation arises from stimulated angiogenesis, a direct assessment of which should shed great insights into the effectiveness of stem cell therapy. Molecular imaging of angiogenesis has been performed with a wide variety of techniques (e.g., PET, SPECT, MRI, and US) and is well described by other reviews.148, 149 Integrin (αvβ3) is by far the most well validated target, and its overexpression on activated endothelial cells plays an important role in angiogenesis. The clinical feasibility of imaging myocardial angiogenesis with PET has recently been demonstrated in a patient with subacute MI using 18F-Galacto-arginine-glycine-aspartic acid (18F-Galacto-RGD) as an integrin (αvβ3)-targeting probe.150 Focal probe uptake reflecting angiogenesis as a natural component of myocardial healing could be clearly visualized in the infarcted myocardium. Angiogenesis has also been noninvasively assessed in porcine models of chronic myocardial ischemia undergoing plasmid-mediated vascular endothelial growth factor (phVEGF165) therapy using 123I-Gluco-RGD, another αvβ3-targeting PET imaging probe.151 Up to 1.7-fold greater probe accumulation could been seen in areas of gene delivery compared to that of saline-injected controls, which may be further enhanced using stem cell-mediated gene transfer in the future.
Besides PET, ultrasound has also been successfully used to image endogenous reactive angiogenesis and the effect produced by fibroblast growth factor (FGF) supplementation in a rat model of chronic hindlimb ischemia. The ability of this technique to detect peak integrin expression 7-10 days before maximal blood flow recovery following FGF administration further lends support to the greater sensitivity of molecular imaging over conventional physiological imaging for assessing therapeutic interventions.152 With further validation in both animal models and humans, these aforementioned imaging technologies could greatly improve clinical treatment planning by helping to predict the long-term benefits of stem cell therapy based on early cell engraftment or local stimulated angiogenesis.
An immense repertoire of molecular imaging techniques have been developed for various cardiovascular targets. The translatability of these techniques hinges mainly on 1) whether these techniques have been previously validated for other human applications, 2) whether they require additional platforms for clinical translation, and 3) whether there are additional safety concerns regarding the use of new imaging probes or other biologics (e.g., stem cells and reporter genes). In these regards, radionuclide-based techniques are most ideal for clinical use in the short-term (next ~1-3 years) because many of the imaging probes (e.g., 18F-FDG) have already been used routinely in the clinics and are now finding newer applications in cardiovascular medicine. The requirement for only trace, non-pharmacological dose of imaging probe, coupled with high imaging sensitivity, further accelerates FDA-approval and the translation of newer radionuclide-based imaging techniques into clinical trials. Although PET/CT is now routinely performed in humans, newer applications that focus on detailed anatomies within the heart (e.g., coronary plaques) may suffer significantly from partial volume errors due to PET's relatively poor spatial resolution. A dedicated PET/CT scanner optimized for heart imaging may be the key to ensuring successful imaging of small and sparse molecular imaging targets within the heart. The MRI-based molecular imaging techniques are likely to be successful in the mid-term (~3-5 years), pending more demonstration of clinical safety for both Gd-based and iron oxide-based compounds, especially after repetitive imaging. As the imaging sensitivity of MRI continues to improve with the optimization of nanoparticulate agents, MRI may become the most clinically versatile modality due to its high spatial resolution and its ability to provide anatomical, physiological, and molecular information, all in one imaging session. Fluorescence imaging will likely take longer for clinical translation (~6-8 years) pending full maturation of a clinically feasible technology (e.g., intravascular near-infrared fluorescence-sensing catheter for coronary imaging). Although FMT cannot currently be applied to humans due to current detector technology, noninvasive imaging of superficial vessels may still be possible. The large number of fluorescent probes already available for cell imaging, however, will likely make in vivo fluorescence imaging very versatile in the long run. Molecular US and molecular CT are currently the two least explored molecular imaging modalities and are expected to take the longest for clinical translation. Imaging applications that involve manipulation of biologics (reporter genes) will also take longer to flourish due to the need to ensure clinical safety from these biologics and the challenges in imaging them. Ultimately, each of the above imaging modalities may find a unique role in clinical molecular imaging that best utilizes its strengths.
As the field of molecular imaging heads forward in parallel with the advancement in molecular biology and human genetics, an increasing number of molecular targets will be discovered, many of which will become worthwhile to noninvasively image. It is likely that a high throughput approach will need to be used to quickly develop and screen for imaging probes with the ideal pharmacodynamic and pharmacokinetic properties, which primarily determines the performance of a molecular imaging technique. In this regard, high throughput screening (HTS) of peptides or small scaffold proteins from either one-bead one-compound combinatorial libraries or phage/yeast display libraries, in conjunction with a high-stringency screening method, represents a promising strategy to quickly identify a few peptide- or protein-based probes (out of billions) that can interact favorably with the imaging target.153 The in vivo pharmacokinetics of these selected peptide or protein probes can then be further fine-tuned by modification of the amino acid sequences after repeated testing in pre-clinical animal models. Such a systematic, highly efficient approach has been demonstrated for the development of peptide-ligands targeted to relevant physiological targets including integrin.154, 155 Other high throughput screening techniques have also been used to identify nanoparticles with improved uptake by different cell types.156 Perhaps, with a high throughput manufacturing scheme like these, it will be possible to satisfy the preclinical and clinical needs for imaging a large pool of new molecular targets.
In conclusion, significant technological development in the field of molecular imaging over the past 2 decades has made molecular imaging a clinically feasible tool for interrogating important disease-related molecular events in a wide spectrum of clinical conditions ranging from atherosclerosis to advance heart failure. As demonstrated in this review, molecular imaging has the potential to 1) detect vulnerable plaques before the clinical manifestation of coronary artery disease in order to guide early intervention, 2) assess the severity and prognosis of CHF via sympathetic innervation and glucose metabolism to help identify patient subpopulations who are more suitable for specific treatments (e.g., ICD and CRT), 3) study post-ischemia cardiomyocyte apoptosis and post-infarct repair so that therapeutic strategies can be designed to prevent ventricular remodeling and heart failure, and 4) monitor novel stem cell therapies to both predict treatment outcomes and guide therapy. As molecular imaging techniques continue to be translated into the clinical setting, future efforts will focus on: 1) standardizing existing imaging protocols for particular applications (e.g., 123I-MIBG imaging) to facilitate comparison across studies, 2) developing generalizable platforms by which molecular imaging probes can be either screened or tailored for optimal pharmacodynamics or pharmacokinetics, 3) exploring the combined use of different imaging modalities to utilize the strength of each technique (e.g., PET/MRI), and 4) combining molecular imaging assays with highly sensitive serum-based assays to maximize the sampling of relevant disease-specific biomarkers (secretory and non-secretory) and to improve their sensitivity and predictability. The proper implementation of these measures will help propel molecular imaging to the forefront of clinical cardiology, where diagnostic imaging continues to be crucial to effective day-to-day patient management.
We thank Blake Wu, Jim Strommer, and Dr. Patricia Nguyen for assistance with preparing the manuscript. We also thank Drs. Craig Levin and Frezghi Habte for helpful discussions.
Sources of Funding: This work was in part supported by HL093172, HL099117, and EB009689 (J.C.W.)