The Task Force of the Society of Nuclear Medicine (SNM) Molecular Imaging Center of Excellence (MICoE) defines molecular imaging as the “visualization, characterization and measurement of biological processes at the molecular and cellular levels in humans and other living systems.”1
In the context of translation into clinical practice, molecular imaging would perhaps better be known as “molecular diagnostic imaging,” to distinguish it from “classical diagnostic imaging.” Unlike the latter, molecular imaging goes beyond structural assessment, and probes disease-specific abnormalities at the molecular level, putting it at the frontiers of biomedical science where new genetic and molecular causes of disease are continually being discovered.
However, molecular imaging is not a new concept. For example, positron-emission tomography (PET) has proved capable of imaging molecular processes such as blood flow in the brain and other organs using O-15 water, as early as the 1970s.2 Yet, molecular imaging has only recently been defined as a separate field. This was probably instigated by the completion of the human genome project in April 20033,4 and the recognition that many diseases have molecular and cellular causes. It is now recognized that imaging of processes intrinsic to metabolism, cell signaling and gene expression is possible. The evolution of our knowledge of molecular mechanisms of disease and the progress of imaging technology are happening rapidly and in parallel, fostering their combined application in molecular imaging.

Molecular imaging systems
Simply put, a molecular imaging system typically consists of a target, an agent and an imaging modality. Molecular imaging necessitates the interaction of the target with a “labeled” agent that can be detected externally by one or more modalities.
Agent-target interaction
Three major ways of agent-target interaction are recognized (Figure 1). Targeted binding: Here, the labeled agent selectively binds to its target, for a long enough period to allow external detection by an imaging modality. One example is neuroreceptor imaging in the brain, such as with 11C-raclopride, which binds to the type-2 dopamine receptor. The interaction is detected externally by PET imaging (Figure 1A).

Imaging agent accumulation in the cell: This usually results from enzymatic action that modifies the structure of the agent. The best example is 18F-fluorodeoxyglucose (FDG), which, once inside the cell, is acted upon by hexokinase, resulting in a phosphorylated version that can neither cross the cell membrane again nor undergo glycolysis. This results in the accumulation of FDG in highly metabolic cells, such as in cancer or infection/inflammation. The interaction can be detected externally by PET imaging (Figure 1B).
Activation of imaging agent by cellular components: The cellular components are usually enzymes, resulting in signal amplification. One example is bioluminescence in which the luciferase enzyme expressed by the target cell acts on injected luciferin. Emitted light is then detected externally by specialized cameras (Figure 1C).
Agent labeling and amplification strategies
Prior to agent-target interaction, substantial work goes into the labeling process of the imaging agent. The most common labeling method is probably the use of modified injectable agents adapted from known drugs/molecules.5-10 For that purpose, the in vivo characteristics of the labeled molecules need to be determined first. Favorable characteristics include, high specific activity of the label (to avoid drug toxicity and preserve tracer characteristics); high site selectivity and specificity; appropriate binding affinity; suitable hydro/lipophilicity and size (which govern transport across barriers); suitable metabolism (metabolites are likely to carry the label but exhibit altered specificity); low immediate excretion (renal, hepatic) or sequestration (drug resistance transporters, phagocytes); low non-specific binding (as the background signal reduces contrast); and, the ability to achieve high local concentrations.11-15
Limitations of the use of injectable molecular imaging agents of different modalities can be explained to a great extent by their pharmacological properties. For example, ultrasound contrast agents (in the form of microbubbles) and magnetic resonance imaging (MRI) iron nanoparticles are relatively bulky, and are restricted by body membranes such as the vascular endothelium, the blood-brain barrier or the cell membrane; they are restricted much more than low-molecular-weight agents.16-25 Radionuclide-based labeling techniques, on the other hand, provide the ability of replacing atoms of biological molecules with their radioisotopes, resulting in radiolabeled injectable agents that ideally are chemically identical (same size, chemistry and charge) to their unlabeled state. Positron-emitting radioisotopes lend themselves to these techniques, because they include the lower-mass-number elements of the periodic system, which are major components of biological molecules, such as carbon, nitrogen, fluorine and oxygen.5,7,8,26-38
Another form of labeling is indirect, using reporter genes. This constitutes the basis of molecular genetic imaging. In this case, a reporter gene (the product of which can be detected externally), is genetically linked (ex vivo) to a promoter of the gene of interest in such a way that when the gene of interest is expressed, the reporter gene product (protein) is produced, enabling imaging (Figure 2). One such example is the use of the herpes simplex virus type 1 thymidine kinase gene (HSV1-tk) as a reporter gene. When the gene of interest is expressed in the cell, the HSV1-tk gene is simultaneously transcribed to HSV1-tk mRNA and then translated to TK. TK can then phosphorylate one of its substrates, such as iodine-124 fluorodeoxy arabinofuranosyl iodouracil (124I-FIAU). The phosphorylated 124I-FIAU cannot cross the cell membrane and is sequestered within the cell, becoming amenable to PET imaging. The phosphorylated 124I-FIAU is thus an indirect indicator of the expression of the gene of interest.39-42 Many new imaging agents for human reporter genes are subject to active research in anticipation of their use in human gene therapy.

In contradistinction to indirect labeling, direct cell-labeling techniques are also used in molecular imaging to introduce a label into cells in vitro, before transplantation. One example is the use of the highly derivatized cross-linked iron oxide nanoparticle (CLIO-HD) to label killer lymphocytes ex vivo prior to reintroduction of the labeled cells into the system.43 These iron oxide particles provide strong negative contrast when imaged with MRI. After reintroduction of the labeled cells, the 3-dimensional distribution of infiltrating T-cells across the whole tumor can be detected using MRI, with simultaneous assessment of both cellular recruitment and therapeutic efficien-cy.43
Molecular imaging modalities
There are 4 main categories of molecular imaging modalities: ultrasound, optical imaging, MRI and nuclear medicine techniques. The choice of the imaging modality is determined based on the temporal and spatial resolution; field of view; sensitivity of the imaging system; depth of the biological process; the molecular or cellular process to image (protein vs. cell); and, the availability of suitable probes and labels that can be delivered to the imaging target.
Ultrasound
Molecular imaging with ultrasound commonly utilizes specialized contrast agents, usually in the form of small acoustically active gas-filled microbubbles that possess high echogenicity, sufficient to elicit signal from as little as a single microbubble (femtoliter ~10-15 liter volume). The microbubbles are often coated with lipids, proteins or polymers, with diameters of >1µm,20 which confines them to the intravascular space. This is why relevant targets are usually those expressed on the endovascular structures or intravascular cells such as angiogenesis, inflammation and intravascular thrombi.16 The use of ultrasound in molecular imaging is largely limited to animal applications (Figure 3). Excellent reviews on the use of ultrasound in combination with targeted microbubbles and ultrasound-induced drug delivery exist.17-20

Optical techniques
Optical techniques include 2 major classes: fluorescence and bioluminescence imaging. Fluorescence refers to the property of certain molecules (fluorophores) to absorb light at a particular wavelength and to emit light of a longer wavelength after a brief interval known as the fluorescence lifetime.44 Fluorescent reporters carry such fluorophores. Two-dimensional planar fluorescence reflectance imaging (FRI) technology has evolved into 3-dimensional fluorescence-mediated tomography (FMT),45,46 which is capable of deep tissue penetration, using sophisticated computational analysis methodology to reconstruct the in vivo distribution of intravenously injected fluorescent probes.47 Photon wavelength influences the depth resolution of optical imaging techniques. Near-infrared photons currently provide the greatest wavelengths (650 nm to 900 nm) and the best depth of penetration (>1 cm).
Bioluminescence imaging uses reporter genes that lead to expression of luciferase proteins. Upon injection of the substrate, luciferin, light is emitted as a result of a chemical reaction involving luciferase, luciferin, oxygen and ATP. The emitted light is detected externally.48 As opposed to fluorescence imaging, no excitation light is required. Luciferase proteins are derived from different organisms, such as bacteria, fireflies, red and green click beetles, and renilla reniformis (sea pansy). Firefly and red click beetle luciferases emit longer wavelength photons.49
Optical imaging techniques are characterized by their spatial resolution varying from several millimeters to micrometer resolution, and by their excellent sensitivity. However, their use is limited to small animals and surface and fiber optic imaging in nonhuman primates and humans (Figure 4).

MRI technique
While MRI provides very high resolution (up to 10 µm) and unlimited depth of penetration, it is, however, limited by low sensitivity, with detectabilities in the milli- to micromolar (10-3 to 10-6) range.50 Therefore, amplification techniques are often needed to image molecular processes in vivo.
The best recognized MRI amplification technique is the use of iron oxide particles as contrast agents. These provide negative MR contrast through local increase of the relaxivity (R1 and R2) of the tissue.51-55 Superparamagnetic iron oxide (SPIO) particles are becoming increasingly popular as they provide the strongest contrast available for MR imaging, while they are biodegradable by cellular enzymes. Their surface coating dextrans facilitate linking to ligands.56 Detection is possible at micromolar concentrations of iron, and sensitivity is sufficient for T2*-weighted imaging.52
Cells labeled with iron oxide are used for monitoring of cell trafficking in vivo. The labeling can be done through systemic IV injection of the particles, which are then incorporated inside macrophages through phagocytosis. The migration of macrophages can then be monitored externally using MRI (Figure 5).57-59 Labeling can also be performed through in situ injection of iron oxide particles near areas of stem-cell formation such as the subventricular zone of the brain.60,61 The most common technique though is in vitro labeling of cells prior to injection.62-64

Nuclear medicine techniques
Nuclear medicine techniques provide practically unlimited depth penetration and have very high sensitivities, in the nanomolar (10-9) range.65,66 Production of radiotracers with high specific radioactivity yields detectable radioactivity while maintaining low pharmacological doses. Although PET and single photon emission computed tomography (SPECT) cause radiation exposure and have relatively low resolution (2 mm to 5 mm for PET, 8 mm to 12 mm for SPECT),65,66 nonetheless they are the most commonly used human molecular imaging modalities. As a result of applicability of the tracer principle to radiopharmaceuticals used in human studies (only a very small quantity of the radiopharmaceutical is introduced, too small to exert any pharmacologic effect), the United States Food and Drug Administration (FDA) approval of new radiopharmaceuticals is generally much less complicated than for other modality imaging agents.
PET tracers are, in general, easier to quantify; reliable kinetic modeling is established for many tracers. PET has approximately 100 times higher sensitivity than SPECT, in large measure due to the ability to avoid using collimators during imaging.67 The most commonly used PET radionuclides are 11C (half-life ≈ 20 min) and 18F (half-life ≈ 110 min). SPECT tracers, in general, have longer half-lives, allowing the study of molecular processes that evolve over longer times. SPECT provides many readily useable radiotracers, and a local cyclotron is not needed for operation.
Small animal molecular imaging
Small animal molecular imaging represents the vast majority of molecular imaging applications at the present time. This mirrors the fact that molecular imaging is becoming an integrated and indispensable tool for modern life sciences. Medical application is expected for the most successful, least harmful and most reliable techniques, and clinical usefulness remains the true final proving ground.
Molecular imaging in small animals is facilitated by the adaptation of the imaging systems. Preclinical small-animal imaging PET, SPECT, MRI and CT scanners68-75 provide high-resolution imaging. Also, there is increasing availability of mouse models of human diseases, such as cancer, atherosclerosis and neurological diseases.76-83
Current applications of molecular imaging
Translational molecular imaging brings promising experimental therapies and diagnostic tests to the clinic, after extensive evaluation in experimental models. Frequently used modalities include PET and SPECT, as well as a few MRI applications. Optical techniques and ultrasound applications remain of limited use (for reasons outlined previously) but they retain high potential. For example, fluorescence imaging is being investigated as an enhancement of fiber-optic detection of colon cancer.84,85
Oncology
FDG is the most widely used molecular imaging agent in oncology. Since its inception as a marker of metabolism in tumor cells, FDG has played a major role in the diagnosis, evaluation and follow-up of different tumors, including lymphoma, lung cancer, brain cancer, head-and-neck tumors, melanoma, and breast cancer (Figure 6).86-108 Unlike normal cells, where glycolysis is inhibited by the presence of oxygen (Pasteur effect),109,110 FDG uptake in tumors is reflective of increased glycolysis, even in the presence of oxygen (aerobic glycolysis or Warburg effect),109 which is facilitated by overexpression of glucose transporters and glycolytic enzymes in malignant cells.111-114

However, FDG-PET has limitations. For example, in brain tumors, FDG use is limited by high background uptake of the tracer, since the brain uses glucose as its main source of energy (Figure 6).115 The use of FDG-PET in prostate cancer is also limited by low tracer uptake of tumor cells which can overlap with uptake in benign prostate hyperplasia (BPH)116-118 and the anatomic location of the prostate gland in close proximity to the urinary bladder.119 Such limitations have instigated the evaluation of alternative molecular and cellular targets, such as amino acid transport, DNA synthesis, fatty acid metabolism and angiogenesis.
Proliferating tumor cells are characterized by increased amino acid transport across the cell membrane120, which can be evaluated using radiolabeled amino acids such as 11C-methionine (MET).120,121 In brain tumors, the superiority of MET over FDG in the evaluation of disease extent, surgical planning, evaluation prior to stereotactic biopsy, follow-up and evaluation for recurrence, has been demonstrated.122-130 This is probably due to the lower background uptake of MET, resulting in higher tumor-to-background ratios and better visualization of the tumor. The main limitation, however, is in the short half-life of 11C prohibiting smooth clinical translation. 18F-labeled amino acids are thus being evaluated, such as 18F-FET, 131,132 18F-DOPA133,134 and 18F-ACBC.135,136
DNA synthesis is another molecular imaging target in tumors. Imaging of thymidine kinase 1 (TK1), an enzyme overexpressed during the DNA synthesis phase of the cell cycle, reflects cellular proliferation.137-140 18F-Fluorothymidine (FLT)29,30,141,142 is a radiolabeled nucleoside analog that is phosphorylated by TK1, rendering it unable to leave the cell. FLT has been successfully used in the original evaluation as well as the follow-up and assessment of treatment response in many tumors such as lung cancer143, lymphoma144, head-and-neck cancer145, and breast cancer (Figure 7).146

Besides nuclear medicine techniques, MR applications of molecular imaging in oncology exist, such as the use of coated iron oxide particles to detect metastatic disease in lymph nodes or the liver.23,147-150 When injected systemically, iron oxide particles that are phagocytosed by macrophages are then transported to the lymph nodes. A metastatic lymph node, in which the normal macrophage population has been replaced by tumor cells, will demonstrate partial or no drop in signal, while a normal lymph node, in which the iron particles have localized, will have decreased signal and provide detailed characterization independent of typically accepted size criteria (Figure 5).23
Inflammation and infection
Increased FDG uptake can be seen in infectious/inflammatory conditions since FDG-PET does not target a molecular process that is specific for neoplasia but rather uses the relative increase in glucose metabolism of neoplastic cells over normal parenchymal cells. The main limitation of using FDG in inflammation imaging, thus, is its lack of specificity. In fact, FDG uptake of benign tumors, inflammatory processes and malignant neoplasms can sometimes overlap.151-153
To improve differentiation between neoplastic and infectious/inflammatory processes, multiple molecular imaging targets of infec-tion/inflammation have been evaluated mostly in animals, with few human applications at this point. Examples of human applications include the use of monoclonal antigranulocyte antibodies, such 99mTc-fanolesomab (NeutroSpec®, Palatin Technologies, Cranbury, NJ), which binds the CD15 antigen expressed on neutrophils.154-158 Antibody fragments are slightly more popular due to lower immunogenicity and faster clearance, such as 99mTc-labelled sulesomab (LeukoScan®, Immunomedics Inc., Morris Plains, NJ) which was found to be as accurate as white blood cell scanning in osteomyelitis and soft-tissue infections159,160 as well as prosthetic joint infections.161,162 Radiolabeled antibiotics, on the other hand, have been used to directly target bacteria, rather than reactive cells. The best known is probably 99mTc-ciprofloxacin (Infecton®, DRAXIMAGE, Kirkland, Quebec, Canada)163,164 which has been evaluated in a multitude of infectious entities such as acute cholecystitis,165 spinal infections,166 and abdominal infections.167
Recently, 124I-FIAU168,169 PET-CT has been found to be useful in imaging musculoskeletal bacterial infections.170 In that study, the substrate specificity difference between bacterial TK and the major human TK was exploited to develop a new imaging technique that can detect the presence of viable bacteria (Figure 8).170

Neuroimaging and neurodegeneration
Alzheimer’s dementia (AD) is the most common cause of dementia and earlier diagnosis is sought for more accurate prognosis and education of the patients and their families. Several gene therapy trials are attempting to halt or even reverse progression of AD.171-175 In AD, decreased FDG uptake is seen, reflective of regional impairment of cerebral glucose metabolism, mostly in neocortical association areas, whereas primary visual areas, the sensorimotor cortex, basal ganglia, and cerebellum are relatively well preserved.176 FDG-PET abnormalities however are not pathognomonic of AD. Alternative molecular imaging targets are sought for higher specificity. The most intuitive targets are the pathologic associates of AD, namely neurofibrillary tangles (NFTs) and amyloid plaques (APs). Novel PET and SPECT ligands for NFTs and APs are under investigation, most of which are derived from known histologic staining agents used in AD such as DDNP,177-179 thioflavins S and T180-185, and stilbenes.178,186-188
Perhaps the best known amyloid ligand currently is the 11C-labeled Pittsburgh compound B (PIB). The development of this compound from the initial investigations in mouse models of AD189-191 to the first applications in humans was extraordinarily fast (Figure 9).192-195 In the earliest studies, AD patients showed increased retention of PIB in association cortex areas known to contain large amounts of amyloid deposits in AD, compared with controls.192 However, “cognitively normal” controls with higher-than-normal PIB uptake were noted,196-199 raising the possibility of those subjects being predisposed to develop AD. If this is proven in prospective larger studies, PIB could potentially be a useful diagnostic tool for detection of disease prior to the onset of symptoms, which can help maximize the benefits of therapy. However, PIB suffers from the short physical half-life of 11C such that 18F amyloid-binding derivatives are actively being pursued. Other ligands investigated as AP and NFT markers include 18F-FDDNP,200,201 6-iodo-2-(4’-dimethylamino) phenyl-imidazo[1,2-a]pyridine (IMPY) derivatives,202,203 and 11C-SB-13.204-205

Conclusion
In summary, molecular imaging is undergoing constant change and is rapidly expanding. It spans all current life sciences and is being utilized at the frontiers of modern research. For the clinical radiologist, the future will bring application of molecular imaging techniques into the standard diagnostic workflow. Radiology will continue to be enhanced as knowledge from molecular biology, genomics and proteomics, neuroscience and molecular physiology continues to be integrated into imaging research and, eventually, practice.





