DOI: 10.1148/radiol.2451060731
(Radiology 2007;245:21-42.)
© RSNA, 2007
Molecular Neuroimaging: From Conventional to Emerging Techniques1
Dima A. Hammoud, MD,
John M. Hoffman, MD, and
Martin G. Pomper, MD, PhD
1 From the Department of Radiology, Johns Hopkins University School of Medicine, 1550 Orleans St, CRB-2, Room 492, Baltimore, MD 21231 (D.A.H., M.G.P.); and Departments of Radiology and Neurology, University of Utah, Salt Lake City, Utah (J.M.H.). Received April 26, 2006; revision requested June 27; revision received October 7; final version accepted November 20.
Address correspondence to M.G.P. (e-mail: mpomper{at}jhmi.edu).
 |
ABSTRACT
|
|---|
The use of molecular imaging techniques in the central nervous system (CNS) has a rich history. Most of the important developments in imaging—such as computed tomography, magnetic resonance imaging, single photon emission computed tomography, and positron emission tomography—began with neuropsychiatric applications. These techniques and modalities were then found to be useful for imaging other organs involved with various disease processes. Molecular imaging of the CNS has enabled scientists and researchers to understand better the basic biology of brain function and the way in which various disease processes affect the brain. Unlike other organs, the brain is not easily accessible, and it has a highly selective barrier at the endothelial cell level known as the blood-brain barrier. Furthermore, the brain is the most complex cellular network known to exist. Various neurotransmitters act in either an excitatory or an inhibitory fashion on adjacent neurons through a multitude of mechanisms. The various neuronal systems and the myriad of neurotransmitter systems become altered in many diseases. Some of the most devastating diseases, including Alzheimer disease, Parkinson disease, brain tumors, psychiatric disease, and numerous degenerative neurologic diseases, affect only the brain. Molecular neuroimaging will be critical to the future understanding and treatment of these diseases. Molecular neuroimaging of the brain shows tremendous promise for clinical application. In this article, the current state and clinical applications of molecular neuroimaging will be reviewed.
© RSNA, 2007
|
INTRODUCTION
|
|---|
The brain has been the great proving ground for new imaging techniques, particularly functional imaging techniques, because of the relative anatomic uniformity of brain parenchyma, the lack of substantial motion artifacts that may confound imaging of other organs, the structural symmetry of the brain, and the inherent fascination with the physiology and neurochemistry of the brain. Despite the presence of the blood-brain barrier, which limits access of many potential molecular imaging agents to the brain, functional and, more recently, molecular neuroimaging have been successfully performed and have provided a conceptual framework within which quantitative imaging techniques can be applied to other organs and diseases.
The use of molecular imaging techniques within the central nervous system (CNS) has a rich history. One must first ask the following questions: What is molecular neuroimaging? Does diffusion-tensor imaging or functional magnetic resonance (MR) imaging constitute molecular neuroimaging? What about autoradiography? In this article, molecular neuroimaging refers to the use of imaging techniques to detect and characterize molecular processes other than water, which is detected with diffusion-tensor imaging and functional MR imaging in vivo. When one thinks of molecular imaging in the CNS, one invariably thinks of probing neurochemistry with positron-emitting or lower-energy gamma-emitting radiopharmaceuticals. In this article, positron emission tomography (PET) and single photon emission computed tomography (SPECT), as they have been used in neurology and psychiatry, are considered conventional molecular neuroimaging techniques because they have been used to quantify neurotransmission and neurochemistry in humans for more than 20 years (1,2). An emerging indication for molecular neuroimaging, both within the brain and within the spinal cord, is following the movement of phagocytic or neural or oligodendrocyte progenitor cells by using MR or optical techniques (3–8). MR and multifunctional nanoparticles have been developed for use in the assessment of brain tumor margins, with MR nanoparticles having seen limited clinical use (9,10). Forays into molecular-genetic imaging within the CNS in human subjects and animal models are under way (11–13) and will be discussed in detail later in this article. MR metabolic imaging, which can be used to investigate a limited number of molecular species in the CNS, has been used in clinical practice for nearly 30 years, beginning with the use of proton and phosphorous spectroscopy to study tumors and stroke in animal models (14,15). Proton MR spectroscopic imaging of the brain has become a standard clinical adjunct to MR imaging of neoplasms, and it is increasingly being applied to other indications, such as pediatric metabolic brain disorders (16,17). Sodium MR imaging of the brain was first described by Hilal et al in 1985 (18); since then, this technique has experienced a renaissance, with the possibility of it being used to study changes in tumor sodium concentration with therapy in experimental models (19). Amide protein transfer imaging, which does not require contrast material administration, has been described for protein imaging in the human brain (20–24).
Conventional radionuclide-based techniques have been performed primarily in humans because the tracer principle (ie, the lack of a need to administer more than 10 µg of contrast material to the subject being examined) makes it relatively easy to gain institutional review board and Food and Drug Administration approval for use of radionuclide-based agents. The ability to perform studies in humans with the Radioactive Drug Research Committee approach will be discussed in detail in the final article of this series. Such a small amount of radiopharmaceutical is subpharmacologic and is therefore harmless. However, the more recent MR and optical molecular imaging strategies have been pursued predominantly in small-animal models of CNS disease. Because MR (except in the instance of cell labeling, as will be discussed later) and optical techniques (at least in their current incarnation) require higher concentrations of detectable probes to be delivered, the majority of studies in which these modalities are used are preclinical. The contribution of small-animal models to CNS molecular imaging will only increase as more sophisticated and physiologically relevant models are developed, such as the commercialization of reporter mice and improved genetic models of glioma (25–27).
Small-animal models have been particularly useful in helping to refine new MR and optical molecular neuroimaging techniques. Although radiopharmaceutical-based molecular neuroimaging is well established in human subjects, the use of this technique in animals is being revisited because of the proliferation of dedicated high-resolution small-animal imaging devices (28,29). The purpose of these small-animal brain imaging studies is to determine whether useful data can be obtained to help in the development of radiopharmaceuticals and drugs, as well as to test the sensitivity and resolution of the new imaging devices being built. One goal is to screen promising new radiopharmaceutical-based brain imaging agents in small-animal models in a semi–high-throughput fashion. Several challenges must be overcome before that approach, generally reserved for use in nonhuman primate studies, can be used. These challenges include assuring that adequate specific radioactivities are achieved so that target sites are not saturated, as well as assuring that appropriate blood samples can be obtained or that a suitable brain region can be used to serve as a surrogate for the input function, which is necessary for full quantification. The contribution of small-animal models to molecular neuroimaging will be discussed later in this article in the context of conventional molecular neuroimaging techniques, cell trafficking, molecular-genetic imaging, and brain metabolic imaging. The Table summarizes the various techniques used for molecular neuroimaging.
|
CONVENTIONAL MOLECULAR NEUROIMAGING
|
|---|
Although receptor-based PET imaging of the brain is methodologically complex and an interdisciplinary team is needed to perform this technique, it is nevertheless considered conventional in light of newer molecular imaging methods that are being applied to neuroimaging. To understand the application of PET and SPECT to the CNS, one must understand certain fundamentals of basic neurochemistry. Neurologic PET and SPECT have proved most useful in the characterization of various neuropsychiatric disorders—such as Alzheimer disease (AD) and other dementias (30); movement disorders, such as Parkinson disease (PD) (31); epilepsy (32); inflammation, such as that seen in patients with human immunodeficiency virus–associated dementia (33); or multiple sclerosis (34)—and brain tumors (35).
|
BASIC NEUROCHEMISTRY AND RECEPTOR-BASED IMAGING
|
|---|
Because molecular imaging is a biologically driven enterprise, one must understand the relevant underlying physiology before attempting to perform this technique. Although images obtained with molecular neuroimaging can be useful for establishing a diagnosis and for therapeutic monitoring, they are often analyzed by applying appropriate pharmacokinetic models and generating parametric maps or images for full quantification (ie, determination of physiologically meaningful rate constants that describe radiotracer kinetics) (36). Conventional radiotracer-type molecular neuroimaging probes are designed to mimic closely the behavior of endogenous transmitters. As such, a radiotracer can be used to investigate neurotransmission by binding to pre- or postsynaptic receptor sites or transporters, or it can function as a substrate for enzymes involved in the synthesis or inactivation of transmitters (37). Some researchers purport to measure second messenger effects relevant to neurotransmission (38). The design, synthesis, and development of molecular neuroimaging probes in the assessment of neurotransmission remains one of the most active areas in molecular imaging research and, to a large degree, parallels the development of new drugs for the CNS (39).
The CNS is different from other organ systems in two critical ways. First, unlike other organs, the brain has a unique highly selective barrier at the endothelial cell level known as the blood-brain barrier. This barrier has long posed a challenge to researchers, particularly in the realm of drug development. Second, the brain is the most complex cellular network known to exist. Transmitters act in either an excitatory or an inhibitory fashion on adjacent neurons through a multitude of mechanisms. They are generally low-molecular-weight organic molecules, such as biogenic amines, amino acids, or peptides; however, they may also be less conventional agents, such as nitric oxide, carbon monoxide (40), or hydrogen sulfide (41). Neurotransmitters achieve their effects primarily by binding to postsynaptic receptors. Transmitters and intracellular signaling are linked by association with small guanosine triphosphate–binding and hydrolyzing proteins, protein kinases, or the receptor itself, in the form of a ligand-gated ion channel, such as the acetylcholine receptor (42,43). There are numerous subtypes of receptors for any given neurotransmitter, and a major goal in molecular neuroimaging is to develop subtype-selective probes.
The ability to image a particular receptor (or enzyme or transporter) site in vivo is largely dependent on the target site concentration and the affinity of the radioligand for its receptor (44). The higher the target site concentration, the lower the affinity needed to affect visualization with molecular imaging techniques. The ratio of target site concentration to the affinity of the radioligand for its receptor represents a critical outcome parameter for receptor-based imaging and is known as the binding potential (45). The criteria for a successful radiopharmaceutical are continually evolving, but they largely parallel those for a successful drug (ie, suitable lipophilicity, simple metabolism, lack of affinity for the multidrug resistance pump, and low molecular weight) (37,46). Where drug and contrast agent development diverge, however, is in the degree of tolerable nonspecific binding. While a drug may bind to many sites, including nontarget sites, as long as the desired effect is evident and there is no substantial toxicity, the drug will succeed. Target site binding is necessary but not sufficient to ensure use of a radiopharmaceutical will be successful, as washout from nontarget sites is also essential.
Rather than provide an exhaustive description of each transmitter and its distribution and function (many excellent reviews already exist [40,42,43]), we will briefly focus on dopamine (DA) transmission as the prototype, as DA transmission has arguably been better characterized with imaging than with any other transmitter system (Fig 1). Selective high-affinity radioligands have been developed that bind not only to postsynaptic sites but also to presynaptic transporters and storage vesicles (vesicular monoamine transporter) or are the substrate for the enzyme that produces DA from DOPA and DOPA decarboxylase. Different radiopharmaceuticals have been used to depict each of those aspects of DA transmission with PET or SPECT (Fig 1). The biologic emphasis on imaging DA transmission resides in the importance of this system not only in movement disorders but also in the pathophysiology of schizophrenia and in mediating the emotional sensation that accompanies reward (47–50). DA neurons originate, for the most part, in the pars compacta of the substantia nigra and project to the striatum. DA neurons are also found in cells in the ventral tegmentum of the midbrain projecting to the cerebral cortex and limbic areas (51). Because so much effort has been expended in studying DA transmission, there is a menu from which to choose a specific imaging agent to be used to assess a particular biologic question. For example, there are many ligands that may be used to study the D2 receptor, including [11C]raclopride (RAC) (52,53), [18F]fluoroethylspiperone (54,55), [11C]N-methylspiperone (2), and [123I]epidepride (56,57). The choice of radioligand with which to assess a particular receptor system depends on the aspect of the system to be studied. For example, to study extrastriatal DA transmission, epidepride may be superior to other D2 radioligands because of its high affinity and the relatively long physical half-life of 123I (approximately 10 hours), enabling it to be used to label only low-concentration target sites while diffusing out of nontarget tissues. To study the release of endogenous DA due to pharmacologic challenge or to calculate receptor occupancy, [11C]RAC is the agent of choice because of its relatively low receptor binding affinity, which allows it to be displaced by an endogenous transmitter or a drug (58–60). With careful selection of the appropriate radiopharmaceutical, much can be learned in vivo that cannot be learned with autoradiography or microdialysis. This is primarily because the network of interactions that comprise neurotransmission is intact in vivo and the lack of invasiveness provides a viable alternative to microdialysis, which is generally reserved for animal studies.
View larger version (55K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1: Diagram of a dopaminergic synapse shows the synthetic pathway of DA. DA precursor tyrosine is transformed first to dihydroxyphenylalanine (DOPA) and then to free DA. Once synthesized, the vesicular monoamine transporter bundles DA into vesicles, where it is stored for synaptic release. The arrival of a nerve impulse produces the release of vesicular DA by exocytosis into the synaptic cleft. In the cleft, DA binds to both presynaptic (D2DR) and postsynaptic (D1DR and D2DR) receptors. Excess DA in the synapse is then transported back into the presynaptic neuron through specific DA transporters. DMFP = desmethoxyfallypride; DTBZ = dihydrotetrabenazine; FECNT = 2ß-carbomethoxy-3ß-(4-chlorophenyl)-8-(2-fluoroethyl)nortropane; FESP = fluoroethylspiperone; FP-CIT = N-(3-fluoropropyl)-2-ß-carbomethoxy-3-ß-(4-iodophenyl)nortropane; IBZM = methoxybenzamide; MNPA = (R)-2-CH3O-N-n-propylnorapomorphine; NMSP = N-methylspiperone; NNC 112 = (+)-5-(7-benzofuranyl)-8-chloro-7-hydroxy-3-methyl-2,3,4,5-tetrahydro-1H-3-benzazepine; SCH 23390 = ((R)-(+)-8-chloro-2,3,4,5-tetrahydro-3-methyl-5-phenyl-1H-3-benzazepin-7-ol); TRODAT-1 = [2-[[2-[[[3-(4-chlorophenyl)-8-methyl-8-azabicyclo[3,2,1]oct-2-yl]methyl](2-mercaptoethyl)amino]-ethyl]amino]ethanethiolate(3-)-N2,N,S2,S2']oxo-[1R-(exo-exo)]; WIN 35,428 = 2ß-carbomethoxy-3ß-(4-fluorophenyl)-tropane.
|
|
|
EXAMPLES FROM COMMON NEUROPSYCHIATRIC DISORDERS
|
|---|
Use of PET, SPECT, and, to a lesser extent, MR spectroscopy in the diagnosis and therapeutic monitoring of patients with a neuropsychiatric disease has been the subject of reviews (61–63). Also abundant are reviews on the use of these techniques in CNS drug development (39,64,65). Thus, only the most fundamental principles will be discussed herein.
Alzheimer Disease
PET and SPECT have long been used to assess various forms of dementia, with AD being the most common subtype. AD is progressive and is commonly unrecognized in its early stages (66). In AD, as in many other neurologic diseases, functional changes usually precede the anatomic changes seen on CT and MR images, attesting to the need for functional and molecular surrogate markers and/or biomarkers (67). The more we understand the pathologic nature of AD, the better equipped we become to design highly selective molecular neuroimaging agents. At pathologic analysis, the number of cortical neurons decreases with concurrent deposition of amyloid plaques and neurofibrillary tangles in patients with AD. Amyloid plaques are extracellular deposits of ß-amyloid protein from abnormal metabolism of amyloid precursor protein, which is found normally in cell membranes and membranes of intracellular organelles (68,69). Neurofibrillary tangles are intracellular aggregates of tau proteins, which are the hyperphosphorylated forms of normal components of the neuroskeleton. Tau proteins interfere with the assembly and stability of the neuroskeleton (70).
The cause of AD remains controversial. Recently, the apolipoprotein 
4 allele, which is a gene on chromosome 19, has emerged as a putative marker for AD, and its presence has been associated with an increase in ß-amyloid plaque deposition (71) and earlier onset of AD (72). Genetic profiling is important so that suitable patients can be chosen to undergo further assessment with noninvasive techniques such as imaging. This becomes important because there are more than 60 causes of dementia, and further supporting evidence of AD may help prevent this entity from being merely a diagnosis of exclusion. For example, patients with the apolipoprotein 4 allele may benefit from early PET performed before the onset of symptoms (67), as it was shown that such patients have a lower regional cerebral glucose metabolic rate than do control subjects (73). As medications used to treat AD continue to emerge and because these medications have demonstrated an ability to improve memory or at least delay the deterioration of cognitive function, imaging could aid in determining patient prognosis (74), disease progression, and efficacy of various medication regimens (75–78).
[18F]Fluorodeoxyglucose (FDG) PET is the mainstay of both oncologic metabolic imaging and evaluation of neuropsychiatric disease with PET. Classically, patients with AD have a pattern of bilateral parietotemporal hypometabolism that is not generally seen in patients with other forms of dementia or in age-matched control subjects. Asymmetry of the metabolic deficits is not uncommon. There is typically sparing of the basal ganglia, thalamus, cerebellum, and primary sensory cortex (79). Even early in the disease process, before the appearance of volume loss, FDG PET has been helpful in diagnosing AD, with a sensitivity and specificity of about 90%, irrespective of the degree of cognitive impairment (80–82). In fact, abnormal parietotemporal uptake patterns were seen on FDG PET images obtained in asymptomatic members of families in which a familial form of early AD was present (83). Similarly, asymptomatic subjects with the apolipoprotein 4 allele were found to have significantly less parietotemporal metabolic activity than those without this allele (84).
FDG PET has also proved useful for following the disease course, with a negative PET study indicating that pathologic progression of cognitive impairment during the 3-year follow-up period was unlikely to occur (84). In another study, investigators were able to predict reliably the clinical course of patients with mild cognitive impairment by using FDG PET images obtained early in the course of disease (85). A wide variety of other established radiopharmaceuticals have been used to study AD and have been discussed elsewhere (30,86,87); however, the use of the isoquinoline [11C]R-PK11195 deserves note (88,89). [11C]R-PK11195 binds to the peripheral benzodiazepine receptor within the outer mitochondrial membrane of activated microglia, which are mediators of CNS inflammation (90). A hypothesis regarding the pathophysiology of AD is that activated microglia may produce cytokines that in turn produce neuronal damage, perhaps mediating neuronal damage caused by ß-amyloid protein. Consistent with this hypothesis are the salutary effects noted with anti-inflammatory drugs in patients with AD and the fact that [11C]R-PK11195 binding is significantly increased in the temporal cortex of patients with AD relative to that in age-matched control subjects (88).
Perhaps more in spirit with current molecular imaging research is the search for newer radiotracers that can be used in the evaluation of patients with AD. Among those agents is the class of compounds that bind to the amyloid plaque. Although the precise binding mechanism has not been determined for any amyloid plaque-binding radioligands, which range in size from small molecules to peptides (91,92), these radioligands tend to behave as receptor-based radiopharmaceuticals.
One compound that has proved useful in this area is 2-[18F]FDDNP (2-(1-{6-[(2-[18F]fluoroethyl)(methyl)amino]-2-naphthyl}ethylidene)malononitrile), which binds to both the plaques and the neurofibrillary tangles (93,94). The relative residence time of the radiotracer in brain regions affected by AD was significantly greater in patients with AD than in control subjects. Consistent with the FDG PET findings, brain regions with low glucose metabolism were generally matched with those that demonstrated higher retention of 2-[18F]FDDNP (95).
Another promising ligand is Pittsburgh compound B (N-methyl-[11C]2-(4-methylaminophenyl)-6-hydroxybenzothiazole) (hereafter, [11C]PIB), which also binds to amyloid plaques (96) (Fig 2). In one study, 16 patients with mild AD showed marked retention of [11C]PIB in areas of association cortex compared with nine healthy control subjects. As with 2-[18F]FDDNP, [11C]PIB retention in the cortex correlated inversely with cerebral glucose metabolism (97).
View larger version (66K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2: Coronal (top), transverse (middle), and sagittal (bottom) volumetric spoiled gradient-recalled MR images (repetition time msec/echo time msec, 25/5; 40° flip angle; 1.5-mm section thickness; no intersection gap) and corresponding standardized uptake value PET images obtained with 11C-PIB in a patient with AD and an age-matched control subject. Standardized uptake values were obtained by normalizing tissue concentration (measured in nanocuries per milliliter [1 nCi = 37 Bq]) by injected dose (measured in nanocuries) and body mass (measured in milliliters). Note that higher 11C-PIB uptake is present in the frontal, parietotemporal, and—to a lesser extent—occipital regions in the patient with AD compared with the control subject. 11C-PIB uptake reflects ß-amyloid deposition. (Image courtesy of C. A. Mathis, PhD, University of Pittsburgh, Pittsburgh, Pa.)
|
|
The initial clinical success of amyloid plaque imaging provides a useful lesson for molecular imaging research. The prototype ligand on which [11C]PIB was based was not blood-brain barrier penetrable and was modeled after a dye used to delineate plaque on pathologic specimens (98,99). Molecular imaging performed with this technique is merely an in vivo extension of the histologic method and proved successful only after structural optimization of chrysamine dyes (100).
The extension of in vitro probes and techniques to in vivo imaging is a recurrent theme in molecular imaging research. Notably, a version of this technology has also been extended to near-infrared fluorescence imaging in small-animal models of AD (101). The importance of this is the relative ease with which near-infrared fluorescence imaging can be performed relative to MR imaging or PET, in addition to an improved signal-to-noise ratio due to less background signal. Currently, near-infrared fluorescence–based plaque imaging is only a preclinical tool; however, it could become clinically viable when coupled with fluorescence molecular tomography.
Parkinson Disease
Unlike AD, the metabolic changes in patients with PD or other movement disorders are less specific and can overlap with findings in patients with AD (102). The DA system has been thoroughly evaluated in patients with PD, since damage to nigrostriatal neurons is the most important element in the pathophysiology of PD (103). Certain aspects of DA transmission seen in patients with PD use the radiotracer [18F]fluoro-DOPA ([18F]FDOPA), which is metabolized by DOPA decarboxylase and stored in presynaptic vesicles (104). Thus, [18F]FDOPA is used to measure three aspects of DA transmission: DA transport to the presynaptic neuron, DOPA decarboxylase activity, and DA storage (Fig 1). Nevertheless, [18F]FDOPA imaging is likely to result in underestimation of the degree of nigrostriatal damage in patients with PD because DOPA decarboxylase activity may initially be increased as an adaptation by the residual surviving cells (105). However, [18F]FDOPA imaging has proved quite useful in the quantification of changes seen in patients with PD, with markedly decreased uptake compared with that in control subjects being noted in the striatum (106,107). Other aspects of DA transmission seen in patients with PD include estimation of presynaptic transporter sites with the tropanes (-)-2ß-carbomethoxy-3ß-(4-fluorophenyl)tropane naphthalenedisulfonate (or CFT) (108) and (-)-2ß-carbomethoxy-3ß-(4-iodophenyl)tropane (or ß-CIT) (109,110).
The use of DA probes in patients suspected of having PD has proved to be important when one attempts to differentiate PD from other causes of parkinsonian symptoms, such as hypermanganesemia. Some investigators have proposed that because the presynaptic neurons of the substantia nigra are normal in manganese toxicity, normal [18F]FDOPA PET images support a diagnosis of hypermanganesemia rather than PD (111,112). This is a contentious finding (113), as few subjects have been studied; however, this example serves to underscore the practical importance of molecular neuroimaging not only in individual diagnoses but also with respect to occupational health.
Other radiotracers that have been used successfully in the evaluation of patients suspected of having PD include [11C]dihydrotetrabenazine (DTBZ), which has been used to estimate the density of presynaptic type II vesicular monoamine transporters and [11C]R-PK11195. Use of [11C]DTBZ has been successful in the evaluation of the severity, progression, and therapeutic response of disease (114,115), while use of [11C]R-PK11195 has enabled researchers to uncover areas of increased uptake in the brainstem that were not previously identified in patients with PD and that possibly point toward new or understudied aspects of pathophysiology (116). This finding is consistent with findings that suggest that PD may actually initiate within the medulla rather than within neurons of the substantia nigra (117).
In brief, as with AD, many aspects of PD, including early detection, follow-up for disease progression, therapeutic monitoring, and differential diagnosis, can be studied quantitatively with PET and SPECT. In the case of PD, this is due in part to 20 years of careful characterization of nearly every aspect of DA transmission with radionuclide-based molecular imaging. Receptor-based imaging for other neurologic disorders, such as epilepsy (118–120), Huntington disease (121,122), and human immunodeficiency virus dementia (33,123), have been described elsewhere.
Psychiatric Diseases
Both radionuclide and MR-based molecular neuroimaging techniques have proved useful in the study of psychiatric diseases, particularly schizophrenia and depression. Excellent reviews exist in which the use of conventional agents in patients with those entities is described, with DA transmission in patients with schizophrenia and serotonergic transmission in patients with depression having been thoroughly evaluated with a variety of radiotracers and imaging modalities (47,124–132). As stated previously, the outcome parameters derived from such studies generally require radiotracer kinetic modeling approaches of varying levels of complexity (36,133). Kinetic modeling was initially complex and, in the case of schizophrenia, may be partially responsible for the contradictory results obtained by various research groups in early studies (134). Modeling techniques that use a surrogate input function or reference tissue method are being increasingly used, allowing easier comparison between research groups (135). Higher-resolution imaging devices and more reliable radioligands with less nonspecific binding, such as [11C]-3-amino-4-(2-dimethylaminomethylphenylthio)benzonitrile (or [11C]DASB), which is used to study serotonin transporters, also are contributing to more robust receptor-based neuroimaging (136,137).
On the horizon for neuropsychiatric imaging is the study of glutamatergic transmission with PET. Promising agents exist for the metabotropic glutamate subtype 5 receptor and for glutamate carboxypeptidase II (138–140). Imaging glutamatergic neurotransmission is important because of the modulatory role suggested for glutamate, which may affect its contribution to the pathogenesis of AD, depression, pain, and schizophrenia (141–146).
|
RECEPTOR OCCUPANCY
|
|---|
Of particular relevance to psychiatry, however, is the ability to measure receptor occupancy with PET (52,60). Receptor occupancy is measured in the pharmaceutical industry to determine dosage regimens for new pharmaceuticals before phase 1 trials and sometimes in phase 1 and phase 2 clinical trials. The principle is simple: The receptor-binding radiopharmaceutical is administered some time after administration of the intended drug. The radiopharmaceutical must bind to the same receptor at which the drug mediates its action. At higher levels of drug receptor occupancy, lower levels of the radiopharmaceutical bind to the target sites. Images obtained at baseline are compared with those obtained at peak drug receptor occupancy. Receptor occupancy is then calculated from the difference in binding potentials between the two sets of images (Fig 3).
View larger version (104K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3: Brain PET images obtained with 11C-RAC before (left image) and after (right image) the administration of the antipsychotic drug aripiprazole at a dose of 30 mg/d for 14 days. The striatal uptake of 11C-RAC was lower after aripiprazole administration than at baseline, indicating the drug occupied D2 and D3 receptors. Treatment with aripiprazole for 14 days resulted in a receptor occupancy of approximately 95%. (Reprinted, with permission, from reference 52.)
|
|
One protocol may involve performing baseline brain imaging with the imaging agent to measure receptor occupancy and then obtaining additional images at three or four time points after drug administration. The drug is generally administered orally, while the radiopharmaceutical is always administered intravenously. Currently, a suitable dose for a new CNS-active agent is determined by administering various doses to patients and determining the plasma drug concentration. However, plasma drug concentration may not reflect the drug concentration in the brain, which is why calculating receptor occupancy at the target site with imaging may be more relevant (147). The receptor occupancy may then be related to the plasma concentration so that patients may be followed up carefully, and effective doses that are smaller than those that may cause serious side effects, such as tardive dyskinesia, may be provided. For most CNS-active agents, receptor occupancy of more than 50% has been determined to provide an effective dose (148). Receptor occupancy can be determined with imaging in particular patients, if one assumes that the pharmacokinetic profile of various drugs will be different from patient to patient. That powerful approach will prove useful in the evolving area of personalized medicine. The calculation of receptor occupancy is one of the most powerful examples of the utility of molecular imaging in CNS drug development on a large scale for clinical trials and on a small scale for personalized patient therapy. All of the principles used to quantify drug delivery to target that have been applied to the CNS for more than 20 years may be extended to cancer, inflammation, and cardiovascular disease.
|
MOLECULAR IMAGING IN NEURO-ONCOLOGY
|
|---|
Since the National Cancer Institute has provided the initial funding for In Vivo Cellular and Molecular Imaging Centers and Small-Animal Imaging Resource Programs, a substantial amount, if not the majority, of molecular imaging research has been undertaken to study cancer. Although many new reagents and techniques continually emerge from these centers and programs, molecular neuroimaging of brain tumors has involved radionuclide- and MR-based techniques for many years (35,149–153).
A variety of radiopharmaceuticals, including FDG, have been used to study brain tumors. FDG demonstrates increased uptake in malignant lesions when compared with surrounding brain parenchyma (154). However, because of the high glucose metabolism of normal gray matter, FDG PET is not part of the routine evaluation of patients who present with signs and symptoms of a brain tumor. Other uses of FDG PET in the evaluation of patients suspected of having a brain tumor include grading, localization for biopsy, differentiation of radiation necrosis from tumor recurrence, therapeutic monitoring, and assessment for malignant transformation of what were originally low-grade gliomas (76,155–158). Despite these uses, FDG PET is infrequently used to study brain tumors in clinical practice. Perhaps the only use of FDG PET in this context worth further discussion is its use in the differentiation between radiation necrosis and tumor recurrence (Fig 4). In 47 patients with a variety of brain tumors, the sensitivity and specificity of FDG PET in the differentiation of a tumor demonstrating elevated FDG uptake from radiation necrosis, in which uptake is diminished, were 75% and 81%, respectively (157). However, in another study, sensitivity and specificity were lower, which led researchers to conclude that FDG PET had a limited role in this indication (159).
View larger version (119K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a: Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (a) MR and (b) FDG PET images of a patient with glioblastoma show abnormal enhancement (arrows) in the left frontal lobe corresponding to increased FDG uptake (arrowhead in b) along the posterolateral aspect of the lesion. These findings are compatible with tumor recurrence. Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (c) MR and (d) FDG PET images of another patient with anaplastic oligodendroglioma show an abnormally enhancing lesion in the right frontal lobe (arrowheads in c) with no appreciable corresponding FDG uptake. These findings are compatible with radiation necrosis. (PET images courtesy of H. A. Jacene, MD, Johns Hopkins University.)
|
|
View larger version (109K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b: Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (a) MR and (b) FDG PET images of a patient with glioblastoma show abnormal enhancement (arrows) in the left frontal lobe corresponding to increased FDG uptake (arrowhead in b) along the posterolateral aspect of the lesion. These findings are compatible with tumor recurrence. Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (c) MR and (d) FDG PET images of another patient with anaplastic oligodendroglioma show an abnormally enhancing lesion in the right frontal lobe (arrowheads in c) with no appreciable corresponding FDG uptake. These findings are compatible with radiation necrosis. (PET images courtesy of H. A. Jacene, MD, Johns Hopkins University.)
|
|
View larger version (119K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4c: Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (a) MR and (b) FDG PET images of a patient with glioblastoma show abnormal enhancement (arrows) in the left frontal lobe corresponding to increased FDG uptake (arrowhead in b) along the posterolateral aspect of the lesion. These findings are compatible with tumor recurrence. Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (c) MR and (d) FDG PET images of another patient with anaplastic oligodendroglioma show an abnormally enhancing lesion in the right frontal lobe (arrowheads in c) with no appreciable corresponding FDG uptake. These findings are compatible with radiation necrosis. (PET images courtesy of H. A. Jacene, MD, Johns Hopkins University.)
|
|
View larger version (93K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4d: Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (a) MR and (b) FDG PET images of a patient with glioblastoma show abnormal enhancement (arrows) in the left frontal lobe corresponding to increased FDG uptake (arrowhead in b) along the posterolateral aspect of the lesion. These findings are compatible with tumor recurrence. Postcontrast transverse T1-weighted (550/9, 5-mm section thickness, 5-mm intersection gap) (c) MR and (d) FDG PET images of another patient with anaplastic oligodendroglioma show an abnormally enhancing lesion in the right frontal lobe (arrowheads in c) with no appreciable corresponding FDG uptake. These findings are compatible with radiation necrosis. (PET images courtesy of H. A. Jacene, MD, Johns Hopkins University.)
|
|
Clearly, superior radiopharmaceuticals are needed for neuro-oncology. Among the most promising radiopharmaceuticals are those that target amino acid transport or are incorporated within DNA. Unlike FDG uptake, which reflects the increased glucose metabolic rate of malignant cells, radiolabeled amino acids can be used to assess protein synthesis within tumors and the integrity of the blood-brain barrier (154). One such tracer is L-[methyl-11C]methionine ([11C]MET). However, [11C]MET, like tyrosine, can also be incorporated into nonprotein products, adding to the complexity of radiotracer kinetic modeling of these agents (160). The primary advantage of amino acid–based agents is the fact that their uptake in a normal brain is low, providing enhanced tissue contrast between tumor and normal parenchyma, unlike FDG. Many studies have shown [11C]MET to have high sensitivity in the detection of brain tumors, with levels as high as 98% having been documented (154,161–163) (Fig 5). [11C]MET can also help in the differentiation of (a) neoplastic from nonneoplastic lesions (164) and (b) radiation necrosis from tumor recurrence (156,165). Tsuyuguchi et al (166) documented 100% sensitivity, 60% specificity, and 82% overall accuracy in the detection of tumor recurrence with [11C]MET PET.
View larger version (135K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5: [11C]MET PET, transverse enhanced MR, and FDG PET images of two patients with glioma. Top: [11C]MET PET image shows increased uptake (white arrows), but FDG PET image shows iso- and hypometabolism (black arrows) corresponding to the left-sided enhancing tumor (open arrows). Bottom: Low-grade glioma treated with local resection and chemotherapy. MR image shows residual nonenhancing tumor in the right insula (open arrows). [11C]MET PET image shows a focal area of focally increased uptake (white arrow). This area was confirmed to be anaplastic transformation with a relatively higher proliferative index. FDG PET findings were discordant with [11C]MET PET findings, which demonstrated hypometabolism (black arrows) similar to that of white matter, consistent with a low-grade glioma. (Reprinted, with permission, from reference 161.)
|
|
Another agent that has shown promise in the assessment of the proliferation of tumor cells and that may be useful in the assessment of brain tumors is 3'-deoxy-3'-[18F]fluorothymidine ([18F]FLT). Although [18F]FLT is not incorporated into DNA, it is trapped in the cell because of phosphorylation by thymidine kinase, which is a part of the proliferation pathway. As such, it has the potential to enable imaging of a proliferating tumor in proportion to the DNA synthesis rate. In the case of the brain, [18F]FLT is not taken up in normal brain cells due to essentially no proliferative activity and an intact blood-brain barrier. This gives [18F]FLT a distinct advantage compared with FDG, as it will only be concentrated in a proliferating tumor and thus allows for improved differentiation of necrosis from recurrence and differentiation of low- and high-grade brain tumors. Recently, several reports have been published that show the potential clinical utility of this compound in brain tumors (167–169).
The use of SPECT in brain tumor imaging is limited. Thallium 201 (201Tl), which is an analog of the sodium ion, is the most studied radiotracer, with the results of some studies demonstrating direct correlation between 201Tl uptake and tumor grade (170). However, there is marked overlap between tumor uptake and histologic grades, making 201Tl, like FDG, not particularly useful for single-agent diagnosis. Like FDG, 201Tl may be used to differentiate tumor recurrence from radiation necrosis (171,172).
As stated previously, MR-based conventional molecular neuroimaging has been applied to brain metabolism and tumors for more than 20 years, dating back to early studies of phosphorous spectroscopy (15,173). More frequently, and almost routinely in clinical practice, 1H MR spectroscopy and MR spectroscopic imaging are used to evaluate brain tumors. In this case, the critical marker is an elevated tumor choline concentration (152,153,174). Although the MR-based techniques are capable of extremely high spatial resolution (ie, on the order of 10 µm for MR microscopy [Table]), they suffer from a lack of sensitivity in the detection of molecular species, where millimolar (millimole per liter) concentrations are generally required. Furthermore, only a limited number of nuclei can be studied. Although proton spectroscopy can be used to grade tumors, the technique used to do so is somewhat difficult to perform routinely (175). However, grading is not necessarily the most important aspect of in vivo tumor assessment. It is likely that the ability to distinguish tumors from radiation necrosis, benign lesions, or edema is more important. With the advent of MR spectroscopic imaging, the tumor margin can be carefully studied. Tumor and vasogenic edema are often indistinguishable on images acquired with a long repetition time, so the ability to investigate those peritumoral T2 high-signal-intensity regions metabolically is particularly desirable. Brain MR spectroscopic imaging is being pursued avidly at many clinical centers for use in the diagnosis of disease and the direction of tumor biopsy and treatment planning (176,177). Conventional MR-based techniques, including spectroscopy, that are used to image brain tumors have been reviewed extensively (178,179).
Sodium imaging is a less common but equally clinically applicable technique used to image brain tumors (180). Sodium imaging has been performed for nearly 20 years (18); however, with the advent of new pulse sequences, such as twisted projection imaging (181,182), sodium imaging may provide useful new information about the tumor and its surrounding tissue. Brain tumors are believed to demonstrate increased levels of intracellular sodium such that they appear bright on sodium MR images (180). Recently, Schepkin et al (19) revisited sodium imaging for therapeutic monitoring in a rat glioma model. In that study, animals were examined serially at baseline and after treatment with the established antiglioma therapy, 1,3-bis(2-chloroethyl)-1-nitrosourea (or BCNU). Similar to the principle of diffusion-weighted imaging, in which the water content of the tumor increases as tumor cells die, the hypothesis was that the cells would release sodium as they died, which would produce an increase in the extracellular sodium concentration detected with MR imaging. The importance of using sodium imaging for therapeutic monitoring is that early changes in tumor sodium content may be identified before anatomic changes, as has been shown for apparent diffusion coefficient mapping (183). Validation of this technique under controlled conditions in an animal model against apparent diffusion coefficient mapping, which is a well-established clinical tool, lends further support to the utility of small-animal molecular neuroimaging for improving or extending techniques that have been applied clinically.
|
SMALL-ANIMAL MOLECULAR NEUROIMAGING
|
|---|
Conventional Probes and Techniques
Thus far, we have focused on molecular neuroimaging in humans because the aforementioned conventional techniques have been performed in humans for more than 20 years. Because of the development of high-resolution and multimodality imaging devices for animal imaging and the presence of many valuable and relevant animal models of human diseases, such as AD, stroke, and PD, we are seeing increasing application of small-animal neuroimaging in which new radiotracers and methods can be tested and optimized before clinical use. Relevant genetic models have been used in radiopharmaceutical development previously; however, instead of being used merely to show access of new agents to the target sites, they are now used in the actual quantification and calculation of binding affinities and receptor concentrations (184,185).
Forsback et al (186) performed small-animal brain PET studies with [18F]FDOPA. Their results demonstrated that rodents can be used not only in ex vivo biodistribution assays but also in imaging studies. Small-animal imaging has also been used to assess the viability of DA neuronal transplants in rodent models of neurodegenerative disorders such as Huntington disease (187) (recently performed in humans [188]), for absolute quantification of glucose metabolic rate in rat brain (189), and to study the effects of anesthesia on radiopharmaceutical uptake (190). Toyama et al (191) used positron-emitting ß-amyloid probes, such as [11C]OH-BTA-1 ([N-methyl-11C]2-(4'-methylaminophenyl)-6-hydroxybenzothiazole), to study transgenic AD mice. The results of their study demonstrated less binding to plaque than expected. They surmised that the model did not reflect human AD and produced far fewer plaques or that [11C]OH-BTA-1 had lower affinity for the plaques produced. Ma et al (192) used [11C]N-methylpyrrolidinyl benzilate to determine if rodent PET could be used to detect changes in endogenous acetylcholine concentrations. The study of Ma et al (192) is particularly important because the findings demonstrated that it may be more difficult to detect subtle findings, such as endogenous neurotransmitter changes, in vivo with current commercially available scanners as compared with use of traditional ex vivo biodistribution assays.
Hume et al (185) discussed the need for high specific radioactivity of radiopharmaceuticals for quantitative brain PET imaging in rodents so that the tracer principle will not be violated, since rodents have brain regions with a limited number of receptors that may be readily saturated by material produced in low specific radioactivity. Such theoretical treatments have led to a resurgence and interest in the development of materials with high specific radioactivity (185). Besides target saturation, another potential problem with small-animal PET involves limits in spatial resolution due to positron path range. After correcting for noncolinearity, positron path range, and other factors that degrade the spatial resolution of PET studies, it is widely believed that the maximum spatial resolution achievable with PET will be on the order of 0.5 mm. Although that spatial resolution is relatively high, it is not nearly as high as that obtainable with MR imaging and CT and may limit the usefulness of molecular neuroimaging with PET in small-animal models to study subregions of the brain, such as the hippocampus or the superior colliculus. Small-animal brain SPECT imaging is not hampered by such problems, as there is no annihilation event that needs to be detected. In fact, Acton et al (193) demonstrated submillimeter spatial resolution of mouse brain SPECT images with the presynaptic transporter ligand [99mTc]TRODAT ([2-[[2-[[[3-(4-chlorophenyl)-8-methyl-8-azabicyclo[3,2,1]oct-2-yl]methyl](2-mercaptoethyl)amino]-ethyl]amino]ethanethiolate(3-)-N2,N,S2,S2']oxo-[1R-(exo-exo)]) with full quantification. A drawback of the SPECT studies, however, is that relatively large quantities of radioactivity must be administered, the biologic effects of which are currently unknown and understudied. To avoid the spatial resolution limitations of small-animal radionuclide-based studies, some researchers have turned to detection of the actual ß particles emitted by the radiotracer; however, the ß microprobe, which is used to detect such particles, provides only regional radioactivity counts (194) and does not provide images. Small-animal brain imaging in conjunction with microdialysis is particularly powerful in the validation of neurochemical physiology, as demonstrated by Schiffer et al (195), who investigated the threshold dose at which a pharmacologic response to radiotracer occurred.
Cell Trafficking
Perhaps the most rapid growth in molecular neuroimaging has occurred in the area of cell trafficking (5,7,196,197). A wide variety of techniques are currently available that can be used to introduce suitable labels to cells that can be detected with one of the molecular imaging modalities. The cells, labels, and modalities used are complementary and are chosen with respect to the indication pursued (198,199). The probes to be detected can be divided into roughly two categories: passive and active labeling versus genetically modified. The first category of passive labeling can take advantage of either radioactive or MR-detectable probes that are introduced into the cell in vitro and can then be studied in vivo. That would be considered a passive cellular labeling approach. Detectable particles may also be introduced into cells in vivo according to the so-called active labeling paradigm for robustly phagocytic cells, such as macrophages. The idea behind these techniques is not new and has been used in nuclear medicine clinical practice for many years to observe red blood cells in gastrointestinal bleeding or the disposition of white blood cells within an inflammatory focus (200). What is new, however, is the introduction of MR and optical imaging to the area of cell trafficking in vivo, as well as the proliferation of these techniques to applications within the CNS.
Aside from active (phagocytic) or passive (external cellular) labeling, cells can also be labeled by using a genetic mechanism. With this method, a transgene can be introduced that will be detected after exogenous administration of a radiolabeled probe or light-emitting compound that serves as a substrate for the product of the introduced transgene. Alternatively, cells may be modified in the germ line during the production of transgenic reporter mice. These cells may harbor tissue and even cell-specific promoters so that only a small fraction of the cell population may be tracked in vivo without needing to administer an exogenous probe. Reporter mice have been engineered to study a variety of genetic events (201), including cytochrome p450 3A4, which is important in drug metabolism (202). Such mice are now commercially available (25). Because of the lack of background activity, bioluminescence of transgenic cells within the brains of intact animals may be detected easily. Initially, the gene-encoding firefly luciferase was introduced to glioma cells, which were then orthotopically transplanted and followed in vivo for therapeutic monitoring of antineoplastic drugs (203). That technology has been extended to depiction of the trafficking of transgenic neural stem cells as a result of the presence of a brain tumor both for imaging and for therapeutic purposes, as will be discussed. The use of fluorescent optical reporters has required the use of a cranial window for imaging intracranial cellular trafficking; however, when this technique is used with near-infrared fluorescence probes and intravital microscopy, tremendous detail can be obtained, as in the study of a new near-infrared fluorescence probe for use in the detection of venous thrombi (204) (Fig 6).
View larger version (77K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6: Colocalization of A15 near-infrared fluorescence signals and superior sagittal sinus thrombus. Intense fluorescence signal observed in the intravital fluorescence microscopy image 25 minutes after A15 injection was shown in a color map (measured in arbitrary units to represent the pixel intensity range of the overlay fluorescence imaging superimposed on a white-light image). The representative hematoxylin-eosin–stained (inlet) transverse section (vertical gray line) of the superficial brain at the location of the frontal tip of the thrombus signal (white arrow) shows corresponding thrombus filling the dural venous sinus (black arrow). In addition to the main body of the thrombus within the superior sagittal sinus, several scattered thrombi are observed in the adjacent vessels. (Reprinted, with permission, from reference 204.)
|
|
Refinements in MR imaging for cell trafficking over the past decade have culminated in two clinical trials outside of the CNS, with applications in the CNS soon to follow (205,206). Basic questions must be answered before this technique can be considered reliable. How can the label be introduced to the cell? Does labeling interfere with the differentiation, repair capacity, or other relevant function of these cells? A variety of methods can be used to introduce the label to the cells, including targeting transferrin receptor (5), linking with the human immunodeficiency virus–Tat protein (207), encapsulation within dendrimers (4), in vivo fluid-phase pinocytosis (208), use of transfection agents such as protamine sulfate (209), and—most recently—rediscovery of electroporation with a technique referred to as magnetoelectroporation (210). In the latter study, Walczak et al (210) demonstrated that nonphagocytic cells can be instantly labeled with ferumoxides, without the need to use adjunct transfection agents and prolonged incubation, thereby facilitating clinical implementation. Alternatively, the use of transfection agents, such as the Food and Drug Administration–approved protamine sulfate, along with superparamagnetic iron oxide nanoparticles (ie, ferumoxides) to label stem cells has been successfully attempted and appears promising (211).
Cells labeled for CNS applications include neural stem cells, oligodendrocyte precursors, and macrophages (3,5,212). The need for stem cell grafting derives from the limited regenerative capacity of neural tissue and the interest in stem cell therapy for replacing such damaged tissue. The MR-based approach is sensitive after optimization of the correct particles to use and pulse sequences with which to image. It is widely believed that iron oxide–containing particles tend to be more stable and provide stronger contrast than gadolinium-based agents, despite the positive MR signal provided by the latter. Superparamagnetic iron oxide particles are also nontoxic, and because of their 20–100-nm diameter, they provide large field gradients easily seen on gradient-echo images. The MR technique is also amenable to quantification through T2* relaxometry, and the application of new pulse sequences, such as fast imaging with steady-state acquisition, is continuing to improve the sensitivity of detection (3,213). Perhaps most important, the MR technique is readily translatable and can be performed with clinical magnets, the standard of which is increasing in field strength to 3 T, which will further improve the sensitivity for detection. In fact, one cell can be imaged in vivo with a clinical magnet and superparamagnetic iron oxide particles with an optimized imaging sequence (214).
One rationale for performing such in vivo trafficking studies derives from the clinical need to optimize stem cell therapy protocols. Optimization of those protocols involves knowing how many cells to administer and determining how many cells will eventually survive and function. The route of cell delivery (ie, intracerebroventricular vs intraparenchymal administration), can also be investigated with these techniques. Histologic follow-up of these techniques would be prohibitively complex and would require many animals and many brain slices at great expense over multiple time points. Furthermore, the MR-based techniques do not generally label all of the cells, particularly for the so-called active labeling techniques, further rendering histologic approaches impractical. This topic has been reviewed (215).
Bulte et al (5) used MR technology in cell trafficking. They observed prelabeled oligodendrocyte precursors along the axons in the spinal cord in a rodent model of dysmyelinating disease. These progenitor cells could be followed for several weeks after administration with this technique. Similarly, Schwann cells labeled ex vivo with fluid-phase pinocytosis of superparamagnetic iron oxide particles were shown to persist for up to 4 weeks (208). Neuroinflammatory cells such as macrophages can be labeled with an active technique, as performed by Oweida et al (3), in which the inflammatory response in a mouse model of demyelination could be studied with a clinical 1.5-T magnet. One issue that needs to be addressed is how to get these nanoparticles across the blood-brain barrier if they are to be administered intravenously rather than introduced directly into the parenchyma. Veiseh et al (10) developed a dual-modality MR-optical nanoprobe that targets glioma and is internalized via the chlorotoxin receptor. Their findings show the emerging relationship between cell trafficking and nanobiotechnology. In summary, the MR-based techniques are attractive because of clinical translatability, high sensitivity, and lack of toxicity or radiation dose to the patient. Nevertheless, such techniques suffer from the potential to alter the function of cells containing the label, as shown in one study (216), as well as the dilution of the label as cells differentiate. The optical and radionuclide-based (transgenic) techniques used for cell trafficking have the potential to obviate at least the latter problem.
Bioluminescent and fluorescent transgenic approaches to cell trafficking are based on the principle of reporter gene imaging, which we will elaborate on in the next section. Similar to techniques used for years in vitro, production of a protein that can be imaged, such as enhanced green fluorescent protein, ß-galactosidase, choline phenylace-tyltransferase or luciferase, is linked stoichiometrically to that of a gene of interest (217,218). For the purpose of cell trafficking, the gene to be imaged need only be expressed constitutively. If that gene has been stably transfected into the cell to be imaged, such as a neural stem cell, a constitutive promotor, such as that from a cytomegalovirus or elongation factor 1, may be used to affect not only trafficking but also clonal expansion of suitably transfected cells. Investigators at the Memorial Sloan-Kettering Cancer Center (New York, NY) have transduced human embryonic stem cell–derived neural precursor cells with a triple-reporter construct that produces firefly luciferase and green fluorescent protein (219). Serial bioluminescence imaging revealed cell viability after 2 months in NOD/SCID mice. Immunohistochemistry revealed that cells that were grafted to the striatum of mice not only remained viable but also migrated to the subventricular zone, which is a site of neurogenesis (Fig 7).
View larger version (66K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7: Images obtained with in vivo long-term monitoring of striatal graft function after stereotactic implantation of human embryonic stem cell–derived neural precursor cells transduced with a triple reporter lentiviral construct containing luciferase and green fluorescent protein. A, Serial bioluminescence image of striatally-implanted luciferase and human embryonic stem cell–derived neural precursor cells over a 2-month period in NOD/SCID mice (black circles with black or gray dashed lines), with additional transduced cells injected into the right thigh region as a reference standard. B–D, Corresponding images of a representative mouse are shown at various times (inset in B). Data are the average rate of photon emission ± standard deviation (measured in photons per second). Two months after transplantation, graft immunohistochemical analysis was performed. A high-power fluorescence image of the intact graft after staining with antibodies to green fluorescent protein (B, green) and human nuclear antigen (B, red) demonstrates a viable graft with many positive colabeled cells (inset). Migration of a small number of cells to the, C, subventricular zone, a site of neurogenesis, as well as within the, D, corpus callosum was noted within the host tissue. (Images courtesy of M. Bradbury, MD, PhD, Memorial Sloan-Kettering Cancer Center, New York, NY.)
|
|
Optical techniques have also been used to evaluate cell-based imaging and therapeutic techniques. In one indirect case, a dual-modality MR optical probe was used to delineate the tumor margins for intraoperative guidance (220). This is particularly important in the case of brain tumors because of their inherently infiltrative nature and the fact that recurrence tends to occur in the surgical bed. In this case, Kircher et al (220) used a nanoparticle known as Cy5.5-CLIO that emits in the near-infrared range and, by virtue of the CLIO moiety, can be detected with MR imaging. Animals with orthotopic green fluorescent protein–expressing 9-L glioma tumors underwent craniotomy to expose tumor and surrounding tissue 24 hours after intravenous injection of Cy5.5-CLIO. In this active cell–labeling approach, phagocytic cells, such as microglia at the tumor margin, were used to sequester the probe, enabling both preoperative MR and intraoperative near-infrared fluorescence imaging. This technique is eminently translatable to the clinic, again because of the lack of toxicity of the administered probes and the fact that devices are currently being developed for intraoperative surgical guidance in a limited field (221). A second example of the optical approach to cellular trafficking by the same group blurs the boundaries between diagnosis and therapy. Shah et el (8) developed a method by which neural progenitor cells travel from one side of the brain toward a tumor in the contralateral hemisphere. Because the cells were also concurrently expressing the S-TRAIL protein and because the glioma was expressing Renilla luciferase, the decrease in tumor burden could be detected when treated with this novel cellular therapy (Fig 8).
TOP
ABSTRACT
INTRODUCTION
