Brain Receptor Imaging*

Receptors have a prominent role in brain function, as they are the effector sites of neurotransmission at the postsynaptic membrane, have a regulatory role on presynaptic sites for transmitter reuptake and feedback, and are modulating various functions on thecellmembrane.Distribution, density,andactivityofreceptors in the brain can be visualized by radioligands labeled for SPECT and PET, and the receptor binding can be quantiﬁed by appropri-ate tracer kinetic models, which can be modiﬁed and simpliﬁed for particular application. Selective radioligands are available for the various transmitter systems, by which the distribution of these receptors in the normal brain and changes in receptor binding during various physiologic activities or resulting from pathologic conditions can be visualized. The quantitative imaging for several receptors has gained clinical importance—for example, dopamine (D 2 ) receptors for differential diagnosis of movement disorders and for assessment of receptor occupancy by neuroleptics drugs; serotonin (5-hydroxytryptamine, 5-HT) receptors and the 5-HT transporter in affective disorders and for assessment of activity of antidepressants; nicotinic receptors and acetylcholinesterase as markers of cognitive and memory impairment; central benzodiazepine-binding sites at the g -aminobutyric acid A (GABA A ) receptor complex as markers of neuronal integrity in neurodegenerative disorders, epilepsy, and stroke and as the site of action of benzodiazepines; peripheral benzodiazepine receptors as indicators of inﬂammatory changes; opioid receptors detecting increased cortical excitabil-ity in focal epilepsy but also affected in perception of and emotional response to pain; and several receptor systems affected in drug abuse and craving. Further studies of the various trans-mitter/receptor systems and their balance and infraction will im-prove our understanding of complex brain functions and will provide more insight into the pathophysiology of neurologic and psychiatric disease interaction.

membranes that, after interaction with specific ligands (first messenger transmitters), trigger a signal causing defined responses mediated by secondary messengers (G-proteincoupled receptors) or ion channels (ligand-gated ion channels) (1). Receptors can be characterized by their affinity and density; as proteins, they are degraded after a functional period by specific proteases. The function of receptors is obvious in direct neurotransmission, where the interaction of a presynaptically released transmitter with the postsynaptic receptor causes depolarization or hyperpolarization of the postsynaptic membrane. Receptors can also be located on the presynaptic membrane-important for negative feedback and for reuptake of transmitters-and on cell bodies. Receptors involved in direct neurotransmitter function are found in high density in specific brain areas and are affected by many pathologic conditions; therefore, they form a primary target for imaging studies. Other receptors have a modulating effect on signal transduction. They can affect cell membrane enzymes and transcription factors of proteincoding genes. These receptors are distributed in lower density and their in vivo detection is still limited but might gain importance in the future. Imaging of the regional distribution of receptors provides a relevant insight into the organization of functional networks in the brain, which cannot be achieved by morphologic investigations or imaging of blood flow and metabolism. It must be kept in mind that histochemical receptor studies have demonstrated a complex laminar distribution of several receptor types in individual cortical areas, which contribute to the neurochemical organization of intracortical and cortical-subcortical networks (2). This multireceptor organization of functional networks can only be visualized by postmortem autoradiography but is not accessible for in vivo imaging studies.
Distribution, density, and activity of receptors in the brain can be visualized and quantified by radioligands (Table 1), which must fulfill several criteria to be successful for PET or SPECT (3,4): stability of labeling; sufficient affinity and high selectivity for the specific receptor combined with low nonspecific binding to brain tissue not containing the receptor of interest; rapid permeation through the bloodbrain barrier permitting high access of tracers to receptors. Additionally, there should be as few as possible metabolites of labeled ligand. If metabolites are generated in brain tissue, concentrations should be low with rapid clearance, whereas systemic metabolites present in plasma should be polar so that they cannot cross the blood-brain barrier and enter the brain.

IMAGING AND QUANTITATIVE ANALYSIS OF RADIOLIGAND STUDIES
Radioligand studies permit measurement of the time course of uptake and clearance of specific tracers but, for the assessment of pharmacokinetic parameters from these time-activity curves, a tracer kinetic model is needed, which usually requires an input function indicating delivery of tracer to the tissue. In some applications for receptor studies, an arterial input function can be replaced by a reference to tissue devoid of specific binding sites (5)(6)(7)(8)(9)(10)(11). At present, PET represents the most selective and sensitive (pico-to nanomolar range) method for measuring receptor density and interactions in vivo. For many receptor studies, SPECT can also be applied but detector efficiency is much lower and it provides less quantitative accuracy than PET. SPECT isotopes typically have longer physical half-lives than most PET tracers, which may partially compensate for their disadvantages if measurement times over several hours are required to get rid of nonspecific binding and reach equilibrium.
Imaging of radioligand distribution at some stage after intravenous injection shows the pattern of relative uptake in different regions. In some instances, this pattern represents receptor density but usually the regional signal does not only represent specific receptor binding but also contains contributions from nonspecific binding, free ligand in tissue, and intravascular activity, and all these components vary with time after tracer injection. Therefore, kinetic studies are necessary to differentiate the various components and to extract the compartment of specific binding. This implies acquisition of multiple scans for measuring cerebral uptake, clearance of the tracer, and assessment of the tracer concentration in arterial plasma over a period of 30-120 min. In addition, with some tracers metabolites must be measured for correction of plasma curves.

DETERMINATION OF RECEPTOR BINDING
For the determination of the dynamics of receptor binding, a 3-compartmental model ( Fig. 1) can be applied, with the free exchangeable ligand in plasma, the free-to-bind ligand in tissue, and the specifically bound ligand forming the compartments and kinetic constants describing the interaction among  Tracer  Abbreviation  Target   11 C-Cocaine, 11 C-methylphenidate  DAT  11 C-Nomifensine  DAT  11 C-WIN-35428, 11 C-PE2I DAT 18 F-2b-Carbomethoxy-3b-(4-fluorophenyl)tropane CFT DAT 11 C-Dihydrotetrabenazine DTBZ VMAT 2 11 C-SCH-23390 D 1 11  these compartments. Differential equations describe the change of concentrations in these compartments over time: In these equations C f denotes the fraction of tissue ligand free to bind to receptor, C a is the fraction of blood ligand free to cross the blood-brain barrier, and C b is the fraction of tissue ligand bound to receptor. Whereas K 1 and k 2 describe the transport of ligand to brain and into brain tissue and back, the rate constants k 3 and k 4 relate to receptor-binding parameters ( Fig. 1). This simple model obviously is only an approximation to reality, and even more complicated models have been developed but usually are too complex and poorly defined to be of practical use. B max denotes the concentration of receptors that are available for binding, k on is the receptor-ligand association rate, k off is the corresponding dissociation rate constant, and f 2 is the fraction of free ligand that is available for specific binding. As unlabeled (''cold'') ligand is present in preparations of many radioligands, which compete for binding sites with the labeled tracer, the proportion of labeled tracer to total amount of injected ligand (specific activity, SA) must also be considered in the equations: If the amount of total receptor-binding ligand, C b /SA is very small relative to the binding capacity, B max , it can be disregarded and k 3 is then simply the product of k on , f 2 , and B max . With k 3 being constant in this linear model, a separation of receptor density (B max ) from the association kinetic constant (k on ) is not possible. If receptor occupancy by the ligand cannot be disregarded, k 3 is not constant and the individual variables and constants must be determined separately by complicated curve-fitting and analytic procedures. Only this complicated analysis, usually requiring multiple measurements with different SAs, provides quantitative measures of B max and the dissociation constant (K d ) (5 k off /k on ) (12). In most clinical applications, no attempt is made to determine B max and K d separately but the binding potential (BP 5 B max /K d ) is accepted as adequate (13). In the case of negligible occupancy of receptors by the tracer and coinjected cold ligand, the BP can be calculated from measured parameters of a kinetic study (14) as: Because accurate measurement of f 1 in plasma by clinical means is problematic, quite frequently a measure closely related to the BP, denoted BP9, is used. It assumes that nonspecific binding (as considered by f 2 ) in tissue is a constant and can, therefore, be disregarded. BP9 can be directly calculated from rate constants k 3 and k 4 by: Full kinetic fits to determine B max and K d are usually too complicated for routine use. Therefore, equilibrium approaches are applied to determine receptor binding. These compare the target region containing the receptor of interest and a reference region where tissue activity is free of specific receptor binding. Under equilibrium conditions the volumes of distribution can be obtained, typically by simple measurement of tissue activity C b in target and reference region. Ideally, the 2 regions are different with respect to bound tracer C b , but both contain the same activity of free tracer C f . The modified binding potential (BP9) is related to the different distribution volumes, DV rec in target tissue and DV ref in reference tissue, and to the ratio between tissue activity (C t ) and free tracer activity: Because it is often difficult to achieve actual equilibrium, an approximation is used from a graphic representation of kinetic data, the Logan plot (15) (Fig. 2): when the integral of regional activity over current regional activity is plotted versus the integral of plasma activity over regional activity, the slope of the curve approximates the regional tracer DV. By comparing the slopes for the target (DV rec ) and the reference region (DV ref ), the BP can be calculated. If it can be assumed that k 2 is not changed by the experiment in the target or reference region, it is even possible to avoid the use of a plasma input function, and a transformed time axis involving the integral of tissue activity in the reference region is used. However, although the linear part of this plot approximates BP9, this convenient variant of the Logan plot must be used with caution because the results might be biased by changes in plasma binding or nonspecific tissue binding as well as changes in blood flow. Since the fitting in the Logan plot approach can be done by simple linear regression, this method is well suited for generation of parametric images.

RADIOLIGANDS FOR NEURORECEPTOR IMAGING BY PET AND SPECT
A large number of radioactive ligands labeled for receptors has been developed, but most of them were used only in vitro or in experimental animals and only few reached application in clinical nuclear medicine (1,16). For SPECT, most of the ligands tested in humans were labeled with 123 I (17 ). For PET, the ligands applied in humans were usually labeled with 11 C or 18 F (18) ( Table 1).

DOPAMINE SYSTEM
Dopaminergic neurotransmission has a central role in many brain functions. It is necessary for proper movement coordination, and degeneration of the nigrostriatal dopamine system causes Parkinson's disease and is involved in multisystem atrophy. Pulsatile dopamine secretions elicit pleasant feeling and form a strong reward system that appears to play also a central role in drug abuse. The presynaptic nigrostriatal projection is the main location of the pathologic process in Parkinson's disease and, therefore, the assessment of disturbed dopamine synthesis is the main target for clinical studies (19). Postsynaptic receptors may also be involved in neurodegenerative disorders; they are functionally changed in the course of Parkinson's disease and by treatment with antiparkinson drugs, and they play an eminent role in schizophrenia and the effect of neuroleptics. There are 2 major dopamine receptor types (D 1 and D 2 ); the 3 other types (D 3 , D 4 , and D 5 ) detected in molecular genetic studies are related to these families-D 3 and D 4 are close to D 2 and D 5 is close to D 1 . Only D 1 and D 2 receptors have been imaged in humans, but the applied radioligands usually bind also to the other subtypes.
Several preclinical and clinical studies have applied labeled benzazepines as potent and selective D 1 antagonists. Using these D 1 receptor tracers ( 11 C-SCH 23390, 11 C-NN112, and related compounds), the distribution of D 1 receptors known from autoradiographic investigations could be replicated by PET (striatum . kinetic regions 5 neocortex . thalamus) (20), and an age-dependent decrease (7% per decade) was observed (21). Studies with these radioligands indicated a significant decrease in D 1 receptor density in the striatum and frontal cortex of patients with bipolar affective disorders and in the prefrontal cortex of patients with schizophrenia (22,23). These tracers were also applied in competition studies to examine the binding of neuroleptic drugs to D 1 receptors (24) and for studies in Parkinson's disease.
There is a strong correlation between the antipsychotic potency of the classical neuroleptics and their in vitro affinity to and in vivo occupancy of D 2 receptors (25,26). Therefore, labeled neuroleptics were applied to map brain receptors, and the first visualization of receptors in living men was obtained with 11 C-N-methylspiperone (27,28). Several spiperone derivates, labeled for SPECT ( 77 Br-spiperone) or PET ( 11 C-N-methylspiperone, 76 Br-spiperone, 18 F-fluoroethylspiperone), have been introduced to image the physiologic and pathologic changes of the dopamine receptors in human striata. These tracers have a high affinity to D 2 receptors (K d , 0.1 nmol/L) but also a high affinity to (5-hydroxytryptamine, 5-HT) 5-HT receptors, which are abundant in the cortex (29). Additionally, the binding of spiperone derivatives to D 2 receptors is very tight-essentially irreversible for practical reasons-allowing only saturation or competitive pretreatment studies. To permit equilibrium studies and facilitate replacement by other drugs, more specific D 2 ligands with higher selectivity to D 2 receptors, rapid association, and reversible binding were developed. The gold standard PET tracer for D 2 receptors today is the benzamide 11 C-raclopride (30), which has permitted the selective analysis of central D 2 receptors in primates and humans. For SPECT, 123 I-iodobenzamide is the most widely used D 2 receptor tracer (31). 11 C-Raclopride has a medium affinity to D 2 receptors (K d , 1.2 nmol/L), is displaced more easily, and does not bind to 5-HT receptors. Benzamide derivatives with picomolar affinityfor example, 18 F-fallypride-are especially suitable for investigation of extrastriatal D 2 receptors (32). With these FIGURE 2. Schematic Logan plot for ligands that approach equilibrium during measurement time. By integral transformation of tissue and blood activity (as indicated at the axes), data points approach a straight line, whose slope equals BP9.
tracers binding capacity is measured after equilibrium with plasma levels has been reached, making these tracers suitable to study receptor occupancy by neuroleptics or altered dopamine release.
These specific radioligands have been widely applied for the investigation of D 2 receptors in various physiologic conditions, in different neurologic and mental disorders, and in clinical pharmacology. The densities of D 2 receptors measured in vivo using equilibrium or kinetic models were comparable with the values from in vivo determinations, but differences in B max between raclopride (30-40 nmol/L) and N-methylspiperone (15-25 nmol/L) were observed (33,34). A decrease of D 2 receptors with age has been observed in humans (35), and the D 2 receptor BP appears to be significantly lower in women than in men.
The binding capacity of striatal D 2 receptors is increased in idiopathic Parkinson's disease (Fig. 3). The increase is most pronounced in the posterior putamen, where the dopamine deficit is most severe, reflecting a change in receptor affinity (36,37). With progression of disease, striatal D 2 receptor activity returns to normal or even falls below the normal value (Fig. 3). However, agonist treatment affects receptor binding and possibly prevents supersensitivity phenomena expected in this disorder. After transplantation of medullar or fetal mesencephalic grafts to the putamen of patients with Parkinson's disease, no postoperative alteration of receptor density was observed despite some clinical improvements and increase in 6-fluoro-L-DOPA (FDOPA) uptake (38). Despite remarkable and long-lasting improvement of deficits during continuous high-frequency stimulation of basal nu-clei, especially the subthalamic nucleus, a permanent effect on D 2 receptor activity was not observed (39). In progressive supranuclear palsy, a loss of striatal D 2 receptors, correlated with postmortem in vitro binding assays, has been reported (40). This loss of receptors was accompanied by a significant decrease in 18 F-labeled 3,4-dihydroxyphenylalanine ( 18 F-DOPA) uptake, indicating both pre-and postsynaptic neuronal degeneration. These findings support the role of D 2 receptor imaging in the differentiation of idiopathic Parkinson's disease from parkinsonian syndrome caused by multisystem atrophy (Fig. 3), which may account for up to 20% of cases presenting with parkinsonian symptoms. These patients, however, usually do not respond satisfactorily to treatment. The selective damage of postsynaptic striatal neurons can also be demonstrated by D 2 receptor imaging in Huntington's chorea, where presynaptic dopamine turnover is normal (41,42).
The findings with regard to D 2 receptor activity are controversial in schizophrenia. An increase in D 2 receptor binding capacity that was observed originally with 11 C-N-methylspiperone and competition with haloperidol (43) could not be replicated in drug-naïve patients with the more specific ligand raclopride (33). This controversy has not been entirely resolved, but the finding with spiperone seems not to be specific for schizophrenia (44). However, even when significant changes in striatal D 2 receptors could not be proven, changes of D 2 receptors in extrastriatal brain regions have been demonstrated, suggesting a relationship of positive symptoms with decreased D 2 receptor binding in the anterior cingulate cortex and in the thalamus (45,46).
In clinical pharmacology, specific labeling of striatal receptors with radioligands has been used to determine receptor occupancy during treatment with antipsychotic drugs. The classical neuroleptics occupy 60%-85% of striatal dopamine receptors at clinically effective doses (47). It has been estimated that antipsychotic action requires a receptor occupancy of 60%, whereas extrapyramidal side effects occur at $80% receptor occupancy (48). Overall, the results show an absence of direct correlations between D 2 receptor blockade and the disappearance or reappearance of psychotic symptoms. Additionally, long-lasting remission in schizophrenia patients after drug withdrawal was not related to persistent receptor occupation. However, even the most potent atypical neuroleptics-for example, clozapine and quietapine-show a lower occupancy of D 2 receptor than the typical antipsychotics with considerable interindividual variability (49). They may exert a more complex action on the imbalance of the dopaminergic system in schizophreniacharacterized by hyperstimulation of striatal D 2 receptors and understimulation of prefrontal D 1 receptors involving prefrontal glutamate transmission at N-methyl-D-aspartate (NMDA) receptors. The improved therapeutic efficacy of these atypical neuroleptics probably derives from a more moderate D 2 receptor blockade (hence, decreased extrapyramidal side effects) and increased prefrontal D 1 and NMDA-receptor transmission (50). Binding of the reversible ligand raclopride to striatal D 2 receptors is sensitive to endogenous levels of dopamine and, therefore, the BP can be reduced by direct competition of the tracer with dopamine at the D 2 receptor. This effect can be used to study pharmacologic challenges (amphetamine, methylphenidate, cocaine, dopamine reuptake inhibitors, or tetrabenazine), which lead to increased synaptic levels of dopamine and, as a consequence, result in decreased raclopride binding. Changes in dopamine concentration and resulting raclopride binding, however, were also observed after a large variety of drugs-including ketamine, lorazepam, fenfluramine, nicotine, and alfentanil, to name a few-which do not all influence dopamine reuptake or synthesis directly but increase dopamine release indirectly via other complex neuronal circuits. Further studies in which dopamine release during physiologic activation can be assessed, as reduced raclopride binding, are of great interest and of importance: Functional stimulation by playing a video game (51) may elicit dopamine release, which can also be observed during repetitive transcortical magnetic stimulation of the prefrontal cortex, pointing to the close functional connections between these brain structures (52). The dopamine system is also activated by placebo infusion due to the expectation of improvement of extrapyramidal symptoms and pain (53,54) and as a reward system in substance abuse and addiction (55) as well as in obesity (56).
The neural dopamine transporter (DAT) is a membranebound presynaptically located protein that regulates the concentration of dopamine in nerve terminals. Several compounds have been shown to be antagonists of the monoamine reuptake system and, as nomifensine and substituted analogs of cocaine, they can be labeled for PET or SPECT studies ( 11 C-nomifensine, 11 C-cocaine, 11 C-methyl-phenidate, as well as the tropane-type derivatives of cocaine: 11 C and 123 I-CIT 5 2b-carboxymethoxy-3b-(4-iodophenyl)tropane and 123 I-or 11 C-CIT-FE and CIT-FM as analogs) (1,8). Labeling of the reuptake sites with such tracers provides a measure of the density of dopaminergic nerve terminals. Studies in parkinsonian patients have demonstrated a reduction of the uptake of these tracers that correlated with the decrease in 18 F-DOPA uptake (Fig. 3). Therefore, these tracers can be used to demonstrate presynaptic involvement, especially in Parkinson's disease. Additionally, putamen and caudate were equally affected in multisystem atrophy, and a decrease of the transporter was also observed in Lesch-Nyhan disease. Other studies focused on drug abuse, where a relationship of the time course of the DAT blockade to cocaine abuse was found and a long-acting DAT inhibitor was postulated for antagonizing the pleasurable and additive effect of this drug (57).

SEROTONIN (5-HYDROXYTRYPTAMINE, 5-HT) SYSTEM
5-HT is the transmitter in the central nervous system involved in sleep, eating, sexual behavior, impulse control, circadian rhythm, and neuroendocine function, and it is also the precursor of the hormone melatonin. Serotonergic neurotransmission is altered in many neurologic and psychiatric disorders-in particular, in depression and compulsive disorders-but also in Alzheimer's and Parkinson's disease, autism, and schizophrenia. There is a great heterogeneity of the postsynaptic 5-HT receptors, which belong to 7 major classes and contain .16 subtypes. Suitable radioligands are only available for 5-HT 1A and 5-HT 2A receptors.
5-HT 1A receptors are present in high density in the hippocampus, septum, amygdale, hypothalamus, and neocortex of human brain. On stimulation, they open K 1 channels and in some areas also inhibit adenylate cyclase (58). Carbonyl-11 C-N-(2-[4-C2-methoxyphenyl]-1-piperazinyl)ethyl-N-(2-pyridinyl)cyclohexane carboxamide ( 11 C-WAY-100635, WAY) has high affinity for this receptor. 18 F-Labeled analogs of WAY (e.g., 18  Abnormalities of 5-HT 1A receptors have been described in affective disorders, especially in depression, where a substantial reduction in BP was found in the raphe and in the mesotemporal cortex, and to a less extent also in the cortical areas (59); these changes are most prominent in bipolar depressives and in unipolar depressives with bipolar relatives. The reduction does not depend on previous treatment with 5-HT reuptake inhibitors (60). Some b-adrenergic receptor blockers that might be useful for augmentation of antidepressive therapy compete with 5-HT 1A ligands at the receptor. It must be kept in mind that displacement of ligands by a 5-HT agonist is age dependent and is also affected by anxiety (61). 5-HT 2A receptors are present in all neocortical regions, with lower densities in hippocampus, basal ganglia, and thalamus. The cerebellum and striatum of the brain stem are virtually devoid of 5-HT 2A receptors. A quantitation of 5-HT 2A receptors has been possible with 18 F-altanserin, 18 F-setoperone, and 123 I-2-ketanserin. These ligands are selective 5-HT 2A antagonists; they show good reproducibility. There is a sex difference in 5-HT 2A receptor density and a decrease in binding with age (62). Reduced 5-HT 2A receptor binding may be related to the pathophysiology of anorexia nervosa, as it remains affected long after weight restoration (63). There is also a significant binding of the D 2 receptor ligand 18 F-N-methylspiperone to 5-HT 2A receptors, which can be displaced by anxiolytic drugs (64).
The 5-HT transporter (5-HTT) belongs to the same family as the transporter for dopamine and norepinephine, all of which provide reuptake of neurotransmitters into presynaptic neurons. Interest in 5-HTT is based on the essential role of the 5-HT system in depression (65) and the fact that tricyclic and nontricyclic antidepressants belong to the pharmacologic class of inhibitors of 5-HT reuptake, which enhance extracellular 5-HT levels. Many antidepressants were labeled, but their high lipophilicity induced high nonspecific binding in the brain and they have not been successful for PET or SPECT studies in humans.

CHOLINERGIC SYSTEM
Acetylcholine receptors were originally subdivided into 2 main classes, nicotinic and muscarinic, but multiple heterogeneous subtypes have been characterized. Nicotinic receptors belong to the ligand-gated ion channels, and muscarinic receptors operate via several second messengers.
Nicotinic receptors have been implied in many psychiatric and neurologic diseases, including depression and cognitive and memory disorders, such as Alzheimer's and Parkinson's disease. Thus, 11 C-labeled nicotine was used to visualize and quantify nicotinic receptors in the brain. High binding was reported in several cortical and subcortical regions, and low density was found in the pons, cerebellum, occipital cortex, and white matter. Higher uptake was observed in smokers than in nonsmokers, indicating increased density of nicotinic binding with chronic administration of nicotine. However, the high nonspecific binding and its rapid washout have limited the clinical application of this tracer. More specific ligands, such as epibatidine and derivatives labeled with 11 C or 18 F, showed high uptake in thalamus and hypothalamus or midbrain, intermediate uptake in the neocortex and hippocampus, and low uptake in the cerebellum, which corresponds with the known distribution of nicotinic receptors (70).
Early in the course of Alzheimer's disease, a reduced uptake of radioligands to nicotinic receptors in frontal and temporal cortex has been observed in comparison with that of age-matched healthy control subjects (71). However, the quantitative measurement of the acetylcholinesterase activity in the brain by radiolabeled acetylcholine analogs (N-methyl-3-piperydyl-acetate [MP3A], N-methylpiperidin-4-yl-acetate [MP4A], and N-methylpiperidin-4-yl-propionate [PMP]) has found broader clinical application, because these tracers easily enter the brain, are selectively hydrolyzed, and then are trapped in the brain, permitting quantitation in a 2-tissue-compartment model (72). Studies in patients with Alzheimer's disease demonstrated a widespread reduction of acetylcholinesterase activity in the cerebral cortex (73-75) (Fig. 4) and allow differentiation of Parkinson's disease and progressive nuclear palsy. Additionally, the early loss of cholinergic transmission in the cortex could be shown with these tracers, which precedes the loss of cholinergic neurons in the nucleus basalis of Meynert (76) (Fig. 4). The inhibition of acetylcholinesterase, resulting from treatment with specific drugs, such as donepezil, can also be measured with these tracers (77).
Imaging muscarinic receptors, which are the dominant postsynaptic cholinergic receptors in the brain, has been pursued by many tracers for SPECT and PET ( 123 I-and 11 Cquinuclidinyl-benzylate [QNB], 11 C-2a-tropanyl-benzylate [TKB], 11 C-N-methylpiperidyl-benzylate [NMPB], 11 Cscopolamine, and 11 C-benztropine), but all of these radiopharmaceuticals lack selectivity for receptor subtypes. The highest binding for these ligands is usually found in the cortex, striatum, thalamus, and pons. Normal aging was associated with reduction of muscarinic receptor binding in neocortical regions and thalamus (78). Whereas no regional changes were observed in Alzheimer's disease, hypersensitivity of muscarinic receptors was found in the frontal cortex of patients with Parkinson's disease.

g-AMINOBUTYRIC ACID (GABA) SYSTEM
GABA is the most important inhibitory neurotransmitter. Its transmission is altered in epilepsy and is also altered in anxiety and other psychiatric disorders. Because the GABA receptor is abundant in the cortex and is very sensitive to damage, it represents a reliable marker of neuronal integrity-for example, in ischemic brain damage and in various neurodegenerative diseases. Part of the GABA A receptor complex, which gates the Cl 2 channel, is the central benzodiazepine receptor, which specifically mediates all pharmacologic properties of the benzodiazepines (sedative, anxiolytic, anticonvulsant, myorelaxant). The tracer most widely used for central benzodiazepine-binding sites is the antagonist flumazenil labeled with 11 C or 123 I.
Because of low nonspecific binding, quantitation can be achieved in equilibrium by a rather simple 1-tissuecompartment model (79). Owing to a high extraction rate, initial tracer uptake represents regional cerebral perfusion, with subsequent rapid approximation of the DV that represents receptor binding. For clinical use to detect focal alterations in flumazenil activity, images obtained 10-20 min after injection are essentially equivalent to parametric images of the DV. A differentiation between receptor density and affinity can be achieved by kinetic analysis in combination with various amounts of unlabeled tracer.
The highest degree of binding is observed in the medial occipital cortex, followed by other cortical areas-the cerebellum, thalamus, striatum, and pons-with very low binding in the white matter (which can be used as a reference region). Flumazenil binding is age dependent, with highest values reached at approximately 2 y and a subsequent decline by 25%-50%, until adult values are reached at age 14-22 y (80). A partial inverse agonist, 11 Clabeled RO-15-4513, which binds preferentially to the a 5 receptor subtype, showed uptake that was greater in limbic areas-in particular, in the anterior cingulate cortex, hippocampus, and insular cortex-but lower in the occipital cortex and cerebellum in comparison with flumazenil, which binds mainly to the a 1 subtype (81). This compound could be of particular interest for studies of memory function and memory-enhancing drugs. In contrast to other receptors, this radioligand does not show an age-dependant effect on receptor binding.
In epilepsy, a reduced benzodiazepine receptor density has been demonstrated in epileptic foci, and flumazenil was found to be more sensitive and more accurate for focus localization than 18 F-FDG (82,83). Flumazenil is a biochemical marker of epileptogenicity and neuronal loss; benzodiazepine receptor-density changes are more sensitive than 18 F-FDG in detecting hippocampal sclerosis or microdysgenesis, and benzodiazepine receptor studies were useful in the selection of patients for targeted surgery and for predicting outcome of these procedures (84,85).
The benzodiazepine-binding sites are affected in a variety of degenerative disorders, indicating a loss of GABA A receptor-carrying neurons: This is the case in Huntington's disease, where a reduction in the benzodiazepine receptor density was observed in the caudate and putamen, which, however, was preceded by impairment of neuronal metabolism assessed by FDG (86). In progressive supranuclear palsy, the global reduction in benzodiazepine-binding sites in the cortex suggests loss of intrinsic neurons and is related to the cognitive impairment accompanying this disorder (87). In contrast, in early Alzheimer's disease, preserved benzodiazepine-binding sites relative to glucose hypometabolism suggests intact cortical neurons. In some cerebellar ataxias (e.g., SCA6), benzodiazepine receptor density in the cerebellum is reduced early in the course, indicating loss of cortical neurons (88,89). In patients with hepatic encephalopathy, an increase in cortical flumazenil binding was observed, which may account for the hypersensitivity to benzodiazepines observed in this disorder (90).
In ischemic stroke, flumazenil is used to detect irreversible tissue damage: it is the most sensitive early predictor of eventual ischemic infarction and, therefore, is useful for the selection of patients who can benefit from invasive therapeutic strategies (91). In subacute and chronic states after stroke, functional impairment without morphologic lesions on CT could be attributed to silent infarctions characterized by significant reduction of benzodiazepine binding (92).
In schizophrenia, a correlation between reduced benzodiazepine receptor binding in limbic cortical regions and psychotic symptoms was found (93). Reduced receptor binding in the temporal, occipital, and frontal lobes was also observed in panic and anxiety disorders (65).
Studies of brain receptor occupancy by different drugs acting at the benzodiazepine receptor sites have been used to demonstrate the pharmacologic activity of various benzodiazepines and to investigate noninvasively in vivo the activity of new anxiolytic, anticonvulsant, and sedative drugs in relation to their molecular interaction at the benzodiazepine receptor (94). These studies yield important insight into drug action, despite the fact that flumazenil-binding parameters in the neocortex and cerebellum were not found to be related to anxiety trait or state (95).
Another binding site for benzodiazepine in the brain is the peripheral-type receptor, which is not located on the membrane of neurons but, rather, in the mitochondrial and nuclear subcellular fraction. This peripheral benzodiazepine receptor may play a role in oxidative metabolism and ion fluxes. The tracer most frequently used as the ligand for the peripheral benzodiazepine receptor is the isoquinoline 11 C-PK-11195, which binds to activated, but not resting, microglia. It has been used as a marker of disease activity in inflammatory diseases, such as multiple sclerosis (96), and as an indicator of glia-macrophage activation in ischemia (97), Alzheimer's disease (98), and brain tumors (99).
As an antagonist of GABA, glutamate is the main excitatory neurotransmitter in the cortex, and alterations of glutamatergic neurotransmission are associated with many neurologic diseases. At higher concentrations, glutamate has neurotoxic potential and is involved in the cascade of neuronal damage in ischemia and degenerative disorders. So far, the tracers used to study this system ( 11 C-MK 801, 18 Ffluoroethyl-TCP, 11 C-ketamine, and 18 F-memantine) have only a low specificity to the NMDA receptor, and relevance for clinical studies has not been established (100).

ADENOSINE RECEPTORS
The adenosine receptors (A 1 and A 2A ) play a role in neuromodulation and are functionally altered in epilepsy, stroke, movement disorders, and schizophrenia. The labeled xanthine analogs 18 F-CPFPX and 11 C-MPDX are ligands for the A 1 receptor, which shows high density in the putamen and mediodorsal thalamus, intermediate density in most cortical regions, and low density in the midbrain, brain stem, and cerebellum (101)(102)(103). The A 2A receptors have also been visualized recently in human brain (104). These tracers might have a potential for prediction of severe tissue damage in early states of ischemic stroke.

OPIOID RECEPTORS
Morphine, codeine, heroin, and pethidine were labeled with 11 C but, because of the complex metabolism of these compounds and nonspecific binding, these tracers are not well suited for opioid receptor imaging. Successful PET tracers have been obtained with 11 C-carfentanyl, a potent synthetic m opiate antagonist, and 11 C-diprenorphine and 18 F-cyclofoxy, which bind to all types of opioid receptors. 11 C-Carfentanyl, the ligand for the m receptors, which are the primary site for pleasurable reward feeling, reaches the highest concentrations in the basal ganglia and thalamus (105). With 11 C-diprenorphine, the ligand for m and non-mreceptor sites, high naloxone-replaceable concentrations can be obtained in the striatum and in the cingulate and frontal cortex, including the cortical projections of the medial pain system (106). 18 F-Cyclofoxy, an antagonist to m and k subtypes, is distributed similarly to the distribution of naloxone in the caudate, amygdala, thalamus, and brain stem. In some areas, opioid receptors are at a higher concentration in women, in whom also an age-related decline in receptor density was found (107).
In temporal lobe epilepsy, 11 C-carfentanyl binding was increased in the amygdala and temporal cortex, suggesting an upregulation of opioid receptors or decreased occupancy by endogenous peptides (108). Increased 11 C-carfentanyl binding was also observed with cocaine craving in addicted men (109). The opioid receptor system plays a central role in perception and emotional processing of pain, and changes in receptor availability are related to chronic pain, suggesting increased endogenous release or downregulation of receptors (110).
Regional m receptors were differentially regulated according to sensory and affective or emotional dimensions of pain (111), as decreased 11 C-carfentanyl BP varied directly with ratings of pain intensity. Decreased 11 C-diprenorphine binding during trigeminal neuralgic pain was also explained with increased occupancy by endogenous opioid peptides (112). In several degenerative disorders (Parkinson's disease, striatonigral degenerations, progressive supranuclear palsy, olivopontocerebellar atrophy, and Tourette's syndrome), the pattern of opioid receptor binding is altered (113).
Although there are numerous other ligands and other receptors in the human brain, this review concentrated on those that have already gained application in several centers and are thus beginning to have broader clinical impact. Further work on these tracers, and on the new ones to emerge, is likely to deeply influence development of new drugs and to provide highly specific diagnostic imaging for more effective treatment in clinical neurology.

CONCLUSION
Receptors have a central role in neurotransmission and neuromodulation and are involved in all brain functions, ranging from motor performance to memory, emotion, and pain. The imaging of the distribution, density, and activity of various receptors therefore permits insight into the physiologic activities of functional networks and their disturbance by neurologic and psychiatric disorders and can be used for studying selective drug action.