Innovative Molecular Imaging for Clinical Research, Therapeutic Stratification, and Nosography in Neuroscience

Abstract

Over the past few decades, several radiotracers have been developed for neuroimaging applications, especially in PET. Because of their low steric hindrance, PET radionuclides can be used to label molecules that are small enough to cross the blood brain barrier, without modifying their biological properties. As the use of 11C is limited by its short physical half-life (20 min), there has been an increasing focus on developing tracers labeled with 18F for clinical use. The first such tracers allowed cerebral blood flow and glucose metabolism to be measured, and the development of molecular imaging has since enabled to focus more closely on specific targets such as receptors, neurotransmitter transporters, and other proteins. Hence, PET and SPECT biomarkers have become indispensable for innovative clinical research. Currently, the treatment options for a number of pathologies, notably neurodegenerative diseases, remain only supportive and symptomatic. Treatments that slow down or reverse disease progression are therefore the subject of numerous studies, in which molecular imaging is proving to be a powerful tool. PET and SPECT biomarkers already make it possible to diagnose several neurological diseases in vivo and at preclinical stages, yielding topographic, and quantitative data about the target. As a result, they can be used for assessing patients' eligibility for new treatments, or for treatment follow-up. The aim of the present review was to map major innovative radiotracers used in neuroscience, and explain their contribution to clinical research. We categorized them according to their target: dopaminergic, cholinergic or serotoninergic systems, β-amyloid plaques, tau protein, neuroinflammation, glutamate or GABA receptors, or α-synuclein. Most neurological disorders, and indeed mental disorders, involve the dysfunction of one or more of these targets. Combinations of molecular imaging biomarkers can afford us a better understanding of the mechanisms underlying disease development over time, and contribute to early detection/screening, diagnosis, therapy delivery/monitoring, and treatment follow-up in both research and clinical settings.

Introduction

Molecular imaging is the visualization, characterization, and measurement of biological processes at the molecular and cellular levels in humans and other living systems (1). Over the past few years, rapid improvement in molecular imaging has led to gain in specificity and quantification helpful for early diagnosis and disease follow-up, particularly within the field of neurology. A key advantage of in vivo molecular imaging is its ability to identify pathological processes without the need for invasive biopsies or surgical procedures (2).

This imaging technique is currently performed with positron emission tomography (PET) and single-photon emission tomography (SPECT). Several PET and SPECT radiotracers have been developed for neuroimaging applications. The first ones, namely 123I-labeled amines, 99mTc-hexamethylpropyleneamine-oxime (99mTc-HMPAO), and 99mTc-ethyl cysteinate dimer (99mTc-ECD), were developed in the 1990s to measure regional cerebral blood flow in the presurgical evaluation of patients with refractory partial epilepsy (3). The 2000s saw the advent of PET with the use of fluorine-18 fluorodeoxyglucose ([18F]FDG) in clinical routine, for the assessment of cerebral glucose metabolism. As such, it has also been used in the preoperative evaluation of partial epilepsy, but its indications equally include the early diagnosis and differential diagnosis of dementing disorders, differential diagnosis of cerebral space-occupying lesions, detection of viable tumor tissue (recurrence), non-invasive grading, and differentiation between Parkinson's disease and atypical Parkinsonian syndromes (4).

During the past decade, advances in molecular imaging have enabled scientists to focus on specific brain targets, such as receptors, neurotransmitter transporters, or abnormal protein deposits. There are a growing number of radiotracers, which are regarded as valuable tools for many medical imaging applications, including early detection, diagnosis, and treatment follow-up (2). New imaging biomarkers (e.g., amyloid peptide) allow for the diagnosis of neurological diseases at an early stage, thus contributing to the emergence of the concept of preclinical disease (5, 6). Several PET and SPECT radiotracers are used for both routine clinical applications and research that aim to improve the prevention, diagnosis and treatment of brain diseases. For instance, molecular imaging biomarkers can be used for treatment follow-up, or for selecting patients to be included in clinical trials, or for exploring the neurobiological underpinnings of disease progression.

The aim of the present review was to map out the main innovative radiotracers used in neurology, and explain their role in clinical research. We did not explore 11C-labeled tracers in any depth, as they are not widely used for clinical purposes, owing to their short half-life (20 min). We classified the radiotracers according to their target.

Dopaminergic System

Today, the main class of radiotracers targeting neurotransmission is the one that enables the dopaminergic pathways to be explored (7). These molecules allow for the imaging of nigrostriatal neurons and dopamine receptors. They are used as PET or SPECT radiotracers and assist with the diagnosis of Parkinson's disease (PD), other Parkinsonian syndromes, and Lewy body dementia (LBD) (8).

The first radiotracer to be introduced for the non-invasive assessment of nigrostriatal terminals was [18F]-DOPA in 1983 (9). This radiotracer reflects the activity of aromatic amino acid decarboxylase (AADC), an enzyme that converts L-DOPA to dopamine, through its subsequent accumulation in the dopamine neurons (10). Striatal F-DOPA uptake has been found to be closely related to the nigral cell count (11), except at the beginning of the disease as a consequence of functional compensation (F-DOPA uptake is preserved while motor symptoms can be already presents) (12). This molecule has a history of more than 30 years in clinical research and for the diagnosis of PD. However, in the past decade, the clinical practice led to prefer instead tracers targeting the plasma membrane dopamine transporter (DAT). The latter is easier to use and has a high sensitivity for detecting presynaptic dopaminergic degeneration at early-stage of PD. F-DOPA has recently regained interest in the context of regenerative therapy for PD such as the implantation of dopamine cells or the infusion of drugs with regenerating effects into the striatum (13, 14). The purpose of this therapy is to regenerate the dopaminergic presynaptic function by converting L-DOPA to dopamine. In that cases, DAT tracers are considered to be less relevant for measuring therapeutic response than F-DOPA.

As mentioned above, the second presynaptic dopaminergic target is the DAT, located on dopamine nerve cell terminals. In contrast to the AADC, the DAT is only expressed within dopamine neurons. However, the ligands used for its imaging may also bind to related transporters, such as the serotonine reuptake transporter (SERT) or the norepinephrine reuptake transporter (10). In SPECT imaging, several radiotracers have been developed. The most commonly used are the two cocaine derivatives: [123I]-βCIT and [123I]-FPCIT (8). Compared with [123I]-βCIT, [123I]-FPCIT has better selectivity for DAT vs. SERT, and due to its lower DAT affinity, it has better kinetic properties, with a striatal peak time at 148 min after intravenous injection (15). Although direct comparison of FP-CIT SPECT and F-DOPA PET has shown that both FP-CIT SPECT scans and F-DOPA PET scans are able to distinguish patients with PD from healthy controls with high levels of sensitivity and specificity, the decrease in [123I]-βCIT binding more closely mirrors the reduction in dopaminergic neurons than the decrease in F-DOPA uptake does, suggesting that β-CIT binding is a better index of dopaminergic neuron loss (16). These different sensitivity of the two tracers to a reduction in dopamine transmission is linked to differing degrees of decrease in the striatal uptake of the two tracers, with less striatal FP-CIT uptake than F-DOPA uptake at the early phase of disease (17). [123I]-FPCIT was licensed as DaTSCAN (Amersham Health) in Europe in 2000, and is now a frequently used SPECT radioligand in clinical routine, particularly as an ancillary tool for diagnosing patients with movement disorders, but also in clinical research (15). In the latter context, [123I]-FPCIT has been used in numerous studies seeking to determine the sensitivity and specificity of this tracer in the differentiation of several causes of dementia (18), as well as to study variations in DAT density after different treatments, such as antipsychotics in patients with schizophrenia (19), or psychotherapy in individuals with depression (20).

Tropane derivatives have also been labeled with 99mTc: TRODAT-1 has been compared with F-DOPA in patients with PD (21), and may represent a reliable alternative. 99mTc-labeled ligands are less expensive, and may therefore be more easily accessible, and more suitable for routine use (22–24).

Another tracer has been developed to image the DAT: PE2I. Like FP-CIT and β-CIT, this molecule is a cocaine derivative, which can be labeled with iodine-123 or−125, carbon-11, or tritium (25). This ligand has about a 30-fold higher affinity for DAT than for SERT, and its lower affinity for DAT makes [123I]-PE2I kinetics better than that of [123I]-FPCIT, with a striatal peak time of 30–60 min. However, despite its favorable properties, [123I]-PE2I is not currently licensed as a SPECT radioligand for clinical use (15).

The excellent properties of PE2I mentioned above recently were exploited to develop a new DAT tracer: LBT-999, exploited by Zionexa, which could be used in future PET explorations using fluorine-18 (26–28). Because of its higher resolution, PET imaging is more useful than SPECT for accurate in vivo quantification of DAT density. LBT-999 is a phenyltropane derivative that has demonstrated its suitability for in vivo quantification of DAT in non-human primates (29). An in vivo kinetic study in baboons confirmed that LBT-999 brain uptake is fast, high, and mainly located in the putamen and caudate, with peak uptake in these regions at 30 min postinjection.

A third way of investigating the function of dopamine terminals is to measure the density of vesicular monoamine transporter (VMAT2), which is responsible for taking up neurotransmitters into presynaptic secretory vesicles. Although a majority of VMAT2 are expressed in dopaminergic terminals, this transporter is also located in various monoaminergic neurons, and is involved in the vesicular trapping of a wide variety of neurotransmitters including dopamine, serotonin, norepinephrine, and epinephrine. This target can be investigated with [11C]-DTBZ, or more recently with fluorinated analog [18F]-AV-133, by PET (30). This presynaptic marker follows very typical patterns in several neurodegenerative diseases affecting dopaminergic function, such as PD, LBD, multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and corticobasal syndrome (CBS). Their uptake/binding is altered in several brain areas, depending on the disease and its stage (31). In contrast to AADC activity or DAT binding, it has been suggested that VMAT2 activity is less inclined to changes induced by medication or compensatory mechanisms. However, VMAT2 activity can be impacted by the amount of vesicular dopamine, competing at the recognition site. Hence, the level of VMAT2 binding may decrease with levodopa administration (32). These tracers have a future in early detection/screening, diagnosis, and neuroprotective treatment follow-up of these neurodegenerative diseases, as well as in the monitoring of neural grafted cells after transplantation (8).

Dopaminergic neurotransmission can also be explored by visualizing postsynaptic D2 receptors. The binding potential of these receptors can be assessed using SPECT with the ligands [123I-IBZM and 123I]-IBF, as well as PET with [11C]-raclopride and [18F]-fallypride as radiotracers (Figure 1) (8, 31). The concomitant study of DAT and D2 receptors may improve the diagnostic value of molecular imaging in differentiating between PD and other parkinsonian syndromes (33, 34). Nowadays, however, the measurement of cardiac [123I]-MIBG uptake remains the most frequently used technique to differentiate PD and MSA (35). Molecular imaging of dopamine D2 receptors has also been used to study dopamine's role in drug abuse and addiction (36), and to evaluate several neuropsychiatric disorders (37).

Figure 1.

Comparison of [123I]-IBZM image (A) and [18F]-fallypride image (B) within the same individual.

Key features of all these tracers are summarized in Figure 2 and Table 1.

Figure 2.

Schematic illustration of PET and SPECT techniques for assessing presynaptic and postsynaptic dopaminergic targets. L-DOPA is converted to dopamine by DOPA decarboxylase, then stored in vesicles by a vesicular monoamine transporter. Dopamine reuptake into presynaptic neurons occurs via a dopamine transporter (DAT). Two different types of dopamine receptors are expressed on postsynaptic neurons: D1 and D2.

Table 1.

Main SPECT and PET dopaminergic tracers, molecular strucures, pharmacological properties, and examples of clinical studies.

Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies	Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies
[18F]-DOPA	PET	AADC activity	Uptake	PD (10, 13, 14) LBD (38, 39) MSA (40) PSP (41)	[11C]-DTBZ	PET	VMAT2 density	Ki = 2 (42)	PD (43, 44) LBD (44, 45) MSA (46, 47)
[123I]-βCIT	SPECT	DAT density	Ki = 27 ± 2 (DAT) Ki = 3 ± 0.2 (SERT) Ki = 80 ± 28 (NET) (48)	PD (49–52) LBD (53) MSA (34)	[18F]-AV133	PET	VMAT2 density	Kd = 0.19 (striatum) Kd = 0.25 (hypothalamus)(54)	PD (55) LBD (56)
[123I]-FPCIT	SPECT	DAT density	Ki = 3.5 (DAT) Ki = 9.7 (SERT) (48)	PD (57–59) LBD (60) MSA (61) PSP (58)	[123I]-IBZM	SPECT	D2 receptors density	Kd = 3.1 ± 0.62 Ki = 0.32 (D2) Ki = 4143 (D1) (62)	MSA (63, 64) PSP (63, 65)
[99mTc]-TRODAT-1	SPECT	DAT density	Ki = 14.1 ± 2.1 (DAT) Ki = 360 ± 44 (SERT) (23)	PD (21, 66)	[123I]-IBF	SPECT	D2 receptors density	Kd = 0.106 ± 0.015 Ki = 0.015 ± 0.002 (D2) Ki = 820 ± 164 (D1) (67)	MSA (68) PSP (68, 69)
[123I]-PE2I	SPECT	DAT density	Ki = 17 ± 7 (DAT) Ki = 500 ± 30 (SERT) Ki >1000 (NET) (25)	PD (70, 71) LBD (72) PSP (73)	[11C]-raclopride	PET	D2 receptors density	Ki = 7.5 (74)	MSA (75, 76)
[18F]-LBT-999	PET	DAT density	Kd = 9.15 ± 2.8 (DAT) IC50 >1000 (SERT and NET) (26)	PD (77)*	[18F]-fallypride	PET	D2 receptors density	IC50 = 0.6 (78)	Epilepsy (79)

Kd, dissociation constant; Ki, inhibition constant; IC50, half maximal inhibitory concentration.

^*Preclinical study.

Amyloid Imaging

β-amyloid (Aβ) plaques in the brain are one of the key histopathologic lesions of Alzheimer's disease (AD) (80). Advances in the understanding of the physiopathology of AD suggest that progressive amyloid accumulation begins during the presymptomatic phase, followed by synaptic dysfunction, tau-mediated neuronal injury, a reduction in brain volume, and finally the emergence of cognitive symptoms, followed by a clinical syndrome of overt dementia (81). This suggest that Aβ imaging is a critical step for the early diagnosis of AD.

These deposits were first imaged in PET in 2002, using a thioflavin-T derivative: 11C-Pittsburgh compound B ([11C]-PIB) (82). Although this is the best known compound, its use is restricted to the research field, owing to the short half-life of 11C. Numerous studies have showed that [11C]-PIB binds to Aβ plaques in several cortical regions in patients with AD (82–84). [11C]-PIB binding is correlated with a reduction in cerebrospinal fluid Aβ42 (85), cerebral atrophy (86), and episodic memory impairment in apparently healthy elderly individuals and those with mild cognitive impairment (MCI) (87). These studies have paved the way for the development of several Aβ plaque PET tracers labeled with 18F. To date, three radiopharmaceuticals with equivalent diagnostic performances have been authorized by the European Medicines Agency and the US Food and Drug Administration: 18F-florbetapir, 18F-florbetaben, and 18F-flutemetamol (88) (Table 2).

Table 2.

Main amyloid PET tracer, molecular structures, pharmacological properties, and examples of clinical trials in AD.

Compounds	Target/ measure	Affinity (nM)(88)	Clinical studies in AD
[11C]-PIB	Aβ plaques (fibrillar oligomer)	Ki = 0.9	(82–84)
[18F]-florbetapir	Aβ plaques (aggregated form)	Ki = 2.2	(89–92)
[18F]-flutemetamol	Aβ plaques (soluble form)	Ki = 0.7	(93–95)
[18F]-florbetaben	Aβ plaques (aggregated form)	Ki = 2.4	(96–99)

The clinical criteria that are currently used for AD diagnosis have variable specificity and sensitivity, with pooled averages of 70 and 81% (100). A recent review assessing studies published from January 1980 to March 2014 on the diagnostic utility of these three radiotracers demonstrated a pooled weighted sensitivity and specificity of 89.6% and 87.2% for florbetapir, and 89.3 and 87.6% for florbetaben in differentiating patients with AD from age-matched normal controls (101). These results suggest that 18F-labeled tracers have better sensitivity and specificity than clinical diagnosis and other biomarkers commonly used in practice (89), and are comparable to 11C-PiB. They have also been shown to have good patient tolerability (96). However, the extent and distribution of Aβ plaques and amyloid PET tracer binding in patients are only moderately correlated with patterns of neurodegeneration and cognitive deficits (102–104). This suggests that Aβ deposition, which is a prerequisite for diagnosing AD, is just the starting-point of a cascade of other neuropathological events, rather than the actual driver of neurodegeneration and clinical disease progression (105).

In this respect, these tracers are chiefly useful for their good negative predictive value. A negative scan (i.e., amyloid burden undetectable or extremely low) is considered to be incompatible with a diagnosis of AD. Although a moderate-to-high amyloid plaque density may point to AD, a positive test is not sufficient to diagnose this disorder, especially in elderly participants. It was in this context that the Society of Nuclear Medicine and Molecular Imaging and the Alzheimer's Association delineated “appropriate use criteria” in 2013, identifying three clinical circumstances in which amyloid PET imaging is recommended to clarify the diagnosis: “Patients with persistent or progressive unexplained mild cognitive impairment”, “Patients satisfying core clinical criteria for possible (as opposed to probable) Alzheimer's disease (i.e., atypical clinical course or etiologically mixed presentation)”, and “Patients with atypically young-onset dementia” (106).

In spite of its excellent diagnostic capacity, the use of amyloid PET imaging in clinical practice is still limited. However, this technique has proved extremely useful in clinical trials. Currently, the treatment options for AD are limited to symptomatic drugs, with no attenuation of the ultimate prognosis (107). Numerous studies are being conducted to find new treatments, as well as to better understand the physiopathology of AD. One of the research approaches to develop new treatments involves targeting the two pathological features associated with AD, namely senile plaques (Aβ) and neurofibrillary tangles (NFTs) composed of aggregates of hyperphosphorylated tau protein in paired helicoid filaments (PHF). According to the amyloid cascade hypothesis, toxic plaques are the earliest manifestation of the disease, a notion supported by evidence of Aβ up to 20 years prior to the onset of symptoms (107). Two main classes of medication are under development as a result: monoclonal anti-amyloid antibodies, and inhibitors of pathogenic cleavage of the amyloid precursor protein (APP). PET amyloid radiotracers in clinical trials evaluating the therapeutic potential of these medications are used for selecting and including patients with significant Aβ, or monitoring disease progression under treatment (108). For example, in an amyloid-based immunotherapy study, PET imaging used for treatment follow-up suggested that anti-amyloid antibodies were more effective in the early stages of amyloid accumulation (108). Soon after this discovery, another study was therefore conducted to study the effect of this class of medication in patients with few or no symptoms (MMSE 20–26) but positive amyloid PET imaging (109). This study failed to show a significant difference in cognitive outcomes between the study group and asymptomatic controls; however other drug studies with similar design using amyloid tracer PET imaging in asymptomatic patients with AD are ongoing.

Tau Imaging

As previously indicated, several studies have reported that Aβ burden is only moderately correlated with glucose hypometabolism, disease severity, progression, and clinical presentation. Furthermore, clinical trials assessing monoclonal anti-amyloid antibodies have mostly failed to show a clinical benefit in AD. The other main histopathological figure of AD, abnormal tau protein aggregates, has therefore be considered with much interest. Several PET radiopharmaceuticals have therefore been developed to accurately target abnormal tau protein conformations. NFTs composed of aggregated hyperphosphorylated tau in paired helicoid filaments are one of the two key neuropathological substrates of AD, along with Aβ plaques (110). Whereas, Aβ levels stabilize at an early stage, the presence and extent of NFTs and neuronal injury increase in parallel with disease duration and severity of symptoms (111). Moreover, tau has been found to be more closely related to memory decline in post mortem studies of AD than amyloid pathology (112). Abnormal aggregation of tau protein has also been observed in the pathophysiology of other neurodegenerative diseases, including frontotemporal dementia (FTD), CBS, PSP and, to a smaller extent, LBD; the abnormal conformation of tau in these diseases are distinct from that observed in AD which involves paired helicoid filaments (PHF). These pathologies are collectively known as tauopathies. These tauopathies differ by the isomeric form and ultrastructural morphology of aggregated tau, affected brain regions, and spatial patterns of tau accumulation (110).

Over the past few years, six promising tau imaging agents have been developed: [11C]-PBB3, [18F]-AV-1451 (or flortaucipir, previously known as T807), [18F]-T808, and the THK family [18F]-THK523, [18F]-THK5105, and [18F]-THK5351. These radiotracers have been synthesized, using structure–activity relationship software, from N-benzylidene-benzohydrazide compounds used for the detection of tau-paired helical filament (PHF) (88).

One of the first radiotracers developed for tau imaging was [18F]-FDDNP. This tracer is rapidly metabolized in hydrophilic compounds that cross the blood brain barrier (BBB), resulting in non-specific binding and therefore significant background noise. Furthermore, this tracer is not specific to NFTs, but also has an affinity for Aβ plaques, meaning that it is not the best choice for tau assessment (88, 113, 114).

The first tau-selective radioligand, [18F]-THK523 was synthesized by Okamura et al. (115), and its selectivity for phosphorylated tau was confirmed in post mortem studies, as well as in several in vitro, ex vivo, and in vivo experiments (116). However, this tracer is not able to bind to tau aggregates in non-AD tauopathies such as PSP and CBD, and is characterized by high retention in white matter (117, 118). New THK compounds have since been developed: [18F]-THK5105, [18F]-THK5117, and [18F]-THK5351. The latter has better kinetics, less white matter binding, and a higher affinity for tau than [18F]-THK523 (119). However, it also binds to MAO-B sites, and has a lower binding level in AD than AV-1451 does (110).

[11C]-PBB3 is another tau radiotracer with a high affinity for NFTs, a low level of white matter binding, good BBB penetration and rapid washout. The peculiarity of [11C]-PBB3 is its affinity for the tau isoforms of several non-AD tauopathies. However, it metabolizes to a radiolabeled compound that can cross the BBB, thus limiting its quantification (110).

[18F]-T807 ([18F]-AV1451 developed by Lilly Research Laboratories) and [18F]-T808 belong to the benzimidazole pyrimidine family. They have a nanomolar affinity for the tau PHF found in AD, and are 25 times more selective for tau PHF than for Aβ (120, 121). Today, [18F]-AV-1451 is the most widely used tau radioligand. Like [11C]-PBB3, it has low retention in white matter. Several clinical studies have shown a close correlation between [18F]-AV1451 binding and the neuropathological stages of tau (122), cognitive decline and tau levels in cerebrospinal fluid (123, 124). However, a recent autoradiographic evaluation of AV1451 reported a lower level of binding in non-AD tauopathies, as well as off-target binding in the basal ganglia and substantia nigra in the absence of tau pathology (125).

Recently, another radioligand ([18F]MK-6240, developed by Merck laboratories) was administered to patients with AD with promising results. This tracer showed a high specificity and selectivity for NFTs, good pharmacokinetic properties, and no apparent off-target binding, in contrast to [18F]-AV-1451 (110, 126–128).

As a link has been demonstrated between NFTs and AD symptoms, tau PET tracers are increasingly being used in AD clinical trials, especially those investigating drugs to reduce the tau or Aβ burden (129), such as Aβ monoclonal antibodies. The indirect effect of reducing Aβ on the rate of PHF deposition downstream further supports the amyloid hypothesis, and tau PET imaging may highlight the presumptive disease-modifying impact of these drugs. Furthermore, as tau monoclonal antibodies are designed and investigated, tau PET imaging will be helpful in demonstrating and quantifying the engagement of the molecular target. Many trials currently use cerebrospinal fluid (CSF) biomarkers of tau and phosphorylated tau to detect target engagement, but there are few data on how CSF biomarkers and tau PET imaging correlate. Tau PET imaging may also help to confirm that changes in tau deposition are correlated with clinical disease progression (130). Several tau vaccines have shown efficacity and safety in animal models (131). In a recent study, an anti-tau drug exhibited a good safety profile and even stimulated a positive immune response in human patients (132). Several other early-phase trials of drugs that target tau protein are currently underway, although the results are yet to be published (133).

In this context, like amyloid tracers, tau radioligands (summarized in Table 3) have an important role to play in clinical studies assessing new treatments and measuring disease progression.

Table 3.

Main tau PET tracers, molecular structures, pharmacological properties, and examples of clinical studies.

Compounds	Target/measure	Affinity (nM)	Comments	Clinical studies
[18F]-flortaucipir(AV1451, T807)	PHF-tau	Kd = 14.6 (88)	25 time more selective for tau PHF than for Aβ. Low retention in white matter. Off-target binding has been reported in the basal ganglia and substantia nigra in the absence of tau pathology.	AD (134–137)
[18F]-T808	PHF-tau	Kd = 22 (138)	Slow metabolic defluorination (139)	AD (120)
[18F]-THK523	PHF-tau	Kd = 86 (88)	12-fold selectivity for tau over Aβ. High retention in white matter.	AD (118)
[18F]-THK5117	PHF-tau	Kd = 5.19 (88)	High binding selectivity to tau over Aβ. Substantial white matter binding.	AD (140)
[18F]-THK5105	PHF-tau	Kd = 2,63 (88)	Higher binding affinity to tau fibrils than to Aβ1–42 fibrils (Kd = 35.9 nM) (141) Substantial white matter binding.	AD (115, 142)
[18F]-THK5351	PHF-tau	Kd = 2.9 (88)	Low binding affinity for white matter, and rapid pharmacokinetics. It also bind to MAO-B sites (110)	AD (119)
[11C]-PBB3	PHF and non-PHF tau	Kd = 100 (88)	40–50 fold higher affinityfor NFTs than for Aβ, rapid washout, minimal white matter binding, but it metabolizes to a radiolabeled compound that cross the BBB (110)	AD (143), PSP (144), Amyotrophic lateral sclerosis/parkinsonism dementia complex [ALS/PDC (145)]
[18F]-MK6240	PHF-tau	Ki = 0.36 ± 0.8 (88)	Poor affinity for Aβ plaques (Ki = 10 μM) (127) No apparent off-target binding.	AD (128)

Neuroinflammation

Neuroinflammation is an inflammatory and adaptive response within the central nervous system, and depends on several processes mediated by neuronal cells such as astrocytes, as well as by non-neuronal cells such as the brain's resident macrophages and microglia.

Although initiation of an inflammatory response may be beneficial in response to injury of the nervous system, chronic or maladaptive neuroinflammation can have harmful outcomes in many neurological diseases. During inflammatory processes, cytokines, chemokines and reactive oxygen species (ROS) are produced by glial cells, and all these molecules can be targeted by molecular imaging (146).

The main target for imaging neuroinflammation is currently translocator protein (TSPO) overexpression in activated microglia. TSPO is a highly hydrophobic protein that is mainly situated in the outer mitochondrial membrane. Classically not present in healthy brain parenchyma, TSPO has been widely identified in microglial cells in dementia neuropathology, which involves neuroinflammatory processes and microglial activation. The most widely used TSPO PET radiopharmaceutical tracer used to be [11C]-(R)-PK11195. A new generation of fluorinated tracers has been developed in the past decade (147, 148), with different compound families such as phenoxyarylacetamides derivatives ([18F]-FEDAA1106, [18F]-FEPPA, [18F]-PBR06), imidazopyridine derivatives ([18F]-PBR111), and pyrazolopyrimidine derivatives ([18F]-DPA-714) (Figure 3). However, while these fluorinated compounds have turned out to be more sensitive and specific, with a clear improvement in the signal-to-noise ratio, a major additional problem has been identified, in the shape of a polymorphism in the TSPO gene (rs6971) that affects TSPO binding, with a significant impact on its visualization and its quantification. To circumvent this drawback, a new generation of rs6971-insensitive TSPO radioligands have been developed, such as flutriciclamide ([18F]-GE180) (149), and this latest generation of tracers is currently under evaluation (150).

Figure 3.

[18F]-DPA-714 images obtained from two clinical studies: (A) Comparison between Amyotrophic Lateral Sclerosis (ALS) patients and healthy individuals, and (B) stroke patient.

Other PET tracers of gliosis have been tested, such as [11C]-DED, which binds to MAO-B, and some results in transgenic animals (151) seem to indicate that gliosis occurs early in AD and precedes the deposition of Aβ senile plaque. Cyclooxygenase was also investigated by Shukuri et al. (152), who showed that [11C]-ketoprofen methyl ester, a specific tracer of COX1, is useful for imaging cerebral inflammation in injured rats, with very different kinetics from TSPO tracers. However, a study in humans with this ketoprofen derivative in 2016 (153) failed to yield positive results, suggesting that COX1 expression is more specific for acute inflammation than for chronic inflammation.

Recently, researchers have shown increasing interest in the ROS system. In cardiology, [18F]-DHMT makes it possible to visualize early ROS activation prior to ventricular function deterioration induced by doxorubicin toxicity (154). In neurology, [18F]-ROStrace, a tracer trapped in the brain when it is metabolized by ROS is currently being assessed in models of AD, PD and other neurodegenerative diseases (155).

These tracers are summarized in Table 4.

Table 4.

Main PET tracers for neuroinflammation imaging, molecular structures, pharmacological properties, and examples of clinical studies.

Compounds	Target/measure	Affinity (nM)	Clinical studies	Compounds	Target/ measure	Affinity (nM)	Clinical studies
[11C]-(R)-PK11195	TSPO density	Ki = 9.3 in rat (156)	AD (157, 158), PSP (157), multiple sclerosis (MS), PD, ALS, HI, Rasmussen's encephalitis, Herpes encephalitis, Schizophrenia (156)	[18F]-DPA714	TSPO density	Ki = 7.0 in rat (156)	AD (159, 160)
[18F]-FEDAA1106	TSPO density	Ki = 0.078 in rat (156)	AD (161) MS (162)	[18F]-GE180	TSPO density	Kd = 0.87 in rats (163)	MS (164)
[18F]-FEPPA	TSPO density	Ki = 0.07 in rat (156)	AD (165)	[11C]-DED	MAO-B activity	NA	AD (166)
[18F]-PBR06	TSPO density	Ki = 0.30 in monkey (156)	MS (167)	[11C]-ketoprofen methyl ester	COX-1	IC50 = 47 (COX-1) IC50 = 2.9μM (COX-2) (152)	AD (153)
[18F]-PBR111	TSPO density	Ki = 3.70 in rat (156)	MS (168) Schizophrenia (169) Epilepsy (170)	[18F]-ROStrace	ROS activity	NA	Preclinical studies

Glutamate Receptors

Glutamate is the most abundant excitatory neurotransmitter, and glutamate receptors (GluRs) are implicated in plenty of neurological functions within the central nervous system (CNS). GluRs are classified into two groups: ionotropic receptors (iGluRs) and metabotropic receptors (mGluRs). iGluRs form ligand-gated ion channels and are divided into three subtypes based on their pharmacological properties: NMDA (N-methyl-D-aspartate receptors, NMDARs), AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-proprionic acid) receptors, and kainate receptors. mGluRs are G-protein coupled receptors and include eight receptor subtypes, classified into three groups according to their sequence homology, signal transduction, and pharmacological profiles. Group I is comprised of mGluR1 and mGluR5, group II includes mGluR2 and mGluR3, and group III contains mGluR4, mGluR6, mGluR7, and mGluR8 (171). A dysfunction of these receptors may be involved in the pathophysiology of numerous brain disorders. Several PET and SPECT probes have been developed for GluRs imaging (Table 5).

Table 5.

Main SPECT and PET glutamatergic tracers, molecular structures, pharmacological properties, and examples of clinical studies.

Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies	Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies
[123/125I]-CNS-1261	SPECT	NMDARs density	Ki = 4.2 (171)	Schizophrenia (172, 173)	[18F]-FIMX	PET	mGlu1Rs density	IC50 = 1.8 (171)	(174)
[18F]-GE-179	PET	NMDARs density	Ki = 2.4 (171)	-	[18F]-FPEB	PET	mGlu5Rs density	Ki = 0.2 (171)	PD (175), alcohol dependence (176), depression (177), autism (178)
[18F]-FNM	PET	NMDARs density	Ki = 3500 (179)	Tourette's syndrome (GlutaTour project, ToNIC TMBI)	(E)-[11C-]ABP688	PET	mGlu5Rs density	Kd = 5.7 (171)	Cocaine addiction (180), depression (181), FTD (182), alcohol dependence (183)
[11C]-ITMM: R = O11CH3 [11C]-ITDM: R = 11CH3	PET	mGlu1Rs density	Ki = 12.6 ([11C]-ITMM) (171)	(184)	[18F]-PSS232	PET	mGlu5Rs density	Ki = 1 (E-isomer) (185)	(186)
			Ki = 13.6 ([11C]-ITDM) (171)	Preclinical studies	[11C]-JNJ42491293	PET	mGlu2Rs density	IC50 = 9.2 (171)	(187)

NMDARs

Linked to ligand- and voltage-gated ion channels, NMDARs play an important role in many biological functions, including neurotransmission, neuroprotection, neurodegeneration, long-term potentiation, memory, and neurogenesis (188). These receptors are heteromeric multimers composed of one GluN1 (NR1 subunit) and combinations of GluN2 (NR2 subunits) (189) and GluN3 (NR3 subunits) (190). NR2 subunits come in four subtypes (A D) that determine the type of receptor, with A and B being the most widespread. NR2B subunits, preferentially expressed on primary afferent fibers (PAFs), play a particular role in the transmission of pain messages (191). NMDARs activation requires several types of agonists interacting in cooperation and the simultaneous presence of strong membrane depolarization. Furthermore, NMDARs activation is modulated by extracellular Mg2+, which exerts a voltage-dependent blockade of the open ion channel (192). First, two co-agonists, glutamate and glycine, have to simultaneously bind to their respective sites. Membrane depolarization then causes the release of Mg²⁺ from the channel to allow for the intraneuronal entry of calcium, the starting point for the synthesis of second and third messengers [e.g., prostaglandins and nitric oxide (NO)] (193). Under physiological conditions of synaptic transmission, NMDARs are activated for only brief periods of time. However, in pathological circumstances, their overactivation causes excessive Ca²⁺ influx into nerve cells, and can lead to cell death (194). This abnormal mechanism mediates excitotoxic neuronal injury after acute brain damage (195) and is thought to contribute to disorders of neuronal hyperexcitability (e.g., epilepsy) and chronic neurodegenerative (e.g., AD, Huntington's) (196) and psychotic (197) disorders. Several tracers have been synthesized in order to better understand the physiopathology of these diseases. Most of them are phencyclidine site ligands (PCP) that selectively bind to ion channels in the open and active state. These tracers thus make it possible to visualize only activated NMDARs. Several 123I-, 125I-, 11C-, or 18F-labeled SPECT/PET radiotracers have been developed, based on phencyclidine (PCP), thienylcyclohexyl piperidine (TCP) (198, 199), ketamine (200), memantine (201, 202) or MK-801 (203, 204), as these ligands are known to inhibit the intrachannel PCP sites of NMDARs. Although most of these radiotracers have been found to cross the BBB, none of them have detectable specific binding in vivo, owing to high non-specific binding, poor brain retention, or insufficient affinity for the small number of specific binding sites (205, 206). To our knowledge, only few NMDARs radiotracers have been used in human studies. The diarylguanidine analog, [123I]-CNS-1261 exhibited limited success in a clinical study of patients with schizophrenia (207). In PET imaging, despite encouraging results (208), a recent preclinical study using [18F]GE-179 was unable to demonstrate displaceable in vivo binding that would have been evidence of an in vivo activity-dependent NMDA signal in rats and primates (209–211). Recently, a new [18F]-labeled derivative of memantine, [18F]-fluoroethylnormemantine ([18F]-FNM), was synthesized. In vivo evaluation of this novel PET tracer has yielded encouraging results (179, 212), and it had been injected for the first time into humans, in a pilot study to explore the glutamatergic system in patients with Tourette syndrome (GlutaTour project, ToNIC TMBI) (Figure 4).

Figure 4.

Images from first-in-man injection of [18F]-FNM in a Tourette's syndrome patient (GlutaTour project).

Other NMDAR binding sites, such as the glycine and NR2B sites located on the receptor's extracellular domain, have been the subject of various studies aimed at developing new tracers. However, radiotracer development for these targets has so far been unsuccessful, owing to the ligands' suboptimum physiochemical and pharmacological characteristics, such as affinity, lipophilicity, stability, BBB penetration and pharmacokinetics (171, 205, 206, 213–215).

mGluR

Group I

Group I mGluRs, predominantly expressed postsynaptically, are involved in modulation of synaptic plasticity, and their activation leads to increased neuronal excitability. They are implicated in the physiopathology of several neurological and psychiatric disorder, such as PD, motor dysfunction, multiple sclerosis, epilepsy and stroke, and are the target of recently developed PET probes (171).

mGluR1

mGluR1 are found extensively throughout the brain, but are highly expressed in the cerebellar cortex, hippocampus and thalamus. mGluR1 antagonists have shown promising anxiolytic and antidepressant effects, whereas positive modulators of mGluR1 have been reported to be useful for the treatment of schizophrenia (171). Among all developed molecules to image them, only two radioligands have been injected into humans. The first is [11C]-ITMM. In vitro and preclinical studies found that this ligand had high affinity and selectivity for mGluR1, and displayed high brain uptake, with highest uptake in the cerebellum (richest mGluR1 area). This cerebellar uptake has also been observed in human PET studies, however, [11C]-ITMM showed relatively low uptake in the brain regions with modest expression of mGluR1, such as thalamus, hippocampus, and cerebral cortex, making it difficult to examine target density in these regions (184). Nevertheless, [11C]-ITMM could be used to evaluate alterations in cerebellar mGluR1 under pathological conditions, and further clinical studies may be needed to assess the usefulness of this radioligand as a PET probe for mGluR1 quantification. [11C]-ITDM, an analog of ITMM, was considered superior to [11C]-ITMM after in vivo studies in monkeys because of its higher regional distribution volume in the mGluR1-rich region (216). To our knowledge, clinical PET studies with this radiotracer have not been published.

Finally, [18F]-FIMX, is the second high affinity mGluR1 radioligand injected into humans. The rank order of this tracer uptake correlated well with mGluR1 expression levels in the human brain, with a highest uptake in the cerebellum (174).

mGluR5

mGluR5 are found in the cerebral cortex, hippocampus, accessory olfactory bulbs, and nucleus accumbens (171). In physiological conditions, mGluR5 activates an intracellular cascade by second messenger processes and modulates functions as diverse as memory, anxiety, or learning. It has been demonstrated that the disruption of brain homeostasis in pathological conditions causes hyperactivation of mGluR5, which then contributes to excitotoxicity. mGluR5 dysregulation is therefore implicated in a broad variety of neuropsychiatric disorders and mGluR5 is recognized as a relevant molecular biomarker of glutamate pathology in these diseases. PET imaging of mGluR5 has expanded in recent years and has contributed to go deeper in the pathophysiology of brain diseases and to better evaluate new treatment strategies. Several PET radioligands targeting mGluR5 have been synthetized (205, 217) and the most promising candidates are currently being investigated in several preclinical and clinical studies.

[18F]-FPEB has been developed by Merck Research Laboratories and, regarding its high specificity and selectivity for mGluR5, together with a suitable brain kinetics (218, 219), has been extensively used to investigate mGluR5 density in neurological disorders. In neurology, [18F]-FPEB has shown mGluR5 upregulation in Parkinson's Disease (220), but recent main contributions of [18F]-FPEB imaging are about psychiatry and addictions. Thus, Leurquin-Sterk et al. studied the effects of acute alcohol intake on the glutamatergic system (221), and demonstrated that mGluR5 availability was lower in limbic regions of alcohol-dependent subjects than in healthy controls, suggesting that limbic mGluR5 was involved in a compensatory mechanism helping to reduce craving during abstinence (176). The alteration of mGluR5 availability was also demonstrated in posttraumatic stress disorder, with a higher cortical [18F]-FPEB in vivo binding that was positively correlated with avoidance symptoms (222). Besides, [18F]-FPEB PET imaging did not find any mGluR5 contribution in Major Depressive Disorder (177), whereas, considering neurodevelopmental diseases, an increased [18F]-FPEB binding was observed in postcentral gyrus and cerebellum of male individuals with autism Specter disorder (178).

[11C]ABP688 is a selective, high-affinity mGluR5 antagonist widely used in mGluR5 clinical PET imaging (223, 224). Recently, [11C]ABP688 revealed in vivo evidence of reduced availability of mGluR5 in behavioral variant frontotemporal dementia (182) and in focal cortical dysplasia, in tissue resected from epilepsy patients (225). Whereas, Akkus et al. reported no significance difference in [11C]ABP688 binding in individuals with schizophrenia compared with healthy controls (226), a multi-modal imaging approach, combining mGluR5 PET imaging with [11C]ABP688 together with fMRI reported a lower mGluR5 availability and related functional connectivity alterations in drug-naïve young adults with major depression (227). Esterlis et al. confirmed this hypothesis and objectified an antidepressant response of ketamine through a change in [11C]ABP688 binding that was associated with a significant reduction in depressive symptoms following ketamine administration (228). In alcohol consumption abuse, [11C]ABP688 evidenced altered mGluR5 signaling in the amygdala, that was correlated with the temptation to drink (183).

Regarding the limitations in clinical availability of [11C]ABP688, due to the short physical half-life of carbon-11, fluorinated ABP688 derivatives have been proposed, including the promising radioligand [18F]PSS232. After a preclinical validation evidencing specific and selective in vitro and in vivo properties (185), Warnock et al. reported recently the first-in-human evaluation of this tracer, highlighting in healthy volunteers a favorable brain uptake pattern and kinetics of [18F]PSS232 (186).

These clinical studies, with sometimes ambiguous or even discordant results, must be put in perspective with regard to the influence of the intrasynaptic concentration in endogenous glutamate on the binding of radioligands. For that purpose, pharmacological challenges have been performed in both preclinical and clinical settings, using several glutamate modulators, including ceftriaxone, a potent GLT-1 activator that decreases extracellular levels of glutamate, N-acetylcysteine (NAC), a promoter of the cysteine–glutamate antiporter that increases extrasynaptic glutamate release, and ketamine, an NMDA glutamate receptor antagonist, that increases glutamate release when administered at subanesthetic doses. To date, these pharmacological explorations remain equivocal according to: 1- the pharmacological compound used; 2- the tested radioligand; 3- the studied species (rodents, non-human primates, or human subjects). Thus, whereas ketamine administration decreases [11C]ABP688 binding in vivo in human subjects (229), this result has not been confirmed in rats (230). On the other hand, [18F]PSS232 binding appears to be not impacted to neither acute glutamate shifts after stimulation with N-acetylcysteine (NAC) in human (231) nor ketamine and ceftriaxone infusions in the rat brain (232). This parameter has to be considered carefully to accurately quantify mGluR5 expression in vivo using PET.

Group II and III

Group II and III mGluRs are mostly located within presynaptic regions and involved in the inhibition of neurotransmitter release. Of all the subtypes, only an mGluR2 tracer has been the subject of a human PET study. [11C]JNJ42491293 is a selective, high-affinity radioligand for the positive allosteric modulator (PAM) site of mGluR2. This site is a potential target for treating anxiety, schizophrenia or addiction. In the first human study, its in vivo distribution was consistent with known mGluR2 expression patterns (highest uptake in the striatum and cerebellum) (187). Unfortunately, recent experiments showed an off-target binding in vivo and [11C]JNJ42491293 was considered unsuitable for in vivo imaging of mGluR2 (233).

Cholinergic System

The cholinergic system is well known to be involved in cognitive function, and cholinergic dysfunction has been shown to play a key role in the physiopathology of dementia. Targets have been identified by post mortem studies, which have highlighted alterations in functional components of the cholinergic system (234). These include both presynaptic dysfunction [e.g., in acetylcholinesterase (AChE) or vesicular acetylcholine transporters (VAChTs)] and postsynaptic dysfunction [e.g., in nicotinic acetylcholine receptors (nAChR) or muscarinic acetylcholine receptors (mAChR)] (235, 236). Several radiotracers (summarized in Figure 2) have been developed for each of these targets.

There are two PET tracer substrates for AChE: [11C]-PMP and [11C]-MP4A. These have been used in several clinical studies over the past two decades to highlight modifications in AChE activity in patients with AD, PD, PSP or LBD (237–242). [11C]MP4A has a high specificity for AChE, but also a high rate of hydrolysis by this enzyme, and radioligand uptake in regions with high AChE activity is therefore strongly dependent on the rate of transport into the brain (243). By contrast, [11C]PMP exhibits a hydrolysis rate that is three to four times slower than that of [11C]MP4A, allowing for more precise estimates of AChE activity in regions of moderate-to-high AChE concentration (244). Presynaptic cholinergic terminal density can also be assessed with selective radioligands for presynaptic VAChTs. This has been done in clinical studies with [123I]-IBVM (237, 245) and, more recently, in PET imaging with [18F]FEOBV (246). [18F]FEOBV exhibits lower binding in the mesopontine junction and medulla than [123I]IBVM, providing a robust index of VAChT binding (247).

Postsynaptic cholinergic dysfunction has been assessed in patients with AD, using (S)-[11C]nicotine (248–250). However, these [11C]nicotine studies were hindered by high levels of non-specific binding, rapid metabolism, and washout from the brain, as well as a strong dependence on cerebral blood flow (234). New PET and SPECT radioligands have recently been developed to target α4β2 nAChR, which is the most severely affected receptor subtype in AD, with reductions of up to 50% in the neocortex, entorhinal cortex and hippocampus (251). Some clinical studies using either the SPECT tracer [123I]-5IA, or the PET tracer [18F]-2FA, in patients with AD have highlighted significant reductions in α4β2 nAChR in several brain areas, correlated with cognitive impairment (252, 253). Furthermore, another study found a negative correlation between α4β2 nAChR availability and Aβ load (measured by [11C]-PIB), suggesting that Aβ deposition induces the degeneration of cholinergic neurons (254). It was suggested 10 years ago that the α7 nAChR subtype plays a neuroprotective role, by modulating the neurotrophic system that is needed to maintain cholinergic neuron integrity, and by stimulating signal transduction pathways that support neuron survival. In AD, α7 nAChR is implicated in Aβ toxicity and tau phosphorylation (255). Moreover, deletion of the α7 nAChR gene has been shown to reduce cognitive impairment in animal models of AD (256). Further PET studies using radioligands specific to the α7 nAChR, such as [18F]ASEM, are needed to determine the relationship between α7 nAChR and AD pathology (234).

In PD, LBD or PSP, mAChR has also been imaged with [123I]QNB and [11C]NMPB (257, 258), which are high-affinity mAChR antagonists with similar chemical structures and regional brain distributions. These radiotracers are able to penetrate the BBB efficiently, but non-specifically in relation to the mAChR subtype (234).

All these cholinergic tracers are resumed in Figure 5 and Table 6.

Figure 5.

Schematic illustration of the main cholinergic PET and SPECT radioligands, and their presynaptic or postsynaptic targets. Acetylcholine (ACh) is synthesized by choline acetyltransferase from choline and acetylCoA. ACh is released into the synaptic cleft, where it can bind to two types of receptors expressed on postsynaptic neurons: nicotinic receptors (nAChR) and muscarinic receptors (mAChR). ACh is degraded to choline and acetate by acetylcholinesterase (AChE). The reuptake of choline into presynaptic neurons occurs via a choline transporter. Choline is recycled within presynaptic neurons to form ACh, and stored in vesicles by a presynaptic vesicular ACh transporter (VAChT).

Table 6.

Main SPECT and PET cholinergic tracers, molecular structures, pharmacological properties, and examples of clinical studies.

Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies	Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies
[11C]-PMP	TEP	AChE activity	NA	AD (237, 239, 240, 259), PD (239)	[123I]-5IA	SPECT	α4β2 nAChR density	Kd = 0.011 in rats (260)	AD (252)
[11C]-MP4A	TEP	AChE activity	NA	AD (242), PD (238, 241), LBD (261), PSP (238)	[18F]-2FA	TEP	α4β2 nAChR density	Ki = 0.046 in rats (262)	AD (253)
[123I]-IBVM	SPECT	VAChT density	IC50 = 2.5 ± 0.2 in rats (263)	MSA (245)	[123I]-QNB	SPECT	mAChR density	IC50 = 0.8 in mouse (264)	AD (265), PD, LBD (258)
[18F]-FEOBV	TEP	VAChT density		AD (266), LBD (246)	[11C]-NMPB	TEP	mAChR density	IC50 = 1.8 in mouse (264)	AD (267), PD, PSP (257)
[11C]-nicotine	TEP	α4β2 nAChR density	Kd = 2.4 in rats (268)	AD (248–250)	[18F]-ASEM	TEP	α7 nAChR density	Ki = 0.3 in HEK293 cells stably transfected with rat α7 nAChR (269)	Schizophrenia (270)

GABA_A Receptors

γ-Amino butyric acid (GABA), is the predominant inhibitory neurotransmitter in the central nervous system. This neurotransmitter is able to bind to two types of receptor: ionotropic GABAA/C and metabotropic GABAB. GABAA receptors, also known as the central benzodiazepine receptor, are found on most neurons in the brain, and are part of a superfamily of ligand-gated ion channels. They have a primary binding site for GABA, as well as multiple allosteric modulatory sites. When benzodiazepines, or other allosteric modulators such as barbiturates, bind to GABAA receptors, conformational changes increase the permeability of the central pore to chloride ions, resulting in a chloride flux that hyperpolarizes the neuron (271). GABAA receptors can be composed of several subunit isoforms (272), but only pentamers containing α1, α2, α3, α4, or α5 subunits are benzodiazepine sensitive. These various subunits have a region-specific distribution in the brain, and are believed to subserve different functional and physiological roles and mediate a variety of pharmacological effects. Impairment of GABAA receptor function is increasingly recognized to play a major role in the pathophysiology of several neuropsychiatric diseases such as AD, epilepsy, panic disorders, major depression, cortical brain damage following an acute stroke, anxiety disorders, and chronic alcohol dependency (273). Radiotracers that bind to benzodiazepine sites on GABAA receptors (GABAA-BZ sites) have been shown to be useful for investigating these disorders (274). The first molecules developed for GABAA receptor imaging was carbon-11 labeled benzodiazepines such as [11C]flunitrazepam, [11C]diazepam, or [11C]fludiazepam, but the lack of specificity and in vivo affinity of these ligands (Kd ≥ 10 nM) did not allow accurate determination of GABAA receptor density (275). The triazolobenzodiazepine [11C]alprazolam have also been investigated. Despite an increased affinity (Kd = 3.4 nM), PET studies in six healthy volunteers showed a low extraction into brain (<1% of injected dose), and a substantial depot effect probably into the lungs (276). Finally, the imidazobenzodiazepine flumazenil (Ro 15-1788 or N-methyl-11C]flumazenil), became the most commonly used radioligand for GABAA receptor imagingand is still extensively used to quantify benzodiazepine binding in the human brain (277–279). It was used to measure changes in GABA levels (280), as well as to quantify BZ receptors density in the epileptic foci of patients with partial epilepsy (281–283), in schizophrenic patients (284), neuronal loss in stroke (285), and more recently as a tool in clinical research to evaluate GABAA receptor occupancy using molecules with potential anxiolytic properties (286). [123I]iomazenil, a iodo-analog of flumazenil with very similar binding profile, has also been widely used in clinical studies (287–289).

[11C]Ro15-4513 is a partial inverse agonist at the GABAA-BZ site, preferentially targeting α5 subunits (290, 291). Like the previous ones, this tracer has also been used in clinical studies to understand the precise involvement of GABAA receptors in different neuropsychiatric diseases and the relationship between GABAA receptor density and clinical symptoms (292, 293).

Several attempts of fluorine-18 labeling of flumazenil were performed. Thus, [18F]-FEF, [18F]-FFMZ, and [18F]-flumazenil have been tested. Studies have demonstrated the superiority of [18F]-flumazenil because of a higher affinity and lower levels of radiometabolites in brain (275, 294). Because of the longer half-life of the isotope, this tracer could become the “gold standard” in benzodiazepine PET studies.

The development of GABAA radioligands (summarized in Table 7) need several improvements. Several improvements are needed. First, is to develop receptor subtype specific radioligands such as [11C]Ro15-4513. Radioligands specific for all the GABAA receptor subtypes would be of great importance to PET imaging. The second important enhancement is to develop and apply full agonist radioligands sensitive to changes in endogenous neurotransmitter levels. Finally, development of radiotracers specific to other sites than the BZ binding site will be important in order to further investigate GABAA pharmacology as well as to investigate the role of GABAA receptors in various disease staFinally, development of radiotracers specific to other sites than the BZ binding site will be important in order to further investigate GABAA pharmacology as well as to investigate the role of GABAA receptors in various disease states (275).

Table 7.

Main radioligands for GABAA receptors imaging, molecular structures, pharmacological properties, and examples of clinical studies.

Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical trials
[11C]-FMZ	PET	GABAA-BZ sites (α1, α2, α3, and α5 subunits)	Ki ≈ 1.3 (BZRs containing α1, α2, α3, or α5 subunits) Ki ≈ 150 (BZRs containing α4, or α6 subunits) (295)	Epilepsy (281–283) Stroke (285) Schizophrenia (284)
[123I]-IMZ	SPECT	GABAA-BZ sites	Ki = 0.47 (in primates) (296)	Stroke (287) Epilepsy (288) Anorexia nervosa (289)
[11C]Ro15-4513	PET	GABAA-BZ sites α5 subtype	Ki = 0.3 (BZRs containing α5 subunits) (290)	Alcohol dependence (292) Schizophrenia (293) Autism (297)
[18F]-flumazenil	PET	GABAA-BZ sites	–	Epilepsy (298, 299)

Serotoninergic System

The serotonergic system plays an important modulatory role in many central nervous system functions. It is the target of many drugs commonly used to treat brain disorders, either through reuptake blockade or via interactions with serotonin (5-HT) receptors. Serotonergic dysfunction has been involved in the etiology of many psychiatric disorders, including depression, anxiety and schizophrenia, as well as neurological diseases such as AD and epilepsy. Currently available radiotracers for in vivo brain imaging of the 5-HT system in humans include radioligands for the 5-HT1A, 5-HT1B, 5-HT2A and 5-HT4 receptors, and for the 5-HT transporter (SERT) (300).

The 5-HT1A receptor is one of the most extensively studied receptors in the serotonergic family. Like most 5-HT receptors, it is a G protein-coupled receptor (GPCR) with seven membrane-spanning domains. It serves as an inhibitory autoreceptor in the raphe nuclei, and is targeted by serotonin reuptake inhibitors. It also plays a role with 5-HT4 and 5-HT6 receptors in learning and memory (301, 302). Several radioligands have been synthesized up to now, but only three are in frequent use in clinical studies. The two most widely used are [carbonyl-11C]WAY-100635 and [18F]MPPF (300). These two radioligands are selective and high-affinity 5-HT1A receptor antagonists with a high target-to-background ratio. These tracers have been used in numerous studies of patients with psychiatric disorders such as panic disorder (303), bipolar depression (218) and anorexia nervosa (304), as well as in neurological disorders such as epilepsy, cognitive impairment, AD and migraine (305–311). The third 5-HT1A antagonist radioligand used in clinical studies is [18F]-FCWAY (312), a fluorinated analog of WAY-100635, which also has high 5-HT1A affinity and a high hippocampal-to-cerebellar binding ratio (313–317). However, this compound undergoes high defluorination in vivo, leading to high bone radioactivity uptake. Although this radiodefluorination has been prevented in humans by preadministering disulfiram, this drawback may explain why its use has not been expanded beyond a single PET center (300). A novel and promising 18F-labeled radiotracer, [18F]MefWAY, that is thought to be resistant to defluorination in vivo was recently administered to healthy humans, but no clinical study has yet been published (318). There has been recent interest in the use of 5-HT1A agonists to study variations in endogenous 5-HT levels. [11C]CUMI-101 shows high affinity, but its sensitivity to endogenous 5-HT variations in vivo has not yet been reported (319).

Because they are involved in the etiology and treatment of many psychiatric disorders, 5-HT2A receptors have also been imaged. Five specific radioligands of this receptor have successfully been used in clinical studies: [123I]-R91150, and the PET radioligands [18F]setoperone, [18F]altanserin, [18F]deuteroaltanserin, and [11C]MDL 100, 907. Despite its low signal-to-noise ratio, [123I]-R91150 has often been used in drug occupancy studies, on account of the widespread availability of SPECT (320, 321). It has also been used to study changes in 5-HT2A receptor density that are implicated in various diseases, including cognitive decline (322), suicidal behavior (323), and anorexia nervosa (324). [18F]altanserin is the most frequently used PET tracer. Although it is metabolized to lipophilic radiometabolites, which contribute to non-specific binding, like the previous one, this tracer has been used to determine 5-HT2A receptor density in relation to several psychiatric diseases, such as depression (325), cognitive decline (326), Tourette's syndrome (327), schizophrenia (328) and other neuropsychiatric disorders (329, 330).

Another target allowing for serotoninergic system imaging is the SERT. Interest in SERT imaging has been stimulated by the success of serotonin reuptake inhibitors. The three most widely used belong to the diarylsulfide family: [11C]-DASB, [11C]-MADAM, [123I]-ADAM (300). These radiotracers have been successfully used to estimate SERT occupancy by selective serotonin reuptake inhibitors (331–337), in order to demonstrate changes in SERT density in several neuropsychiatric disorders and throughout their treatment (338–344), as well as in healthy individuals to investigate physiological variations such as personality traits (345) or seasonal changes (346). Other specific radiotracers for this target are still being developed: 4-[18F]ADAM has yielded promising results (347, 348).

All these serotoninergic tracers are summarized in Table 8.

Table 8.

Main serotoninergic radioligands, molecular structures, pharmacological properties, and examples of clinical studies.

Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies	Compounds	Imaging modality	Target/ measure	Affinity (nM)	Clinical studies
[11C]-WAY-100635	PET	5-HT1A density (antagonist)	Kd = 0.2–0.4 (300)	PD (306), depression (349), panic disorder (303), social anxiety disorder (302), anorexia nervosa (304)	[18F]altanserin	PET	5-HT2A density	Ki = 0.13 (300)	AD (326), depression (325), schizophrenia (328), Tourette's syndrome (327), anorexia nervosa (329), obsessive compulsive disorder (330)
[18F]MPPF	PET	5-HT1A density (antagonist)	Kd = 0.3 (300)	AD (308), epilepsy (282, 307, 311), migraine (309, 310)	[18F]deuteroaltanserin	PET	5-HT2A density	–	AD (350)
[18F]FCWAY	PET	5-HT1A density (antagonist)	Ki = 0.25 (300)	Epilepsy (313–315), panic disorder (316)	[11C]MDL-100,907	PET	5-HT2A density	Kd = 0.14–0.19 (300)	Depression (351), obsessive compulsive disorder (352)
[18F]MefWAY	PET	5-HT1A density (antagonist)	IC50 = 26 in rats (353)	–	[11C]DASB	PET	SERT density	Ki = 0.97 ± 0.07(354)	Depression (340), schizophrenia (341), alcohol dependence (343), obsessive compulsive disorder (342), bipolar disorder (344)
[11C]CUMI-101	PET	5-HT1A density (partial agonist)	Ki = 0.15 (300)	Measure of endogenous changes in serotonergic neurotransmission (355)	[11C]MADAM	PET	SERT density	Kd = 0.02 (356)*	–
[123I]R91150	SPECT	5-HT2A density	Kd = 0.11 (300)	AD (322), anorexia nervosa (324), suicidal behavior (323)	[123I]ADAM	SPECT	SERT density	Kd = 0.03 (356)	Depression (338), migraine (339)
[18F]setoperone	PET	5-HT2A density	Kd = 0.7 in rats (357)	AD (358), migraine (359), stroke (360), depression (361)	4-[18F]ADAM	PET	SERT density	Ki = 0.081 (362)	Depression (348)

^*Determined for [3H]MADAM.

α-Synuclein

α-synuclein (α-Syn) is a phosphoprotein found in Lewy bodies (LBs), pathological inclusions that are the hallmark of PD and LBD, as well as in the glial cytoplasmic inclusions (GCIs) that are typical of MSA. All these diseases fall now under the heading of synucleinopathies (363). α-Syn aggregates might induce mitochondrial and proteasomal dysfunction, and interfere with vesicular trafficking within dopamine neurons, leading to their degeneration (364). These protein aggregates have been shown to spread from cell to cell via the extracellular space, and the presence of α-Syn has been demonstrated in extracellular matrices such as plasma, conditioned cell media, and cerebrospinal fluid (365, 366). It is thought that occult α-Syn deposition may occur years before the onset of motor symptoms. Hence, accurate and early detection of premotor synucleinopathies may benefit more from α-Syn imaging, rather than from evidence of dopaminergic changes (367, 368). Although several molecules are able to bind to aggregated α-Syn, a selective imaging biomarker has not been found yet. A sensitive and specific α-Syn radiotracer would have to fulfill several criteria. First, α-Syn exist in different forms, including soluble and insoluble oligomers. An imbalance between these two species led to the formation of pathologic aggregates (369, 370), which have to be recognized by the tracer. Secondly, α-Syn aggregates have distinct cellular localization patterns according to the synucleinopathy, with intraneuronal aggregates (e.g., LBs) in PD, and oligodendrocytic aggregates (e.g., GCIs) in MSA. The ideal α-Syn radiotracer would be able to detect and differentiate these different locations, thereby providing a potential tool for differential diagnosis. Third, colocalization between α-Syn aggregates and other aggregating proteins, such as tau and Aβ (371), has frequently been reported. The optimum tracer would have to be able to specifically detect α-Syn with regard to other deposits, despite their small size and low density. Finally, α-Syn undergoes various posttranslational modifications, such as oxidative modification (372), phosphorylation (373, 374), and N-terminal acetylation, all of which the tracer should be able to detect (363).

As explained above, several molecules are able to cross the BBB and bind to aggregated α-Syn. Unfortunately, these molecules also tend to bind to other aggregated proteins, including Aβ plaques. In this context, diverse Aβ-binding compounds have been investigated for potential affinity for α-Syn, such as [11C]-PIB (375), and more especially [18F]-BF227. In vitro binding studies indicate that [18F]-BF227 binds with high affinity to two binding sites on Aβ1–42 fibrils, and to one class of binding site on α-Syn fibrils. [18F]-BF227 has been found to bind to Aβ-containing AD brain, but failed to bind to Aβ-free LBD or age-matched control homogenates. Furthermore, [18F]-BF227 labeled both Aβ plaques and LBs in an immunohistochemical/fluorescence analysis of human AD and PD brain sections (376). [18F]-BF227 has also been reported to stain GCIs in post mortem tissues, and [11C]-BF227 PET was used to measure the aggregated α-Syn load in eight cases of probable MSA (377). This study demonstrated high signals in GCI-rich brain regions, including subcortical white matter and the putamen, globus pallidus, primary motor cortex, and anterior and posterior cingulate cortex. However, a very recent autoradiography study failed to support binding of [18F]BF-227 to CGI at concentrations typically achieved in PET experiments (378). The lack of specificity and affinity of [18F]-BF227 means that it cannot be used to diagnose synucleinopathies, although it could, theoretically, still be used to monitor changes in α-Syn aggregate load after interventions such as immunotherapy. Levels of other aggregated proteins, such as Aβ, would first have to be independently determined (368).

The last reported α-Syn radioligand is [125I]-SIL23 (379). This tracer has been found to bind to α-Syn fibrils in post mortem brain tissue from patients with PD, as well as to α-Syn in a transgenic mouse model for PD. However, the affinity of SIL23 for α-Syn vs. Aβ and tau fibrils is not optimum for imaging fibrillar α-Syn in vivo. Moreover, high non-specific binding, including non-specific binding in white matter liable to be secondary to lipophilic interactions, also appeared to limit autoradiography with SIL23 in preliminary experiments.

To conclude, the development of an α-Syn PET radiotracer is particularly challenging, and although several studies have tried to develop suitable PET α-Syn radiotracers (380), the ideal candidate remains elusive. These three radiotracers and their main properties are resumed in Table 9.

Table 9.

Main PET and SPECT radiotracers relevant to α-Syn imaging, molecular structures, pharmacological properties, and examples of clinical trials. ^*determined for [3H]-PIB.

Compounds	Imaging modality	Affinity for α-Syn fibrils (nM)	Affinity for Aβ fibrils (nM)	Clinical trials
[11C]-PIB	PET	Kd = 4.16* (381)	Kd1 = 0.71* Kd2 = 19.80* (Aβ1−42 fibrils) (382)	Not used in clinical trials for α-Syn imaging
[18F]-BF227	PET	Kd = 14.03 ± 43.52 (380)	Kd1 = 0.82 ± 1.08 Kd2 = 125.2 ± 29.05 (Aβ42 fibrils) (380)	MSA (377)
[125I]-SIL23	SPECT	Kd = 148 (379)	Kd = 635 (379)	–

Discussion

Molecular imaging agents have evolved from non-specific agents to ligands with very high selectivity for specific brain targets such as receptors, neurotransmitter transporters, or abnormal protein deposits over the last decades. Through the nine targets mentioned above, we have seen that the specificity of the ligands for their target is of paramount importance. Indeed, cross binding affinities of several radioligands could reduce the specificity of the results and may interfere with diagnosis.

More and more the diagnosis of dopaminergic disorders is sustained by molecular imaging combined with clinical examination and have been included in guidelines (383, 384). Thus, molecular imaging is used as an ancillary tool when clinical symptoms are insufficient to confirm a diagnosis. Dopaminergic imaging rests on F-DOPA, but mostly on DAT imaging (especially [123I]-FPCIT), which is considered more relevant to evaluate dopaminergic neuron loss. Thus, LBT-999 could be of great interest in the future because of its better sensitivity, and the higher resolution of PET imaging. In parallel, the increase in attempted to graft dopaminergic neurons may drive up F-DOPA imaging to monitor cell survival. An interesting target remain particularly challenging: indeed, to date, α-Syn cannot be specifically detected with existing radiotracers. This target constituting the hallmark of PD, LBD and MSA, its early visualization could be considerably helpful for diagnosis.

In regards to AD imaging, the first investigations was the assessment of cerebral perfusion. Then, [18F]FDG has allowed to assess cerebral glucose metabolism, and remains a widely prescribed exam at present. Within the last decades, amyloid imaging became the most specific examination because of its excellent negative predictive value, and allow therapeutic stratification in clinical trials. In 2007 (later updated in 2010), Dubois and al. published revised criteria for AD that for the first time included AD biomarkers (amyloid PET and CSF Aβ42) as a supportive criteria. However, Aβ plaques are not correlated with cognitive decline, therefore, clinical research is increasingly turning to tau and neuroinflammation imaging to assess new treatments and follow-up disease progression. Further radiotracers targeting other mechanisms, such as [18F]FNM or [18F]-2FA, could be used in AD studies to improve understanding of the cascades of events leading to neurodegeneration.

Psychiatric diseases diagnosis does not call for molecular imaging in clinical routine. However, in psychiatry, physiopathological modifications behind the symptoms remain not well known and understood. Hence, PET and SPECT radioligands such as, serotonergic, GABAergic or glutamatergic tracers, are a powerful tool to improve psychiatric nosography. Nowadays, it is possible to quantify receptors and transporters imbalances in numerous psychiatric diseases including depression, anxiety and schizophrenia, and explore different treatments options. Moreover, several hypothesis suggest a potential link between excitotoxicity and psychiatrics disorders especially schizophrenia. The hypothesis suggest that progressive excitotoxic neural cell death in hippocampal and cortical areas occurs via “disinhibition” of glutamatergic projection to these areas. Disinhibited glutamatergic activity could result from inhibition of glutamate-mediated neurotransmission and a consequent failure to stimulate inhibitory GABAergic neurons, and/or degeneration of inhibitory GABAergic interneurons (385). Unfortunately, too few studies have been performed yet to highlight this hypothesis. Today, more tracers are be needed to explore glutamatergic and GABAergic systems.

Conclusion

After several decades of research, some radiotracers targeting a hallmark of a disease are valuable diagnostic tools in clinical routine and research, and are used on a large scale. Recently, numerous radiotracers have been developed in order to detect primary changes in brain tissue, and improve our understanding of physiopathological mechanisms of neuropsychiatric diseases. These radioligands provide quantitative and topographical information on the evolution of their target during the course of the disease. More than diagnostic tools, they are one of the only ways to better understand the functioning of the brain in the healthy man and in pathological conditions. Their future usefulness is more focused on therapy monitoring than on the diagnosis itself. As in oncology, molecular neuroimaging is now becoming a therapeutic assistance tool, for screening patient's eligibility for drugs and monitoring the proper functioning of therapy. These new companion drugs are a new challenge for molecular imaging, and quantitative and kinetic analyzes seem to be increasingly relevant for image interpretation. Further development in understanding radiotracer metabolization, binding characteristics, BBB crossing, and clinicopathologic correlations of all these imaging probes will assert their clinical utility, and will lead to the development of more neuroimaging probes in the future.

Author Contributions

This review was written by MB, A-SS and NA. Correction was made by MR, PD, FL, J-FD, and PP. PP was also involved in the plan development.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.