Advances of Molecular Imaging in Epilepsy

Positron emission tomography (PET) is a neuroimaging method that offers insights into the molecular functioning of a human brain. It has been widely used to study metabolic and neurotransmitter abnormalities in people with epilepsy. This article reviews the development of several PET radioligands and their application in studying the molecular mechanisms of epilepsy. Over the last decade, tracers binding to serotonin and γ-aminobutyric acid (GABA) receptors have been used to delineate the location of the epileptic focus. PET studies have examined the role of opioids, cannabinoids, acetylcholine, and dopamine in modulating neuronal hyperexcitability and seizure termination. In vivo analyses of drug transporters, e.g., P-glycoprotein, have increased our understanding of pharmacoresistance that could inform new therapeutic strategies. Finally, PET experiments targeting neuroinflammation and glutamate receptors might guide the development of novel biomarkers of epileptogenesis.


Introduction
Positron emission tomography (PET) is one of the first neuroimaging techniques that offered unprecedented insights into the molecular functioning of a living human brain. It involves the injection of a positron-emitting radioligand (tracer) and the detection of coincident gamma waves within a scanner. The resulting image represents the spatial distribution of the tracer within the brain.
The 1980s saw the first major implementation of PET in epilepsy after the development of [ 18 F]fluorodeoxyglucose (FDG) to measure local brain glucose metabolism [1]. It was soon recognized that focal interictal hypometabolism correlated with the localization of the epileptic focus, and the method was widely implemented in presurgical epilepsy evaluation [2]. Although FDG remains the workhorse radioligand used in clinical PET imaging, the relevance of FDG PET has slowly diminished after the advent of high-resolution magnetic resonance tomography (MRI).
The last decade brought a gradual shift to the development of receptor-specific ligands that would reveal the in vivo neurochemistry of the epileptic brain. Only few tracers have demonstrated acceptable brain penetration, low-nonspecific binding, high affinity, and target selectivity to progress into clinical studies (Table 1). This review highlights the latest major trends, focusing on PET ligands, which have been used in vivo in people with epilepsy.

γ-Aminobutyric Acid
Since the late 1990s, a large number of PET studies have described the use of [ 11 C]flumazenil PET in epilepsy. This radioligand binds to the benzodiazepine site on the γaminobutyric acid (GABA A ) receptor complex and has shown promising results in the localization and lateralization of the epileptic focus [3]. Its binding is reduced in hippocampal sclerosis and vascular lesios but can be elevated in areas of dysgenesis.
An interesting finding has recently been obtained with [ 11 C]flumazenil PET in a mixed group of nonlesional focal epilepsy patients. Increased seizure frequency was inversely correlated with uptake in the frontal piriform cortex [4]. This result is particularly intriguing, because it was independent of the site of seizure onset and has been reproduced with EEG-fMRI and morphometric MRI [4,5]. Similarly, a crucial epileptogenic area has been described in the prepiriform cortex of rats and monkey and termed Barea tempestas^ [6]. Nevertheless, the tracer has not seen wide application in clinical routine, mainly due to its short half-life. The recent development of an [ 18 F] alternative might overcome these restrictions [7].

Glutamate
Glutamate is the main excitatory neurotransmitter of the central nervous system and is released shortly before and during epileptic seizures [8].
The N-methyl-D-asparate (NMDA) glutamate receptor is not only involved in long-term potentiation and learning but has also been implicated in excitotoxic neuronal damage and epileptogenesis [9,10]. Multiple attempts of developing suitable PET ligands for imaging NMDA receptors in epilepsy failed [6]. Experiments in temporal lobe epilepsy (TLE) with these tracers, e.g., with [ 11 C]-labeled ketamine, usually showed reduced tracer uptake, possibly reflecting either reduced NMDA receptor density, reduced perfusion, or focal atrophy [11].
We developed [ 18 F]GE-179, which binds to the phencyclidine site within the NMDA ion channel pore, thus indicating the activated state of the receptor [12]. In a pilot study, eight epilepsy patients not taking antidepressants had a significantly increased global radioligand binding and four had focally increased signal compared to controls suggesting increased NMDA receptor activation ( Fig. 1). This could point to ongoing epileptogenesis in the group of refractory epilepsy patients [13••]. The most surprising finding was a markedly decreased global binding in three epilepsy patients taking antidepressants, which fell below that of healthy controls. The potential influence of antidepressants and, possibly, depression on activation of NMDA receptors warrants further research.

Multidrug transporters
Around one third of epilepsy patients are resistant to treatment with antiepileptic drugs (AEDs). In these patients, pharmacoresistance is not limited to a single drug but affects drugs acting through diverse mechanisms. Multidrug efflux transporters such as P-glycoprotein might limit intracellular AED concentration by pumping them from the cell and thus contribute to drug resistance [14]. Pharmacoresistant epilepsy was associated with a genotype leading to increased expression of P-glycoprotein [15], and increased P-glycoprotein expression has been observed in hippocampal tissue of refractory epilepsy patients and in rat epilepsy models [16,17].
To study the contribution of P-glycoprotein to pharma coresistance in vivo, an experimental PET protocol involving a radiolabeled P-glycoprotein substrate, [ 11 C]verapamil, and partial, half-maximum P-glycoprotein blockage with tariquidar has been proposed [14]. The brain uptake of [ 11 C]verapamil would correlate with the magnitude of P-glycoprotein action at the blood-brain barrier. Studies in healthy controls confirmed a uniform distribution of P-glycoprotein activity throughout the brain without regional differences [18,19]. A pilot project in seven epilepsy patients found a trend for lower [ 11 C]verapamil influx constants, indicating increased Pglycoprotein function, in the temporal cortex ipsilateral to the epileptic focus [20]. Although these results were obtained in a small sample and failed to reach significance, they informed further evaluation of the tracer in 14 drug refractory epilepsy patients, 8 seizure-free patients, and 13 healthy controls [21••]. Pharmacoresistant patients had a lower baseline influx rate constant (K 1 ) corresponding to a higher P-glycoprotein activity. These results were significant in the bilateral hippocampus, ipsilateral amygdala, fusiform gyrus, inferior temporal gyrus, and middle temporal gyrus. Seizure frequency was positively correlated with P-glycoprotein activity in the hippocampus and on a whole brain level. The findings of [ 11 C]verapamil PET were validated in ex vivo specimen of five patients who underwent epilepsy surgery. Equivalent results were externally reproduced in a subsequent study of 11 epilepsy patients [22]. A follow-up [ 11 C]verapamil project analyzed Pglycoprotein activity in seven patients undergoing epilepsy surgery [23]. Increased P-glycoprotein function before surgery and a decrease postoperatively were associated with optimal surgical outcome. The results also indicated that Pglycoprotein expression responded dynamically to therapeutic procedures or changes in seizure frequency.
These studies provide intriguing evidence for an involvement of multidrug efflux transporters in pharmacoresistant epilepsy. If confirmed, [ 11 C]verapamil PET could be used to identify patients with P-glycoprotein overactivity who might benefit from novel treatment strategies aimed at inhibiting or modulating P-glycoprotein activity [21••]. The main limitation is, however, that the authors could not distinguish whether P-glycoprotein overactivity was a cause or consequence of increased seizure frequency [24]. Another difficulty arose from high uptake in the choroid plexus, which complicates quantification of PET tracer uptake in the hippocampus. This particular effect was used to outline the choroid plexus using a different P-glycoprotein substrate tracer, [ 11 C]-Ndesmethyl-loperamide, in a PET study assessing translocator protein (TSPO) binding in the hippocampus [25].

Inflammation
Inflammatory mediators were found in human and rodent epileptic tissue and it was demonstrated that experimental seizures caused a rapid and pronounced inflammatory reaction [26][27][28]. Additionally, some antiepileptic drugs were shown to have antiinflammatory properties [29]. a PET/MR fusion image (left) and MR FLAIR sequence (right) displaying parts of the lesion (empty arrow) and the hippocampus (full arrow). b Bilateral comparison of time-activity curves (TACs) in the inferior temporal cortex (above) and the hippocampus (below). Both visual analysis and TACs show reduced tracer uptake in the lesioned temporal cortex and slightly increased uptake in the ipsilateral hippocampus. Hypothetically, the extralesional increase of NMDAreceptor activation in the ipsilateral hippocampus could point to ongoing epileptogenesis and prospective studies will be needed to prove this assumption Translocator protein positron emission tomography Several PET tracers have been developed to target neuroinflammation [30]. The most commonly used radioligands are [ 11 C]PK11195 and [ 11 C]PBR28 that bind to TSPO, a marker of activated microglia [26]. The concentration of TSPO is very low in healthy brain tissue; however, a marked increase can be observed in experimental models inducing neuroinflammation in the form of microglial activation [31,32]. This increase can be reliably measured with TSPO PET tracers [31,32]. In rat models of epileptogenesis, increased TSPO expression pointing to neuroinflammation can be demonstrated with PET [28]. The inflammatory process peaks 2 weeks after initial status epilepticus but limbic activation of microglia persists into the chronic phase [26].
The first human investigations with TSPO PET were performed in single subjects or small case series with presumed inflammatory epileptic encephalopathies. In an elegant study, unilaterally increased TSPO expression has been demonstrated with 11C-PK11195 PET in two patients with histologically confirmed Rasmussen's encephalitis, substantiating the role of neuroinflammation in this syndrome [33]. Similarly, increased tracer binding has been demonstrated in a case with seizures due to cerebral vasculitis and in another patient with an epileptic encephalitis of unknown etiology [34,35].
Two recent well-powered studies in temporal lobe epilepsy showed increased TSPO radioligand uptake, pointing to activation of microglia, in temporal regions ipsilateral to the epileptic focus and, to a lesser extent, in the ipsilateral thalamus and contralateral temporal lobe [36••, 37]. However, concordance of increased TSPO PET signal with the ictal-EEG onset zone has not been evaluated and the role of this technique as a diagnostic localizing tool is uncertain. Nevertheless, these in vivo findings give support to a local neuroinflammation in human epileptic cortex. Although the causal role of brain inflammation in generating seizures still needs to be confirmed, these observations might provide a rationale for antiinflammatory treatment in some epilepsy patients.

Serotonin
The influence of serotonin (5-HT) on hyperexcitability has been studied for decades. Several animal models demonstrated that elevating extracellular serotonin levels inhibits seizures, mainly acting through the 5-HT 1A receptor subtype [38]. Conversely, increased levels of serotonin metabolites were found in resected epileptic tissue [39]. The interest in serotonin has led to the development of several suitable PET tracers that study three different aspects of cerebral serotonin function. α-[ 11 C]methyl-L-tryptophan positron emission tomography Originally, increased uptake of α-[ 11 C]methyl-L-tryptophan ([ 11 C]AMT) was thought to reflect increased serotonin synthesis. The downside of this approach is that serotonin metabolism might be disturbed in disease and the findings are difficult to interpret in a pathophysiological context [40]. Additionally, recent evidence in patients with tuberous sclerosis complex (TSC) points to an increased tryptophan metabolism via the kynurenine pathway in the presence of neuroinflammation, leading to the production of proconvulsants [40]. Hence, [ 11 C]AMT PET might also reflect the degree of inflammation in neuronal tissue.
The differentiation of epileptogenic and non-epileptogenic tubers was extensively examined with [ 11 C]AMT PET and increased tracer binding was consistently demonstrated in tubers that colocalized with ictal EEG findings [41][42][43]. A recently published large series of 191 TSC patients demonstrated excellent agreement of [ 11 C]AMT PET with ictal EEG findings [44••]. Moreover, [ 11 C]AMT PET supplied localizing information in more than half of patients with inconclusive EEG. The results were independent of underlying TSC mutation. A smaller series of 12 TSC patients demonstrated low sensitivity (12 %) but high specificity (100 %) of [ 11 C]AMT PET in the prediction of epileptogenic tubers [45•]. Although this series is relatively small, it is one of few [ 11 C]AMT PET studies performed outside of Detroit, providing valuable data for external validation of this tracer.
[ 11 C]AMT PET has also been applied to intractable childhood epilepsy of other causes [46,47]. Again, sensitivity for correct localizing information was low, but nearly perfect specificity surpassed that of FDG PET. However, the limiting factor for the clinical use of this promising tracer is the difficult synthesis and short half-life (20 min). Production is currently restricted to a few dedicated centers worldwide, which severely impacts the application of these findings in clinical practice.

5-HT 1A Receptor Ligands and Tracer for Serotonin Transporter
The density of 5-HT 1A receptors has been measured with several PET tracers that differ in their pharmacological properties (  [50][51][52][53][54][55][56]. The decreases were more pronounced in the hippocampus and in areas involved in seizure generation [52,55]. These findings fit well with the concept of a proconvulsive effect of serotonin depletion.
The use of 5-HT 1A receptor PET in presurgical epilepsy evaluation has only been tested in small patient samples. 5-HT 1A receptor PET showed decreased temporal binding in more than 80 % of these cases and all pathological decreases were congruent with the lateralization of the ictal onset on EEG [54,57]. All patients with lateralizing [ 18 F]MPPF PET became seizure-free after surgery [57]. These results suggest a higher specificity than FDG PET; however, they remain to be reproduced in larger patient samples. The sensitivity of 5-HT 1A receptor PET can be further improved by comparing tracer binding between both cerebral hemispheres using asymmetry indices. This increases the sensitivity to above 90 % with a specificity of 88 % [58].
Some studies have reported HT 1A abnormalities beyond the temporal lobe, describing decreased [ 18 F]FCWAY binding in the insular cortex and anterior cingulate [51,52]. Such changes in the limbic areas were significantly more common in epilepsy patients with concomitant depression compared to those without mood disturbances [51,53]. Also, the magnitude of hippocampal binding inversely correlated with depressive symptoms [59]. These results suggest a common pathomechanism of epilepsy and comorbid depression due to a decrease of serotonin receptors that extends beyond the temporal lobe and affects limbic structures.
A seemingly contradictory finding was demonstrated using [ 18 F]MPPF PET, showing increased uptake in insula and raphe nuclei of depressed epilepsy patients [60]. However, this increased binding most likely reflects a decreased extracellular serotonin concentration in epilepsy patients with depression resulting in increased 5-HTT availability.
A recent study examined serotonin transporter 5-HTT function using [ 11 C]DASB PET in 13 TLE patients and 16 controls [49•]. There were no regional differences in 5-HTT function between patients and controls. However, epilepsy patients with history of depression had a relatively reduced 5-HTT activity in the ipsilateral insula compared to those without depression. A reduced transporter activity would decrease serotonin reuptake and might represent a compensation mechanism to increase extracellular serotonin concentration. Insular 5-HTT activity correlated with 5-HT 1A receptor density measured with [ 18 F]FCWAY-PET, indicating that a reduction of 5-HT 1A receptors may be associated with decreased reuptake of serotonin. However, this study creates more questions than it answers. It remains unclear why epilepsy patients without depression do not show reduced 5-HTT function and why depressed patients develop mood disorders despite these compensatory mechanisms. The results need to be interpreted with caution especially because contradictory findings were reported in patients with major depression [61].

Dopamine
BThe role of dopamine in epilepsy is intriguing, complex, and unresolved^ [62]. Initial input came from experimental data suggesting that dopaminergic neurons in the striatum and substantia nigra were involved in seizure termination [63,64]. These dopamine-rich nigrostriatal areas are thought to modulate thalamocortical projections to regions involved in epilepsy syndromes [62].
The diversity of methodical approaches and patient populations underlines the robustness of these results. However, the detection of dopaminergic abnormalities independent of the underlying type or pathology of epilepsy suggests that these alterations are a nonspecific reaction to seizures [76].
Taken together, an altered dopaminergic system is commonly seen in epilepsy and might act to impair termination of seizures. Also, the findings highlight the importance of subcortical structures in epilepsy. However, the basal ganglia likely represent only the tip of the iceberg within a more widespread dopamine dysregulation. Nevertheless, it is difficult to ascertain significant extrastriatal changes with PET because cortical binding of dopamine is generally low [62,76].

Cannabinoids
Most PET ligand development in this group has focused on cannabinoid receptor 1 (CB 1 ), the major cannabinoid receptor in the central nervous system. Several lines of evidence indicate that CB 1 has anticonvulsant properties that mainly rely on the induction and modulation of protective neuronal mechanisms [77,78]. A downregulation of CB 1 during epileptogenesis might impair these mechanisms and facilitate seizures [79,80]. However, experimental research is not entirely conclusive as others have reported that activation of CB 1 receptors might be proepileptogenic as CB 1 antagonists prevented the development of epilepsy [81].
Goffin et al. recently examined the activation of the endocannabinoid system in temporal lobe epilepsy with the CB 1 radioligand [ 18 F]MK-9470 [82]. They found an increased uptake in the temporal lobe ipsilateral to the epileptic cortex and a bilateral decrease in insular CB 1 binding. However, the authors could not differentiate whether increased CB 1 availability in the epileptic cortex was an abnormal proconvulsive or a protective anticonvulsive phenomenon.
Quantification of CB 1 receptor availability using a different radioligand, [ 11 C]MePPEP, recently showed good to excellent reproducibility [83] with investigation of CB 1 receptormediated neurotransmission in epilepsy currently under way.

Opioids
Increased release of endogenous opioids has previously been demonstrated during provoked absences and reading-induced seizures [84,85]. These findings have underlined the potential Increased TSPO expression points to activation of microglia and an inflammatory reaction in epilepsy patients that could induce epileptogenesis.

Serotonin, inflammation
Tryptophan metabolism Increased uptake in epileptic vs. non-epileptic brain tubers in TSC. Increased uptake in the epileptic focus of children with intractable epilepsy. Low sensitivity but high specificity of these findings [44••, 45•, 46, 47] [ 11 C]AMT-PET adds valuable information on the location of the epileptic focus. It might reflect increased tryptophan metabolism that indicates the local production of proconvulsants.
Cannabinoids CB 1 receptor Increased uptake in ipsilateral temporal lobe; decreased uptake in bilateral insula [82] Supports dysregulation of cannabinoids in epilepsy that could represent a pro or anticonvulsive phenomenon. Opioids μ, δ. and κ opioid receptors Reduced radioligand uptake during absence and reading-induced seizures. Increased uptake 8 h after spontaneous seizures [84][85][86][87] Opioid release during seizures might contribute to seizure termination. This is likely followed by an early interictal overexpression of opioid receptors.

Acetylcholine Nicotinic ACh receptor
Increased uptake in epithalamus, ventral mesencephalon, and cerebellum in ADNFLE patients. Decreased uptake in prefrontal cortex [90] Thalamic and mesencephalic findings may indicate a unique mechanism of nocturnal seizures in ADNFLE. Reduced prefrontal receptor density could be due to neuronal loss. role of opioids in the termination of epileptic seizures.

[ 11 C]diprenorphine
A recent study investigated opioid receptor binding using [ 11 C]diprenorphine PET after spontaneous seizures [86]. Hammers et al. scanned nine refractory temporal lobe epilepsy patients within hours after spontaneous seizures and rescanned them during the later interictal phase. After a median postictal interval of 8 h, increased binding was found in the ipsilateral temporal pole, fusiform gyrus, and, after correction for the partial volume effect, also in the parahippocampal gyrus [87]. There was a negative correlation of ligand uptake with the postictal interval, pointing to an early increase in binding with a gradual return to normal. These findings might either indicate an increased expression of opioid receptors, a reduced endogenous opioid tone or both. In context with previous research, the authors favored the first explanation. The most likely sequence after spontaneous seizures starts with a release of endogenous opioids leading to decreased [ 11 C]diprenorphine binding [84,85]. This is followed by a rapid recovery of available receptors and an overshoot in receptor expression, depicted as increased binding after 8 h on PET [86,87]. Finally, there is a gradual return to normal or lownormal levels during the interictal phase.

Acetylcholine
ADNFLE is an uncommon familial epilepsy syndrome associated with hypermotor seizures occurring during sleep. Two causative genes have been described, most likely leading to a gain of function of the α4β2 subtype of the nicotinic acetylcholine receptor (nAChR) [88,89].
Picard et al. examined eight ADNFLE patients using PET with [ 18 F]fluoro-A-85380, a high affinity agonist at the α4β2 nAChR [90]. Increased tracer uptake was demonstrated in the epithalamus, ventral mesencephalon, and cerebellum, whereas decreased binding was found in the right dorsolateral prefrontal region. The mesencephalic and thalamic increases point toward an overactivated cholinergic pathway ascending from the brainstem. These findings are particularly interesting in the context of a sleep disorder and might indicate a unique pathogenesis of nocturnal seizures. Conversely, decreased receptor density in the prefrontal region could be due to neuronal loss as part of the frontal lobe epilepsy.
However, some questions remain unanswered. It remains unknown whether the observed changes are a hallmark of ADNFLE or a non-specific consequence of seizures. Moreover, although the authors demonstrated spatial changes in receptor density, the functional consequences of the mutation on nAChR activity remain unclear.

Conclusions
PET studies have successfully demonstrated a number of molecular functional abnormalities in epilepsy (Table 2). These insights have not only increased our understanding of the underlying mechanisms of seizures but will also improve the diagnostic evaluation of MRI negative patients. Increasing efforts are being put into translating the novel results into the development of new therapeutics. These will specifically target the neurotransmitters and molecular mechanisms demonstrated in PET studies.
A central challenge will be to tackle the problem of epileptogenesis. The long latent period between an initial insult and the subsequent development of seizures offers plenty of time for diagnostic and therapeutic interventions [91]. PET might play an important role as a biomarker by revealing the molecular processes involved in the development of epileptogenesis.

Compliance with Ethical Guidelines
Conflict of Interest Marian Galovic and Matthias Koepp declare that they have no conflicts of interest.
Human and Animal Rights and Informed Consent This article does not contain any studies with human or animal subjects performed by any of the authors.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.