Accepted: 22 February 2018 DOI: 10.1111/epi.14059
FULL-LENGTH ORIGINAL RESEARCH
Positron emission tomography imaging of cerebral glucose metabolism and type 1 cannabinoid receptor availability during temporal lobe epileptogenesis in the amygdala kindling model in rhesus monkeys Evy Cleeren1,2
| Cindy Casteels3,4
Koen Van Laere3,4
| Karolien Goffin3,4
| Peter Janssen1
| Michel Koole3,4
|
| Wim Van Paesschen2
1
Laboratory for Neuro- and Psychophysiology, KU Leuven, Leuven, Belgium 2
Laboratory for Epilepsy Research, KU Leuven, Leuven, Belgium 3
Nuclear Medicine & Molecular Imaging, Department of Imaging and Pathology, KU Leuven, Leuven, Belgium 4
Molecular Small Animal Imaging Center (MoSAIC), KU Leuven, Leuven, Belgium Correspondence Evy Cleeren, Laboratory for Epilepsy Research, KU Leuven, Leuven, Belgium. Email:
[email protected] Funding information Scientific Research Flanders, Grant/Award Number: G.0745.09; Programma Financiering, Grant/Award Number: PFV/ 10/008; FWO Vlaanderen (Fund for Scientific Research, Flanders); Merck Inc, USA
Summary Objective: We investigated changes in the endocannabinoid system and glucose metabolism during temporal lobe epileptogenesis. Methods: Because it is rarely possible to study epileptogenesis in humans, we applied the electrical amygdala kindling model in nonhuman primates to image longitudinal changes in type 1 cannabinoid receptor (CB1R) binding and cerebral glucose metabolism. Two rhesus monkeys received [18F]-MK-9470 and fluorodeoxyglucose–positron emission tomography ([18F]-FDG -PET) scans in each of the 4 kindling stages to quantify relative changes over time of CB1R binding and cerebral glucose metabolism in vivo. We constructed z-score images relative to a control group (n = 8), and considered only those changes measured in both kindled animals by calculating the binary conjunction image per kindling stage. Results: The seizure-onset zone exhibited an increased CB1R binding and a decreased glucose metabolism, which both aggravated gradually in extent and intensity throughout kindling. The ipsilateral thalamus and insula showed hypometabolism that coincided with an increase and a decrease in CB1R binding, respectively. These changes also became gradually more severe throughout kindling and overlapped with ictal perfusion changes during the final stage of amygdala kindling, with hyperperfusion in the ipsilateral thalamus and hypoperfusion in the ipsilateral insula. Significance: The observed changes in CB1R binding may reflect a combination of a protective mechanism of neurons against seizure activity that becomes stronger over time to combat more severe seizures, and on the other hand, a process of epileptogenesis that facilitates seizure activity and generalization, depending on the cell type involved in those specific regions. This study provides unique evidence that the CB1R is dynamically and progressively involved from the start of mesial temporal lobe epileptogenesis. KEYWORDS amygdala kindling, interictal metabolism, mesial temporal lobe epilepsy, positron emission tomography, type 1 cannabinoid receptor
---------------------------------------------------------------------------------------------------------------------------------------------------------------------This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2018 The Authors. Epilepsia published by Wiley Periodicals, Inc. on behalf of International League Against Epilepsy Epilepsia. 2018;1–12.
wileyonlinelibrary.com/journal/epi
|
1
2
1
|
CLEEREN
ET AL.
| INTRODUCTION
Epileptogenesis is not well understood and practically impossible to study in patients because patients’ history before the first symptoms is frequently unknown. In contrast, the occurrence of epileptic seizures in the amygdala kindling animal model is highly controlled. In the rhesus monkey, seizure progression during amygdala kindling is slow, providing an ideal opportunity to study several systems believed to play a role in epileptogenesis longitudinally and within the same subjects.1,2 For centuries, cannabis has been used to treat epilepsy.3 The endocannabinoid system (ECS) is a neuromodulatory system consisting of the cannabinoid receptors, the endocannabinoids (eCBs), and the mechanisms of their synthesis and degradation. Evidence of the role of the ECS in epilepsy is supported mainly by animal data and case studies, because very few clinical trials exist.4,5 In vitro and in vivo animal models of epilepsy have shown that cannabinoid agonists have anticonvulsant properties, and that the ECS plays a role in controlling excitability.6–8 Studies with selective knock-out mice implied that the protective effect of the ECS is mediated by the type 1 cannabinoid receptor (CB1R) on the glutamatergic principal neurons.8,9 Functional imaging studies using N-[2-(3-cyano-phenyl)-3-(4-(2-[18F]fluoroethoxy)phenyl)-1methylpropyl]-2-(5-methyl-2-pyridyloxy)-2-methylproponamide ([18F]-MK-9470)10 as a positron emission tomography (PET) radiotracer to study the CB1R, found an increase in CB1R binding in the seizure-onset zone (SOZ) and a decrease in the ipsilateral insula in patients with refractory mesial temporal lobe epilepsy with hippocampal sclerosis (mTLE-HS).11 The CB1R changes in the SOZ correlated negatively with the latency of the PET scan after a seizure, indicating that dynamic changes in CB1R availability occur in the chronic phase of this disorder.11,12 PET imaging of interictal cerebral glucose metabolism using fluorine-18-labeled fluorodeoxyglucose ([18F]-FDG) is widely used in the presurgical workup of epilepsy patients. In patients with mTLE-HS, [18F]-FDG-PET imaging revealed interictal hypometabolism in the SOZ, hypometabolism in the frontal lobes, and a lesser degree of hypometabolism in the ipsilateral thalamus, basal ganglia, occipital and parietal lobes, and in the contralateral temporal lobe.13,14 In a previous study, we used the amygdala kindling model in rhesus monkeys to study longitudinal changes in ictal brain perfusion.1 Here, we aimed to determine the possible role of the CB1R and its relation to brain glucose metabolism and ictal perfusion changes during temporal lobe epileptogenesis. To that end, we assessed the course of interictal CB1R availability and cerebral glucose metabolism throughout amygdala kindling in the same animals using PET imaging.
Key Points
• • • •
2 2.1
The role of cannabinoids in epilepsy and during epileptogenesis remains disputed We applied PET brain imaging longitudinally during amygdala kindling in rhesus monkeys to study CB1R availability and glucose metabolism We report an early and progressive involvement of the CB1R during temporal lobe epileptogenesis This CB1R change is paralleled by cerebral glucose metabolism and ictal perfusion changes in the same subjects
| METHODS | Animals
All experimental procedures were performed in accordance with EU Directive 2010/63/EU and were approved by the ethical committee at the KU Leuven. Eight male rhesus monkeys (Macaca mulatta, weighing 5.0-8.5 kg) participated in the experiments. Six monkeys were used to obtain baseline imaging values. Two monkeys underwent an amygdala kindling protocol, as described previously1 and as summarized in the Appendix S1. We stimulated the amygdala in 358 sessions in Monkey S and 344 sessions in Monkey K. Both animals progressed from stage I seizures of short duration with minimal behavioral signs to stage IV seizures of longer duration and bilateral motor signs, as described by Wada and colleagues.2
2.2
| PET radiotracer synthesis
All radiotracers were prepared and labeled in-house. CB1R imaging was performed in all animals using the radioligand [18F]-MK-9470.10 The precursor for tracer synthesis was obtained from Merck Research Labs (West Point, NY, USA), and labeling was performed using an alkylation with 2-[18F]fluoroethyl bromide as described previously.10 The final product was obtained after high-performance liquid chromatography (HPLC) purification and had a radiochemical purity >95%. Specific activity was always higher than 90 GBq/lmol. The tracer was administered in a sterile solution of 5 mmol/L sodium acetate buffer pH 5.5 containing 6% of ethanol. The radiotracer [18F]-FDG was routinely prepared inhouse using a Cyclone 18/9 cyclotron (IBA, Louvain La Neuve, Belgium) and a routine FDG GE Tracerlab (GE Healthcare, Brussels, Belgium) synthesis module.
CLEEREN
2.3 | PET data acquisition and reconstruction PET acquisitions were performed after overnight fasting. The 2 animals used in the kindling protocol were scanned with each radiotracer at baseline, 6 weeks after electrode implantation, and during the 4 clinical stages. During amygdala kindling, we elicited a seizure 3 hours before [18F]-MK-9470 and 30 minutes before [18F]-FDG tracer injection. The animals were sedated with a 50/50% mixture of ketamine and medetomidine prior to scanning. Blood pressure, heart rate, and saturation were monitored continuously. During scanning, the monkeys were wrapped in a heat blanket. There was no statistically significant difference in weight (6.2 1.3 kg vs 6.4 078 kg) and injected activity ([18F]-MK-9470: 171 15.8 MBq vs 176.2 43.4 MBq; [18F]-FDG: 183.7 14.9 MBq vs 192 20.1 MBq) for the baseline group vs the kindled monkeys, respectively. PET imaging was performed using an LSO-detector-based FOCUS 220 system (Siemens/Concorde Microsystems, Knoxville, TN), which has a nominal transaxial resolution of 1.35 mm full-width at half-maximum (FWHM). Data were acquired in a 128 9 128 9 95 matrix with a pixel width of 1.27 mm and a slice thickness of 0.8 mm. The coincidence window width was set at 6 ns. For quantification purposes, PET scans were reconstructed using an iterative maximum a posteriori probability algorithm, and scans were corrected for attenuation using a 86 Ge transmission source. Two hours after tracer injection, 30-minute static [18F]MK-9470 acquisitions were obtained. Tracer injection (169.6 34.5 MBq, mean standard deviation [SD]) took place after sedating the monkey for the purpose of quantification by venous blood sampling 2 to 150 minutes postinjection, as described in the Appendix S1. Baseline scanning could be performed for 7 monkeys only. Due to technical failure, we were not able to reconstruct the CB1R PET scan during stage III of kindling for Monkey K. [18F]-FDG (187.4 17.2 MBq [mean SD]) was injected in the saphenous vein of the awake monkey. During the 30-minute uptake phase of the tracer,15 the animal was placed in a dark room to standardize behavior during the uptake period, after which sedation was administered and a 20-minute static [18F]-FDG acquisition was started 1hour postinjection.
2.4
|
ET AL.
technologies, Zurich, Switzerland). Correction for motion between scan frames was performed using the realignment module of SPM8. PET images were coregistered to a masked T1-weighted magnetic resonance image (MRI) of the monkey using an automatic registration algorithm based on mutual information implemented in SPM8. The MRI and the coregistered PET frames of each monkey were spatially normalized to a template MRI.16
2.4.2
2.4.3
Scans were preprocessed using SPM8 (Statistical Parametric Mapping, Wellcome Department of Imaging Neuroscience, London, UK) and PMOD v3.1 (PMOD
| Volume of interest analysis
We used a volume of interest (VOI)–based analysis to evaluate the intersubject and test-retest variability in our group. Therefore, we constructed a VOI map consisting of 34 volumes (17 left and 17 right hemisphere) that were drawn manually on the segmented gray matter of the template MRI,16 as described previously.17,18 The size of every VOI was at least 3 times the volumetric resolution of the PET camera in order to reduce partial volume effects. This VOI map was loaded on all parametric images to permit calculation of the average parametric values within each VOI (PMOD Inc., Zurich, Switzerland). The VOI map was adjusted based on the individual MR of the monkeys and transferred to all rigidly aligned functional scans. For each tracer, we calculated the test-retest value between the baseline and postimplantation scan of Monkey K and Monkey S, considering only the VOIs that were not affected by the implantation of the electrodes. The time between the 2 scans (≥5 weeks) was sufficient to rule out changes due to edema or subacute inflammation. The testretest variability for the radioligand parameter Ri in VOI region i was determined as: |Ri,b - Ri,p|*2/(Ri,b + Ri,p), where the indices b and p refer to the baseline and postimplantation scan, respectively. To measure intersubject variability, we calculated the variation coefficient for each of the 34 VOIs as the SD/mean value over baseline measurements of all monkeys (N = 8).
2.4.4
| Spatial preprocessing
| Quantification-parametric maps
Absolute CB1R binding by [18F]-MK-9470 was expressed as fractional uptake ratio (FUR) values, as described in the Appendix S1. In addition, we investigated relative regional [18F]-MK-9470 binding normalized to whole-brain uptake. Relative regional glucose metabolism was determined by normalizing [18F]-FDG data to the whole-brain uptake.
| PET data processing
2.4.1
3
| Z-score analysis
Because the number of subjects in each stage of the kindling group (n = 2) was insufficient to perform a sensitive voxel-based analysis with sufficient degrees of freedom, we calculated a z-score image for each scan in each
4
|
CLEEREN
kindling stage, as follows: zPET = (PETmonkey_stageX - mean (PETcontrols))/SD(PETcontrols). Here PETmonkey_stageX is the relative and smoothed (2.5 mm FWHM) PET image of Monkey K or Monkey S in each of the 4 clinical stages during kindling. The mean and SD are the mean and SD images across the relative and smoothed PET images of the control group. For each clinical stage, we calculated a binary conjunction image for the z-score images of Monkey K and Monkey S, representing the voxels that showed an increase or decrease in CB1R binding or glucose metabolism in both monkeys at a 1 and 2 SD threshold. These conjunction images were used for further analysis.
2.5
| Ictal [99mTc]-ECD SPECT imaging
Ictal single-proton emission computed tomography (SPECT) perfusion imaging was performed using 99mTechnetium-labeled ethyl cysteinate diethylester ([99mTc]ECD) as radiotracer. A seizure was elicited by electrical stimulation of the amygdala 10 seconds after injection of the radiotracer while the monkey was awake. SPECT images were analyzed using the Subtraction Ictal SPECT co-registered to MRI (SISCOM) technique.19 Details of this series of imaging experiments are described in Cleeren et al1 and are summarized in the Appendix S1. To evaluate the common changes of both kindled monkeys, we calculated the binary conjunction image in every clinical kindling stage (at 1 SD threshold). This image was used for the comparison with the [18F]-FDG and [18F]-MK-9470 parametric maps.
3
| RESULTS
3.1 | Variability and reproducibility of [18F]MK-9470 and [18F]-FDG imaging Eight monkeys were scanned for each tracer to determine the normal variability among naive subjects. Absolute [18F]-MK-9470 FUR values showed a high intersubject (52%-73%) and test-retest variability (36%-63%). Because we were unable to estimate global changes due to this variability in the 2 experimental animals, we assessed regional changes by relative scaling to the global mean brain uptake. For this regional mean relative uptake, intersubject and test-retest variability for both tracers are listed in Table 1. There were no formal outlier values in the relative VOI values for the baseline monkeys for each radiotracer. Intersubject variability of the relative regional CB1R binding ranged from 2.42% (contralateral cerebellum) to 15.98% (ipsilateral globus pallidus) (Table 1). The prefrontal lobe showed the highest [18F]-MK-9470 binding, and the thalamus the lowest, which is consistent with the reported [18F]-MK-9470 uptake.10 Test-retest variability
ET AL.
was in the same range as the intersubject variability, ranging from 0.56% (ipsilateral cerebellum) and 0.18% (contralateral visual portion of temporal lobe) to 9.94% (contralateral auditory portion of the temporal lobe) and 17.67% (ipsilateral thalamus) in Monkey K and Monkey S, respectively. As expected, regional relative [18F]-FDG uptake showed a homogenous pattern in the monkey brain. The highest relative uptake was seen in the prefrontal lobe and the lowest uptake in the globus pallidus (Table 1). Intersubject variation ranged from 2.75% (contralateral parietal lobe) to 12.09% (ipsilateral globus pallidus). Test-retest variability for relative [18F]-FDG values for Monkey K ranged from 0.69% in the ipsilateral insula to 13.81% in the ipsilateral thalamus, and from 0.75% in the contralateral parietal lobe to 14.26% in the contralateral hippocampus for Monkey S. Note that the test-retest variability was not corrected for VOI size.
3.2 | Changes in the interictal relative regional CB1R binding during amygdala kindling The conjunction of the z-score images of the relative regional changes thresholded at 1 and 2 SD of both kindled monkeys is depicted in Figure 1, overlaid on the MRI of Monkey S. Amygdala kindling induced widespread significant changes in CB1R binding in every clinical stage. An increased CB1R binding was observed bilaterally in the mesial temporal lobe (mainly in stage II; white arrows in Figure 1), thalamus (orange arrows in Figure 1), cerebellum, parietal, and occipital cortex, and ipsilaterally in the caudate nucleus and contralaterally in the insula. The largest cluster of increased CB1R binding was seen in the ipsilateral limbic temporal lobe, extending to the thalamus. When applying the anatomical VOI of the limbic temporal lobe to these conjunction images (thresholded at 1 SD), we observed that the number of voxels that showed an increased CB1R binding ipsilaterally peaked in stage II, but was still 3 times higher in stage IV as compared to stage I (Figure 2A). Furthermore, the increase in CB1R binding became lateralized to the site of seizure onset in stage IV. The intensity of the increased binding also rose gradually in the ipsilateral limbic temporal lobe (Figure 2B). Within the anatomical VOI of the thalamus, the number of voxels with an increased CB1R level grew gradually over the different stages of amygdala kindling in the ipsilateral hemisphere (Figure 2C). The mean z-score within the thalamus was about 3 times higher in stage IV compared to stage I, but peaked in stage II (Figure 2D). We also observed decreases in CB1R binding throughout the amygdala kindling process in, for example, the ipsilateral insula (cyan arrows in Figure 1) and lateral temporal
Cingulate cortex
Parietal lobe
Visual temp lobe
Auditory temp lobe
Limbic temp lobe
Frontal lobe
Prefrontal lobe
Insula
Thalamus
Caudate
Globus pallidus
Putamen
0.92 1.04
C
4.13
C
I
4.14
2.02
I
2.06
C
1.70
C
I
1.63
1.25
C
I
1.24
I
2.41
C
3.06 2.43
C
I
3.04
0.90
I
0.73
C
0.43
C
I
0.44
0.47
C
I
0.45
I
0.18
C
0.56 0.18
C
I
0.55
0.31
I
0.31
C
C
I
0.26
I
Amygdala
Hippocampus
0.26
Hemisphere
Area
Volume (cc)
1.12
1.13
1.06
1.06
0.85
0.87
0.93
0.93
0.69
0.70
1.13
1.11
1.21
1.22
1.00
1.01
0.66
0.65
1.03
1.05
0.83
0.75
1.18
1.20
0.73
0.71
0.69
0.70
Mean
Baseline
0.08
0.06
0.03
0.03
0.05
0.04
0.03
0.04
0.03
0.03
0.05
0.05
0.05
0.04
0.04
0.05
0.05
0.05
0.07
0.09
0.08
0.12
0.06
0.07
0.03
0.05
0.04
0.04
SD
7.55
5.20
3.02
2.56
6.14
4.21
3.25
4.04
4.78
4.60
4.08
4.10
4.30
3.16
4.33
4.88
7.99
7.41
7.07
8.30
10.01
15.98
5.42
5.88
4.45
7.63
5.70
5.77
Intersubject variability (%)
[18F]-MK-9470 relative uptake
1.77
NA
7.76
3.97
1.72
0.63
9.94
1.87
NA
NA
NA
NA
NA
3.43
NA
3.24
3.56
6.71
1.50
NA
NA
NA
NA
NA
NA
NA
NA
NA
Monkey K
Test-retest (%)
2.76
NA
3.65
NA
0.18
1.11
1.32
1.09
8.21
NA
4.85
NA
10.38
NA
2.43
8.21
12.87
17.67
8.20
NA
0.36
NA
2.74
NA
7.83
NA
16.71
NA
Monkey S
1.05
1.02
1.04
1.04
0.93
0.94
0.97
0.96
0.71
0.73
1.08
1.06
1.18
1.18
0.98
1.03
0.91
0.89
1.05
1.07
0.70
0.67
1.09
1.12
0.74
0.75
0.70
0.71
Mean
Baseline
0.05
0.06
0.03
0.05
0.03
0.06
0.03
0.05
0.05
0.04
0.05
0.03
0.05
0.09
0.04
0.03
0.04
0.08
0.10
0.11
0.08
0.08
0.04
0.06
0.04
0.04
0.05
0.02
SD
5.10
5.42
2.75
5.08
3.35
6.93
2.88
5.35
6.59
6.01
4.34
3.08
4.11
7.22
4.06
3.06
4.45
9.33
9.12
10.52
10.87
12.09
4.09
5.42
5.32
5.52
7.44
3.05
Intersubject variability (%)
[18F]-FDG relative uptake
T A B L E 1 Mean relative uptake, intersubject variability (%), and test-retest variability (%) obtained using predefined VOI analysis
3.96
NA
5.99
1.42
3.15
2.57
1.40
13.65
NA
NA
NA
NA
NA
3.54
NA
0.69
8.57
13.81
7.88
NA
NA
NA
NA
NA
NA
NA
NA
NA
Monkey K
Test-retest (%)
ET AL.
(Continues)
4.01
NA
0.75
NA
1.44
5.02
4.94
3.95
10.74
NA
4.74
NA
6.31
NA
2.27
3.31
6.01
5.56
1.20
NA
5.72
NA
1.25
NA
14.26
NA
7.55
NA
Monkey S
CLEEREN
| 5
| I, ipsilateral from the kindling site; C, contralateral from the kindling site; cc, cubic centimeters; NA, not applicable, test-retest could not be calculated between baseline and postoperative scans in those VOIs that were affected by electrode implantations (note that Monkey K had bilateral implanted electrodes and the ones in the left hemisphere were not working).
9.40 6.55 0.91 C
4.43
0.02
2.42
0.84 2.76 1.51
0.06
9.83
8.14 7.52 0.92 Cerebellum
I
4.68
0.03
3.56
0.84 10.21 0.56
0.06
11.17
1.51
2.41
4.86
12.75
4.15
0.05 1.03 1.98 3.27 4.38
3.04
0.94
0.03 0.97
C
5.87
Occipital lobe
I
5.88
0.04
1.05 1.52 1.36
0.04
10.77
9.86 12.23 6.45 0.70 C
0.41
0.05
7.29
0.74 12.73 15.61
0.05
Monkey S Monkey K
10.89 6.91
SD Mean
6.21 8.89
Mean
I Entorhinal cortex
0.72
Hemisphere Area
0.40
0.06
SD
0.76
Monkey S Monkey K
8.83
0.05
Test-retest (%)
Intersubject variability (%) Baseline Test-retest (%)
Intersubject variability (%) Baseline
Volume (cc)
[18F]-FDG relative uptake [18F]-MK-9470 relative uptake
11.90
CLEEREN
T A B L E 1 (Continued)
6
ET AL.
F I G U R E 1 Interictal changes in CB1R binding as measured with [18F]-MK-9470. This figure depicts the conjunction image of the z-score images of Monkey K and Monkey S. Z-score images were thresholded at either 1 SD with respect to the control group, where red depicts an increase and blue a decrease in CB1R binding, or at 2 SD, where orange depicts an increase and cyan a decrease in CB1R binding. The different rows illustrate the different clinical stages during amygdala kindling. Coronal slices are shown in radiological convention. The orange and white arrows depict an increase of CB1R binding in the thalamus and mesial temporal lobe bilaterally, respectively. In fact, these 2 areas are part of 1 large cluster of increased CB1R binding, which is further extending to parietooccipital areas. A decrease in CB1R binding was observed in for example, the ipsilateral insula, as marked with a cyan arrow
lobe and bilaterally in the putamen, and (pre)frontal, motor, and somatosensory cortex (Figure 1). The ipsilateral insula contained an increasing number of voxels showing a downregulation of the CB1R throughout the kindling process (Figure 2E) and became lateralized to the ipsilateral hemisphere from stage II on. The mean z-score in this anatomical VOI showed a small upregulation of the CB1R in stage I, but a gradual decrease thereafter over time reaching a mean z-score of 0.31 in stage IV (Figure 2F).
3.3 | Changes in interictal relative glucose metabolism during the amygdala kindling Figure 3 illustrates the conjunction of the [18F]-FDG zscore images of the 2 kindled subjects in the different clinical stages at the 1 SD and 2 SD thresholds. Throughout kindling, we observed a widespread bilateral decrease in brain glucose metabolism. The largest cluster of hypometabolism was seen in the insula and in the lateral temporal cortex ipsilateral to the kindling stimulation site during stage IV seizures (white circles in Figure 3). Other areas showing hypometabolism included the ipsilateral (orbito) frontal cortex, the putamen, the contralateral motor cortex,
CLEEREN
ET AL.
|
7
F I G U R E 3 Interictal changes in glucose metabolism as measured with [18F]-FDG. This figure depicts the conjunction image of the z-score images of Monkey K and Monkey S. Z-score images were thresholded at either 1 SD with respect to the control group, where red depicts hypermetabolism and blue hypometabolism, or at 2 SD, where orange depicts hypermetabolism and cyan hypometabolism. The different rows illustrate the different clinical stages during amygdala kindling. Coronal slices are shown in radiological convention. White circles highlight the largest cluster of hypometabolism comprising the ipsilateral temporal lobe and insula; orange arrows depict the hypometabolic areas in the thalamus
F I G U R E 2 Changes in CB1R binding in the mesial temporal lobe, ipsilateral thalamus, and insula. A, C, Bar plot expressing the course of the number of voxels that show an increase in CB1R binding during the different stages of amygdala kindling in the limbic temporal lobe (A) and thalamus (C). E, Progression of the voxels showing a decrease in CB1-receptor binding in the insula. The VOIs were anatomically defined. B, D, F, Progression of the mean z-score in these anatomical VOIs. The mean z-score is calculated as the mean z-score of the nonthresholded z-score image in these VOIs, averaged over both kindled monkeys. In every panel, blue bars indicate the ipsilateral VOIs, while the orange bars indicate the contralateral VOIs
the posterior insula, the caudate, the cerebellum, and the thalamus bilaterally (orange arrows in Figure 3). We also observed few areas that showed an increase in glucose metabolism over the course of kindling: ipsilaterally in the cingulate cortex and mesial temporal lobe, contralaterally
in the posterior hippocampus, and bilaterally in the somatosensory cortex and early visual areas. These clusters of hypermetabolism seemed to be less consistent over clinical kindling stages. The most striking area of hypometabolism in stage IV of kindling was the ipsilateral insula and temporal lobe. Figure 4A and B show the progression of the number of hypometabolic voxels in the anatomical VOIs of these areas at the 1 SD threshold. Both areas were hypometabolic from stage I on, but the number of the hypometabolic voxels increased gradually during the course of kindling. Lateralization of this hypometabolism was most apparent from stage III. In addition to an increase in the extent of the hypometabolic cluster, we also observed that the mean z-score within these VOIs became more hypometabolic (Figure 4D,E). Another area with consistent hypometabolism was the thalamus (Figures 3 and 4C). The mean z-score of the ipsilateral thalamus decreased throughout the 4 stages of kindling (Figure 4F). The average increase in metabolism during stage I in the ipsilateral thalamus (Figure 4F) seems contradictory with the decrease shown in Figure 3. However, because the anatomical VOI of the thalamus is larger than the thresholded changes displayed in Figure 3, this effect is likely diluted over this whole volume. Note that we observed hypometabolism in the lateral temporal cortex, not in the amygdala itself.
8
|
CLEEREN
ET AL.
F I G U R E 4 Changes in metabolism in the temporal lobe, insula, and thalamus. Bar plots showing the progression of the number of hypometabolic voxels in the different clinical stages of amygdala kindling in the anatomical VOI of the temporal lobe (A), insula (B), and thalamus (C). D, E, F, Progression of the mean z-score in these anatomical VOIs. The mean z-score is calculated as the mean z-score of the nonthresholded z-score image in these VOIs, averaged over both kindled monkeys. In every panel, blue bars indicate the ipsilateral VOIs, while the orange bars indicate the contralateral VOIs
3.4 | Joint changes in interictal CB1R binding, glucose metabolism, and ictal perfusion The brain areas that are involved during an epileptic seizure can be imaged by ictal brain perfusion imaging using [99mTc]-ECD SPECT imaging. Both kindled monkeys were subjected to ictal brain perfusion imaging throughout the kindling process, as described earlier.1 Here, we compared the conjunction of this ictal perfusion network of both monkeys (at 1 SD) with the results of the PET experiments. During stage IV, 3 clusters survived the 3-fold conjunction (changes in perfusion, CB1R binding, and metabolism at 1 SD; minimum cluster size 50 voxels). The ipsilateral thalamus showed a combined ictal hyperperfusion and interictal hypometabolism and increased CB1R binding. The ipsilateral insula showed a decrease in ictal perfusion, interictal metabolism, and CB1R binding. In the contralateral premotor cortex, we found a cluster of ictal hyperperfusion, and a decrease in both interictal metabolism and CB1R binding. In the early kindling stages, we observed only one cluster showing ictal hyperperfusion, interictal hypometabolism, and a decreased CB1R binding in the ipsilateral orbitofrontal cortex during stage II.
4
| DISCUSSION
We investigated changes in interictal metabolism and cannabinoid type 1 receptor expression in a nonhuman primate model of temporal lobe epileptogenesis. During
amygdala kindling, the monkeys presented with evoked seizures, arising from the right amygdala, that slowly increased in duration and semiology.1,2 It is this progressive nature, among other reasons, that makes this model valid and interesting for studying temporal lobe epileptogenesis. In every stage of the model, we quantified interictal metabolism and CB1R binding using PET imaging. We found widespread changes in metabolism and CB1R binding, many of which were consistent over time, which is in agreement with the results of our ictal SPECT experiments in nonhuman primates in which we described a common network of perfusion changes.1 Type 1 cannabinoid receptors are expressed both on glutamatergic excitatory and GABAergic (gamma-aminobutyric acid) inhibitory neurons. This means that the ECS reduces overactive neurotransmission of both excitatory and inhibitory neurotransmitters. In patients, CB1R availability causes opposite changes in different brain regions involved in temporal lobe seizures,11,12 which is most likely due to cell-type–specific modifications. However, in patients it is impossible to study functional changes early during epileptogenesis. Therefore, we imaged longitudinal changes, starting from a naive brain, in CB1R availability during induced temporal lobe epileptogenesis in an animal model. Remarkably, we observed changes in CB1R availability in the earliest stage of kindling. CB1R availability in the SOZ (right amygdala) and the ipsilateral thalamus increased from stage I on and remained elevated. Moreover, this CB1R upregulation gradually increased throughout kindling. In contrast, previous studies in rodents showed that CB1Rs are
CLEEREN
ET AL.
downregulated in the SOZ in the acute phase of an epilepsy model but upregulated in the chronic phase.20–22 In the ipsilateral insula, the number of voxels showing a decreased CB1R availability was gradually growing throughout kindling. Although PET imaging allows for in vivo imaging of the CB1R in behaving animals, it cannot distinguish between the different subpopulations of neurons on which these receptors are present. Therefore, our observed gradual upregulation of the CB1R in the SOZ and the ipsilateral thalamus and the gradual downregulation in the ipsilateral insula can be interpreted in several ways. First, it may reflect an antiepileptic and neuroprotective mechanism, which can be an acute effect by activation of CB1Rs on glutamatergic neurons, leading to reduced excitability. During seizures, axon terminals release an excessive amount of glutamate, which will activate the receptor-driven 2-AG (2-arachidonoyl glycerol, an endocannabinoid) synthesis pathway.12 This will activate the CB1R, suppress glutamate release, and prevent the overexcited circuits from undergoing uncontrolled hyperexcitability.12 That way, an eCB-mediated negative feedback control system is activated. Indeed, the levels of eCBs were elevated after acute insults.8,22,23 In addition, administering a CB1R agonist had anticonvulsant effects in several animal models by increasing the eCB concentration.6,7 In our previous studies1,17 we reported that the ipsilateral thalamus shows ictal hyperperfusion and is effectively connected to the kindled amygdala, indicating that seizure activity reaches the thalamus fast. Therefore, the same mechanism of acute protection against hyperexcitability could have taken place in the thalamus. As seizures became more severe, this protection mechanism should become stronger, leading to an increasing upregulation of the CB1R on glutamatergic terminals over time. Neuroprotection can also be the result of a long-term downregulation of the CB1R on GABAergic neurons, which will lead to an increased inhibition, thereby contributing to the prevention of seizure propagation. Second, gradual changes in the CB1R levels could reflect a proepileptic effect. A decreased CB1R availability on glutamatergic neurons results in a decrease of depolarization induced suppression of excitation (DSE), yielding hyperexcitable neurons. The eCB-mediated negative feedback control system could be functionally compromised in chronic epilepsy.24 Moreover, administration of an indirect CB1R agonist can delay kindling progression through principal glutamatergic neurons in rats.25 In addition, our data suggest that the “failure” of this control system became apparent early during epileptogenesis. According to this interpretation, this control system will no longer be able to prevent the generation and generalization of seizures.12 In addition, a long-term upregulation of CB1Rs on GABAergic neurons promotes seizure activity by an increased
|
9
suppression of inhibition, which renders neurons more excitable. Indeed, in the hippocampi of patients with epilepsy, CB1R expression on inhibitory axon terminals was upregulated.26 The gradual rise of CB1R availability on GABAergic neurons during epileptogenesis could therefore be responsible for creating a hyperexcitable network in which seizure activity can be aggravated and spread. Third, and most likely, upregulation of CB1Rs could reflect a combination of the previous 2 phenomena. In any case, it is clear that progressive changes in CB1R availability took place throughout the brain during mesial temporal lobe epileptogenesis and that these changes already occurred at the earliest stage. Therefore, an early intervention targeted at the cell-specific CB1R to avoid repeated seizures should be a desirable goal to prevent a kindlinglike phenomenon in patients. Interictal [18F]-FDG-PET imaging in our animal model revealed similar changes during stage IV of amygdala kindling as compared to patients with mTLE-HS.14 In contrast to patient studies, we studied the metabolic changes longitudinally starting from the naive brain and reported early and gradual increases in relative hypometabolism in the ipsilateral temporal lobe, insula, and thalamus. In addition to relative changes in CB1R availability and metabolism in and around the SOZ, the ipsilateral mediodorsal thalamus and the insula seemed to be the key nodes of the epileptogenic network. During stage IV, in which the animals experienced secondarily generalized seizures, these interictal changes also overlapped with the ictal perfusion changes seen during SPECT imaging. The insula lies between the temporal and frontal lobes, which showed hyperperfusion and hypoperfusion, respectively, both during complex partial seizures in patients27 and in our subjects.1 Furthermore, the amygdala has dense connections to the insula,28 as confirmed by electrical microstimulation during functional MRI (EM-fMRI) in our subjects.17 Seizure activity originating in the temporal lobe often spreads to the insular cortex.29 In the insular cortex, CB1Rs are expressed in moderate levels (in humans30) and are mainly present on GABAergic nerve terminals (in rats31). A decreased level of CB1Rs on these nerve terminals may thus result in an increased level of inhibition, which might contribute to prevention of seizure propagation. This reduced CB1R availability coincided with ictal hypoperfusion and interictal hypometabolism. We therefore hypothesize that this growing CB1R decrease could reflect surround inhibition,32 which is in accordance with previous studies.11,14,27 Our baseline group had a high intersubject variability for both tracers. The range of test-retest and intersubject variability for relative [18F]-FDG values was comparable to that for rodent studies.33 Because there are known interspecies differences in CB1R expression,34 the normal
10
|
distribution and variation for a specific radiotracer targeting the CB1R should be considered species-specific. To date, we are the first to report the intersubject variability in a group of rhesus monkeys using [18F]-MK-9470, so reference values are missing. We ruled out all technical factors (tracer injection, blood sample collection, HPLC separation. . .) in our absolute quantification of [18F]-MK-9470 uptake. One could argue that an [18F]-FDG PET and an [18F]-MK-9470 scan 30 minutes and 3 hours after a seizure may still be considered a postictal scan that could be different from a scan obtained days or weeks after a seizure, as is generally done in clinical practice. We applied this time interval for several practical and logistical reasons (clinical tracer production schedules and use of sedatives with potential antiepileptic effects). Furthermore, we believe that tracer injection was always truly interictal in our subjects because electroclinical findings in our subjects pointed to a postictal period of maximum 75 seconds. Because of our small number of subjects, we calculated the conjunction of the changes seen in both monkeys, and we are thus displaying only those changes that are similar in both subjects. This might be more stringent than calculating group averages, while neglecting possible subtle alterations. Data on receptor functioning, signal transduction, and activation by eCBs are lacking and beyond the scope of our study. If we want to draw solid conclusions regarding therapies that tackle the ECS, information on receptor distributions alone are not sufficient. G-protein– coupled receptors, like the CB1R, are assumed to act in a compensatory manner to the endogenous ligand concentrations. In the pilocarpine model of epilepsy, for example, increased levels of hippocampal 2-AG were measured during status epilepticus.22 However, in the cerebrospinal fluid (CSF) of drug-naive patients with TLE, the levels of anandamide, but not 2-AG was reduced.35 Some studies found CB1R desensitization and internalization following chronic tetrahydrocannabinol (THC) treatment36; others observed an increase in CB1R density after chronic exposure to THC and anandamide.37 Furthermore, it was reported that a large CB1R reserve is located in intracellular vesicles.38 It is unknown whether [18F]-MK-9470 binds to these intracellular CB1Rs. Future studies applying microdialysis in a target area would deliver useful information about synaptic activity. The role of cannabis compounds for the treatment of epilepsy is currently receiving attention, with a special focus on cannabidiol. However, the effects in patients with different types of pharmacoresistant epilepsy, especially mTLE, are not conclusive. It was shown that some effects of cannabidiol are effectuated by an indirect effect on the CB1R.39,40 From our data, we speculate that a nonspecific cannabinoid drug could have different and possibly opposing effects within different parts of the epileptic network,
CLEEREN
ET AL.
which appears to be a major challenge for the development of eCB signaling system–based treatments for epilepsy.12 This study provides unique evidence that the ECS and more specifically the CB1R, is involved in the epileptogenesis of mTLE at a very early stage and becomes gradually more involved as seizure severity increases, resulting in chronic changes in CB1R availability. Use of an animal model has the advantage of starting from a naive brain and can therefore document a causal relationship between epileptogenesis and changes in the ECS, in contrast to patient studies, which show correlational changes in a chronic stage. Our results during the final stage of kindling were comparable with the changes seen in patients with chronic mTLE-HS, providing further support for the validity of the amygdala kindling model in rhesus monkeys for the study of temporal lobe epileptogenesis. Specifically targeted interactions with the ECS early during epileptogenesis could, therefore, be beneficial for the prevention of a kindling-like increase in seizure severity in patients. ACKNOWLEDGMENTS This work was supported by the Fund for Scientific Research Flanders (G.0745.09) and the Programma Financiering (PFV/10/008). Koen Van Laere is senior clinical fellow of the FWO Vlaanderen (Fund for Scientific Research, Flanders). The precursor to [18F]-MK-9470 was kindly donated by Merck Inc, USA. The authors thank Ann Van Santvoort, Peter Vermaelen, Ivan Sannen, Sofie Celen, Kwinten Porters, Johan Nuyts, and the Leuven PET radiopharmacy team for technical support during the PET experiments. Technical and administrative assistance of Christophe Ulens, Inez Puttemans, Piet Kayenbergh, Gerrit Meulemans, Jan Lenaerts, Stijn Verstraeten, Wouter Depuydt, Marc De Paep, Astrid Hermans, and Sara De Pril is greatly acknowledged. DISCLOSURE None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. ORCID Evy Cleeren https://orcid.org/0000-0003-0592-1354 Cindy Casteels https://orcid.org/0000-0003-0134-2232 Karolien Goffin http://orcid.org/0000-0002-7453-0229 Michel Koole https://orcid.org/0000-0001-5862-640X https://orcid.org/0000-0001-5200-7245 Koen Van Laere Peter Janssen https://orcid.org/0000-0002-8463-5577 Wim Van Paesschen https://orcid.org/0000-0002-85351699
CLEEREN
ET AL.
REFERENCES 1. Cleeren E, Casteels C, Goffin K, et al. Ictal perfusion changes associated with seizure progression in the amygdala kindling model in the rhesus monkey. Epilepsia. 2015;56:1366–75. 2. Wada JA, Mizoguchi T, Osawa T. Secondarily generalized convulsive seizures induced by daily amygdaloid stimulation in rhesus monkeys. Neurology. 1978;28:1026–36. 3. Friedman D, Sirven JI. Historical perspective on the medical use of cannabis for epilepsy: ancient times to the 1980s. Epilepsy Behav. 2017;70:298–301. 4. Rosenberg EC, Tsien RW, Whalley BJ, et al. Cannabinoids and epilepsy. Neurotherapeutics. 2015;12:747–68. 5. Devinsky O, Cross JH, Laux L, et al. Trial of cannabidiol for drug-resistant seizures in the dravet syndrome. N Engl J Med. 2017;376:2011–20. 6. Blair RE, Deshpande LS, Sombati S, et al. Activation of the cannabinoid type-1 receptor mediates the anticonvulsant properties of cannabinoids in the hippocampal neuronal culture models of acquired epilepsy and status epilepticus. J Pharmacol Exp Ther. 2006;317:1072–8. 7. Wallace MJ, Martin BR, DeLorenzo RJ. Evidence for a physiological role of endocannabinoids in the modulation of seizure threshold and severity. Eur J Pharmacol. 2002;452:295–301. 8. Marsicano G, Goodenough S, Monory K, et al. CB1 cannabinoid receptors and on-demand defense against excitotoxicity. Science. 2003;302:84–8. 9. Monory K, Massa F, Egertova M, et al. The endocannabinoid system controls key epileptogenic circuits in the hippocampus. Neuron. 2006;51:455–66. 10. Burns HD, Van Laere K, Sanabria-Bohorquez S, et al. [18F]MK9470, a positron emission tomography (PET) tracer for in vivo human PET brain imaging of the cannabinoid-1 receptor. Proc Natl Acad Sci USA. 2007;104:9800–5. 11. Goffin K, Van Paesschen W, Van Laere K. In vivo activation of endocannabinoid system in temporal lobe epilepsy with hippocampal sclerosis. Brain. 2011;134:1033–40. 12. Soltesz I, Alger BE, Kano M, et al. Weeding out bad waves: towards selective cannabinoid circuit control in epilepsy. Nat Rev Neurosci. 2015;16:264–77. 13. Van Paesschen W. Qualitative and quantitative imaging of the hippocampus in mesial temporal lobe epilepsy with hippocampal sclerosis. Neuroimaging Clin N Am. 2004;14:373–400, vii. 14. Nelissen N, Van Paesschen W, Baete K, et al. Correlations of interictal FDG-PET metabolism and ictal SPECT perfusion changes in human temporal lobe epilepsy with hippocampal sclerosis. NeuroImage. 2006;32:684–95. 15. Itoh T, Wakahara S, Nakano T, et al. Effects of anesthesia upon 18F-FDG uptake in rhesus monkey brains. Ann Nucl Med. 2005;19:373–7. 16. McLaren DG, Kosmatka KJ, Oakes TR, et al. A population-average MRI-based atlas collection of the rhesus macaque. NeuroImage. 2009;45:52–9. 17. Cleeren E, Premereur E, Casteels C, et al. The effective connectivity of the seizure onset zone and ictal perfusion changes in amygdala kindled rhesus monkeys. Neuroimage Clin. 2016;12: 252–61. 18. Styner M, Knickmeyer R, Joshi S, et al. Automatic brain segmentation in rhesus monkeys. Proc. of SPIE 2007;6512.
|
11
19. O’Brien TJ, O’Connor MK, Mullan BP, et al. Subtraction ictal SPET co-registered to MRI in partial epilepsy: description and technical validation of the method with phantom and patient studies. Nucl Med Commun. 1998;19:31–45. 20. Bhaskaran MD, Smith BN. Cannabinoid-mediated inhibition of recurrent excitatory circuitry in the dentate gyrus in a mouse model of temporal lobe epilepsy. PLoS ONE. 2010;5:e10683. 21. Falenski KW, Carter DS, Harrison AJ, et al. Temporal characterization of changes in hippocampal cannabinoid CB(1) receptor expression following pilocarpine-induced status epilepticus. Brain Res. 2009;1262:64–72. 22. Wallace MJ, Blair RE, Falenski KW, et al. The endogenous cannabinoid system regulates seizure frequency and duration in a model of temporal lobe epilepsy. J Pharmacol Exp Ther. 2003; 307:129–37. 23. Panikashvili D, Simeonidou C, Ben-Shabat S, et al. An endogenous cannabinoid (2-AG) is neuroprotective after brain injury. Nature. 2001;413:527–31. 24. Ludanyi A, Eross L, Czirjak S, et al. Downregulation of the CB1 cannabinoid receptor and related molecular elements of the endocannabinoid system in epileptic human hippocampus. J Neurosci. 2008;28:2976–90. 25. von Ruden EL, Bogdanovic RM, Wotjak CT, et al. Inhibition of monoacylglycerol lipase mediates a cannabinoid 1-receptor dependent delay of kindling progression in mice. Neurobiol Dis. 2015;77:238–45. 26. Magloczky Z, Toth K, Karlocai R, et al. Dynamic changes of CB1-receptor expression in hippocampi of epileptic mice and humans. Epilepsia. 2010;51(Suppl 3):115–20. 27. Van Paesschen W, Dupont P, Van Driel G, et al. SPECT perfusion changes during complex partial seizures in patients with hippocampal sclerosis. Brain. 2003;126:1103–11. 28. Mufson EJ, Mesulam MM, Pandya DN. Insular interconnections with the amygdala in the rhesus monkey. Neuroscience. 1981;6: 1231–48. 29. Isnard J, Guenot M, Ostrowsky K, et al. The role of the insular cortex in temporal lobe epilepsy. Ann Neurol. 2000;48:614–23. 30. Van Laere K, Koole M, Sanabria Bohorquez SM, et al. Wholebody biodistribution and radiation dosimetry of the human cannabinoid type-1 receptor ligand 18F-MK-9470 in healthy subjects. J Nucl Med. 2008;49:439–45. 31. Hill EL, Gallopin T, Ferezou I, et al. Functional CB1 receptors are broadly expressed in neocortical GABAergic and glutamatergic neurons. J Neurophysiol. 2007;97:2580–9. 32. Schwartz TH, Bonhoeffer T. In vivo optical mapping of epileptic foci and surround inhibition in ferret cerebral cortex. Nat Med. 2001;7:1063–7. 33. Casteels C, Vermaelen P, Nuyts J, et al. Construction and evaluation of multitracer small-animal PET probabilistic atlases for voxel-based functional mapping of the rat brain. J Nucl Med. 2006;47:1858–66. 34. Herkenham M, Lynn AB, Little MD, et al. Cannabinoid receptor localization in brain. Proc Natl Acad Sci USA. 1990;87:1932–6. 35. Romigi A, Bari M, Placidi F, et al. Cerebrospinal fluid levels of the endocannabinoid anandamide are reduced in patients with untreated newly diagnosed temporal lobe epilepsy. Epilepsia. 2010;51:768–72. 36. Breivogel CS, Childers SR, Deadwyler SA, et al. Chronic delta9tetrahydrocannabinol treatment produces a time-dependent loss of
12
37.
38.
39.
40.
| cannabinoid receptors and cannabinoid receptor-activated G proteins in rat brain. J Neurochem. 1999;73:2447–59. Romero J, Garcia L, Fernandez-Ruiz JJ, et al. Changes in rat brain cannabinoid binding sites after acute or chronic exposure to their endogenous agonist, anandamide, or to delta 9-tetrahydrocannabinol. Pharmacol Biochem Behav. 1995;51:731–7. Coutts AA, Anavi-Goffer S, Ross RA, et al. Agonist-induced internalization and trafficking of cannabinoid CB1 receptors in hippocampal neurons. J Neurosci. 2001;21:2425–33. Hayakawa K, Mishima K, Hazekawa M, et al. Cannabidiol potentiates pharmacological effects of Delta(9)-tetrahydrocannabinol via CB(1) receptor-dependent mechanism. Brain Res. 2008;1188: 157–64. Devinsky O, Cilio MR, Cross H, et al. Cannabidiol: pharmacology and potential therapeutic role in epilepsy and other neuropsychiatric disorders. Epilepsia. 2014;55:791–802.
CLEEREN
ET AL.
SUPPORTING INFORMATION Additional Supporting Information may be found online in the supporting information tab for this article.
How to cite this article: Cleeren E, Casteels C, Goffin K, et al. Positron emission tomography imaging of cerebral glucose metabolism and type 1 cannabinoid receptor availability during temporal lobe epileptogenesis in the amygdala kindling model in rhesus monkeys. Epilepsia. 2018;00:1–12. https://doi.org/10.1111/epi.14059
