Brain Advance Access originally published online on March 9, 2005
Brain 2005 128(6):1330-1343; doi:10.1093/brain/awh470
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Seizure-related short-term plasticity of benzodiazepine receptors in partial epilepsy: a [11C]flumazenil-PET study
1 EA1880, Federal Institute of Neurosciences, 2 CERMEP and the 3 Department of Functional Neurology and Epileptology, Neurological Hospital, Lyon and 4 CEA, Frédéric Joliot Hospital, Orsay, France
Correspondence to: Professor P. Ryvlin, Service de Neurologie Fonctionnelle et d'Epileptologie, Hôpital Neurologique, 59 Bd Pinel, 69003 Lyon, France E-mail: ryvlin{at}cermep.fr
Received October 14, 2004. Revised January 20, 2005. Accepted January 31, 2005.
 |
Summary |
|---|
 Top Summary IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
We have undertaken a test–re-test [11C]flumazenil (FMZ) PET study in 10 drug-resistant epileptic patients, including six with a mesiotemporal epilepsy (MTE), and 10 normal controls, in order to investigate seizure-related short-term plasticity of benzodiazepine (BZD) receptors. All subjects underwent two FMZ-PET scans at a 1 week interval. Patients benefited from a concurrent video-EEG monitoring which allowed determination of the duration of the interictal period (IP) preceding each PET. Test–re-test whole brain B'max variations, evaluated with a partial-saturation injection protocol, were similarly observed in patients and controls, suggesting a physiological modulation of BZD receptors. Five patients (50%), but no controls, also demonstrated clinically significant test–re-test FMZ-PET variations in the mesial temporal region. This was observed in all three patients with MTE and no hippocampal atrophy in whom only the PET study associated with the shortest IP correctly identified the epileptogenic zone. Statistical analysis revealed a significant effect of IP duration on BZD receptor B'max in MTE patients, suggesting that the shorter the IP, the lower the B'max in the epileptogenic hippocampus. FMZ-PET appears to be an interesting tool for investigating both normal and abnormal short-term modulations of the BZD receptor system, and should ideally be performed within a few days following a seizure in patients with MTE and a normal MRI.
Key Words: epilepsy; PET; flumazenil; benzodiazepine receptors; hippocampus
Abbreviations: BZD = benzodiazepine; FMZ = flumazenil; IP = interictal period; MTE = mesiotemporal epilepsy; ROI = region of interest; SPM = statistical parametric mapping; TLE = temporal lobe epilepsy
|
Introduction |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
[11C]Flumazenil (FMZ) and PET have been used for more than a decade to study benzodiazepine (BZD) receptors in epileptic patients (Savic et al., 1988
), as well as in various neurological disorders, partly based on the assumption that BZD receptor density offers a reliable marker of neuronal loss (Henry et al., 1993; Sette et al., 1993; Heiss et al., 2000). As a matter of fact, decreased [11C]FMZ binding was consistenly observed in the atrophic hippocampus in temporal lobe epilepsy (TLE) (Savic et al., 1993; Koepp et al., 1996; Debets et al., 1997; Ryvlin et al., 1998), and was found to correlate with autoradiographic measurements of [3H]FMZ binding performed on surgically removed tissue (Koepp et al., 1998). Correlations were also reported between in vitro [3H]FMZ binding and neuronal cell count in human epileptogenic hippocampi (Burdette et al., 1995). However, these correlations were not replicated when in vivo PET measurement of [11C]FMZ distribution volume was compared directly with the histologically assessed neuronal loss (Koepp et al., 1997), suggesting that part of the FMZ-PET abnormalities observed in epileptic patients might reflect functional changes of BZD receptors. In epileptic patients, the extent of FMZ-PET abnormalities was found to correlate positively with seizure frequency (Savic et al., 1996), whereas a normalization of such abnormalities was observed after successful epilepsy surgery in non-resected brain regions corresponding to projection areas of the epileptogenic zone (Savic et al., 1998). A focal increase in FMZ binding has also been reported in patients with cryptogenic neocortical epilepsy, suggesting an upregulation of BZD receptors (Hammers et al., 2003). Finally, we and others have observed falsely lateralizing mesial temporal FMZ-PET asymmetries in some patients with TLE and no hippocampal atrophy (Ryvlin et al., 1998; Koepp et al., 2000), suggesting either a downregulation of BZD receptors contralateral to seizure onset or an ipsilateral upregulation. These abnormalities spontaneously resolved in two of our patients who underwent a second FMZ-PET study 3 months after the first investigation (Ryvlin et al., 1999). Overall, long-term functional changes of the human BZD receptor system, which might occur over a period of a few months, appear to be a likely phenomenon in human partial epilepsy.
In addition, experimental data suggest the possibility of seizure-induced short-term plasticity of BZD receptors (McNamara et al., 1980; Valdes et al., 1982; Shin et al., 1985; Nobrega et al., 1989; Schmitz et al., 1991; Rocha et al., 1994). Whether patients with partial epilepsy exhibit comparable short-term BZD receptor changes, over days or weeks, as well as the magnitude and clinical significance of such variations remain unknown.
These issues have two major impacts: first, they question the clinical reliability of FMZ-PET investigations in the pre-surgical evaluation of drug-resistant partial epilepsy (Ryvlin et al., 1999; Koepp et al., 2000; Hammers et al., 2002); and, secondly, they might offer new insights into the pathophysiology of human epilepsy, and the physiological role of the BZD allosteric site of GABAA receptors.
In order to evaluate the possibility of physiological and seizure-induced short-term fluctuations of FMZ binding in humans, we have undertaken a prospective test–re-test PET study in 10 patients and 10 matched normal subjects, who underwent two FMZ-PET investigations at a 1 week interval.
|
Material and methods |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Patients
Ten epileptic patients suffering from refractory partial seizures were included in this study prospectively. They were selected over a period of 20 months from among a group of patients undergoing long-term video-EEG monitoring in our epilepsy surgery unit according to the following criteria: (i) no barbiturate, vigabatrin or tiagabine treatment during the 2 months, and no BZD during the 2 weeks preceding the study, in order to avoid major therapeutic interference with the BZD and GABAergic systems; (ii) patients gave their informed consent to participate in this study; and (iii) on the day preceding the first FMZ-PET, we judged that it was clinically appropriate to proceed to another week of video-EEG monitoring without tapering medication, in order to ensure a stable drug regimen between the two PET scans. According to French legislation, the study was approved by the local ethics committee.
Clinical data are presented in Table 1. Four men and six women (mean age ± SD: 35.4 ± 10.5 years) participated in the study. The mean age at onset of epilepsy was 9 years (range: 1–16) and the mean duration of epilepsy prior to PET examination was 26 years (range: 5–37). All patients underwent a pre-surgical evaluation, including a long-term video-EEG monitoring during which the two FMZ-PET investigations and an MRI were performed. In addition, six patients subsequently benefited from an intra-cranial EEG investigation.
|
Control subjects
Ten normal individuals were selected in order to match each of our patients on an individual basis, with respect to gender and age. Accordingly, there were four men and six women (mean age ± SD: 36 ± 9.5 years), whose age never differed by >4 years from that of their corresponding patient. All control subjects were free of BZD, barbiturate or any other anti-epileptic treatment.
Video-EEG monitoring
Long-term video-EEG monitoring lasted from 2 to 3 weeks, during which seizures were counted precisely. It was started at least 6 days before the first FMZ-PET study, and ran at least until the second PET investigation. Patients were recorded 5 days a week from 8.30 a.m. to 4.30 p.m. (8 h) and three nights a week from 9.30 p.m. to 6.30 a.m. (9 h). Recordings were stopped during the weekends, but patients stayed under medical control in the neurological ward of our institution, enabling the medical staff to be aware of any seizures occurring, either observed by proxy or recollected by the patient.
MRI
MRI was performed on a 1.5 T device (Magnetom 63SP; Siemens, Erlangen, Germany). A 3D mpr T1 sequence [TR (repetition time) 9.7 ms, TE (echo time) 4 ms], providing 1 mm thick slices, a spin echo T2 sequence (TR 2260 ms, TE 45 and 90 ms) parallel to the hippocampal plane, as well as a turbo-spin echo T2 sequence (TR 3000 ms, TE 16 and 98 ms) perpendicular to the bihippocampal plane, were acquired.
[11C]Flumazenil PET (FMZ-PET): acquisition protocol
The two FMZ-PET studies were performed in every subject at exactly a 1 week interval, at around 10 a.m. FMZ (RO15-1788) was labelled with 11C, using the methylation process described by Mazière et al. (1984). We used a high resolution tomograph (Exact ECAT HR+, Siemens, Erlangen, Germany), allowing a dynamic 3D acquisition and providing 12 consecutive frames of 63 contiguous 2.42 mm thick slices, with an isotropic spatial resolution near 5 mm3 FWHM (full width at half-maximum) measured with 18F according to the NEMA protocol (Brix et al., 1997). A thermolabile plastic face mask ensured the stable position of the head in the scanner gantry. Attenuation correction was measured in each individual using a 68Ge transmission scan. We used a partial saturation protocol consisting of a single intravenous injection of a mixture of 5 mCi of [11C]FMZ and 0.01 mg/kg of unlabelled FMZ, followed by a 55 min acquisition of the emission data. As described by Delforge et al. (1995, 1996, 1997), this single injection allows the calculation of B'max parametric images without arterial blood sampling.
PET data analysis
Co-registration of PET and MRI data
The two PET acquisition series (12 scans) were co-registered with the Woods criteria (Woods et al., 1993) implemented in a home-made program (ACTIVIS, Cognitive Science Institute, Bron, France). Target and object images were the summed frames 8–12 (corresponding to the acquisition period from 20 to 55 min post-injection) of the first and second series, respectively. The MRI subsequently was realigned to static images of the PET volumes (REGISTER, Montreal Neurological Institute, McGill University, Montreal, Canada). We judged the quality of the co-registration procedure by looking at the superimposed PET and MRIs, using the Surface Matching function of ANALYZETM (Biomedical Imaging Resource, Mayo Foundation, Rochester, MN), and by visually comparing the two sets of printed PET images obtained in each individual. In addition, we evaluated the anatomical matching of pontine and hippocampal regions of interest (ROIs) placed on MRIs and subsequently transferred on the corresponding FMZ-PET slices, provided that both the pons and the hippocampus can be easily identified on FMZ-PET images. Overall, the quality of the co-registration procedure and of MRI-guided placement of ROIs on FMZ-PET images proved satisfactory in all cases.
Calculation of B'max parametric images
Circular ROIs, 10 mm in diameter, were first placed over the midportion of the pons, on the 7–9 consecutive MRI slices displaying that structure, the latter being considered as an adequate reference for the calculation of the non-specific FMZ binding (Delforge et al., 1995). These ROI's were then transferred onto the corresponding FMZ-PET slices. In addition, a single circular ROI, 15 mm in diameter, was directly placed onto the PET volume, over the occipital cortex which usually displays a high concentration of BZD receptors. A partial-saturation model, previously validated (Delforge et al., 1997), was then applied to obtain B'max parametric images for each FMZ-PET scan. This model, based on a Scatchard plot, estimates the free ligand concentration (F) in the pons, and the range of the bound ligand concentration (B) from the natural decrease of the latter observed in the occipital cortex. The (F, B) pairs obtained at various times provide a straight line whose intercept with the B-axis represents the B'max parameter, and the slope of which corresponds to 1/KdVR (Delforge et al., 1997). Ultimately, we obtained, for each PET volume, a set of 63 contiguous 2.42 mm thick parametric images of BZD receptor B'max.
Spatial normalization and reorientation of B'max parametric images
B'max parametric images, as well as the corresponding MRIs were spatially normalized onto the ICBM (International Consortium for Brain Mapping) template using SPM 99 (Statistical Parametric Mapping, Wellcome Neurological Institute, London, UK). The latter were used in subsequent SPM analyses, detailed below. For the other analyses, all sets of images were reoriented in: (i) 6 mm thick axial slices oriented in the hippocampal planes; and (ii) 5 mm thick coronal slices, perpendicular to the latter.
Calculation of whole brain B'max test–re-test variation index
The whole brain (wb) average B'max (wbB'max = mean B'max over intracranial voxels) was computed with SPM 99 for each scan. The test–re-test variation index (VI) of the wbB'max was calculated in each subject as follows:
 |
Visual analysis of focal B'max test–re-test variations
The previously reconstructed 6 mm thick axial slices and 5 mm thick coronal images were printed on high quality A4 colour paper to display the entire brain on a single sheet. B'max data from the two PET studies performed in each patient were printed separately, but using an identical colour scale, taking advantage of the whole brain activity ratio calculated between the two PET studies. The printed PET images were labelled anonymously and placed in a random order, then qualitatively reviewed by two independent PET experts blinded to all other data. These investigators were asked to report the precise anatomical location of visually significant abnormalities. Ultimately, the two experts met and agreed on a final consensual report for their only two discrepancies. Except for patient 6 who presented with a post-traumatic lesion easily identifiable on PET images, investigators could not readily match the two PET images from the same individual during this reviewing process. Therefore, a bias in the visual analysis of PET data related to the recognition of a previously assessed brain was unlikely.
ROI-assessed quantitative analysis of hippocampal B'max variations
This analysis was prompted by the results of our visual assessment of PET data, showing that short-term focal changes of BZD receptors were most frequently observed in the mesial temporal structures. It was performed by an investigator blinded to all other data. We manually traced an anatomical ROI over the hippocampus using the 6 mm thick hippocampal-oriented MRI which best displayed that structure. In order to reduce the subjective bias related to the manual tracing and the partial volume effect related to brain structures surrounding the hippocampus within the plane of interest, we placed four contiguous circular ROIs, 13 mm in diameter, centred within the anatomical ROI. We then transferred these ROIs on the FMZ-PET B'max parametric image to obtain hippocampal B'max (hipp B'max) values and calculated the following indexes:
 |
 |
 |
 |
Statistical analysis of whole brain and ROI-derived data
Hippocampal asymmetry index values of individual patients were considered as abnormal when varying by >2 SD from the mean control values. For this analysis, patients' PET 1 and PET 2 values were compared with those derived from the first and second PET of control subjects, respectively. In fact, since only minor test–re-test differences were observed in the hippocampal asymmetry index of normal individuals (see Table 2), all of the hippocampal abnormalities detected in individual patients proved significant, using either one or the other set of normal data (i.e. PET 1 or 2).
|
We investigated a possible scan order effect by comparing the whole brain B'max and the ROI-derived hippocampal B'max values of the first and second FMZ-PET scans performed in each individual using a non-parametric test (Wilcoxon). Analysis of hippocampal data was performed using either the normalized or non-normalized hippocampal B'max values for PET 2.
We then investigated whether patients showed larger test–re-test changes of whole brain B'max and hippocampal test–re-test asymmetry variation index than controls, using the Student t test.
According to the result of this analysis, we then searched for seizure-related hippocampal B'max interscan variations. We first evaluated the duration of the interictal period (IP) preceding each PET study, and defined in each patient which of his two FMZ-PET evaluations was associated with the shortest IP. In order to ensure the reliability of these data, we only considered seizures which occurred once the patients were hospitalized in our epilepsy surgery unit. Six patients did not experience a seizure during the 6 days separating the onset of the video-EEG monitoring from the first PET, including five subjects who presented one or several seizures during the week separating the two FMZ-PET scans, and one patient who only suffered seizures 2 days after the second PET study. Though we could not calculate precisely all IP durations in those patients, we could nevertheless ascertain which of the two PET scans was associated with the shortest IP based on the available data (see Table 2). This was the rationale for not using the exact delay between PET and seizures, but the more robust information as to which of the two FMZ-PET scans was associated with the shortest IP, in all the following statistical analyses. We used a non-parametric test (Wilcoxon) to compare in each patient the hippocampal B'max values of the two FMZ-PET scans associated with the shortest and the longest IP, respectively.
This comparison was performed for the hippocampi ipsilateral and contralateral to seizure onset separately. Due to the heterogeneity of our population, we restricted this analysis to the subgroup of patients with mesiotemporal epilepsy (MTE). We therefore included data from the six patients whose ictal onset zone proved to be mesial temporal (patients 1, 2, 3, 5, 7 and 10), and excluded those from patients 4, 8 and 9, whose ictal onset zone was not well defined but most probably was heterogeneous and partly extra-temporal, as well as those of patient 6 who presented with an extensive left postero-lateral temporal post-traumatic lesion and EEG focus. This analysis was carried out using either normalized or non-normalized hippocampal B'max values for PET 2.
SPM analysis of seizure-related focal B'max and variation
To evaluate further the possibility of seizure-related FMZ-PET changes in MTE, we performed a voxel-based analysis of covariance (ANCOVA) using SPM. Among the six patients with a mesiotemporal focus, four presented with a left-sided, and two with a right-sided ictal onset zone. As reported in previously published series (Bouilleret et al., 2002; Merlet et al., 2004), we flipped the PET volumes of the two patients with a right TLE in order to pool the six unilateral TLE patients in the same analysis, where all epileptogenic zones appeared as being left temporal. Their six matched control subjects were also included in this analysis. Accordingly, we flipped the PET volume of the two control subjects who matched the two right TLE patients. The IP classification, which depends on the delay between the PET and the last seizure, could not be applied directly in normal individuals. Thus, we opted to use in each control subject the same classification as that of their matched TLE patient. For example, in patient 1, the second PET study was associated with the shortest IP; likewise, the second PET of his matched control subject was also classified as the shortest IP scan. This procedure actually allowed us to control for a spurious scan order effect, related to the IP duration classification.
We restricted this SPM analysis to the epileptogenic hippocampus based on the following rationale: the small number of scans available for this analysis dramatically limited its statistical power using a whole brain approach and a correction for multiple comparisons. This problem could be partly circumscribed by narrowing the analysis to a much smaller number of voxels. Seizure-related changes of BZD receptors are expected to occur primarily within the epileptogenic zone. This hypothesis was supported by our previous observations of focal FMZ binding variations in TLE (Ryvlin et al., 1999), and was confirmed by the visual and ROI-derived analyses previously performed in this study. The spatially normalized MRI data of all patients and controls were averaged and the boundaries of the hippocampus were manually outlined. The resulting selected volume of 462 voxels was applied as an inclusive mask onto the B'max parametric images. The design matrix of the ANCOVA was formed by two factors: the whole brain B'max (mean B'max over voxels) treated as a covariate of non-interest, and a dummy covariate representing the duration of the IP preceding each PET study (classified as either the shortest or the longest), treated as the factor of interest. The variances of these regressors were then estimated and compared with the residual variance. For any voxel whose variance test was significant (P < 0.001), a post hoc analysis was performed. This post hoc analysis consisted of a Student t test of the B'max parametric images contrasting the shortest IP versus the longest IP in patients relative to the same comparison in controls. As a result of this test, we obtained an SPM map representative of a seizure-related modulation of BZD receptor density, independent of the whole brain B'max variation. All voxels exceeding a Z score of 2.3 (P < 0.01 uncorrected for multiple comparisons) were displayed on the resulting SPM map, before applying correction for multiple comparisons with the small volume correction (SVC) formula of SPM.
Other pre-surgical and surgical procedures
After the above-described protocol was completed and the electroclinical and imaging data reviewed, six patients underwent an intra-cranial EEG investigation (stereo-EEG) to ensure the localization of the ictal onset zone. In three TLE patients (patients 1, 2 and 10), stereo-EEG was primarily performed because of a strictly normal MRI. Intra-cranial EEG recordings were considered mandatory in patient 5 in whom an amygdalo-hippocampal radiosurgery was contemplated. In patient 7 with left TLE and early ictal speech disorders, stereo-EEG was performed in order to exclude a posterior temporal ictal onset zone, and to tailor the surgical resection with respect to the language area if necessary. None of these TLE patients were suspected to suffer from a bi-temporal focus. Conversely, patient 8 underwent a stereo-EEG with the hope of delineating a single focus within its bilateral perisylvian polymicrogyria.
Seven patients had a surgical cure of their epilepsy; these comprised five anterior temporal lobectomies, one amygdalo-hippocampal radiosurgery followed by an anterior temporal lobectomy, and one callosotomy.
|
Results |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Clinical data (see Table 1)
Based on all available electroclinical and MRI data, we concluded that the ictal onset zone was: (i) mesial temporal in six patients (patients 1, 2, 3, 5, 7 and 10), half of whom presented with MRI signs of hippocampal sclerosis, and the other half with a normal MRI; (ii) postero-lateral temporal in patient 6 whose epilepsy was symptomatic of an extensive post-traumatic porencephalic lesion; (iii) bilateral temporo-occipital with a left-sided predominance in patient 8 who suffers from a bilateral perisylvian polymicrogyria; and (iv) unlocalized, with a left-sided predominance of scalp-EEG abnormalities in the two remaining patients (patients 4 and 9) whose MRI was normal.
Regarding the MTE group, the origin of seizures was demonstrated by intra-cranial EEG recordings in five patients and appeared very likely in the sixth patient (patient 3) who presented with typical electroclinical features and MRI signs of mesial temporal sclerosis. Concerning patient 6, the left postero-lateral temporal origin of seizures was suggested by the occurrence of aphasia as the first ictal sign, and a posterior temporal ictal onset on scalp-EEG recordings, consistent with the location of his post-traumatic lesion. Regarding patient 8, the bilateral temporo-occipital seizure onset with a left-sided predominance was evidenced by intra-cranial ictal EEG recordings. Finally, patients 4 and 9 presented no localizing clinical or scalp-EEG abnormality during their seizures, even though the ictal EEG discharge predominated over the left hemisphere.
The five TLE patients who underwent a standard anterior temporal lobectomy have all been seizure free since the operation, with a postoperative follow-up ranging from 21 to 45 months (32 ± 11). Patient 5 who underwent an amygdalo-hippocampal radiosurgery has not significantly improved after 4 years of follow-up, and subsequently benefited from a successful anterior temporal lobectomy. Finally, the callosotomy performed in patient 8 was ineffective.
Altogether, we could reliably conclude that patients 1, 2, 3, 5, 7 and 10 suffered from MTE, whereas the other patients most probably presented another form of epilepsy.
Whole brain BZD receptor B'max test–re-test variation index (see Table 2)
The whole brain B'max variation index between the two PET studies ranged from –29 to +13% (mean ± SD = –3 ± 12%) in patients, and from –6 to +23% (mean ± SD = 5 ± 9%) in normal subjects, and did not prove to be significantly different between the two populations. No scan order- or seizure-related effect was detected by non-parametric test, suggesting that these factors do not primarily account for the whole brain B'max test–re-test variations observed in this study.
Visual analysis of focal B'max test–re-test variations (see Table 2)
All 10 patients demonstrated one or several areas of focal decreased B'max. Eight of these 10 patients showed visually detectable variations between the two FMZ-PET scans, that either affected part (n = 5) or all (n = 3) of the abnormalities observed on each individual scan. Among the three patients whose entire areas of focal decreased B'max were detected on only one of the two PET studies, two presented with abnormalities of opposite lateralization on their first and second evaluations, respectively (patients 9 and 10: Fig. 1), while the third patient (paient 1) had one of his two PET studies considered as normal (Fig. 2).
|
|
The clinical impact of focal FMZ-PET test–re-test variations differed according to the type of epilepsy. In the three TLE patients with MRI signs of hippocampal sclerosis (patients 3, 5 and 7), the two PET studies consistently showed an ipsilateral mesial temporal decreased B'max, the magnitude of which varied between scans. The other test–re-test variations observed in these three patients consisted of focal abnormalities that were detected in only one of the two FMZ-PET scans, and which always affected brain regions ipsilateral to the ictal onset zone, such as the anterior temporal, the parietal or the temporo-occipital cortex, but also the hippocampus proper in one patient (Fig. 3).
|
In the three TLE patients with a normal MRI (patients 1, 2 and 10), FMZ-PET test–re-test focal variations had a more dramatic clinical impact. Indeed, in each of these patients, only one of the two PET studies demonstrated an abnormal pattern that pointed unequivocally to the epileptogenic temporal lobe. In patient 1, the first PET proved normal whereas the second demonstrated an obvious mesial temporal decreased B'max ipsilateral to seizure onset (Fig. 2). In patient 2, one of the two PET scans showed bilateral asymmetric abnormalities (Fig. 4), whereas unilateral findings were observed on the other. In patient 10, the two PET scans demonstrated opposite mesial temporal asymmetries, which were falsely lateralizing for the first evaluation and correctly pointed to the epileptogenic zone for the second (Fig. 1).
|
In the two patients with unlocalized epilepsy (patients 4 and 9), a mesial temporal decreased B'max was observed in one of the two PET studies and not in the other. In one of these patients, this PET finding pointed to the side contralateral to that of predominating EEG abnormalities (patient 9).
ROI-assessed quantitative analysis of hippocampal B'max variations (see Table 2)
The mean (±SD) hippocampal asymmetry index of normal individuals was 5.3 ± 11.3% for PET 1, and 6.8 ± 12.6% for PET 2. As illustrated in Table 2, seven patients demonstrated a significantly abnormal hippocampal assymetry index on either their first or their second FMZ-PET study.
Large test–re-test variations of the hippocampal B'max asymmetry index were observed in the majority of our patients, whose magnitude proved to be significantly higher than in controls. Indeed, the mean test–re-test asymmetry variation index (±SD) was 24 ± 13% in patients, whereas it was 6 ± 3% in controls (P < 0.0005). As a result of these variations, five patients demonstrated a hippocampal asymmetry which significantly differed from control values in one of their studies, but not in the other. In patients with an obvious unilateral TLE, test–re-test changes in asymmetry index mostly reflected hippocampal B'max variations ipsilateral to seizure onset. However, less marked changes could also be observed in the contralateral hippocampus and contributed in some patients to the asymmetry index variation. In patients with unlocalized or bilateral epilepsy, large B'max variations affected either the hippocampus ipsilateral or contralateral to the side of predominating EEG abnormalities depending on whether we would consider normalized or non-normalized data for the second PET study.
We did not find any significant scan order effect on the hippocampal B'max values. Conversely, we observed a significant seizure-related effect in the epileptogenic hippocampus of TLE patients with lower hippocampal B'max values for the PET study associated with the shortest IP, as compared with the other PET. This seizure-related effect was demonstrated using either normalized or non-normalized data for the second PET study (P < 0.01 and P < 0.005, respectively). This effect was not observed in the contralateral hippocampus.
SPM analysis of seizure-related focal B'max variation (Fig. 5)
The ANCOVA-designed SPM map demonstrated a cluster of 63 voxels which varied as a function of the IP duration independently of the whole brain mean B'max variation (P < 0.01 uncorrected for multiple comparisons at the voxel level). After correcting for multiple comparisons, this focal variation remained significant within the epileptogenic hippocampus (P = 0.001 at the cluster level). In this area, B'max values were lower for PET studies associated with the shortest IP than for those associated with the longest IP. This result is consistent with that provided by the ROI-assessed quantitative analysis, and suggests the possibility of a transient post-ictal decreased hippocampal BZD receptor B'max, which would progressively resolve over time.
|
|
Discussion |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
This test–re-test FMZ-PET study conducted 1 week apart demonstrates that significant seizure-related focal variations of BZD receptor B'max frequently occur in patients suffering from drug-resistant partial epilepsy. In the subgroup of patients with MTE, these focal variations predominantly affected the epileptogenic hippocampus, with an inverse relationship between the intensity of the decreased hippocampal B'max and the duration of the IP preceding the PET study. In TLE patients whose MRI was normal, as well as in patients with unlocalized or multifocal epilepsies, hippocampal B'max variations were likely to result in misleading indications regarding the side of ictal onset.
Several methodological issues must be addressed before discussing the pathophysiological and clinical impacts of our findings. Both the calculation of B'max parametric images and the comparison of the two FMZ-PET volumes obtained in each of our patients strongly rely on the quality of the MRI–PET, and PET–PET co-registration. We have used a validated method of co-registration (Woods et al., 1993), and have carefully evaluated that co-registered images precisely fitted with each other. We also visually verified that the position of the pontine and hippocampal ROIs were equally centred within their corresponding anatomical brain structures, as displayed on MRIs and PET images. Overall, the potential issue of misaligned PET data appears unlikely to play a significant role with respect to the test–re-test B'max variations observed in our patients.
Another important issue relates to the evaluation of the IP duration preceding each PET study. Indeed, since we could not obtain reliable IP values for all PET studies, but could ascertain in each patient which of the two PET studies was associated with the shortest IP, we used that latter information as a surrogate marker to look for seizure-related FMZ-PET changes. In addition, MTE patients might have presented subclinical hippocampal ictal discharges, undetected by scalp-EEG monitoring, which could also interact with FMZ-PET findings. Though these limitations do not invalidate our statistical results, they prohibit a precise temporal analysis of the dynamics of epilepsy-driven BZD receptor changes.
Large whole brain B'max variations, exceeding 10%, were observed in three of our 10 patients and two of our 10 normal subjects, reaching 29% in one patient. We did not find any satisfactory explanation for these variations. A scan order effect might have occurred, through changes in the level of anxiety or state of wakefulness between the two scans, but this hypothesis was not supported by our statistical analysis. In patients, the role of anti-epileptic drugs also appeared unlikely, since a stable regimen was maintained between the two PET studies, and none of the prescribed medications directly interfere with the BZD-GABAergic system. A seizure-related effect was also ruled out, based on the analysis performed in this study which only considered which of the two FMZ-PET scans was associated with the shortest and the longest IP, respectively. However, we cannot exclude that part of the whole brain B'max test–re-test variations might be related to other seizure-related factors, such as the exact delay between the PET study and the last previously occurring seizure, or the type and duration of that seizure. Uncertainties regarding these latter parameters, together with the small number of patients included in our study, precluded the feasibility of a multivariate analysis that would take into account all potentially relevant factors.
Random variations, related to the intrinsic limitations of the methods used in our protocol, might also contribute to the test–re-test whole-brain B'max findings. For instance, head motions which necessarily occur during any 55 min scan, regardless of the immobilization procedure, are likely to vary in direction and magnitude between two different PET sessions. This phenomenon might have a significant impact on the [11C]FMZ activity measured in the pontine ROIs, which represent one of the major parameters in the calculation of B'max (Delforge et al., 1997). To date, only the distribution volume and the transport rate of [11C]FMZ have been studied twice at rest in the same normal individuals (Holthoff et al., 1991), with an interscan delay of only 120 min. The mean test–re-test change in global cortex [11C]FMZ distribution volume was –6.4% in this study, a difference which was not significant. Interestingly, these results are consistent with our own data in epileptic patients, where the mean test–re-test whole brain B'max variation index was –3%, with no significant scan order effect. It is therefore possible that the whole brain B'max changes observed in our patients only reflect physiological fluctuations and the degree of FMZ-PET measurement errors, using the quantification method used in this study. Whatever the causes of the whole brain B'max variations, these did not have a significant clinical impact.
Conversely, test–re-test variations in focal B'max and mesial temporal asymmetry index were clinically relevant in our patients. Importantly, both the calculation of asymmetry index and the visual analysis of PET data are insensitive to whole brain B'max. All but two of our patients demonstrated transient focal FMZ-PET abnormalities, which were detected in only one of their two PET studies. These transient findings were of various degrees of significance, depending on the type of epilepsy.
In the three patients with an MTE and hippocampal sclerosis on MRI, the two PET studies showed a clearcut mesial temporal decreased B'max ipsilateral to seizure onset, though in one patient a transient abnormality was observed in the hippocampus proper. Less marked transient abnormalities were also observed in distant ipsilateral regions, such as the anterior lateral temporal, the temporo-occipital and the posterior parietal cortex. These regions might represent areas of seizure propagation, and previous series have pointed to the possibility of a transient decreased FMZ binding in projection areas of the epileptogenic zone (Savic et al., 1998). Specifically, frontal FMZ-PET abnormalities, which were observed in TLE patients preoperatively, proved to resolve after a successful anterior temporal lobectomy (Savic et al., 1998). The same authors also found that TLE patients with extra-temporal decreased FMZ binding had more frequent seizures than those without, supporting the view that these extra-temporal findings represent a seizure-related functional change of the BZD receptor system (Savic et al., 1996). We, and others, have also demonstrated the possibility of temporo-polar, lateral temporal and extra-temporal decreased FMZ binding in patients with TLE (Ryvlin et al., 1998; Hammers et al., 2001; Bouilleret et al., 2002), which might be indicative of an epileptogenic zone extending outside the mesial temporal structures (Ryvlin et al., 1998). The fact that such abnormalities transiently occur in our patients with pure MTE clearly questions the clinical value of these FMZ-PET findings. Regarding the mesial temporal B'max asymmetry, which was consistently demonstrated by visual and ROI analyses in this population, we observed that the intensity of this asymmetry varied greatly between the two PET evaluations in some patients. This type of variation also questions the relationship previously reported between the degree of decreased FMZ binding and that of the histologically assessed neuronal loss within epileptogenic hippocampi (Koepp et al., 1997). Based on our findings, it seems reasonable to consider that hippocampal cell loss only represents one of the factors contributing to the FMZ-PET abnormalities observed in patients with TLE and MRI signs of hippocampal sclerosis.
In all three TLE patients with a normal MRI, the two PET studies showed strikingly different patterns, and these variations had a significant clinical impact. In patient 1 with a right TLE, the first PET was normal whereas the second showed a clear-cut right mesial temporal abnormality. In patient 2 with a left TLE, the second PET demonstrated an ipsilateral mesial temporal decreased B'max, whereas the first PET showed bilateral asymmetric temporo-limbic abnormalities which could not help in lateralizing the seizure focus. In the third patient, the two PET studies showed a mesial temporal asymmetry of opposite direction. In all three cases, the PET study which provided the most pertinent information was that associated with the shortest IP duration, ranging from 4 to 5 days. FMZ-PET has been recognized as poorly sensitive, and potentially misleading in TLE patients with a normal MRI (Ryvlin et al., 1998, 1999; Koepp et al., 2000). We have also previously reported striking mesial temporal asymmetry variations over a period of 3 months in two other TLE patients with no hippocampal sclerosis (Ryvlin et al., 1999). The present study confirms these findings, and suggests that the FMZ-PET abnormalities observed in such patients are primarily functional and transient. It also provides evidence for performing FMZ-PET early after a seizure in order to obtain reliable data in this clinical situation. However, avalaible data from this and other series do not provide a clear indication as to the range of IP duration which could be at risk of resulting in misleading FMZ-PET findings.
In the three patients with unlocalized or bilateral epilepsy, it remains difficult to conclude on the reliability of FMZ-PET abnormalities. However, since these patients did not show any hippocampal abnormality on MRI, it was interesting to note that two of them demonstrated a mesial temporal asymmetry that was observed in only one of their two PET studies.
The mechanisms underlying the short-term unilateral changes of mesial temporal B'max observed in our epileptic patients can hardly be explained by physiological fluctuations. As for the whole brain B'max, we have searched for a scan order effect, which could be mediated by the varying degree of anxiety during the two PET studies, and failed to demonstrate such an effect. Conversely, in the homogeneous group of TLE patients, we found a significant seizure-related effect, with greater mesial temporal asymmetries for PET studies associated with the shortest IP duration. This effect was demonstrated using a non-parametric comparison of the ROI-derived hippocampal asymmetry index, as well as an SPM analysis centred over the epileptogenic hippocampus. This latter analysis showed a seizure-related effect, independent of the whole brain B'max changes, suggesting that the closer the PET study to the last previously occurring seizure, the lower the BZD receptor B'max in that region. Despite the very small number of scans included in this study, this effect reached significance after applying a correction for multiple comparisons. It was limited in size, however, a finding which is likely to reflect the heterogeneous spatial distribution of the hippocampal involvement among our six TLE patients.
The seizure-related hippocampal B'max fluctuations observed in TLE patients reinforces the view that the visually detected mesial temporal changes observed in our patients cannot be explained solely by physiological fluctuations of BZD receptors, or by the technical limitations of our PET studies. At the present time, there are no other data in human epilepsy regarding the short-term plasticity of BZD receptors as studied with FMZ-PET. In amygdala kindled rats, a 22–31% increase of BZD receptor density was demonstrated in the dentate gyrus 24 h after a stage 5 seizure (Mc Namara et al., 1980; Valdes et al., 1982; Shin et al., 1985; Nobrega et al., 1989), but not 2 or 4 weeks later (Shin et al., 1985; Nobrega et al., 1989; Schmitz et al., 1991; Rocha et al., 1994). A more prolonged increase of flunitrazepam binding was reported in the kaïnate model of epilepsy (Rocha et al., 1999). Conversely, a downregulation of BZD receptors was evidenced in kindled rabbits with no detectable underlying hippocampal neuronal loss, 2 weeks after the last seizure (Kurokawa et al., 1994). In our patients, the short-term BZD receptor changes observed in the epileptogenic hippocampus clearly pointed to a transient decreased expression or availability of these receptors during the days following a seizure. With respect to these observations, several important differences between human and animal studies must be stressed. Our patients suffered from longstanding drug-resistant TLE, in contrast to the short lasting epileptic condition obtained in the various animal models that have been studied. The in vitro quantification of BZD receptor density in animal studies, using autoradiography, does not distinguish functional receptors located on the outer surface of the plasmic membrane from internalized receptors, and cannot provide information regarding the receptor occupation by its endogeneous ligand. Conversely, in vivo PET studies, using [11C]FMZ, will only label externalized BZD receptors, and might potentially reflect the degree of receptor occupancy by an endogeneous ligand (Delforge et al., 2001), raising the issue that variations in that ligand concentration might partly account for the test–re-test FMZ-PET focal changes observed in our patients. Though this issue remains highly speculative according to our current knowledge of the endogeneous BZD neurotransmission system (Costa et al., 1991; Do-Rego et al., 2001), it relates to recent advances in the molecular genetics of generalized epilepsy with febrile seizures plus (GEFS+) indicating that a loss of function of the BZD receptor system, due to a mutation in the GABARG2 gene coding for the 
2 subunit of the GABAA receptor, can provoke an epileptic disorder (Wallace et al., 2001).
In conclusion, the demonstration of a seizure-related short-term plasticity of the BZD receptor system in human partial epilepsy opens new avenues in the fields of epilepsy research and neuroscience, suggesting that the regulatory mechanisms of this neurotransmitter system can be studied in humans with FMZ-PET. Our results also indicate that the degree of FMZ binding might not be a reliable marker of neuronal density, at least in epileptic patients. Finally, our findings in TLE patients with a normal MRI suggest that FMZ-PET should be performed within a few days following a seizure, in order to optimize the sensitivity and the specificity of this presurgical investigation.
|
Acknowledgements |
|---|
We wish to thank Christian Pierre for his technical assistance, Isabelle Merlet for critical reading of the manuscript, and Dr W. Hunkeler at F. Hoffman-La Roche AG (Basel, Switzerland) for the generous gift of RO15-1788 and precursors. This study was supported by grants from the Hospices Civils de Lyon, the Université Claude Bernard Lyon 1 and the Programme Hospitalier de Recherche Clinique.
|
References |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Bouilleret V, Dupont S, Spelle L, et al. Insular cortex involvement in mesiotemporal lobe epilepsy: a positron emission tomography. Ann Neurol 2002; 51: 202–8.[CrossRef][ISI][Medline]
Brix G, Zaers J, Adam LE, et al. Performance evaluation of a whole-body PET scanner using the NEMA protocol. National Electrical Manufacturers Association. J Nucl Med 1997; 38: 1614–23.
Burdette DE, Sakurai SY, Henry TR, et al. Temporal lobe central benzodiazepine binding in unilateral mesial temporal lobe epilepsy. Neurology 1995; 45: 934–41.[Abstract]
Costa E, Guidotti A. Diazepam binding inhibitor (DBI): a peptide with multiple biological actions. Life Sci 1991; 49: 325–44.[CrossRef][ISI][Medline]
Debets RM, Sadzot B, Van Isselt JW, et al. Is [11C]Flumazenil PET superior to 18FDG PET and 123I-iomazenil SPECT in presurgical evaluation of temporal lobe epilepsy? J Neurol Neurosurg Psychiatr 1997; 62: 141–50.[Abstract]
Delforge J, Pappata S, Millet P, et al. Quantification of benzodiazepine receptors in human brain using PET, [11C]flumazenil, and a single-experiment protocol. J Cereb Blood Flow Metab 1995; 15: 284–300.[ISI][Medline]
Delforge J, Spelle L, Bendriem B, et al. Quantitation of the benzodiazepine receptor in human brain using a partial-saturation method. J Nucl Med 1996; 37: 5–11.
Delforge J, Spelle L, Bendriem B, et al. Parametric images of benzodiazepine receptors concentration using a partial-saturation injection. J Cereb Blood Flow Metab 1997; 17: 343–55.[CrossRef][ISI][Medline]
Delforge J, Bottleander M, Pappata S, et al. Absolute quantification by positron emission tomography of the endogenous ligand. J Cereb Blood Flow Metab 2001; 21: 613–30.[ISI][Medline]
Do-Rego JL, Mensah-Nyagan AG, Beaujean D, et al. The octadecaneuropeptide ODN stimulates neurosteroid biosynthesis through activation of central-type benzodiazepine receptors. J Neurochem 2001; 76: 128–38.[CrossRef][ISI][Medline]
Hammers A, Koepp MJ, Labbé C, et al. Neocortical abnormalities of [11C]-flumazenil PET in mesial temporal lobe epilepsy. Neurology 2001; 56: 897–906.
Hammers A, Koepp MJ, Hurlemann R, et al. Abnormalities of grey and white matter [11C]-flumazenil binding in temporal lobe epilepsy with normal MRI. Brain 2002; 125: 2257–71.
Hammers A, Koepp MJ, Richardson MP, et al. Grey and white matter flumazenil binding in neocortical epilepsy with normal MRI. A PET study of 44 patients. Brain 2003; 126: 1300–18.
Heiss WD, Kracht L, Grond M, et al. Early [(11)C]Flumazenil/H(2)O positron emission tomography predicts irreversible ischemic cortical damage in stroke patients receiving acute thrombolytic therapy. Stroke 2000; 31: 366–9.
Henry TR, Frey KA, Sackellares JC, et al. In vivo cerebral metabolism and central benzodiazepine receptor binding in temporal lobe epilepsy. Neurology 1993; 43: 1998–2006.
Holthoff VA, Koeppe RA, Frey KA, et al. Differentiation of radioligand delivery and binding in the brain: validation of a two-compartment model for [11C]-flumazenil. J Cereb Blood Flow Metab 1991; 11: 745–52.[ISI][Medline]
Koepp MJ, Richardson MP, Brooks DJ, et al. Cerebral benzodiazepine receptors in hippocampal sclerosis: an objective in vivo analysis. Brain 1996; 119: 1677–87.
Koepp MJ, Richardson MP, Labbé C, et al. 11C-flumazenil PET, volumetric MRI, and quantitative pathology in mesial temporal lobe epilepsy. Neurology 1997; 49: 764–73.
Koepp MJ, Kieran SP, Hand BS, et al. In vivo [11C]flumazenil-PET correlates with ex vivo [3H]flumazenil autoradiography in hippocampal sclerosis. Ann Neurol 1998; 43: 618–26.[CrossRef][ISI][Medline]
Koepp MJ, Hammers A, Labbé C, et al. 11C-Flumazenil PET in patients with refractory temporal lobe epilepsy and normal MRI. Neurology 2000; 54: 332–9.
Kurokawa K, Jibiki I, Matsuda H, et al. Comparison of benzodiazepine receptor and regional cerebral blood flow imaging of epileptiform foci in hippocampal kindled rabbits: a study with in vivo double tracer autoradiography using 125I-iomazenil and 99mTc HMPAO. Brain Res 1994; 642: 303–10.[CrossRef][ISI][Medline]
Mazière M, Hantraye P, Prenant C, et al. Synthesis of ethyl 8-fluoro-5,6-dihydro-5-(11C)methyl-6-oxo-4H-imidazo(1,5-a) benzodiazepine-3-carboxylate (Ro 15-1788): a specific radioligand for in vivo study of central benzodiazepine receptors by positron emission tomography. Int J Appl Radiat Isot 1984; 35: 973–8.[CrossRef][Medline]
McNamara JO, Peper AM, Patrone V. Repeated seizures induce long term increase in hippocampal benzodiazepine receptors. Proc Natl Acad Sci USA 1980; 77: 3029–32.
Merlet I, Ryvlin P, Costes N, et al. Statistical parametric mapping of 5-HT(1A) receptor binding in temporal lobe epilepsy with hippocampal ictal onset on intracranial EEG. Neuroimage 2004; 22: 886–96.[CrossRef][ISI][Medline]
Nobrega JN, Kish SJ, McIntyre Burnham W. Autoradiographic analysis of benzodiazepine binding in entorhinal kindled rat brains. Brain Res 1989; 49: 315–22.
Rocha L, Ackermann RF, Chugani HT, Engel Jr. Chronic pretreatment with naloxone modifies benzodiazepine receptor binding in amygdaloid kindled rats. Epilepsy Res 1994; 17: 135–43.[CrossRef][ISI][Medline]
Rocha L, Ondarza-Rovira R. Characterization of benzodiazepine receptor binding following kainic acid administration: an autoradiography study in rats. Neurosci Lett 1999; 262: 211–4.[CrossRef][ISI][Medline]
Ryvlin P, Bouvard S, Le Bars D, et al. Clinical utility of flumazenil-PET versus 18F-fluorodesoxyglucose-PET and MRI in refractory partial epilepsy: a prospective study in 100 patients. Brain 1998; 121: 2067–81.
Ryvlin P, Bouvard S, Le Bars D, Mauguière F. False lateralization of 11C-flumazenil PET in temporal lobe epilepsy with no hippocampal atrophy. Neurology 1999; 53: 1882–5.
Savic I, Roland P, Sedvall G, et al. In vivo demonstration of reduced benzodiazepine receptor binding in human epileptic foci. Lancet 1988; 8616: 863–6.
Savic I, Ingvar M, Stone Elander S. Comparison of [(11)C]flumazenil and [(18)F]FDG as PET markers of epileptic foci. J Neurol Neurosurg Psychiatry 1993; 56: 615–21.[Abstract]
Savic I, Svanborg E, Thorell JO. Cortical benzodiazepine receptor changes are related to frequency of partial seizures: a positron emission tomography study. Epilepsia 1996; 37: 236–44.[CrossRef][ISI][Medline]
Savic I, Blomqvist G, Halldin C, et al. Regional increases in [11C]flumazenil binding after epilepsy surgery. Acta Neurol Scand 1998; 97: 279–86.[ISI][Medline]
Schmitz E, Loscher W, Honack D, et al. Kindling does not induce persistent changes in fluorographic labeling patterns of benzodiazepine binding proteins in various rat brain regions. Epilepsy Res 1991; 9: 105–12.[CrossRef][ISI][Medline]
Sette G, Baron JC, Young AR, et al. In vivo mapping of brain benzodiazepine receptor changes by positron emission tomography after focal ischemia in the anesthetized baboon. Stroke 1993; 24: 2046–58.
Shin C, Pedersen HB, McNamara JO. Gamma aminobutyric acid and benzodiazepine receptors in the kindling model of epilepsy: a quantitative radiohistochemical study. J Neurosci 1985; 5: 2696–701.[Abstract]
Valdes F, Dasheiff RM, Birmingham F, et al. Benzodiazepine receptor increases after repeated seizures: evidence for localization to dentate granule cells. Proc Natl Acad Sci USA 1982; 79: 193–7.
Wallace RH, Marini C, Petrou S, et al. Mutant GABA(A) receptor gamma2-subunit in childhood absence epilepsy and febrile seizures. Nat Genet 2001; 28: 49–52.[CrossRef][ISI][Medline]
Woods RP, Mazziotta JC, Cherry SR. MRI–PET registration with automated algorithm. J Comput Assist Tomogr 1993; 17: 536–46.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Didelot, P. Ryvlin, A. Lothe, I. Merlet, A. Hammers, and F. Mauguiere PET imaging of brain 5-HT1A receptors in the preoperative evaluation of temporal lobe epilepsy Brain, September 12, 2008; (2008) awn220v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Hammers, M.-C. Asselin, R. Hinz, I. Kitchen, D. J. Brooks, J. S. Duncan, and M. J. Koepp Upregulation of opioid receptor binding following spontaneous epileptic seizures Brain, April 1, 2007; 130(4): 1009 - 1016. [Abstract] [Full Text] [PDF]
|
|