Topography of brain glucose hypometabolism and epileptic network in glucose transporter 1 deficiency

Epilepsy Research (2015) 110, 206—215

journal homepage: www.elsevier.com/locate/epilepsyres

Topography of brain glucose hypometabolism and epileptic network in glucose transporter 1 deficiency Cigdem Inan Akman a,b,∗, Frank Provenzano c, Dong Wang a, Kristin Engelstad a, Veronica Hinton a, Julia Yu a, Ronald Tikofsky c, Masonari Ichese c, Darryl C. De Vivo a a

Department of Neurology, Division of Pediatric Neurology, Colleen Giblin Research Laboratory, Columbia University College of Physician & Surgeons, United States b Department of Neurology, Comprehensive Epilepsy Center, Columbia University College of Physician & Surgeons, United States c Department of Radiology, Kreitchman PET Center, Columbia University College of Physician & Surgeons, United States Received 30 April 2014; received in revised form 21 September 2014; accepted 11 November 2014 Available online 11 December 2014

KEYWORDS Glucose transporter deficiency; Positron emission tomography; Brain glucose metabolism; Statistical parametric mapping

Summary Rationale: 18 F fluorodeoxyglucose positron emission tomography (18 F FDG-PET) facilitates examination of glucose metabolism. Previously, we described regional cerebral glucose hypometabolism using 18 F FDG-PET in patients with Glucose transporter 1 Deficiency Syndrome (Glut1 DS). We now expand this observation in Glut1 DS using quantitative image analysis to identify the epileptic network based on the regional distribution of glucose hypometabolism. Methods: 18 F FDG-PET scans of 16 Glut1 DS patients and 7 healthy participants were examined using Statistical parametric Mapping (SPM). Summed images were preprocessed for statistical analysis using MATLAB 7.1 and SPM 2 software. Region of interest (ROI) analysis was performed to validate SPM results. Results: Visual analysis of the 18 F FDG-PET images demonstrated prominent regional glucose hypometabolism in the thalamus, neocortical regions and cerebellum bilaterally. Group comparison using SPM analysis confirmed that the regional distribution of glucose hypo-metabolism was present in thalamus, cerebellum, temporal cortex and central lobule. Two mildly affected

∗ Corresponding author at: Columbia University, College of Physicians & Surgeons, 180 Fort Washington Avenue, New York, NY 10032, United States. Tel.: +1 212 305 7549. E-mail address: [email protected] (C.I. Akman).

http://dx.doi.org/10.1016/j.eplepsyres.2014.11.007 0920-1211/© 2014 Published by Elsevier B.V.

Epilepsy and Regional glucose hypometabolism in Glut1 DS

207

patients without epilepsy had hypometabolism in cerebellum, inferior frontal cortex, and temporal lobe, but not thalamus. Glucose hypometabolism did not correlate with age at the time of PET imaging, head circumference, CSF glucose concentration at the time of diagnosis, RBC glucose uptake, or CNS score. Conclusion: Quantitative analysis of 18 F FDG-PET imaging in Glut1 DS patients confirmed that hypometabolism was present symmetrically in thalamus, cerebellum, frontal and temporal cortex. The hypometabolism in thalamus correlated with the clinical history of epilepsy. © 2014 Published by Elsevier B.V.

Introduction

glucose hypometabolism in an effort to map the epileptic network in Glut1DS.

Glucose transporter 1 deficiency syndrome (Glut1DS) is a genetically determined developmental encephalopathy resulting from insufficient transport of glucose into the brain (De Vivo et al., 1991). Cardinal clinical features include infantile-onset seizures, acquired microcephaly, ataxia, dysarthria, dystonia, intellectual disability, and motor retardation (Pearson et al., 2013). The majority of Glut1 DS patients present with seizures in infancy. However, seizure onset beyond the first year of life has been also described in the number of studies with normal intelligence and alteration in SLC2A1 gene (Suls, Pong). In contrast to the earlier definition of Glut1 DS, majority of these children with late seizure onset did not have the cardinal features of the syndrome that have lead to the expansion of the clinical spectrum. Absence seizures and generalized tonic clonic seizures are the most common seizure type in this syndrome (Leary et al., 2003). Despite the progress in clinical spectrum, diagnosis and recognition of the syndrome even in the individuals with milder phenotype; the pathophysiology underlying the epileptogenesis remains obscure. Brain glucose metabolism in this clinical condition has been studied using 18 F-FDG-PET brain imaging. Qualitative analysis based on the visual interpretation of 18 F-FDG-PET data revealed a global decrease in glucose metabolism (Pascual et al., 2002). Regional hypometabolism was also noted in thalamus, cerebellum and neocortical regions. Visual interpretation of 18 F-FDG-PET is the most traditional method for qualitative analysis. Introduction of quantitative methods to analyze 18 F-FDG-PET data has refined this information and provided a more precise topographical understanding of the regional vulnerability and severity of the metabolic insult in various neurological disorders and in focal epilepsy (Cummings et al., 1995; Duncan et al., 1997; Engel, 1984; McMurtray et al., 2008; Rintahaka et al., 1993; Schapiro et al., 1992). Quantitative analysis of imaging data can be achieved by Statistical Parametric Mapping (SPM), an effective, objective, and reliable method that supplements visual interpretation (Salek-Haddadi et al., 2003; Swartz et al., 1999). This quantitative method provides a voxel based analysis of metabolic activity that permits whole brain global analysis. We hypothesized that the degree of glucose hypometabolism would vary from one region to another based on the clinical phenotype, and the regional distribution of glucose uptake would correlate with the clinical features. In this study, we searched for a correlation between the epilepsy history and regional vulnerabilities to

Methods Participants Clinical features of this patient cohort were reported in our earlier study (Pascual et al., 2002). Sixteen patients diagnosed with Glut1 DS underwent 18 F-FDG-PET imaging. The study was approved by the Institutional Review Board of Columbia University. Informed consent was obtained from patients and their parents. Mean age at the time of the imaging was 12.4 ± 9.9 years (range: 1.3—39). Except for two patients, cerebrospinal fluid (CSF) glucose concentration was less than 40 mg/dl. One patient was not tested and the other had a CSF glucose concentration of 52 mg/dl. Pathogenic GLUT1 mutations were demonstrated by gene sequence analysis in all patients. The demographic information, history of seizures, antiepileptic drug (AED) treatment, EEG findings and neuroimaging studies were reviewed. The Columbia Neurological Score (CNS) was used to assess the phenotypic severity of patients diagnosed with Glut1 DS (Kaufmann et al., 2004). This is a semi-quantitative tool that scores the following physical examination domains: (1) height, weight, and head circumference; (2) general medical exam; (3) funduscopic exam; (4) cranial nerves; (5) stance and gait; (6) involuntary movements; (7) sensation; (8) cerebellar function; (9) muscle bulk, tone and strength; (10) tendon reflexes, (11) Babinski sign; and (12) other findings. Results for each of these domains were scored as ‘‘normal’’ or ‘‘abnormal’’ and summarized in the CNS, ranging from 0 to 76, with 76 being perfect. It was previously shown that this instrument has good inter-rater reliability and correlates with other measures of disease severity. Based on CNS scores, the neurological phenotype was described as severe (CNS 40—49); moderate (CNS 50—59) or mild (CNS 60—69). The minimal phenotype (CNS 70—76) merged with the control subjects. The 18 F-FDG-PET images of 7 healthy adults were used as the control group. Informed consent was obtained in every case. Mean age of the healthy participants was 44 ± 10.9 years (range 36—58). No neurological or other medical problems were reported in this group, and their CNS scores were normal (70—76).

PET scanning protocol: image acquisition An interictal 18 F-FDG-PET scan was obtained in patients following a bolus injection of 10 mCi 18 F-FDG-PET after a 6-h

208

C.I. Akman et al.

fast. Scans were obtained using a Siemens ECAT EXACT 47 scanner (in plane spatial resolution 5.8 mm, axial resolution 4.3 mm FWHM at the center) 30 min after the injection. A 35—40 min emission scan was acquired with attenuation correction applied to all the images.

Analysis of

18 F-18 F-FDG-PET

images

Statistical analysis SPM analysis. 18F-FDG-PET data were analyzed using SPM (SPM 2, Wellcome Department of Imaging Neuroscience, University College London, UK). The analysis was performed on a Dell workstation, using MATLAB 7.0 (Mathworks, Sherborn, MA, USA). Prior to statistical analysis, images were spatially normalized to the standard brain template from the Montreal Neurological Institute (MNI). Images from the two patients who were younger than age 6 years were excluded from SPM analysis. The reason for exclusion was an earlier report describing a significant artifact in SPM analysis associated with spatial normalization of pediatric brains to an adult brain template in children younger than 6 years of age (Muzik et al., 2000). To remove interparticipant anatomical variability, voxels with dimensions of 2 mm × 2 mm × 2 mm were re-sampled. Spatially normalized images were smoothed by convolution, using an isotropic Gaussian kernel with 12 mm FWHM. The aim of smoothing was to increase the signal-to-noise ratio to permit evaluations of the subtle anatomic variations. The effect of global activity was removed by normalizing the counts of each voxel to the global count of the whole brain using proportional scaling. The general linear model was employed to perform the appropriate voxel-by-voxel statistical test in the form of group comparisons (two-sample t-test): using this approach three analyses were performed: Analysis 1: A comparison between the patient group and healthy participants; Analysis-2: A comparison between subgroups, (a) Glut1 DS with epilepsy, (b) Glut1 DS without epilepsy history versus and (c) healthy participants. Analysis-3: SPM analysis using single subject covariate model to study the correlation between the decreased glucose uptake and clinical variables such as disease duration, CNS score, head circumference, CSF glucose value at the time of initial diagnosis (p < 0.05, corrected with extend threshold of 100 voxels). Each analysis produced a t statistic for each voxel which constituted the statistical parametric map, SPM {t}. The resulting SPM {t} was then transformed into a normal distribution, SPM {Z}, using a voxel height reaching to the significance p < 0.05 (corrected) with clusters consisting of a minimum of 100 contiguous voxels (extend threshold (ke): 100 voxels). The spatial coordinates of the significant voxels were applied to identify the corresponding brain areas based on the Talairach and Tournoux atlas using the Talairach Daemon software (Research Imaging Center, UTHSC, San Antonio) (Talairach and Tournoux, 1988). Region of interest analysis. This analysis was performed to validate the results obtained from SPM

analysis. The anatomical regions which suggested glucose hypometabolism by SPM analysis were selected as regions of interest (ROI). These regions were intuitively drawn on axial slices to an SPM-normalized template brain using the Fusion tool in PMOD software (PMOD software, version 2.5, Zurich, Switzerland). The volume of interest (VOI) map was saved and then applied to both control and Glut1 DS patient groups (Ichise et al., 2006). Averaged counts for each ROI (right and left regions of hippocampus, temporal neocortex, thalamus, cerebellum, and lentiform nucleus) were then obtained including a whole brain value. Prefrontal cortex demonstrated insignificant glucose hypometabolism compared to the healthy control based on SPM analysis of the 18 F-FDG-PET. Therefore, frontopolar prefrontal cortex voxel counts were chosen as the reference region to compare with voxel counts in other ROIs. The average glucose uptake for the region including the entire brain was used as a denominator for ratio comparison. Percent differences were calculated between Glut1 DS and Control groups and an independent sample t-test was performed between the two groups. A Bonferroni correction for multiple comparisons was applied for a p < 0.05. This method creates ratios that can be quantitatively compared within and between groups while accounting for unique metabolic differences between patients and tracer dose.

Results Clinical features Clinical features and pathogenic mutations are shown in Table 1 and Fig. 2. Severity of the clinical phenotypes as determined by the CNS scores ranged from severe (3 patients), moderate (3 patients), mild (9 patients) to minimal (1 patient). All but two patients (Table 1) had a history of seizures (87%). Seizures were reported as the initial clinical manifestation, leading to the diagnosis in 8 patients (50%). Absence and myoclonic seizures were the most common seizure types. Epilepsy duration was 10.3 ± 7 years (range: 1—31) and 8 patients were on a ketogenic diet at the time of imaging. Earlier EEG recordings demonstrated a background activity of 4—6 Hz superimposed on a 14—16 Hz activity without hemispheric asymmetry. Previous EEGs from the patients with a history of seizures reported generalized spike and polyspike discharges (2—3 Hz; amplitude of 200—300 ␮V). EEG recordings were not obtained at the time of the PET scan. However, the patients were observed closely (by KE) throughout the imaging procedure for any signs of clinical seizures or behavioral changes suggestive of seizures. Other neurological symptoms included dystonia in 11 (68%), ataxia in 13 (81%), dysarthria in 12 (75%), paroxysmal dyskinesia in 5 (31%) and opsoclonus in 4 (25%) patients. Two patients had no clinical seizures or other cardinal neurological signs of Glut1 DS (Table 1). Except for these 2 patients, the other 14 patients had more than 2 cardinal features. Intellectual function ranged from Average Ability to Moderate Intellectual Disability, as determined by averaging performance on the Peabody Picture Vocabulary Test (a

Clinical features of patients with Glut1 DS

Patient

Mutation

CSF glucose

RBC uptake

CNS

PET age (y)

KD at PET

1st event age (m)

Earliest symptom

Seizure

Microcephaly

Dystonia

Dysartria

Ataxia

PED

A

c.87delC, c.88A > T p.R333W p.R126H p.R153C p.T310I c.562-563 in C, c.189-190in TCCTGCCCACCACGCTCACACC c.790delC, c.504506delCTC p.R333W p.E146K p.Y449X p.R93Q p.R126C p.R93Q c.907 del G

29

47

49

10.40

Yes

1.5

Seizure

Yes

No

No

Yes

Yes

No

30 38 36 27 32

40 63 51 43 59

53 54 43 47 55

10.90 16.8 6.25 9.20 6.20

Yes No Yes Yes Yes

0.5 18 2 2 3

Seizure Seizure Seizure Seizure Opsoclonus

Yes Yes Yes Yes Yes

No No Yes Yes Yes

Yes Yes Yes Yes No

Yes Yes Yes Yes Yes

Yes Yes Yes Yes Yes

Yes No No No No

39

26

60

9.20

Yes

3

Opsoclonus

Yes

Yes

Yes

Yes

Yes

No

26 39

51 55

64 63

1.20 6.60

Yes Yes

1.5 24

Seizure DD

Yes Yes

Yes Yes

Yes Yes

Yes No

Yes No

No Yes

38 32 25 52 na 34 26

43 46 43 77 47 59 45

67 63 61 73 68 64 60

11.70 11.80 12.70 39.00 32.00 7.60 7.00

Yes Yes Yes No No Yes Yes

2 8 2 na 4.5 60 3

Apnea Seizure Seizure LD Seizure PED Opsoclonus

Yes Yes Yes No Yes No Yes

No Yes Yes No No No Yes

Yes Yes Yes No No No Yes

No Yes Yes No No Yes Yes

No Yes Yes No Yes Yes Yes

Yes No No No Yes Yes No

B C D E F G

H I J K L M N O P

Epilepsy and Regional glucose hypometabolism in Glut1 DS

Table 1

PED: paroxysmal dyskinesia; DD: developmental delay; KD: ketogenic diet; LD: learning disability.

209

210 measure of receptive vocabulary) with performance on the Ravens Colored Matrices (a measure of spatial reasoning). Individual raw scores on these tests were converted to age standardized scores with a mean of 100 and a standard deviation of 15. Scores can range from Very Superior (>130) to Severe Intellectual Disability (<35). One participant was too young to be assessed. Among the 15 who were evaluated, four were Average (between 90 and 109), three were Low Average (between 80 and 89), two were Borderline Impaired (between 70 and 79), four were Mildly Intellectually Disabled (between 50 and 70), and one was Moderately Intellectually Disabled (between 35 and 49). A total of six participants had scores less than 70, and were considered to have Intellectual Disability. Adaptive behavioral function ranged from Average to Moderately Deficient, as determined by parent/caregiver report on the Vineland Adaptive Behavior Scales, a semi structured interview probing for information about daily function in Communication, Daily Living Skills and Socialization Skills. Individual raw scores on the test are converted to age standardized scores with a mean of 100 and a standard deviation of 15. Adaptive Behavior Composite scores can range from High Adaptive Level (>130) to Profoundly Deficient Adaptive Skills (<25). Information was collected on 13 of the 16 participants who had parents or caregivers available for interview. Two had Adequate adaptive skills (between 85 and 115), three had Moderately Low scores (between 70 and 84), two scored in the Mild Deficit range (between 55 and 70) and six scored in the Moderate Deficit range (between 35 and 55). Nine participants had Adaptive Behavior Deficits, with scores less than 70. Visuomotor integration, the ability to coordinate visual information with motor output, was measured using the Beery—Buktenica Developmental Test of Visual-Motor Integration to assess ability to reproduce simple line drawings that increase in complexity. Age standardized scores (with a mean of 100 and a standard deviation of 15, and a range of 55—145) were obtained on 14 of the 16 participants. One scored in the Low Average range (80—89), three scored in the Borderline Impaired range (70—79), and ten scored in the Mild Deficit range (55—70). Notably, five participants received the lowest possible score on this measure. Ten participants scored lower than 70, indicating deficient visual-motor integration.

C.I. Akman et al.

Visual interpretation of

18 F-18F-FDG-PET

All 18 F-18F-FDG-PET scans were reviewed by an experienced nuclear medicine physician, and the qualitative results were previously published by our group (Pascual et al., 2002). Glucose hypometabolism, as previously reported, was apparent in bilateral thalamus and neocortical regions in all patients (Fig. 1). Glucose metabolism was relatively preserved in bilateral lentiform nuclei, occipital, prefrontal and mesial frontal cortices. Brain MRI was available for eight patients. Two were abnormal, one with mild frontal lobe atrophy and the other with atrophy of the cerebellar vermis. There were no signs of clinical seizures during PET scanning. Therefore, we assumed that the PET imaging reflected the interictal state.

Quantitative analysis of

18 F-FDG-PET

SPM analysis Analysis-1: Glut1 DS versus controls: For all Glut1 DS patients, symmetrical glucose hypometabolism was reported in the cortical and subcortical regions of cerebrum, and cerebellum (Table 2). Despite the symmetrical appearance of glucose hypometabolism in Glu1 DS, inter- and intra-regional differences were reported in thalamus, frontal cortex, and temporal cortex. In frontal cortex, glucose hypometabolism was significant in the dorsolateral aspect, including precentral gyrus and inferior frontal gyrus. However, glucose uptake in mesial frontal and frontopolar cortical regions did not differ significantly from controls. In temporal cortex, the hypometabolism was more prominent in mesial cortical regions including hippocampus and parahippocampal gyrus in addition to neocortex, middle and superior temporal gyrus. In cerebellum, hypometabolism was seen in the posterior and anterior lobe (Table 2). There was no evidence of increased global or regional glucose uptake in any Glut1 DS or control participants. Analysis-2: Glucose hypometabolism and the clinical phenotype: PET images of the fourteen patients with seizure history were compared to the images of the control participants. Four anatomic regions showed the significant hypometabolism; thalamus, cerebellum, temporal lobe and central lobule. Lower FDG-uptake was observed in the left

Figure 1 Example of 18F-FDG-PET image of a patient with Glut1 DSGlut1 DS, axial, coronal, and sagittal view. Hypometabolism was seen in neocortical region and mesial temporal structures (green arrows), thalamus (purple arrow) and cerebellum (yellow arrow).

Epilepsy and Regional glucose hypometabolism in Glut1 DS

211

Figure 2 The mutations are displayed on this figure depicting the genetic alteration reported in Table 1 and their effect on GLUT1 protein sequence. Black bar represents the mRNA extending from 5 to 3 . Protein sequence was divided into boxes that were colored according to the cellular location of amino acids [trans-membrane (yellow), cytosolic (pink), extracellular (blue)]. Blue arrow points the mutation reported in two patients without history of epilepsy. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

and right thalamus (z-scores: 5.43 and 5.57); the right and left middle temporal gyrus (z-scores: 4.61 and 4.44); the right and left superior temporal gyrus (z-scores: 4.97 and 4.73); left inferior parietal lobule (z-score: 4.48), and the right superior frontal gyrus (z-score: 5.03) (p < 0.05 corrected) (Supplementary Table 1, Fig. 3a). In the two patients

Table 2

without epilepsy history, glucose hypometabolism was not present in thalamus; but was present predominantly in the right and left cerebellum, the right inferior frontal gyrus (z: 4.20), the right and left uncus (z: 4.09 and z: 4.26), and the right inferior parietal lobule (z: 5.12) (p < 0.05 corrected) (Supplementary Table 2, Fig. 3b).

Brain regions with glucose hypometabolism: All GLUT1 DS compared to control.

Side

Anatomical location

L L R R

Thalamus, Thalamus, Thalamus, Thalamus,

L R R L

Mid-frontal gyrus Mid-frontal gyrus Inferior frontal gyrus Inferior frontal gyrus

R L R

Cerebellum, posterior lobe Cerebellum, posterior lobe Cerebellum, anterior lobe

R L L R R L R L

Hippocampus Hippocampus Parahippocampal gyrus Parahippocampal gyrus Superior temporal gyrus Superior temporal gyrus Middle temporal gyrus Middle temporal gyrus

R L R R

Precentral gyrus Postcentral gyrus Postcentral gyrus Inferior parietal lobule

BA

p-corrected

z

Coordinates x, y, z

<0.001 <0.001 <0.001 <0.001

6.26 5.39 5.74 5.34

−20, −28, 0 −14, −24, 6 24, −2, 4 14, −16, 6

0.015 0.014 0.001 0.001

4.30 3.61 5.18 4.57

−30, 40, 12 34, 44, 6 28, 24, −24 −30, 8, −16

0.001 0.001 0.001

5.69 5.55 4.40

54, −56, −28 −34, −70, −26 8, −30, −26

20 20 27 28 38 38 21 21

0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001

5.35 4.68 5.48 5.34 4.97 4.73 4.61 4.44

20, 2, −24 −14, 4, −24 −20, −32, −2 16, −12, −8 26, 6, −46 −44, 24, −20 44, 2, −28 −64, −44, −10

4 3 3 40

0.002 0.001 0.002 0.0001

3.49 3.78 3.57 4.48

60, −2, 16 −58, −10, 22 60, −14, 30 −50, −54, 40

sub-lobar VPLN sub-lobar VPLN 10 10 47 47

BA: Brodmann area. Areas are best estimate based on the Talairach and Tournoux atlas. z-Score gives the magnitude of statistical difference (p < 0.05, corrected).

212

C.I. Akman et al.

Discussion

Figure 3 SPM results of the patients analyzed based on the epilepsy history in Glut1 DS: the patients with (a) and without (b) epilepsy history versus controls; (p < 0.05, corrected, threshold for voxel clusters, kE 100). Hypometabolism is significant in thalamus, neocortex, mesial temporal cortex with epilepsy history (a). Without epilepsy, hypometabolsim was limited to frontal neocortex, sparing thalamus and mesial temporal structures (b).

Analysis-3: Correlation with the other clinical features: There were no significant correlations between FDG uptake and: age when the initial symptom occurred, CSF glucose concentration at the time of diagnosis, CNS score, RBC glucose uptake, or head circumference (data not shown). Regions of interest analysis The ratio of glucose uptake between the ipsilateral prefrontal cortex and other ROIs was performed for patient groups and healthy participants and validated SPM results. Hypometabolism was determined to be significant in the thalamus, left and right cerebellum, hippocampus, temporal neocortex and left lentiform nucleus of Glut1 DS patients (p < 0.005) (Table 3).

In this study, we examined the relationship between the classical clinical phenotype of Glut1 DS and the regional brain glucose metabolism. Voxel based analysis of 18 F-FDG-PET imaging documented significant glucose hypometabolism in neocortex, thalamus and cerebellum in Glut1 DS patients. Furthermore, inter-regional hypometabolic differences correlated with the clinical history of epilepsy. In contrast, we found no correlation between glucose hypometabolism and CSF glucose concentration, head circumference, or chronological age. Thalamocortical network: The quantitative evaluation of FDG PET data reinforce and strengthen our original qualitative report describing the distribution of glucose hypometabolism in thalamus, mesial temporal structures, and other neocortical regions (Pascual et al., 2002, 2007). Application of voxel based analysis confirmed the bilateral involvement clearly demonstrated in cerebellar hemispheres in Glut1 DS. Although brain hypometabolism was widespread, regional differences existed. The most pronounced differences were observed in the cerebellum, thalamus, temporal, and frontal cortical regions. Hypometabolism was most profound in cerebellum and thalamus. Hypometabolism in thalamus appeared to be nearly homogenous with symmetric involvement of most thalamic nuclei. Moreover, the regional distribution of glucose hypometabolism correlated with the clinical phenotype that was dominated by seizures. In the presence of epilepsy history, hypometabolism was prominent in thalamus, central lobule and temporal lobe, and specifically the mesial temporal cortex. There was no correlation between increasing age and glucose hypometabolism (data not shown) that highlights the fact that expression of Glut 1 may lead to a fixed functional imprint in the pattern of glucose hypometabolism. In other words, hypometabolism pattern may not show a progressive changes with increasing age or clinical course of seizures. Seizures represent the most striking clinical feature leading to the diagnosis of Glut1 DS. At least 90% of 150 patients, in our experience, have clinical seizures. Most seizures present in early infancy, and are refractory to conventional antiepileptic drug treatment (De Vivo et al., 2002; Leary et al., 2003). The majority of patients present more than one seizure type. Generalized tonic—clonic seizures and absence seizures are the seizure types most commonly reported. Improvement in the clinical and EEG features follow the consumption of a carbohydrate-enriched meal or a ketogenic diet (Akman et al., 2010; von Moers et al., 2002). However, most patients with Glut1 DS are initially diagnosed with ‘‘idiopathic’’ childhood absence epilepsy and treated with antiepileptic medications (Pong et al., 2012). The central role of the thalamus has been studied extensively in absence seizure models. Absence seizures are associated with synchronous burst firing of a network involving the thalamic reticular nucleus, thalamic relay neurons and cortical pyramidal neurons. The components of this thalamic network include glutamatergic fibers between neocortex and nucleus reticularis; and gamma-aminobutyric acid (GABA)-ergic fibers from nucleus reticularis and thalamic relay neurons. Any alteration in this network may lead to sustained thalamocortical oscillatory burst-firing and

Epilepsy and Regional glucose hypometabolism in Glut1 DS Table 3

213

Region of interest analysis: the difference in metabolic activity between the groups.

ROI

Control

Glut1-DS

% Decrease

p

Left thalamus Right thalamus Left lentiform Right lentiform Left hippocampus Right hippocampus Left lateral temporal Right lateral temporal Left cerebellum Right cerebellum

0.93685 0.961486 1.041152 1.090558 0.809816 0.770573 0.895248 0.951566 1.051299 1.039381175

0.828406 0.812495 1.049221 1.073046 0.657609 0.643001 0.869446 0.891539542 0.82986366 0.837510372

11.5 15 0.7 1.6 18 16.5 13 17 22 20

<0.005 <0.005 <0.005 ns <0.005 <0.005 <0.005 <0.005 <0.005 <0.005

ns: not significant.

result in synchronized spike and wave discharges regardless of the underlying cause (Banerjee and Snead, 1994; Coenen et al., 1992; Coulter et al., 1990; Crunelli and Leresche, 2002; de Curtis and Avanzini, 1994; Liu et al., 1991, 1992; McCormick and Contreras, 2001; Niedermeyer, 1996; Richards et al., 1995; Shouse et al., 1988; Snead, 1995; Snead et al., 1990; Tucker et al., 2007; von Krosigk et al., 1993; Wallenstein, 1994). Furthermore, with advances in imaging modalities, thalamic activation during absence seizures has been reported using 18 F-FDG-PET imaging or functional brain MRI (Chan et al., 2006; Labate et al., 2005). Morphometric analysis of brain gray and white matter volume has been performed on subjects with CAE and has demonstrated volume reduction both in thalamus and the subcallosal gyrus (Chan et al., 2006). These reports support earlier published findings suggesting the key role of the thalamus and the thalamocortical network in the epileptogenic state underlying absence seizures. These neurophysiological findings and functional imaging data in CAE and Glut1 DS are congruent and consistent with the speculation that thalamocortical network dysfunction is a shared pathophysiological mechanism of epileptogenesis in all clinical settings. What may account for the regional differences of glucose hypometabolism in Glut1 DS? These regional differences might be explained by the regional differences in Glut1 protein density. Researchers have demonstrated that Glut1 protein density in microvessels supplying the cerebral cortex, hippocampus, and cerebellum is higher when compared to other brain regions (Choeiri et al., 2002; Dobrogowska and Vorbrodt, 1999; Zeller et al., 1997). In cerebral cortex, Glut1 density is particularly high in the central lobule (Choeiri et al., 2002). Ontogeny: Maturational changes in regional cerebral glucose utilization have been described in and correlated with functional development of the organism. In newborn rats, glucose utilization is 50% of the adult values in both thalamus and hypothalamus, and reaches adults values by 21 days of age (Neglig et al., 1988). In hippocampus and cerebral cortex, glucose utilization develops slowly and reaches 75% of the adult value by 21 days of age. In newborns, cerebral glucose uptake is significantly lower in basal ganglia and cortical brain regions (Kinnala et al., 1996). By 5 weeks of life, glucose uptake reaches its zenith in the sensorimotor cortex, thalamus, mid-brain, brainstem and cerebellar vermis (Chugani, 1998; Kinnala et al., 1996). Glucose uptake

expands to the temporal—parietal and frontal cortices by the end of the first year of life (Bentourkia et al., 1998) and reaches maximum values between ages 4 and 12 years (Bentourkia et al., 1998). It is apparent that regional glucose utilization undergoes significant post-natal changes that correlate with the functional demands of the developing organism. In our present study, we did not demonstrate any correlation between the pattern or severity of glucose hypometabolism and patient age at the time of PET imaging. This observation is congruent with the clinical fact that Glut1DS is a non-progressive encephalopathy. Previously, we compared the PET imaging findings in a young patient with Glut1 DS to another patient diagnosed with infantile hypoglycemia secondary to nesidioblastosis. These two conditions result in hypoglycorrhachia. Surprisingly, 18 F-FDG-PET images demonstrated a similar distribution of glucose hypometabolism in both patients (Pascual et al., 2007). This anecdotal experience suggests that brain glucose hypometabolism, present in infancy, leaves an imprint that persists even when the precipitating condition causing the hypoglycorrhachia is corrected. We have speculated that hypoglycorrhachia during a vulnerable period in early infancy stunts the maturation of the developing thalamocortical network and synaptic connections, leaving an indelible mark on selective cerebral circuits and networks. Application of quantitative analysis of metabolic data is the strength of this study. However, the limited number of 18 F-FDG-PET images from patients and control participants potentially could impact the statistical power of the analysis. Moreover, age matched comparisons between the healthy participants and patients with Glut1 DS were not performed. This is a continuing limitation for any imaging study using PET methodology because of the understandable concerns regarding unnecessary exposure to radiopharmaceuticals in the pediatric age group. We attempted to perform PET imaging during the interictal state; however, we were unable to obtain EEG recordings at the time of the imaging. Therefore, the interictal state was assessed only by clinical inspection as to whether any signs of seizures or dystonia were present 30 min prior to or after the administration of the radiotracer. Furthermore, there was no evidence of increased brain regional glucose uptake in any study, an expected finding if the PET studies were performed during an ictus. These findings support the clinical

214 contention that the PET images were obtained during the interictal state. In this cohort, majority of the patients presented with three or more cardinal features. There were only 2 individuals, one without any cardinal features and the other with two cardinal features representing milder phenotype. One can speculate whether or not lack of seizure history or the absence of other cardinal features plays a role to explain with milder pattern of hypometabolism in Glut1 DS.

Conclusion Brain glucose metabolism of Glut1 DS patients is compromised globally, and is particularly pronounced in the thalamocortical network and cerebellum. Distinctive brain regional abnormalities, as visualized by PET imaging, are uniquely correlated with the clinical diagnosis of Glut1 Deficiency. The regional differences and the glucose hypometabolic topography permit us to map the affected brain network and anticipate the pathophysiology of epileptogenesis in this syndrome; namely, infantile-onset seizures. The pattern of glucose hypometabolism, and the lack of correlation with increasing age, also implies that the functional imprint of this condition is established early in life and remains relatively immutable throughout later development.

Acknowledgements This work was supported in part by the Colleen Giblin Charitable Foundation, Will Foundation, Milestones for Children and the United States Public Health Service [NS37949-01; RR00645 to DCD]. We remain indebted to the patients and their families for their continuing involvement and support of these investigations. Authors also thank Dr. Orhan Akman for his help in creating Fig. 2.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.eplepsyres.2014.11.007.

References Akman, C.I., Engelstad, K., Hinton, V.J., Ullner, P., Koenigsberger, D., Leary, L., et al., 2010. Acute hyperglycemia produces transient improvement in glucose transporter type 1 deficiency. Ann. Neurol. 67, 31—40. Banerjee, P.K., Snead 3rd., O.C., 1994. Thalamic mediodorsal and intralaminar nuclear lesions disrupt the generation of experimentally induced generalized absence-like seizures in rats. Epilepsy Res. 17, 193—205. Bentourkia, M., Michel, C., Ferriere, G., Bol, A., Coppens, A., Sibomana, M., et al., 1998. Evolution of brain glucose metabolism with age in epileptic infants, children and adolescents. Brain Dev. 20, 524—529. Chan, C.H., Briellmann, R.S., Pell, G.S., Scheffer, I.E., Abbott, D.F., Jackson, G.D., 2006. Thalamic atrophy in childhood absence epilepsy. Epilepsia 47, 399—405.

C.I. Akman et al. Choeiri, C., Staines, W., Messier, C., 2002. Immunohistochemical localization and quantification of glucose transporters in the mouse brain. Neuroscience 111, 19—34. Chugani, H.T., 1998. A critical period of brain development: studies of cerebral glucose utilization with PET. Prev. Med. 27, 184—188. Coenen, A.M., Drinkenburg, W.H., Inoue, M., van Luijtelaar, E.L., 1992. Genetic models of absence epilepsy, with emphasis on the WAG/Rij strain of rats. Epilepsy Res. 12, 75—86. Coulter, D.A., Huguenard, J.R., Prince, D.A., 1990. Differential effects of petit mal anticonvulsants and convulsants on thalamic neurones: GABA current blockade. Br. J. Pharmacol. 100, 807—813. Crunelli, V., Leresche, N., 2002. Childhood absence epilepsy: genes, channels, neurons and networks. Nat. Rev. Neurosci. 3, 371—382. Cummings, T.J., Chugani, D.C., Chugani, H.T., 1995. Positron emission tomography in pediatric epilepsy. Neurosurg. Clin. N. Am. 6, 465—472. de Curtis, M., Avanzini, G., 1994. Thalamic regulation of epileptic spike and wave discharges. Funct. Neurol. 9, 307—326. De Vivo, D.C., Leary, L., Wang, D., 2002. Glucose transporter 1 deficiency syndrome and other glycolytic defects. J. Child Neurol. 17 (Suppl. 3), 3S15—3S23; discussion 3S24—3S25. De Vivo, D.C., Trifiletti, R.R., Jacobson, R.I., Ronen, G.M., Behmand, R.A., Harik, S.I., 1991. Defective glucose transport across the blood-brain barrier as a cause of persistent hypoglycorrhachia, seizures, and developmental delay. N. Engl. J. Med. 325, 703—709. Dobrogowska, D.H., Vorbrodt, A.W., 1999. Quantitative immunocytochemical study of blood—brain barrier glucose transporter (GLUT-1) in four regions of mouse brain. J. Histochem. Cytochem. 47, 1021—1030. Duncan, J.D., Moss, S.D., Bandy, D.J., Manwaring, K., Kaplan, A.M., Reiman, E.M., et al., 1997. Use of positron emission tomography for presurgical localization of eloquent brain areas in children with seizures. Pediatr. Neurosurg. 26, 144—156. Engel Jr., J., 1984. The use of positron emission tomographic scanning in epilepsy. Ann. Neurol. 15 (Suppl.), S180—S191. Ichise, M., Vines, D.C., Gura, T., Anderson, G.M., Suomi, S.J., Higley, J.D., et al., 2006. Effects of early life stress on [11C]DASB positron emission tomography imaging of serotonin transporters in adolescent peer- and mother-reared rhesus monkeys. J. Neurosci. 26, 4638—4643. Kaufmann, P., Shungu, D.C., Sano, M.C., Jhung, S., Engelstad, K., Mitsis, E., Mao, X., Shanske, S., Hirano, M., DiMauro, S., De Vivo, D.C., 2004. Neurol. 62 (8), 1297—1302. Kinnala, A., Suhonen-Polvi, H., Aarimaa, T., Kero, P., Korvenranta, H., Ruotsalainen, U., et al., 1996. Cerebral metabolic rate for glucose during the first six months of life: an FDG positron emission tomography study. Arch. Dis. Child. Fetal Neonatal Ed. 74, F153—F157. Labate, A., Briellmann, R.S., Abbott, D.F., Waites, A.B., Jackson, G.D., 2005. Typical childhood absence seizures are associated with thalamic activation. Epileptic Disord. 7, 373—377. Leary, L.D., Wang, D., Nordli Jr., D.R., Engelstad, K., De Vivo, D.C., 2003. Seizure characterization and electroencephalographic features in Glut-1 deficiency syndrome. Epilepsia 44, 701—707. Liu, Z., Vergnes, M., Depaulis, A., Marescaux, C., 1991. Evidence for a critical role of GABAergic transmission within the thalamus in the genesis and control of absence seizures in the rat. Brain Res. 545, 1—7. Liu, Z., Vergnes, M., Depaulis, A., Marescaux, C., 1992. Involvement of intrathalamic GABAB neurotransmission in the control of absence seizures in the rat. Neuroscience 48, 87—93. McCormick, D.A., Contreras, D., 2001. On the cellular and network bases of epileptic seizures. Annu. Rev. Physiol. 63, 815—846.

Epilepsy and Regional glucose hypometabolism in Glut1 DS McMurtray, A.M., Licht, E., Yeo, T., Krisztal, E., Saul, R.E., Mendez, M.F., 2008. Positron emission tomography facilitates diagnosis of early-onset Alzheimer’s disease. Eur. Neurol. 59, 31—37. Muzik, O., Chugani, D.C., Juhasz, C., Shen, C., Chugani, H.T., 2000. Statistical parametric mapping: assessment of application in children. Neuroimage 12, 538—549. Neglig, A., De Vasconcelos, A.P., Boyet, S., 1988. Quantitative autoradiographic measurement of local cerebral glucose utilization in freely moving rats during postnatal development. J. Neurosci. 8, 2321—2333. Niedermeyer, E., 1996. Primary (idiopathic) generalized epilepsy and underlying mechanisms. Clin. Electroencephalogr. 27, 1—21. Pascual, J.M., Van Heertum, R.L., Wang, D., Engelstad, K., De Vivo, D.C., 2002. Imaging the metabolic footprint of Glut1 deficiency on the brain. Ann. Neurol. 52, 458—464. Pascual, J.M., Wang, D., Hinton, V., Engelstad, K., Saxena, C.M., Van Heertum, R.L., et al., 2007. Brain glucose supply and the syndrome of infantile neuroglycopenia. Arch. Neurol. 64, 507—513. Pearson, T.S., Akman, C., Hinton, V.J., Engelstad, K., De Vivo, D.C., 2013. Phenotypic spectrum of glucose transporter type 1 deficiency syndrome (Glut1 DS). Curr. Neurol. Neurosci. Rep. 13, 342. Pong, A.W., Geary, B.R., Engelstad, K.M., Natarajan, A., Yang, H., De Vivo, D.C., 2012. Glucose transporter type I deficiency syndrome: epilepsy phenotypes and outcomes. Epilepsia 53, 1503—1510. Richards, D.A., Lemos, T., Whitton, P.S., Bowery, N.G., 1995. Extracellular GABA in the ventrolateral thalamus of rats exhibiting spontaneous absence epilepsy: a microdialysis study. J. Neurochem. 65, 1674—1680. Rintahaka, P.J., Chugani, H.T., Messa, C., Phelps, M.E., 1993. Hemimegalencephaly: evaluation with positron emission tomography. Pediatr. Neurol. 9, 21—28. Salek-Haddadi, A., Lemieux, L., Merschhemke, M., Friston, K.J., Duncan, J.S., Fish, D.R., 2003. Functional magnetic

215 resonance imaging of human absence seizures. Ann. Neurol. 53, 663—667. Schapiro, M.B., Haxby, J.V., Grady, C.L., 1992. Nature of mental retardation and dementia in Down syndrome: study with PET, CT, and neuropsychology. Neurobiol. Aging 13, 723—734. Shouse, M.N., Stroh, P.J., Vreeken, T., 1988. Temporal lobe and petit mal antiepileptics differentially affect ventral lateral thalamic and motor cortex excitability patterns. Brain Res. 473, 372—379. Snead 3rd., O.C., 1995. Basic mechanisms of generalized absence seizures. Ann. Neurol. 37, 14657. Snead 3rd, O.C., Hechler, V., Vergnes, M., Marescaux, C., Maitre, M., 1990. Increased gamma hydroxybutyric acid receptors in thalamus of a genetic animal model of petit mal epilepsy. Epilepsy Res. 7, 121—128. Swartz, B.E., Thomas, K., Simpkins, F., Kovalik, E., Mandelkern, M.M., 1999. Rapid quantitative analysis of individual (18) FDGPET scans. Clin. Positron Imaging 2, 47—56. Talairach, J., Tournoux, P., 1988. Co-planar Stereotaxic Atlas of the Human Brain. Thieme, Stuttgart. Tucker, D.M., Brown, M., Luu, P., Holmes, M.D., 2007. Discharges in ventromedial frontal cortex during absence spells. Epilepsy Behav. 11, 546—557. von Krosigk, M., Bal, T., McCormick, D.A., 1993. Cellular mechanisms of a synchronized oscillation in the thalamus. Science 261, 361—364. von Moers, A., Brockmann, K., Wang, D., Korenke, C.G., Huppke, P., De Vivo, D.C., et al., 2002. EEG features of glut-1 deficiency syndrome. Epilepsia 43, 941—945. Wallenstein, G.V., 1994. The role of thalamic IGABAB in generating spike—wave discharges during petit mal seizures. Neuroreport 5, 1409—1412. Zeller, K., Rahner-Welsch, S., Kuschinsky, W., 1997. Distribution of Glut1 glucose transporters in different brain structures compared to glucose utilization and capillary density of adult rat brains. J. Cereb. Blood Flow Metab. 17, 204—209.