Epilepsia, 45(3):223–229, 2004

Blackwell Publishing, Inc.


C 2004 International League Against Epilepsy

Areas of Interictal Spiking Are Associated with Metabolic

Dysfunction in MRI-negative Temporal Lobe Epilepsy

∗ Jerry J. Shih, †Michael P. Weisend, ‡Jeffrey Lewine, §John Sanders, Jamie Dermon,
and ¶Roland Lee
∗ Departments of Neurology & Neurosciences, University of New Mexico School of Medicine, and †Department of Radiology,
Veterans Administration Medical Center Albuquerque, New Mexico; ‡Hoglund Brain Imaging Center, Kansas City, Kansas;
§Radiology Imaging Associates, Denver, Colorado; //Stanford University, Stanford, California; and ¶ Department of Radiology,
University of New Mexico Health Sciences Center, Albuquerque, New Mexico, U.S.A.

Summary: Purpose: The objective of our study was to determine Results: Fifteen of 20 subjects had a lower NAA/Cho ratio in
noninvasively whether metabolic dysfunction is present in focal the MEG spike zone compared with the contralateral homolo-
areas of interictal electrophysiologic abnormality and whether gous region. NAA/Cho was significantly decreased in the MEG
metabolic dysfunction correlates with frequency of spiking. spike zone (p < 0.01). NAA/Cho ratios were not significantly
Methods: We used a prospective, power analysis–driven, age- different between the hippocampus ipsilateral and contralateral
matched design to study 20 subjects with nonlesional tempo- to the spike activity, or from control hippocampi. NAA/Cho ra-
ral lobe epilepsy by using magnetoencephalography (MEG) and tios did not correlate with spike frequency.
proton magnetic resonance spectroscopy (1 H-MRS). MEG was Conclusions: Metabolic dysfunction is present in focal areas
used to localize the source area of interictal spikes. 1 H-MRS mea- of interictal spiking in nonlesional temporal lobe epilepsy. These
sured integrated peak areas for N-acetyl compounds (NAA) and findings confirm that functional abnormalities can be detected
choline-containing compounds (Cho) in both hippocampi, the in vivo in radiographically normal-appearing cortex exhibiting
MEG spike zone, and the region contralateral to the MEG spike abnormal excitability. Key Words: Interictal spikes—MEG—
zone in all subjects. 1 H-MRS was performed in seven controls. MRS—Temporal lobe epilepsy.

In localization-related epilepsy, the hallmark of an area (7,8), thus making MEG one of the most accurate in vivo
of abnormal increased excitability in the cortex is the inter- localization tools for localizing electrophysiologic abnor-
ictal spike discharge often seen in EEG. Cortical irritabil- malities such as interictal spike discharges. The areas of
ity is postulated to be an imbalance between excitatory electrophysiologic abnormality as determined by MEG
and inhibitory forces due to underlying perturbations at correlate with areas of epileptogenicity as determined by
the molecular, cellular, and neuronal circuitry level (1). electrocorticography and epilepsy surgery outcome (6,9).
Although metabolic dysfunction exists in epileptogenic Proton magnetic resonance spectroscopy (1 H-MRS) is
tissue (2,3), it is unclear whether focal areas of cortical a noninvasive method to detect certain brain metabo-
irritability exhibit metabolic dysfunction. lites in vivo. The major metabolites of the brain exam-
Magnetoencephalography (MEG) is a noninvasive ined by 1 H-MRS include N-acetyl compounds (NAA),
imaging modality that provides characterization of brain choline-containing compounds (Cho), and creatine- and
electrophysiology by measuring the magnetic field gen- phosphocreatine-containing compounds (Cr). Several
erated by intracellular neuronal currents in the brain. In- studies provide evidence that NAA is a specific neuron
terictal MEG dipole source-modeling studies have shown marker not found within mature glial cells, and thus a re-
that MEG can accurately localize sources of epileptiform duction in the NAA signal or in its ratio to other metabo-
discharges (4–6). MEG is commonly believed to local- lite signals is thought to indicate neuronal loss (10,11).
ize an interictal discharge to within 3–8 mm of its source Metabolic dysfunction also may cause a reversible de-
crease in NAA signal or its ratio to other metabolite
Accepted October 21, 2003. signals, and has been documented in a variety of disor-
Address correspondence and reprint requests to Dr. J. J. Shih at Uni- ders such as demyelinating diseases (12), amyotrophic
versity of New Mexico School of Medicine, Department of Neurology,
MSC10 5620, 1 University of New Mexico, Albuquerque, NM 87131- lateral sclerosis (13), mitochondrial encephalopathy (14),
0001, U.S.A. E-mail: jshih@salud.unm.edu and epilepsy after successful surgical intervention (15,16).

223
224 J. J. SHIH ET AL.

A decrease in NAA or its ratio to other metabolites is useful Clinical evaluation

in lateralizing the side of temporal lobe epilepsy (2,3,17). Medical and neurologic history and physical examina-
Garcia et al. (18) also found the degree of metabolic abnor- tion were performed for all subjects. Seizure-onset age,
mality correlated with the frequency of seizure activity. aura type, frequency of seizures, number of medications
The primary objective of our study was to determine if tried, and length of epilepsy were determined. Results of
metabolic dysfunction is present in focal areas of interictal routine, surface video-EEG and intracranial video-EEG
spiking. We also assessed whether metabolic dysfunction were recorded (Table 1).
correlated with spike frequency or length of epilepsy.
Magnetoencephalography
We documented no seizures for ≥16 hours before the
SUBJECTS AND METHODS MEG study. MEG was performed by using a 122-channel
Subjects and controls whole-head biomagnetometer (Neuromag Ltd., Helsinki,
We used a prospective, power analysis–driven, age- Finland). This system consists of a head-shaped array of 61
matched design to study 20 consecutive patients (10 men pairs of orthogonal planar gradiometers, each gradiome-
and 10 women); age range, 16 to 56 years (mean age, 30 ter coupled to a superconducting quantum interference
years) with medically refractory temporal lobe epilepsy device (SQUID). EEG was always recorded simultaneous
who were being evaluated for possible epilepsy surgery. with MEG by using a custom electrode cap with 20 MEG-
All were diagnosed as having nonlesional temporal lobe compatible contacts. Attached to the EEG cap were three
epilepsy based on magnetic resonance imaging (MRI) energizable coils, the positions of which were determined
scans, on clinical semiology indicative of complex par- relative to a head-centered coordinate frame (defined by
tial seizures of temporal lobe onset, and on ictal EEG the nasion and right and left preauricular points) by using
showing onset in the temporal lobe. The control group a 3D-digitizer. Each subject was studied in the biomagne-
was composed of five men and two women (age range, 21 tometer for 90 min, with the position of the sensor coils
to 41 years; mean age, 31 years). All subjects and controls relative to a head-centered coordinate frame checked ev-
signed an Institutional Review Board consent form. ery 10 min through triangulation of signals generated by

TABLE 1. Subject clinical history, course and electrophysiologic results

Years #meds Sz freq
Subject Age w/ szs Meds tried per month Aura EEG V-EEG Intracranial EEG Course

1 27 26 DPH,CBZ 4 8 Epigastric discomfort BT Not done Not offered Experimental AED,

1 sz/2 months
2 17 4 CBZ/VPA/TPM 4 4 Nausea BT LT onset Declined Lost to f/u
3 20 12 CBZ 3 10 Fear BT RT onset (R) mesial temp (R) ATL, Engel II
4 15 9 CBZ/LMT 4 3 None BT LT onset Not offered AEDs
5 33 28 CBZ 3 4 Dizziness LT LT onset No szs over 10 days Awaiting repeat
Phase 2
6 25 20 DPH/NT 5 2 Epigastric rising BT RT onset Depths nonlocalizing AEDs
7 29 25 CBZ/DPH 4 4 None BT RH onset (R) mesial temp (R) ATL, Engel I
8 38 16 LMT 5 3 Fear, sulphur smell LT LT onset Declined AEDs
9 30 3 DPH/VPA 3 30 Dizziness/disorientation RT RT onset Declined Refused further w/u,
on AEDs
10 56 1 CBZ 1 0.25 None RT Not done Not offered 1 sz/yr on CBZ after
optimization
11 20 16 CBZ/NT 4 12 Blowing sensation LT LH onset (L) post temp onset MSPT (L) post
temp-improved
12 37 33 CBZ 6 1 Déjà vu, disorientation LT No szs Nonlocalizing AEDs
13 16 5 DPH/LMT 6 6 Heat sensation (L) body RT RT onset (R) mesial temp (R) ATL, Engel I
14 37 17 DPH 5 4 None BT LT onset (L) mesial temp (L) ATL, Engel Ic
15 31 15 PB 6 2 Strange taste/smell LT Not done Not offered 1 sz/yr on PB/LT
combo
16 41 30 DPH/PRM 7 4 Depersonalized feeling LT LH onset (L) mesial temp (L) ATL, Engel I
17 49 20 CBZ/PB 6 8 Musicogenic BT LH onset (L) mesial temp (L) ATL, Engel III
18 18 2 CBZ 5 8 None BT LT onset (B) temp onsets AEDs
19 43 25 DPH/LMT/KLP 5 2 Strange sensation in head BT Nonloc Declined AEDs
20 21 15 CBZ/VPA 2 20 Nausea RT Nonloc Pending Undergoing presurg
eval

DPH, phenytoin; CBZ, carbamazepine; VPA, valproic acid; TPM, topiramate; LMT, lamotrigine; NT, gabapentin; PB, phenobarbital; PRM, primidone;
KLP, clonazepam; BT, bitemporal spikes; RT, right temporal spikes; LT, left temporal spikes; RH, right hemisphere; LH, left hemisphere; AED,
antiepileptic drug; ATL, anterior temporal lobectomy; MSPT, multiple subpial transection; Engel, Engel classification for surgical outcome; Nonloc,
nonlocalizing.

Epilepsia, Vol. 45, No. 3, 2004

 SPIKING AND METABOLIC DYSFUNCTION IN TLE 225

FIG. 1. Three plane view of hippocampal voxel placement in MRS in subjects and controls.

three coils on the electrode cap. Continuous MEG and field of view (FOV), 256 mm; slice thickness, 1.5 mm)
EEG data were collected at a digitization rate of 300 Hz was acquired and reformatted to give an orthogonal set
and stored for off-line analysis. Off-line analyses began of scout images. This data set was used by the MEG
with digital high-pass (6 Hz) filtering of the data to elim- system for mapping the source locations directly onto
inate low-frequency drift and contaminating slow waves. the scout images. This mapping is done by using three
The entire recording was then visually examined for in- fiducial landmarks (nasion and right and left preauricu-
terictal epileptiform spikes. When an interictal spike was lar points) that are landmarked by the MEG system and
identified, multiple-dipole, spatiotemporal modeling pro- are identified within the volumetric image data. Spectra
cedures were used to define the location, orientation, and from four VOIs were acquired during the examination.
time course for that set of dipole currents that best de- Placements in the left and right hippocampi were done
scribed (in a least-squares statistic sense) the recorded according to the anatomy displayed in three orthogonal
magnetic signal. A spherical volume-conductor model images (Fig. 1). The third VOI was positioned to enclose
was used in all calculations. In cases in which multiple the most spatially clustered MEG source locations, and
dipoles with overlapping activation signals were needed an additional VOI was positioned directly contralateral
to account for a spike, the dipole source with earliest acti- within a similar tissue region. A PRESS (21) sequence
vation was taken to represent the spike trigger zone (19). (TR, 1,500/TE, 135) was preceded by a four-pulse water-
suppression radiofrequency and spoiler gradient train. One
Magnetic resonance imaging hundred twenty-eight averages were obtained of the water-
Magnetic resonance imaging of the brain was per- suppressed signal, and 16 averages were obtained of a
formed in all subjects and controls with a 1.5-T Picker non–water-suppressed signal. The unsuppressed spectra
Edge MRI System (Cleveland, OH, U.S.A.). The imag- were used to phase-correct the suppressed spectra for
ing sequence incorporated tilted coronal 3-D fast-spin system eddy currents. The original 2,048 data samples
echo T2 -weighted images (repetition time (TR)/echo time were zero-padded to 16,384, and a 3-Hz line broadening
(TE), 3,500/133; slice thickness, 1.5 mm; zero skip), was applied. Residual water signal was removed by high-
tilted coronal 3-D gradient-echo T1 -weighted images pass filtering with gaussian smoothing with a window of
(TR/TE, 15/4.4; flip angle, 25 degrees; slice thickness, 32 ms.
1.0 mm; zero skip), axial T2 -weighted images (TR/TE, For the quantitation needs of this study, magnitude spec-
2,674/30/90); slice thickness, 5.0 mm; 1.0-mm skip), tral results were used so as to avoid difficulties (primarily
sagittal T1 -weighted images (TR, 450/8.0; slice thickness, software) in accurately phase- and baseline-correcting the
4.0 mm; 1.0-mm skip) and sagittal 3-D gradient-echo T1 - data. The shim quality and signal-to-noise ratio obtained
weighted images (TR/TE, 15/4.4; flip angle, 25 degrees; were generally very good. By using the NAA resonance
slice thickness, 1.5 mm; zero skip). All scans were inter- (2.02 ppm) as a shift standard, the primary and secondary
preted by a board-certified neuroradiologist. A diagnosis spectral peaks were numerically fit by using a Levenberg-
of normal MRI was based on clinically accepted parame- Marquardt algorithm. A model peak shape consisting of
ters for visual analysis (20). 5% lorentzian and 95% gaussian was used. These peaks
were manually modeled to provide starting values used
1
H-MRS examinations in the numeric optimization. Integrated peak areas are re-
The single-voxel proton MR spectroscopy was per- ported for NAA and Cho.
formed on a 1.5-T Picker Edge MRI system obtaining
measurements from a 15 × 15 × 15 mm3 volume of in- Statistical analysis
terest (VOI). A whole-head volumetric 3D FLASH T1 - The power analysis is derived from the data of Connelly
weighted data set (TR, 15/TE, 4.4; flip angle, 25 degrees; et al. (3). Metabolite ratios between MEG spike zone and

Epilepsia, Vol. 45, No. 3, 2004

226 J. J. SHIH ET AL.

FIG. 2. Individual epileptiform discharges from Subject 4 detected by MEG are displayed as circles in left lateral and mid-temporal
lobe.The 4 squares represent placement of voxels corresponding to the hippocampi,the epileptic spike region and the region homologous
to the spike region.

the contralateral homologous region, ipsilateral and con- RESULTS

tralateral hippocampi in epilepsy subjects, and right and
left hippocampus in controls were analyzed with a paired Metabolite ratios in the MEG spike zone
t test based on ratio (M1 /M2 -1) compared with zero. The Fifteen (75%) of 20 subjects had a lower NAA/Cho ra-
NAA/Cho ratios of subjects with infrequent MEG spiking tio in the voxel containing the centroid of MEG spiking
(defined as ≤10 spikes) and subjects with frequent spik- compared with the voxel from the homologous contralat-
ing (defined as >10 spikes) were compared by using an eral region (Table 2, Fig. 2, Fig. 3). NAA/Cho ratios were
independent-samples t test. NAA/Cho ratios were corre- significantly lower in regions of interictal spiking com-
lated with monthly seizure frequency by using a Pearson pared with the contralateral homologous region (1.37 ±
correlation. 0.39:1.71 ± 0.42; p < 0.01). NAA/Cho ratios in regions of

FIG. 3. Individual epileptiform discharges from Subject 20 detected by MEG are displayed as circles in right posterior lateral and superior
temporal lobe.The 4 squares represent placement of voxels corresponding to the hippocampi, the epileptic spike region, and the region
homologous to the spike region.

Epilepsia, Vol. 45, No. 3, 2004

 SPIKING AND METABOLIC DYSFUNCTION IN TLE 227

TABLE 2. Subject MEG and MRS results

MEG NAA/Cho NAA/Cho NAA/Cho NAA/Cho
Subject Spike # MEG Spike locations MRS Voxel Location MEG Contra MEG Ipsi Hippo Contra Hippo

1 >10 (R) mid-lat temp cluster, scattered (L) temp (R) mid-lat temp 1.14 1.76 0.87 0.96
2 >10 (L) ant-mesial temp cluster, scattered (B) frontal (L) ant-mesial temp 0.78 1.48 0.46 1.83
3 >10 (R) ant-mesial temp cluster, rare (L) temp (R) ant-mesial temp 1.48 1.61 0.80 1.13
4 >10 (L) mid-lat temp > (R) temp (L) mid-lat temp 1.03 1.29 1.03 1.05
5 7 (L) ant-mesial temp (L) ant-mesial temp 1.24 2.13 1.57 1.29
6 >10 (L) ant-mesial >(R) ant mesial (L) ant-mesial temp 0.86 2.02 1.12 1.35
7 >10 (R) mid-postlat temp, rare (L) sup-post temp (R) mid-postlat temp 1.30 1.50 1.05 0.94
8 8 (L) ant-mesial temp (L) ant-mesial temp 1.74 1.27 0.95 1.27
9 >10 (R) ant-mesial temp (R) ant-mesial temp 1.82 1.38 1.25 1.07
10 7 (R) mid-lat temp (R) mid-lat temp 1.88 2.68 1.04 1.55
11 >10 (L) sup-post temp cluster (L) sup-post temp 1.81 1.66 1.12 0.86
12 >10 (L) post-lat temp (L) post-lat temp 2.14 2.60 1.27 1.32
13 >10 (R) ant-mesial temp cluster (R) ant-mesial temp 1.22 1.98 1.06 1.32
14 >10 (L) ant-mesial temp cluster >(R) ant mesial cluster (L) ant-mesial temp 1.42 1.77 1.21 1.51
15 8 (L) ant-mesial temp cluster (L) ant-mesial temp 1.21 1.36 Not done Not done
16 >10 (L) med-post temp cluster (L) med-post temp 1.32 1.06 1.52 1.09
17 10 (L) post-sup temp (L) post-sup temp 1.84 1.57 1.32 1.68
18 >10 (R) mid-lat temp >>> (L) ant-mesial temp (R) mid-lat temp 1.32 2.05 1.54 0.89
19 >10 (L) mesial temp > (R) mesial temp (L) mesial temp 1.07 1.46 1.20 1.35
20 >10 (R) post-lat temp spreads to (R) frontal (R) post-lat temp 0.78 1.46 1.31 1.21

MEG, magnetoencephalography-identified spike zone(s); MRS, magnetic resonance spectroscopy; contra MEG, homologous region in contralateral
temporal lobe to MEG spike zone; ipsi hippo, hippocampal voxel ipsilateral to MEG spike zone; contra hippo, hippocampal voxel contralateral to
MEG spike zone; NAA/Cho, N-acetyl compounds/choline-containing compounds ratio.

interictal spiking did not differ significantly between sub- Metabolite ratios in control hippocampus
jects with infrequent spiking on MEG (n = 5; NAA/Cho, NAA/Cho was measured in the hippocampi of seven
1.58 ± 0.33) versus those with a higher frequency of spik- nonepileptic controls. NAA/Cho in the right hippocampus
ing (n = 15; NAA/Cho, 1.30 ± 0.39). NAA/Cho ratios did was 1.24 ± 0.15. NAA/Cho in the left hippocampus was
not correlate with seizure frequency. Except for two sub- 1.21 ± 0.17.
jects with 20 and 30 seizures per month, respectively, most
patients’ seizure frequency was within a narrow range of MRS and surgical outcome
two to eight seizures per month. The area of predominant Six subjects underwent anterior temporal lobectomy to
spike activity was located in the anteriomesial aspect of the remove the epileptogenic region. One subject underwent
temporal lobe in 10 subjects, and in the posterior or lateral multiple subpial transection in a left posterior temporal
aspect of the temporal lobe in 10 others. The MRS vox- lobe region determined by electrocorticography to sub-
els encompassing spikes in the anteromesial temporal re- serve language. All four resective surgery subjects with
gion overlapped but were not identical to the hippocampal lower NAA/Cho in the MEG spike zone had Engel I/II out-
voxels. NAA/Cho ratios were similar in the anteromesial comes at 1 year. One of two resective surgery subjects with
voxels (1.28 ± 0.34) compared with those in the postero- disconcordant MEG/MRS data (NAA/Cho lower in the
lateral region (1.45 ± 0.44). The contralateral temporal temporal lobe contralateral to predominant MEG spiking)
lobe also showed similar NAA/Cho in the anteromesial had an Engel I/II outcome. Three subjects had predomi-
region (1.65 ± 0.31) compared with the posterolateral re- nant MEG spiking in the anteromesial temporal region (all
gion (1.79 ± 0.52). with Engel I/II outcome); three subjects had predominant
MEG spiking in the lateral or posterior temporal lobe (two
Hippocampal metabolite ratios in patient subjects of three with Engel I/II outcome).
Nineteen of 20 subjects underwent single-voxel exam-
ination of the hippocampi. In one subject, we were able to
DISCUSSION
obtain data only from the spike zone and the contralateral
homologous region before the subject terminated testing. Our results show that NAA/Cho is decreased in the
Twelve subjects had lower hippocampal NAA/Cho in the area of electrophysiologic spike activity in temporal lobe
side ipsilateral to predominant spiking, but no difference epilepsy patients with a normal brain MRI. The metabolite
was noted between ipsilateral and contralateral NAA/Cho ratios did not differ between hippocampi of epilepsy sub-
ratios (1.14 ± 0.27:1.24 ± 0.27), or compared with control jects, despite 12 of 19 having lower NAA/Cho in the hip-
hippocampi. pocampus ipsilateral to predominant electrophysiologic

Epilepsia, Vol. 45, No. 3, 2004

228 J. J. SHIH ET AL.

spike activity. If we applied a different statistical approach structures in both epilepsy patients and controls have been
(3), and used a cut-off of 95% confidence interval based shown with MRSI (29). Our anteromesial voxels often
on our control data, four subjects (20%) would have an encompassed hippocampal structures, whereas the pos-
abnormally low NAA/Cho ratio in the ipsilateral hip- terolateral voxels sampled mainly neocortex. The lower
pocampus. Our results are similar to data from Woermann NAA/Cho values in the anteromesial voxels may be ex-
et al. (22) showing 27% of MRI-negative temporal lobe plained by reduced neuronal densities in the hippocam-
epilepsy patients have abnormally low NAA in the ipsi- pal and parahippocampal gyrus when compared with
lateral hippocampus. neocortex (30).
We chose MEG to localize spike activity because of The decreased NAA/Cho in the area of spike activity
its superior spatiotemporal resolution. Phantom and sim- most likely does not represent neuronal loss. We did not
ulation studies demonstrated the ability of MEG to lo- obtain cell counts from areas of spiking and thus cannot
calize dipole sources accurately (7,8). Human intracranial definitively rule out neuronal loss. However, no structural
EEG studies in adults (6,23) and children (24) confirm abnormalities are seen on MRI in nonlesional temporal
that MEG is able accurately to localize spike activity in lobe epilepsy, and the most common histopathologic ab-
temporal lobe epilepsy. For determination of NAA and normality in this population is hippocampal sclerosis (31).
Cho, either single-voxel or spectroscopic imaging would One would not expect our findings to be explained by neu-
be sufficient. In our study, we chose single-voxel MRS ronal loss in the hippocampus, because many of the MRS
because the typical advantages of spectroscopic imaging voxels that corresponded to the location of MEG spiking
were mitigated by two factors. We anticipated imaging in did not encompass hippocampal structures. We note that
the region of the anterior temporal lobe where spectro- others (32) have found decreased NAA concentrations in
scopic imaging is technically challenging for others and extrahippocampal tissue in patients with medial temporal
us (2). We also designed a study to evaluate voxels that lobe epilepsy without documenting extrahippocampal cell
were often not within a two-dimensional plane within the loss.
temporal lobe. Single-voxel MRS is appropriate for this We speculate the decreased NAA/Cho may be a result of
study design (25). oxidative stress in neurons exhibiting increased electrical
Our findings are in agreement with and, in some in- irritability and leading to dynamic changes in NAA. One
stances, extend the work of others. Serles et al. (26) found a study (33) demonstrated oxidative stress in areas of elec-
trend toward higher interictal EEG spike frequencies over- trophysiologic spiking by finding increased lactate levels
lying areas of metabolic abnormality or dysfunction. Our in the MRI-normal occipital cortex of patients with photo-
study also showed a trend toward higher NAA/Cho with sensitive epilepsy. Elevation of lactate in a localized tissue
lower spike numbers. Our original power analysis was de- belies a region of oxidative stress (34), and the level of
rived for the purpose of determining whether NAA/Cho is lactate increase correlated well with other markers of dis-
decreased in the spike zone. It is possible that having more ease severity. Another line of evidence consistent with the
subjects in the high/low-frequency spiking groups would hypothesis of oxidative stress causing decreased NAA is
demonstrate a difference between groups. The EEG spik- the finding of reversible NAA decrease in a case of mito-
ing from the Serles et al. study was determined from eight chondria encephalopathy (14). In this particular case with
electrode positions on the scalp and correlated with an paroxysmal neurologic deficits, the mechanism of NAA
assigned anatomic brain subregion from where metabo- recovery is related to resumption of normal mitochondrial
lite values were measured. Our study used 122 recording function and the resolution of oxidative stress. In addition
sensors and a localization algorithm superior to the EEG to oxidative stress, subtle structural changes such as a re-
International 10–20 System. Our finding that seizure fre- duction in dendritic branching (35) may lead to a reduction
quency did not correlate with NAA/Cho ratios is supported in NAA.
by an EEG/MRSI study (27) showing no significant dif- In summary, our study demonstrated that metabolic
ferences in seizure frequency or age at onset when com- dysfunction is present in focal areas of electrophysi-
paring temporal lobe epilepsy patients with or without ologic spike abnormality in nonlesional temporal lobe
evidence of metabolic dysfunction. However, other work- epilepsy. Whether the electrophysiologic abnormality and
ers (18,28) have found a correlation between increased metabolic dysfunction are causally related or epiphenom-
seizure frequency and evidence of metabolic dysfunction. ena of another process remains to be elucidated. The
We note that our subject group had a relatively narrow potential clinical implication of our findings is that this
range of seizure frequencies, which may have made it dif- method also may be valuable in the evaluation of extratem-
ficult to demonstrate a difference. Our study also showed poral epilepsy in which there is often no gross structural
lower NAA/Cho ratios in the anteromesial temporal lobe abnormality detectable on MRI. In these cases, lateral-
as compared with ratios from the posterolateral temporal ization can be aided by techniques such as MEG-guided
1
lobe, although the difference was not statistically differ- H-MRS that detect pathology below the threshold for
ent. Anteroposterior NAA differences from hippocampal MRI-detectable pathology.

Epilepsia, Vol. 45, No. 3, 2004

 SPIKING AND METABOLIC DYSFUNCTION IN TLE 229

Acknowledgment: This work was supported by NIH/NCRR 17. Kuzniecky R, Hugg J, Hetherington H, et al. Predictive value of
1 H MRSI for outcome in temporal lobectomy. Neurology
M01 RR00997-20S3 (J.J.S.), the UNM General Clinical Re-
search Center, and NIH/NCRR COBRE 5P20RR15636 (J.J.S., 1999;53:694–8.
M.W., R.R.L.). We thank John Davis, Ph.D., for technical MEG 18. Garcia PA, Laxer KD, van der Grond J, et al. Correlation of
analysis. We thank David Blum, M.D., and Bruce Enos, M.D., seizure frequency with N-acetyl-aspartate levels determined by 1 H
magnetic resonance spectroscopic imaging. Magn Reson Imaging
for referral of study patients. 1997;15:475–8.
19. Lewine JD, Morrell F, Orrison WW Jr. Neuromagnetic evaluation of
normal and abnormal brain function in epilepsy. Int J Neuroradiol
REFERENCES 1995;1:182–98.
20. Cascino G, Jack CR. Neuroimaging in epilepsy: principles and prac-
1. Engel J, Pedley TA. Epilepsy : a comprehensive textbook. Philadel- tice. Boston: Butterworth-Heinemann, 1996.
phia: Lippincott-Raven, 1998. 21. Bottomley PA. Spatial localization in NMR spectroscopy in vivo.
2. Cendes F, Andermann F, Preul MC, et al. Lateralization of tem- Ann N Y Acad Sci 1987;508:333–48.
poral lobe epilepsy based on regional metabolic abnormalities 22. Woermann FG, McLean MA, Bartlett PA, et al. Short echo time
in proton magnetic resonance spectroscopic images. Ann Neurol single-voxel 1 H magnetic resonance spectroscopy in magnetic res-
1994;35:211–6. onance imaging-negative temporal lobe epilepsy: different bio-
3. Connelly A, Van Paesschen W, Porter DA, et al. Proton magnetic chemical profile compared with hippocampal sclerosis. Ann Neurol
resonance spectroscopy in MRI-negative temporal lobe epilepsy. 1999;45:369–76.
Neurology 1998;51:61–6. 23. Wheless JW, Willmore LJ, Breier JI, et al. A comparison of mag-
4. Barth DS, Sutherling W, Engel J Jr, et al. Neuromagnetic local- netoencephalography, MRI, and V-EEG in patients evaluated for
ization of epileptiform spike activity in the human brain. Science epilepsy surgery. Epilepsia 1999;40:931–41.
1982;218:891–4. 24. Minassian BA, Otsubo H, Weiss S, et al. Magnetoencephalographic
5. Sutherling WW, Crandall PH, Cahan LD, et al. The magnetic field localization in pediatric epilepsy surgery: comparison with inva-
of epileptic spikes agrees with intracranial localizations in complex sive intracranial electroencephalography. Ann Neurol 1999;46:627–
partial epilepsy. Neurology 1988;38:778–86. 33.
6. Knowlton RC, Laxer KD, Aminoff MJ, et al. Magnetoencephalog- 25. Constantinidis I. MRS methodology. Adv Neurol 2000;83:235–
raphy in partial epilepsy: clinical yield and localization accuracy. 46.
Ann Neurol 1997;42:622–31. 26. Serles W, Li LM, Caramanos Z, et al. Relation of interictal spike fre-
7. Leahy RM, Mosher JC, Spencer ME, et al. A study of dipole local- quency to 1 H-MRSI-measured NAA/Cr. Epilepsia 1999;40:1821–7.
ization accuracy for MEG and EEG using a human skull phantom. 27. Li LM, Cendes F, Andermann F, et al. Spatial extent of neu-
Electroencephalogr Clin Neurophysiol 1998;107:159–73. ronal metabolic dysfunction measured by proton MR spectroscopic
8. Cohen D, Cuffin BN, Yunokuchi K, et al. MEG versus EEG local- imaging in patients with localization-related epilepsy. Epilepsia
ization test using implanted sources in the human brain. Ann Neurol 2000;41:666–74.
1990;28:811–7. 28. Namer IJ, Bolo NR, Sellal F, et al. Combined measurements of
9. Ko DY, Kufta C, Scaffidi D, et al. Source localization determined by hippocampal N-acetyl-aspartate and T2 relaxation time in the eval-
magnetoencephalography and electroencephalography in temporal uation of mesial temporal lobe epilepsy: correlation with clini-
lobe epilepsy: comparison with electrocorticography: technical case cal severity and memory performances. Epilepsia 1999;40:1424–
report. Neurosurgery 1998;42:414–21; discussion 421–2. 32.
10. Birken DL, Oldendorf WH. N-acetyl-L-aspartic acid: a literature 29. Vermathen P, Laxer KD, Matson GB, et al. Hippocampal struc-
review of a compound prominent in 1 H-NMR spectroscopic studies tures: anteroposterior N-acetylaspartate differences in patients with
of brain. Neurosci Biobehav Rev 1989;13:23–31. epilepsy and control subjects as shown with proton MR spectro-
11. Urenjak J, Williams SR, Gadian DG, et al. Specific expression scopic imaging. Radiology 2000;214:403–10.
of N-acetylaspartate in neurons, oligodendrocyte-type-2 astrocyte 30. Arnold SE, Trojanowski JQ. Human fetal hippocampal develop-
progenitors, and immature oligodendrocytes in vitro. J Neurochem ment, I: cytoarchitecture, myeloarchitecture, and neuronal morpho-
1992;59:55–61. logic features. J Comp Neurol 1996;367:274–92.
12. Davie CA, Hawkins CP, Barker GJ, et al. Serial proton magnetic 31. Tran TA, Spencer SS, Marks D, et al. Significance of spikes
resonance spectroscopy in acute multiple sclerosis lesions. Brain recorded on electrocorticography in nonlesional medial temporal
1994;117:49–58. lobe epilepsy. Ann Neurol 1995;38:763–70.
13. Kalra S, Cashman NR, Genge A, et al. Recovery of N-acetylaspartate 32. Vermathen P, Laxer KD, Schuff N, et al. Evidence of neuronal in-
in corticomotor neurons of patients with ALS after riluzole therapy. jury outside the medial temporal lobe in temporal lobe epilepsy:
Neuroreport 1998;9:1757–61. N-acetylaspartate reductions detected with multisection proton MR
14. Kamada K, Takeuchi F, Houkin K, et al. Reversible brain dysfunc- spectroscopic imaging. Radiology 2003;226:195–202.
tion in MELAS: MEG, and (1)H MRS analysis. J Neurol Neurosurg 33. Chiappa KH, Hill RA, Huang-Hellinger F, et al. Photosensitive
Psychiatry 2001;70:675–8. epilepsy studied by functional magnetic resonance imaging and
15. Cendes F, Andermann F, Dubeau F, et al. Normalization of magnetic resonance spectroscopy. Epilepsia 1999;40(suppl 4):3–7.
neuronal metabolic dysfunction after surgery for temporal lobe 34. Lin DDM, Crawford TO, Barker PB. Proton MR spectroscopy in
epilepsy: evidence from proton MR spectroscopic imaging. Neu- the diagnostic evaluation of suspected mitochondrial disease. AJNR
rology 1997;49:1525–33. Am J Neuroradiol 2003;24:33–41.
16. Serles W, Li LM, Antel SB, et al. Time course of postoperative 35. Multani P, Myers RH, Blume HW, et al. Neocortical dendritic
recovery of N-acetyl-aspartate in temporal lobe epilepsy. Epilepsia pathology in human partial epilepsy: a quantitative Golgi study.
2001;42:190–7. Epilepsia 1994;35:728–36.

Epilepsia, Vol. 45, No. 3, 2004