Gender-specific Differences of Hypometabolism in mTLE: Implication for Cognitive Impairments


Address correspondence and reprint requests to Dr. R.J. Seitz at University-Hospital Düsseldorf, Department of Neurology, Moorenstrasse 5, D-40225 Düsseldorf, Germany. E-mail:


Summary: Purpose: To determine gender differences of hypometabolism and their implications for cognitive impairment in patients with medically refractory mesial temporal lobe epilepsy (mTLE).

Methods: Regional cerebral glucose metabolism (rCMRGlu) was studied in 42 patients (21 male, 21 female) with either left- or right-sided mTLE (22 left, 20 right) and in 12 gender- and age-matched healthy controls during resting wakefulness and in 12 sex- and age-matched healthy controls. Clinical characteristics were balanced across the patient subgroups. All patients were subjected to neuropsychological assessment: 41 patients had histologic changes of definite or probable hippocampal sclerosis.

Results: Data analysis based on pixel-by-pixel comparisons and on a laterality index of regions of interest (ROIs) showed significant depressions of the mean rCMRGlu extending beyond the mesiotemporal region and temporolateral cortex to extratemporal regions including the frontoorbital and insular cortex in mTLE patients. Extramesiotemporal hypometabolism prevailed in the male patients. Metabolic asymmetry in temporal and frontal regions was related to performance in the Trail-Making Test and WAIS-R subitems.

Conclusions: Our data showed a gender-specific predominance of extramesiotemporal hypometabolism in male patients with mTLE related to abnormalities of temporal and frontal lobe functions.

Mesial temporal lobe epilepsy (mTLE) is the most common and best-defined syndrome of symptomatic localization-related epilepsy (1). Positron emission tomography (PET) using 18-fluoro-deoxyglucose (FDG-PET) was the first functional neuroimaging method allowing location of regions generating epileptiform EEG changes noninvasively (2) and has been established as a sensitive and reliable method for presurgical evaluation of TLE patients (3–5). FDG-PET has been used for lateralization of the seizure-onset zone, especially in those cases in which surface EEG and magnetic resonance imaging (MRI) provided discordant findings. However, FDG-PET revealed not only hypometabolism in the epileptogenic zone in mTLE, but also extratemporal metabolic disturbances (6–8).

These “remote depressions” can affect the ipsi- and contralateral neocortical part of the temporal lobe, the frontoorbital cortex, and the thalamus. These areas are supposed to be affected secondarily by ongoing subclinical seizure activity, inhibition, and deafferentation rather than by morphologic tissue damage (8,9). The degree of hypometabolism in these regions was shown to correlate well with neuropsychological impairments such as verbal memory function or cognitive abilities (6,7).

Apart from its use in the assessment of neuropsychiatric abnormalities, FDG-PET also has the capacity to provide physiological parameters of brain function such as gender-specific differences of cerebral metabolism in healthy volunteers (10–12). However, so far it remains unclear whether the metabolism in mTLE patients and thus brain function is affected differently in male and female subjects.

The goal of this study was therefore to investigate whether gender-specific differences exist in the magnitude and distribution of hypometabolism in mTLE patients and, if so, whether these findings are related to clinical and neuropsychological abnormalities.



Of >500 consecutive epilepsy patients who had been scanned with FDG-PET since 1991 in the PET center at the University Hospital Düsseldorf, 42 patients were selected (see Table 1) who fulfilled the following five criteria:

Table 1. Clinical data of the mTLE patients
 Left maleLeft femaleRight maleRight female
  1. Values are shown as mean (SD); no significant differences after Bonferroni correction for multiple comparisons. Outcome is according to the classification of Engel. Mean global CMRGlu of the control group was 31.1 (8.1) μmol/100 g/min.

  2. CPS, complex partial seizure; CPS with p./c.j., CPS with posturing/clonic jerks of an extremity or extremities; AEDs, antiepileptic drugs; CBZ, carbamazepine; PHT, phenytoin; VPA, valproate; VGB, vigabatrin; PRM, primidone; CLB, clobazam; LTG, lamotrigine; GBP, gabapentin; OXC, oxcarbazepine; PB, phenobarbital; LEV, levopropylhexedrin; [n], number of patients; CMRGlu, cerebral metabolism rate of glucose; HS, hippocampal sclerosis; glob. TLS, global temporal lobe sclerosis.

Patients (n)1111 10 10 
Age yr—(±SD)28.1 (4.9) 29.5 (10.0)29.0 (6.6) 31.7 (11.8)
Onset of TLE at yr—(±SD)8.9 (4.4)8.6 (6.1)9.9 (6.4)13.4 (10.3)
Duration of TLE yr—(±SD)19.2 (6.0) 20.8 (10.2)19.1 (5.6) 18.3 (12.3)
Seizures/mo n—(±SD)10.8 (10.2)15.1 (12.6)11.5 (22.4)5.5 (2.8)
Type of aura (n)
 Epigastric (sole or leading)  6487
 Somatosensory  21
 Psychic  2421
 Mixed (without epigastric)21
 No aura  11
Habitual seizure (n)
 CPS  711 96
 CPS with additional p./c.j.  414
Secondary generalization  433
 regularly (n patients)
 CBZ (mg/d) [n]1,600.0 (495.0) [5]1,750.0 (480.6) [6]2,033.3 (557.4) [6]1,900.0 (141.4) [2] 
 PHT (mg/d) [n]  435.0 (89.4) [5]  283.3 (28.9) [3]    300.0 (139.3) [3]383.3 (28.9) [3]
 VPA (mg/d) [n]1,050.0 (636.4) [2]  825.0 (318.2) [2]
 Other AEDs [n]VGB, LTG, PRM, GBP [4]VGB, PRM, CLB [3]VGB, LTG, PB [4]LTG, GBP, OXC, PB,
   LEV, PRM [6]
Monotherapy (n)  8965
Combined therapy (n)  3245
Surgery (n)1111 10 9
 Proved HS only  6884
  + glob. TLS  4325
 Compatible with HS/TLS  11
Follow-up time (mo)  7.5 (1.4)7.0 (1.5)7.3 (1.0)8.1 (1.9)
 Class I  710 76
 Class II  12
 Class III1
 Class IV  21
 Not available  1112
Mean global CMRGlu27.8 (6.9)28.7 (7.2) 28.7 (7.1) 32.3 (9.2) 
  • 1All patients had seizures of unilateral mesiotemporal origin (mTLE), as demonstrated by continuous interictal and ictal video-EEG monitoring during the preoperative phase I studies at the Bethel epilepsy center. FDG-PET was carried out as part of this noninvasive study protocol.
  • 2Patients with a history of cerebral trauma, meningitis, dysplasia, hamartomas, vascular malformations, tumors, and former neurosurgical treatment were excluded.
  • 3In all patients, MRI scans showed no pathology other than ipsilateral hippocampal atrophy or sclerosis.
  • 4After the PET study, 41 of the 42 patients underwent temporal lobectomy. All patients completed the first routine follow-up protocol ∼7.5 months after surgery. This included neurologic examination, latest medical history for seizures/medications, and neuropsychological retesting.
  • 5All patients were right-handed, and for all patients, data from preoperative neuropsychological testing were available.

The patients were grouped into four subgroups according to gender and lateralization of the mTLE origin. Some overlap of patients in this study occurs with former studies from our group (6,7), but criteria for inclusion into this study were more strict, and new patients were included.


Twelve healthy gender- and age-matched, right-handed controls with a mean age of 30.6 years (SD, 7.8 years), comprising six women (mean age, 29.7 years; SD, 10.3 years) and six men (mean age, 31.5 years; SD, 5.2 years) who had no evidence of structural brain abnormality and a normal FDG-PET scan.

The subjects were told that the purpose of the study was to investigate the regional cerebral metabolism rate of glucose (rCMRGlu) pattern during resting wakefulness. Written informed consent was obtained in accordance with guidelines of the Declaration of Human Rights, Helsinki 1975, and approved by the Ethics Committee of the Heinrich-Heine-University Düsseldorf.


Positron emission tomography

PET scanning was carried out with the Scanditronix PC4096/7WB PET camera with a field of view of 92 mm, as described in detail elsewhere (13). In short, subjects were placed comfortably on the scanner bed, and the gantry of the PET camera was aligned with the orbitomeatal line. Room noise was minimized, and the lights were dimmed. The subjects were asked not to move or speak, and to keep their eyes open. All subjects were observed by a neurologist during scanning to monitor the resting wakefulness and clinical state. None of the patients had an epileptic seizure 24 h before or during the examination.

A bolus of 200 MBq 2-[18F]fluoro-2-deoxy-d-glucose (FDG) was injected through a small cannula in a brachial vein. The FDG-input function was obtained from arterialized blood samples with an oxygen saturation of >90% after the subject's left or right hand was warmed to ∼40°C for 45 min before PET scanning. Calculation of the rCMRGlu was performed as described by Phelbs et al. (14). The kinetic constants and the lumped constant of 0.52 were taken from the findings of Reivich et al. (15). The reconstructed PET images had an in-plane resolution (full width, half maximum; FWHM) of 7.1 mm. The distance of the axial images was 6.5 mm.

Reconstructed FDG-PET images were analyzed by using the computerized brain atlas program (CBA) of Bohm and Greitz, which has been described in detail elsewhere (13). In short, this program allows spatial standardization of PET images of different subjects by using translations and linear and nonlinear transformations. This compensates for different brain shapes and sizes and enables us to evaluate these images statistically on a pixel-by-pixel basis. The spatially standardized parametric images of the CBA had an axial separation of 6.5 mm and a pixel size of 1.27 mm2 (giving a pixel volume of 10.5 mm3). Spatially standardized and metabolically normalized rCMRGlu images (reference: mean global metabolism of the control group) were used in comparisons within and between control and patient groups. Comparisons were performed by using t-maps calculated according to a random-effects model on a pixel-by-pixel basis, as previously described (6). The t-maps thresholded at p < 0.01 were analyzed for clusters of ≥12 pixels, which partially corrected for spatial correlation and multiple comparisons (6,13). If this thresholding cancelled out all differences, t-maps were reevaluated at p < 0.05. Any significant mean rCMRGlu depressions were localized by using anatomic templates of the computerized brain atlas (16) and by plotting the centers of gravity of the significant mean rCMRGlu depressions into the stereotactic coordinates of the Talairach and Tournoux atlas (17), obtained from the CBA. For visualization, the significant hypometabolic regions were superimposed on a spatially standardized MRI dataset of a healthy control by using the CBA. This brain had been determined to be the most representative of 28 healthy male volunteers (18).

In addition, the data of each patient and each control subject were evaluated in regions of interest (ROIs) that covered the temporal, frontal, and insular cortex (see Fig. 1 for details; the ROIs of the insular cortex in slices 10 and 11 are small and not shown in this figure). All ROIs were drawn on homologous regions of both hemispheres, which were displayed on a spatially standardized MRI dataset of the reference brain and projected on the PET images. The mean CMRGlu of each ROI was normalized to the mean global metabolism of each patient before comparison with controls and across patient subgroups.

Figure 1.

Location of the regions of interest (ROIs). The slice-numbers (7-12) refer to the slices of the standardized CBA-datasets: slice 1 (lowest z-value) to slice 21 (highest z-value).

The ROI data also were used for calculations of asymmetry indices,


A value <0 indicated a lateralization of hypometabolism to the left hemisphere, and a value >0, a lateralization to the right hemisphere. The metabolism in a pair of ROIs was taken as lateralized when the AI was >10%, because no gender differences >10% were observed in the controls, and the 10% cut-off is well known from other studies (19). Moreover, for correlation analysis with the neuropsychological data, squared values of the AIs were used to find out if metabolic asymmetry per se plays a role (7).

Neuropsychological assessment

The neuropsychological assessment was performed within the preoperative phase-I studies and at follow-up. The tests included the German version of the Wechsler Memory Scale–Revised (WMS-R) with the Logical Memory Test (Story B, delayed recall) and the Wechsler Adult Intelligence Scale–Revised (WAIS-R). Scores of the following subtests are reported in this study: Verbal/Performance Full-Scale-IQ, Information, Comprehension, Similarities, Digit Symbol, Picture Completion, Block Design, and the Digit Span (forward) (20,21).

Memory performance was tested further by using the California Verbal Learning Test (delayed recall and percentage recall consistency across trials) (22), the Rey Visual Design Learning Test (correct/errors delayed recall) (23), and the Corsi Block-Tapping Span (24) for visuospatial short-term memory.

Verbal fluency was assessed by the Controlled Oral Word Association Test (letters b, f and l) (25) and Category Naming (“animals”). This assessment examines categorical verbal fluency, an impairment that is associated with frontal lobe damage (26). The Trail-Making Test (TMT, frontal lobe time) (27) was used to explore visuoconceptual tracking. The performance in visual search, mainly a right frontal lobe activity, was measured by the Two-and-Seven Test (digits/letters) (28). Details of the tests used for the neuropsychological evaluation of the patients are described in detail elsewhere (7,29).

Statistical analysis

The statistical analysis for the ROI/AI metabolism data, the neuropsychological tests, and the correlation analysis was performed by using the German version of SPSS for Windows 10.0.5 (SPSS Inc., 1989–1999). Group comparisons were calculated by using two-tailed t tests (Fisher) and an analysis of variance (ANOVA). A correlation analysis between metabolic values obtained in the ROIs and the subunit scores of the presurgical neuropsychological tests was performed by using the two-tailed nonparametric Kendall's W (Kendall's tau b in the German version). For the statistical analysis, the significance level was set at p < 0.05 with post hoc correction for multiple comparisons (Bonferroni). The neuropsychological scores (NPSs) were either lower (+) or higher (–), as depicted in Fig. 2. As explained earlier, the metabolism asymmetry index (AI) was negative for a left-lateralized hypometabolism and positive for a right-lateralized hypometabolism. Thus a positive Kendall correlation indicated a lower NPS with a left-lateralized hypometabolism or a higher NPS and a right-lateralized hypometabolism. In contrast, a negative Kendall correlation indicated a higher NPS with a left-lateralized hypometabolism or a lower NPS and a right-lateralized hypometabolism (see Fig. 2).

Figure 2.

Correlation of FDG-PET asymmetry and neuropsychological testing (see text for further details).


Clinical data

Age, TLE onset, disease duration, and seizure frequency were similar for the four subgroups of patients (Table 1). Although the mean seizure frequency of the right-sided women was lower than that of the left-sided men, this difference did not reach significance, probably because the standard deviations were relatively large in all groups. Importantly, the types of aura, showing a preponderance for epigastric and psychic (anxiety, déjà vu) aurae in each subgroup, the type of seizure semiology, and the occurrence of secondarily generalized seizures, were similar between the subgroups. As an exception, the subgroup of the right-sided male patients had no regular experience of generalized seizures at the time before data acquisition (Table 1).

The examination of the resected tissue available from the 41 patients that underwent temporal lobectomy after the PET study revealed sclerosis of the hippocampal formation in all cases. In 39 cases, the diagnosis of hippocampal sclerosis was proven, and in two patients, the results were compatible with this diagnosis. The resected specimen of one third of the patients presented evidence for an additional gliosis in lateral parts of the temporal lobe. At the time of the follow-up examinations (mean intervening period, 7.5 months; range, 3–12 months), 30 patients (75% of sample) remained seizure free after surgery (class I, according to the classification of Engel) (30). No significant differences were found between subgroups of gender or side of TLE.

As the patients had medically refractory mTLE, the antiepileptic drug (AED) treatments were quite variable across individuals. Several patients were treated medically with a combined-therapy approach, and apart from the most common medication with carbamazepine (CBZ) or phenytoin (PHT), several other AEDs were used alone or in combination with CBZ or PHT. However, the median dosages for CBZ and PHT were not different among groups, based on a one-way ANOVA with post hoc Bonferroni correction for multiple comparisons (p value of the CBZ ANOVA, 0.539; p values of the PHT ANOVA, 0.109; all p values for post hoc tests, >0.20). A subgroup t test for the left-sided men showed comparable IQ scores (full-scale, verbal, and performance IQ) and WAIS-R subscores between those patients with (n = 5) and without (n = 6) PHT therapy (p > 0.20). Thus we did not expect any differential influence of the PHT therapy on the mean group neuropsychological performance or cerebral metabolism.

Cerebral metabolism


The global metabolism (CMRGlu) of male and female controls was identical with a mean global CMRGlu of 31.1 (8.1) μmol/100 g/min. Direct pixel-by-pixel comparison of the male and female controls revealed a relative hypometabolism (p < 0.01) in the left medial and inferior frontal gyrus (CBA slices 10 and 13) in the men and bilaterally in the lingual and posterior medial temporal and occipital gyrus (CBA slices 9 to 14) in women (Table 2). No overlap was seen between these regions and those found in the patients (see later).

Table 2. Regions with gender-specific relative hypometabolism in the controls
Anatomy, L/RCoordinate (x; y; z)SliceSizeng-rCMRGlu (μmol/100 g/min) Δ rCMRGlu %
  1. Data normalized to the mean global cerebral glucose metabolism (CMRGlu) of the control group. Results are shown at p < 0.01 for both groups (male and female controls). Slice numbers refer to the computerized brain atlas (CBA). Size is expressed in the number of pixels in a slice.

Male controls
 Gyrus frontalis medius, L −31 +68   −2101227.834.4
 Gyrus frontalis medius/inferior, L −39 +21 +16134535.148.5
Female controls
 Gyrus lingualis, L>R   −4 −115    −6  97033.145.8
 Gyrus lingualis, cuneus, R>L +12 −97   +110107 39.542.5
 Cuneus, R +12 −105    +2112127.653.8
 Gyrus temporalis/occipitalis medius, L −32 −88 +18131730.237.8
 Gyrus temporalis/occipitalis medius, L −37 −83 +23141536.032.2

The controls did not show any lateralization (AI <10%) in those ROIs that were different in the patient groups (Table 4). Comparison of the AIs between male and female controls found significant differences (p < 0.05; AI, >10%) in the frontolateral (CBA slices 10 to 12) and frontomesial cortex (CBA slice 10), with the women being more asymmetric in terms of a relative right-hemispheric hypermetabolism (mean AI, ≥10%) compared with the men (mean AI, <5%). No significant differences were observed for the squared AIs.

Table 4. Metabolic asymmetries in mTLE and correlations with neuropsychology
mTLE subgroupLeft maleLeft femaleRight maleRight femaleControlsNeuropsychology
  1. Values are shown as mean (SD); in the first column, the anatomic localization of the regions of interest (ROIs; temp, temporal lobe; front, frontal lobe; mes, mesial; lat, lateral) used for the calculation of the asymmetry indices (AIs) is given; the slice numbers refer to the 21 slices of the standardized computerized brain atlas (CBA) datasets: slice 1 (lowest z-value according to the Talairach–Tournoux system) to slice 21 (highest z-value). AIs with significant group differences (p < 0.05) after Bonferroni correction for multiple comparisons are listed (m, male; f, female; r, right; l, left); + or − before the digits indicates lateralization of hypometabolism to the left/right hemisphere, respectively. The last column shows significant (p < 0.01) Kendall's correlation (CC, correlation coefficient) with presurgical and postsurgical (before and after) Trail-Making test (TM) and WMS-R Logical Memory–Delayed Recall (WMS-R-LM), and WAIS-R Picture Completion (WAIS-R-PC) tests.

N patients1111101012Difference p < 0.05SubtestCC
Temp-mes (slice 7) −11.9 (9.0)  −20.1 (9.6)   +19.6 (10.1) +19.7 (13.0) +3.2 (4.7)All vs. contralateral/controls
Temp-lat (slice 7) −14.6 (10.2) −7.6 (24.0) +21.5 (15.1) +18.0 (21.0) +1.4 (8.1)All vs. contralateral
Temp-mes (slice 8) +2.4 (6.7) −5.1 (9.9)  +17.9 (11.2) +13.9 (12.9) +3.9 (7.5)M vs. f, rm vs. left/controls
Temp-lat (slice 8) −12.7 (13.1) −7.2 (13.5) +21.4 (13.1) +21.2 (15.2) +6.7 (9.8)All vs. contralateral,TM (post) +0.315
Insula (slice 8)   −1.0 (12.3) −9.2 (9.3)    +8.4 (13.3) +14.1 (10.4)   +2.5 (11.1)Lf vs. right, rf vs. left
Temp-lat (slice 9) −14.7 (13.2) −6.5 (14.8) +16.7 (11.6) +18.3 (10.9) +4.8 (7.4)All vs. contralateral, lm vs. controlsTM (post) WMS-R-LM (post) +0.313 +0.296
Front-mes (slice 9)   −2.5 (13.2) −0.9 (14.5)      3.0 (12.6)    7.5 (6.0)    5.7 (8.1)WAIS-R-PC (pre) −0.329
Front-lat (slice 9) +4.5 (9.3) +8.2 (12.4)   +6.4 (10.3) +19.6 (14.1)     +7.3 (12.2)Lm vs. rf
Insula (slice 11) −2.4 (9.2) −6.9 (11.0) +7.7 (7.4)   +4.6 (14.4) −0.2 (7.0)Lf vs. rm


The global mean CMRGlu was slightly reduced in the patients compared with controls, except for the subgroup of the right-sided female mTLE patients (Table 1). However, none of these differences reached significance.

The regional pixel-based analysis showed that the left-sided (n = 22) and the right-sided (n = 20) mTLE patients had a mesiotemporal and temporolateral hypometabolism, with more widespread changes in left-sided mTLE substantiating our earlier observations (6). The hypometabolism in the mesial temporal lobe involved more than twice as many pixels (384 vs. 164) in the left-sided mTLE group than in the right-sided. Additionally, an extratemporal remote depression was present in the contralateral thalamus in left mTLE patients. No hypermetabolic area was found in the patients. The gender-specific subgroups showed patterns of regional hypometabolism that were generally more extensive in the male patients (Fig. 3, Table 3).

Figure 3.

Patterns of mean hypometabolism in male and female mTLE patient groups in comparison to the controls (p < 0.01). The hypometabolism in the right female patients was less pronounced (p < 0.05). The slice numbers (7-12) refer to the slices of the standardized CBA-datasets: slice 1 (lowest z-value) to slice 21 (highest z-value). The left-sided male group is on the top left side, left-sided females on the bottom left side, right-sided males on the bottom right side and right-sided females on the top right side.

Table 3. Areas of gender-specific hypometabolism in mTLE
Anatomy, L/RCoordinate (x; y; z)SliceSizeng-rCMRGlu (μmol/100 g/min) Δ rCMRGlu %
  1. Data normalized to the mean global cerebral glucose metabolism (CMRGlu) of the control group. Results are shown at p < 0.01 for the groups of left-sided mTLE (males and females) and right-sided male mTLE. The results for the group of right-sided women were obtained at p < 0.05. Slice numbers refer to the computerized brain atlas (CBA). Size is expressed in the number of pixels in a slice.

Male left-sided mTLE
 Gyrus frontalis inferior, L −21 +21 −1571237.223.4
 Mesencephalon, L   −6 −19 −1571311.340.0
 Hippocampus + gyrus parahippocampalis, L −38 −54 −1578625.626.0
 Gyrus temporalis superior/medius, L −51 −61 −157109 30.628.4
 Gyrus cinguli > frontalis inferior, L −17 +21 −1283434.022.9
 Mesencephalon, L   −7 −15 −1281920.331.1
 Mesencephalon, mesial + R +10 −21 −1282925.726.1
 Hippocampus + gyrus parahippocampalis, L −31 −51 −1283823.625.8
 Gyrus temporalis superior/medius, L −58 −29 −128103 33.927.2
 Gyrus temporalis superior/medius, L −64 −24   −694239.323.1
 Thalamus/nucleus lentiformis, R +16 −17   −693326.127.8
 Gyrus frontalis medius, L −41 +27 +1312 1621.038.7
Female left-sided mTLE
 Gyrus temporalis superior, L −56    0 −1573432.131.6
 Amygdala, hippocampus, gyrus −48 −30 −167156 27.535.7
     parahippocampalis/temporalis medius, L
 Gyrus parahippocampalis, L −24 −41 −1671328.133.6
 Hippocampus + gyrus parahippocampalis, L −23 −30 −1283820.830.0
 Gyrus temporalis medius/fusiformis, L −48 −75 −1281538.125.4
Male right-sided mTLE
 Gyrus temporalis medius, R +65   −8 −1577328.226.7
 Gyrus temporalis medius, R +62 −56 −1571537.823.0
 Gyrus temporalis superior, R +54 +18 −1574624.338.5
 Gyrus subcallosus, R   +7   +4 −1572618.838.4
 Nucleus accumbens septi, L −10   −8 −1281219.429.6
 Mesencephalon   −2 −45 −1282128.427.4
Female right-sided mTLE
 Hippocampus, R +29 −27 −1571924.720.1

When compared with controls, the left-sided men showed the most widespread hypometabolism (p < 0.01). The hypometabolism extended over three consecutive slices (CBA slices 7–9) in the left temporal lobe including the temporal pole and mesial and lateral regions (Fig. 3, Table 3). The maximal metabolic depression was found in the medial and superior temporal gyrus (x − 51/y − 61/z − 15) and involved additional remote areas such as the ipsilateral frontoorbital cortex and the contralateral thalamus. The female left-sided patients showed a pattern of hypometabolism (p < 0.01) that was restricted to two consecutive slices of the left temporal lobe (CBA slices 7 and 8) and included the temporal pole and mesial and lateral regions, with a large mesiotemporal region (x − 48/y −30/z − 16). The male right-sided patients showed a hypometabolism (p < 0.01) that included mesial and lateral temporal regions, the temporal pole, a dorsal part of the midbrain, and an additional region in the mesial frontoorbital region in CBA slices 7 and 8 (Fig. 3, Table 3). The largest hypometabolic region was located in the medial temporal gyrus (x + 65/y − 8/z − 15). The hypometabolism was less pronounced in the subgroup of female right-sided mTLE patients. Even at a lower level of significance (p < 0.05), only one small mesiotemporal region in CBA slice 7 near the hippocampus (x + 29/ y − 27/z − 15) showed a hypometabolism.

Direct comparisons of male and female metabolic patterns confirmed a more widespread hypometabolism in male patients both for right and left mTLE. It became evident that in male patients, the extramesiotemporal hypometabolism extended bilaterally into the frontal lobes. This is illustrated in Fig. 4 for the group of left-sided male mTLE patients by direct comparison with the group of left-sided female patients.

Figure 4.

Areas of gender-specific mean hypometabolism in male left mTLE-patients (direct comparison of left-sided male vs. left-sided female mTLE-patients in groups; p < 0.05) shown as categorical comparison across groups. The slice numbers (7-12) refer to the slices of the standardized CBA-datasets: slice 1 (lowest z-value) to slice 21 (highest z-value).

Regional metabolic asymmetry

Taking the AIs into account, the mesial temporobasal metabolism distinguished all subgroups from the controls and the contralateral groups (Table 4). The AI of lateral temporal lobe metabolism running from slice 7 to slice 9 of the CBA also facilitated the differentiation between left- and right-sided mTLE.

Additionally, a trend toward right hemispheric insular hypometabolism was present for both left- and right-sided male patients. In contrast, the right-sided females showed a less extensive hypometabolism, and metabolism of the left-sided female patients was normal. The difference between the latter and the right-sided men reached significance.

Analysis of the squared AIs of the frontal ROIs demonstrated consistent gender differences mainly for the frontolateral cortex. A larger amount of asymmetry was seen in the female patients than in the males (squared AIs in CBA slices 8–10, p < 0.05), obviously due to a more unilateral metabolic impairment than seen in the men (Table 5 and Fig. 4).

Table 5. Gender-specific asymmetries of rCMRGlu as indicated by squared metabolic asymmetries and neuropsychology
 Squared AIsNeuropsychology
Squared AIFemale mTLEMale mTLEDifferenceIQ scoreCC
  1. Values are shown as mean (SD); in the first column, the anatomic localization of the regions of interest (ROIs; temp, temporal lobe; front, frontal lobe; mes, mesial; lat, lateral) used for the calculation of the squared asymmetry indices (AIs) is given; the slice numbers refer to the 21 slices of the standardized computerized brain atlas (CBA) datasets: slice 1 (lowest z-value according to the Talairach–Tournoux system) to slice 21 (highest z-value); NS, not significant. The last column shows the significant Kendall's W correlation with the pre- and postsurgical neuropsychology (ap < 0.05; bp < 0.01). Fs IQ, full-scale IQ; Verb IQ, verbal IQ; Perf IQ, performance IQ.

Front-mes (slice 7)183.5 (230.1)201.0 (279.5)NS
Front-lat (slice 7)1,399.5 (1,941.1)854.5 (938.7)NS
Front-mes (slice 8)262.8 (281.1)379.4 (601.7)NS
Front-lat (slice 8)568.0 (594.7)217.0 (281.3)<0.05
Insula (slice 8)229.1 (249.4)182.0 (216.9)NSVerb IQ (pre)0.247a
Front-mes (slice 9)142.7 (203.9)159.2 (174.7)NS
Front-lat (slice 9)376.3 (463.9)116.4 (151.3)<0.05Fs IQ (pre) −0.286b
 Fs IQ (post) −0.228a
 Verb IQ (pre) −0.231a
Insula (slice 9)188.3 (193.8)257.6 (384.4)NSPerf IQ (pre)0.403b
 Perf IQ (post)0.367b
Front-mes (slice 10)322.6 (258.5)173.3 (287.5)NS
Front-lat (slice 10)668.6 (717.8)299.2 (295.0)<0.05Fs IQ (pre) −0.237a
 Verb IQ (pre) −0.226a
Front-mes (slice 11)133.5 (161.4)  98.4 (135.7)NS
Front-lat (slice 11)597.6 (686.2)354.1 (322.3)NS
Insula (slice 11)181.3 (245.5)  94.8 (111.2)NSPerf IQ (pre)0.239a
Front-mes (slice 12)  95.4 (111.0)  91.3 (128.8)NS
Front-lat (slice 12)367.4 (440.1)238.1 (218.2)NS

Neuropsychological assessment

The educational level and IQ scores were similar across patient groups. NPS showed great interindividual differences across all patients. Significant group differences (p < 0.05) on the neuropsychological performance were found by using the t test. The group of male right-sided mTLE patients achieved significantly higher scores on several subunits of the WAIS-R than did the female right-sided mTLE patients.

In addition, a clinically relevant impairment (>1 SD from normal controls; 1 SD being equivalent to 15 points for the WAIS-R) occurred in the mean scores of the WAIS-R full-scale IQ for the right-sided female group (86.0 ± 13.3) in comparison with the right-sided males (107.4 ± 13.8), which reached statistical significance (p < 0.05). In the verbal IQ, the right-sided male patients (110.5 ± 11.6) performed better (p < 0.05) than all other subgroups, without relevant differences among the others (86.5 ± 15.4 to 94.2 ± 16.5).

Findings were different for the performance IQ, which showed normal values for the left-sided women (106.4 ± 21.1) and the right-sided men (99.8 ± 18.6), and slightly reduced values for the left-sided men (91.1 ± 17.6) and the right-sided women (86.4 ± 11.1). However, these differences failed to reach a statistically significant level (each p > 0.05).

At the time of the postoperative follow-up assessment, all subgroups showed improved performance in some tests, but most scores remained slightly reduced as compared with healthy controls. Interestingly, the left-sided women exclusively showed a tendency toward a postoperative decline of performance for verbal and performance IQ, whereas all other subgroups showed a trend toward improved IQ subscores.

A postoperative increase in performance IQ scores for the left-sided male patients (103.3 ± 21.1 vs. 91.1 ± 17.6) reached statistical significance (p < 0.05). The same subgroup also improved significantly in the WAIS-R subtests Picture Completion and Block Design. Left-sided women and right-sided men improved significantly in the Digit-Symbol subtest, and right-sided women improved in the Similarities subtest. In contrast, a decline was detected for the subgroup of the left-sided women in the picture-completion part (all p < 0.05).

Correlation of hypometabolism and neuropsychology

A correlation analysis was performed to detect a possible relation between presurgical neuropsychological scores (subset and IQ scores) and the AI metabolic data, given that the nature of the ROI-based data allowed a comparison across the different patient groups.

Correlation analysis between AIs and NPS showed that poor performance in the Trail-Making Test and the WMS-R Logical-Memory Test correlated with a temporolateral hypometabolism in the right hemisphere (CBA slice 8 and 9, Kendall's correlation coefficients, +0.296 to +0.315; p < 0.01). In contrast, good scores in the WAIS-R Picture Completion subset correlated with a relatively preserved right-hemispheric frontomesial metabolism (CBA slice 9, Kendall's correlation coefficient, –0.329; p < 0.01; Table 4).

In addition, correlations between the neuropsychological scores and the squared AIs of the regional cerebral metabolism were performed, because the use of squared AIs has been shown to be more sensitive than the raw AIs (7). This analysis eliminated the algebraic sign of the AI and indicated the relation of neuropsychological performance and cerebral metabolic asymmetry, with a greater numeric value suggesting greater metabolic asymmetry (Fig. 2)

The most outstanding result of this analysis was that full-scale IQ scores and verbal IQ scores were negatively related to an asymmetric metabolism in the frontolateral brain (CBA slices 9 and 10, Kendall's correlation coefficients, –0.226 to –0.286; p < 0.05, except for the full-scale IQ in slice 9 with p < 0.01). This means that higher frontal metabolic asymmetry was related to worse IQ scores. In contrast, a significant positive correlation was found of the performance IQ score with metabolism in the insular cortex (Kendall's correlation coefficients, +0.367 to +0.403; p < 0.01, in CBA slice 9, and +0.239; p < 0.05, in CBA slice 11), indicating a relation between good performance and an asymmetric insular metabolic pattern (Table 5).


We report a gender-specific pattern of hypometabolism in mTLE patients. We found male patients to have more widespread regions of hypometabolism than did the female patients. This included remote hypometabolism in extratemporal cortex in the frontal lobe, both in left and right mTLE.

Remarkably, no significant hypometabolism was seen for the group of female right-sided mTLE patients at the standard threshold (p < 0.01). However, a hypometabolic region occurred in the contralateral thalamus of the male left-sided mTLE patients, with no evidence for an ipsilateral thalamic impairment, which would be suggested by other metabolic (31) and structural (32) imaging studies.

Hypometabolism in the mesial temporal region was not different between men and women. A misleading picture of less extensive mesial temporal hypometabolism in men (124 vs. 207 pixels for left TLE and 0 vs. 19 pixels for right TLE) was found on summation of the pixels of the significant mesiotemporal hypometabolic regions from the CBA analysis. However, this pseudo-difference was due to a very large region (156 pixels, CBA slice 7) in the female patients, which exceeded the mesiotemporal region including also extramesiotemporal cortex (left TLE).

Based on the AI values, gender-specific differences of the asymmetry in the temporal lobes were not significant. Frontal metabolism was significantly more asymmetric in female patients. This may have been caused by a more bilateral frontal hypometabolism in male patients, in addition to a relative right-hemispheric frontal hypermetabolism with consecutive asymmetry seen in our female controls.

Our findings confirm and extend earlier FDG-PET studies on the distribution of hypometabolism and seizure spread in mTLE, as shown by pixel-to-pixel comparison (6), ROI-based analysis (7,33), and EEG (34). It has been argued that remote depressions of metabolism do not simply reflect regional atrophy (35). Rather they represent a functional deficit related to the structural abnormalities in the hippocampal formation.

As found by EEG, seizure activity may spread from the seizure-onset zone in the temporal lobe to the ipsilateral frontal lobe in mTLE (34). The activity typically crosses over to the contralateral hemisphere via frontal commissures. From there, it finally reaches the contralateral temporal lobe via the contralateral frontal lobe. Although this was observed in the majority of cases, it is not the only possible route of spread. The functional and structural changes in the mesial temporal regions correlate with seizure frequency (33,36,37). The most likely explanation for the reduced metabolism in remote brain areas is a functional inhibition of the remote regions by ongoing, subclinical epileptic activity in the epileptogenic mesial temporal lobe (9).

As has been shown in animal models, remote hypometabolism is a basic effect of epileptic activity (38,39), which is associated with reduced synaptic activity and tonic hyperpolarization of neurons. These experiments also showed that because of the functional trait in the absence of any structural abnormalities, the hypometabolism may be reversible (40).

Thus the neocortical areas that were found to be functionally deficient in terms of hypometabolism (6,7) probably represent the “functional deficit zone” as proposed by Lüders (41,42). That remote hypometabolism in mTLE represents a functional abnormality and is potentially reversible is shown by postoperative normalization after epilepsy surgery (43,44).

In a previous report, Savic and Engel (45) reported gender-specific differences in FDG-PET count-rate images in mTLE patients. They described a higher incidence of frontal lobe hypometabolism ipsilateral to the hemisphere of habitual seizure onset in the male patients combined with a higher incidence of spread of epileptiform activity to the ipsilateral frontal lobe, as recorded in the scalp sphenoidal EEG during a habitual seizure. They interpreted these findings as evidence for an underlying inherent gender dimorphism in cerebral connectivity. Our PET findings are in accordance with these results. However, we were not able to confirm the former finding of more contralateral temporal abnormalities in FDG-PET of female patients. The correlation with neuropsychological data emphasizes the clinical impact of the gender-specific pattern of hypometabolism in male mTLE patients.

Because of the strict clinical inclusion criteria, a prerequisite for the detection of gender differences in cerebral metabolism in the imaging analysis, the sample size was probably too small to reveal intergroup differences in neuropsychological test performance or subgroup-specific metabolic correlations.

Nevertheless, our data supported the model of malperformance in mTLE (7). Specifically, we found right hemispheric temporolateral hypometabolism to be associated with worse results on logical memory tasks. In contrast, good scores in visual attention and memory (WAIS-R Picture Completion) were associated with nondisturbed frontomesial metabolism. WAIS-R IQ scores correlated inversely with an asymmetry of metabolism in the frontolateral cortex (full-scale and verbal IQ). This suggests the importance of a more counterbalanced metabolism between both hemispheres and therefore a vulnerability toward metabolic disturbances in the frontal lobe as a remote effect mainly in male mTLE. For the performance IQ, better scores correlated with a rather asymmetric remote metabolism in the ipsilateral insular cortex.

Gender differences also have been reported for structural MRI findings, both in healthy volunteers and in patients with TLE. In a study using MRI and gray-matter volume measurements (46), women were found to have higher gray-matter percentages in language-related cortical regions (in the superior temporal gyrus and in the dorsolateral prefrontal cortex).

On this basis, one could assume a greater reserve or potential for plasticity in language regions in women. Thus the same progressive harm (for example, mTLE with recurrent seizures and consecutive release of toxic doses of glutamate) could lead to more functional and/or structural disturbances, resulting in regional cortical hypometabolism in men. Taking into account these considerations and the assumption of a larger gray-matter volume in the superior temporal gyrus in women, our findings of a larger temporolateral hypometabolism in our male mTLE patients is remarkable.

MRI volumetric calculations demonstrated a larger vulnerability to seizure-associated progressive brain damage in male compared with female mTLE patients (47). After a mean duration of >20 years after seizure onset, men had significantly more relative hemicranial volume loss than did women (both calculated in comparison to healthy men/women). This was present ipsilateral (12%) and contralateral (7%) to the hemisphere of seizure onset in men, but only ipsilateral and to a lesser degree in women (4%). This is in agreement with our findings of the more bilateral hypometabolism in male patients.

Recently, animal studies of gender differences in epilepsy research demonstrated the outstanding role of sex hormones. Studies of pharmacologically induced epilepsy (kainic acid and pilocarpine), revealed that male rats were more susceptible to convulsant agents and showed a higher seizure frequency. Testosterone was suggested to mediate these gender-specific effects on seizures (48), indicating a higher vulnerability toward hormonal triggers of epilepsy in men. This also could play a role in the gender-specific metabolic disturbances reported in this study, where seizure frequency and onset of epilepsy were not significantly different among our male and female patients. In another study, female rats were found to show more frequent and prolonged slow spike-and-wave discharges than males after treatment with the cholesterol synthesis inhibitor AY-9944. Sex hormones were assessed as an important factor for intermediation, this time suggesting the effect of a higher vulnerability in females (49). Together, these animal data show that sex hormones may play a relevant role in epilepsy; however, a single-edge disadvantage for one of the sexes seems unlikely.

Finally, functional magnetic resonance imaging (fMRI) showed that during a spatial task (virtual navigation through a three-dimensional maze), men showed activation of the left hippocampus and right parahippocampal gyrus, whereas women activated right-hemispheric regions in the frontal and parietal lobe (50). This suggests that functional disturbances of the temporal lobe could result in a gender-specific decline of neuropsychological performance. Gender-dependent performance was also different, with men finding their way out of the maze significantly faster. With fMRI, estrogen has been shown to alter brain-activation patterns in postmenopausal women in specific brain regions during working-memory task performance (51). This suggests a modulating role of sex-hormones for cognitive performance as an extracerebral trigger of brain function. In another fMRI study (52), gender differences were reported during frequency-dependent pattern processing, with men showing a more global (right-hemispheric) information processing than women. In addition to the expected differences for left- and right-sided TLE patients, an fMRI study about phonologic and semantic processing in TLE patients (53) reported a greater spatial extent of activation for men in the left middle temporal gyrus during the phonologic-processing condition. In women, a greater spatial extent of activation was found in the right inferior frontal cortex and the left hemisphere during the semantic processing task. These studies suggest a wider distribution of cerebral activity in women than in men during cognitive performance, which may render cognitive functions in men more vulnerable.

In conclusion, our findings of a gender-specific pattern of hypometabolism in mTLE may be the result of physiologic gender-specific differences in functional cerebral connectivity and discrete physiologic differences in brain structures between male and female subjects. These factors may build the framework for epilepsy-specific changes that affect the sexes differentially and males more severely.


Acknowledgment:  We thank Prof. Dr. R. Lahl and Dr. R. Villagran from the Neuropathological Institute of the “von Bodelschwinghschen Anstalten Bethel” in Bielefeld for histological evaluation of the resected specimen and for making the results available to this study. We are grateful to Dr. A. Wirrwar, Department of Nuclear Medicine, University-Hospital Düsseldorf, for expert technical support during the PET scans. We thank Dr. Leeanne Carey for the helpful comments and language editing of this manuscript. This research was funded by the Sonderforschungsbereich (SFB) 194 “Strukturveränderung und Dysfunktion im Nervensystems” of the Deutsche Forschungsgemeinschaft (DFG) and the DFG-Teilprojekt “Epilepsieforschung” (No. 701010198).