Address correspondence and reprint requests to Dr. W. H. Theodore at NIH/NINDS/ERB, Bldg. 10, Room 5N-250, 10 Center Dr., MSC 1408, Bethesda, MD 20892–1408, U.S.A. E-mail: firstname.lastname@example.org
Summary: Purpose: Reports conflict on the relation of glucose metabolism to hippocampal volume in temporal lobe foci. Previous studies usually have used side-side ratios rather than regional metabolic rates.
Methods: We measured hippocampal volume and glucose metabolism in 37 patients with temporal epileptogenic zones identified by ictal video-EEG telemetry. Metabolic rates were normalized to global brain mean.
Results: Both 18-fluoro-2-deoxyglucose-PET and volumetric MRI lateralized the epileptic focus determined by ictal video-EEG. There were significant correlations between left-right metabolic asymmetry and hippocampal formation volume left-right ratios. Comparisons between normalized metabolism and hippocampal formation volume, ignoring the side of the epileptic focus, showed significant relations between left hippocampal volume and left inferior lateral temporal metabolism, right hippocampus and right inferior mesial temporal, and left hippocampus and left inferior mesial temporal metabolism. In contrast, when normalized metabolism was compared with hippocampal volume in the epileptic focus, no relation was found.
Conclusions: Our study suggests that the relation between hippocampal volume and glucose metabolism breaks down in epileptic foci and that hypometabolism is not dependent on neuronal loss. It is consistent with data suggesting that hypometabolism is an independent predictor of surgical outcome.
The origin of hypometabolism in temporal lobe foci is uncertain. Commonly, in patients with mesial temporal sclerosis the hypometabolic zone extends beyond the region of increased hippocampal formation (HF) signal on magnetic resonance imaging (MRI) and cell loss found at surgery (1). Hypometabolism is prominent in lateral as well as mesial temporal cortex and may extend into other regions, including frontal lobes and thalamus (2).
However, neuronal loss could still be a contributing factor. In a previous study, we found a significant correlation between asymmetry (asymmetry index [AI]) in HF volume as measured on MRI and AIs for both mesial and lateral temporal metabolic rate for glucose (CMRglc) (3).
Other investigators have also addressed this issue. Semah et al. (4) found a significant correlation between HF volume and “temporo-limbic” asymmetry. O'Brien et al. (5) found a significant correlation between HF volume and mesial and lateral temporal asymmetry when the epileptic focus was compared with the contralateral side but not when the smaller side was compared with the larger side. Neither of these investigators studied the relation of HF volumes to regional metabolic rates rather than AIs.
The relation of HF volume to CMRglc may have clinical implications for patients with “temporal lobe epilepsy” and for understanding the place of 18-fluoro-2-deoxyglucose (FDG)-PET in presurgical evaluation. Neocortical and extratemporal hypometabolism may be associated with neuropsychological dysfunction or affect surgical outcome (6,7).
We studied 37 patients who had been referred to the National Institute of Neurological Disorders and Stroke Clinical Epilepsy Section for uncontrolled seizures. Temporal lobe epileptogenic zones were identified by ictal video-EEG telemetry.
Each patient had a coronal MR scan with contiguous slice thickness of 1.5 mm obtained with a General Electric Signa 1.5-T MR scanner (GE Medical Systems, Milwaukee, WI, U.S.A.). Scanning sequence was TR of 24 ms, TE of 5 ms with a flip angle of 45 degrees, and field of view of 24 × 24 cm, with a matrix of 256 × 256. The voxel size was 0.9375 × 0.9375 × 2 mm3. The images were transferred to a VAX station and analyzed using the Medical Imaging Retrieval, Analysis and Graphics (MIRAGE, NIH) software. HF volumes were drawn on consecutive images as described previously (8). The hippocampus was outlined in its entire extent, from the pes hippocampi anteriorly to the fascicular gyrus. The posterior limit of the hippocampus corresponded anatomically to the region of the pulvinar. The margins of the body of the hippocampus were defined superiority as the choroidal fissure curving laterally along the medial boundary of the temporal horn and then medially along the grey matter of the hippocampus, including the medialmost part of the subiculum where the hippocampus joins the parahippocampal gyrus. Anteriorly, the pes was distinguished from the amygdala, which is close to the anterosuperior aspect of the pes near the tip of temporal horn and the uncus. Pixel points were summed from successive pictures to calculate structure volumes. HF volume images had 15 slices.
PET studies were performed on the 15-slice Scanditronix PC 2048–15B (GE Medical Systems) scanner (axial and in-plane resolution, 5.5 mm). Patients were in the awake state with ears plugged and eyes patched. A thermoplastic face mask minimized head movement. EEG was monitored during radiotracer uptake period to detect ictal activity. Tomographic images were oriented parallel to the canthomeatal plane. After transmission scanning for attenuation correction, 5 mCi 18F-FDG was injected. Scanning began after a 40-min uptake period. We used a standard template of 188 regions grouped into 14 paired anatomic areas for PET data analysis (3). Regional values were normalized by dividing by the global mean. We used Systat (SSPS Inc., Chicago, Ill) for statistical comparisons.
Both FDG-PET and volumetric MRI lateralized the epileptic focus determined by ictal video-EEG. HF volume ipsilateral to the focus was 2.7 ± 0.10 (mean ± SEM) vs. 3.2 ± 0.08 cm3 contralateral to the focus; inferior lateral temporal (ILT) metabolism was 0.78 ± 0.02 vs. 0.91 ± 0.01; and inferior mesial temporal (IMT) 0.63± 0.03 vs. 0.78 ± 0.01 (all p < 0.001). Lateral temporal metabolic asymmetry was 0.19 ± 0.04, and mesial temporal was 0.17 ± 0.02. Mean HF asymmetry was 0.196 ± 0.04.
There was a significant correlation between left-right CMRglc asymmetry and HF volume left-right ratio in mesial (r = 0.427, p < 0.02) and lateral temporal cortex (r = 0.508, p < 0.01).
Regional comparisons between normalized CMRglc and HF volume, ignoring the side of the epileptic focus, showed significant relations between left hippocampal volume and left inferior lateral temporal CMRglc (r = 0.47, p < 0.01), right hippocampus and right inferior mesial temporal CMRglc (r = 0.36, p < 0.05), and left hippocampus and left inferior mesial temporal metabolism (r = 0.36, p < 0.05) (Fig. 1).
In contrast, when normalized CMRglc was compared with HF volume in the epileptic focus, no relation was found (Fig. 2).
Our study suggests that there is an overall relation between hippocampal volume and glucose metabolism in mesial and lateral temporal lobes. This relation may break down in epileptogenic regions, however, suggesting that the hypometabolism is not due mainly to neuronal loss. Studies of HF resections have not found significant CMRglc-neuronal density relations (9,10). These results suggest in addition that partial volume effects are not the main contributors to hypometabolism. However, the degree of volume loss was comparable to the degree of hypometabolism.
Semah et al. (4), who also used normalized values, found a correlation between CMRglc and volume only in HF and temporal pole—not in other regions. Dlugos et al. (11) found a correlation between HF density and bilateral thalamic, putamen, and globus pallidus CMRglc; they suggested thalamic hypometabolism could be due to decreased efferent activity from HF to thalamus and basal ganglia. Jokiet et al. (6) reported that the duration of epilepsy controlled for age at investigation, side of seizure origin, underlying cause, and sex were negatively correlated with ipsilateral and contralateral hippocampal volume, hippocampal metabolism, and Wada hemispheric memory performance. O'Brien et al. (5) confirmed our original observation that CMRglc and HF right-left AI were correlated but found no relation between “absolute” CMRglc and volume asymmetries; they suggested that decreased synaptic activity or areas of microdysplasia not detected on MRI could lead to hypometabolism. In this study, we confirmed that CMRglc and HF volume measurements reflect similar focus to contralateral AI. We also found that CMRglc within epileptic foci is not closely correlated to HF volume.
The investigations that have been performed have all involved approximately the same number of patients in each study, and the variability of the results may be due to relatively modest statistical power. Factors independent of HF volume, such as ictal clinical features or the time since the last seizure, may affect both the pattern and degree of hypometabolism (12,13). In addition, it may be harder to detect a correlation between HF and CMRglc when both are reduced, because the possible range of values is constricted. Antiepileptic drugs are unlikely to have affected the results, however, because we normalized PET data to the global mean.
In general, brain volume has relatively little effect on CMRglc. In healthy aging subjects 21–83 years old, cerebral atrophy, measured as increased cerebrospinal fluid volume, correlated significantly with global CMRglc but accounted for only 13% of the variance in CMRglc measurements (14).
Other pathological conditions also show reduced CMRglc out of proportion to atrophy. Patients with Alzheimer's disease have a true glucose metabolic rate reduction per gram of tissue, which persists even after partial volume correction (15).
The basis of hypometabolism in epileptic foci remains uncertain, but it appears to be less closely related to structural than to functional derangements. This is consistent with the presence of hypometabolism in extratemporal regions and the finding that PET hypometabolism is an independent predictor of outcome after temporal lobectomy (7).