ELSEVIER Electroencephalography and clinical Neurophysiology 103 (1997) 692-697
Topographic distribution of seizure onset and hippocampal atrophy: relationship between MRI and depth EEG
David King a'*, Richard A. Bronen b, Dennis D. Spencer c, Susan S. Spencer a
aDepartment of Neurology, Yale University School of Medicine, New Haven, CT 06510, USA bDepartment of Diagnostic Imaging, Yale University School of Medicine, New Haven, CT 06510, USA
CDepartment of Neurosurgery. Yale University School of Medicine. New Haven, CT 06510, USA
Mesial temporal sclerosis (MTS) is the most common pathological substrate underlying medically refractory med- ial temporal lobe epilepsy (MTLE). MTS is characterized by hippocampal cell loss and reorganization of the dentate gyms (de de Lanerole et al., 1992). Although the relation- ship between these pathological changes and the mechanism of epileptogenicity is not known, studies by Babb et al. (1984) suggested correlations between the topographic dis- tribution of cell loss and site of epileptogenesis in patients with MTS. More recently, however, Baulac et al. (1994) demonstrated only weak correlations between the degree of hippocampal damage and hyperexcitabili ty, concluding that 'the segmental distribution of MRI abnormalities can-
* Corresponding author. Department of Neurology, University of Penn- sylvania Medical Center, 3 W. Gates Bldg., 3400 Spruce St., Philadelphia, PA 19104, USA. Tel.: +1 215 3495166; fax: +l 215 3495733.
not be considered a reliable marker of precise seizure origin. ' Babb et al. (1984) analyzed volumetric cell densi- ties, while Baulac et al. (1994) used magnetic resonance imaging (MRI) volumetry to demonstrate the spatial distri- bution of hippocampal involvement.
MRI is a useful technique to assess hippocampal atrophy, both for research and for clinical purposes. Hippocampal atrophy has been shown to correlate with the presence of MTS, degree of hippocampal cell loss and postoperative seizure control (Jack et al., 1990; Bronen et al., 1991; Cas- cino et al., 1991; Lencz et al., 1992; Cook et al., 1992). We studied patients with MTLE who were investigated with depth EEG recordings and segmental MRI to establish the relationship between spatial distribution of hippocampal atrophy and sites of seizure onset. At our Center, the hippo- campus is studied with a depth electrode placed stereo- tactically along its longitudinal axis, providing spatially extensive and precise electrographic sampling capabili ty when intracranial recording is necessary.
EEG 97029
D. King et al. / Electroencephalography and clinical Neurophysiology 103 (1997) 692-697 693
2. Methods and materials
2.1. Patient selection and evaluation
Patients with refractory MTLE who were considered for anteromedial temporal lobectomy between 1987 and 1993 and evaluated by depth EEG, were retrospectively reviewed. Because of conflicting data between surface EEG recordings, clinical findings, MRT and functional ima- ging and/or neuropsychological results, these patients were studied with bilateral hippocampal depth electrodes and variable combinations of subdural strips for the purpose of localization of the epileptogenic focus. Only patients with (at least) bilateral hippocampal depth electrode study, MRI evidence of unilateral or bilateral hippocampal atrophy, hip- pocampal seizure onset and subsequent anteromedial tem- poral lobectomy were included in this study. Details of the initial evaluation, which included long-term surface EEG recordings, neuropsychological testing, intracarotid amo- barbital evaluation of speech and memory, and imaging studies, can be found elsewhere (Spencer, 1993). The inter-electrode distance of the 12 contacts of the hippocam- pal depth electrodes was 1 cm. Placement of electrodes and the anatomical position of each electrode within the hippo- campus was confirmed by post-implantation MRI. Antic- onvulsants were tapered and in some cases withdrawn until at least 3 spontaneous seizures were recorded. Patients were excluded if the depth electrode was not confirmed to be within the substance of the hippocampus or if EEG recordings were not available for review.
2.2. Surgery, pathology and outcome
All patients underwent anteromedial temporal lobectomy with amygdalo-hippocampectomy. Neuronal cell counts (Kim et al., 1995) and immunohistochemistry (de Lanerole et al., 1992) were performed on a segment of the most anterior portion of the hippocampal body. MTS was defined as greater that 50% cell neuronal loss in CA1 with evidence of reorganization of the dentate gyrus (sprouting of mossy fibers into the inner molecular layer of the dentate). Non- specific cell loss was defined as less than 30% cell loss in CA1. Control values for neuronal cell counts were obtained from patients without a history of epilepsy. Follow-up was determined at the latest clinic visit or through telephone conversation. Mean follow-up was 38 months (range 11- 72 months). Outcome was defined according to Engel's classification (Engel, 1987) when follow-up was at least 12 months.
matrix of 256 × 128, 195 or 256 and 3 mm slice thickness patients with a gap of 0.9-3 mm. Tl-weighted coronal images were obtained with the following parameters: 400-600/20/4 (TR/TE/NEX), 256 × 128 matrix 16 cm FOV and 5 mm contiguous slices (N = 12 patients) or 3D- SPGR imaging with 25/5/45/2 (TR/TE/flip angle/NEX), 16 cm FOV, 256 × 192 matrix and 3 mm thick contiguous slices (N = 15 patients).
Visual analysis was performed by a neuroradiologist blinded to clinical information, noting the presence or absence of atrophy and signal changes in defined segments of medial temporal structures (Bronen et al., 1991 Bronen et al., 1995). The hippocampus and amygdala were divided into 8 segments from anterior to posterior, based on the following landmarks: anterior pituitary (segment 1), poster- ior pituitary (segment 2), suprasellar cistern (segment 3), basilar artery (segment 4), interpeduncular cistern (segment 5), red nucleus (segment 6), 5 mm posterior to red nucleus (segment 7), and superior colliculus (segment 8). These landmarks showed a consistent relationship to the amygdala (segments 1-3), hippocampal pes (segments 4, 5), body (segments 6, 7) and tail (segment 8) (Bronen and Cheung, 1991). The location of the 12 contacts of the hippocampal depth electrode was then correlated with the same land- marks along the medial structures of the temporal lobe using the post-implantation MRI (see Figs. 1 and 2).
2.4. EEG analysis
Depth EEG recordings were obtained from bilateral 12 contact depth electrodes that were stereotactically inserted medially through occipital areas and directed toward medial temporal regions. On average, contacts 1-3 sampled the amygdala (segments 1-3), 4 -5 the pes (segments 4-5), 5 -7 the body (segments 5-7) and 7 -8 the tail (segments 7-8); contacts 9-12 were in the white matter of the occipital
MR landmark
Segment #
Electrode contact #
J ' "
AP PP SS BA I [P ~N [ PR SC [C PC
$1 $2 $3 $4 $5 $6 I $7 $8 $9 $10
1.9 2.5 3.0 3,7 4.3 5.2 I 5.7 6.7 7.1 7.7 I
2.3. Imaging technique
MRI measurements were obtained from coronal images using a 1.5 T magnet. Axial and coronal long TR images were obtained with the following parameters; 2000-3000/ 20-30.80-100/1-2 (TR/TE/NEX), 20-24 cm FOV, a
Fig. 1. Diagram of the anatomical divisions of the amygdala and hippo- campus based on the following landmarks: anterior pituitary (segment 1), posterior pituitary (segment 2), suprasellar cistern (segment 3), basilar artery (segment 4), interpeduncular cistern (segment 5), red nucleus (seg- ment 6), 5 mm posterior to red nucleus (segment 7), and superior colliculus (segment 8). Electrode contact number refers to the average electrode number, across all patients, per anatomical segment.
694 D. King et al. / Electroencephalography and clinical Neurophysiology 103 (1997) 692-697
contacts, a total of 623 segments were involved at ictal onset: segment 1 was involved in 64 seizures, segment 2 in 86 seizures, segment 3 in 121 seizures, segment 4 in 117 seizures, segment 5 in 111 seizures, segment 6 in 77 sei- zures, segment 7 in 33 seizures, and segment 8 in 14 sei- zures. Thus, of the 147 seizures analyzed, 44% involved segment 1, 58% involved segment 2, 82% segment 3, 80% segment 4, 76% segment 5, 52% segment 6, 22% seg- ment 7, and 10% segment 8.
Of the 21 seizures with focal onset, 47% arose from the amygdala/pes, 47% arose from the pes/anterior portion of the hippocampus and 5% from the anterior portion of the body.
3.2. Distribution of atrophic segments
Fig. 2. TI weighted sagittal MRI showing the location of the 12 contact depth electrode along the long axis of the hippocampus. Each contact was correlated with the anatomical landmarks used in Fig. 1.
lobe (see Figs. 1 and 2). Various combinations of subdural electrode strips were used to sample temporal, frontal and/or occipital neocortical regions. Seizures with simultaneous onset in the hippocampal depth electrode and any subdural strip were not included in the analysis. Ictal recordings were analyzed visually for the distribution of seizure onset along the long axis of the hippocampus. Ictal discharge was defined as a localized and sustained rhythmic discharge >3 Hz, distinct from the background activity and followed by clinical symptomatology. Focal onsets were defined as rhythmic discharges arising from one or two contacts along the hippocampal depth electrode; regional onsets involved 3 or more contacts. For the purpose of analysis, we included only clinical seizures arising from the hippocampus without considering subclinical seizures, interictal or preictal abnormalities. At least 3 clinical seizures were analyzed per patient (range 3-9).
3. Results
3.1. Seizure onset distribution
Twenty-seven patients met the criteria for inclusion: 25 with unilateral and two with bilateral hippocampal atrophy. One hundred and forty-seven seizures were analyzed: 21 (in 7 patients) showed focal onsets and 126 showed regional onsets. Twenty patients showed strictly unilateral medial temporal onsets and 7 showed bilateral independent onsets. Seizures starting in the non-atrophic hippocampus were not included in the analysis.
Given that most seizures involved at least 3 or more
Most patients had two or more atrophic segments; only two patients had a single segment with atrophy. Therefore, of the total 110 segments with atrophy, 5 (19%) involved segment 2, 8 (30%) involved segment 3, 12 (44%) involved segment 4, 19 (70%) involved segment 5, 23 (85%) involved segment 6, 22 (81%) involved segment 7, and 21 (78%) involved segment 8. No patient had atrophy of seg- ment 1.
3.3. Relationship between the distribution of segmental atrophy and site of ictal onset
Overall, 80% of ictal onsets involved the anterior half of the hippocampus and amygdala (segments 1-5); 40% of patients had atrophy of these 5 segments. Twenty percent of ictal onsets were in the posterior hippocampus, with 60% of patients showing atrophy of this area. In 42% of the patients the predominant or most frequent site of ictal onset was anterior to the atrophic segments; in 36% it was predominantly anterior to, but with some overlap with, the atrophic segments; in 22% it paralleled the atrophic seg- ments. Thus, most ictal onsets involved non-atrophic areas (see Fig. 3).
In the two patients with bilateral hippocampal atrophy, exclusive seizure onset occurred in the most atrophic hip- pocampus; ictal onsets were predominantly anterior, with minimal overlap in the atrophic hippocampus.
Of the 7 patients with focal onset, 3 had a strong correla- tion between the site of ictal onset and atrophic segments and 4 had a poor correlation. Given the low number of patients with focal onset, it is difficult to conclude that a focal onset correlates with segmental hippocampal atrophy.
3.4. Pathology and outcome
Pathological examination showed MTS in 22 patients and non-specific neuronal loss in 5 patients. All of the patients with focal ictal onsets showed MTS and all had grade I outcome. Of the 20 patients with regional onsets, 12 (7 with MTS) had grade I outcome, 3 (all with MTS) had
D. King et al. /Electroencephalography and clinical Neurophysiology 103 (1997) 692-697 695
== o= m
D.
[ ] atrophic segments
==segments wi h sz onse
$1 $2 $3 $4 $5 $6 $7 $8
Number of amygdela end hippocampa! segments
Fig. 3. Relationship between the distribution of the atrophic hippocampal and amygdala segments and the site of ictal onset. Results show the cumulative percentage of involvement of each segment.
grade II, one with MTS had grade III, 4 had insufficient follow-up.
4. Discuss ion
Our results suggest that the area of ictal onset in patients with medial temporal seizure onset as demonstrated by depth EEG does not parallel the segments with hippocampal atrophy; it is usually anterior or has minimal overlap, with such segments. In this group of patients, 78% of ictal onsets was anterior to, or had minimal overlap with the atrophic segments; only 22% of ictal onset was documented to over- lap the predominantly atrophic, posterior hippocampus.
The topographic distribution of hippocampal ictal onsets in patients with MTLE was first described by Babb et al. (1984). The UCLA group studied 12 patients with bilateral, stereotaxically placed, orthogonal depth electrodes in amyg- data, anterior, mid and posterior sites of the hippocampal pes and (para) hippocampal gyms. Following temporal lobectomy, correlations were made between the antero-pos- terior distribution of cell densities and zones of ictal onset. Patients were divided into 3 groups: (1) 3 patients with exclusive ictal focal onsets (defined as beginning in one electrode site) arising from the anterior segment of the hip- pocampus; (2) 4 patients with partial focal onsets involving, although not restricted to, the anterior hippocampus; and (3) 4 patients with regional onsets (defined as involving two or more electrode sites simultaneously). When cell densities were compared among the 3 groups of patients, those with (exclusive) anterior focal onsets had significant greater loss of hippocampal pyramidal cells in the anterior compared to their posterior hippocampus, and compared to the anterior hippocampus of the control group. When patients with par- tially anterior focal onsets were compared with patients with regional onsets, anterior cell damage was significantly greater in the former group. Although mean cell densities
of the anterior hippocampus were not significantly different among the 3 groups, posterior hippocampal cell densities were significantly reduced in the partially focal and regional onset groups relative to the exclusively anterior onset group. Thus, the topographic distribution of ictal onsets duplicated the loss of pyramidal cells along the axis of the hippocam- pus.
Babb et al. (1984) used cell counts to quantify degree of cell loss. Even though the use of cell counts provides an accurate estimate of hippocampal damage, the analysis was restricted to only one portion of the most anterior and pos- terior hippocampal segments, thus limiting the spatial ana- lysis of the degree of cell loss along the long axis of the hippocampus. Our MRI protocol allows for the sampling of the whole hippocampal volume; qualitative detection of atrophy has been shown to correlate with the degree of cell loss (Bronen et al., 1991; Kuzniecky et al., 1996). Furthermore, Babb et al. (1984) used 2-3 orthogonal depth electrodes sampling the amygdala and pes. The rou- tine use of a longitudinal depth electrode, providing us with more complete and continuous spatial sampling along the long axis of the hippocampus, might contribute to the dis- crepancy of our findings with theirs.
In the study with the closest parallel to ours, Baulac et al. (1994) studied 18 patients with MTLE using quantitative MRI volumetry of the hippocampus and correlated the find- ings with depth electrode recordings. At least 3 depth elec- trodes were implanted stereotaxically, orthogonal to the hippocampal axis, sampling (1) the amygdala and hippo- campal pes, (2) the anterior and (3) posterior segments of the hippocampus. MRI images were obtained through sec- tions matching the position of the electrodes. Diffuse atro- phy of the hippocampus was present in 9 out of 18 (50%) patients with hippocampal sclerosis; 39% of patients had segmental atrophy. Overall, the anterior portion of the body and pes of the hippocampus was normal in 17% and in 22% the posterior portion was normal. Only two patients
696 D. King et al. / Electroencephalography and clinical Neurophysiology 103 (1997) 692-697
had restricted damage confined to the amygdala and pes without involvement of the body (although, in both, the posterior portion of the body was atrophic). In 13 patients, several segments showed ictal onsets simultaneously, with predominance in the amygdala and hippocampal head; 4 had posterior body onsets. In patients with onsets restricted to a single segment, the amygdala and hippocampal head were involved 5 times, the anterior portion of the hippocam- pus 3 times and the posterior portion one time. When cor- related with MRI evidence of segmental loss, ictal onsets occurred in the area of atrophy in 71% of cases with amyg- dala/pes atrophy, in 64% of cases with anterior hippocampal body atrophy and in 31% of cases with posterior atrophy. Conversely, 50% of segments without atrophy gave rise to seizure onsets. There were no differences in the degree of atrophy among segments with and without ictal onsets. They concluded that only a weak correlation between inten- sity of hippocampal damage and hyperexcitability is pre- sent.
In agreement with the results of Baulac et al. and with the additional advantage of longitudinal depth electrodes which improve the extent of sampling of the hippocampal forma- tion when compared with orthogonally placed depth elec- trodes, our series demonstrates that the area responsible for the majority of seizure onsets is anterior to, or includes only a small portion of, the hippocampal segments with atrophic changes. In a prior study we reported that only about 20% of patients with unilateral temporal lobe epilepsy had an abnormal electrical focus identified in the posterior hippo- campus (Spencer et al., 1982).
The distribution of atrophy in these patients is consistent with prior reports, both with quantitative and qualitative MRI analysis. While we did not obtain quantitative seg- mental volumes of the anterior half of the hippocampal, results from several series suggest that the brunt of atro- phic damage usually involves mid to posterior portions of the hippocampus. Bronen et al. (1995) reported that in patients with MTLE and hippocampal sclerosis, 37% have MRI changes that involved the entire structure and an additional 16% had MRI changes throughout the hippocam- pus with extension to the amygdala; non-uniform involve- ment of the hippocampus occurred in 44%; in 9% the anterior segments (amygdala-pes-body) were involved exclusively and in 14% the posterior segments (body-tail) were affected exclusively. No patient had findings affecting the anterior limbic structures (amygdala or pes) without affecting also the hippocampal body. Similarly, Kuzniecky et al. (1996) studied the patterns of mesial temporal atrophy by qualitative segmentation MRI analysis in 47 patients with pathologically proven MTS. All patients had hippo- campus body atrophy: 70% had in addition hippocampus tail atrophy, 68% with hippocampus head and 23% with amygdala atrophy; 10% had focal segmented hippocampal body atrophy. Kim et al. (1994) have also shown that seg- mental MR measurements of the body of the hippocampus are as accurate as measurements of the whole hippocampus
for lateralizing the epileptogenic focus in patients with tem- poral lobe epilepsy.
Electrophysiological data from animal models support the concept of regional differences in the epileptogenicity of the hippocampal formation. In the rat hippocampus, the temporal pole is more vulnerable to the development of kindling than more posterior areas. Gilbert et al. (1985) compared the field potentials evoked in the CAI subfield by stimulation of Shaffer collateral in slices from ventral and dorsal hippocampus of rats. A significantly greater ten- dency to generate burst responses was found in ventral com- pared to dorsal slices. Work by Cavazos et al. (1992) has demonstrated variation in the synaptic transmission of path- ways in hippocampal slices prepared from normal and kindled rats. By using Timm staining to characterize the septotemporal variation of reorganization of mossy fiber pathways, more prominent projections onto the supragranu- lar region (of the dentate gyrus) were demonstrated in the temporal pole versus more posterior regions. These results suggest that the anterior segments of the hippocampus, at least in the rat, have a greater propensity to amplify and sustain epileptiform discharges than posterior segments.
Our results indicate that most ictal onsets arise in hippo- campal segments anterior to the areas with the greatest degrees of atrophy, corresponding to segments with lesser degrees of cell loss. A lesser degree of cell loss in conjunc- tion with reorganization of the dentate gyrus may provide the circuitry for initiation and propagation of spontaneous synchronized epileptiform activity. The profound degree of pyramidal cell loss in more atrophic segments might exceed the ability to initiate or sustain ictal discharges. Alterna- tively, the process of seizure generation may involve mechanisms independent of the observed hippocampal atro- phy and the cell loss it represents.
References
Babb, T.L., Lieb, J.P., Brown, W.J., Pretorius, J. and Crandall, P.H. Dis- tribution of pyramidal cell density and hyperexcitability in the epileptic human hippocampal formation. Epilepsia, 1984, 25(6): 721-728.
Baulac, M., Saint-Hilare, J.M., Adam, C., Martinez, M., Fontaine, S. and Laplane, D. Correlation's between magnetic resonance imaging-based hippocampal sclerosis and depth electrode investigation in epilepsy of the mesiotemporal lobe. Epilepsia, 1994, 35(5): 1045-1053.
Bronen, R.A. and Cheung, G. Relationship of hippocampal and anlygdala to coronal MRI landmarks. Magn. Reson. Imaging, 1991, 9: 449-457.
Bronen, R.A., Cheung, G., Charles, J.T., Kim, J.H., Spencer, D.D., Spen- cer, S.S., Sze, G. and McCarthy, G. Imaging findings in hippocampal sclerosis: correlation with pathology. Am. J. Neuroradiol., 1991, 12: 933-940.
Bronen, R.A., Fulbright, R.K., Kim, J.H., Spencer, S.S., Spencer, D.D. and Al-Rodhan, N.R.F. Regional distribution of MR findings in hippocam- pal sclerosis. Am. J. Neuroradiol., 1995, 16:1193-1200.
Cascino, G.D., Jack, Jr., C.R., Parisi, J.E., Sharbrough, F.W., Hirschorn, K.A., Meyer, F.B., Marsh, W.R. and O'Brien, P.C. Magnetic resonance imaging-based volume studies in temporal lobe epilepsy: pathological correlation's. Ann. Neurol., 1991, 30: 31-36.
Cavazos, J., Golarai, G. and Sutula, T.P. Septotemporal variation of the
D. King et al. /Electroencephalography and clinical Neurophysiology 103 (1997) 692-697 697
supragranular projection of the mossy fiber pathway in the dentate gyrus of normal and kindled rats. Hippocampus, 1992, 2 (4): 363-72.
Cook, M.J., Fish, D.R., Shorvon, S.D., Straughan, K. and Stevens, J.M. Hippocampal volumetric and morphometric studies in frontal and tem- poral lobe epilepsy. Brain, 1992, 115: 1001-1015.
de Lanerole, N.C., Brines, M.L., Kim, J.H., Williamson, A., Philips, M. and Spencer, D.D. Neurochemical remodeling of the hippocampus in human temporal lobe epilepsy. Epilepsy Res., 1992, 9 (Suppl. l): 205- 219.
Engel, Jr., J., Outcome with respect to epileptic seizures, In Engel, Jr., J. (Ed.), Surgical Treatment of the Epilepsies. Raven Press, New York, 1987, pp. 553-571.
Gilbert, M., Rancine, R.J. and Smith, G.L. Epileptiform burst responses in ventral vs. dorsal hippocampal slices. Brain Res., 1985, 361: 389-391.
Jack, Jr., C.R., Sharbrough, F.W., Twomey, C.K., Cascino, G.D., Hirsch- corn, K.A., Marsh, W.R., Zinsmeister, A.R. and Scheithauer, B. Tem- poral lobe seizures: lateralization with MR volume measurements of the hippocampal formation. Radiology, 1990, 175: 423-429.
Kim, J.H., Tien, R.D., Felsberg, G.J., Osumi, A.K. and Lee, N. MR mea- surements of the bippocampus for lateralization of temporal lobe epi-
lepsy: value of measurements of the body vs. the whole structure. Am. J. Radiol., 1994; 163: 1453-1457.
Kim, J.H., Kraemer, D.L. and Spencer, D.D. The neuropathology of epi- lepsy. In: A. Hopkins, S. Shorvon and G. Cascino (Eds.), Epilepsy, 2nd edn. Chapman and Hall, London, 1995, pp. 243-265.
Kuzniecky, R.I., Burgard, S., Bilir, E., Morawetz, R., Gillianl, F., Faught, E., Black, L. and Palmer, C. Qualitative MRI segmentation in mesial temporal sclerosis: clinical correlations. Epilepsia, 1996, 37 (5): 433- 439.
Lencz, T., McCarthy, G., Bronen, R.A., Scott, T.M., Inserni, J.A., Sass, K.J., Novelly, R.A., Kim, J.H. and Spencer, D.D. Quantitative magnetic resonance imaging in temporal lobe epilepsy: relationship to neuro- pathology and neuropsychological function. Ann. Neurol., 1992, 31: 629-637.
Spencer, S.S., Spencer, D.D., Williamson, P.D. and Mattson, R.H. The localized value of depth electroencephalography in 32 patients with refractory epilepsy. Ann. Neurol., 1982, 12: 248-253.
Spencer, S.S. Temporal lobectomy: selection of candidates. In: E. Wyllie (Ed.), The Treatment of Epilepsy: Principles and Practice. Lea and Febiger, Malvem, 1993, pp. 1062-1074.
