Search

Summary: Purpose: Medically intractable temporal lobe epilepsy (TLE) due to hippocampal sclerosis (HS), with or without cortical dysplasia (CD), is associated with atrophy of the hippocampal formation and regional fluorodeoxyglucose positron-emission tomography (FDG-PET) hypometabolism. The relation between areas of functional and structural abnormalities is not well understood. We investigate the relation between FDG-PET metabolism and temporal lobe (TL) and hippocampal atrophy in patients with histologically proven isolated HS and HS associated with CD. Methods: Twenty-three patients underwent en bloc resection of the mesial and anterolateral neocortical structures. Ten patients were diagnosed with isolated HS; 13 patients had associated microscopic CD. Temporal lobe volumes (TLVs) and hippocampal volumes were measured. Magnetic resonance imaging (MRI) and PET were co-registered, and regions of interest (ROIs) determined as gray matter of the mesial, lateral, and anterior temporal lobe. Results: All patients (HS with or without CD) had significant ipsilateral PET hypometabolism in all three regions studied (p < 0.0001). In patients with isolated HS, the most prominent hypometabolism was in the anterior and mesial temporal lobe, whereas in dual pathology, it was in the lateral temporal lobe. TLVs and hippocampal volumes were significantly smaller on the epileptogenic side (p < 0.05). The PET asymmetries ipsilateral/contralateral to the epileptogenic zone and TLV asymmetries correlated significantly for the anterior and lateral temporal lobes (p < 0.05) in the HS+CD group, but not in the isolated HS group. Mesial temporal hypometabolism was not significantly different between the two groups. Conclusions: Temporal neocortical microscopic CD with concurrent HS is associated with more prominent lateral temporal metabolic dysfunction compared with isolated HS in TL atrophy. Further studies are needed to confirm these findings and correlate the PET hypometabolic patterns with outcome data in patients operated on for HS with or without CD.

The EEG characteristics of isolated hippocampal sclerosis (HS) and HS associated with other types of temporal lobe pathology are not well defined. The pathologic substrate may be an important variable in determining seizure-free outcome. The objective of this study was to define the distribution of epileptiform discharges in patients with HS and HS associated with microscopic dysplasia, and to examine their relationship with hippocampal atrophy and cell loss. Thirty-four patients (15 women and 19 men; mean age, 30.6 +/- 11.2 years), all with good outcomes after temporal lobectomy (Engel classes I and II), were included. The characteristics studied were frequency and distribution of spikes, MRI-based hippocampal volume ratios, and quantitative hippocampal cell density in various subregions. The isolated HS group showed a trend to a higher percentage of epileptiform discharges maximal at the anterior temporal electrodes (89.87 +/- 17.0%; 79.5 +/- 28.2% in the dual-pathology group). The isolated HS group had, on average, significantly more cell loss (P < 0.001). There was a significant negative correlation between the amount of cell loss in the CA1 area and both anterior temporal spikes and hippocampal ratios (P < 0.05). Isolated HS and dual pathology show minimal differences in interictal spike distribution and frequency. More widespread spike distributions in severe isolated HS compared with patients with less cell loss is probably the result of less organized limbic circuitry.

Objective: To characterize the clinical, EEG, MRI, and histopathologic features and explore seizure outcome in pediatric candidates for epilepsy surgery who have temporal lobe epilepsy (TLE) caused by hippocampal sclerosis (HS). Methods: The authors studied 17 children (4 to 12 years of age) and 17 adolescents (13 to 20 years of age) who had anteromesial temporal resection between 1990 and 1998. Results: All patients had seizures characterized by decreased awareness and responsiveness. Automatisms were typically mild to moderate in children and moderate to marked in adolescents. Among adolescents, interictal spikes were almost exclusively unilateral anterior temporal, as opposed to children in whom anterior temporal spikes were associated with mid/posterior temporal, bilateral temporal, extratemporal, or generalized spikes in 60% of cases. MRI showed hippocampal sclerosis on the side of EEG seizure onset in all patients. Fifty-four percent of children and 56% of adolescents had significant asymmetry of total hippocampal volumes, whereas the remaining patients had only focal atrophy of the hippocampal head or body. Subtle MRI abnormalities of ipsilateral temporal neocortex were seen in all children and 60% of adolescents studied with FLAIR images. On histopathology, there was an unexpectedly high frequency of dual pathology with mild to moderate cortical dysplasia as well as HS, seen in 79% of children and adolescents. Seventy-eight percent of patients were free of seizures at follow-up (mean, 2.6 years). A tendency for lower seizure-free outcome was observed in patients with bilateral temporal interictal sharp waves or bilateral HS on MRI. The presence of dual pathology did not portend poor postsurgical outcome. Conclusions: TLE caused by HS similar to those in adults were seen in children as young as 4 years of age. Focal hippocampal atrophy seen on MRI often was not reflected in total hippocampal volumetry. Children may have an especially high frequency of dual pathology, with mild to moderate cortical dysplasia as well as HS, and MRI usually, but not always, predicts this finding. Postsurgical seizure outcome is similar to that in adult series.

Purpose: The cellular mechanisms that may contribute to epilepsy in resected human cortical dysplasia (CD) were compared with the in utero radiated rat CD model. In human and rat focal hippocampal epilepsy, postsynaptic N-methyl-D-aspartate receptors are up-regulated and presynaptic axon collaterals hyperinnervate them. We hypothesized that in both human and rat CD: (a) the N-methyl-D-aspartate receptor subunits NR1 and NR2A/B would be increased and coassembled, and (b) aberrant axons would be in regions of CD. Methods: Tests for presynaptic and postsynaptic changes in human and rat CD included the following: (a) cytology, (b) immunocytochemistry, (c) coimmunoprecipitation, (d) double-labeled immunofluorescence, and (e) Timm histochemistry of hippocampal mossy fibers. Within-patient comparisons were made between epileptic tissue, identified by subdural electroencephalographic seizure onsets, and nonepileptic tissue remote from the focus but within the therapeutic resection. Rats were radiated at embryonic day 17, and offspring were studied postnatally. Statistical comparisons were made against normal rats matched for age and tissue processing. Results: In focal CD patients, NR2A/B subunits and their coassemblies with NR1 were increased significantly more than for the remote nonepileptic cortex. Confocal microscopy showed that NR1-NR2A/B colabeled single dysplastic neurons in both human and rat. In CD rats, mossy fibers innervated the anomalously oriented hippocampal neurons. Conclusions: Human epileptic CD exhibits a spectrum of abnormal cell orientations and laminations that must require plastic axodendritic changes during development. These altered circuits and receptors could account for the seizures and cognitive deficits found in patients with CD. The radiated rat CD model with cortical dyslaminations and NR2A/B subunit increases would allow the development and testing of drugs targeted at only the NR2A/B subunit or at decoupling the NR1-NR2 coassembly, which could provide a specific antiepileptic drug for dysplastic circuits without inducing general depression of all brain neurons.

Summary: Purpose: This study was designed to quantify the relation between expressions of NMDA receptor (NMDAR) subunits (1 and 2A/B) and the epileptogenicity in human focal cortical dysplasia. Methods: Immunoblotting and immunoprecipitation were used to quantify these receptor subunits in tissue resected from EEG-verified epileptic and distal nonepileptic frontal cortical areas in each of three patients as determined by chronic subdural electrode recordings. In each patient, adjacent sections were immunostained to verify that the numbers of dysplastic neurons were greater in epileptic than in nonepileptic cortex. Results: In all patients, NMDAR2A/B expressions and their coassemblies with NMDAR1 were increased in epileptic dysplastic cortex compared with the relatively normal appearing nonepileptic cortex. For all three patients, there were no significant differences in NMDAR1 protein expressions between the two EEG groups. Conclusions: These results suggest that increased NMDAR1-NMDAR2A/B coassembly contributes to hyperexcitability in dysplastic cortical neurons and focal seizure onsets.

Neurofilament proteins (NFPs), the cytoskeletal proteins that are essential for axogenesis and maintenance of neuron shape in the nervous system, were studied for their spatial distributions at nine postnatal days (PN 3, 5, 7, 10, 14, 17, 21, 28, and 120). Simultaneously non-phosphorylated (SMI-32; 150/200 kDa; Sternberger) and phosphorylated (SMI-31; 200 kDa) NFP immunoreactivity in the entire developing rat hippocampus was studied, quantified, and compared to that of mossy fiber (MF) axons and terminals using Neo-Timm's histochemistry, the most selective, sensitive, and reproducible technique. Differential developmental expressions were observed between the two NFP states. SMI-32 was initially expressed on PN 3 only in the perikarya of pyramidal neurons in CA3. As early as PN 5, SMI-31 appeared in the MF pathway, in parallel to the growth of MF axons. By contrast, SMI-32 did not appear at any age in the MF pathway, including the MF terminal zone of stratum lucidum. At PN 14, the distribution of both NFPs in the MF system (MFs and their target neurons, i.e., CA3/CA4 pyramidal neurons and hilar neurons) was nearly complete; however, the peak densities of SMI-32 and SMI-31 were later at PN 21 and statistically equal to the most adult level (PN 120). The temporal regulation and maximal levels of SMI-32 and SMI-31 expressions on MF target neurons (CA3: SMI-32) and in the MF terminal zone (stratum lucidum: SMI-31) were nearly parallel to the progressive and rapid PN growth of the MF axons and terminals occurring between PN 14 and PN 17, suggesting that the mechanisms for maturation of MF synaptogenesis occur after PN 17.

Kainic acid (KA)-induced status epilepticus (SE) in adult rats results in extensive neuronal damage throughout the limbic system and the loss of selectively vulnerable neuronal populations, particularly CA3 neurons. We investigated the effects of a short episode of seizure activity on neuronal death elicited by a subsequent prolonged SE episode. A short episode of seizure activity was produced by sub-cutaneous (s.c.) injection of KA followed after 1 h by pentobarbital administration. Twenty-four hours later, KA was administered again, and animals were sacrificed 3 days later. Neuronal damage was estimated by visual analysis of neuronal density. Our results show that a short episode of seizure activity did not produce neuronal damage but almost completely protected vulnerable neurons from KA-induced neuronal damage. These results extend to epileptic tolerance the notion of tolerance previously described in the case of ischemia.

Following intrahippocampal (hilar) kainic acid (KA) lesions in rats, NMDAR2A/B receptor proteins are upregulated significantly in the inner molecular layer (IML) of the dentate gyrus by post-injection day 5. By contrast, the aberrant mossy fibers which reinnervate the IML remained in the subgranular zone before sprouting and synapsing in the IML, which occurs at approximately post-KA day 17. For 40 days thereafter, this mossy fiber ingrowth progressed, while the increased NMDAR2A/B (receptors) immunoreactivity remained at the same densities. These results suggest that new NMDAR2A/B proteins in granule cell dendrites are limited to the IML, which is the eventual site for MF hyperinnervation, neosynaptogenesis, and recurrent synaptic hyperexcitability.

Kainic acid (KA) lesions of the rat hilus model hippocampal sclerosis and temporal lobe epilepsy. Unilateral hilar cell loss denervates the associational afferents normally projecting to the inner molecular layer (IML) granule cell dendrites, followed by ipsilateral mossy fiber (MF) sprouting. Hilar neurons also project through the hippocampal commissure to the contralateral IML. This study compared densities of IML MF sprouting following unilateral versus bilateral low dose KA lesions, using Neo-Timm stain 30 days later. Unilateral KA (0.4 microg) caused only dense ipsilateral MF sprouting. Bilateral lesions with lower doses of KA (0.1 with 0.2 or 0.3 microg) induced dense bilateral MF sprouting. However, the same low doses of KA injected unilaterally did not induce significant sprouting ipsilaterally or contralaterally. These data show that denervations of both associational and commissural afferents to the same IML dendritic zones of granule cells induce non-linear, additive bilateral MF neoinnervations.

This study determined whether there were differences in hippocampal neuron loss and synaptic plasticity by comparing rats with spontaneous epilepsy after limbic status epilepticus and animals with a similar frequency of kindled seizures. At the University of Virginia, Sprague-Dawley rats were implanted with bilateral ventral hippocampal electrodes and treated as follows; no stimulation (electrode controls; n=5): hippocampal stimulation without status (stimulation controls; n=5); and limbic status from continuous hippocampal stimulation (n=12). The limbic status group were electrographically monitored for a minimum of four weeks. Four rats had no recorded chronic seizures (status controls), and all three control groups showed no differences in hippocampal pathology and were therefore incorporated into a single group (controls). Eight limbic status animals eventually developed chronic epilepsy (spontaneous seizures) and an additional eight rats were kindled to a similar number and frequency of stage 5 seizures (kindled) as the spontaneous seizures group. At the University of California (UCLA) the hippocampi were processed for: (i) Niss1 stain for densitometric neuron counts; (ii) neo-Timm's histochemistry for mossy fiber sprouting; and (iii) immunocytochemical staining for glutamate decarboxylase, N-methyl-D-aspartate receptor subunit 2, AMPA receptor subunit 1 and the GABA(A) receptor. In the fascia dentata inner and outer molecular layers the neo-Timm's stain and immunoreactivity was quantified as gray values using computer image analysis techniques. Statistically significant results (P

This study determined in temporal lobe epilepsy patients if there were correlations among hippocampal granule cell expression of neurotrophin mRNAs, aberrant supragranular mossy fiber sprouting, and neuron losses. Consecutive surgically resected hippocampi (n = 9) and comparison tissue from autopsies (n = 3) were studied for: 1. Granule cell mRNA levels using in situ hybridization for brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), and neurotrophin-3 (NT-3); 2. neo-Timm supragranular mossy fiber sprouting; and 3. Ammon's horn neuron densities. Clinically, patients were classified into those with hippocampal sclerosis (HS; n = 7) and non-HS cases (i.e., mass lesions and autopsies; n = 5). Results showed that compared to non-HS cases, HS patients showed increased granule cell mRNA levels for BDNF, NGF, and NT-3 (p = 0.035, p = 0.04, p = 0.045 respectively; one-tail directional test). Moreover, granule cell BDNF mRNA levels correlated inversely with Ammon's horn neuron densities (p = 0.02) and correlated positively with greater supragranular mossy fiber sprouting (p = 0.02). NGF mRNA levels correlated inversely with Ammon's horn neuron densities (p = 0.02), and NT-3 mRNA levels correlated inversely with age at surgery (p = 0.04) and correlated positively with greater mossy fiber sprouting (p = 0.026). These results indicate in the chronically damaged human hippocampus that granule cells express neurotrophin mRNAs, and mRNA levels correlate with either hippocampal neuron losses or aberrant supragranular mossy fiber sprouting. These data support the hypothesis that in the epileptic human hippocampus, there may be pathophysiologic associations among mossy fiber synaptic plasticity, hippocampal neuron damage, and granule cell mRNA neurotrophin levels.

Summary In this study, we determined whether childhood seizures were associated with hippocampal neuron loss and mossy fibre synaptic reorganization and if hippocampal sclerosis evolved from longer seizure histories. Children undergoing surgical treatment for catastrophic epilepsy were grouped into the following pathology categories: (i) those with generalized seizures and extra-hippocampal congenital pathologies (i.e. prenatal cortical dysplasia; n = 17); (ii) cases of generalized seizures and extra-hippocampal acquired lesions (i.e. postnatal ischaemic injuries and encephalitis: n = 7); (iii) children with complex partial hippocampal epilepsy (n=4). Further, to determine whether the epileptogenic location influenced hippocampal pathology, the seizure focus was classified as (i) hippocampal, (ii) temporal (n = 13) or (iii) extra-temporal (n = 11). Surgical and autopsy (n = 23) hippocampi were studied for (i) fascia dentata (FD) and Ammon's horn (AH) neuron densities: (ii) thickness, height or length of the FD molecular layer, stratum granulosum (SG) and stratum pyramidale: and (iii) grey value (GV) densities of supragranular neo-Timm's staining. Statistically significant results (P

This study determined differences of fascia dentata (FD) peptide and inhibitory neuroanatomy between patients with epileptogenic hippocampal sclerosis (HS), those with extrahippocampal seizure pathologies, and autopsy comparisons. Surgically treated temporal lobe epilepsy patients were clinically classified into two pathogenic categories: (1) HS with focal mesial temporal neuroimaging and histories of initial precipitating injuries to the brain (n = 18) and (2) non-HS patients with extrahippocampal mass lesions or idiopathic seizures (i.e., without lesions or HS; mass lesion/idiopathic; n = 9). The hippocampal sections were studied for (1) granule cell, hilar, CA4, and CA3 neuron densities; (2) hilar densities and the percentage of neurons immunoreactive (IR) for neuropeptide Y (NPY), somatostatin (SS), and glutamate decarboxylase (GAD); (3) densities of GAD neurons in the lower granule cell and infragranular zone (basket-like cells); (4) the semiquantitative pattern of IR peptides/GAD FD molecular layer axon sprouting; (5) IR gray values (GV) of the FD molecular layers; and (6) the thickness of the supragranular molecular layer. Results showed the following. (1) Compared to autopsies, both HS and mass lesion/idiopathic patients showed less granule cell and CA3 neuron densities, but there were no statistical differences between the latter two pathogenic categories. (2) By contrast, compared to autopsies and mass lesion/idiopathic cases, HS patients showed less hilar and CA4 neuron densities, and there were no differences between autopsies and mass lesion/idiopathic. (3) Compared to autopsies, the NPY and SS hilar neuron densities in HS patients, but not mass lesion/idiopathic cases, were less. (4) Compared to autopsies, the hilar GAD neuron densities for HS and mass lesion/idiopathic patients were not less. (5) In HS patients the averaged percentages of hilar SS neurons were less than autopsies, and no other differences of IR hilar percentages were found. (6) The densities of GAD basket-like neurons and the thickness of the supragranular molecular layer were not different between any combination of pathogenic categories and autopsies. (7) By semiquantitative visual assessments, peptides/GAD axon sprouting into the FD was greater in HS compared to mass lesion/idiopathic or autopsies. (8) Compared to mass lesion/idiopathic cases, in HS NPY outer molecular layer GVs were lower, SS GVs were not different, and GAD inner molecular layer GVs were higher. (9) Analyses comparing the two pathogenic categories and neuron densities with peptides/GAD axon sprouting found six comparisons that correlated sprouting with hilar and CA4 neuron losses, and four comparisons showing greater sprouting in HS compared to mass lesion/idiopathic.(ABSTRACT TRUNCATED AT 400 WORDS)

Hyppocampal epilepsy is a recently defined syndrome occurring in 65% of all temporal lobe epilepsies as defined by: 1) electrographic (EEG) onset in the hippocampus (HC) prior to EEG seizures elsewhere, 2) post-resection hippocampal sclerosis and mossy fiber synaptic reorganizations and 3) relief of typical complex partial seizures after surgical resection of the hyppocampus. We used intrahippocampal kainic acid injections V2 in rats at different developmental ages (postnatal 7 through adult) to develop long term spontaneous HC EEG spikes, EEG seizures, and behavioral seizures. Split-screen video/EEG monitoring demonstrated that this intrahippocampal kainic acid model produced progressive development of: 1) ipsilateral interictal spikes, 2) later polyspike complexes, 3) bilaterally-asynchronous EEG spiking, 4) unilateral HC EEG seizure onsets with occasional secondarily generalized spread to apposite HC and motor cortex to elicit complex partial seizures, and 5) in all seizing rats there was mossy fiber synaptic reorganization, even when injected at age 7 days. These results indicate that the intrahippocampal kainic acid injection model is similar to human hippocampal epilepsy.

A retrospective study was carried out to determine whether a prior cerebral injury or medical illness was associated with hippocampal sclerosis in intractable, surgically treated temporal lobe epilepsy (TLE), or whether there was evidence for progressive hippocampal neuron damage from repeated seizures. Temporal lobe epilepsy patients (n = 162) from one epilepsy centre were retrospectively and blindly catalogued into groups based on the presence or absence of an initial precipitating injury (IPI) and whether, when an IPI was present, it had involved seizures (independent variables). Patients were catalogued into four groups: (i) non-seizure IPIs (Group A; n = 54); (ii) IPIs with a prolonged seizure (Group B; n = 66); (iii) IPIs with repetitive non-prolonged seizures (Group C; n = 20); (iv) or no IPIs and idiopathic TLE (Group D; n = 22). The dependent variables were: the differences in the time course of clinical seizures, and quantified hippocampal neuron counts and seizure outcomes. Statistically significant (ANOVA at least P < 0.05) results showed the following. (i) Patients with IPIs (Groups A, B and C) had hippocampal sclerosis, while those with idiopathic TLE (Group D) showed fewer neuron losses and worse post-resection seizure relief. (ii) Patients with non-seizure IPIs (Group A) were on average older at injury; had a longer latent period; showed less neuron losses in Ammon's horn, CA1 and prosubiculum than seizure associated IPIs (Groups B and/or C). (iii) Initial precipitating injury patients with repetitive non-prolonged seizures (Group C) showed the shortest latent period, earliest age of TLE onset, and less CA2 damage than the other IPI groups. Other findings that were statistically significant by analysis of covariance along with the IPI category included the following. (i) CA1 (P = 0.0097) and prosubiculum (P = 0.0089) neuron losses were greater in patients when their TLE was longer than 22 years. (ii) IPIs after age 4 years were associated with latent periods shorter than 10 years compared with variable and longer latent periods of IPIs before age 4 years (P = 0.0015). These results indicate that in surgically treated TLE, hippocampal sclerosis and good seizure outcomes are associated with IPIs. Most of the hippocampal damage found at surgery and the clinical time course of the habitual TLE are influenced by the pathogenic IPI mechanism. However, some secondary neuron losses were associated with longer TLE seizure histories.(ABSTRACT TRUNCATED AT 400 WORDS)

✓ Rarely are both sides of the hippocampus available for pathological study in a patient with intractable temporal lobe epilepsy (TLE). The authors report a patient with TLE investigated with bilateral depth electrodes who had an episode of postictal psychosis. The patient died 4 weeks after temporal lobectomy of unknown reasons, despite complete postmortem examination and clinical evidence of postsurgery seizure control. Pathological examination of surgical and autopsy hippocampal specimens found bilateral asymmetric neuron losses. However, only the resected epileptogenic hippocampus showed the profile of neuron loss typical of mesial temporal sclerosis (MTS) and abnormal mossy fiber synaptic reorganization. Quantitative depth electroencephalographic (EEG) analysis of the postictal psychotic event showed that it was not associated with a cluster of seizures, increased postictal depth EEG spike activity, or insufficient antiepileptic medication. These results support the hypothesis that ipsilateral hippocampal epileptogenesis is associated with MTS and mossy fiber sprouting. The results also suggest that the etiology of postictal psychosis in this patient was initiated by an ictal event and the behavior apparently depended on seizure propagation outside the hippocampus. The relevance of these two findings to the literature is discussed.

We used the pilocarpine model of chronic spontaneous recurrent seizures to evaluate the time course of supragranular dentate sprouting and to assess the relation between several changes that occur in epileptic tissue with different behavioral manifestations of this experimental model of temporal lobe epilepsy. Pilocarpine-induced status epilepticus (SE) invariably led to cell loss in the hilus of the dentate gyrus (DG) and to spontaneous recurrent seizures. Cell loss was often also noted in the DG and in hippocampal subfields CA1 and CA3. The seizures began to appear at a mean of 15 days after SE induction (silent period), recurred at variable frequencies for each animal, and lasted for as long as the animals were allowed to survive (325 days). The granule cell layer of the DG was dispersed in epileptic animals, and neo-Timm stains showed supra- and intragranular mossy fiber sprouting. Supragranular mossy fiber sprouting and dentate granule cell dispersion began to appear early after SE (as early as 4 and 9 days, respectively) and reached a plateau by 100 days. Animals with a greater degree of cell loss in hippocampal field CA3 showed later onset of chronic epilepsy (r = 0.83, p < 0.0005), suggesting that CA3 represents one of the routes for seizure spread. These results demonstrate that the pilocarpine model of chronic seizures replicates several of the features of human temporal lobe epilepsy (hippocampal cell loss, supra- and intragranular mossy fiber sprouting, dentate granule cell dispersion, spontaneous recurrent seizures) and that it may be a useful model for studying this human condition. The results also suggest that even though a certain amount of cell loss in specific areas may be essential for chronic seizures to occur, excessive cell loss may hinder epileptogenesis.

Previous histological and immunocytochemical studies suggest that reorganization of the dentate granule cell axons, the mossy fibers, can occur in epileptic human hippocampus (Sutula et al., 1989; Houser et al., 1990; Babb et al., 1991) and in animal models of epilepsy (Tauck and Nadler, 1985; Sutula et al., 1988; Cronin et al., 1992). However, neuroanatomical analyses of the trajectory and morphology of reorganized axons are not yet available. The present study was conducted to investigate single dentate granule cell axonal systems in human epileptic hippocampus. Individual mossy fibers were directly visualized by injecting a tracer (biocytin or Lucifer yellow) intracellularly in hippocampal slices prepared from temporal lobes that were surgically removed from patients for treatment of intractable epilepsy. Two major arborization patterns were identified: (1) the parent axons extended to and coursed through the hilus toward CA3, leaving collaterals along their paths in the hilus (N = 19 neurons); (2) in addition to the aforementioned axonal system, collateral(s) branched from the parent axon near the soma and projected to the granule cell layer and molecular layer, forming an aberrant axonal pathway (N = 9 neurons). These aberrant collaterals bore large boutons similar to those of the hilar axons and formed extensive plexuses in the granule cell layer and/or in the molecular layer. The summed length of collaterals in the granular/molecular layers was 1110.8 microns on average, which was one-fourth of the total summed length of the mossy fibers (3698.5 microns on average). The size of the somata in neurons that had aberrant collaterals was significantly larger than that of neurons without such collaterals (p < 0.025). In four cases, filopodium- like fine processes were present near the axon hillock and proximal parts of the parent axon, suggesting that the aberrant collateral formation might be an ongoing process in these tissues. The lack of control slices from normal living human hippocampus makes it difficult to assess to what extent the present findings are epilepsy associated. However, the presence of aberrant mossy fiber collaterals in the hippocampi used in the present study has been confirmed by Timm's staining and/or dynorphin immunohistochemistry in comparison with nonepileptic autopsy material, indicating its relation to epilepsy (Babb et al., 1991, 1992). At present, there seems to be a consensus that the projection of mossy fiber collaterals to the supragranular layer is a rare occurrence in normal rats (Lorento de No, 1934; Claiborne et al., 1986; Seress et al., 1991; present study), normal monkeys (Seress et al., 1991), and normal humans (Houser et al., 1990).(ABSTRACT TRUNCATED AT 400 WORDS)

✓ The authors reviewed 149 patients who underwent standard anterior temporal lobectomies for intractable complex partial epilepsy with a mean follow-up period of 5 years. Quantitative analyses of hippocampal neuron loss showed that all patients had some cell loss compared to control hippocampi obtained at autopsy. The average hippocampal cell loss was categorized as severe (> 30% of autopsy control levels) or mild. Analysis of hippocampal and extrahippocampal pathologies showed that in 109 cases (73%, the hippocampal lesion group) hippocampal cell loss was mild in 17 cases (16%) and severe in 92 cases (84%); in the remaining 40 cases (27%, the extrahippocampal structural lesion group) hippocampal cell loss was mild in 24 cases (60%) and severe in 16 (40%). The first index of surgical outcome was worthwhile seizure reduction, which occurred in 94 cases (86%) with mild or severe hippocampal lesions and in 33 cases (82%) with extrahippocampal pathology. In the hippocampal lesion group, worthwhile seizure reduction occurred in 90% of cases with severe and in only 65% of cases with mild hippocampal cell loss (p = 0.015). In the extrahippocampal pathology group, worthwhile seizure was not statistically different, whether hippocampal cell loss was severe (94% of cases) or mild (75% of cases). The second index of surgical outcome was the occurrence of residual seizures in the patients with worthwhile seizure reduction, which would indicate remaining epileptogenic tissue. In the hippocampal lesion group, the incidence of residual seizures was not statistically different whether hippocampal cell loss was severe (24% of cases) or mild (45% of cases). However, in the extrahippocampal pathology group, residual seizures occurred in 53% of cases with severe cell loss (dual pathology) but in only 11% of cases with mild cell loss (p = 0.025). Worthwhile seizure reduction can be predicted by the presence of either severe hippocampal cell loss or an extrahippocampal structural lesion. However, residual seizures more frequently follow in cases with a combination of both (extrahippocampal pathology associated with severe hippocampal cell loss, or dual pathology), suggesting that epileptogenic tissue more likely extends outside the boundaries of a standard temporal lobectomy.

Patients with drug-refractory temporal lobe epilepsy can be treated successfully with surgical resection of one temporal lobe, especially when the resection includes the hippocampus. Although intrahippocampal recordings usually localize seizure onsets to one hippocampus, there are bilaterally-independent interictal spikes, occasional contralateral seizure onsets and post-resection seizures that implicate contralateral damage and epileptogenicity. Post-mortem onquantified studies of both hippocampi in epileptics have revealed incidences of bilateral hippocampal sclerosis, mostly being asymmetric. The present paper reports on two post-mortem cases of bilateral, asymmetric cell loss in patients with physiologically-verified hippocampal epilepsy. In one patient the damage was severe bilaterally, but only slightly greater damage in one hippocampus. In the second patient, damage in one hippocampus was as severe as in the first patient; however the contralateral hippocampus appeared undamaged. However, cell counts revealed losses of over 30% in three different hippocampal subregions, indicating a mild form of asymmetric bilateral damage in patient two.

In a previous investigation of functional limbic pathways in the human mesial temporal lobe, we found evidence for strong connections between ipsilateral mesial temporal structures, but none for contralateral functional connections (Wilson et al. 1990). In the present study, we focused specifically upon the question of functional commissural linkages between these structures by systematic stimulation of a total of 390 electrode placements in 74 epileptic patients with temporal lobe depth electrodes implanted for surgical diagnosis. Eight standard electrode placement regions were targeted: amygdala, entorhinal cortex, anterior, middle and posterior hippocampus, subicular cortex, middle parahippocampal gyrus, and posterior parahippocampal gyrus. Three to six electrodes were implanted bilaterally in each patient, and each electrode was individually stimulated while recording from all the other sites. Out of the 390 electrodes stimulated, 78% were effective in evoking clear responses in adjacent ipsilateral structures, and 75% of 581 ipsilateral recording sites were responsive to stimulation. Only one of the stimulated electrode sites was effective in evoking responses in contralateral recording sites, and only two of 511 contralateral recording sites were responsive to that stimulation. The effective stimulation site was in presubicular cortex, and the responsive contralateral recording sites were in entorhinal and presubicular cortices. Response to this stimulation site was intermittent and variable in latency. The relative ease of obtaining functional verification of significant ipsilateral anatomical pathways in the human limbic system, and the sharply contrasting difficulty of functionally activating commissural pathways to contralateral limbic sites are discussed in the context of decreases in hippocampal contribution to commissural pathways in the primate brain compared to sub-primate mammals, and the significance of this change to normal limbic system function as well as to mechanisms of seizure spread in epilepsy.

The present study examined the bilateral synaptic rearrangements of presumed gamma-aminobutyric acid (GABAergic) inhibitory axons and mossy fiber (presumed excitatory) recurrent collaterals following intrahippocampal kainic acid (KA) injection. Glutamate decarboxylase immunoreactivity (GAD-IR) was used to study inhibitory axon terminal sprouting, following 0.5 microgram KA/0.2 microliter injected unilaterally into the posterior hippocampus of rats (n = 16), with survival periods of 14, 28, and 120 days. The age-matched control animals (n = 9) received intrahippocampal 0.2 microliter saline (sham, n = 4) or no injection (normal, n = 5). To study mossy fiber synaptic rearrangements, 0.5 microgram KA/0.2 microliter volumes were injected unilaterally into the posterior hippocampus of rats (n = 10), with survival periods from 14, 28, and 120 days, and Timm sulfide-stained tissue sections were compared to age-matched sham (n = 4) or normal controls (n = 4). At 14 through 120 days after posterior KA injection, GAD-IR puncta were significantly increased in the ipsi- and contralateral inner molecular layers (IML) of the fascia dentata (FD) when compared to sham or normal controls. KA lesion also induced mossy fiber recurrent collateral sprouting into the ipsi- and contralateral FD IMLs. The loss of both the commissural and ipsilateral associational afferents to the FD apparently induced sprouting into their ipsi- and contralateral termination zones by granule cell mossy fibers and GAD-IR axons, thus establishing an abnormal circuitry near the observed pathology in the kainate model of epilepsy. Although reactive synaptogenesis of mossy fibers producing monosynaptic excitation may be one mechanism for KA epileptogenicity, the concurrent sprouting of GABAergic terminals in the same IML zone of the FD suggests that anomalous inhibitory synapses may contribute to chronic KA hippocampal hyperexcitability.

The present study used Nissl stains and glutamate decarboxylase immunoreactivity (GAD-IR) to quantify the acute and chronic toxicity of kainic acid (KA) on focal and remote hippocampal principal neurons (i.e., pyramidal and granule cells) and on putative inhibitory neurons (GAD-IR or GABAergic) following intrahippocampal KA administration. Concentrations of 0.5, 1.0, 1.25 or 1.5 micrograms KA/0.2 microliters were injected unilaterally into the posterior hippocampus of rats (n = 32), with survival periods of 1, 3, 5, 14, 21, 30 and 60 days. The age-matched control animals (n = 10) received an intrahippocampal injection of 0.2 microliter saline (sham control, n = 4) or no injection (normal, n = 6). The ipsilateral (KA+) cell counts demonstrated a selective vulnerability of CA3 and CA4 pyramidal neurons which was maximal at 14 days and unchanged to 60 days. However, in the same region, putative inhibitory (GAD-IR) neurons were resistant to the neurotoxic effects of KA. Contralateral (KA-) pyramidal cell and GAD-IR neuron densities were equivalent to controls. The present data demonstrate a selective resistance to KA by GABA neurons compared to the vulnerability of pyramidal neurons. Because GABA neurons are relatively spared in the KA focus, loss of GABAergic inhibitory neurons is probably not a mechanism for the seizure sensitivity in the KA model.

The NMDA receptor is one of the ionotropic glutamate receptors essential for excitatory neurotransmission and synaptic plasticity. The present study examined the ontogenies of NMDAR1 and NMDAR2B during development (embryonic days: (E) 14, 15, 16, 17 and postnatal days: (P) 0, 5, 10, 20, and 60). The immunoreactivities of both NMDAR1 and NMDAR2B were visualized and semi- quantified simultaneously using batch-matched procedures during the developmental period. In addition, the heteromeric coassembly of NMDAR1 (NR1) and NMDAR2B (NR2B) was analyzed throughout development using immunocoprecipitation blots. Differential developmental expressions were observed between the two NMDA receptor subunits. Although both NR1 and NR2B were already expressed on cortical neurons at E14, NR1 immunoreactivity was lower than NR2B but increased up to P20 then was lower at the last sampling day (P60). By comparison, the high NR2B immunoreactivity at E14 increased until P5, and until the next sample day (P60). Despite the early embryonic protein expressions in prenatal cortex both NR1 and NR2B immunocoprecipitation analyses showed that NR1–NR2B coassembly was not initially detectable until birth (P0), and increased up to P20. This suggests that the physically interactive NMDAR1-2B complex may contribute to the plasticity of early postnatal development characterized by axonal dendritic synaptogenesis.

Hippocampal sclerosis (HS) is characterized by hippocampal atrophy and increased signal on T2-weighted images and on fluid-attenuated inversion recovery (FLAIR) images.To quantitate cell loss and compare it with signal abnormalities on FLAIR images.Thirty-one patients with temporal lobe resection, pathologically proven HS, and Engel class I and II outcome were included: 20 with HS only and 11 with HS associated with pathologically proven cortical dysplasia (dual pathology). The signal intensity on FLAIR was rated as present or absent in the hippocampus and correlated with the neuronal losses in the hippocampus.FLAIR signal increases were present in 77% (24/31) of all patients studied. In patients with isolated HS, 90% (18/20) had ipsilateral signal increases, but in patients with dual pathology, only 55% (6/11; p0.02) showed FLAIR signal increase. Hippocampal cell losses were significantly higher in the isolated HS group. The average cell loss in patients with FLAIR signal abnormalities was 64.8 +/- 8.0% as compared with only 32.7 +/- 5.1% in patients with no FLAIR signal abnormalities. There was a significant positive correlation between the presence of signal abnormality and average hippocampal cell loss in both pathologic groups.Ipsilateral FLAIR signal abnormalities occur in the majority of patients with isolated HS but are less frequent in those with dual pathology. The presence of increased FLAIR signal is correlated with higher hippocampal cell loss.

We examined the time course of NMDAR1 (NR1) immunoreactivity (IR) in the rat inner molecular layer of the dentate gyrus following unilateral intrahippocampal (hilar) kainic acid (KA) lesions and compared them to progressive aberrant mossy fiber (MF) sprouting into the inner molecular layer (IML). The results demonstrated that NR1 receptors in the IML of the KA side were decreased as early as 3 days after KA-induced denervation, then significantly increased at postinjection day (PID) 7. The densities of NR1 IR in the IML continued to increase up to 5 months. By comparison, MF sprouting did not occur significantly in the IML until PID 17, 10 days after NR1 IR was significantly increased. Recurrent MF-IML neoinnervation significantly increased on days 17, 60, and 150. This progressive MF innervation was significantly correlated with NR1 increases. These results suggest that NR1 receptors were decreased soon after KA-induced deafferentation of granule cell dendrites in the IML; however, they were replaced by new NR1 receptors at increased densities in the granule cell dendrites, which may have released neurotrophic factors to stimulate growth cones of MFs to reinnervate the IML. The progressive increases of NR1 and MFs in the IML suggest that such neosynaptogenesis would contribute monosynaptic recurrent excitatory mechanisms for focal hippocampal hyperexcitability and seizure onsets.

We determined whether age or seizure types were associated with hippocampal neuron loss, mossy fiber (MF) and GABAergic synaptic reorganizations or postsynaptic receptor densities. Children and adolescents were grouped into: (1) nonhippocampal sclerosis (non-HS; n = 11) and (2) hippocampal sclerosis (HS; n = 11). The most important results showed that: (1) regardless of the etiology of the seizures, there were greater cell losses in Ammon’s horn with older ages in years; in the non-HS group, cell losses were greater with the older ages or with longer epilepsy durations; however, in the HS patients, the cell losses were not related to the patients’ ages or epilepsy durations; (2) in both HS and non-HS, CA1 had greater cell losses than CA4; (3) in HS, CA1 and CA4 had greater cell losses than those in non-HS; (4) in non-HS, MF sprouting was greater with ages or with longer epilepsy durations; by contrast, in HS, MF sprouting was not related to the patients’ age or epilepsy duration; (5) densities for AMPA GluR1, GABA-Aβ and for GABA axonal terminals were positively increased with age. These findings support the hypothesis that hippocampal cell losses and aberrant synaptic reorganizations are greater in the hippocampi of adolescents than in children, even for non-HS pathologies.

NR1 and NR2 are the two gene families for the NMDA receptor. In vitro studies show that while NR2 alone is nonfunctional, NR1 alone produces weak currents to glutamate or NMDA. We previously showed by immunocytochemistry (ICC) that in normal appearing, nonepileptic human cortical neurons, only NR1 and not NR2 proteins were expressed, in contrast to the presence of both NR1 and NR2 in normal rat cortical neurons. We also showed, in dysplastic epileptic cortex, that both NR1 and NR2 were highly expressed using ICC on adjacent 30-μm sections. However, the relative coexpressions of NR1 and NR2 proteins in single neurons in single sections of human epileptic cortex were unknown. In this study, we used double-labeled immunofluorescence and confocal microscopy to examine the distribution and coexpression of subunit proteins for NR1 and NR2A/B in both nondysplastic (control comparison) and dysplastic regions of human brain resected for the treatment of intractable epilepsy (11 patients). In nondysplastic regions, cortical neurons did not have immunoreactivity (ir) for NR2A/B, whereas NR1-ir was abundant. By contrast, dysplastic neurons in the regions with epileptic cortical dysplasia showed intense NR2A/B-ir in the somata and their dendritic processes. These same NR2A/B-ir dysplastic neurons were colabeled by NR1. These results demonstrate directly that dysplastic neurons express both NR2A/B and NR1 proteins, whereas nondysplastic cortical neurons express only NR1 proteins. Selective coexpression of NR2A/B and NR1 in dysplastic neurons suggests that NR2A/B may form heteromeric NR1–NR2 coassemblies and hyperexcitability in dysplastic neurons that could contribute to focal seizure onset.

The N -methyl- d -aspartate receptors (NMDAR) produce physiologically functional channels for enhanced excitatory neurotransmission when they exist as heteromeric complexes containing the NMDAR1 subunit combined with NMDAR2. We examined the expressions of NMDAR1 and 2A/B protein in the kainic acid induced rat chronic epileptic hippocampus. Immunoreactivities of both NMDAR1 and NDMAR2A/B were increased in the inner molecular layer of the dentate gyrus, while they were decreased in the hilar and CA3/4 pyramidal zones. Immunoblot analysis demonstrated that the overall level of NMDAR1–2A/B coassembly was increased in the whole hippocampus. These results indicate that the increase of the NMDAR1–2A/B complex in the inner molecular layer is a significant cellular mechanism that contributes to focal hyperexcitability in rat chronic hippocampal epilepsy.

We examined the time course of expression of glial cell line-derived neurotrophic factor (GDNF) protein in the granule cells of the dentate gyrus following unilateral intrahippocampal injection of kainic acid (KA). Recurrent behavioral seizures were observed approximately 1 h after KA injection, which lasted for 4-6 h. GDNF immunoreactivity began to increase bilaterally in the granule cells within 3 h after KA injection, continued to increase until post-injection day (PID) 4, and returned to the control level by PID 7. The results suggest that the increase of GDNF protein in the granule cells may be ascribable to seizures induced by the KA injection. The increase of GDNF protein might promote survival of the granule cells after the intrahippocampal KA injection.

Publisher Summary This chapter discusses various aspects of hippocampal neurophysiology in humans. The human hippocampus occupies less than 1% of the forebrain, which is a significantly smaller fraction than in the basal insectivores or in monkeys. The hippocampus and amygdala may increase with phylogeny, only the olfactory bulbs actually decrease in absolute size in primates relative to lower mammals, suggesting that man does not depend much on olfaction. Large, postoperative quadranopsias and occasional partial hemianopsias following standard anterior temporal lobectomies have confirmed the findings that in man, there are variable asymmetric geniculocalcarine pathways in the Meyer's loop of the human temporal lobe. The hippocampus, although a relatively small cortex with fewer than six layers, is nevertheless anatomically well connected for feedback and feedforward excitation and inhibition. The local circuit neurons are interactive with such complexity that even a small simultaneous loss of recurrent inhibition may be sufficient to create hyperexcitability. It is suggested that interactive physiology of inhibition and excitation within the hippocampal network poses the major challenge for understanding the many crucial, unique functions of the hippocampal formation, which includes short-term memory and focal epileptogenesis.

Developmental disorders of neuronal migrations in the human brain are referred to as `cortical dysplasia', and current knowledge of cortical dysplasia is limited to varied pathologic descriptions which lack specific investigations of glutamate receptor mechanisms. In this study, immunocytochemistry was used to study the expressions of glutamate receptor subunit proteins for NMDAR2A/B, NMDAR1 and AMPA Glu-R2/3 in human brain resected for intractable epilepsy associated with cortical dysplasia. Seventeen patients were studied with batch-matched glutamate subunit reagents on adjacent 30-μm sections. The most striking microscopic abnormalities identified in cresylecht violet stains were cortical dyslaminations, disoriented neurons, and unexpectedly, very dark Nissl body staining of those dysplastic neurons. NMDAR2A/B intensely labeled dysplastic neurons, showing staining in both the cell bodies and dendritic profiles. However, non-dysplastic neurons were not immunoreactive to NMDAR2A/B. Dysplastic neurons were also labeled by antibodies selective to NMDAR1. Both dysplastic neurons and non-dysplastic neurons were immunoreactive to AMPA GluR2/3. Our results suggest that the epileptic hyperexcitability of dysplastic cortical regions may result, at least in part, from the heteromeric coassembly and expressions of NMDAR2A/B subunits with selectively expressed NMDAR1 splice variants in dysplastic neurons. AMPA receptors are probably also essential but not sufficient to explain the `epileptic' properties of these dysplastic neurons. A longer, detailed report of some of these findings have been previously published ( Ying et al., 1998 . J. Neuropathol. Exp. Neurol. 57, 47–62).

Paired pulse stimulation has commonly been employed to investigate changes in excitability in epileptic hippocampal tissue employing the in vitro slice preparation. We used paired pulse stimulation in the intact temporal lobe of patients with temporal lobe seizures to compare the excitability of pathways in the epileptogenic hippocampus (located in the temporal lobe in which seizures arise) with those in the non-epileptogenic hippocampus of the contralateral temporal lobe (in the hemisphere to which seizures spread). A total of 20 patients with temporal lobe seizure onsets were studied during chronic depth electrode monitoring for seizure localization. Intracranial in vivo stimulation and recording sites included the hippocampus, entorhinal cortex, subicular cortex and parahippocampal gyrus. A comparison of all hippocampal pathways located in the temporal lobe where seizures typically started (n = 37) with those in temporal lobes contralateral to seizure onset (n = 53) showed significantly greater paired pulse suppression of population post-synaptic potentials on the epileptogenic side (F(1,87) = 6.1, P < 0.01). Similarly, mean paired pulse suppression was significantly greater for epileptogenic perforant path responses than for contralateral perforant path responses (F(1,13) = 7.5, P < 0.01). In contrast, local stimulation activating intrinsic associational pathways of the epileptogenic hippocampus showed decreased paired pulse suppression in comparison to the epileptogenic perforant path. These results may be a functional consequence of the formation of abnormal recurrent inhibitory and recurrent excitatory pathways in the sclerotic hippocampus. Enhanced inhibition may be adaptive in suppressing seizures during interictal periods, while abnormal recurrent excitatory circuits in the presence of enhanced inhibition may drive the hypersynchronization of principal neurons necessary for seizure genesis.

In human hippocampal epilepsy, there is a consistent pathology of cell loss and reactive synaptic reorganization of 'excitatory' mossy fibers (MF) into the inner molecular layer (IML) of the fascia dentata (FD). In this study, neo-Timm's histochemistry of MFs and immunocytochemistry of GluR1 were used to determine, in patients with or without hippocampal sclerosis (HS), if there was a correlation between aberrant supragranular (IML) mossy fiber sprouting and increased densities of AMPA GluR1 subunit proteins in the IML of the FD. Computerized quantified densitometric grey values of Timm and GluR1 densities were corrected for the densities of granule cell losses using cell counts. In the IML of the HS group, despite the losses of granule cells, mossy fiber sprouting was significantly greater (P

This study determined neuron losses, mossy fiber sprouting, and interictal spike frequencies in adult rats following intrahippocampal kainic acid (KA) injections during postnatal (PN) development. KA (0.4 micrograms/0.2 microliters; n = 64) was injected into one hippocampus and saline into the contralateral side between PN 7 to 30 days. Animals were sacrificed 28 to 256 days later, along with age-matched naive animals (controls; n = 20). Hippocampi were studied for: (1) Fascia dentata granule cell, hilar, and CA3c neuron counts; (2) neo-Timm's stained supragranular mossy fiber sprouting; and (3) hippocampal and intracerebral interictal spike densities (n = 13). Mossy fiber sprouting was quantified as the gray value differences between the inner and outer molecular layer. Statistically significant results (p0.05) showed the following: (1) Compared to controls, CA3c and hilar neuron counts were reduced in KA-hippocampi with injections at PN 7-10 and PN 12-14 respectively and counts decreased with older PN injections. Granule cell densities on the KA-side and saline injected hippocampi were not reduced compared to controls. (2) In adult rats, supragranular mossy fiber sprouting was observed in 2 of 7 PN 7 injected animals. Compared to controls, increased gray value differences, indicating mossy fiber sprouting, were found on the KA-side beginning with injuries at PN 12-14 and increasing with older PN injections. On the saline-side only PN 30 animals showed minimal sprouting. (3) Mossy fiber sprouting progressively increased on the KA-side with longer survivals in rats injured after PN 15. Sprouting correlated positively with later PN injections and longer post-injection survival intervals, and not with reduced hilar or CA3c neuron counts. (4) On the KA-side, mossy fiber gray value differences correlated positively with in vivo intrahippocampal interictal spike densities. These results indicate that during postnatal rat development intrahippocampal kainate excitotoxicity can occur as early as PN 7 and increases with older ages at injection. This rat model reproduces many of the pathologic, behavioral, and electrophysiologic features of human mesial temporal lobe epilepsy, and supports the hypothesis that hippocampal sclerosis can be the consequence of focal injury during early postnatal development that progressively evolves into a pathologic and epileptic focus.

The present study examined the relationship between the patterns and densities of glutamate AMPA receptor sub-units GluR1 and GluR2/3 in the molecular layer of the fascia dentata and aberrant mossy fiber neoinnervation in human and kainate rat hippocampal epilepsy. Because AMPA sub-units modulate the fast glutamate synaptic transmission, we hypothesized that the AMPA receptor densities would be related to the glutamate-secreting mossy fibers, which could then contribute to seizure generation. In human hippocampal epilepsy, we found that the immunocytochemical labeling of GluR1 and GluR2/3 dendrites was positively related to the densities and spatial locations of the densest, aberrant neo-Timm stained supragranular mossy fibers. We used quantitative densitometry for the mossy fibers. However, the relatively faint and punctate immunocytochemical staining of the receptors did not allow true quantitative densitometry of the dendritic trees because in human epilepsy granule cell densities were decreased on average 50% of normal. Nevertheless, visual observations did confirm spatial relations between dense fascia dentata inner molecular layer mossy fibers and dense AMPA receptor staining. In the outer molecular layer, the mossy fibers were present only in the lower portion, were not densely-stained, and the AMPA receptors were only faintly-labeled. Nevertheless, outer molecular layer AMPA receptor densities were usually present more distally than were the mossy fibers. Experiments were done using intrahippocampal kainate epileptic rats to test the time courses for the changes in mossy fibers and AMPA receptors. The upregulation of inner and outer molecular layer AMPA receptors occurred maximally within 5 days post-kainate injection, prior to any mossy fiber supragranular ingrowth. One hundred and eighty days after ipsilateral kainate the AMPA receptors were increased bilaterally in the inner and outer molecular layers despite the fact that the contralateral aberrant supragranular mossy fibers were minor in comparison to the dense ipsilateral mossy fiber hyperinnervation. These results suggest that in hippocampal epilepsy AMPA receptor numbers increase throughout the length of the molecular layer dendrites; however the AMPA receptor densities are greater in rough relation to the greatest aberrant mossy fiber presynaptic inputs. Interestingly, the receptor upregulation precedes the mossy fiber ingrowth and may play a role in initiating axonal sprouting or in maintaining the aberrant mossy fiber synapses.

This study determined in temporal lobe epilepsy patients and rats injected with intrahippocampal kainate (KA) whether fascia dentata molecular layer mossy fiber sprouting was associated with increases in NMDAR2 immunoreactivity (IR). Patients with hippocampal sclerosis (n = 11) were compared with those with temporal mass lesions (n = 7) and material obtained at autopsies (n = 4); and unilateral KA-injected rat hippocampi (n = 7) were compared with the contralateral saline-injected side and non-lesioned animals (n = 7; control). Hippocampi were studied for neo-Timm's stained mossy fiber sprouting and NMDAR2 IR. The staining was quantified as gray values (GV) using computer image analysis. Hippocampal sclerosis patients and KA-injected rats showed the greatest inner molecular layer (IML) mossy fiber sprouting and NMDAR2 staining. Compared with autopsies and patients with mass lesions, hippocampal sclerosis patients had greater IML neo-Timm's (p = 0.0018) and NMDAR2 staining (p = 0.0063). Similarly, compared with controls and saline-injected rats, KA-injected hippocampi showed greater IML mossy fiber sprouting and NMDAR2 IR (p = 0.0001). Furthermore, IML mossy fiber sprouting positively correlated with greater IML NMDAR2 staining in both human and experimental rat groups (p < 0.0099). These results support the hypothesis that in severely damaged hippocampi abnormal mossy fiber sprouting and concordant increases in IML NMDAR2 receptor staining may contribute or partially explain granule cell hyperexcitability and the pathophysiology of hippocampal epilepsy.

In temporal lobe epilepsy (TLE) lesion patients the pathology, location of intracranial ictal EEG onsets, and hippocampal neuron losses were compared. Patients (n = 63) were classified into: (1) Tumors (n = 26, e.g. astrocytomas, gangliogliomas); (2) vascular (n = 9, e.g. cavernous and venous angiomas); (3) developmental (n = 17, e.g. cortical dysplasia, heterotopias); or (4) atrophic (n = 11, e.g. cortical or white matter encephalomalacia). Other variables were; (1) the location of the temporal lesion in the mesial to lateral, and anterior to posterior plane, (2) a clinical history of an initial precipitating injury (IPI) prior to the onset of TLE (e.g. prolonged first seizure, head trauma), (3) hippocampal neuron densities, (4) focal or regional location by intracranial depth EEG of ictal onsets, and (5) seizure outcomes. Results showed that severe hippocampal neuron losses were associated with two statistically significant findings. First, patients with mesial lesions in or adjacent to the body of the hippocampus had greater neuron losses compared to mesial lesions anterior or posterior to the hippocampus (P = 0.04). Second, lesion patients with an IPI history had greater Ammon's horn (AH) neuron losses compared to those without IPI histories (P = 0.0005), and the profile of loss was similar to hippocampal sclerosis (HS). Granule cell losses correlated in a complex manner in that; 1) by regression analysis densities decreased with longer intervals of TLE (P = 0.006), (2) tumor patients with IPIs had less granule cell loss compared to those without IPIs intervals of TLE (P = 0.006), (2) tumor patients with IPIs had less granule cell loss compared to those without IPIs (P = 0.05), and (3) developmental patients with IPIs had greater granule cell loss than patients without IPIs (P = 0.009). Mesial-temporal depth EEG electrodes were the first areas of ictal activity in 15 of 16 patients (94%), and greater hippocampal neuron losses were not associated with focal mesial-temporal EEG onsets. Seizure outcomes were worse in tumor patients compared to HS patients (P = 0.01), and patients with post-resection seizures had incomplete resections of their lesions and/or hippocampi. These results indicate that in TLE lesion patients the amount and pattern of hippocampal neuron loss depends on the location of the lesion, the pathologic classification, and a history of an IPI. Further, despite variable neuron losses, in temporal lesion patients the hippocampus was nearly always involved in the genesis or propagation of the chronic seizures.

✓ The type of initial precipitating injury and the age at which it occurred in 20 patients with nonlesional temporal lobe epilepsy (TLE) were related to clinical features, presurgical neuroimaging, quantified hippocampal pathologies, and seizure outcomes. Clinical data, neuroimaging records, and seizure outcomes were abstracted from medical records and confirmed with patient and family contacts. Hippocampal neuron losses and mossy fiber reactive synaptogenesis were quantified independently. Results showed that the type of initial precipitating injury and the patient's age at which it occurred were related to the clinicopathological features of TLE. An initial precipitating injury occurred in 18 patients (90%), all of whom had mesial temporal sclerosis (MTS). Patients with a prolonged initial seizure or a nonseizure initial precipitating injury before age 5 years were significantly more likely to have unilateral hippocampal atrophy (p < 0.05) shown on magnetic resonance (MR) imaging, and had significantly greater inner molecular layer mossy fiber puncta densities (p < 0.001) than patients with nonprolonged childhood initial precipitating injuries and/or seizures after age 5 years. Furthermore, nonseizure injuries in patients before age 5 years had significantly longer latent periods (p < 0.05), and the patients did not respond to surgical treatment as well as other MTS patients. Those with an initial precipitating injury after age 5 years had MTS but showed significantly less inner molecular layer mossy fiber sprouting (p < 0.05) than patients whose injuries appeared before age 5 years. Patients without an initial precipitating injury (idiopathic TLE) had significantly fewer neuron losses (p < 0.05) and inner molecular layer mossy fiber puncta densities (p < 0.05) and had worse outcomes following en bloc temporal lobectomy compared to patients with MTS who had experienced initial precipitating injuries. Patients with unilateral hippocampal abnormalities on MR imaging did not show significant differences in neuron losses or aberrant mossy fiber puncta densities compared to patients without asymmetry. These results support the hypothesis that the type of initial precipitating injury and the age at which the injury occurred initiates and influences the pathophysiological process that eventually develops into MTS. These data support the notion that the pathophysiology of hippocampal damage and mossy fiber sprouting after an initial precipitating injury may be a progressive process.

The pathophysiology of widespread interictal hypometabolism in temporal lobe epilepsy is unknown but might reflect neuronal loss and diaschisis. We found no significant correlation between any cortical region's metabolism on preoperative { 18 F}fluorodeoxyglucose positron emission tomography and neuronal density of resected hippocampi in 40 patients. We conclude that hippocampal neuronal loss and diaschisis cannot account for the regional interictal hypometabolism of temporal lobe epilepsy

This study determined differences in clinical-pathologic characteristics of intractable temporal lobe epilepsy (TLE) patients whose mechanism of cerebral injury and chronic seizures involved a prior history of cerebral trauma compared to those with non-traumatic initial injuries. TLE patients (n = 120) from a single epilepsy center were retrospectively and blindly catalogued into pathogenic groups (independent variables) based on if there was a significant Birth injury (n = 11) or Cerebral trauma (n = 26). These two 'trauma' categories were compared to TLE patients with non-seizure non-trauma histories (Non-Sz/Non-Trauma; n = 17), or a first Prolonged seizure (n = 66). The four groups were compared for differences in the time course of their clinical injuries and seizures, quantified hippocampal neuron counts, other temporal neocortical pathologies, and seizure outcomes (dependent variables). Between group statistically significant (at least P0.05) results showed: (1) In Birth injury, 33% had Ammon's Horn (AH) neuron loss under 50%, 54% had other temporal neocortical pathologies, they showed the most CA4 neuron loss, and the worse seizure outcomes. (2) Cerebral trauma were older when injured, 29% had AH loss under 50%, 50% showed other pathologies, and they had the best seizure outcomes. (3) Non-Sz/Non-Trauma showed the least AH and CA4 neuron losses, only 12% had other temporal pathologies, and they had seizure outcomes that were intermediate. (4) Prolonged seizure showed the youngest age of habitual TLE onsets, the greatest AH, CA1, and prosubiculum neuron loss, only 11% had other temporal pathologies, and their seizure outcomes were excellent. These results indicate that in intractable surgically treated TLE, a history of cerebral trauma or birth injury as the pathogenic mechanism of their seizures show different clinical-pathologic features and seizure outcomes compared to non-trauma patients. This supports the notion that in TLE there are different pathogenic mechanisms associated with different types of initial injuries and that patients will have different responses to surgical therapy.

Surgically resected hippocampi from children with extrahippocampal seizures and structurally non-atrophic brains were examined to determine the relationship of neuron losses and aberrant mossy fiber (MF) sprouting to the postnatal migration and differentiation of the fascia dentata (FD) granule cells (GC). Percent neuron loss compared to age-matched autopsy controls was determined by quantitative cell densities, and aberrant MF sprouting by neo-Timm histochemistry. Postnatal immature GC migration and differentiation was demonstrated by the transient but GC-specific expression of the immature form of neural cell adhesion molecule (NCAM-H). Results showed that the hippocampi from children with seizures appeared microanatomically intact without focal areas of damage. However, significant neuron losses were found by neuron counts in the fascia dentata (P < 0.01), CA4 (P < 0.01), and CA2 (P < 0.05). Aberrant supragranular inner molecular layer MF sprouting was found in hippocampi of children with seizures, and the MFs showed smaller puncta in specimens resected under 2 years of age (n = 3) compared to the larger puncta in older children (n = 5). Hippocampi from children under 2 years of age also demonstrated NCAM-H positive primitive cells in the infragranular and stratum granulosum of the fascia dentata consistent with the postnatal migration and differentiation of GCs, the parent neurons of the MFs. These results indicate that seizures in the immature but structurally intact human hippocampus are associated with decreased neuron densities and aberrant MF sprouting very early in postnatal development. The data also show that aberrant MF sprouting is found during postnatal migration, differentiation and axogenesis of GCs. These results suggest that the timing of seizures during hippocampal development influences the amount of neuron loss and of MF sprouting, which may contribute to features similar to those of adult hippocampal sclerosis.

Previously, Mathern et al. (1992) demonstrated progressive mossy fiber (MF) sprouting in the intrahippocampal rat kainate seizure model. This study looked at the time course of EEG hyperexcitability and spontaneous seizure activity in the same in vivo model to determine if seizures were associated with MF sprouting. Results showed that animals progressed through 4 distinct EEG and behavioral phases and that in the chronic phase (greater than 90 days post kainate) MF sprouting was strongly associated with hippocampal epileptogenesis. Progressive MF sprouting into the inner molecular layer (IML) of the fascia dentata paralleled the EEG and behavioral appearance of independent hippocampal interictal epileptiform transients and chronic seizures. Hippocampi from chronic animals that demonstrated unilateral MF IML sprouting were observed to have interictal epileptiform transients and spontaneous seizures that lateralized to the hippocampus with synaptically reorganized MFs. Chronic animals with bilateral MF sprouting were observed to have bilateral independent EEG and behavioral hyperexcitability. Control animals and kainate treated animals that lacked hippocampal cell loss and MF sprouting did not show signs of chronic hippocampal EEG hyperexcitability or chronic seizures. These data support the idea that MF sprouting contributes to chronic hippocampal seizures by feedback excitation which leads to the excitability and synchronization required for a damaged hippocampus to become an epileptic focus.

• Objective. —To assess the relationship of hippocampal neuron loss to intellectual and memory measures before and after temporal lobe surgery. Design. —Pyramidal cell loss, as determined on the resected tissue, of hippocampal subregion CA1 correlated highest with other subregional cell loss and thus was used as the primary indicator of hippocampal neuron loss. Groups of patients with left and right temporal lobe seizures were subdivided according to degree of CA1 neuron loss. Behavioral performances of patient groups were compared before and after surgery. Setting. —Patient data were obtained from a university program of surgery for epilepsy. Cases. —Twenty-five patients who had intractable epilepsy. Main Outcome Measures.—Wechsler Adult Intelligence Scale IQ scores, verbal and nonverbal memory measures adapted from the Wechsler Memory Scale, and the Rey-Osterrieth recall score. Results. —Degree of hippocampal cell loss selectively related to learning of unrelated word pairs, both preoperatively and postoperatively, in patients with left but not right temporal lobe seizures. Patients with severe loss of left hippocampal neurons performed worse than those with mild-moderate neuron loss both before and after surgery. Immediate recall of logical prose did not relate to hippocampal neuron loss, although scores decreased following left temporal lobe surgery. Conclusion. —These findings support a role for the left hippocampus in rote verbal memory, ie, learning of unrelated word pairs. Semantically complex verbal learning, ie, recall of logical prose, is more dependent on extrahippocampal temporal lobe regions. Finally, patients with severe as compared with minimal left hippocampal neuron loss may be at risk for lower memory functioning postoperatively.

✓ The authors present their review of 178 patients who underwent en bloc temporal lobectomies as surgical treatment for intractable epilepsy. Hippocampal cell density was quantitatively analyzed and the histology of the anterior temporal lobe was reviewed. Fifty-four patients (30.3%) had evidence of extrahippocampal lesions in addition to neuronal cell loss within the hippocampus (the dual pathology group). The pattern of cell loss was analyzed in the remaining 124 cases (69.7%) with no extrahippocampal pathology, and compared with that of the dual pathology group and a control group of four nonepileptic patients. Hippocampal cell loss was found in almost all epileptic patients compared to the control group. Severe cell loss greater than 30% of control values was found in 88.7% of patients without extrahippocampal lesions, but in only 51.8% of patients with dual pathology. The difference between these two groups was statistically significant (p < 0.001). In the dual pathology group, lesions of different pathology had a significant relationship with the degree of hippocampal cell loss: all 12 patients with glioma had mild cell loss, whereas all 13 patients with heterotopia were associated with severe cell loss. Severity of hippocampal cell loss was also analyzed in relation to seizure history: a prior severe head injury was associated with severe cell loss. Other factors such as seizure duration, secondary generalization, or family history of seizures were not associated with hippocampal damage. Dual pathology may produce a combination of neocortical and temporolimbic epilepsies that warrants a precise definition of the true epileptogenic area prior to surgical treatment.

This study was designed to identify whether synaptic reorganizations occur in epileptic human hippocampus which might contribute to feedback excitation. In epileptic hippocampi, (n = 21) reactive synaptogenesis of mossy fibers into the inner molecular layer of the granule cell dendrites was demonstrated at the light microscopic and electron microscopic levels. There was no inner molecular layer staining for mossy fibers in autopsy controls (n = 4) or in controls with neocortex epilepsy having no hippocampal sclerosis (n = 2). Comparing epileptics to controls, there were statistically significant correlations between Timm stain density and hilar cell loss. Since hilar neurons are the origin of ipsilateral projections to the inner molecular layer, this suggests that hilar deafferentation of this dendritic zone precedes mossy fiber reafferentation. Quantitative Timm-stained electron microscopy revealed large, zinc-labelled vesicles in terminals with asymmetric synapses on dendrites in the inner molecular and granule cell layers. Terminals in the middle and outer molecular layers did not contain zinc, were smaller and had smaller vesicles. These histochemical and ultrastructural data suggest that in damaged human epileptic hippocampus, mossy fiber reactive synaptogenesis may result in monosynaptic recurrent excitation of granule cells that could contribute to focal seizure onsets.

Connections in the human mesial temporal lobe were investigated using brief, single pulses of electrical stimulation to evoke field potential responses in limbic structures of 74 epileptic patients. Eight specific areas within these structures were stereotactically targeted for study, including amygdala, entorhinal cortex, presubiculum, the anterior, middle and posterior levels of hippocampus and the middle and posterior levels of parahippocampal gyrus. These sites were studied systematically in order to quantitatively assess the response characteristics and reliability of responses evoked during stimulation of pathways connecting the areas. Specific measures included response probability, amplitude, latency and conduction velocities. The results are assumed to be representative of typical human limbic pathways since all recordings were made interictally and response probabilities across sites were not found to differ significantly between non-epileptogenic vs. identified epileptogenic regions. Field potentials ranging in amplitude from less than 0.1 to greater than 6.0 mV were evoked ipsilaterally, with mean onset latencies and conduction velocities ranging from 4.4 ms and 3.64 m/s in the perforant pathway connecting entorhinal cortex to anterior hippocampus to 24.8 ms and 0.88 m/s in the pathway connecting the amygdala and middle hippocampus. Stimulation of presubiculum and entorhinal cortex were most effective in evoking widespread responses in adjacent limbic recording sites, whereas posterior parahippocampal gyrus appeared functionally separated from other limbic sites since its probability of influencing ipsilateral sites was significantly lower than any other area. It was particularly noteworthy that stimulation did not evoke responses in any sites in contralateral hippocampal formation; even though a large number of sites were tested with bilateral implantation of homotopic electrodes. The absence of evidence for a functional contralateral limbic projection in the human brain stands in marked contrast to the anatomical and physiological evidence in lower animals for strong contralateral connections between subfields of the hippocampus via the hippocampal commissure. In addition, it correlates well with anatomical evidence for reduced hippocampal commissural connections in lower primates.