Temporal Lobe Sclerosis Associated with Hippocampal Sclerosis in Temporal Lobe Epilepsy: Neuropathological Features

INTRODUCTION

In the earliest descriptions of hippocampal sclerosis (HS) in temporal lobe resection specimens, it was recognized that ‘diffuse and disseminated lesions’ may be present in association with this pathology involving the uncus, amygdala and lateral temporal neocortex (1). The term ‘mesial temporal sclerosis’ (MTS) in preference to HS was introduced in recognition of the frequent involvement of mesial limbic structures adjacent to the hippocampus (2). The cause, severity, frequency and distribution of neocortical neuronal loss and gliosis (temporal lobe sclerosis [TLS]) has been little documented in the neuropathology literature. It remains relatively undefined and is likely to be under-reported.

Attention has recently refocused on the presence of additional pathology in the temporal neocortex in patients with HS in relation to its pathogenesis, the process of epileptogenesis and possible explanations for poor outcomes following surgery. Aside from the identification of a second distinct lesional (or dual) pathology in addition to HS, modern neuroimaging studies identify temporal lobe abnormalities (3-6) and even extra-temporal neocortical changes in patients with HS that may involve hippocampal projection pathways (7-9). The precise pathological correlate of these more widespread neocortical changes remains unclear, as does its significance in relation to generation of seizures. There is also a recognized association between maldevelopmental abnormalities of the temporal lobe, including focal cortical dysplasias, and HS. Furthermore, previous descriptions of TLS-like changes have also alluded to features of cortical dysplasia (10, 11). Therefore, there are fundamental questions regarding the relationship of TLS to HS as well as to cortical dysplasia.

Our aim was to review and clarify the neuropathological features and criteria for TLS in a large series of temporal lobe resections for HS and to study the clinical and neuroimaging features to address some of these issues.

MATERIALS AND METHODS

RESULTS

We identified 30 cases from 272 temporal lobe resections (11%) with the features of TLS as based on previous descriptions (1, 10). Diagnostic criteria for TLS were then formulated based on the observations as detailed below and summarized in Table 1. In the remaining reviewed cases, the temporal lobe neocortex showed an ordered, preserved hexalaminar architecture with no definite neuronal loss and only varying degrees of cortical gliosis. Diffuse gliosis of the temporal lobe white matter and variable degrees of perivascular white matter atrophy were invariable findings in lobectomies for HS; these features were not included in the criteria for TLS (Table 1).

DISCUSSION

Atrophy of the temporal neocortex in patients with temporal lobe epilepsy was recognized in the earliest surgical series from the 1950s (1, 17). Based on these early reports and our current findings we have designated the term ‘temporal lobe sclerosis (TLS)’ and outlined the diagnostic criteria for this pathology in a large series of patients with HS and TLE. TLS was identified in 11% of surgically treated patients in our series, indicating that this pathology is relatively frequent. Mesial temporal lobe sclerosis (MTS) was introduced as a term to encompass sclerosis extending beyond the hippocampus (HS) to involve adjacent medial structures, including the amygdala. We consider that TLS involving the lateral temporal lobe represents part of the spectrum or extension of this process of “mesial temporal lobe sclerosis,” rather than a distinct separate or ‘dual’ pathology.

The hallmark pathological features of TLS in all of our cases are bare areas of neuronal loss in the deeper parts of layer II and upper parts of layer III, as described and illustrated in early studies (17). This is accompanied by laminar gliosis in this region, suggesting that this likely represents an acquired, non-developmental process. The pathology, particularly in more subtle cases, was best appreciated with GFAP and NeuN immunostains. TLS is reminiscent of the laminar atrophy observed at postmortem examination in patients following status epilepticus in whom neuronal loss and gliosis involving ‘middle cortical layers’ (18), or occasionally layers V and VI (19), is observed. In experimental models of TLE, such as the pilocarpine model, neuronal loss is also apparent in layers II/III and in layer IV (20). Neocortical atrophy was a relatively frequent finding in one of the first temporal lobe surgical series of Cavanaugh and Meyer (1), present in 17 of 19 cases with HS. In their series, 13 had a history of status epilepticus, which was thought to be important in the causation of both the HS and the TLS. Furthermore, in that series, IPIs were documented, some of which occurred early (1). In the present series, although episodes of status epilepticus were less frequent, IPIs were frequently recorded; the majority (73%) were febrile seizures in early life; 27% of these were prolonged (status epilepticus). Overall, a history of FS in this series was significantly more frequent in cases with than in cases without TLS. One possibility therefore, is that the changes of TLS are a manifestation of an enhanced vulnerability of superficial cortical neurons to an early cerebral event in the maturing neocortex of a subset of patients.

Some of the architectural features we observed in cortical layer II of TLS cases suggest a disturbance in normal cortical development or maturation as occurs in cortical malformations in human and animal models (15, 21-23). Indeed, we and others have previously interpreted these features as a form of cortical dysplasia or microdysgenesis (10, 11); they do not, however, easily fit into current classification schemes for cortical dysplasia (15). Layer II hypercellularity may be observed in more severe cortical malformations, including hemimegalencephaly (24, 25), polymicrogyria (26), FCD II (27) (Fig. 2h) and experimental models of cortical dysplasia (28, 29). Layer II is the last cortical lamina to be formed during development. Maturation of both the hippocampus (30) as well as the neocortex continues following birth, likely continuing into adolescence (31). It is conceivable that an early IPI or seizure event could interfere with such a process of late cortical development and maturation, resulting in layer II architectural disorganization in addition to neuronal injury, TLS and HS. In our experience and based on this series, this type of layer II architectural disorganization is always observed in the setting of TLS and HS and not in isolation. We also noted that numbers, overall orientation, morphology and distribution of inhibitory interneurons, including calretinin-positive cells that predominate in the upper cortical layers, appear to be preserved in this condition, as noted in previous studies (10, 11). This suggests that, within the superficial cortical layers, specific subsets of late, radially migrating neurons are preferentially affected. As such we consider the layer II architectural abnormalities as ‘acquired’ or re-organizational dysplasia or cortical dysmaturation that is entwined with the process of TLS/HS and distinct from other FCD types. We therefore advocate the routine use of NeuN for the assessment of such cytoarchitectural abnormalities in the temporal lobe, for the identification of TLS and its distinction from other cortical dysplasia types.

We noted variability in the distribution of specific features among TLS cases. One common pattern showed an anterior-posterior gradient, the temporal pole being more severely affected. In previous MRI studies, temporopolar or anterior temporal lobe abnormalities in HS, including volume loss, increased T2 signal and loss of demarcation between the gray and white matter in this region have been identified (3-6). The pathological correlates of these changes remain to be clearly characterized but an abnormality of white matter myelination has been suggested (3, 5, 6) . In vivo qualitative MRI identified temporal lobar abnormalities in only 1 patient in the present series, suggesting that the cortical atrophy associated with TLS may be subtle and require quantitative MRI for detection. For example, a previous sub-regional quantitative MRI study in HS demonstrated the greatest volume reduction to involve the ipsilateral temporal pole compared to other temporal gyri (32). We have, however, recently been unable to detect any temporal lobe changes in individual patients in a subgroup of the TLS patients included in the current study using voxel-based-morphometry and statistical parametric mapping (Eriksson et al, in press).

EEG studies using depth electrodes have also supported a pivotal role for the temporopolar cortex in TLE with preceding or concurrent discharges with those from the hippocampus arising from this site (33). Preferential involvement of temporopolar structures in TLS/HS may relate to known connections between the temporal pole and the hippocampus (34); these pathways have also been implicated in extra-temporal cortical abnormalities in HS as seen on MRI (8). The temporopolar cortex has a distinct cytoarchitecture corresponding to Brodmann area 38 and has connections with the perirhinal, entorhinal cortex and parahippocampal gyrus (35) and amygdala, which are also regions known to show variable sclerosis in patients with HS/MTS (36, 37). In the present series there were limited samples of amygdala and parahippocampal gyral tissue available for any assessment of associated pathology of these regions. The hippocampus, however, more often showed severe degrees of sclerosis in TLS patients that could indicate greater susceptibility to neuronal injury in this subset. We also noted sulco-gyral variations in the extent of neuronal loss in some cases, as also noted in early reports (17). This pattern was initially interpreted as compatible with ‘anoxia’ and comparisons drawn with the ulegyric distribution of cortical infarction seen in neonates. This may indicate common pathogenic pathways for ulegyria and TLS as a result of an early insult. An alternative hypothesis is that it reflects chemoarchitectonic patterns, for example, the sulcal-gyral variation in N-methyl-D-aspartic acid (NMDA) receptors on layer II pyramidal neurons that we have previously shown (38), or regional variations in vulnerability to neuronal injury.

In addition to anatomical distribution within the temporal lobe, the severity of TLS may also be variable; therefore, it is conceivable that subtle cases of TLS with mild focal neuronal loss are overlooked. Quantitative methods (as used in the present study) may prove to be more sensitive and objective methods for the routine detection of the abnormal cytoarchitecture in subtle TLS, which could then also be more precisely localized and correlated with pre-operative quantitative neuroimaging (39). Here, we selected the middle temporal gyrus 1 cm caudal to the pole as a comparable anatomical area for measurement between cases and controls, as in previous studies (38). Since this region was not always affected by TLS, some cases failed to show a reduction with field fraction measurements. Therefore, targeting these quantitative measures to the cortical region of maximal perceived neuronal loss by qualitative inspection (with comparison to control values for that region) may be more appropriate for assessment of individual cases. At the other end of the spectrum, some cases showed severe and extensive neuronal loss throughout the temporal lobe. It is conceivable that this process of laminar sclerosis may extend to extra-temporal regions and this would be relevant to continuation of seizures after anterior temporal lobe resection. In the present series, however, there was no suggestion of poorer post-surgical outcome in patients with TLS, with 73% being seizure-free at 2 years.

There are several limitations to this pathological study. Although we reviewed a large series of cases, the number of TLS cases identified is relatively small. In addition, it is possible that episodes of status epilepticus as well as IPI were inaccurately or under-reported. As with all quantitative and immunohistochemical studies there is an assumption that fixation, processing and staining methods are comparable between cases; wherever possible we have aimed to standardize these steps.

In summary, we report the neuropathological features of TLS in 11% of patients with TLE and HS. We consider this to represent part of the spectrum of MTS rather than a second or dual pathology. The associated layer II dysplastic features are likely to be re-organizational changes as part of the process of TLS rather than a primary or separate cortical malformation. In this series, childhood febrile seizures were a frequent event that could be of etiological significance. The presence of TLS does not appear to influence outcome after surgery but these initial clinical, radiological and pathological correlations could be addressed in future larger series of patients.