Sclerosis of the hippocampus is one of the most common and best characterized pathologies identified at both postmortem (Margerison & Corsellis, 1966; Meencke et al., 1996) and in surgical series of patients with epilepsy, and is particularly associated with the syndrome of mesial temporal lobe epilepsy (MTLE; Blumcke, 2009). In surgical series, hippocampal sclerosis (HS) is typically encountered in specimens from young adults, in the context of refractory seizures, with sclerosis visible on magnetic resonance imaging (MRI) and confirmed in resected specimens (Wieser, 2004). Neuropathology studies have, from the outset, recognized that more extensive, albeit more subtle, pathology may accompany HS (Cavanagh & Meyer, 1956). In recent years, clinical and neuroimaging studies have also moved away from a “hippocampocentric” view of MTLE (Thom et al., 2010a; Engel & Thompson, 2012). The additional structures altered include those anatomically linked to, or in proximity of, the hippocampus, such as the amygdala (Bernasconi et al., 2003), entorhinal cortex (Bernasconi et al., 1999; Jutila et al., 2001; Bernasconi et al., 2003; Bonilha et al., 2004; Keller et al., 2004), temporal pole (Coste et al., 2002; Sankar et al., 2008), and cingulate gyrus (Bernhardt et al., 2008; Bonilha et al., 2010a). This wider network of disease may correlate with an extended zone of epileptogenesis and explain poor outcomes in some patients following focal surgery, as well as offer an explanation for comorbidities associated with HS as progressive memory decline.
Neuropathologic study of the thalamus is merited in patients with HS for several reasons. The hippocampus has important reciprocal connections to the thalamus (Hirai & Jones, 1989; Herrero et al., 2002). The main output pathway of the hippocampus (fornix) projects through the medial mamillary nucleus to the anteroventral thalamic nucleus (AV); the anterior and lateral dorsal thalamic nuclei are both reciprocally connected with the limbic cortex; the mediodorsal nucleus (MD) also receives input from the amygdala, entorhinal cortex, and temporal pole (Nieuwenhuys et al., 2008a). A number of experimental studies have indicated that the thalamus plays a crucial role in seizure initiation and modulation, thereby strongly suggesting that it may be an important part of the substrate for MTLE (Bertram et al., 2008; Sloan & Bertram, 2009; Sloan et al., 2011). In addition, there are data from electroclinical studies supporting synchronization of activity in the thalamus in association with temporal lobe seizures (Guye et al., 2006). The neuroimaging literature also points toward volume loss occurring in the thalamus in TLE, although there are inconsistencies between studies regarding its association with HS, whether greater volume loss is observed ipsilateral to side of HS, as well as the regional distribution of pathology within the thalamus (DeCarli et al., 1998; Dreifuss et al., 2001; Behrens et al., 2003; Natsume et al., 2003; Bonilha et al., 2004; Labate et al., 2008; Kim et al., 2010; Mueller et al., 2010; Alhusaini et al., 2012).
Detailed histologic studies of the thalamus in TLE are lacking. Neuronal loss has been demonstrated in experimental models (Bertram & Scott, 2000; Bertram et al., 2001). In a single postmortem study from 1966, thalamic damage was frequent in a series of patients with TLE and HS (Margerison & Corsellis, 1966), but not further detailed. Our aim was to investigate the extent and regional distribution of thalamic pathology in a postmortem epilepsy series using quantitative immunohistochemistry.
- Top of page
There is converging clinical, experimental, and neuroimaging evidence that hippocampal sclerosis in epilepsy may represent part of a wider network of pathologic changes, including within the thalamus. In this first detailed neuropathologic study of the thalamus in epilepsy, we have demonstrated that, in contrast to the stereotypical patterns of sclerosis (neuronal loss and gliosis) observed in the hippocampus, cellular alterations within subthalamic nuclei are milder, less clearly defined, varying both in regional distribution, nature, as well as severity. It remains to be established whether any of the extended regional pathologies accompanying HS are as a result of the same initial “insult,” represent secondary progressive changes, such as retrograde/anterograde degeneration along known projections, or arise independently as a result of the seizures. Indeed all of these mechanisms may operate. We employed three standard neuropathologic measurements in this initial assessment of the thalamus to quantify total synaptic density as a measure of connectivity (synaptophysin), neuronal number (cresyl violet), and gliosis (GFAP) in selected subnuclei. These measurements did not correlate with each other (except for the LD nucleus), suggesting that the pathologic processes of astrocytic gliosis and alterations in synaptic and neuronal density may not occur synchronously.
We did not demonstrate significant differences between all the patients with epilepsy compared to controls. Indeed for some subnuclei, higher neuronal densities were noted in epilepsy groups than in controls. A possible explanation for this is that gliosis (and therefore tissue volume reduction) may predominate over neuronal loss, resulting in a paradoxical increase in cell density (cell number per area). When considered as groups, based on the presence or not of HS, some patterns did emerge. Several observations pointed to regional pathology preferentially involving the MD in UHS, with evidence of lateralization to the side of sclerosis. We demonstrated significantly lower mean neuronal density ipsilateral to side of sclerosis in UHS cases in the MD compared to the contralateral side. Within individual HS cases, greater GFAP, lower synaptophysin and neuronal density between sides were more consistently seen in the MD than other nuclei, lateralizing to the side of sclerosis. Such comparisons within individual cases may prove a more reliable measurement than comparison between postmortem groups, as technical variations in staining dependent on tissue quality and fixation times are circumvented; in addition, the anatomic coronal level through thalamic subnuclei are more exactly matched when sampling tissue blocks. These findings could support the notion that pathology in the MD is more consistently present in HS associated with epilepsy.
The MD receives incoming networks from the amygdala, entorhinal cortex, and temporal pole (Nieuwenhuys et al., 2008a) and is connected to the prefrontal cortex; it has physiologic roles in cognition and working memory (Watanabe & Funahashi, 2012). There is experimental evidence supporting a role for the MD in epilepsy.
Cassidy and Gale reported that pretreatment of the MD with inhibitory agents protected against the development of seizures (Cassidy & Gale, 1998). Bertram also demonstrated seizure activity within the midline thalamus occurring simultaneously with the onset of seizures evoked by hippocampal stimulation in rats (Bertram et al., 2001), as well as induction of hippocampal seizures through stimulation of the MD (Bertram et al., 2008). This was found to be specific to the MD, implicating this thalamic region in the initial stages of limbic seizures. Furthermore, the seizures stimulated from MD were found to quickly generalize to surrounding areas, supporting a role for the MD in seizure spread (Bertram et al., 2008). Histologic studies in experimental models of MTLE have also demonstrated significant neuronal loss in the MD compared to controls (Bertram et al., 2001). In surgical series of patients with TLE, examination of the thalamus is clearly not possible and human histologic studies are dependent on postmortem series. In the only previous postmortem study by Margerison & Corsellis (1966) in 55 patients with TLE, thalamic “scarring” was seen in 25% of cases, more often in association with HS. In this study, they noted that the severity and distribution of damage varied, but with no predilection for a particular nuclear group. However, in their study, quantitative immunohistochemistry was not carried out as in the present study, which may have the power to detect and evaluate more subtle pathology.
Stereoelectroencephalography studies in human TLE have shown synchrony between the thalamus and cortex in patients with TLE (Guye et al., 2006); it has been suggested that the thalamus could act as a seizure amplifier and synchronizer of ictal activity (Guye et al., 2006). Magnetic resonance imaging (MRI) studies have greatly advanced our understanding of associated alterations of the thalamus in TLE. Using manual segmentation methods (Moran et al., 2001; Natsume et al., 2003), voxel based morphometry (VBM) (Bonilha et al., 2004), and diffusion tensor imaging (DTI) (Kim et al., 2010) volume reductions in the thalamus have been shown ipsilateral to the side of seizure onset (DeCarli et al., 1998; Dreifuss et al., 2001; Moran et al., 2001; Natsume et al., 2003; Bonilha et al., 2004; Kim et al., 2010; Mueller et al., 2010; Bonilha et al., 2010b; Alhusaini et al., 2012) associated both with (Alhusaini et al., 2012; Kim et al., 2010; Mueller et al., 2010) and without HS (Natsume et al., 2003). Some MR studies support greater involvement of the anterior thalamus (Bonilha et al., 2005; Mueller et al., 2010), and there are contradictory studies suggesting relative differences in the extent of contralateral thalamic atrophy associated with right- versus left-sided HS (Morgan et al., 2012; Pail et al., 2010). We failed to detect any differences in the lateralization of thalamic pathology between right- or left-sided CHS cases. A more recent MRI study, has mapped and localized thalamic atrophy to the ipsilateral medial thalamic surface (Bernhardt et al., 2012), which is comparable to our identification of pathology preferentially localizing to the MD nucleus. Positron emission tomography (PET) studies have also localized hypometabolism in the ipsilateral MD subnuclei in patients with TLE (Juhasz et al., 1999).
Clinical and imaging studies have also been evaluated in an attempt to gain further insight of the cause and progression of thalamic pathology. Natsume showed a negative correlation between thalamic volume and duration of TLE in 40 patients with TLE (Natsume et al., 2003). Duration of epilepsy was also shown to correlate with progressive thalamic changes observed in patients with HS using DTI (Keller et al., 2012). Not all studies, however, have shown a relationship between loss of thalamic volume and duration of epilepsy (Alhusaini et al., 2012). The question of progressive changes becomes more difficult to address in postmortem tissues from patients mainly at the end-stage of disease, but we did identify an association between the extent of gliosis in the VL and AV and total duration of epilepsy. A similar association was not seen for age at death, suggesting that this may represent a seizure-related effect.
A further proposal, that thalamic pathology represents transsynaptic degeneration following secondary deafferentation in hippocampal networks, was supported in one recent study combining volumetric MRI study with tractography, suggesting a relationship between alterations observed with these methods (Bonilha et al., 2010b). The evidence for loss of thalamic volume associated with HS in patients with mild (drug-sensitive) TLE (Labate et al., 2008) also supports this hypothesis of atrophy secondary to degeneration in the network rather than a direct seizure effect. As the main output pathway of the hippocampus, via the fornix and medial mamillary nucleus, is to the AV, we might anticipate greater atrophy in this nucleus. Indeed, stimulation of the AV has recently been trialed as an effective treatment in refractory epilepsy of temporal origin (Fisher et al., 2010), with these direct connections between the anterior nucleus and the hippocampus implicated in this therapeutic effect (Fridley et al., 2012). However, in our small series we failed to confirm significant pathology preferentially involving the AV, either in single cases or in groups of HS cases. Finally, MRI studies have shown smaller thalamic volumes with a history of febrile seizures as a potential precipitating insult for thalamic pathology (Natsume et al., 2003); it was not possible to investigate the contribution of initial injury to thalamic pathology in the present clinical series because of the lack of sufficient early historic data in many cases.
There are several limitations in the present study. One of the main ones being that this postmortem group represents a more heterogenous cohort than surgical epilepsy series, particularly in terms of the epilepsy syndrome in that not all patients have MTLE, which may partly explain the lack of consistent thalamic pathology observed. Tissue processing and differences in fixation times can influence the intensity of immunostaining. We attempted to overcome this problem by comparing the left and right sides within individual cases in addition to mean values between groups. Although we aimed to exclude confounding pathologies, we cannot exclude, particularly in older patients, the contribution of subclinical cerebrovascular disease to thalamic gliosis and neuronal loss. Not all thalamic subnuclei were represented in each case, and the thalamic coronal levels at which subnuclei were sampled, varied between cases, which could influence data. In addition, we did not subdivide the nuclei further, which may reduce the sensitivity of detecting focal pathology. A further prospective PM study with sampling through the entire thalamus, utilizing three-dimensional stereologic analysis, may be required to further study the MD in detail.
In summary, our PM study demonstrates that stereotypical pathologic changes, as seen in HS, are not consistently or clearly defined in the thalamus. We did identify a predilection for neuronal and synaptic loss and gliosis in the MD, in keeping with previous clinical and experimental data, supporting the potential role of the MD in epilepsy. This is an area that warrants further detailed investigation.