Article Text

This article has a correction. Please see:

Evolution of MRI changes and development of bilateral hippocampal sclerosis during long lasting generalised status epilepticus
  1. B Pohlmann-Eden1,2,
  2. A Gass1,
  3. C N A Peters1,
  4. R Wennberg2,
  5. I Bluemcke3
  1. 1Department of Neurology, Mannheim Hospital, University of Heidelberg, Germany
  2. 2Krembil Neuroscience Centre, Toronto Western Hospital, University of Toronto, Canada
  3. 3Department of Neuropathology, University of Erlangen, Germany
  1. Correspondence to:
 Dr B Pohlmann-Eden
 Toronto Western Hospital, University of Toronto, 5W-443, 399 Bathurst St, Toronto, ON M5T 2S8, Canada;


This report describes a previously healthy 28 year old patient with a 5 month period of intractable generalised status epilepticus (SE) of unknown aetiology with fatal outcome. Repeated magnetic resonance imaging (MRI) showed no pre-existing abnormality, but did show progressive cortical and hippocampal atrophy and T2 hyperintensity in both hippocampal formations, suggestive of progressive tissue damage. Post-mortem histopathological analysis revealed substantial neuronal cell loss including CA1 and CA4 sectors of the hippocampus compatible with bilateral hippocampal sclerosis. There was no evidence of systemic complications including arterial hypotension and hypoxia, hypoglycaemia, hyperpyrexia, or other confounding factors to account for these findings. This case provides further evidence of SE induced hippocampal damage in humans.

  • status epilepticus
  • magnetic resonance imaging
  • hippocampal sclerosis
  • GTCS, generalised tonic−clonic seizure
  • HC, hippocampus
  • HS, hippocampal sclerosis
  • MRI, magnetic resonance imaging
  • SE, status epilepticus
  • TLE, temporal lobe epilepsy

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Hippocampal sclerosis (HS) is the most common pathological finding in chronic temporal lobe epilepsy (TLE). Whether hippocampal atrophy and neuronal cell loss is the cause or consequence of repeated seizure activity has been a controversial debate for almost a century. Experimentally, Meldrum and coworkers1 first described selective neuronal damage in the hippocampus (HC) in a model of SE induced by bicculline, and showed that prolonged seizure activity alone could lead to excitotoxic cell death. Meanwhile, numerous animal studies using different SE models such as kainic acid,2 pilocarpine,3 or electrical stimulation4 demonstrated that long lasting seizure activity resulted in selective neuronal loss in the HC with a similar distribution as in humans, mostly affecting CA1 and CA3 regions of the HC (for overview see Lado et al,5 Blümcke et al6). According to new experimental data, even repeated brief seizures may induce neuronal loss and a pattern of hippocampal damage similar to that found in TLE.7

Thorough documentation of patients with SE induced brain damage is still rare. A recent magnetic resonance imaging (MRI) study in a small series of children provided some evidence that prolonged febrile seizures may lead to hippocampal atrophy, although the role of pre-existing traumatic, developmental, and genetic factors remained uncertain.8 In a retrospective, neuropathologically confirmed study of three cases with HS and SE lasting up to 3 days, where confounding factors could be ruled out, Fujikawa et al9 described widespread neuronal loss in the HC, amygdala, dorsomedial thalamus and cerebellar Purkinje cells, and the entorhinal and periamygdaloid cortex. A small number of clinically and radiologically well documented case studies using MRI changes over time have provided further evidence that HS may occur as a result of neuronal damage induced by seizures during SE.10–14 These studies describe an initial pattern of T2 signal increase and associated swelling in the HC without contrast enhancement, interpreted as focal peri-ictal oedema.8,10–17 Although progressive hippocampal atrophy on follow up MRI scans was only found in single cases,10–13 three patients had additional pathological proof of neuronal loss in the HC, amygdala, and claustrum.11,12,14 Some of these reported cases suggest that HS can develop within 2 months of SE and may further progress during the following 3 to 4 years.11,13 Even repeated single brief generalised tonic−clonic seizures were reported to lead to HS according to MRI criteria.18


We report a case of a 28 year old man with no previous neurological or medical history, no contact with infectious diseases, and no history of drug or alcohol withdrawal. He was a product of a normal delivery. The family history for epilepsy was negative. There was no evidence of delay in psychomotor development, previous head trauma, encephalitis, febrile convulsions, or other predisposing conditions. He presented to the emergency room with a first generalised tonic−clonic seizure (GTCS), followed by a series of GTCS on day 1, developing into a drug resistant generalised SE on day 2 that prompted treatment with general anaesthesia. The initial post-ictal clinical examination revealed no focal signs. Long term continuous EEG monitoring, immediately started with the diagnosis of SE, repeatedly showed generalised seizure activity consisting of asynchronous irregular 4–5/s spike wave and sharp wave transients with shifting predominance without clear focal onset. Medical treatment included the entire range of established standard anti-epileptic strategies (benzodiazepines, phenytoin, valproate, all barbiturates, lidocaine, propofol, ketamine chloride, clomethiazol, and oral lamotrigine, carbamazepine, topiramate, all at high doses, as single treatment or as combination therapy) as well as other therapeutic strategies such as high dose corticosteroids and even intermittent electroconvulsive therapy in the final stage of the SE. The SE could be only controlled by deep GA paralleled by classical burst suppression EEG. At no point did the patient develop significant hyperpyrexia, arterial hypotension, hypoxaemia, or hypoglycaemia. The patient died 5 months later most likely due to ICU complications including pancytopenia, with no evidence of the symptomatic aetiology underlying his SE having been discovered.

Extensive serum analyses including thorough screening for toxic agents such as domoic acid and repeated CSF samples were unrevealing. Autoimmunological parameters such as anti-nuclear antibodies, anti-neutrophil cytoplasmic antibody, and myelin antibodies were negative, as were all antibodies and complement binding reactions against tuberculosis, Borrelia, Toxoplasma, Candida and Actinomyces, human immunodeficiency virus, herpes simplex virus, cytomegalovirus, Epstein-Barr virus, varicella zoster virus, polio, rabies, and rubella. Initial CSF cell count increased from 9 cells/mm3 to 37 cells/mm3 during three follow up investigations.

MRI scans (Proton Density, FLAIR, T1 and T2 weighted (+ contrast)) were performed four times during the hospital stay. Initial MRI on the day of admission was normal. Repeat MRIs after 4, 10, and 16 weeks demonstrated the development of generalised atrophy (fig 1A–C) and T2 hyperintense signal changes after 4 weeks in the hippocampal formations (fig 1D).

Figure 1

Demonstration of the development of generalised, predominantly supratentorial, cortical atrophy, on T2 weighted MRI over 3 months. (A) Week 4, (B) week 16. FLAIR images at week 10 show evidence of flattened small hippocampus in coronal plane (C) and hyperintensity in the head of the hippocampus in the transverse plane (D).

Neuropathological investigation demonstrated widespread neuronal cell loss and astrogliosis in neocortex, basal ganglia, cerebellum, and especially in the hippocampi bilaterally, most pronounced in the CA1 and CA4 regions (fig 2). In contrast, dentate gyrus granule cells and CA2 pyramidal cells were not affected. Hippocampal pathology was reminiscent of an early stage of Ammon’s horn sclerosis in patients with intractable temporal lobe epilepsy. Thorough examination of all brain structures revealed no specific findings in regard to the presumed symptomatic aetiology of the SE such as encephalitis, traumatic scar formation, malformations, or neoplasia.

Figure 2

Left hippocampus (Nissl staining): note the extensive neuronal cell loss in CA4 (A) and CA1 (B), whereas granule cells and CA2 pyramidal cells (C) are preserved. Scale bar = 150 μm.


The hippocampal subfields CA1 (Sommer’s), CA3, and CA4 appear particularly susceptible to seizure induced damage, which is presumed to result from excessive presynaptic release of excitatory neurotransmitters, specifically glutamate and to a lesser extent aspartate.19 Abundant glutamate binding sites in rat CA1 and CA3 regions support this concept of glutamate mediated excitotoxicity.20 In the kainic acid animal model of SE (kainic acid is an excitotoxic analogue of glutamate), a time dependency and threshold phenomenon of neuronal damage has been demonstrated. Neuronal cell loss occurred as early as 30 minutes after SE and showed its greatest topographical distribution in long lasting SE.21 Experimental imaging studies showed that intrahippocampal injection of low doses of kainate in mice reproduced the MRI pattern of HS with T2 signal increase, interpreted to result from both the excitotoxic effect of kainate and the ongoing seizure activity.22 A recent lithium−pilocarpine model of SE demonstrated as the earliest MRI change, after 2 hours, a blood−brain barrier breakdown in the thalamus, followed by oedema in the amygdala and the piriform and entorhinal cortices after 24 hrs. Histopathologically confirmed hippocampal cell loss correlated significantly with the evolution of the T2 signal increase.23

In humans, the most frequently observed MRI change in the early stage of SE is either asymmetrical or bilateral T2 signal increase in the HC,10–17 sometimes found after as little as 24 hours,10 often encompassing the entire length of the HC and interpreted as focal transient cytotoxic and vasogenic oedema. High resolution MRI demonstrated involvement of the CA subfields within the HC in one case.12 Documentation of an associated early swelling of the affected HC was shown using different methods of hippocampal volumetry,8,13,16 and by signal increase on diffusion weighted imaging together with focally reduced apparent diffusion coefficients.16 T2 signal increase seems to disappear within a few days after short term SE,16 but may persist in the HC for several weeks and spread to other structures such as the fornices in long term refractory SE.12 Development of hippocampal atrophy proven by MRI in a previously unremarkable HC was reported in both children10 and adults11,13 following long lasting seizure activity.

Not surprisingly, there is a lack of MRI data in the chronic stage of ongoing SE in both animals and humans. As the first follow up MRI on our patient was performed almost 4 weeks after the first normal MRI, it is likely that the initial T2 increase and swelling of the HC had already disappeared and the excitotoxicity related neuronal damage had already begun to occur by this time. At this time point, we found signal increases without swelling in both hippocampal formations, at a maximum in the vicinity of the head of the HC. Serial coronal T1 weighted and FLAIR images showed significant volume loss in both hippocampi between week 4 and 10, most likely representing neuronal loss and astrogliosis as the correlate of the later histologically proven HS. The selective vulnerability of the hippocampal formations is compatible with the outlined concept of excitotoxicity with a critically lowered threshold for glutamate toxicity in this stage of ongoing and widespread seizure activity. However, this topographical pattern is unspecific and shares certain characteristics with observations in post-hypoxic damage,24 such that a mechanism of repeated hypoxic injury due to ictal overactivation and metabolic compromise in a vulnerable tissue compartment could also be considered.

In summation, both the neuropathological findings and corresponding MRI features in our case further support the hypothesis that neuronal cell loss in HS may occur as the result of prolonged severe seizure activity.


Supplementary materials

  • Publisher Correction
    Please note that there is an error in the author list. The correct listing is shown here:

    B Pohlmann-Eden, A Gass, C N A Peters, R Wennberg, and I Blumcke

    The error is much regretted. A print erratum will be published shortly.


  • Competing interests: none declared

Linked Articles

  • Correction
    BMJ Publishing Group Ltd