What is the evidence that SE kills neurons?

Animal Studies

The first evidence that electrographic seizure discharges alone could kill neurons and that SE-induced neuronal death is not simply the result of secondary systemic complications was obtained by Meldrum and colleagues (87, 89), who induced SE with the GABA_A receptor antagonist bicuculline in paralyzed, artificially ventilated adolescent baboons (89). The baboons were killed immediately after 3.4–7.5 hours of SE, and cresyl violet and hematoxylin-eosin staining of brain sections revealed “ischemic cell change” in layers 3, 5, and 6 of neocortex, the CA1 pyramidal cell layer and hilus of the hippocampus, and anterior, dorsomedial, and ventrolateral nuclei of the thalamus. Brown (15) subsequently summarized studies showing “ischemic cell change” (neuronal necrosis) in hypoxia-ischemia by light and electron microscopy, and we and others have shown that the same changes occur in SE (37, 46, 49, 50, 66, 104, 125, 131). These changes are described later (see discussion under What Type of Neuronal Death Is Produced?) and are shown in Figures 36.1, 36.3, and 36.4.

Figure 36.1.  

Acidophilic neurons, indicative of neuronal necrosis, may be found after only 20 minutes of pilocarpine-induced SE in the rat, and increasing the duration of SE produces progressively more necrotic neurons and neuropil edema. (A) Ventral hippocampal CA1 pyramidal cell layer after 20 minutes of SE. Two necrotic neurons can be seen (arrows). (B–F) Ventral hippocampal CA1 pyramidal cell layer after 40 minutes of SE (B, arrows point to necrotic neurons), 60 minutes of SE (C), 3 hours of SE (D), 3 hours of SE with a 24-hour survival period (E), and 3 hours of SE with a 72-hour survival period (F). Progressively more necrotic neurons and more neuropil edema are found with increasing seizure duration. Scale bar = 30µm. These are previously unpublished photomicrographs of hematoxylin-eosin-stained brain regions used in the data analysis by Fujikawa (44).

At the same time, John Olney and colleagues were investigating the neurotoxic effects of exogenous administration of glutamate analogues both in vitro and in vivo (103, 105, 106, 109); the best studied of these analogues, kainic acid, was found to damage neurons in limbic structures by inducing SE (10, 84, 94, 125, 131). Shortly thereafter the muscarinic cholinergic agonist pilocarpine, alone (142) or in combination with lithium chloride pretreatment, which permits a 13-fold lower pilocarpine dose to be used (63), was also found to induce SE that damages neurons in limbic structures; the extent and severity of the neuronal damage were the same whether or not lithium chloride pretreatment was given (24). Once started, SE could not be stopped by administration of the anticholinergic drug scopolamine, suggesting that the seizure discharges had spread beyond the cholinergic system. Finally, sustained, prolonged electrical stimulation of the perforant path (PPS) was found to damage postsynaptic hippocampal neurons, providing conclusive evidence that it is excessive presynaptic activity that kills neurons, and not the neurotoxic effects of chemical convulsants (104, 128).

Human Studies

The evidence that SE kills neurons in humans is in general limited by the lack of information regarding the duration of SE, the presence or absence of complicating systemic factors such as hypotension, hypoxemia, and hypoglycemia, and the duration of survival after SE. In addition, preexisting epilepsy and brain lesions may complicate interpretation of study results. With this in mind, we examine three studies.

In 1964, Norman (101) reported on 11 children, ages 1.4–6 years, who had survived 1.5–14 days after SE; 8 had preexisting epilepsy. Acidophilic neurons were found in cerebral cortex, hippocampal CA1, CA3, and hilus, amygdala, thalamus, cerebellum, and striatum. In this study the duration of SE was not specified and the presence or absence of complicating systemic factors was not known. In 1983, Corsellis and Bruton (27) reported 20 cases, 8 in children and 12 in adults. Six of the 8 children who died were less than 3 years old, and had neuronal loss in hippocampus, cerebellum, thalamus, striatum, and neocortex, especially the middle cortical layers. Only 4 of the 12 adults showed variable degrees of neuronal loss. In these cases the duration of SE and the presence or absence of systemic complications were also unspecified.

We reported on three men, ages 44, 48, and 56, none with preexisting epilepsy, all of whom had the onset of SE in the hospital, with subsequent ICU monitoring (46). Two of the three had no underlying brain pathology; the third had diffuse carcinomatous meningitis. All three patients were unresponsive, with electrographic SE and focal motor manifestations. The duration of SE was 3 days, 8.8 hours, and 2 days in the three patients, and survival after SE was 15, 11, and 27 days. None of the three had hypotension, hypoxemia, or hypoglycemia until the terminal event. Neuronal loss and gliosis were found in all three cases in hippocampal CA1, 2, and 3 pyramidal cell layers, dentate hilus, corticomedial and basolateral amygdaloid nuclei, layers 2–4 of periamygdaloid (piriform) cortex, layers 5 and 6 of neocortex, and Purkinje cells of the cerebellum. The dorsomedial thalamic nucleus showed neuronal loss and gliosis in two of the cases, and layers 2–6 of entorhinal cortex were involved in one patient. The regional distribution of neuronal loss was remarkably similar to the neuronal necrosis we have found in pilocarpine- and lithium-pilocarpine-induced SE and kainic acid–induced SE in rats (43–45, 47, 49, 50).