Abstracts

Rationale: Studies involving mathematical analysis of the EEG signal currently allow no definite conclusions as to whether seizures are predictable by prospective algorithms with no evidence for above chance prediction (Mormann F et al. Brain 2007, 130, 314-333). Subsequently, with intracranial electrodes, seizure prediction algorithms provided above chance predictions of an impending seizure in 9 of 15 patients with variable sensitivity. Progress is being made in seizure prediction using large EEG databases and applying a variety of mathematical techniques.

In patients with focal seizures, changes in cerebral blood flow/oxygenation occur minutes before seizure onset, demonstrated in SPECT, fMRI, subdural thermal diffusion flowmetry and near-infrared spectroscopy studies. Transcranial magnetic pulse stimulation shows increased preictal cortical excitability. Animal models of focal seizures show preictal increases in brain temperature.

Heat is released during neuronal activation and reflected in tympanic membrane temperature change. For example, cognitive tasks result in temperature changes detected at the tympanic membrane in chimpanzees. The middle ear is heated passively by the brain and tympanic membrane temperature provides a surrogate for brain temperature change.

For patients with uncontrolled epilepsy, the unpredictability of seizures is particularly disabling. Risk of serious injury and feeling of helplessness impacting daily life. If seizure occurrence could be predicted, it might allow for therapeutic interventions that could have a major impact on the patient’s quality of life.

Seizure prediction requires that a reliable, timely, signal be provided to the patient of an imminent seizure. The methodology would have to be relatively simple, affordable, noninvasive for most patients, and usable in the ambulatory setting.

We therefore sought to determine whether a preictal temperature change could be detected by monitoring epitympanic temperature.


Methods: Patients with focal frontotemporal seizures undergoing VET in the EMU were recruited. A thermistor-based ear canal temperature sensor was placed ipsilateral to the presumed seizure focus. Temperatures were recorded at 1 second intervals synchronously with VET data.


Results: Preictal temperature recordings were available for 22 seizures in 11 patients with a mean age of 37.5 years. Mean baseline ear-canal temperature was 33.94 ± 1.224 °C. Mean peri-ictal peak temperature was higher (p< 0.001) by 0.433 ± 0.348°C (range 0.11- 1.44). Mean temperature at seizure onset was higher than baseline (p< 0.001) by a mean of 0.321 ± 0.258°C (range 0.16-0.76). Increase in temperature started 841 seconds before seizure onset (range -102 to 2816 seconds). 5 patients were awake at seizure onset and 17 asleep.