Abstracts

Rationale: Rodent models of Temporal Lobe Epilepsy (TLE) show marked deficits in episodic memory as shown by performance in spatial navigation tasks which are accompanied by disruptions at the single-unit level. Patients with TLE also have cognitive deficits, and in order to test whether these deficits parallel what is seen in rodents we have developed virtual spatial navigation tasks that mimic tasks performed in rodent models. These virtual navigation tasks allow us to investigate behavioral performance in spatial tasks while recording both local EEG and single units.Methods: Local EEG activity and single units were recorded from depth electrodes in patients with TLE while they were performing spatial navigation tasks. We designed two virtual environments to mimic classical rodent spatial navigation tasks, coin chasing and virtual water maze (vMWM). The first task, coin chasing, consisted of a rectangular room with a prominent visual cue on one side. The virtual arena contained 64 coins that were spawned one at a time in one of the possible 16 spawn points. This allowed the patient to explore the entirety of the virtual space as would be done in a rodent pellet chasing tasks that are performed to record place cell activity. The second task was the vMWM, which consisted of a rectangular environment with 4 prominent buildings and a large circular pool in the middle. The player was virtually dropped into the pool and asked to find a hidden platform submerged in the water four consecutive times for three blocks followed by a 2-minute probe trial. Performance in the vMWM was also analyzed on control subjects. Single unit firing and local field potentials were recorded from up to 32 microwires implanted in the mesial temporal structures during the virtual navigation tasks to investigate spatial firing patterns.Results: Recordings from 8 patients that were implanted either bilaterally (n=4) or unilaterally (n=4) in the mesial temporal structures were analyzed for both tasks. All patients were able to complete the coin chasing and the virtual MWM task. Cox regression analysis shows that controls were 3.6 times (95% CI: 1.7 to 8.3) more likely than patients to find the platform (p=0.001). Concurrent EEG recordings from depth electrodes showed increased power in the theta and alpha bands but not the delta band during navigation in the virtual environment. Preliminary analysis of single unit firing shows that units were scarce during the spatial navigation tasks with ~6 units per depth electrode per patient. . There have been no clear spatial firing patterns of units that have been analyzed thus far. Conclusions: Patients show a robust deficit in performance in the vMWM that mimics results seen in rodent models of TLE. Changes were seen in theta and alpha bands of local EEG. Single units were recorded during these spatial tasks; however, a relationship between spatial coordinates and single unit activity is difficult to establish. Studying virtual navigation in patients with intracranial electrodes and recording single unit activity is feasible.