Abstracts

Rationale: Experimental evidence from animal models demonstrates that disruptions in thalamic reticular nucleus (TRN) function during wakefulness reduces selective gating of redundant sensory inputs. This "sensory gating" (SG) safeguards the integrity of higher cognitive functions from irrelevant stimuli. The TRN also initiates sleep spindles, cardinal sleep oscillations during stages 2 and 3 non-rapid eye movement (NREM) sleep. Whether sensory gating can be measured in the human thalamus and how it relates to attention is not known. We evaluated electrophysiological responses to a paired auditory stimulus in the thalamus and evaluated whether sensory gating predicted performance on an attention task in patients with drug refractory epilepsy.

Methods: Subjects undergoing direct thalamic recording during epilepsy surgery evaluation were prospectively enrolled (n=26). Thalamic stereoEEG (SEEG) electrode localization was confirmed by manual co-registration of post-operative head CTs or MRIs and pre-operative MRIs and Freesurfer segmentation. Thalamic SEEG recordings from the pulvinar (Pul), centromedian nucleus (CM) and anterior nucleus (ANT) were analyzed referenced to ipsilateral white matter. Scalp EEG recordings from FZ were analyzed referenced to the second cervical spinous process. Subjects passively listened to 240-300 paired auditory stimuli comprised of an initial auditory click, stimulus 1 (S1), and a second click, stimulus 2 (S2) separated by 500 msec, with intertrial intervals of 8-10 sec. Data were bandpass filtered (6.0-50.0 Hz) and epoched by trial. Trials were rejected if the peak-to-peak amplitude exceeded ±100 µV, the remaining trials were averaged, and the latency and amplitude of the P50 auditory evoked response was recorded. Sensory gating was computed as the ratio of the P50 response to S2 versus S1. The Connors Continuous Performance Task (CCPT) was performed following the SG task. Sleep spindle rates were estimated during stages 2 and 3 NREM sleep from the overnight record prior to the SG task using a validated automated sleep staging algorithm and spindle detector.

Results: A clear thalamic P50 response to S1 preceded the scalp response by 18 ms on average (p< 0.001; by thalamic region: CM -18 ms, p< 0.001; ANT -19 ms, p< 0.001; Pul -25 ms, p< 0.001). SG was detected (ratio< 1) in the scalp (0.58, p< 0.001) and in the thalamus (0.60, p< 0.001; CM 0.64, p< 0.001; ANT 0.50, p< 0.001; Pul 0.72, p=0.024). The SG ratio in the ANT trended to be lower than in the Pul (p=0.040). The auditory SG ratio decreased as the distance from the thalamic medial geniculate nucleus increased (p=0.015). Thalamic SG (p=0.048), but not scalp SG (p=0.14), predicted accuracy on the CCPT. Both thalamic (p=0.019) and scalp (p=0.027) overnight sleep spindle rate predicted accuracy on the CCPT. Thalamic SG predicted thalamic spindle rate (p=0.003), but scalp SG did not predict scalp spindle rate (p=0.4).


Conclusions: These results demonstrate that thalamic reticular nucleus function mediates performance on an attention task and can be assessed using thalamic SG or non-invasive sleep spindle measures.

Funding: NIH NINDS R01NS115868