Sensory gating in intracranial recordings — The role of phase locking

Abstract

For patients with schizophrenia, a deficient gating (or filtering) of sensory input has been described. One major approach to study this sensory gating is to measure event-related potentials (ERPs) in response to paired clicks. In these experiments, sensory gating is quantified as amplitude reduction of the ERP components P50 and N100 from the 1st to the 2nd stimulus. In ERP studies brain electrical signals are averaged over single trials. Alterations in phase locking might be one factor contributing to the observed deficits in sensory gating, but findings have been inconclusive as yet. In particular, the contribution of different frequency bands to the deficit required further investigation. We studied N100 gating by intracranial recordings in a sample of epilepsy patients and subdivided the group into good and poor gators of the intracranial ERP component N100. Data were evaluated by frequency specific wavelet-based phase and power analyses. Poor N100 gators had an increased phase locking in the frequency range from 6.0–15.1 Hz after the 2nd stimulus, as compared to good gators. Other group differences were apparent already before the 2nd stimulus. Poor gators had less phase locked beta band activity (20.2–30.0 Hz) than good gators 200–315 ms after the onset of the 1st stimulus. Within the group of poor gators, lower values of phase locking in this frequency range were also associated with lower gating ratios. The reduced beta band response in response to the 1st stimulus may reflect poorer memory encoding of the 1st stimulus in poor gators. This in turn might lead to increased demands to process the 2nd stimulus.

Introduction

Sensory gating has been conceptualized as the ability to screen out and filter excess and trivial stimuli (Clementz et al., 1997, Freedman et al., 1997). Deficits in sensory gating, as investigated by event-related potential (ERP) studies in paired clicks experiments, have repeatedly been described in schizophrenia (Bramon et al., 2004). In these experiments, sensory gating is commonly quantified as suppression of ERP amplitudes by the stimulus repetition. The deficit in schizophrenia patients is manifested usually as a reduced suppression of the P50 component, but more recently also as a reduced suppression of the N100 component, as compared to healthy populations (Freedman et al., 1983, Blumenfeld and Clementz, 1999, Young et al., 2001, Boutros et al., 2004, Brockhaus-Dumke et al., 2008). In the current study, we aimed to investigate the neurophysiological basis of poor N100 gating by intracranial recordings in epilepsy patients.

In animal experiments, sensory gating was shown for single unit activity and local field potentials in various brain regions, as the hippocampus, prefrontal cortex and striatum, (Moxon et al., 1999, Mears et al., 2006, Cromwell et al., 2007). However, the relationship between both kinds of activity was weak or absent (Mears et al., 2006, Cromwell et al., 2007). Thus, animal data suggests that the reduction of ERPs by stimulus repetition cannot be referred primarily to changes in single unit activity. Since ERPs are, by definition, identified by averaging the single trial responses of the electroencephalography (EEG) signal after the stimulus, the amplitude of the ERP represents changes in the EEG that are phase locked to the stimulus. Impaired gating might therefore be related to alterations in phase locking.

In fact, when the temporal variability of the single trial response of schizophrenia patients was compared to control subjects, there was greater temporal variability for the P50 elicited by 1st stimulus (S1) (Jin et al., 1997, Patterson et al., 2000). By calculating the phase locking of on-going EEG oscillations for single frequencies, it was found that schizophrenia patients produced significantly less phase locking after the 1st stimulus in the theta and alpha band range than healthy controls did (Jansen et al., 2004, Brockhaus-Dumke et al., 2008).

Although all these studies reported a similar loss of phase locking in response to S1 in schizophrenia patients as compared to healthy controls, studies differed with regard to the reported ERP amplitudes. While P50 amplitudes to S1 were reduced in patients in two of the studies (Jin et al., 1997, Patterson et al., 2000), they were virtually unaffected in the two other studies (Jansen et al., 2004, Brockhaus-Dumke et al., 2008). Furthermore, the reduction of P50 amplitudes to S1 and a deficit in gating are not necessarily associated (Freedman et al., 1983, Moxon et al., 2003) and a reduced P50 gating in schizophrenia patients can be observed in absence of a reduction of P50 amplitudes to S1. Thus, the variance in the ERP of schizophrenic patients makes it difficult to assign relationships between gating and other measures of neuronal activation.

To overcome the variability when comparing schizophrenic patients to normal controls, Young et al. (2001) divided their schizophrenia patient sample into those with good and poor N100/P200 gating in order to address this issue. Young et al. (2001) determined the N100 peak latencies in single trials. The standard deviation of the latency measurements across trials was defined as N100 latency jitter. The authors reported that subjects with good gating had less N100 latency jitter to S1 stimuli as compared to subjects with poor gating and, moreover, that this latency jitter was inversely correlated to the S1 N100/P200 amplitude. However, similar as from the above cited studies, it is difficult to deduce from this study what determines poor gating in absence of S1 deficits.

Further complicating the picture, a loss of phase locked activity after S1 was found to contribute to the reduced amplitude of the evoked potential after S2: Hong et al. (2004) specifically explored evoked beta (14–26 Hz) and gamma (30–50 Hz) activity after S1 and their relationship to the P50 response after the 2nd stimulus (S2), without directly addressing the issue of phase locking. The authors found that 59% of the S2 P50 variance in patients was explained by gamma and beta activity after S1. Thus, the magnitude of the P50 S2 response in patients was determined in major parts by activity before S2 onset.

In the current study, we examined the correlates of poor gating in more detail. We focused on N100 gating because of the increasing interest of clinical researchers in this functional measure and also because of the fact that N100 activity dominates auditory ERPs. In this study, we took advantage of intracranial recordings which have a superior signal-to-noise ratio as compared to scalp recordings. In addition to phase locking and the amount of event-related (phase locked) activity, we analyzed induced activity which is event-related, but not phase locked. Because of its lack of phase locking, induced activity is not reflected in the ERP. The relation between induced activity and gating of ERP components has not been explored as yet. However, it appears possible that poor gating is associated with increased levels of induced activity, indicating a systematic stimulus-related hyperactivation.

In this context we were also interested in induced high frequency gamma band activity (GBA), comprising activity up to 200 Hz, which has only been observed in intracranial recordings (e.g. Crone et al., 2001, Trautner et al., 2006). We hypothesized that induced high frequency GBA reflects “ripple” activity. Ripples are believed to be due to simultaneous excitation of pyramidal cells and interneuronal networks and represent IPSPs on the somata of the pyramidal cells (Buzsaki et al., 1992, Chrobak and Buzsaki, 1996). Therefore, ripples comprise an important inhibitory component and might presumably also be related to sensory gating.

For the purpose of the current study, we analyzed N100 activity, recorded from the posterolateral surface of the temporal lobe of epilepsy patients who underwent invasive presurgical evaluation. We divided the sample in good and poor N100 gators. Time-frequency transforms were calculated in order to compare event-related (phase locked) activity, phase locking and induced activity between the two groups. Areas of significant differences were subjected to a correlation analysis, in order to further understand the relation between frequency specific signals and N100 gating.