|Home | About | Journals | Submit | Contact Us | Français|
Background: The P50 event–related potential sensory gating deficit, a failure to inhibit responses to repeated stimuli, is a leading endophenotype for schizophrenia (SZ). Both gamma and beta event–related oscillations (EROs) are major contributors to the auditory P50 response. However, the topographic distribution of gamma and beta ERO responses to initial (S1) and repeat (S2) stimuli and the association of these oscillations with P50 sensory gating are not clear. Methods: A total of 51 schizophrenic patients, 25 unaffected first-degree relatives, and 34 healthy comparison subjects were tested using a paired-click paradigm. Evoked power of gamma- and beta-band responses using wavelet analyses to S1 and S2 stimuli and gating of EROs and P50 were the main outcome measures. Results: A P50 gating deficit was found in patients (P < .001) and at a trend level in relatives (P = .087). Patients showed widely distributed reductions in gamma and beta EROs to S1 stimuli and S2 stimuli, respectively, and impaired gating in both frequencies. Reduced gamma and beta ERO activity in patients was associated primarily with age of onset. Relatives did not differ significantly from control subjects in either EROs power or gating. Gating of P50, gamma, and beta were not significantly correlated (r = .18–.19, P > .05). Conclusions: These results suggest that ERO deficits in gamma to S1 and beta to S2 stimuli and impaired ERO gating are associated with SZ, but are not related to genetic liability for the illness. The components of information processing assessed by gamma- and beta gating appear to be independent from those mediated by P50 suppression.
Sensory gating deficit is a robust finding in patients with schizophrenia (SZ) and a leading endophenotype for the illness.1–3 A “paired-stimulus” paradigm, in which 2 identical auditory stimuli are presented 500 milliseconds (ms) apart, is commonly used to evaluate sensory gating. The P50 auditory event–related potential (ERP) response (ie, amplitude) to the second stimulus (S2) is normally reduced compared with that of the first stimulus (S1). This suppression, also termed “P50 gating,” is conventionally measured as the ratio of S2:S1 amplitude and is thought to reflect an individual’s ability to screen out, or “gate,” trivial or repetitive stimuli in order to protect against information overload.4,5 Patients with SZ and a significant proportion of their clinically unaffected relatives exhibit reduced P50 suppression, suggesting that compromised ability of the brain to filter out extraneous sensory information is associated with the disease and may be an endophenotype.1,2,6–9 Twin data support a significant genetic association between the P50 sensory gating deficit and SZ.8 At molecular level, the P50 sensory gating deficit has been linked to the alfa-7 nicotinic receptor (CHRNA7) locus on chromosome 15q.10,11
The majority of auditory sensory gating studies have used time domain analysis to extract P50 ERPs from the electroencephalogram (EEG). However, the auditory stimulus used to elicit P50 ERPs also evokes brain activity in a wide spectrum of frequency bands. Recent studies employing spectral frequency analyses document the contributions of specific frequency bands to auditory sensory gating processing in healthy subjects and in patients with SZ. Auditory P50 ERPs overlap morphologically with evoked gamma frequency (~40 Hz) activity.12,13 Response to S1 stimuli in the low beta frequency range (~16 Hz) is significantly negatively associated with the P50 response to S2 stimuli.14 Further, both gamma and beta responses to S1 stimuli are significantly correlated with P50 S1 amplitudes.15 These studies suggest that gamma-band (35–45 Hz) and beta-band oscillations (13–30 Hz) contribute to auditory P50 ERP responses, although the precise mechanisms remain to be determined.
It is important to decompose the specific frequency components of ERPs because each event-related oscillation (ERO) may reflect different neuronal properties and associate with different aspects of information processing. ERO in the beta-band has been shown to be a sign of the recognition that a novel or salient stimulus warrants monitoring.15–17 ERO in the gamma-band is thought to reflect the “binding of spatially distinct neuronal activities within and between brain regions” necessary for coherent sensory perception and processing.16,18 Both gamma and beta EROs can be (1) automatically generated in response to repetitive stimulation at that particular frequency (ie, the gamma/beta driving or steady-state response); (2) evoked by specific unique stimuli; and 3) cognitively elicited or induced using a variety of experimental paradigms.
Several findings suggest a dysfunction in the detection and encoding of salient sensory information in SZ. Using the standard paired-stimulus paradigm, SZ patients showed decreased beta activity to S1 stimuli19 as well as reduced activity at low frequencies that include beta.20,21 Further, reduced gamma activity to S1 stimuli has been observed in some samples of SZ,21 but not all,12,20 suggesting a disrupted or inefficient formation of neural assemblies for registering sensory input.
Gating measures can be derived from responses to gamma and beta EROs and have been studied in patients and relatives in one study. Hong and colleagues22 reported that SZ patients exhibited a gamma gating deficit, but not a beta gating deficit. However, they observed no differences between clinically unaffected relatives and healthy controls (HCs) in gating of gamma or beta responses. The findings in relatives suggest that neither measure meets the familiality criterion for an endophenotype.23 Because both gamma and beta EROs potentially contribute to the generation of P50 ERPs, the presence of normal gamma and beta gating in relatives suggests that the underlying cognitive functions measured by the ERO gating responses may differ from those tapped by P50 suppression. However, Hong and colleagues examined only a single scalp location (Cz). Thus, little is known about the topographic distribution of gamma or beta EROs and their associated gating measures in patients and relatives over other scalp locations.
The primary aims of the present study were to address 3 critical issues relevant to the contributions of gamma and beta EROs to P50 sensory gating deficits in SZ: (1) Are gamma and beta EROs to S1 and S2 stimuli reduced in SZ and their unaffected relatives compared with controls, and do the topographic distributions of these ERO responses differ between groups; (2) is gating of gamma and beta EROs impaired in SZ patients and their relatives, and do the topographic distributions of these ERO gating measures differ from those of controls; and (3) are P50, gamma, and beta ERO gating measures significantly correlated.
Groups included 51 individuals who met the criteria for the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition; DSM-IV) for SZ (n = 20; 5 female) or schizoaffective disorder (SA, n = 31; 13 female; 10 patients [6 female] with the depressed subtype and 21 patients [7 female] with the bipolar subtype), 25 of their nonpsychotic biological relatives, and 34 HCs. Demographic characteristics of the sample are presented in table 1. Patients were chronically ill outpatients (mean duration of illness: 18.2 years, range 0.13–40.1 years) and were moderately symptomatic (mean Brief Psychiatric Rating Scale [BPRS] score: 45.0, range 23–85). All but 3 patients were taking antipsychotic medication at the time of testing: 76.5% were on atypical antipsychotics and 23.5% on typical antipsychotics; mean daily dose in chlorpromazine (CPZ) equivalents was 605.3 mg (SD = 485.4). SZ and SA patients did not differ in mean age of onset defined as age at first hospitalization (P = .40), mean duration of illness (P = .33), mean BPRS score (P = .55), or mean CPZ equivalent daily dose (P = .42). The relatives group included 12 relatives of schizophrenic patients and 13 relatives of schizoaffective patients and came from 15 families. Only relatives who did not meet diagnostic criteria for a lifetime diagnosis of psychotic disorder, bipolar disorder, or an SZ spectrum personality disorder were included in this study. HCs met the same inclusion criteria as relatives and also did not have a first- or second-degree relative with a history of psychosis, psychiatric hospitalization, or suicide. The following exclusion criteria applied to all participants: lack of fluency in English, history of serious head trauma or organic brain disease, history of substance abuse or dependence during the past 12 months or previous chronic dependence, and hearing loss by audiometry. All participants had an estimated verbal IQ of 85 or greater based on the vocabulary subtest of the Wechsler Adult Intelligence Scale—Revised.24 Structured diagnostic interviews were performed by experienced interviewers using the Structured Clinical Interview for DSM-IV.25 Consensus diagnoses were assigned using best estimate methods26 by an independent group of senior clinicians after review of the interview materials and all available hospital records. The interviews and the diagnostic evaluations were performed blind to group membership and to the results of the experimental procedures. This study was approved by the McLean Hospital Institutional Review Board. After complete description of the study to the subjects, written informed consent was obtained.
The subject groups did not differ in socioeconomic status (table 1). Relatives had significantly more years of education (P = .01) and a larger proportion of females (P = .02) than HCs and patients, who did not differ from each other. The HCs were significantly younger than both patients (P = .001) and relatives (P = .01), who did not differ from each other. Among smokers, patients smoked significantly more cigarettes per day than HCs (P < .001) or relatives (P = .05), who did not differ from each other.
The EEG was recorded with Neuroscan Synamp amplifiers (0.01–100 Hz, 500 Hz digitization rate) with sintered Ag/AgCl electrodes in an electrode cap at 60 scalp sites, the nose tip, and the left mastoid, referenced to the right mastoid. The forehead (AFz) served as ground. Bipolar vertical and horizontal electro-oculograms were recorded from electrodes above and below the right eye (VEOG) and at the left and right outer canthi (HEOG). Electrode impedances were below 5 kΩ. Subjects were not allowed to smoke for a minimum of 40 minutes prior to recording. Auditory stimuli were generated using Superlab software and delivered through intra-aural earphones. In all, 160 paired-click stimuli (5-ms duration; 2-ms rise/fall period; 500-ms interclick interval; 10-s intertrial interval) were presented in 4 blocks (40 paired-click stimuli/block). Each block was separated by a 1-minute break. Hearing thresholds were ascertained before the recording and stimulus intensity was adjusted to 50 dB above each individual’s hearing threshold. Data were acquired from 100 ms before to 400 ms after each click. Subjects were asked to fix their gaze on a target and to avoid making eye movements and blinking during the presentation of clicks.
EEG signal processing was performed blind to group membership. P50 ERPs were processed using the same procedures described previously.27 P50 processing was performed off-line using NEUROSCAN (4.3) software. Continuous EEG from each channel was epoched (from −100 to 400 ms relative to stimulus onset), creating 320 sweeps, and a 1-Hz high-pass filter (24 dB/oct) was applied. Epochs were then baseline corrected using the prestimulus interval. An automatic ocular artifact rejection procedure identified and rejected any sweeps ±35 μV in the Cz, HEOG, or VEOG channels between 0 and 75 ms poststimulus (to capture blinks and other slow-wave activity). Sweeps without artifacts were averaged for each of the 4 blocks of 40 trials for S1 and S2 separately. Average waveforms were then digitally filtered using a 10-Hz high-pass filter (24 dB/oct) with zero phase shift, and a 7-point moving average was applied twice.
P50 waveforms were measured at 4 electrode sites: Fz, Cz, FC1, and FC2. At each site, the S1 response was identified as the most prominent peak in the 40- to 80-ms poststimulus window. The preceding negative trough was used to calculate the S1 amplitude.28,29 For the S2 response, the positive peak with latency closest to that of the S1 peak was selected. The S2 amplitude was determined in the same way as the S1 amplitude. P50 suppression was calculated as the ratio of the mean value of the S2 amplitude to the mean value of the S1 amplitude (S2:S1) at Cz. In order to maximize reliability, we also quantified P50 suppression as the ratio across all 4 sites: averaged S2 amplitude across all 4 sites to averaged S1 amplitude across all 4 sites. Lower ratio values signify more inhibition of the S2 response relative to the S1 response. The number of artifact-free trials in P50 suppression or responses to S1 and S2 stimulus did not differ significantly (all P > .08) between the groups.
Raw EEG recordings were processed off-line using Brain Vision Analyzer software (Brain Products, Munich, Germany). EEG signals were first filtered between 10 and 80 Hz, segmented from −100 to 400 ms relative to stimulus onset, and then baseline corrected using the 100-ms prestimulus interval. Epochs containing artifacts ±50 μV at Fz, Cz, or Pz were then removed. Individual trials were exported for time-frequency analysis. The number of artifact-free trials in frequency responses to S1 and S2 stimulus did not differ significantly between the groups (all P > .10).
The time-varying spectral power was computed for gamma by convolving the average of individual trials with a complex-valued Morlet’s wavelet with Morlet’s constant σf/f = 6 and a fixed cycle length of 6 over the 20- to 80-Hz range with 11 frequency bins. Baseline correction for each frequency bin used the 100-ms prestimulus interval. Evoked power was derived from the squared amplitude coefficient of the wavelet transform. Gamma was identified as the evoked power over 35–46 Hz.30 A similar procedure was used for beta, convolving with a complex Morlet wavelet with Morlet’s constant σf/f = 6 and a fixed cycle length of 6 over the 7.5- to 60-Hz range with 16 frequency bins. The same baseline correction procedure was used, and beta was defined as the evoked power over 13–30 Hz. Wavelet analysis was done in Matlab utilizing software provided by C. Torrence and G. Compo (available at http://atoc.colorado.edu/research/wavelets).
The patterns of gamma and beta ERO responses to S1 and S2 stimuli and their gating measures were assessed by grouping the electrodes into 13 regions of interests (ROI): FZ, CZ, PZ, left frontal (F1, F3, FC1, FC5), left temporal (T3, T5, T9, TP7), left central parietal (C1, C3, C5, CP1, CP3, CP5), left parietal (P1, P3, P5, P9), left occipital (PO3, PO7, PO9, O1), right frontal (F2, F4, FC2, FC6), right temporal (T4, T6, T10, TP8), right central parietal (C2, C4, C6, CP2, CP4, CP6), right parietal (P2, P4, P6, P10), right occipital (PO4, PO8, PO10, O2). For each ROI, the averaged response across sites was calculated. Prior to statistical analyses, extreme outliers for each ROI were identified as values greater than 3 SD of the group mean. Extreme outliers were excluded from statistical analyses. There were no more than 5 outliers in any given ROI. We calculated the S1-S2 amplitude difference as an expression of the gating measure for the gamma and beta EROs because some individual’s responses to S1 or S2 stimuli in time frequency analyses were negative values, making S2:S1 ratio measure difficult to interpret and impractical in statistical analysis.
The dependent measures (S1 response, S2 response, S2:S1 suppression ratio, and S2-S1 difference) were compared between groups using regression analyses. Linear regression analyses using SEs that are robust against nonindependence of observations from individuals within families (clusters) and against departures from normality assumptions were carried out with the regress command and the “robust” and “cluster” options in STATA (version 10; Stata Corp., College Station, TX). An advantage of this approach is that it maintains correct type I error rates, given cluster-correlated data. Gender and age were included as covariates. In the patient group, clinical variables (medication dose in CPZ equivalents, age of onset, and BPRS score) were also included as covariates. In order to account for the effect of S1 on S2 activity, the S1 response was included as an additional covariate in all regression analyses of S2 responses. For each of the gamma or beta responses, separate regression analyses were performed for each ROI; a Bonferroni correction (P < .004) was applied as the threshold for statistical significance. For P50 waves, years of education correlated marginally with P50 suppression (partial r = −.17; P = .07) and was included as a covariate in the analyses. A 0.05 level of significance was used for the analyses of P50 ERPs indexes (S1 and S2 amplitude, and suppression ratio). Pearson pairwise correlations correcting for multiple comparisons were used to assess associations among the 3 gating measures (P50, gamma, and beta) and between P50 ERPs and frequency responses to the S1 and S2 stimuli at Cz. Partial correlation coefficients based on regression coefficients from analyses that included age and sex as covariates assessed associations between clinical variables (age of onset, BPRS score, and CPZ equivalent) and the ERP/ERO-dependent variables in the patient group. A 0.05 level of significance was used in this analysis.
Scatterplots of S1 and S2 amplitudes, and suppression ratios averaged across all 4 sites for P50 waves are presented in figures 1 and and2.2. A significant main effect of group on P50 suppression was observed (F3,82 = 2.94, P = .04). SZ patients had a significantly higher P50 S2:S1 ratio (mean = 0.76, SD = 0.35) than HCs (mean = 0.56, SD = 0.31; P < .001). P50 ratio showed a trend to be higher in relatives (mean = 0.68, SD = 0.46) than in HCs (P = .09). Patients (mean = 2.22, SD = 1.11; P = .57) and relatives (mean = 2.47, SD = 1.46; P = .58) did not differ from HCs (mean = 2.41, SD = 1.48) in amplitude of the S1 wave. However, patients (mean = 1.58, SD = 0.91) and relatives (mean = 1.55, SD = 1.06) had larger S2 amplitudes than HCs (mean = 1.37, SD = 1.18) but only patients differed significantly from controls (P < .01). In the patient group, illness-related factors, smoking status, and number of daily cigarettes were not significantly associated with P50 ratio, S1 amplitude, or S2 amplitude (all P values >.05).
Table 2 shows group means and SDs of gamma- and beta-evoked power to S1 and S2 stimuli at each ROI. Widely distributed reduced gamma power to S1 stimuli (in 10 out of 13 ROIs) was found in patients compared with HCs (P < .001 in all, figure 3). Gamma power to S2 stimuli was significantly reduced at Pz (P = .001), left temporal (P = .002), and left parietal (P < .001) ROIs (figure 4). HCs and relatives did not differ significantly in gamma power at any ROI (figures 3 and and44).
Means and SDs of gamma gating (S2-S1 difference) at each ROI for each group are shown in table 3. Compared with HCs, patients showed gamma-band gating deficits at Fz, left temporal, Pz, left parietal, left occipital, and right occipital ROIs. HCs and relatives did not differ significantly in gamma gating at any ROI.
Partial correlation coefficients revealed that age of onset was significantly associated with gamma EROs. For the S1 response, age of onset was significantly associated with gamma power at the Fz (r = .48, P = .001), Pz (r = .32, P = .04), bilateral frontal (LF: r = .48, P = .001; and RF: r = .45, P = .003), and bilateral central-parietal (LCP: r = .51, P = .001; and RCP: r = .51, P = .001) sites. For the S2 response, age of onset was significantly associated with gamma power at the Fz (r = .43, P = .005), Cz (r = .49, P = .001), and bilateral frontal (LF: r = .47, P = .002; and RF: r = .37, P = .02) sites; the younger the age of onset, the smaller the gamma ERO response. Medication dose in CPZ equivalents was modestly but significantly negatively associated with gamma gating at the left (r = −.31, P = .05) and right occipital (r = −.33, P = .04) ROIs; the higher the dose, the less the gating deficit.
Compared with HCs, patients showed significantly reduced beta power activity to S1 stimuli at the Pz, right temporal, and left parietal sites (P < .001 in all, figure 3). Widely distributed reduced beta power to S2 stimuli (in 8 out of 13 ROIs) was found in the patient group (P < .001 in all, figure 4). In the relatives, beta power to S1 and S2 stimuli was not reduced at any ROI (figures 3 and and44).
Means and SDs of beta gating for each group at each ROI are shown in table 3. Compared with HCs, patients showed deficits in 8 of 13 ROIs, with differences most pronounced at temporal and posterior areas, including: Pz, left temporal, bilateral central parietal, bilateral parietal, and bilateral occipital (table 3). HCs and relatives did not differ significantly at any ROI.
Age of onset was significantly associated with beta EROs; the younger the age of onset, the smaller the beta ERO response. For the S1 response, age of onset was significantly associated with beta ERO at the right frontal (r = .35, P = .03), Pz (r = .58, P < .001), bilateral central-parietal (LCP: r = .37, P = .02; and RCP: r = .37, P = .02), bilateral parietal (LP: r = .56, P < .001; and RP: r = .58, P < .001), and bilateral occipital (LO: r = .42, P = .005; and RO: r = .40, P = .008) sites. For the S2 response, age of onset was significantly associated with beta ERO at Cz (r = .40, P = .009). Beta gating was significantly associated with age of onset in 10 of 13 ROIs (r = .32–.57, all P values <.05) except left temporal, left occipital, and left frontal ROIs, indicating that the younger the age of onset, the greater gating deficit. Medication effects were not significantly associated with beta EROs.
Table 4 shows the correlations of P50 ERPs with gamma and beta responses for each group at Cz. P50 gating was not significantly correlated with either gamma or beta gating at Cz in all 3 groups. No significant association between the P50 ratio and either gamma or beta responses to S1 or S2 stimuli was found either. Gamma and beta gating were not significantly correlated with each other in HCs or relatives but were positively correlated in patients (r = .47, P = .04). In HCs, P50 S1 amplitude was not significantly associated with beta or gamma responses but was significant with beta response to S1 (r = .81, P < .001) in relatives and S2 (r = .46, P = .04) in patients. P50 S2 amplitude was significantly associated with beta power in response to both S1 and S2 stimuli (S1 r = .60; S2 r = .61), and with gamma power in response to S1 stimuli (r = .71) in HCs. Patients did not show such associations (r = .09–.29).
In this study, we examined (1) evoked power and the topographic distributions of gamma and beta EROs to S1 and S2 stimuli in SZ patients, their relatives, and HCs and assessed whether responses to either type of stimulus met the familiarity criterion for being endophenotypes; (2) amplitude difference and the topographic distributions of gamma and beta gating measures in SZ patients, their relatives, and HCs and assessed whether gamma or beta gating measures met the familiality criterion for being endophenotypes; and (3) the relation between P50, gamma, and beta ERO gating measures.
This study is the first to examine the topographic distributions of gamma and beta EROs to S1 and S2 stimuli and their gating measures in SZ patients and their clinically unaffected relatives, although other studies have used a paired-click paradigm to examine oscillatory activity in response to the S1 and S2 stimuli in SZ. Johannesen et al21 found that SZ patients showed smaller gamma power to the S1 stimulus compared with HCs at frontal-central regions. Brenner et al19 reported that SZ patients exhibited less evoked beta power in response to salient S1 stimuli at Cz.
Compared with control subjects, SZ patients showed widely distributed reductions in gamma power to S1 stimuli and in beta power to S2 stimuli (figures 3 and and4).4). Significantly reduced gamma power to S2 and beta power to S1 stimuli were also observed in patients in some areas of the brain. These findings are consistent with the hypothesis that the core pathophysiology of SZ is in part related to disturbed neural synchrony at high oscillations across widely separated regions of the brain.16 Neural oscillatory activity in the beta/gamma frequency ranges is thought to reflect different aspects of early sensory information processing,17,31,32 with beta activity implicated in novelty detection and salience encoding and gamma oscillation associated with immediate registration of sensory stimuli.17,33 Deficits in both frequency ranges would result in weak neuronal signals and propagation across broadly distributed networks, potentially leading to impaired perceptual experiences and cognitive dysfunction.17,31–33
We also found significant correlations between gamma and beta power to both S1 and S2 stimuli, between P50 amplitudes and beta power to both S1 and S2 stimuli, and between P50 S2 amplitude and gamma S1 power (table 4). These results replicate those of Kisley and Cornwell,15 who reported significant positive correlations between P50 amplitudes and beta/gamma power to S1 and S2 stimuli, and those of Hong et al14, who documented significant positive correlations between gamma and beta responses. The positive relation between beta and P50 responses to both S1 and S2 stimuli (correlations between .34 and .45) suggests a contribution of beta frequency to P50 ERP activity in general and the positive relation between S1 gamma and S2 P50 amplitude (r = .45) suggests a contribution of gamma S1 activity to P50 S2 ERP. These results are consistent with the notion that both gamma and beta EROs contribute to the generation of P50 ERPs. The study by Hong and colleagues14 reported that, in patients with SZ, gamma positively contributed to the P50 S2 response, whereas beta negatively contributed to the S2. The results of our study are consistent with those of Hong et al in showing a significant positive contribution of gamma S1 activity to the P50 S2 response (βgamma = .09, t = 2.6, P = .01) and negative relationship of beta S1 to the P50 S2 (βbeta = −.01, t = −0.28, P > .05) in SZ. However, contribution of beta was not significant in our study.
Clinical variables, particularly age of onset, had significant effects on gamma and beta ERO responses to both S1 and S2 stimuli and on beta gating. Medication dose, however, was associated with gamma gating at bilateral occipital regions but was not significantly associated with gamma/beta ERO S1/S2 activity. Younger age of onset was associated with greater reductions in gamma/beta activity and more impaired gating of beta frequency–specific EROs. Age of onset, defined as the age of first hospitalization, was significantly correlated with duration of illness (r =−.43, P < .01). Using duration of illness in the analyses produced results similar to those obtained for age of onset (but weaker P values). These findings suggest that the magnitude of the reduction in gamma/beta oscillatory activity and impaired gating varies as a function of illness-related factors. However, whether the specific clinical association reflects illness progression/chronicity or increased severity as a function of earlier onset could not be determined in this cross-sectional study. Future studies using a longitudinal design would help to clarify the relation between specific illness-related variables and severity of symptoms and gamma/beta ERO activity.
The topographic distributions of gamma and beta responses to S1 and S2 stimuli in the relatives were similar to those of HCs. In fact, mean gamma responses to S1 and S2 stimuli at Fz and Cz in the relatives group were greater (although nonsignificantly) than those in the HCs (figures 3 and and4).4). Taken together, these results suggest that deficits in gamma and beta EROs to S1 and S2 stimuli are associated with SZ, but are not related to genetic liability for the illness.
Our results show that SZ patients had significantly poorer gamma gating compared with control subjects in 6 of 13 ROIs (table 3). For beta gating, poorer gating was widely distributed across the brain in 8 of 13 ROIs and was most pronounced at temporal and central-posterior areas (table 3). Although it is not entirely clear as why these brain areas are particularly vulnerable to beta gating deficits, there was a trend toward increasing correlations of beta gating deficits with age of onset at central-posterior regions compared with those of frontal regions. Correlations with age of onset at frontal regions were between .27 and .39, whereas at central-parietal areas they were between .38 and .50 and at parietal areas they were between .48 and .57. Other clinical variables were not associated with beta gating deficit. It is possible that younger age of onset results in more progressive brain degeneration, leading to more pronounced beta gating deficits. This hypothesis is consistent with the finding of greater-than-normal age-related cortical gray matter volume reductions in the posterior superior temporal region bilaterally in chronic SZ.34,35 Whether reduced beta ERO responses and gating deficits show associations with progressive reduction of cortical gray matter will need to be examined in a future study.
Medications are an unlikely explanation for the observed gamma/beta gating deficits. The only significant relationship between dose of antipsychotic medication and the magnitude of gamma gating deficit was the inverse correlations at bilateral occipital regions. Thus, if anything, medication may have normalized gamma gating, rather than worsened it, suggesting that the significant difference between patients and controls is a conservative indication of the magnitude of group differences.
In the relatives, no evidence of altered frequency-specific gating responses was observed across the entire scalp compared with HCs. These results replicate the report of Hong et al22 in showing that neither gamma nor beta gating measures are abnormal in relatives. These results are consistent with the interpretation that abnormalities in gamma and beta oscillatory responses and gating do not meet the familiality criterion for an SZ endophenotype.
The beta and gamma gating measures were not significantly correlated with P50 gating as measured by P50 S2:S1 suppression ratio (table 4). An alternative gating measure is the P50 S1-S2 difference.19,21 The P50 S2:S1 suppression ratio was significantly negatively correlated with the P50 S1-S2 difference score (r = −.76, P < .001). Analyses based on the difference score produced results similar to those reported in the text (data available upon request). In contrast, P50 S1-S2 gating was modestly and significantly correlated with beta gating at CZ (r = .31, P = .05) but not at FZ (r = .27, P = .22), and not with gamma gating at either CZ or FZ. The association of P50 and beta S1-S2 gating may be related to overall shared variance between P50 and beta responses to S1 and S2 stimuli. However, the nonsignificant association at FZ suggests that the 2 gating measures are only weakly correlated.
The correlation patterns between the S1 and S2 P50 amplitudes and gamma and beta power were different in HCs and patients (table 4). P50 S1 amplitude was not significantly associated with beta or gamma responses in HC, but it was significantly associated with the beta response to S2 (r = .46, P = .04) in patients. In HCs, P50 S2 amplitude was significantly associated with beta power in response to both S1 and S2 stimuli (S1 r = .60; S2 r =.61; both P < .01) and with gamma power in response to S1 stimuli (r = .71, P < .01). Patients did not show such significant associations (r = .09–.29). It is possible that there are fundamental differences in the generation of evoked responses in patients with SZ, but it is also possible that these differences are the result of illness-related factors, including treatment with psychotropic medications.
The trend level P50 gating deficit found in the relatives contrasts with their normal gamma or beta gating responses. Although the distribution of P50 ratio values shown in figure 1 suggests a possible bimodality in the relatives group, the sample size was too small to have adequate power to detect mixture or to estimate its parameters with acceptable precision. Inspection of the histogram of the data (square root transformation) indicated no bimodality (skewness = −.33), and skewness kurtosis test for normality was nonsignificant (P = .1). These findings indicate that bimodality is unlikely. A subgroup of relatives, however, relatives of SA-bipolar type, continue to have significantly higher mean P50 ratios than HCs, whether transformed (P = .03) or nontransformed (P = .02) data were used. There was no evidence of gamma or beta gating deficits in the subgroup of relatives compared with HCs (P > .05 in all). This combination of findings suggests that P50 sensory gating is functionally distinct from frequency-specific gating responses. That is, the components of information processing assessed by these frequency-specific gating measures appear to be, for the most part, independent of those mediated by P50 suppression. This conclusion is also consistent with the results of Clementz and Blumenfeld20 in showing a significant P50 gating deficit in SZ patients but normal gamma gating in these patients and the independence of a significant theta/alfa-band gating deficit and P50 gating in SZ patients and in their relatives observed by Hong et al22. We constrained our analyses to the beta/gamma frequency range and did not measure theta/alfa EROs because studies of P50 ERPs typically apply a 10-Hz high-pass filter to exclude low-frequency bands2,20 and because the major contributors to auditory P50 ERP responses appear to be in the gamma and beta frequency range.12,15,36
Theta- and alfa-bands may play a role in attentional modulation of the P50 ERP and have been shown to be abnormal in SZ on P50 tasks.20,21 The timing of the theta/alfa activities tend to occur after the gamma/beta activities, suggesting additional information processing. The theta/alfa-band gating deficit observed by Hong and colleagues22 may be, to a large extend, due to top-down modulation via feedback neuronal projection loop beyond local cortical oscillations.37
Although the control group had an equivalent representation of males and females, the patient group was predominantly male and the relatives group was predominantly female. Control subjects were younger than the relative and patient groups. To address these differences in group composition, we included both sex and age as covariates in the analyses. Patients with SZ and SA did not differ in demographic characteristics or in brain activity measures. The sample size of the relatives group was too small for statistical comparison of the subgroups of relatives or test for admixture of the physiological deficit in relatives. The question of whether a subgroup of relatives has reduced gamma/beta activity and gating deficits will be addressed in a larger family study in the future.
In summary, the present study indicates that deficits in frequency-specific oscillatory responses are observed in SZ patients but not in their clinically unaffected relatives. Moreover, information processing assessed by P50 suppression appears to be independent of gamma or beta gating. Neither the absolute power nor gating of gamma and beta EROs appears to satisfy criteria for being endophenotypes.
Supplementary material is available at http://schizophreniabulletin.oxfordjournals.org.
Kaplen Fellowship, Harvard Medical School [to M.-H.H.]; National Alliance for Research on Schizophrenia and Depression Sidney R. Baer, Jr. Foundation [to M.-H.H. and D.L.L.]; Adam Corneel Young Investigator Award [to M.-H.H.]; National Institutes of Mental Health [T32 MH016259-29A1 fellowship to M.-H.H., MH RO1-58704 to D.F.S., and R01 MH071523 and MH31340 to D.L.L.]; Rappaport Mental Health Research Scholar Award, McLean Hospital [to M.-H.H.]; Essel Foundation [to D.L.L.].
We thank all participants. We thank Nancy Mendell for statistical consultation; both Toni Mahowald and Kate Tyler for help with EEG recording; Verena Krause, Olga Krastoshevsky, and Anne Gibbs for help with data management and subject recruitment; Michael Coleman for clinical interviews; and Lenore Boling, Alexander Bodkin, Jan Lerbinger, Frederick Johnson, and Jonathan Cole for DSM diagnoses.
Conflict of interest: None