Contributions of subtype and spectral frequency analyses to the study of P50 ERP amplitude and suppression in schizophrenia

Abstract

Background

Poor suppression of P50 event-related potential (ERP) amplitudes to paired-click stimuli may indicate genetic liability for schizophrenia and weak “sensory gating.” Evidence suggests, however, that P50 amplitude is selectively impaired in nonparanoid, but not paranoid, schizophrenia subtypes. Furthermore, paired-click suppression can appear deficient in schizophrenia due to smaller evoked responses to the first stimulus (S1), rather than larger, less effectively “gated” responses to the second (S2). Finally, the P50 ERP is comprised of activity from at least two frequency components that may be distinctly impaired: the gamma band, associated with sensory registration, and the low frequency response, associated with attention/encoding processes. P50 and related frequency subcomponents were examined as a function of illness subtype to further integrate these concepts.

Method

The standard paired-click paradigm was administered to 38 schizophrenia (27 paranoid, 11 nonparanoid) and 38 age-matched healthy control participants. P50 amplitudes and spectral power of gamma band (GBR; 20–50 Hz) and low frequency (LFR; 1–20 Hz) responses were analyzed.

Results

P50 analyses revealed smaller S1 amplitude and normal S2 in schizophrenia participants collectively, but no differentiation of schizophrenia subtypes. Spectral analyses revealed smaller magnitude S1 and normal S2 responses in schizophrenia across both the GBR and LFR. The LFR, but not GBR, was found to distinguish nonparanoid from control groups, while paranoid participants evidenced no impairment in either frequency domain. LFR amplitude values correlated with clinical ratings of cognitive symptomatology.

Conclusions

ERP deficits in the dual-click paradigm were specific to S1 amplitudes and most prominent in the low frequency response. These results replicate previous findings and extend their relevance to schizophrenia subtype distinctions. Implications for the recurrent inhibition model of sensory gating are discussed.

Introduction

Inhibitory properties of the central nervous system have long been examined using conditioning-testing paradigms, under which the amount of attenuation in the neural response to the second of two identical stimuli indexes the strength of the inhibitory pathway (e.g., Eccles, 1969). This paradigm has been adapted in psychophysiological research as a test of “sensory gating” and has provided basic support for theories postulating that individuals with schizophrenia demonstrate less effective regulation over the influx of sensory information to the brain (Adler et al., 1982, Braff et al., 1995). Specifically, decrement in the amplitude of the P50 event-related potential (ERP) evoked by the second (S2) relative to the first (S1) of two auditory “clicks,” commonly expressed as the suppression ratio S2 / S1, is smaller in schizophrenia and thought to reflect weak inhibition or gating of the repeated stimulus (Freedman et al., 1987).

The P50 suppression deficit may indicate genetic liability for schizophrenia (Clementz et al., 1998, Freedman et al., 2003) and is, according to one meta-analytic study, one of the strongest and most reliable findings in the schizophrenia literature (Heinrichs, 2004). However, while highly important, these findings are not without limitation. For instance, clear relationships between this psychophysiological abnormality and symptoms of schizophrenia have eluded investigations to date. Although a relationship with negative symptoms was supported in one study (Ringel et al., 2004), others suggest that this deficit does not relate to the severity of negative symptoms (Adler et al., 1990) or those in any specific symptom domain (Boutros et al., 2004). Divergent findings of this nature are not uncommon and likely arise, in part, as a consequence of the clinically heterogeneous presentation of this illness. Relatively modest effect size differences observed throughout the schizophrenia neuroscience literature indicate considerable overlap between patients and healthy comparison groups (Heinrichs, 2004), and suggest that as with other indices of brain function, P50 suppression deficits apply to only a subset of patients studied. Consistent with this interpretation, only 50% of patients interviewed in one study reported experiencing anomalies of perception and attention related to sensory gating (Bunney et al., 1999).

As originally conceptualized, schizophrenia represents a diverse taxon with three primary illness variants, paranoid, hebephrenic, and catatonic (Kraepelin, 1919); these appear in similar form today as conventional DSM-IV subtypes (American Psychiatric Association, 1994). While other subtype classifications have been proposed (e.g., Andreasen et al., 1995, Carpenter et al., 1988, Crow, 1985), a paranoid–nonparanoid dichotomy is perhaps the most consistently supported, wherein the latter class represents a relatively more severe course and symptom presentation (Nicholson and Neufeld, 1993). When schizophrenia subgroups have been investigated according to this dichotomy, P50 amplitudes appear normal in paranoid schizophrenia (Boutros et al., 1993, Boutros et al., 1991). Alternatively, nonparanoid individuals produced smaller amplitude responses to the first click stimulus (S1) and, consequently, relatively less attenuation of P50 amplitude at S2. Similar results are found in comparison of “systematic” (i.e., disorganized) patients to healthy normal participants and patients with “unsystematic” schizophrenia or cycloid psychosis (Ringel et al., 2004). Considered in light of relatively worse neuropsychological performance in patients with severe negative and disorganized symptoms than those characterized as paranoid (Hill et al., 2001), questions arise as to whether cognitive disturbances unrelated to sensory processing per se mediate P50 amplitude in schizophrenia. Evidence for relationships between P50 and attentional impairments in schizophrenia (Cullum et al., 1993, Erwin et al., 1998, Yee et al., 1998) and the attentional manipulation of P50 in healthy adults (Guterman et al., 1992) support this possibility. Taken together, there is good reason to believe that clinical features most related to P50 are those pertaining to cognitive or “disorganization” dimension.

Most studies of auditory suppression in schizophrenia have relied on information obtained by amplitude measurements of specific peaks in the time domain (e.g., P50, N100). There is evidence, however, that at least two broadly defined frequency bands, each associated with a distinct role in information processing, contribute to these auditory-evoked response amplitudes (Clementz and Blumenfeld, 2001). The gamma-band response (GBR; 20–50 Hz), characterized by high frequency oscillations emerging approximately 20 ms post stimulus (Pantev et al., 1991), is associated with the initial neural registration of a sensory stimulus (Karakas and Basar, 1998). Although the P50 ERP is thought to be a subcomponent of gamma (Basar et al., 1987, Clementz et al., 1997), auditory suppression deficits in schizophrenia might, to an even greater extent, reflect poor activation of a low frequency response (LFR) occupying the 1–20 Hz range (Clementz and Blumenfeld, 2001). The LFR oscillates at theta frequency, generally associated with new information encoding (Klimesch, 1999) and working memory functions (Grunwald et al., 1999, Jensen and Tesche, 2002). Additionally, the initiation and maintenance of stimulus-locked hippocampal theta might facilitate the processing of novel occurrences in the sensory scene (Tesche and Karhu, 2000, Grunwald et al., 1999). Taken together, the integration of these frequency components seems necessary for normal cognitive function: gamma synchrony establishes the neural communication necessary to “bind” sensory components into a unified percept (Lee et al., 2003), after which a gamma to low frequency shift occurs, theoretically denoting a modification of neural circuitry accompanying activation by new or particularly salient stimuli (Traub et al., 1999). In examination of these frequency domain responses under the paired-click paradigm, individuals with schizophrenia have previously been found to exhibit deficient LFR in the presence of normal GBR, in which case LFR magnitude to S1 but not S2 distinguished patients from healthy participants (Clementz and Blumenfeld, 2001). Compared to conventional time-domain analysis of P50 and N100, which also detected smaller S1 amplitude responses in schizophrenia, group separations appeared largest for the LFR.

In view of the centrality of sensory gating research to the study of schizophrenia, three critical questions concerning the psychophysiological measurement and conceptualization of this construct are addressed in the present study. First, do cortical ERP indices of sensory gating vary as a function of clinical subtypes of schizophrenia? Second, in which frequency band of the auditory ERP are information processing abnormalities most prominent? That is, which electroencephalogram (EEG) frequencies of the neural response, and related neural functions, are most affected in the disorder and its subtypes? Third, from what source are neurophysiological abnormalities in the dual-click paradigm derived in schizophrenia, in terms of S1 and S2 responses? Specifically, is poor P50 suppression the result of weak recurrent inhibition of the S2 response, as suggested by the prevailing “gating” model, or is this phenomenon better explained by compromised recruitment and engagement of substrates normally activated at S1? Answers to these questions are critical to clarifying the nature of information processing deficits associated with schizophrenia, especially with respect to extant sensory gating, or inhibitory, models of the disorder.

Consideration and integration of the relevant P50 and frequency domain research led to the following predictions: (1) “nonparanoid” participants would demonstrate impaired P50 suppression relative to healthy controls, while “paranoid” participants would not differ from healthy controls; (2) concerning the frequency domain analyses, between-group comparisons would yield larger effect size differences for the low frequency than the gamma band response, and the low-frequency response would best discriminate nonparanoid from healthy control participants, and; (3) where differences in suppression occur, these differences would be attributed to lower S1 amplitude in the impaired group, rather than higher S2 response.