|Home | About | Journals | Submit | Contact Us | Français|
A large percentage of patients with schizophrenia are characterized by an abnormality in P50 sensory gating. This abnormality has been shown to be genetically linked to the α-7 nicotinic receptor and is transiently reversed by acute nicotine administration. These observations have led to the development of pharmacological treatments designed to improve sensory gating. However, if normalization of P50 gating abnormalities is to guide drug development, then it becomes important to delineate the clinical correlates of enhanced P50 gating. We conducted a review of all available articles through March 2005 that have examined this issue. We found that, despite the prominent role that P50 abnormalities have played in our understanding of schizophrenia, there is a relative dearth of data examining P50 clinical correlates. There is evidence suggestive of an association between P50 and measures of attention, and multiple studies have failed to document a cross-sectional or longitudinal relationship between P50 and positive, negative, or other symptoms. These results suggest that considerably more work needs to be done to understand and validate the clinical significance of this impairment.
Sensory gating refers to the pre-attentional habituation of responses to repeated exposure to the same sensory stimulus. The inhibition of responsiveness to repetitive stimulation provides humans with the ability to negotiate a sensory-laden environment by blocking out irrelevant, meaningless, or redundant stimuli. P50 is an electroencephalogram (EEG) event-related potential waveform used to assess sensory gating.
There is a large body of evidence to suggest that a significant proportion of patients with schizophrenia have sensory gating impairments.1 These studies have formed the foundation for several hypothetical constructs of the pathophysiology of the symptoms of schizophrenia.2,3 Some theorists suggest that the positive and perhaps the negative symptoms of schizophrenia are a result of sensory overload and/or an impairment in the response to sensory input within the central nervous system, a theoretical concept that originated with Venables.4 Sensory gating abnormalities have also been hypothesized to be associated with various cognitive impairments.5–9
Efforts to delineate the neuroanatomy and neurochemistry of the P50 measure of sensory gating have yielded inconsistent findings, across both species and measurement techniques. Integrating work from human and animal studies, Freedman and colleagues8 suggested that the hippocampus was the primary region implicated in sensory gating, and that multiple neurotransmitters were involved, including the cholinergic, dopaminergic, GABAergic, glutamatergic, noradrenergic, and serotonergic systems. Other recent attempts to localize P50 have relied on magnetoencephalography (MEG) and intracranial recording. MEG-based source analyses have localized P50 generators in several different regions, including Heschl's gyrus10 and the superior temporal gyrus.11 Another set of putative sources for scalp-measured P50 has emerged from intracranial recordings, including prefrontal cortex,12 rhinal cortex,13 and hippocampus,12,14 Gating in the hippocampus, however, was temporally distinct from P50, occurring at a longer latency,12 and may not be reflected in P50. Given the disparate findings across localization studies in humans, it is not possible to delineate clearly the neural circuitry responsible for P50 gating.
Pharmacological studies, however, have demonstrated the importance of the cholinergic system in regulating the diminished response to repeated stimuli, through stimulation of the α-7 nicotinic receptor.8,9,15–19 In patients with schizophrenia, nicotine can transiently reverse the P50 sensory gating deficit.8,20,21 This finding suggests that pharmacological treatments could be developed that modulate P50 with the potential of symptomatic improvement for patients with schizophrenia.
More recently, a marker of the gene for the α-7 nicotinic receptor has been linked to schizophrenia and P50 abnormalities.16 There is also evidence to suggest that polymorphisms in the promotor region of the gene for the α-7 nicotinic receptor may be related to abnormal sensory gating.22 Further supporting a genetic basis of a sensory gating abnormality in schizophrenia has been evidence that family members of people with schizophrenia also have deficits in P50 sensory gating inhibition at higher rates than baseline populations.20,23–26 Additional evidence for a genetic underpinning is provided by patients with schizophrenia spectrum disorders, such as schizotypal personality disorder, who also demonstrate this deficit in gating.27
The hypothesized association of P50 abnormalities with various domains of schizophrenia pathology, the ability to pharmacologically modify the P50 abnormality, and the genetic underpinnings of the P50 abnormality have stimulated the development of novel drugs designed to reverse P50 abnormalities in schizophrenia. If P50 is to be used as a target to guide drug development, then questions arise as to what other illness components are related to P50 abnormalities and what are the clinical implications of enhanced P50 performance. The purpose of this review is to ascertain the clinical significance of P50-related gating abnormalities in patients with schizophrenia. This review will look at studies that have examined the relationship of P50 to neuropsychological measures, symptoms, treatment, and other electrophysiological markers.
PubMed and Medline databases between 1970 and March 2005 were searched using the following terms: P50 or sensory gating and schizophrenia, pre-pulse inhibition, PPI, genetics, animal models, nicotine, neuropsychological, negative symptoms, positive symptoms, BPRS, SANS, SAPS, typical antipsychotics, conventional antipsychotics, atypical antipsychotics, second generation antipsychotics, clozapine, risperidone, olanzapine, or eye-tracking. All authors of relevant articles were subsequently used as search terms. Finally, the references of each article were reviewed. Studies were included that reported data on the association between P50 and clinical correlates. Only studies that had subjects with schizophrenia were included.
P50 is an EEG-based averaged event-related potential (ERP) that can be elicited in the context of either the “paired click” paradigm or the “steady state” paradigm. In the paired click paradigm, two auditory clicks are presented within 500 msec of each other. The first click is commonly referred to as the “conditioning stimulus (C)” or S1, and the second click is called the “testing stimulus (T)” or S2, because it is designed to test the degree to which an individual can gate S2 responses following S1 exposure. EEG responses, usually measured at vertex, are averaged for S1 and S2 separately following baseline correction and artifact rejection. The magnitude of gating is evaluated with composite scores that relate S1 and S2 amplitudes in either ratios (T/C ratio, suppression ratio) or differences (S1 – S2). The most commonly used sensory gating index is the T/C ratio, in which the relative average amplitude of the P50 wave generated in response to the T (S2) stimuli is compared to the average amplitude generated in response to the C (S1) stimuli, the T/C (S2/S1) ratio. A T/C ratio below 50% is the usual definition of normal sensory gating. For longitudinal studies, a decrement in the T/C ratio due to decreased T (S2) amplitude is thought to reflect improvement in sensory gating processes. The physiological significance of T/C ratio decrements due to increases in S1 amplitude is not known. Validation studies are required to determine whether changes in T/C ratios that are achieved by changes in S1 or S2 amplitude measure the same underlying construct.
In the steady state paradigm, auditory clicks are presented at a continuous rate (e.g., 10 clicks/sec9). The average P50 amplitude across all trials serves as an indicator of sensory gating, with larger amplitudes suggesting impaired sensory gating.9,28
Data reduction for both paradigms, however, is not completely automated. For example, subjective judgments are required to finalize the selection of trials to be included in the S1 and S2 averages. If subjective inclusion criteria are used by raters who are not blinded with respect to experimental design features, the opportunity exists for biased results.
There are four major measures derived from the paired click paradigm: S1 amplitude, S2 amplitude, S1 latency, and T/C ratio. The ability to interpret the significance of associations between P50 and clinical measures is dependent, in part, on the reliability of these measures. In general, studies suggest that the P50 waveform can be reliably identified28,29 and that S1 and S2 amplitudes and S1 latency can be reliably measured.30–35 Specifically, Boutros et al. demonstrated a 96% inter-rater agreement in the identification of P50 waveforms.28 Johannesen and colleagues reported intraclass correlations of 0.975–0.995 for S1 and S2 amplitudes.29 Other studies have found test-retest correlations over varying time periods (up to 10 days) ranging from 0.41 to 0.91 for S1 amplitude, with most of the test-retest correlations greater than 0.60.30–35 The test-retest correlations for S2 amplitude are of similar magnitude,31–35 except for one study that found a negative correlation in patients with schizophrenia (r = −0.24; 30). In contrast, the test-retest reliability of the T/C ratio has tended to be more problematic and has ranged from r = −0.22 to 0.47,30–35 Smith et al. argue that the poor reliability of the T/C ratio is due to greater variability in S2 than S1 and poor correlation in variance between S1 and S2.36
In terms of studying the clinical correlates of P50 gating, the reduced reliability of the T/C measure may lead to the failure to detect relationships when they actually exist, especially if this is the only measure used to examine the relationship between P50 gating and the clinical correlate. The differential reliability of the P50 measures suggests that studies should examine the association of S1 and S2 amplitude and S1 latency to the clinical correlate of interest. In our review of the data, we will present information, when available, on these four P50 measures.
Of the available studies that examined the clinical correlates of P50, three broad categories emerged: neuropsychological measures, symptoms, and treatment. The articles will be discussed according to their relevance to each of these three categories and relevant subcategories.
We organized our discussion of the relationships between P50 abnormalities and neuropsychological test performance according to the categories of cognitive impairment proposed by the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) project (37). The MATRICS project is formulating research standards for neuropsychological impairments relevant to schizophrenia. The MATRICS cognitive categories are attention/information processing, reasoning and problem solving, social cognition, processing speed, verbal learning and memory, visual learning and memory, and working memory. Neuropsychological measures that could not be placed into one of these categories were placed into the “other” category. No studies were found that addressed the relationship between P50 and social cognition.
P50 investigations that have compared neuropsychological measures of attention, the impact of varying attentional load, and distraction tasks have provided the strongest evidence of a relationship between P50 and cognition. Cullum et al.5 used the paired click paradigm to examine the relationship between P50 gating and digit vigilance, a neuropsychological measure of sustained attention, in medicated patients (n = 14) treated with conventional antipsychotics and in normal controls (n = 15). Patients were significantly more impaired on both measures. In the patients, but not in the normal controls, larger P50 T/C ratios were associated with worse digit vigilance scores (r = 0.67, p < 0.004). In a group of unmedicated patients, Erwin et al.9 divided subjects into high P50 abnormality (n = 15) and low P50 abnormality (n = 16) groups based on their performance on the steady state P50 paradigm, and then compared the two groups on the Gordon CPT, a measure of sustained attention. The high abnormality group performed significantly worse on the Gordon CPT distraction measure (number correct), and there was a trend for the high abnormality group to do worse on the vigilance measure (number correct).
In contrast, Guterman and Josiassen6 found no evidence of an effect of increased attentional load on sensory gating in patients with schizophrenia. They compared medicated subjects on conventional antipsychotics (n = 10) and normal controls (n = 10) across two different P50 conditions. The first condition was the usual paired click condition. However, in the second condition, subjects had to attend to high pitch or low pitch clicks, by counting either the high pitch or low pitch clicks, a modification that places increased load on the attentional system. Normal controls, but not patients, performed more poorly during the increased attentional load task than they did during the simple paired click task. The lack of an effect in the patient group may have been due to the marked impairment during the low attentional load task, which may have precluded the ability to detect any impact of increased attentional load in the patients.
In summary, the two studies that directly examined P50 gating abnormalities and measures of sustained attention or vigilance found a relationship between impaired sensory gating and impaired attention in patients with schizophrenia. Alteration of attentional load did not appear to effect sensory gating, but study limitations preclude any definitive statement about the impact of such a manipulation. None of the studies were conducted in patients who were treated with second-generation antipsychotics.
Cullum et al.5 failed to find any relationship between P50 ratio and measures of processing speed. In contrast, Erwin et al.9 found that the high sensory gating abnormality group performed significantly worse on the trails B, with a trend for the high abnormality group to do worse on the digit symbol measures of processing speed. The high abnormality group also differed on the “color” but not on the “word” portion of the Stroop.
Two studies have examined the relationship between P50 gating and Wisconsin Card Sorting Test (WCST). Louchart-de la Capelle et al.38 examined the relationship between P50 T/C ratio and WCST performance in patients with schizophrenia (n = 124) and normal controls (n = 100) and failed to find a significant relationship between the two measures. Erwin et al.9 also failed to find a significant relationship between P50 gating and WCST performance, nor any relationship between P50 gating and WAIS-R similarities.
Three studies have examined the relationship between sensory gating and verbal or visual memory. Cullum et al.5 and Erwin et al.9 failed to find any relationship between P50 sensory gating indices and the California Verbal Learning Test. Hsieh et al.39 examined the relationship of P50 gating and both explicit and implicit memory. They failed to find a relationship between impaired P50 gating and measures of verbal or visual explicit memory in either patients with schizophrenia (n = 10) or normal controls (n = 10). However, they did find that patients with schizophrenia with impaired sensory gating performed significantly worse on a visual, but not verbal, implicit memory task. This relationship was not observed in the normal controls.
Cullum et al.5 found a significant correlation between P50 ratio and the digit span (forward direction only) measure of working memory. No other studies have examined this relationship.
In thirteen patients with schizophrenia, Vinogradov et al.7 reported a relationship between worse P50 ratios and impaired intracategory semantic priming. However, the semantic priming measures were obtained while patients were off medications, whereas the P50 measures were collected while patients were taking medications, which may confound the interpretation of the results.
The most striking observation about the relationship between P50 sensory gating abnormalities and cognitive abnormalities is the relative dearth of studies that have examined this issue. Only one study has been conducted with an adequate sample size.38 There are some data to suggest that P50 deficits may be related to impairments in sustained attention and vigilance, though the small number of total subjects included in the two studies5,9 and the lack of any recent studies with patients treated with second-generation antipsychotics preclude a definitive judgment. There is also the suggestion that P50 abnormalities may be related to processing speed,9 which would be consistent with the hypothesis that the cholinergic system may play a role in the pharmacology of processing speed. Other cognitive dimensions have only been minimally studied or not studied at all. The lack of relationships between P50 and measures of explicit verbal and visual memory is somewhat surprising in light of the role of the hippocampus in memory. The absence of a relationship raises the possibility that multiple independent processes are affecting hippocampal function.
Sensory gating abnormalities have been hypothesized to underlie the production of positive symptoms,4 which suggests the co-occurrence of these illness manifestations in patients with schizophrenia. However, abnormal P50 sensory gating is also observed in the family members of patients with schizophrenia,20,23–26 which suggests that P50 gating abnormalities may occur in the absence of overt psychosis. The delineation of the relationship between P50 gating abnormalities and positive symptoms may help to clarify whether abnormal sensory gating is a necessary or sufficient condition for the expression of positive symptoms. The relationship between P50 gating abnormalities and positive symptoms and other symptoms has largely been evaluated in the context of cross-sectional studies. The assessment of P50/symptom relationships is organized according to the following symptom domains: negative symptoms, positive symptoms, global severity score, and other symptoms.
Five studies have failed to find a significant correlation between Scale for the Assessment of Positive Symptoms (SAPS) total score9,40,41 or Positive and Negative Syndrome Scale (PANSS) positive symptom subscale38,42 and P50 T/C ratio. Baker et al.43 found that higher scores on the (BPRS) thought disorder subscale were associated with shorter S1 latency (r = −0.92, p < 0.03), but no other P50 measure.
The majority of studies have failed to demonstrate a relationship between P50 sensory gating and negative symptoms. Seven studies failed to find significant correlations between P50 T/C ratio and other P50 measures and SANS total score9,40,41,44–46 or the PANSS negative symptom subscale scores.38 Adler et al.44 divided patients into two groups: one with negative symptoms and the other without negative symptoms. T/C ratio, S1 latency, and S1 and S2 amplitudes were not significantly different between the two groups; nor were any of these measures related to the Scale for the Assessment of Negative Symptoms (SANS) total score. Light et al.40 found that SANS total score accounted for no more than 2% of the variance in P50 suppression. In the Arnfred study, negative symptom severity was mild, making it difficult to draw conclusions from the data presented.41 In contrast, Ringel et al.42 reported that higher PANSS negative symptom subscale scores were correlated with higher or more abnormal P50 T/C ratios. However, they also reported that patients who met criteria for disorganized schizophrenia had the highest ratios. The authors noted that the PANSS negative symptom subscale includes items that overlap, in part, with the definition of disorganized schizophrenia, which raises the possibility that the observed association between the PANSS negative symptom subscale and P50 T/C ratio was driven by their shared association with the disorganized form of schizophrenia.
Global symptom severity measures, such as BPRS total score and PANSS total score, may be conceptualized as measures of illness severity. Three studies failed to find significant correlations between BPRS total score and any of the examined P50 measures.9,47,48 Boutros et al.49 failed to find any significant correlations among P50 measures and PANSS total score. The sample size in these studies ranged from 2349 to 50.48
Jin et al.50 utilized self-reports of disturbances in auditory or visual perception and found that, contrary to expectation, patients who reported perceptual disturbances (n = 16) had T/C ratios that were not significantly different from those of normal controls (n = 16). In contrast, patients who did not report perceptual disturbances (n = 16) had significantly higher T/C ratios than controls. Jin and colleagues argued that these results do not support the hypothesis that abnormal P50 sensory gating reflects sensory overload or abnormal response to sensory stimuli in schizophrenia.50 However, Light and Braff51 argue that self-reports of perceptual anomalies reflect a voluntary process that is distinct from the pre-attentive process of P50 sensory gating and thus may be unrelated.
Yee et al.45 identified patients with no (n = 15) or mild to severe impairment (n = 7) on the SANS attention scale and demonstrated significant differences in P50 T/C ratios between the “no” and “severe” impairment groups, with worse sensory gating in the severe impairment group. They also found that T/C ratio and S2 amplitude positively correlated to BPRS scores of anxiety-depression (T/C ratio: r = 0.50, p < 0.05; S2 amplitude: r = 0.59, p < 0.01). There were no significant P50 T/C ratio differences between any of the patient groups and the normal controls. The patients were taking either risperidone or fluphenazine decanoate. Baker et al.43 found that higher scores on BPRS anxiety-tension negatively correlated to S1 amplitude (r = −0.99, p < 0.03), but not to either T/C ratio or S2 amplitude. Freedman et al.47 failed to find a significant correlation between P50 and the New Haven Schizophrenia Index.
Finally, Myles-Worsley et al.52 compared P50 T/C ratios in genetically high-risk subjects (n = 44) to those from a group of subjects with prodromal symptoms (n = 43), who were not at high genetic risk for schizophrenia and normal adolescent controls (n = 39). Prodromal symptoms were measured utilizing the Personal Assessment and Crisis Evaluation (PACE). Both the genetic high-risk group and the clinically prodromal group had significantly higher T/C ratios than controls. In addition, within the genetically high-risk group, only subjects with prodromal symptoms had abnormal sensory gating.
In contrast to explanatory models that propose sensory overload or abnormal response to sensory input as the cause of positive, and perhaps negative, symptoms in schizophrenia, there was an almost universal lack of evidence to suggest a cross-sectional relationship between abnormal sensory gating and either positive or negative symptom measures. Baker et al.43 actually found an inverse correlation between severity of thought disorder and S1 latency. The sample sizes of these studies ranged from 1241 to 124,38 with the majority of studies including 20–40 subjects.9,40,42,44–46 Thus, although the lack of relationship between sensory gating and positive and negative symptoms is a consistent finding, only one large sample study has addressed the question.
The only evidence that is somewhat supportive of an association between positive symptoms and abnormal sensory gating comes from the Myles-Worsley et al.52 study, which found higher T/C ratios in subjects not diagnosed with schizophrenia but with prodromal symptoms, in which the definition of prodromal symptoms included either low-grade or brief intermittent psychotic symptoms. This result suggests the hypothesis that abnormal sensory gating may be a marker for the potential to develop positive symptoms, rather than related directly to the severity of their expression.
The majority of studies also failed to find a cross-sectional relationship between P50 and global measures of clinical severity (i.e., BPRS and PANSS total score), though in a longitudinal study of clozapine, reduction in P50 gating was only observed in patients whose overall clinical condition improved (see below).53,54
In order to understand the potential pathophysiological significance of abnormal P50 gating, the effects of antipsychotic and other medications on sensory gating need to be examined. The studies that have examined the relationship between medication treatment and P50 can be described and understood from several different perspectives. We have organized our discussion according to class of medication: conventional; second-generation antipsychotics (SGA); and other. Direct comparisons between conventional and SGAs were included in the SGA category.
Three cross-sectional studies have examined the impact of conventional antipsychotics on P50 gating. Freedman et al.47 found that T/C ratios were not significantly different between medicated (n = 15) and unmedicated (n = 14) patients, though medicated patients had a significant improvement in symptoms. Louchart-de la Capelle et al.38 failed to find any relationship between conventional antipsychotic dose and P50 ratio. In the only study that used the steady state paradigm to address medication effects, Boutros et al.28 compared paranoid (n = 13) and non-paranoid (n = 11) patients in unmedicated and medicated states. They found a significant increase in P50 amplitude in the medicated non-paranoid group.
Nagamoto et al.. have published two papers53,54 describing the effects of clozapine on P50 measures. They reported that six patients who responded to clozapine treatment had a significant decrease in their P50 ratio, when compared to their pre-treatment P50 ratio (pre-clozapine: 0.84 + 0.51; clozapine: 0.56 ± 0.17).53 The change in P50 ratio was largely due to increased S1 amplitude (pre-clozapine: 1.9 ± 1.2; clozapine: 3.4 ± 1.7). The three patients who did not respond to clozapine had no change in their P50 ratio. In the second study, with largely the same sample, Nagamoto et al.54 found a significant reduction in T/C ratios, due to a significant increase in S1 amplitude (there were no significant changes in S2 amplitude), when subjects were clinically stabilized on clozapine, compared to their baseline P50 ratio on conventional antipsychotic agents. The T/C ratio reduction was related to a decrease in BPRS total score (r = 0.64; p = 0.06). One subject had a return of his/her sensory gating impairment after clozapine was discontinued. This study also demonstrated a positive correlation between both T/C ratios and BPRS scores to plasma MHPG (p < 0.05 and p < 0.003, respectively). These studies provide the strongest evidence of a treatment effect and/or symptom improvement effect on P50 gating.
In a cross-sectional study, Becker et al.48 compared patients treated with clozapine (n = 25) to patients treated with conventional antipsychotics (n = 25) and normal controls (n = 25). Patients on clozapine had significantly larger S1 amplitudes (clozapine: 6.4 ± 4.0; conventional antipsychotic: 4.3 ± 2.7; p = .04) and lower T/C ratios (clozapine: 0.57 ± 0.41; conventional antipsychotic: 0.82 ± 0.45; p = .04) than patients on conventional antipsychotics. There were no significant differences between the two groups in S2 amplitude (clozapine: 3.7 ± 4.2; conventional antipsychotic: 2.9 ± 1.4; p = .38). There were no significant difference in P50 ratio or any other P50 measure between patients on clozapine and normal controls.
Four studies have compared the effect of SGAs and conventional antipsychotics on P50 gating. Only one of these studies was a double-blind, placebo-controlled, longitudinal study of medications effects.55 In a 16-week study, Arango et al. compared the effect of haloperidol (baseline P50 ratio: 0.60 ± 0.27; end of study P50 ratio: 0.68 ± 0.26) and olanzapine (baseline P50 ratio: 0.61 ± 0.29; end of study P50 ratio: 0.57 ± 0.26) and found no significant group difference in T/C ratios.55 There were 12 subjects in each group. The three cross-sectional studies had similar findings. Yee et al.45 did not find significant differences in T/C ratios between subjects treated with risperidone (n = 17) and those treated with fluphenazine (n = 5). However, the S2 amplitude was significantly larger in the fluphenazine group (p < 0.05). Light et al.40 found significantly reduced T/C ratios in patients treated with clozapine, risperidone, or olanzapine (n = 13) compared to those treated with conventional antipsychotics (n = 13). There were no significant medication group differences in S1 or S2 amplitudes. The sample size was too small to look at the effect of individual SGAs on P50. Adler et al.56 expanded upon the two Nagamoto studies to include patients taking risperidone (n = 22), olanzapine (n = 31), or quetiapine (n = 9), and compared these groups to patients treated with conventional antipsychotics (n = 34) and clozapine (n = 26). Patients treated with clozapine had significantly lower P50 T/C ratios compared to all of the other treatment groups. Patients treated with any of the three other SGAs did not significantly differ from one another in P50 T/C ratios. Of the other SGAs, only patients treated with risperidone had significantly lower P50 T/C ratios than patients treated with conventional antipsychotic medications (p < 0.03). Adler et al.56 also noted significantly lower T/C ratios in patients who smoked versus those who did not smoke. The smoker and nonsmoker group difference was due primarily to reduced P50 ratios in patients taking clozapine who were also smokers.
Finally, in a recent meta-regression analysis of the relationship between medication and P50 ratio, Bramon et al.1 found no significant effect of antipsychotics on P50 T/C ratio, but they did not divide antipsychotics into conventional or SGA.
Donepezil is a selective acetylcholinesterase inhibitor. In a 6-week, open-label study, donepezil treatment only produced a modest reduction in P50 ratio (baseline: 0.74 ± 0.46; end of study: 0.57 ± 0.26). P50 ratio was normalized in 5 of 9 subjects, who had abnormal P50 ratios at baseline.57 P50 T/C ratio reduction was due to a reduction in S2 amplitude (baseline: 2.2 ± 1.7; end of study: 1.8 ± 1.7).
Ondansetron is a highly selective 5HT3 receptor antagonist. The 5HT3 receptor inhibits acetylcholine release, so the antagonism of this receptor should facilitate the release of acetylcholine and potentially enhance sensory gating. In a double-blind, single-dose, pilot study, Adler et al.58 found that ondansetron significantly normalized the P50 ratio (ondansetron: 0.41 ± 0.40; placebo: 0.80 ± 0.21), through a significant decrease in S2 amplitude.
The majority of studies that have examined the effect of antipsychotics on P50 gating are cross-sectional. The non-random allocation of subjects to a particular treatment in these studies precludes any definitive statements about treatment effects. In addition, only a limited number of studies have addressed the issue of medication effects. In light of these considerations, the following tentative conclusions may be made. First, there is little evidence to suggest that conventional antipsychotics adversely effect sensory gating.1,38,45,47 Second, among the SGAs, there is suggestive evidence that clozapine may normalize P50 T/C ratio. However, the effect of clozapine on P50 T/C ratio differs from the P50 T/C ratio effect of nicotine and other agents that enhance cholinergic function. Nicotine lowers P50 T/C ratio through diminished S2 amplitude, which is thought to reflect enhanced sensory gating through activation of the α7 nicotinic receptor.8 In contrast, clozapine appears to exert its effect through increased S1 amplitude, which suggests that clozapine is not enhancing sensory gating, but is exerting an effect through a different pharmacological/physiological mechanism. The clinical significance of increased S1 amplitude is unknown, though it is noteworthy that the effect of clozapine on S1 amplitude was primarily observed in patients whose clinical condition improved and was correlated with change in MHPG levels.54 Finally, there is little evidence that any of the other SGAs exert a restorative effect on sensory gating. The only double-blind, placebo-controlled study, which examined the comparative effect of a conventional and SGA, did not find a medication effect.56 There is clearly a need for more well-designed studies in this area, since the majority of the studies are limited by either their cross-sectional design or small sample sizes.
The most important observation that emerges from the review of the clinical correlates of abnormal sensory gating is the striking lack of data. The evaluation of the neuropsychological correlates of P50 is characterized by multiple small sample studies, with little overlap in the neuropsychological measures used to assess various cognitive domains. The strongest evidence for a relationship to P50 is with neuropsychological measures of vigilance and sustained attention. However, the lack of adequate studies precludes any assertion about the specificity of this relationship. Moreover, there have been no studies that have examined the relationship between change in P50 and the change in cognitive function, which leaves unanswered the question of whether enhanced sensory gating would be reflected in changes in other cognitive functions.
Most studies have failed to find any significant cross-sectional association between abnormal P50 and the presence or severity of positive or negative symptoms. The lack of association suggests that abnormal sensory gating is not directly associated with the expression of these symptoms. Additional support for this viewpoint comes from the work of Jin et al.,50 who failed to show any correlation between those patients reporting perceptual anomalies and abnormal T/C ratios. However, the demonstration that subjects with prodromal symptoms have impaired sensory gating52 suggests that abnormal P50 ratios may be a marker of those individuals who are at increased risk for the development of positive psychotic symptoms.
Among the conventional and second-generation antipsychotics, only clozapine has been shown to normalize P50 ratios.48,54,55 However, the effect of clozapine appears to be mediated through increased response to the conditioning stimulus rather than diminished response to the test stimulus. The latter effect has been suggested to reflect more closely the sensory gating phenomenon, and it is the effect that is produced by nicotine. There has been little attempt to delineate the clinical or physiological significance of increased S1 amplitude or P50 measures other than S2 amplitude.
Several methodological issues deserve comment. First, most studies employed designs in which data were scored by non-blinded raters, who knew both the group membership and/or the condition of the subjects, as well as the study hypothesis. The use of non-automated data reduction methods, coupled with subjective scoring by non-blinded raters, could have biased the data in favor of confirming the hypothesis under question. A second related issue is the lack of reported reliability data. Most studies do not include the reliability of their measures, which is problematic in light of the non-automated and subjective factors involved in the scoring of the P50 waveforms. Future studies should provide reliability data for the P50 measures and interpret accordingly their study results.
Third, most pharmacological effects were evaluated in the context of non-blind, cross-sectional studies in which the treating physician determined the choice of medication. Only one study evaluated P50 drug effects in a longitudinal, double-blind, randomized clinical trial.55
Finally, the sample sizes for most studies were small. This can lead to inadequate power to test study hypotheses and impairs ability to make interpretations. Fourth, nicotine use is rarely reported or analyzed as a potential confounding variable. Noted exceptions are Adler et al.56 and Louchart-de la Capelle et al.38 Though the effects of nicotine on P50 are known to be short lived, and quickly habituated, the failure to take into account potential differences between smokers and non-smokers further complicates the interpretation of study results.
In summary, there is surprisingly little available data to address the questions posed in the Introduction: What illness components are associated with P50, and what is the clinical significance of enhanced sensory gating? Impaired P50 sensory gating may be associated with attention impairments, but there is little data to connect impaired sensory gating with other cognitive domains. Change in P50 has been shown to be associated with change in clinical status, but the mechanism appears not to be related to enhanced sensory gating. If P50 or sensory gating is to be used to guide drug development, then much more research is required to understand and validate the clinical significance of this impairment.
This study was funded by NIMH grant P30 MH068580.