|Additional related information may be found at:|
|Neuropsychopharmacology: The Fifth Generation of Progress|
Physiological Indicators of the Schizophrenia Phenotype
Robert Freedman, Randal G. Ross, and Lawrence E. Adler
PHYSIOLOGICAL TRAITS AND THE GENETICS OF SCHIZOPHRENIA
The use of physiological traits as alternative phenotypes of the genes that carry risk for schizophrenia was suggested by Paul Meehl (70), in his seminal address to the American Psychological Association, "Schizotaxia, Schizotypy and Schizophrenia." Meehl was concerned that genetic risk was under-appreciated as a clue to the origins of schizophrenia. In his opinion, genetic factors had been particularly neglected by those who were interested in psychological theories of its pathogenesis. He proposed the existence of an inherited schizotaxic factor that is generally expressed as schizotypal traits and less frequently develops into the full clinical illness of schizophrenia in the presence of other factors, such as unusual psychological stress. In many ways, biological psychiatry has also been remiss in failing to consider the genetic aspects of schizophrenia as part of the biological investigation of this illness. For many decades after Meehl's article, biological studies of schizophrenia concentrated on first episode, never-medicated patients, in the hope of dissociating the effects of medication and chronic illness from the pure pathological state. These investigations have become progressively more difficult to accomplish in the current health care environment, which mandates rapid treatment at first presentation. Characterization of biological abnormalities in family members who share a genetic risk for schizophrenia, as Meehl had proposed, avoids many of these problems.
Schizophrenia likely has at least three possible types of pathogenesis: 1) one or more major gene defects; 2) polygenic effects, involving the interactions of small variations in many genes; and 3) environmental or non-genetic effects. Any or all of these may be active in a schizophrenic patient, making the neurobiology and genetics likely to be quite complex. An example of a combined pathogenesis would be deficits in one or two neurotransmitter receptors added to additional smaller deficits in neuronal growth factors and diminished neuronal development because of neonatal viral infection. The result would be compromised brain function, primarily caused by the major defect in neurotransmitter receptors, but significantly compounded by the brain's diminished compensatory capability, caused by the polygenic and environmental components. Characterization of each of these deficits neurobiologically in schizophrenics themselves would be quite difficult, because of the interactions of each with the other and with the secondary effects of the illness, such as medication exposure. It would also be difficult to identify a single genetic defect by linkage analysis. Most of the individuals inheriting any particular gene defect would be normal, because they would not have all the additional polygenic and non-genetic elements needed to produce schizophrenia. However, those relatives who are normal, except for the presence of a single genetic defect, would seem to be better suited for the characterization of the phenotype-genotype relationship than the schizophrenic probands themselves. In the relatives, the neurobiological deficits related to single major gene defects would be expressed in isolation from other pathogenic elements. Even if two or three major gene defects are required to produce schizophrenia, these genes would likely segregate independently in the pedigree so that each could be studied in isolation.
Therefore, it has been suggested that the genetic study of schizophrenia include the characterization of specific neurobiological features that might be more closely linked to specific genetic deficits than the illness itself. Meehl proposed that schizotaxia, his term for the biological expression of the genes that convey risk for schizophrenia, was likely to be a synaptic defect that would make neurons more sensitive to stimuli. This hypothesis was based on work of Peter Venables and other psychophysiologists, who noted that increased sensitivity to stimuli was a very elementary psychophysiological dysfunction found in many schizophrenic individuals (109).
Philip Holzman and his collaborators were among the first to find physiological difficulties in the relatives of schizophrenics. Holzman et al. reported that smooth pursuit eye tracking function was abnormal, not only in schizophrenics, but generally in one of their parents and half their siblings as well (49). He further demonstrated that eye tracking abnormalities were found concordantly in monozygotic twins who were discordant for schizophrenia. His work suggested that an eye tracking deficit was an inherited psychological trait that, in the presence of other as yet unknown factors, could be expressed as the clinical illness of schizophrenia. There have been a number of subsequent efforts to find additional phenotypes, as well as further work on eye tracking dysfunction. Reaction time (31), pre-pulse inhibition of startle (14), and various evoked potential measurements have all been investigated as alternative phenotypes to schizophrenia itself. More recent progress in several of these areas will be reviewed in this chapter.
The identification of genes that convey risk for schizophrenia has become a more realistic goal with the development of the human genome map, which has made detailed linkage studies practical. However, linkage studies of the clinical phenotype of schizophrenia itself have not yielded a consistent gene identification across studies. The use of physiological measures as alternative phenotypes for genetic studies is only beginning to be exploited. We will present several examples of the use of physiological phenotypes for segregation and linkage studies of schizophrenia. Linkage of a specific, physiological phenotype to a genetic locus is informative in two ways. First, linkage is an initial step in the identification of a candidate gene and the investigation of mutations that are associated with the illness. Second, linkage not only points to a candidate's gene candidate, but also it confirms that the physiological abnormality is related to a discrete biological abnormality in schizophrenia. A positive linkage generally requires that the physiological phenotype be closely related to a specific neuronal defect that is caused by a genetic abnormality. Because the physiology of living humans neurons cannot be studied directly, inferences about their dysfunction in schizophrenia are difficult to substantiate or refute. Linkage of a physiological measure to a specific genetic locus in the families of schizophrenics thus also provides unique evidence about the identity of biological deficits in schizophrenia.
The goal of understanding the molecular biology and the neurobiology of schizophrenia necessarily requires that the physiological phenotype be quite elementary, so that it reflects a single gene's effect on a single receptor or other protein. As the level of resolution becomes finer, however, its relevance to schizophrenia becomes more difficult to establish. If a schizophrenic person generates an evoked potential with abnormal latency or amplitude or has an intrusive saccade in a smooth pursuit eye movement, the relevance of that deficit to the failure to hold a job or to have intimate personal relationships seems remote. Some neuro-philosophers have proposed that the link between single neuronal functions and higher order concepts of mind and brain functioning, including—ultimately—consciousness, can never breach the Cartesian mind-brain dualism (21). However, at the level of empirical investigation, correlations are increasingly being made between a specific neuronal dysfunction and a clinical outcome. For example, several groups have shown that defects in attentional mechanisms are more closely related to social difficulties than more overt psychotic symptoms (42,81).
Some investigations have been directed to schizotypal individuals, following Meehl's suggestion that these individuals have a partial expression of the phenotype at the clinical level, so that the effects of a deficit can be seen in isolation from the entire clinical pathology of chronic mental illness. Recently, a number of investigators have associated the presence of particular deficits to specific aspects of schizotypy and, in particular, have shown relationships with negative symptoms (97). In some cases, the deficits are found only in schizotypal subjects who also have a presumed genetic risk for schizophrenia, based on their family history (108). A similar strategy has been followed in studies of children at risk for schizophrenia based on their presumed genetic inheritance of that risk from their schizophrenic mother. Some of these children express physiological abnormalities early in development and thus can be followed for extended periods of time to determine which pattern of deficits might result in various clinical outcomes (101).
The technical demands of using a particular physiological abnormality as an alternative phenotype for schizophrenia are considerable. Many of these phenotypes require extensive technical development to ensure test/retest reliability, diagnostic specificity, and heritability in the normal population. The dependence of the measurement on the clinical state of the patient, including medication status, can induce other complexities. Physiological markers by nature are reflective of the state of the individual at a given instant. This state is generally not only dependent upon the genetic and biological predisposition of the individual, but also upon more non-specific factors, such as anxiety, discomfort, level of psychosis, and general cerebral activity. Thus, the determination of standard conditions, the specification of tasks and recording methods, and extensive research to document all the characteristics of a specific measure are necessary (1,36). This review will point to some of that literature, but readers interested in a particular measure may want to consult more technical literature on each of these issues as well.
EYE MOVEMENT DYSFUNCTION
Studies of oculomotor dysfunction in schizophrenia use a variety of tasks to elicit one or both of the two major types of eye movements: saccades and smooth pursuit. Saccadic movements are high velocity eye movements that move the eye to bring the image of a stimulus from the periphery onto the fovea. Smooth pursuit eye movements (SPEM) are slower and function to maintain the image of a slowly moving stimulus on the fovea by matching eye velocity to target velocity. In some experiments, eye movement tasks are used to elucidate the accuracy of memory and information processing through the use of tasks that require the subject to remember or to compute the position of targets that are not visible.
Smooth Pursuit Eye Movements (SPEM)
Almost 90 years ago, Diefendorf and Dodge (32) first described the phenomenon that schizophrenic patients have difficulty accurately maintaining visual gaze on a predictably moving target. Smooth pursuit eye movement abnormalities in schizophrenia were redescribed in the 1930s (27) and independently rediscovered in the early 1970s (50). They have since become a consistently reproducible physiological abnormality in schizophrenia (64) and are independent of neuroleptic medications (38). Research in this area has focused on four general areas: a) the distribution of SPEM abnormalities in relatives of schizophrenic probands and possible genetic linkage, b) the relationship of SPEM abnormalities to schizophrenic spectrum disorder symptomatology, c) the delineation of the various types of errors which may contribute to SPEM dysfunction, and d) the identification of the underlying neuronal pathology associated with abnormal SPEM.
Initially, studies of the familial distribution of SPEM used qualitative assessments of performance. Several authors (see ref. 52 for a review) used qualitative ratings and found that 34–54% of nonschizophrenic first-degree relatives of schizophrenic probands had abnormal SPEM tracking. Similarly, qualitative ratings were investigated in two independent samples of mono- and dizygotic twins discordant for schizophrenia (47,48). Based on qualitative ratings ("good" vs. "bad" tracking), concordance in monozygotic twins is about twice that of dizygotic twins (e.g., 0.77 vs. 0.40). Holzman and colleagues (46,49) used the results of both the unaffected family members and discordant twin studies to suggest that SPEM dysfunction may be a phenotypic marker for a gene conveying risk for schizophrenia. However, Holzman and coworkers were taken with the finding that several unaffected relatives exhibited abnormal SPEM even when the schizophrenic proband had normal SPEM. From these findings, these authors developed the latent trait hypothesis, which proposed that a single dominant gene conveys risk for schizophrenia and abnormal SPEM, but that penetrances of these two dysfunctions are incomplete and unrelated, so that a carrier of the gene may present with abnormal SPEM, schizophrenia, both abnormalities, or neither. This hypothesis argues that the penetrance of SPEM is relatively high, i.e., most people with the abnormal gene present with abnormal SPEM, and that the penetrance of schizophrenia is lower, i.e., only some people with the abnormal gene develop schizophrenia. Thus, SPEM abnormalities may be a more sensitive marker than the clinical syndrome for identifying gene carriers. With a complex segregation analysis, Grove et al. (43) supplied additional evidence that a single major gene accounts for 68% of the variance in eye tracking.
Blackwood and colleagues (12), using the global SPEM measure of natural logarithm of signal to noise ratio, demonstrated similar effects quantitatively (Fig. 1). While schizophrenic and normal groups each demonstrated a normal distribution of performance scores, family members of schizophrenic probands produced a bimodal distribution—with one group demonstrating a mean score close to that of the schizophrenic probands and the other group demonstrating a mean close to that of normals. The natural logarithm of the signal to noise ratio compares eye position to target position and considers deviations between the two as noise, compared to the target position itself, which is the signal. Iacono et al. (53) reported similar findings (Fig. 2). A related measure is gain, which compares the distance traversed by the target and by the eyes at each time point to determine how closely the eyes have pursued the target.
In nonschizophrenic relatives of schizophrenic probands, the smooth pursuit gain of adult relatives of schizophrenics is correlated with social and interpersonal dysfunction, as measured by the Schedule for Schizotypal Personalities (23); those relatives who have abnormal SPEM are more likely that those relatives with normal SPEM to exhibit schizophrenic-like problems in social functioning. Impaired smooth pursuit is also found in schizotypal (59,99) but not in other personality disorders (99). Similarly, college students with poor eye tracking are more likely than those with good eye tracking to exhibit schizotypal traits or schizotypal personality disorder (97,98), but they are not more likely to experience anxiety, tension, or antisocial behavior (98). Recently, it has been shown that SPEM abnormalities are more likely to be found in schizotypal personality disordered patients who have family histories of schizophrenia than in those who do not have a such history (108).
Although there appears to be strong evidence associating SPEM abnormalities to genetic risk for schizophrenia, a specific gene which predisposes to both abnormal SPEM and schizophrenia has not yet been identified. Arolt et al. (9) recently reported positive evidence for linkage between eye tracking dysfunction and markers at chromosome 6p21-23, a region identified as linked with schizophrenia in previous studies (102). Eight families with 62 members were studied; families were selected because they had two or more cases of schizophrenia. Two SPEM measures were made: smooth pursuit gain and number of saccades during the smooth pursuit task. The task (to follow a constant velocity target) was performed at two different velocities, with targets moving across the visual field at 15 or 30 degress of visual angle per second. Individuals were considered abnormal if one of the four resulting parameters was more than 2.5 standard deviations from the normal or if two of the four were more than 2.0 standards deviations from the normal. The lod score method was used to quantify the linkage. The lod score is the common logarithm of the odds ratio of the likelihood that the marker and the phenotype are linked, divided by the likelihood that they are not. Two markers in this region have two-point lod scores over 3, the minimal requirement for linkage, which indicates a 1000:1 ratio of the likelihood of linkage to the likelihood of no linkage. D6S271 had the largest lod score, 3.51 with no recombination. The lod score for schizophrenia was 1.32, which is not significant. Twenty-two of the subjects had schizophrenia or related psychoses, whereas 39 had eye tracking dysfunction. The higher lod score with eye tracking dysfunction may be related to its greater penetrance in these families, which leads to a higher statistical certainty for the linkage results. D6S271 is an anonymous microsatellite marker, whose relationship to a specific gene is unknown. Thus, the genetic mutation responsible for the linkage result has not yet been identified.
The Arolt et al. report (9) did not identify whether gain or saccade count was the more informative measure. While some have argued that global measurement is the best-supported strategy for distinguishing genetic risk (24), it seems that identification of single gene abnormalities in the context of a multifactorial illness may require a more fine-grained dissection of physiological abnormalities such as eye movement dysfunction. Schizophrenic subjects have a greater tendency to let eye velocity slow below target velocity during the pursuit task (86), requiring frequent saccades to correct for this error (84). Context-inappropriate saccades can also intrude upon otherwise normal smooth pursuit. One subset of these intrusive saccades are leading saccades (also referred to as a subset of anticipatory saccades). Since SPEM tasks generally include a target which is moving in a predictable fashion, the subject quickly generates an internal model of target activity, which includes future target location. There is a context-inappropriate tendency to respond to this internally generated schema by generating a saccade ahead of or "leading" current target location and "anticipating" future target location (Fig. 3) [1,89]. Studies of anticipatory saccades suffer from major differences in operational definitions across laboratories; however, it appears that anticipatory saccades are responsible for more disruption of SPEM in schizophrenic subjects than in normals (89).
Adult relatives of schizophrenic subjects have increased levels of all these types of errors, including problems in sensory-motor integration (26), poorer maintenance of attention on the task (86) and increased tendency to anticipate target motion (16,86). Rosenberg, Sweeney, and colleagues (86) found effects of attention on anticipatory saccades. Although relatives of schizophrenic probands generated more anticipatory saccades, this rate of anticipatory saccades normalized when attentional enhancement procedures were used (Fig. 4).
To explore whether specific measures of abnormal SPEM might be more closely associated with the increased genetic risk for schizophrenia, we attempted to identify obligate genetic carriers, based on their position in the family pedigree. Schizophrenics and both their parents were studied. In most families, only one of the parents has an ancestral family history of schizophrenia. Those parents with the ancestral history for schizophrenia were presumed to be the more likely genetic carrier than their spouses who did not have an ancestral history. The disruptive impact of leading (anticipatory) saccades upon smooth pursuit was elevated in the more likely genetic carrier group over their spouses and over normals, while spouses and normals did not differ from each other. Neither measures of ongoing smooth pursuit nor global SPEM performance sorted with presumed genetic risk in a similar fashion. The disruptive impact of leading saccades thus appears more closely associated with presumed genetic risk than more general eye movement dysfunction measurements (87).
The underlying pathophysiology of abnormal increases in leading saccades has not been fully elucidated. Traditional neuroleptics and atypical neuroleptics do not alter anticipatory saccade rates (65). However, because of the physiological findings implicating nicotinic receptors in the P50 gating abnormality and, because of the high rates at which schizophrenic subjects smoke, the possible impact of nicotine on leading saccades has been explored. Schizophrenics show diminishment of leading saccades to normal levels immediately after smoking their usual number of first morning cigarettes (78). The patients generally smoked two to three cigarettes, and the effect was lost after 10 minutes. Similar, but smaller effects were seen after schizophrenics smoked a single cigarette (60). The high dose requirement and short-lasting effect are consistent with known activation and rapid desensitization kinetics of the a7-nicotinic receptor. Abnormal increases in leading saccades have also been tentatively linked to the site of the a7-nicotinic receptor gene on chromosome 15q13-14 in one pedigree, the same locus where the abnormality in P50 gating (described in the next section) has been linked (79).
The leading saccade measurement has also been studied in children, where it can be reliable measured in children as young as eight years old, with an appropriate decrease in target velocity (89). The number of leading saccades is higher in approximately half of the children of schizophrenic mothers (Fig. 5). If increased leading saccades are reflective of a genetically determined neuronal deficit, then it is possible that the effects of the deficit can be observed prospectively, prior to and during the development of more clinically apparent symptoms, which would be expected to occur in about 10% of these children.
Localization of neuronal structures responsible for abnormal SPEM has been attempted using positron emission tomography (PET). Correlations between SPEM performance (as measured by root mean square error) and localized glucose utilization in the frontal eye fields and the caudate have been reported (Fig. 6) . Additionally, caudate and frontal eye field glucose utilization trends toward a significant inverse correlation. It appears that disruption in the oculomotor cortical-striatal-thalamic circuit may be associated with abnormal SPEM. The relationship between activity in these areas and the expression of specific genetic defects in schizophrenia is as yet unknown.
Saccadic Eye Movements
In their original 1908 study, Diefendorf and Dodge (32) noted that saccades generated by schizophrenic patients did not differ in latency or accuracy from those generated by normal individuals. More recent work generally supports this original finding; when presented with a visual stimulus outside of foveal vision, patients with schizophrenia generate saccades toward that stimulus with normal (39,40) to mildly elevated (67,94) latencies. The accuracy of these visually guided saccades is normal (25) to mildly hypometric in schizophrenic subjects (28,92), with some, but not all, of the hypometria due to neuroleptic medications (28).
Schizophrenics' difficulties in saccadic tasks are enhanced when the task requires a subject either to inhibit a saccade to a task-relevant stimulus or to generate volitionally a saccade to a known location but without a visual stimulus. The memory-guided task consists of three sequential steps. First, the subject is asked to fixate visual gaze on a specific spatial location while a cue stimulus is briefly displayed in another location. The subject then must maintain gaze on the fixation point and inhibit eye movement to the cued location during a brief delay period, during which the cue stimulus is no longer present but the fixation stimulus remains on. Finally, when the fixation point disappears, the subject is asked to move visual gaze to the spatial location at which the cue stimulus was previously present. The ability to inhibit response during the delay period (inhibition of response), latency of the delayed response (inversely related to success at initiating a saccade to a known but not visible stimulus), and accuracy of that response are the dependent measures of interest. Park, Holzman, and Goldman-Rakic (80) showed that performance on such a delayed oculomotor response task is diminished in schizophrenics and some of their relatives (Fig. 7). Two-thirds of schizophrenic subjects and 40% of their relatives performed more poorly by greater than two standard deviations more poorly than normal subjects.
In the antisaccade task, subjects are instructed to look at a central fixation point until presented with a peripheral stimulus. With onset of the peripheral stimulus, subjects are not to look at that stimulus, but instead to look an equal distance in the opposite direction. This movement is called an antisaccade. Again, dependent measures are the ability not to look at the visually presented stimulus (inhibition of response), latency of an appropriate antisaccade, and accuracy of the antisaccade. Schizophrenics demonstrate difficulties with inhibition of inappropriate saccades (28,39,40). Although not always found, increased latency (28,40,80,94) and decreased accuracy of appropriate saccades have also been reported (51,80). Nonschizophrenic relatives of schizophrenic probands also show decreased inhibition (25), poorer accuracy (80), and increased latency (80) during these tasks. Finding of abnormalities in nonschizophrenic relatives of schizophrenic probands suggests that saccadic tasks may also have a role in identifying genetic vulnerability to schizophrenia.
Human stroke victims provide a means to assess the contribution of various brain regions to saccadic task performance. As summarized by Pierrot-Deseilligny (83), studies of human stroke patients have suggested that lesions of the frontal eye fields increase latency and decrease accuracy of memory-guided saccades, while lesions of the prefrontal cortex decrease inhibitory ability and decrease the accuracy of memory guided saccades. Positron emission tomography in adults with schizophrenia also supports the involvement of the frontal eye fields in these tasks (77).
Inhibitory Gating of Auditory P50
Initial studies of P50 auditory sensory gating in schizophrenia were undertaken to test the hypothesis that schizophrenic patients are flooded by a surfeit of sensory stimuli because of a deficit in inhibitory neuronal processing. That hypothesis was originally proposed by Venables (109) and other British psychophysiologists, but it had not been tested physiologically. J.C. Eccles (33) demonstrated the existence of specific inhibitory neuronal pathways in animals by stimulating afferent synaptic pathways in close succession and showing that the electrophysiological response of neurons to the second of the two stimuli was decreased, compared to the first, because of inhibitory post-synaptic potentials generated during the response to the first stimulus. The P50 wave of the auditory evoked potential was used for similar studies in humans, because its amplitude was relatively independent of voluntary attention and might therefore more closely reflect the effects of elementary inhibitory mechanisms (6). Normal subjects, but not schizophrenic patients, showed decreased P50 amplitude to the second of two auditory stimuli presented 500 mesc apart (115) [Fig. 8]. The behavior of P50 in this paradigm was expressed as the ratio of the amplitude of the second (test) response to the amplitude of the first (conditioning) response (P50 testing/conditioning ratio or P50 ratio). The P50 ratio was significantly greater in schizophrenic patients than in normal subjects (6,112) [Fig. 9]. Although an inhibitory post-synaptic potential was the mechanism responsible for the inhibition observed in Eccles' studies in animals, the mechanism responsible for the decrement of the P50 response to repeated stimuli in humans is unknown. Furthermore, the relatively long time between stimuli made a simple postsynaptic inhibitory mechanism unlikely, as such mechanisms are generally active for less than 50 msec following the initial stimulus. Therefore, the general term sensory gating was used to describe the phenomenon, rather than the more specific term inhibition.
Similar differences in the response of P50 to repeated stimuli have been found by Erwin et al. (34), Boutros et al. (13), Judd et al. (57), Lamberti et al. (62), Jin et al. (55), and Clementz et al., (22) [Fig. 10]. One group failed to find suppression of the P50 response in normals and thus found no difference between normals and schizophrenics (58). Several stimulation paradigms, including trains of stimuli of different rates, as well as the paired stimulus paradigm, have been employed (Fig. 11) . Responses have been compared by measuring the ratio of the two stimuli, as well as by computing differences in wave amplitude and area (57,100). In addition to the vertex, frontal and parietal sites have been investigated (57). Variants of dipole modeling have also been used (19,20) [Fig. 12]. Another wave that shows similar differences between normals and schizophrenics in the response to repeated stimuli is the auditory N100 (74,91). The P50 sensory gating deficit has been correlated with poorer performance on the Digit Vigilance Test of the Weschsler Adult Intelligence Scale (29) and increased stimulus generalization in a semantic priming task (110).
The P50 auditory sensory gating paradigm demonstrates a deficit in approximately 90% of schizophrenic patients, whether they are acutely unmedicated and psychotic or stabilized on typical neuroleptic medications (6,35). Acutely psychotic patients do have smaller P50 amplitudes and latencies to the first stimulus; these normalize with neuroleptic treatment (38,103). P50 auditory gating is also significantly impaired in patients with both negative and positive symptoms (7). Thus, P50 auditory gating impairments appear to be a trait deficit in schizophrenia. In contrast, P50 gating abnormalities appear to be state-dependent in several nonschizophrenic diagnoses. For example, elevated P50 ratios in mania returned to normal when the patient became euthymic (7).
Animal models were used to suggest specific neurophysiologic mechanisms for alterations in P50 sensory gating. In the rat, an analog to the P50 waveform originates in the hippocampus (11). Lesion of fibers from the medial septal nucleus to hippocampus impairs gating of auditory response, an observation which suggests a possible role for cholinergic mechanisms. Pharmacological blockade of either muscarinic or high-affinity nicotinic receptors does not impair gating of auditory responses in the rat hippocampus. However, a-bungarotoxin, which binds to the low-affinity, rapidly desensitizing a7-nicotinic receptor, does impair gating of auditory response (66). Alpha-bungarotoxin-sensitive receptors excite inhibitory interneurons in the hippocampus, which release g-aminobutyric acid (GABA). The inhibition of response to the second stimulus appears to depend upon a GABAB-mediated presynaptic inhibition of release of glutamate from afferents to hippocampal pyramidal neurons (45). In addition to this nicotinic cholinergic mechanism, catecholaminergic neurotransmission affects P50 and its inhibition. Augmenting noradrenergic neurotransmission by drugs such as the a2-antagonist yohimbine diminished gating of the hippocampal evoked response. Augmenting dopaminergic neurotransmission did not affect the gating of responses to repeated stimuli but did diminish the amplitude of the initial response (5).
Several studies were performed to assess the applicability of these findings to human schizophrenia. Nicotine-containing gum was administered to relatives of schizophrenics who shared the trait of diminished P50 gating. Nicotine administration significantly improved their gating responses. The effect was brief and consistent with rapid desensitization of the a7-nicotinic receptor (2) [Fig. 13) and Fig. 14]. In a more naturalistic design, schizophrenic patients also had enhanced P50 gating immediately after smoking cigarettes (3). This effect also did not persist. Schizophrenic patients thus may be attempting to self-medicate a deficit in their nicotinic cholinergic neuronal transmission by smoking; their usual pattern of chain smoking may reflect their need to overcome the effects of receptor desensitization. Treatment with clozapine also improves P50 auditory gating, in conjunction with clinical improvement observed on the drug (75). Interestingly, treatment with clozapine also reduces smoking in schizophrenics (69).
Catecholaminergic neurotransmission also appears to regulate the gating of P50 in humans, as predicted from the animal model. The state-dependent deficit in P50 sensory gating in mania appears to be correlated with levels of plasma free MHPG (7). In normal subjects, increased plasma MHPG levels during the initial phase of a recording session are also correlated with increased P50 ratios (113). The effect is intensified by strategies that promote anxiety, such as the cold pressor test (56). Elevation of catecholamine levels pharmacologically by yohimbine also elevates the P50 ratio in normals (4). However, elevated P50 ratios in schizophrenics are not dependent upon catecholamine metabolite levels (7), although the reduction in the initial P50 response seen in unmediated schizophrenics is dependent upon increased plasma homovanillic acid (HVA) levels. Thus, in most subjects, including patients with mania, decreased sensory gating is mediated by noradrenergic neurotransmission. In schizophrenics, however, a nicotinic receptor appears to be deficient. During acute psychotic illness, there is an additional dopaminergic mechanism that further alters auditory sensory processing in schizophrenia.
Half the first-degree relatives of schizophrenic patients have a deficit in P50 sensory gating identical to that observed in the schizophrenic patients themselves (96). The probability of having the deficit is related to the degree of apparent genetic risk (111). The deficit in P50 gating is distributed as an apparent autosomal dominant trait and was therefore used as a phenotype for linkage analysis. Nine families selected to contain multiple cases of schizophrenia were studied; 96 members were phenotyped both for the clinical diagnosis of schizophrenia by Research Diagnostic Criteria and for P50 sensory gating status (Fig. 15). As part of a genome-wide survey with highly polymorphic markers at 10 centimorgan intervals, a dinucleotide repeat polymorphism (D15S1360) was isolated from a yeast artificial chromosome containing the a7-nicotinic receptor gene. For this genetic marker, the lod score for the inheritance of abnormal P50 ratio was 5.30 with no recombination (37) [Fig. 16]. This lod score indicates a likelihood ratio greater than 100,000 to 1 that a gene associated with an abnormal P50 ratio is located near D15S1360 and the a7-nicotinic receptor gene. The lod score for schizophrenia is positive at this location, but less than 3, the usual criterion for linkage. The principal difference between the P50 phenotype and the clinical phenotype is that the P50 deficit is almost completely penetrant, i.e., expressed by every individual who carries the abnormal gene, whereas the full syndrome of schizophrenia is expressed by only a minority of gene carriers. Thus, the statistical certainty of linkage is greater for the P50 ratio than for schizophrenia, as it was in the Arolt et al. (9) linkage study of eye movement dysfunction. Whether the convergence of biological and genetic data implicating the a7-nicotinic receptor in an aspect of the pathophysiology of schizophrenia will be confirmed by finding a mutation in this gene will have to be determined by additional studies.
PREPULSE INHIBITION OF ACOUSTIC STARTLE
The prepulse inhibition of startle is similar to P50 inhibition and is also generally described as a sensory gating paradigm. However, unlike the P50 gating model, which is confined to the auditory system, the acoustic startle paradigm activates a more extensive neural circuit mediating a motor response to an auditory stimulus. The subject listens to two auditory stimuli. The second or startle tone is very loud (115-118 dB) and causes a startle response: an eyeblink in the case of a human subject or a generalized neck and limb contraction in an animal. The amplitude of the eyeblink, as measured at the orbicularis oculi by muscle electromyography (EMG) in the human or by whole body startle in an animal model, is proportional to the loudness of the sound. If a less intense tone, the prepulse auditory stimulus (which can be as little as four dB over a white noise [70 dB] background) is presented to the subject 30–120 msec prior to the second stimulus, the amplitude of the startle response is significantly decreased (41). Components of the startle response that are most commonly measured include latency to the beginning of the response (msec from the startle stimulus onset to the onset of the startle response), latency to peak (msec from the startle stimulus to the peak startle amplitude), amplitude (the size of the EMG response in the human subject), habituation (the decrease in amplitude of the startle without a prepulse tone which results solely from hearing the startle stimulus repeatedly over the course of the experiment), and PPI (prepulse inhibition, or the percentage of the reduction in startle amplitude when a prepulse tone is given vs. the amplitude of the startle response when no prepulse is given). Thus, PPI when expressed as a percentage is proportional to the degree of inhibitory processing or gating; lower PPI means that there is less inhibition of the startle response by the prepulse; higher PPI indicates greater inhibition of the startle response by the prepulse. One limitation to the phenomenon in humans is that 30% of subjects do not show a startle response when the louder sound is presented alone (14).
The prepulse inhibition of acoustic startle is deficient in schizophrenic patients and in schizotypals, compared with normal controls (15,18). This difference has been observed in neuroleptic-treated schizophrenic inpatients and in the absence of significant differences in baseline startle amplitude, peak latency, or habituation (14). In general, however, startle habituation is disrupted in schizophrenics, compared with normal controls, especially if the experiment is of longer duration and no prepulse tones are given (14). In normal subjects, PPI impairment correlated with higher scores of "psychosis proneness" as measured by the Goldberg Index of the Minnesota Multiphasic Personality Inventory (MMPI), but it did not correlate with impaired performance in two tests of perceptual-memory processing (the Stroop test or a test of negative priming) . PPI is not associated only with inpatient status or nonspecific symptoms of anxiety or irritability, because it is not impaired in patients with severe post-traumatic stress disorder, although they have increased baseline startle amplitudes, compared with normal controls (17). However, PPI is impaired in obsessive-compulsive disorder (105). Women have somewhat less PPI than men (104). To date, no studies of PPI comparing schizophrenic patients and their first degree relatives have been completed.
PPI has also been studied in animal models (106). The primary acoustic startle pathways have been clearly identified, but the pathways mediating PPI are not as well delineated, possibly because multiple neurotransmitters may influence the response. However, specific findings appear directly related to both schizophrenia and the pharmacological treatment of schizophrenia. For example, the dopamine receptor agonist apomorphine significantly impairs PPI in animals. This effect is ameliorated by conventional neuroleptics and by clozapine (30). Elements of the circuit mediating PPI include the brainstem reticular formation nucleus accumbens, ventral pallidum, dorsomedial thalamus, and the hippocampus—where there is a possible parallel with P50 auditory gating.
Only one study has compared P50 auditory gating and PPI of startle and found a relationship between long-term habituation of the startle response and initial P50 amplitude (93).
COGNITIVE EVOKED POTENTIALS—P300 AND MISMATCH NEGATIVITY
The P300 abnormality is one of the most extensively investigated electrophysiological deficits in schizophrenia (90). The P300 is a positive wave which occurs 300 msec after a stimulus. Unlike the P50 wave, which occurs only in response to auditory stimuli, P300 waves can be elicited by any sensory modality. The unique feature of the P300 wave is that its amplitude depends upon the significance of the stimulus. In one variant of the paradigm, the subject is presented with a series of stimuli, e.g., two tones, one of which occurs with 80% probability and requires no response and another, of slightly different frequency, which occurs with 20% probability and requires a response, which can be as minimal as silent counting of its occurrence. This paradigm is sometimes called the oddball paradigm, because the subject must recognize and act in response to the odd tone. The maximum amplitude for the P300 response elicited by the infrequent, target tone occurs over the parietal lobes. In another variant of the task, the infrequent tone does not require a voluntary response. Rather, the tone occurs after a very long (ca. 10 sec) interstimulus interval and therefore seems unexpected. The tone may even be quite loud and cause an involuntary startle. In this variant, the P300 wave (called P300a as opposed to the P300b elicited in the first oddball task) is maximal over the frontal lobe. Both P300 variants have been reported to be abnormal in schizophrenia (82) [Fig. 17].
Of course, electrode locations do not rigorously indicate the underlying neuronal source of the response. Recordings from a series of patients with strokes involving the medial temporal lobe suggest that P300a may be generated in the hippocampus, which had been lesioned in these patients (60a). P300b has not been as well localized and is known to persist after surgical removal of the hippocampus. Recordings of neuronal activity in laboratory animals performing similar tasks suggest that infrequent stimuli activate the forebrain cholinergic neurons of the medial septal nucleus, which project to the hippocampus, whose pyramidal neurons are similarly activated (71). That physiology would be consistent with the evidence from stroke patients. The behavioral tasks used to activate P300b have been studied in primates and have resulted in activation of the neurons of the nucleus basalis, another forebrain cholinergic nucleus (85). The nucleus basalis projects widely to the cerebral cortex, including the superior temporal gyrus area associated with lower P300b amplitude in schizophrenia. The paradigm generally used to elicit P300a, i.e., essentially a long interstimulus interval, resembles the conditioning-testing paradigm with P50. The P50 occurring after a long interstimulus interval (the conditioning P50) is larger in amplitude than the P50 occurring after a short interstimulus interval (the test P50). As outlined in the previous section, the human P50 has also been associated with cholinergic innervation, with sources reported in both the hippocampus and the superior temporal gyrus. Comparison of the timing of neuronal activity in response to infrequent auditory stimuli in rats with the human evoked potentials and their animal equivalents suggests that the initial excitation of hippocampal pyramidal neurons by septal cholinergic input correlates with the P50 response, whereas a longer term excitation of hippocampal interneurons correlates with the P300a response (72). It is the synaptic activity associated with this cellular discharge that produces the evoked potential recorded at the skull surface. Cigarette smoking, which normalizes P50 inhibitory deficits in schizophrenics, did not alter P300b in schizophrenics, although it decreased the wave in normals (61). Further correlated study in humans and laboratory animals of the possible interrelationships of the various P300 measures and P50 is clearly warranted.
The P300 measure has a number of strengths. First, the initial finding of low amplitude in schizophrenia is quite robust and has been replicated by a number of laboratories (12,16,73,114). Second, the amplitude and the latency of the wave correlate with several other measures. Many, but not all, investigators have shown that the difference over the left temporal area was greater than the difference over the right temporal area (73,82). The left temporal value correlated with decreased volume of the left superior temporal gyrus and with severity rating on a standardized measure of thought disorder (68) [Fig. 18]. Diminished left superior temporal gyrus volume has been correlated with the severity of auditory hallucinations, as well (10). Third, the latency of the wave has been found to increase with age in schizophrenics at a greater rate than it does in normals; this aspect of the measure suggests a possible relationship to the unknown dementing process that characterizes the clinical course of some schizophrenics (76). The major problem with P300 has been specificity. Diminished P300 amplitude is not unique to schizophrenics but has been observed in a number of illnesses, including alcoholism, dementia, post traumatic stress disorder, and depression. In addition, not all schizophrenic patients have P300 abnormalities. Selection of a more neurodevelopmentally compromised group, including those who ultimately develop tardive dyskinesia, increased the likelihood of low P300 amplitude (44).
P300 has been studied in several populations that carry genetic risk for schizophrenia. Abnormal P300 latency has been evaluated as a indicator of genetic risk in multi-affected pedigrees (12) [Fig. 19]. Relatives had abnormal P300 latencies, but the correlation with putative genetic risk was not significant, although relatives with these deficits have a neuropsychological profile similar to the schizophrenics (95). P300 studies in children at risk for schizophrenia have also been performed. Lower P300 amplitude correlated with the eventual expression of poorer global personality functioning (101) [Fig. 20]. One strength of the finding was that the evoked potential deficit was recorded many years before the personality disorder became manifest. However, this deficit was not significantly associated with the eventual development of schizophrenia. However, persistence of low P300 amplitude into adulthood has been associated with "allusive thinking," an apparent subclinical thought disorder in some normal individuals (114) [Fig. 21].
Thus, there is suggestive evidence that P300 abnormalities are associated with the inheritance of risk for schizophrenia, but the specificity of the association is not clear. There are several possible explanations. P300 abnormalities may represent one of several genetic abnormalities in schizophrenia, whose influence is difficult to detect as a single, specific risk factor. Alternatively, diminished left temporal lobe development may be the underlying biological mechanism. Then, diminished P300 amplitude or prolonged latency may occur as a result of the interaction of genetic and environmental factors affecting temporal lobe integrity. In alcoholics, for example, the diminished P300 amplitude may represent the loss of tissue secondary to alcohol abuse. In relatives of schizophrenics, prolonged P300 latency may occur because of one set of genetic influences, whereas diminished P300 amplitude may represent another. Since P300 amplitude does not appear to segregate in relatives, its abnormality in schizophrenia may be more highly related to environmental factors. Latency appears to have an interaction with the aging process, which may further complicate efforts to detect its relationship to the risk for schizophrenia. In children, diminished P300 amplitude could represent genetically determined temporal lobe abnormalities, as in children of schizophrenic mothers, or environmentally determined abnormalities, as in the case of children of normal mothers. In either case, the resultant personality deficit is similar. If ongoing linkage efforts identify a gene in the families of schizophrenics that it is a major determinant of one of the P300 abnormalities, then the linkage could be used to sort genetic from non-genetic factors.
The mismatch negativity is an evoked potential which shares some of the characteristics of P300. Like P300, the mismatch negativity detects a difference between stimuli. However, the mismatch negativity occurs much earlier, approximately 20 msec after onset of the deviant stimulus. Whereas the P300 depends on the significance of the deviant stimulus, the mismatch negativity simply depends on a difference between the stimuli, regardless of significance. Thus, the question of whether the subject is interested in the task, which is a confound for P300 recording (16), is less problematic for the mismatch negativity. The mismatch negativity is diminished in schizophrenics (Fig. 22) , but it has not yet been studied in the relatives.
This chapter has reviewed several of the physiological deficits associated with schizophrenia and—perhaps more important—possibly associated with the genetic risk for the illness. For each measure considered individually, the nature of the evidence supporting the association of the defect with schizophrenia and its genetic risk varies. Most of the measures have been found in schizotypal individuals, as well as schizophrenics themselves, and most have been found in the first-degree relatives of schizophrenics, including children of schizophrenic mothers. For some P300 measures and for some SPEM parameters, various segregation analyses have been used to suggest a possible shared genetic basis for the defect and for schizophrenia. The abnormality in the inhibitory gating of P50 has been linked to the chromosomal locus of a specific candidate gene, the a7-nicotinic receptor gene. Similar linkage studies are being pursued by several groups using P300 and SPEM as phenotypes. Clearly, the search for genetic defects is currently an extremely valuable use of physiological measures of neuronal dysfunction in schizophrenia.
Findings of genetic linkage and, ultimately, of the underlying mutations will engender re-examination of the physiological basis of each of the measures and their involvement in both normal psychological function and the production of the symptoms of schizophrenia. The measures are sometimes termed "physiological markers," analogous to the term "genetic marker." Just as a genetic marker is not the actual gene defect itself, but rather a nearby polymorphism that points to the gene defect, the implication is that the physiological marker itself has no meaning, i.e., schizophrenics are not psychotic because they misprocess tones or have jerky eye movements. Rather, the defect indicates in some way the presence of a brain abnormality that causes psychosis. The relationship of the genetic marker to the gene defect is clear: the marker and the defect are located on the same chromosome, close to each other. The relationship between elementary physiological measures and the generation of psychosis is not as simple to elucidate.
At this stage in the study of these measures, most of the effort of each laboratory has been devoted to the study of a single measure. Such narrowing of focus generally represents a practical limitation of how many measures can be performed on single subjects before they become fatigued or disinterested. Most laboratories choose to have repeated trials of one measure, to ensure that it has been reliably assessed, rather than single trials of many different measures. Narrowing of focus also reflects the worthwhile scientific goal of investigating the biological and clinical aspects of one measure in depth, as opposed to studying many measures more superficially. However, as different measures become established, correlative studies between them may be helpful to determine whether it is likely that they reflect the same brain deficit. If the measures represent the same biological deficit, genetic studies may also make the correspondence evident, because linkage to the same genetic locus would then be found. The finding of significant linkage to the same or to different genetic loci is a critical test for whether a particular physiological measure indeed reflects part of the inherited pathophysiology of the illness. The test of linkage, supported by the development of the molecular tools for human genetic linkage, makes possible the substantiation of the role of a particular deficit in the pathogenesis of schizophrenia, whereas previously the findings in this field could not be validated or invalidated in any rigorous way.
Interpretation of the significance of a physiological deficit is facilitated in a second way by the finding of genetic linkage. The eventual identification of a gene defects leads to consideration of how a functional deficit found in one brain area might be related to functional defects in other brain areas. Most genes are not expressed in only one type of neuron or in only one brain area. Most are expressed at multiple sites, perhaps including sites outside the brain. These sites can be initially identified by the measure itself, e.g., many of the auditory measures appear to involve either the hippocampus or the superior temporal gyrus. Many of the eye movement measures appear to involve the prefrontal cortex. However, once the likely candidate gene is identified, molecular approaches such as receptor autoradiography, protein immunohistochemistry, and in situ hybridization of the gene's mRNA can be used to determine the distribution of the gene in human brain tissue. For example, the a7-nicotinic receptor gene is expressed in the hippocampus, with a particularly heavy concentration of receptor binding sites on cell bodies and processes of putative interneurons. However, it also expressed in a number of other brain areas, with a particularly heavy concentration of receptors on the nucleus reticularis thalami, a thin sheet of inhibitory neurons that covers most of the thalamus and regulates the flow of synaptic activity to many areas of the cerebral cortex (63). Thus, the functional implications of even this single possible gene defect are likely to extend well beyond the measure originally used to find genetic linkage.
The finding of a series of gene defects correlated with schizophrenia does not in itself explain how individuals become psychotic. Schizophrenia itself is not a single gene trait, so that a comprehensive understanding of its pathogenesis will require identification of how multiple defects, some genetic and some not, interact to produce the characteristic symptoms of the disease. For the P50 gating phenotype, siblings with the deficit who are not schizophrenic have larger hippocampal volumes and lower levels of the plasma dopamine metabolite homovanillic acid (HVA) than other siblings. Schizophrenics, on the other hand, have a P50 gating deficit combined with lower hippocampal volume and higher HVA levels (112) [Fig. 23]. Further studies of which combinations of deficits are necessary and sufficient to produce schizophrenia are obviously required. These studies should also make use of the appearance of some of these markers in childhood, where they can identify children whose clinical course can then be studied prospectively.
In summary, the current goal of physiological measurements is to indicate the presence of genetic alterations. Once the genes are identified, the physiological measures can be reconsidered as elements of brain dysfunction. How these elements interact physiologically, which brain areas are involved, and what stage in development is critical are all questions to be addressed. Thus, although the human brain is not currently well enough understood to comprehend completely how a series of deficits produce psychotic symptoms, physiological measures of specific genetic defect are a necessary next step in the characterization of the brain dysfunctions that cause schizophrenia.