Genetics and intermediate phenotypes of the schizophrenia—bipolar disorder boundary

Abstract

Categorization of psychotic illnesses into schizophrenic and affective psychoses remains an ongoing controversy. Although Kraepelinian subtyping of psychosis was historically beneficial, modern genetic and neurophysiological studies do not support dichotomous conceptualization of psychosis. Evidence suggests that schizophrenia and bipolar disorder rather present a clinical continuum with partially overlapping symptom dimensions, neurophysiology, genetics and treatment responses. Recent large scale genetic studies have produced inconsistent findings and exposed an urgent need for re-thinking phenomenology-based approach in psychiatric research. Epidemiological, linkage and molecular genetic studies, as well as studies in intermediate phenotypes (neurocognitive, neurophysiological and anatomical imaging) in schizophrenia and bipolar disorders are reviewed in order to support a dimensional conceptualization of psychosis. Overlapping and unique genetic and intermediate phenotypic signatures of the two psychoses are comprehensively recapitulated. Alternative strategies which may be implicated into genetic research are discussed.

1. Introduction

Insanity was subdivided by Kraepelin in the 1890s into manic depressive psychosis and dementia praecox, distinguished by symptom profile, course of the illnesses and by overall outcome. Since that time, clinical scientists have discussed whether this is a useful division or a false dichotomy. The question remains whether this categorization conforms to a biological distinction between these two syndromes likely to be molecularly based, or not. Formulating answers to this question highlights the current controversy of whether it is more advantageous to utilize traditional diagnostic categories or to pursue dimensional constructs when examining pathophysiology, etiology and treatment of psychiatric diagnoses. Here we explore the biological characteristics of “psychosis” in two chronic psychotic illnesses, schizophrenia (SZ) and bipolar-I disorder (BD).

DSM-IV distinguishes between SZ and psychotic mood disorders, mainly on the basis of psychosis being the core defining feature of SZ; whereas, in mood disorders, psychosis is a “secondary” phenomenon. In fact, there is no DSM-IV diagnostic category for psychotic BD, although psychosis is included as a specifier for severe mood episodes. Psychosis in mood disorders is treated as a fleeting feature, even though recent research suggests that a subpopulation of BD patients manifests psychosis as a consistent syndrome and has distinctive genetic characteristics (Potash et al., 2001). A dimensional approach to categorization has emerged in the face of categorical inconsistencies, driven by clinical observations and research need. Recently, dimensions have been the target for drug development in SZ, with a focus on cognitive dysfunction; a dimensional organization is applicable more broadly, for pathophysiology and etiology as well. The idea is clinically based and practical, taking “component symptom complexes” and targeting these for evaluation, mechanistic hypotheses and therapeutics (Hyman and Fenton, 2003) Component symptom complexes are groups of symptoms which associate in an illness and appear to have a common pharmacology, neural basis and putative pathophysiology. For example, the symptom construct “psychosis” could be supported by a common disease mechanism across different psychiatric diagnoses, a concept which is important for clinical prediction, mechanistic research and drug development. Moreover, the relevance of this simpler approach for developing pertinent animal models is evident.

There is consistent evidence that genes contribute to the etiology of psychosis. Recent findings from genetic studies provide evidence for an overlap in genetic susceptibility across the traditional psychosis categories of SZ and BD. Identified candidate genes show strong associations with component symptom complexes, such as psychosis or mood symptoms, that are not projected directly onto either of the two Kraepelinian disease entities. Genetic studies suggest that psychosis may be conceptualized as a clinical phenotype with at least partially unique genetic etiologies, independent of a formal diagnosis. Hypothetically, genes interacting with environmental factors, may determine vulnerability to psychosis. Depending on additional syndrome-specific genetic influence and environmental interactions, psychosis may co-exist with different clinical phenotypes, e.g., mood symptoms or cognitive dysfunction, generating categorical diagnoses. We review existing epidemiological, molecular genetic and intermediate phenotype studies in this paper to support a dimensional conceptualization of psychosis. Historically, genetic and intermediate phenotypes studies started first in SZ research, and by the present time a large volume of data has been collected, although this has not yet led to a clear understanding of pathophysiology of the illness. More recently there have been attempts to explore similar intermediate phenotypes, as well as putative candidate genes in BD, however these strategies are relatively novel for BD research and the available data are still limited. In each section of this manuscript we will review the data in SZ first and then will present the studies available in BD, with emphasis on overlapping, as well as unique genetic and pathophysiologic characteristics of the two illnesses.

2. The genetics of psychosis: schizophrenia and bipolar disorder

2.1. Linkage studies and studies of candidate genes in schizophrenia

There is consistent evidence that a genetic component contributes to the etiology of SZ, however little is known for certain about particular genomic regions or individual risk genes. Family studies show that SZ and other psychotic disorders aggregate in families. The life-time risk for developing SZ increases approximately 8–12 folds in first-degree biological relatives of SZ probands. The concordance rate for SZ is higher in monozygotic (47–50%) than in dizygotic twins (12–16%), suggesting a strong heritability component for the illness. It has been reported that the more severe the disorder, the more likely it is that twins will be concordant. Some studies report the concordance rates for monozygotic twins at over 80% in cases of severe SZ with typical core symptoms (Franzek and Beckmann, 1996) In the past decade, numerous genetic studies have implicated chromosomal loci and candidate risk genes associated with SZ (Baron, 2001; Harrison and Weinberger, 2005; Owen et al., 2004) Several large meta-analyses have found evidence of numerous genetic linkages of which 6p24–22, 1q21–22, and 13q32–43 are the best supported (Lewis et al., 2003; Owen et al., 2004). In addition, suggestive linkages have been reported in 8p21–22, 6p22, 6q21–25, 22q11–12, 5q21–33, 10p15–11, and 1q42 (Baron, 2001; Berrettini, 2000; Lewis et al., 2003; Owen et al., 2004; Segurado et al., 2003).

Association studies have identified several putative candidate genes for SZ. Some of these risk genes include, but are not limited to, DISC1 (disrupted in schizophrenia 1) on 1q42 (Ekelund et al., 2001, 2004; Millar et al., 2000); COMT (catechol-O-methyltransferase) on 22q11 (Egan et al., 2001b; Malhotra et al., 2002; Shifman et al., 2002; Wonodi et al., 2006); dysbindin (distrobrevin-binding protein 1) on 6p22.3 (Bray et al., 2003; Funke et al., 2004; Numakawa et al., 2004; Schwab et al., 2003; Straub et al., 2002; Williams et al., 2004a); NRG1 (neuregulin 1) on 8p12 (Corfas et al., 2004; Hall et al., 2004; Petryshen et al., 2005; Stefansson et al., 2003; Tosato et al., 2005); DAOA (G72)/G30 (D-amino acid oxidase activator (G72)/G30) on 13q33 (Hall et al., 2004; Korostishevsky et al., 2004; Schumacher et al., 2004; Wang et al., 2004); BDNF (brain derived neurotrophic factor) (Buckley et al., 2007; Gratacos et al., 2007; Ho et al., 2007); RGS4 (regulator of G protein signalling 4) on 1q23 (X. Chen et al., 2004; Chowdari et al., 2002; Morris et al., 2004; Williams et al., 2004b); DRD4 (Glatt and Jonsson, 2006; Shi et al., 2008b); MTHFR, PPP3CC, GABRB2 and TP53 (Shi et al., 2008b); although the reports considerably vary. Protective allele associations in DAO, IL1B, and SLC6A4 were also reported in a recent meta-analysis (Shi et al., 2008b). Recent large scale genetic studies have yielded rather modest results. For example, a recent attempt to establish genome-wide associations for SZ in the CATIE study which involved nearly 750 SZ patients and a similar number of controls, and analyzed almost half a million SNPs showed that not one of the candidate genes met the rigorous statistical requirements needed to show it was a risk factor (Sullivan et al., 2008). As SZ is a relatively rare illness with presumably multiple genes of small effects involved in its ethiology, available to date studies are still statistically underpowered. Geneticists suggest that tens of thousand cases and controls may be needed to find firm associations (Abbott, 2008). Further, understanding the effect of risk genes will undoubtedly be complex. Even though several risk genes have been implicated, the association variations are different in different populations and it is therefore difficult to determine the biologic effect of each risk gene. Interactions between risk genes add to the complexity of the picture. In addition, the phenomenological heterogeneity of psychotic disorders, as well as lack of clear boundaries and biologically based definitions in the existing diagnostic categories may contribute to difficulties in genetic studies.

2.2. Linkage studies and studies of candidate genes in bipolar disorder

The results of genetic studies in affective psychoses are less consistent; nonetheless, the familial aggregation of BD and major depressive disorder has been observed. First-degree relatives of individuals with BD have elevated rates of bipolar I disorder (4–24%), bipolar II disorder (1–5%), and major depressive disorder (4–24%). Genome-wide scans of BD have produced inconsistent evidence for specific linkage, despite interesting leads in earlier studies (e.g., chromosomes 2 (Liu et al., 2003), 11 (Egeland et al., 1987), 16 (Ross et al., 2008), 18 (Berrettini et al., 1994), 20 (Ross et al., 2008), and ‘X’ (Baron and Risch, 1987)), suggesting that the majority of psychiatric diseases, including SZ and BD, are etiologically heterogeneous and polygenic (Berrettini, 1999). Several meta-analyses of BD data sets indicated no significant linkages per a priori criteria, but the most promising linkages were to 16q12, 18q22, 21q21, 4p16, and 12q24 (Berrettini, 2000; Liu et al., 2003; Segurado et al., 2003), 16p (Ross et al., 2008), 13q and 22q (Badner and Gershon, 2002). Recently, the Wellcome Trust Case Control Consortium published results of a gemone-wide analysis which included 2000 cases of BD and 3000 controls with genotyping nearly 500 thousand SNPs (Wellcome Trust Case Control Consortium, 2007). The strongest signal was seen in 16q12 region, with the moderate evidence of association in 2p25, 2q12, 2q14, 2q37, 3p23, 3q27, 6p21, 8p12, 9q32, 14q22, 14q32, and 20p13.

With regard to individual risk genes, recent studies have reported associations between BD and BDNF (Fan and Sklar, 2008; Liu et al., 2008; Muller et al., 2006; Neves-Pereira et al., 2002), with evidence of specific polymorphisms associated with rapid cycling; NRG1 (Georgieva et al., 2008; Goes et al., 2009; Green et al., 2005; Stefansson et al., 2003; Williams et al., 2006), DISC1 (Ekelund et al., 2001, 2004; Macgregor et al., 2004; Millar et al., 2000), dysbindin (Raybould et al., 2005), and DAOA (G72)/G30 (Y.S. Chen et al., 2004; Green et al., 2004; Hall et al., 2004; Hattori et al., 2003; Korostishevsky et al., 2004; Schumacher et al., 2004; Wang et al., 2004). The Wellcome Trust Consortium (2007) reported that there is support for the previously suggested importance of GABA neurotransmission (rs7680321 in GABRB1 encoding a ligand-gated ion channel [GABA A receptor, beta 1]), glutamate neurotransmission (rs1485171 in GRM7 [glutamate receptor, metabotropic 7] and synaptic function (rs11089599 in SYN3 [synapsin III]). Examining intermediate phenotypes in BD has been strongly advocated, although rarely done, as a critical element in identifying informative genetic loci (Glahn et al., 2004; Lenox et al., 2002; MacQueen et al., 2005). Studies in circadian rhythm provide a compelling example of how phenotypic approach can be used to identify genetic risk factors for BD.

Recent reports suggest that CLOCK (Benedetti et al., 2003, 2007; Lamont et al., 2007; McClung, 2007; Shi et al., 2008c), BmaL1 (Mansour et al., 2006; McClung, 2007; Nievergelt et al., 2006), TIMELESS (Mansour et al., 2006; Shi et al., 2008c), and PERIOD1–3 (Nievergelt et al., 2006; Shi et al., 2008c) are candidate loci associated with the circadian rhythm BD phenotype, although majority of reports are preliminary and not all studies confirm these associations (Bailer et al., 2005; Nievergelt et al., 2005; Shiino et al., 2003).

In the recent comprehensive review in genetics of BD (see Kato (2007) for review) additional associations with TRPM2 (21q22.3), GPR50 (Xq28), Citron (12q24), CHMP1.5 (18p11.2), GCHI (14q22–24), MLC1 (22q13), GABRA5 (15q11-q13), BCR (22q11), CUX2, FLJ32356 (12q23-q24), and NAPG (18p11) have been suggested, although future replicating studies are warranted.

2.3. Commentary: genetics in schizophrenia/bipolar disorder psychosis

Recent findings from genetic studies provide substantial evidence for an overlap in genetic susceptibility across these traditional categories, consistent with the overall hypotheses of this paper (Ivleva et al., 2008). Recent large meta-analyses of linkage studies based on the clinical phenotype have identified several loci that overlap between SZ and BD including 1q32, 10p11–15, 13q32, 18p11.2 and 22q11–13 (Badner and Gershon, 2002; Baron, 2001; Berrettini, 2000; Bramon and Sham, 2001; Sklar et al., 2002). Although preliminary, linkage studies of psychotic BP show evidence for suggestive genome-wide linkage to chromosomes 8p and 13q (Goes et al., 2008) and suggestible linkage to 5q33, 6q21, 8q24, 15q26, 17p12, 18q21, and 20q13 (Park et al., 2004). These studies support the hypothesis that psychosis can be conceptualized as a clinical phenotype with at least partially specific genetic etiologies, independent of any traditional categorical diagnosis. Family studies show that SZ and affective psychoses occur together in the same families, suggesting shared familial risk. Recent reports confirm that the increased risk for psychotic illness in relatives of SZ persons is not confined to SZ alone (Arajarvi et al., 2006; Henn et al., 1995). On the other hand, BD has been associated with increased risk of SZ in relatives (Valles et al., 2000), consistent with the hypothesis that the same genes could contribute susceptibility to both illnesses. Further, twin studies suggest that a SZ diagnosis in one twin increases the risk for SZ and affective disorders (including BD) in the co-twin (Cardno et al., 2002; Farmer et al., 1987). A recent large scale study which included almost 36,000 cases of SZ and over 40,000 cases of BD from a multi-generation register in Sweden reported increased risks for SZ in relatives of probands with BD, including adopted children of biological parents with BD. Heritability for SZ and BD was 64% and 59%, respectively. The co-morbidity between disorders was mainly (63%) due to additive genetic effects common to both disorders (Lichtenstein et al., 2009).

Recent reports have shown that the candidate genes originally implicated in SZ may also influence susceptibility to BD. Identified candidate genes show associations with component symptom dimensions, such as psychosis or mood symptoms, in SZ–BD boundary. Recent association studies of psychotic BD and subtypes such as mood-incongruent psychotic BD have revealed modest positive results for several candidate susceptibility genes, including dysbindin, DISC1, and NRG1 (Goes et al., 2008). For instance, Green et al. (2005) found the association of NRG1 in BD with mood-incongruent psychotic symptoms, as well as with SZ with lifetime manic episodes, suggesting that NRG1 may confer susceptibility to a phenotype with combined features of psychosis and mania. A recent study reported additional evidence for association between psychotic BD phenotype and NRG1 (Goes et al., 2009). In addition, studies support that variation in the DISC1 gene influences susceptibility to disorders of the psychosis spectrum, including SZ, schizoaffective disorder and BD (Owen et al., 2007). Dysbindin has been extensively implicated in SZ, and recently associated with BD with recurrent psychotic symptoms (Raybould et al., 2005). These tentative results are consistent with the hypothesis that the sub-phenotype of psychotic BD may represent a clinical manifestation of “overlap” genes between SZ and mood disorder syndromes. A recent study implicated a genetic variation in G72 (DAOA)/G30 in susceptibility for major mood episodes across the traditional SZ and BD categories (Williams et al., 2006), suggesting that even though this locus was originally described as a SZ risk gene, it may be more strongly associated with mood symptoms than with psychosis within the SZ/BD continuum, however not all studies support this finding (Maheshwari et al., 2009; Shi et al., 2008a). In summary, molecular genetic studies, as well as epidemiological and family studies, have shown evidence that SZ and BD partly share a common genetic cause. These data challenge the current nosological dichotomy between the two types of psychosis, and are reflective of the need for reappraisal of these disorders as distinct diagnostic entities.

3. Intermediate phenotypes of psychosis: schizophrenia and bipolar disorder

Intermediate phenotype studies in psychosis were initially implemented in SZ research and have led to the identification of several important neurocognitive, neurophysiological and anatomical illness markers. An ideal intermediate phenotype for a brain disease would be a measure that is associated with the syndrome, heritable, state-independent, and co-segregates in families while being expressed in unaffected family members (Bearden et al., 2006b; Berrettini, 2005; Doyle et al., 2005;Gottesman and Shields, 1973; Skuse, 2001; Waldman, 2005). These include oculo-motor abnormalities; deficits in P50 sensory gating [inhibition of the 2nd positive event-related potentials (ERP) occurring around 50 ms of the event in a paired-click paradigm] and sensory-motor gating [inhibition of the startle response by a prepulse (PPI)]; abnormal neurocognitive performance; as well as structural brain abnormalities. In contrast to the magnitude of the data published in SZ, much less is known about intermediate phenotypes of other psychotic illnesses, although selected candidate phenotypes have recently begun to be explored in psychotic affective disorders (Thaker, 2008). There are specific elements within these neurophysiological, anatomic and cognitive measures which are probable psychosis intermediate phenotypes. We will discuss the similarities and distinctions in intermediate phenoltypes across the two main psychotic illnesses, SZ and psychotic BD.

3.1. Neurocognitive intermediate phenotypes

3.1.1. Neurcognitive deficits associated with schizophrenia

While cognitive difficulties are commonly associated with a large number of psychiatric disorders, disturbances seen in SZ patients are generally more profound and debilitating. The presence of specific cognitive deficits is not yet part of the diagnostic criteria for the SZ but is recognized as a common and consistent feature of the disorder nonetheless (Bleuler, 1950; Kraepelin, 1919). There appears to be general cognitive deficit associated with SZ (Dickinson et al., 2008; Keefe et al., 2006a). However, there is a wide variety of cognitive deficits that have been measured and documented in SZ patients, but the findings are somewhat disparate, leaving only a few promising candidate neurocognitive intermediate phenotypes. Considerable research into the cognitive difficulties associated with SZ has focused on attention (Braff, 1993; Cornblatt and Keilp, 1994; Nuechterlein, 1977; Nuechterlein and Dawson, 1984; Seidman, 1983; Shakow, 1962). Deficits in sustained attention have consistently been found in SZ patients and show a great deal of promise as a candidate neurocognitive intermediate phenotype (Chen and Faraone, 2000; Cornblatt and Malhotra, 2001). The accepted measurement paradigm for sustained focused attention is the use of a continuous performance task (Braff, 1993; Cornblatt and Keilp, 1994). These evaluations assess the ability to detect and respond to a specific target when presented within a series of distracter stimuli. Deficits in working memory also constitute another prominent candidate neurocognitive phenotype in SZ. Working memory is generally defined as a cognitive mechanism which allows information, for a very brief period of time, to be stored and utilized. Working memory is typically assessed using tasks which require the patients to maintain and manipulate, auditory or visual stimulus for a very brief period of time. Such evaluations often include, but are by no means limited to, the Wechsler Memory Scale, Third Edition (WMS-III) (Wechsler, 1997) subtests such as the Digit Span (forward/backward), Number-Letter Sequencing, and Spatial Span, to name a few. Not surprisingly, disturbances in working memory are also associated with deficits in other areas of higher level cognitive processing such as problem solving and language comprehension (Hutton et al., 1998b), as well as being associated with a generally poorer prognosis, with patients having working memory deficits typically showing lower levels of social and occupational functioning (Green et al., 2000; Kopelowicz et al., 2005; Smith et al., 2002).

Verbal declarative memory deficits are prominent and consistently well documented in patients with SZ (Aleman et al., 1999; Cirillo and Seidman, 2003; Saykin et al., 1991), and resultantly, may represent the most promising of the intermediate neurocognitive phenotypes. The verbal declarative memory deficits appear to be a result of difficulty in memory storage and the flexible use of knowledge, specifically encoding; an interactive multi stage process, which, generally speaking, translates perceptual stimulus into meaningful, workable, memory. Deficits in retrieval, as well as, increased rates of forgetting have also been consistently documented in patients with SZ; however, difficulties in encoding are the most prominent. Among the commonly used measures for verbal declarative memory are the California Verbal Learning Test, Second Edition (CVLT-II) (Delis et al., 2000) and the Logical Memory subtest from the WMS-III, which expose the subject to an auditory stimulus, a list of words or a brief paragraph, and requires the subject to recall the material after a short delay. These deficits appear to be relatively stable over time (Asarnow and MacCrimmon, 1978; Nuechterlein et al., 1992; Wohlberg and Kornetsky, 1973). Difficulties in attention, working memory, and verbal declarative memory have been consistently found in first episode, and remitted patients, and are not a function of active symptoms (Heaton et al., 2001; Hill et al., 2004; Saykin et al., 1994; Tyson et al., 2005).

The work examining the co-segregation of neurocognitive deficits within families of SZ patients is not complete. Some studies have found attentional deficits in unaffected relatives at a higher rate than in the general population (Asarnow et al., 2002; Egan et al., 2000; Finkelstein et al., 1997; Nuechterlein et al., 1992; Orzack and Kornetsky, 1966; Saoud et al., 2000; Seidman et al., 1998; Snitz et al., 2006; Sponheim et al., 2006), although there are some inconsistencies in findings (Maier et al., 1992; Mirsky et al., 1992), which again maybe a function of the method of assessment. There has been very little research examining the co-segregation of working memory within families of SZ probands, and no findings are available. A great deal of work has been done in the area of verbal declarative memory, with the most promising findings in the area of encoding deficits. These studies have consistently shown that biological relatives of SZ patients perform worse on verbal declarative memory tasks than community controls (Cannon et al., 1994, 2000; Cirillo and Seidman, 2003; Faraone et al., 1995, 1999; Seidman et al., 2006; Skelley et al., 2008; Snitz et al., 2006).

3.1.2. Neurocognitive deficits associated with bipolar disorder

A great deal of work has begun to focus on identifying intermediate phenotypes for BD (Hasler et al., 2006). Interestingly, patients with BD manifest cognitive deficits similar to those found in SZ patients (Hill et al., 2008; Schretlen et al., 2007; Seidman et al., 2002b), with attention, working memory, and verbal declarative memory difficulties being among the most prominent, and representing the most likely neurocognitive intermediate phenotypes. However, considerable heterogeneity exists within the BD population, with some patients manifesting profound deficits, and others showing little or none in neurocognitive information processing. It appears that the greater the burden of illness associated with the disorder, as manifest in the number of past manic episodes, length of illness, number of past hospitalizations (Robinson and Ferrier, 2006) and history of psychosis (Glahn et al., 2007; Martinez-Aran et al., 2008; Smith et al., 2009), the greater the neurocognitive deficits typically found in patients with BD. BD patients with a history of psychosis appear to have a slightly different profile of neurocognitive deficits than those without a history of psychosis, manifesting more severe impairment on measures of executive functioning and spatial working memory (Glahn et al., 2007). More severe BD patients, especially those with psychosis, begin to resemble SZ patients with regard to the level and profile of neurocognitive deficits.

The most prominent and consistently documented areas of cognitive difficulties in BD are found in attention (Arts et al., 2008; Clark et al., 2002; Swann et al., 2003; Wilder-Willis et al., 2001), working memory (Arts et al., 2008; Gourovitch et al., 1999; McGrath et al., 1997; van Gorp et al., 1999; Zalla et al., 2004; Zubieta et al., 2001), and verbal declarative memory (Antila et al., 2007; Arts et al., 2008; Atre-Vaidya et al., 1998; Bearden et al., 2006a, 2006b; Ferrier et al., 1999; Hill et al., 2008; Kieseppa et al., 2005; van Gorp et al., 1999; Zubieta et al., 2001), of these, working memory and verbal declarative memory may show the most promise as neurocognitive intermediate phenotypes for BD. The neurocognitive deficits associated with BD are generally regarded as state specific, although the deficits typically remain, in a attenuated form, during periods of euthymia (Clark et al., 2002, 2005; Deckersbach et al., 2004; Ferrier et al., 1999; Fleck et al., 2003; Kieseppa et al., 2005; Martinez-Aran et al., 2004; Rubinsztein et al., 2000; Thompson et al., 2005; van Gorp et al., 1998). Deficits in sustained attention (Clark et al., 2002, 2005; Swann et al., 2003; Wilder-Willis et al., 2001), working memory (McGrath et al., 1997; Zalla et al., 2004; Zubieta et al., 2001), and verbal declarative memory (Atre-Vaidya et al., 1998; Ferrier et al., 1999; van Gorp et al., 1998, 1999; Zubieta et al., 2001) have been consistently documented in euthymic patients with BD. As noted earlier, several additional factors appear to impact the manifestation of cognitive deficits.

There is evidence for co-segregation of neurocognitive deficits in biological relatives of probands with BD, albeit limited given the relatively few studies and the inconsistent findings (Arts et al., 2008; Ferrier et al., 2004; Gourovitch et al., 1999; Kieseppa et al., 2005; MacQueen et al., 2004; Pierson et al., 2000; Zalla et al., 2004).

3.1.3. Heritability of neurocognitive abnormalities and genetic correlates

A wide range of heritability estimates for sustained focused attention have been reported for SZ probands, with estimates ranging from 0.48 to 0.62, depending upon the specific testing paradigm (Asarnow et al., 2002; Christensen et al., 2006; Conklin et al., 2000; Gochman et al., 2004; Saoud et al., 2000; Schubert and McNeil, 2005; Sitskoorn et al., 2004; Tuulio-Henriksson et al., 2003). Heritability rates for working memory have been estimated to range from 0.36 to 0.45 for biological relatives of SZ individuals (Conklin et al., 2000; Delawalla et al., 2006; Diwadkar et al., 2001; Johnson et al., 2003; Krabbendam et al., 2001; Park et al., 1995; Snitz et al., 1999; Tuulio-Henriksson et al., 2002, 2003), depending upon the task. Toulopoulou et al. (2003) reported finding a heritability rate for WM of 0.65 in a sample of dizygotic twins. Heritability rates for SZ patients and their first-degree relatives for verbal declarative memory ranging from 0.21 to 0.49 have been reported (Cannon et al., 2000; Christensen et al., 2006; Delawalla et al., 2006; Egan et al., 2001a; Krabbendam et al., 2001; O’Driscoll et al., 2001; Schubert and McNeil, 2005; Toulopoulou et al., 2003). Several studies have reported associations between various candidate genes and neurocognitive intermediate phenotypes in SZ. Gene complex G72/G30 has been associated with working memory and attention in SZ patients and siblings (Goldberg et al., 2006). An association has been reported between chromosomal region 1q, a commonly studied region in SZ, and spatial working memory (Gasperoni et al., 2003). The 1q23 locus has been associated with neurocognitive deficits in SZ probands and their unaffected relatives (Husted et al., 2009). Studies of individual candidate genes have suggested that variant alleles in the DISC1 gene are associated with working memory in SZ patients and their unaffected twins (Burdick et al., 2005; Gasperoni et al., 2003). Verbal declarative memory in SZ patients was associated with DISC1 and translin-associated factor X genes (1q42), as well as with 4q21 (Cannon et al., 2000); whereas visual working memory was linked to 2q (Paunio et al., 2004). The COMT gene has been associated with neurocognitive deficits in SZ probands (Bilder et al., 2002; Burdick et al., 2007; Egan et al., 2001b) and their unaffected relatives (Rosa et al., 2004).

In BD the most robust heritability findings are for working memory (Gourovitch et al., 1999; Kieseppa et al., 2005; Zalla et al., 2004) and verbal declarative memory deficits (Gourovitch et al., 1999; Kieseppa et al., 2005; Zalla et al., 2004), with research on sustained attention outstanding. Interestingly, Antila et al. (2007) noted significant heritability for deficits in attention and working memory (0.64–0.69, with the exception of digit span forward), but failed to find significant heritability for verbal declarative memory. While findings in BD typically have supported the presence of cognitive deficits in the areas of sustained attention, working memory, and to a lesser degree verbal declarative memory, questions of heritability remain, with study results to date being inconclusive. There have been preliminary findings that appear to relate BDNF gene polymorphisms and executive functioning (Rybakowski et al., 2003), but very little is known of the genetics underlying the neurocognitive deficits found in BD specifically. Polymorphisms in the COMT Val158/108Met had shown some promise, but in a recent meta-analysis of studies of psychiatric patients, Barnett et al. (2008) concluded that neurocognitive deficits were not strongly associated with the COMT gene. In a recent study, Burdick et al. (2007) evaluated several SNPs within the COMT gene, concluding that rs165599, but not Val158Met, was associated with neurocognitive deficits, specifically verbal declarative memory, in BD probands.

3.1.4. Effect of medications on neurocognition

Deficits in sustained focused attention (Finkelstein et al., 1997; Keefe et al., 2006b; Sax et al., 1998; Wohlberg and Kornetsky, 1973), working memory (Barch et al., 2001; Carter et al., 1996), and verbal declarative memory (Brebion et al., 2004; Saykin et al., 1994) have been found in medication-naïve individuals with SZ and those no longer taking medications. All three candidate neurocognitive intermediate phenotypes, attention, working memory, and verbal declarative memory, appear to be relatively independent of the effects of medication, being neither exacerbated nor alleviated. There is however, evidence that small improvements in these cognitive deficits may be associated with the some of the second generation antipsychotics (Green et al., 1997; Keefe et al., 2007a,b), as well as typical antipsychotic agents (see Mishara and Goldberg (2004) for review). Recent studies have suggested that galantamine may provide some improvement in neurocognitive functioning, but the findings have been limited to verbal memory (Buchanan et al., 2008; Dyer et al., 2008).

In a recent metal analysis Bora et al. (2009) concluded that the medication or combination of medications used to treat BD appear to have an adverse effect on psychomotor speed and sustained attention. Increased antipsychotic usage in patients with BD has been associated with increased impairment in semantic fluency, verbal learning and recognition memory (Jamrozinski et al., 2009).

3.1.5. Commentary: neurocognitive deficits in schizophrenia/bipolar disorder psychosis

Several cognitive intermediate phenotypes have been identified in SZ and BD which appear to fit, albeit in varying degrees, the model of an intermediate phenotype, namely, associated with the illness, present in all phases of illness, co-segregated in family members with heritability and genetic correlates, and unaffected by medications. The cognitive deficits found in SZ patients appear to be more ubiquitously distributed, temporally stable, and profound than those found in BD patients, with the primary domain candidates being sustained attention, working memory and verbal declarative memory.

In BD, cognitive deficits are more variable, often have a fluctuating course, are less debilitating, with the deficits in verbal declarative memory appearing to be the strongest candidate, followed by working memory difficulties, and to a lesser extent, problems with sustained attention. Because these neurocognitive alterations are characteristically state dependent, and can worsen over time as a function of disease burden, their heritability is less certain and may not have a genetic substrate.

There appears to be a continuum of cognitive dysfunction along which SZ patients consistently manifest the most severe disturbances, and BD, on average, showing a similar, but less severe pattern of dysfunction. In general, the patterns of cognitive deficits in SZ patients appear to represent, given the limitations of the available data, stronger candidate intermediate phenotypes, than found in BD. BD patients may have what appears as an attenuated version of a profile of cognitive deficits similar to those found in SZ. These candidate neurocognitive phenotypes may be more strongly represented in the severe bipolar probands, and be more overtly expressed in families with a greater genetic loading for the disease. The heterogeneity in the BD population gives rise to the question of whether there is a subpopulation of bipolar patients, with more prominent psychosis, that may more closely resemble SZ patients with regard to cognitive deficits. Consequently, the search for neurocognitive intermediate phenotypes for psychosis, and subtypes thereof, may be an important future direction.

3.2. Electrophysiological intermediate phenotypes

3.2.1. Prepulse inhibition

3.2.1.1. Prepulse inhibition (PPI) in schizophrenia.

PPI measures inhibition of the startle reflex. A weak prepulse stimulus normally reduces the magnitude of the startle response to a paired second strong stimulus. PPI is used as an index of sensorimotor gating. PPI deficits are widely reported in SZ (Bolino et al., 1994; Braff et al., 1978, 1992, 2001, 2005; Cadenhead et al., 2000b; Grillon et al., 1992; Karper et al., 1996; Kumari et al., 2000, 2005a; Meincke et al., 2004; Perry and Braff, 1996; Takahashi et al., 2008), in schizotypal personality disorder (Cadenhead et al., 1993) and in biological relatives of SZ probands, both clinically affected and unaffected (Braff et al., 2001; Cadenhead et al., 2000b; Kumari et al., 2005a). Neither gender, nor family psychiatric history affects PPI (Cadenhead et al., 2000b), however higher schizotypal ratings correlate with more severe PPI deficits (Kumari et al., 2005a). While not consistently replicated, (Wynn et al., 2004) found impaired prepulse facilitation (PPF) in the face of normal PPI in both SZ probands and their siblings. Differential sensitivity to prepulse stimuli does not contribute to PPI abnormalities in the illness (Swerdlow et al., 2006). PPI deficits are similar across different modalities of stimuli (auditory, tactile, electrocutaneous), as well as in application of continuous or discrete prepulses (Braff et al., 2001). In contrast to multiple reports documenting PPI deficits, there are several studies which do not replicate PPI findings in SZ (Dawson et al., 1993; Ford et al., 1999b; Miller et al., 1993; Wynn et al., 2004).

3.2.1.2. Prepulse inhibition in bipolar disorder.

There are limited data on PPI abnormalities in BD and the results are controversial. In the few studies conducted on BD samples, PPI abnormalities show state dependent characteristics with “SZ-like” PPI deficits in acute mania and normal PPI during euthymic phases. Perry et al. (2001) found significantly reduced PPI and less startle habituation in patients with acute psychotic mania. They also reported no significant differences between patients with acute psychotic mania and acute SZ psychosis at any of the prepulse conditions. An inverse correlation between PPI and the abnormal thought content item on the Young Mania Rating Scale (YMRS) (Young et al., 1978) obtained in this study suggests a link between reduced PPI and psychotic thought disturbance, independent of formal diagnosis. In contrast, Rich et al. (2005) did not confirm PPI deficit in medicated, euthymic and non-psychotic children with BD, testing the hypothesis that PPI deficits exist independent of clinical state. They also reported no correlation between PPI and mood symptoms and no effect of co-morbid attention deficit hyperactivity disorder on PPI in young BD patients. Barrett et al. (2005) also suggested a state dependent component of PPI deficits in BD. In this study no abnormalities in startle amplitude, onset and peak response latency, as well as no PPI difference were found in adult euthymic BD patients compared to healthy controls. Gender and smoking showed no effect on PPI in this study, however earlier age of onset had a detrimental effect on PPI (Barrett et al., 2005). Similarly, in a recent report by Carroll et al. (2007), PPI did not significantly differ across subjects with manic, mixed stages of BD and control subjects, and the presence of psychosis in the patient sample was not significantly related to PPI levels. However, a recent study by Giakoumaki et al. (2007) reported lower PPI in euthymic BD probands and their unaffected siblings. This finding, although in need of replication, suggests that PPI disruption may represent a trait deficit in BD associated with genetic predisposition.

3.2.1.3. Heritability and genetic correlates of PPI.

A significant heritability component to PPI was reported in the study conducted with 142 young healthy female twins (Anokhin et al., 2003). This study suggested that over 50% of PPI variance may be attributed to genetic factors. A recent report has shown association between a missense mutation on rs3924999 of the NRG1 gene and PPI in SZ and healthy individuals (Hong et al., 2008c). The homozygous for the minor allele A/A carriers showed lowest PPI, A/G carriers had intermediate PPI and homozygous major alleles G/G carriers showed highest PPI. Neither single nucleotide polymorphism was associated with SZ diagnosis in this study. COMT Val(158)Met polymorphism has been also reported to influences PPI in both healthy individuals (Roussos et al., 2008) and SZ patients (Quednow et al., 2008). Patients carrying the Val(158)Met Met/Met allele showed elevated PPI levels (Quednow et al., 2008). These results, although preliminary, suggest that PPI may be regulated by dopaminergic neurotransmission in the prefrontal cortex. A recent report showed association between PRODH haplotypes and attenuated PPI in healthy males (Roussos et al., 2009). In particular, CGA carriers exhibited attenuated PPI and verbal memory, as well as higher anxiety and schizotypy scores. These reports, although preliminary, suggest that PPI is a polygenic trait and support an importance of utilizing the intermediate phenotypes in genetic studies.

3.2.1.4. Clinical correlates of PPI deficits.

Based on the classic criteria for intermediate phenotypes, an optimal candidate phenotype is a stable trait-like characteristic present in clinically symptomatic and stable phases of illness and even in prodromal period. In the majority of reviewed studies SZ probands were acutely symptomatic at the time of PPI recording, however at least a few studies reported PPI deficits in non-psychotic individuals with schizotypal personality disorder (Cadenhead et al., 1993) and in unaffected relatives of SZ probands (Braff et al., 2001; Cadenhead et al., 2000b; Kumari et al., 2005a). These data indicate that the severity of psychosis is not a critical factor in manifestation of PPI deficits in SZ. In contrast, a longitudinal study, conducted by Meincke et al. (2004), suggests some state-dependent characteristics of PPI in SZ. In this study PPI deficits were observed only in acutely psychotic SZ patients, whereas PPI in the clinically stable SZ group was not different from that in healthy controls. A recent report by Quednow et al. (2008) showed that although PPI disruption is already present in a prodromal state of SZ, startle reactivity deficits seem to emerge with the onset of acute psychosis. Several studies reported associations between PPI deficits and formal thought disorder (Meincke et al., 2004; Perry et al., 1999; Perry and Braff, 1994), greater distractibility (Karper et al., 1996) and bizarre behavior (Meincke et al., 2004) in SZ. In BD, no significant correlations were found between PPI and measures of symptom and disease severity (Giakoumaki et al., 2007). In a recent review by Swerdlow et al. (2008) it was re-emphasized that although PPI is a valuable intermediate phenotype for genetic studies, it does not predict clinical course, specific symptoms, or individual medication responses.

3.2.1.5. Effects of antipsychotic and other medications on PPI.

The majority of psychotropic medications, including clonidine, ethanol, diazepam, and caffeine have no effect on PPI in humans (Braff et al., 2001). Similarly, treatment with antipsychotic drugs showed no or very little effect on PPI deficits in SZ. Cadenhead et al. (2000b) reported no differences in PPI performance between SZ probands and their biological relatives both on and off medication. A meta-analysis of 12 studies suggested that antipsychotic medications do not normalize PPI in SZ probands, although may temporarily improve PPI deficits to a certain extent (Hamm et al., 2001). All antipsychotics fail to have an effect on PPI performance in SZ with acute and/or chronic psychosis (Hamm et al., 2001), as well as in drug-naïve first-episode SZ patients (Mackeprang et al., 2002). However, a recent study has shown a significant main effect of nicotine on PPI in that nicotine transiently improved PPI in SZ patients (Hong et al., 2008b). Interestingly, improvement in PPI in response to nicotine significantly correlated with the baseline severity of clinical symptoms, therefore SZ patients with more severe clinical manifestations may benefit more from nicotinic effect on PPI. In BD patients, a recent report by Giakoumaki et al. (2007) showed no significant correlations between PPI and effects of medication. However in a study by Barrett et al. (2005) there was a trend for serum lithium and sodium valproate levels to correlate negatively with PPI.

3.2.1.6. Association between PPI and other intermediate phenotypes.

In a study by Kumari et al. (2005b), no correlation developed between PPI and antisaccade deficits in SZ probands. Based on this report, PPI and antisaccade eye movements are speculated to represent unique intermediate phenotypes, reflecting the functions of distinct neuronal pathways. Later report revealed significant positive correlations between PPI and grey matter volume in the dorso-lateral prefrontal, middle frontal and the orbital/medial prefrontal cortices in stable male outpatients with SZ (Kumari et al., 2008). This finding suggests that compromised neural resources in the frontal cortex contribute to reduced PPI in SZ.

3.2.1.7. Commentary: PPI in schizophrenia/bipolar disorder psychosis.

Sensorimotor gating, measured by PPI, appears to be a stable, trait-like neurophysiologic characteristic of SZ which is not influenced by severity of psychosis, type of symptoms, or antipsychotic medications. Findings of similar PPI deficits in first-degree relatives of SZ probands and persons with schizophrenia spectrum personality disorders suggest a genetic component to the pathophysiology of PPI. Recent studies reported a strong heritable component to PPI, confirmed by a finding of association between PPI and NRG1, COMT, and PRODH in SZ and healthy individuals. These data, although still limited, support PPI as a promising intermediate phenotype for future genetic studies of psychosis. Reports on PPI in BD are currently very preliminary. A few recent reports have shown “SZ-like” sensorimotor deficits in acute psychotic mania, but not in non-psychotic euthymic individuals. In addition, lower PPI was observed in unaffected siblings of BD probands, based on a single report. These data suggest that PPI alterations are biologically linked to psychosis and may be one of the neurophysiological markers of psychosis liability. However, this interpretation should be taken with caution given significant limitations in design and power of currently available studies in BD, e.g., limited sample size, mixed psychotic and non-psychotic probands groups, lack of longitudinal PPI studies in all phases of the illness, and very limited data in biological relatives of BD probands. There is a compelling need for informative studies of PPI focusing on BD subgroups, especially in individuals with and without psychosis.

3.2.2. P50 event-related potentials (P50 ERP)

3.2.2.1. P50 abnormalities in schizophrenia.

In the standard paired-click paradigm developed by Freedman et al. (1996), P50 amplitude (the largest EEG wave within 50 ms after the stimuli administration) is measured in response to each of two auditory clicks. In healthy individuals, the P50 wave following the second stimulus is inhibited, reflecting a normal sensory gating phenomenon. In contrast, SZ patients and their relatives show disrupted sensory gating, as reflected in less suppression of P50 (Baker et al., 1987; Clementz, 1998; Freedman et al., 1996; Jin et al., 1997; Louchart-de la Chapelle et al., 2005a; Siegel et al., 1984; Wegrzyn, 2004; Wegrzyn and Wciorka, 2004). Myles-Worsley (2002) reported a P50 suppression deficit in a proportion of SZ probands (64.7%) and their biological relatives (51.8%). In an early study of Waldo et al. (1995) unaffected parents of SZ probands showed P50 abnormalities similar to their disordered offspring. In this study P50 deficit correlated with the density of psychotic illnesses in families. “SZ-like” P50 suppression deficits have been also reported in subjects with SZ spectrum personality disorders (Cadenhead et al., 2000a), as well as in individuals at high risk for developing SZ (the subjects with several psychotic family members and individuals with prodromal symptoms) (Cadenhead et al., 2005; Myles-Worsley et al., 2004). In opposite, Arnfred et al. (2003) did not observe any P50 abnormalities in unmedicated males with SZ. Even though, in a recent study they reported an attenuated difference between paired P50 responses in patients with SZ and SZ spectrum personality disorders due to lower P50 amplitude following the first acoustic stimulus (Arnfred, 2006). Based on this observation, the authors suggested that the abnormality in early cortical responses may contribute more significantly to P50 distortion, than an abnormality in sensory gating per se.

A growing number of reports suggest a spectral analysis of P50 as a more accurate measure of sensory gating and information processing (Clementz and Blumenfeld, 2001; Johannesen et al., 2005). Two main frequency bands have been identified: gamma band (high frequency) response (GBR; 20–50 Hz) and low-frequency response (LFR; 1–20 Hz). The gamma band is thought to represent cortical activation during the early phases of cortical processing, related to the integration of information (Lee et al., 2003). The LFR may reflect late hippocampal activation and is thought to be associated with new information encoding and working memory functions (Jensen and Tesche, 2002; Klimesch, 1999). Along with these recent reports, several previous studies linked deficient P50 inhibition to hippocampal dysfunction (Freedman et al., 1996; Hsieh et al., 2004; McCarley et al., 2008; Thoma et al., 2008). Interestingly, the LFR was uniquely abnormal in volunteers with disorganized and undifferentiated types of SZ and correlated with higher PANSS ratings on conceptual disorganization, mannerism and posturing, and poor attention; while patients with paranoid SZ showed no impairment in either frequency domain (Johannesen et al., 2005). Hong et al. (2004) examined different frequency components, evoked by stimuli in the P50 paired-click paradigm, such as the early gamma frequency oscillation, followed by beta frequency oscillation. The authors suggest that the gamma-to-beta shift in response to the first stimulus (S1) may contain critical electrophysiological signals that modulate the S2 inhibition. In particular, post-S1 beta frequency response inversely correlated with S2 P50 response in SZ patients, but not in the normal comparison group. The later report by Hong et al. (2008a) showed that SZ probands and their first-degree relatives significantly differ from healthy controls in gating of the theta–alpha-band responses. In addition, the heritability of theta–alpha-band gating was estimated to be at least 4-fold higher than the P50 heritability estimate. This suggest that gating of the theta–alpha-frequency oscillatory signal in the paired-click paradigm is more strongly associated with SZ and may be better suited for genetic studies of the gating deficit in psychosis.

3.2.2.2. P50 abnormalities in bipolar disorder.

Diminished suppression of P50 auditory stimuli has been also reported in BD, although the data are less definitive. Several studies showed P50 sensory gating deficit in acute mania, similar to the gating deficit in SZ (Franks et al., 1983). Baker et al. (1987) observed variable P50 C/T ratios in acutely manic patients with BD, from low values (indicating complete suppression of the test response) to high values (similar to those in SZ). Additionally, P50 values correlated with severity of manic symptoms in this study, suggesting that at least some characteristics of P50 inhibition deficits in BD might be state-dependent. Two earlier studies suggested that P50 gating deficit in acutely manic patients is mediated by noradrenergic mechanisms (Adler et al., 1990; Baker et al., 1990). A few studies have looked at P50 differences in BD subgroups with and without history of psychosis. Carroll et al. (2008) reported that BD subjects without history of psychosis exhibited reduced S1 response magnitudes for the conventional P50 peak-picking and low-frequency response analyses, which may reflect a diminished capacity to selectively attend to salient stimuli as opposed to impairments of inhibitory sensory processes. Olincy and Martin (2005) reported P50 suppression abnormalities in BD patients with a lifetime history of psychosis and schizoaffective disorder, bipolar type, whereas BD subjects with no history of psychosis exhibited normal P50 suppression. A recent study confirmed deficient P50 gating in BD patients with a lifetime history of psychosis (Sanchez-Morla et al., 2008). These data, although preliminary, suggest that P50 may be a specific intermediate phenotype related to psychosis in SZ–BD boundary. Further, pilot family studies in psychotic BD support the notion that P50 intermediate phenotype may reflect the impact of susceptibility genes across psychosis. Schulze et al. (2007) reported diminished P50 suppression in probands with psychotic BD and their unaffected relatives. This finding has been recently confirmed in a large twin and family sample of psychotic BD (Hall et al., 2008).

3.2.2.3. Heritability and genetic correlates of P50.

A heritable component of P50 is suggested by several twin studies (Myles-Worsley et al., 1996; Young et al., 1996). Disrupted P50 gating is associated with a dinucleotide polymorphism at chromosome 15q13–14 (alpha 7 nicotinic receptor locus) in both SZ probands and their biological relatives (Hall et al., 2008), as well as in healthy individuals (Leonard et al., 2002). Although P50 and eye tracking intermediate phenotypes are thought to be highly independent, a linkage of P50 and antisaccade measures to the D22s315 marker on chromosome 22q was found in eight Utah families with multiple cases of SZ (Myles-Worsley et al., 1999).

3.2.2.4. Clinical and cognitive correlates of P50.

The majority of studies report no effect of clinical symptoms on P50 (Clementz et al., 1997; Clementz, 1998; Myles-Worsley, 2002; Thoma et al., 2003). However, several studies have shown an association between disrupted P50 sensory gating and negative symptoms of SZ (Arnfred, 2006; Louchart-de la Chapelle et al., 2005a; Ringel et al., 2004). A single study reported more prominent P50 abnormalities in patients with disorganized SZ (Ringel et al., 2004). A recent report by Brockhaus-Dumke et al. (2008) suggested progressive worsening of P50 in chronic stages of SZ. Several measures of cognitive function were linked to P50 abnormalities in SZ, such as poor attention (Arnfred, 2006) and implicit color learning on Wechsler Memory Scale (Hsieh et al., 2004).

3.2.2.5. Effects of medications on P50.

The majority of studies found no influence of psychotropic agents on P50 amplitude, latency, and sensory gating ratio (Myles-Worsley, 2002; Ringel et al., 2004). No P50 differences were observed in patients with SZ, medicated either with typical or atypical antipsychotics (Arango et al., 2003; Hong et al., 2009). Uniquely, clozapine appeared to improve P50 sensory gating in person with SZ, based on a single report (Adler et al., 2004). In BD, no significant effect of atypical antypsychotics, antidepressants, or mood stabilizers on P50 has been reported (Olincy and Martin, 2005).

3.2.2.6. Association between P50 and other intermediate phenotypes.

The few studies evaluating the concordance rate among putative neurophysiological phenotypes of psychosis failed to find any associations between P50, eye tracking measures and PPI. Two recent studies showed no correlation between P50 and antisaccadic eye movement measures (Louchart-de la Chapelle et al., 2005b; Price et al., 2006). Similarly, Schwarzkopf et al. (1993) did not find any meaningful correlations between P50 and PPI in healthy individuals, a finding confirmed in two additional studies (Brenner et al., 2004; Light and Braff, 2001). This suggests that these neurophysiological deficits likely represent distinct intermediate phenotypes with unique underlying neurobiological mechanisms.

3.2.2.7. Commentary: P50 in schizophrenia/bipolar disorder psychosis.

The P50 sensory gating deficit is a relatively well established neurophysiologic intermediate phenotype of SZ, consistently observed in SZ schizophrenic probands, their biological relatives, and subjects with schizophrenia spectrum personality features. Classic, as well as more recent studies have shown that the P50 deficits are primary and manifest prior to development of clinical symptoms in SZ. Several studies linked the LFR component of P50 to functional abnormalities in MTL and hippocampus, the brain area well associated with SZ in cognitive, imaging, and postmortem tissue studies. The majority of studies have reported no effect of psychosis severity or psychotropic medications on P50. However, there is a report of clozapine uniquely improving disrupted P50 gating in SZ, a finding which needs to be validated. A few genetic loci, such as alpha 7 nicotinic receptor locus in 15q13–14 and D22s315 locus in 22q, have been linked to P50 in SZ probands, their biological relatives and in healthy individuals. Studies of P50 in BD are few to date and less definitive. Most studies have been conducted almost exclusively in acutely manic patients with BD. Nevertheless, they have consistently reported P50 deficits similar to those in SZ probands. Recent studies has attempted to compare P50 characteristics in BD probands with and without history of psychosis, finding P50 deficits in psychotic BD similar to SZ probands, but normal P50 in non-psychotic BD. This suggests that the P50 deficit may be a unique biological marker of psychosis shared between SZ and BD, although future studies are needed to confirm this finding. Pilot family studies reported characteristic P50 deficits in biological relatives of subjects with psychotic BD, suggesting that P50 intermediate phenotype may reflect the impact of susceptibility genes. However, no studies are available to date on specific genetic correlates of P50 in BD. Large scale studies, comparing SZ and BD families, are needed to fully explore heritability of P50 and its potential as a marker of psychosis liability.

3.2.3. P300 event-related potentials (P300 ERP)

3.2.3.1. P300 abnormalities in schizophrenia.

P300 is a positive event-related potential EEG wave, occurring 300 ms after infrequent or unexpected sensory stimulus (an “odd ball” paradigm). P300 ERP was first reported as abnormal, both in latency and in amplitude, in SZ by Roth et al. (1981). These findings were subsequently confirmed by several other laboratories in SZ patients (Faux et al., 1993; Ford et al., 1999a,b, 2000; Morstyn et al., 1983; Strik et al., 1994; Turetsky et al., 1998), first-episode SZ (de Wilde et al., 2008; Salisbury et al., 1998), SZ prodrom (Frommann et al., 2008) and SZ spectrum personality disorders (Salisbury et al., 1996). Several groups showed alterations in P300 in clinically affected and unaffected siblings of SZ probands (Bharath et al., 2000; Blackwood et al., 1991; Condray et al., 1992; Kidogami et al., 1991; Price et al., 2006; Roxborough et al., 1993; Saitoh et al., 1984). A recent family study has shown normal P300 amplitude but significant latency delays in non-psychotic relatives of SZ probands (Bramon et al., 2005). Two recent meta-analyses of P300 (Bramon et al., 2004, 2005) confirmed that probands with SZ and their biological relatives have significantly reduced P300 amplitudes (pooled effect size = 0.85 (p < 0.001) and 0.61 (p < 0.001), respectively. Of note, cross-study variations in filters, task difficulty, antipsychotic medications and duration of illness did not influence P300 amplitudes and latencies in this meta-analysis (Bramon et al., 2004). In contrast, several studies did not replicate the P300 abnormalities in SZ (Ford et al., 1994; Iwanami et al., 2002; Pfefferbaum et al., 1989, 1991) or in their biological relatives (de Wilde et al., 2008; Steinhauer et al., 1991).

3.2.3.2. P300 abnormalities in bipolar disorder.

The P300 intermediate phenotype is better established in BD compared to other neurophysiologic measures. Studies in BD consistently report prolonged P300 latency in BD probands (Lenox et al., 2002; Muir et al., 1991; Schulze et al., 2008; Souza et al., 1995; Strik et al., 1998), but the data on P300 amplitude vary. The majority of studies report decreased P300 amplitudes in BD, similar to SZ (Hall et al., 2007; Muir et al., 1991; O’Donnell et al., 2004; Salisbury et al., 1999; Vilela et al., 1999), while fewer studies show normal P300 amplitudes in BD (Lahera et al., 2009; Schulze et al., 2008; Souza et al., 1995; Strik et al., 1998). Despite the similarity of P300 abnormalities in SZ and BD patients, they each demonstrate a unique P300 topography with maximally reduced P300 amplitudes in the posterior temporal lobe in SZ subjects in contrast to a frontal lobe distribution in subjects with acute mania (Salisbury et al., 1998). Two studies conducted with biological first-degree relatives of BD probands led to contradictory results: while Pierson et al. (2000) reported prolonged P300 latency and decreased P300 amplitudes; more recent study (Schulze et al., 2008) found significantly delayed P300 latency but normal P300 amplitude in biological relatives of BD probands.

3.2.3.3. Heritability and genetic correlates of P300.

Several studies, conducted on healthy twin pairs (Anokhin et al., 2001; Hall et al., 2006), suggested moderate heritability estimates for P300 amplitude. Interestingly, the P300 heritability in females appeared to be lower (45% compared to 56% in males) (Anokhin et al., 2001), suggesting that environmental factors may have a stronger influence on the P300 phenotype in females. Tsai et al. (2003) linked diminished P300 latencies in healthy females to Met/Met COMT genotype, while Gallinat et al. (2003) found lower frontal P300 amplitudes in Met homozygous SZ patients. A recent preliminary report suggested association between NRG 1 polymorphism and P300 latency in psychotic probands and their non-psychotic relatives (Bramon et al., 2008).

3.2.3.4. Clinical correlates of P300 deficits.

Diminished P300 amplitude and increased latency have been related to the presence of negative symptoms (Eikmeier et al., 1992; Ford et al., 1999a; Mathalon et al., 2000; McCarley et al., 1991; McConaghy et al., 1993; Turetsky et al., 1998) and, to a lesser extent, to positive symptoms (Egan et al., 1994; McCarley et al., 1993; O’Donnell et al., 1993), illness duration and early age of onset (Iwanami et al., 2002; Mathalon et al., 2000). Prolonged P300 latency was linked to deficits in attention and working memory, while decreased P300 amplitude was taken as suggestive of parieto-temporal dysfunction, affecting audition and language (O’Donnell et al., 2004). However, two studies did not find any correlations between P300 and neurocognitive characteristics in patients with BD (Souza et al., 1995; Vilela et al., 1999).

3.2.3.5. Effects of medications on P300.

Existing data on medication effects on P300 are contradictory. No effect of medication on P300 amplitudes was found in an early study (Roth et al., 1981); this was confirmed in a comprehensive meta-analysis of P300 in SZ patients (Bramon et al., 2004). In contrast, at least two later studies reported an effect of antipsychotic drugs on P300. Iwanami et al. (2002) observed lower P300 amplitudes and significantly delayed P300 latency in SZ patients treated with high doses of antipsychotics. Wang et al. (2005) demonstrated improved P300 amplitude in first-episode patients after treatment with haloperidol and bromperidol, especially in patients with shorter period of psychosis prior to the treatment. In BD, auditory and visual P300 evoked potentials were not a good predictive factor for response to valproate or lithium (Reeves and Struve, 2005).

3.2.3.6. Correlations between P300 and different intermediate phenotypes.

Price et al. (2006) attempted to identify associations between P300 and other neurophysiological intermediate phenotypes (P50, antisaccadic eye movements, and mismatch negativity) in a cohort of SZ families. While no correlation between the phenotypes was found, a multivariate intermediate phenotype, including a weighted combination of the individual phenotypes in the logistic regression model, provided greater diagnostic classification power than any single intermediate phenotype.

3.2.3.7. Commentary: P300 in schizophrenia/bipolar disorder psychosis.

P300 is a promising intermediate phenotype for SZ. Specifically, prolonged P300 latency and decreased P300 amplitude have been consistently found in first-break and chronic SZ patients, their biological relatives and individuals with SZ spectrum personality disorders. Some studies suggest associations between P300 and clinical and cognitive characteristics of SZ, mainly, negative symptoms and deficits in attention and working memory, although neuronal circuits underlying the P300 intermediate phenotype remain unknown. Likewise, specific genes involved in P300 heritability are unidentified to date, although a few preliminary reports linked P300 to COMT and NRG1 variants. Although, there are fewer studies in BD, alterations in P300 amplitude and latency in BD probands appear to parallel those in patients with SZ suggesting underlying biological similarity between these two psychosis variants. Moreover, unaffected biological relatives of BD probands with psychosis show a delayed P300 latency. These observations, although preliminary, give support to delayed P300 latency as an intermediate phenotype linked to psychosis independent of diagnostic categories.

3.3. Eye movement intermediate phenotypes

3.3.1. Eye tracking abnormalities in schizophrenia

Abnormalities in smooth pursuit eye movement (SPEM) performance have been well documented in SZ and their first-degree relatives and mark genetic liability for the disease (see (Levy et al., 1994; Thaker, 2000) for a review). Studies have also observed abnormalities in saccadic inhibition and oculo-motor delayed response tasks (which assess spatial working memory) in SZ probands and their relatives (Clementz et al., 1994; Crawford et al., 1998; Curtis et al., 2001; Katsanis et al., 1997; McDowell et al., 1999; Thaker et al., 2000). A study by Boudet et al. (2001) did not find significant differences in SPEM performance between patients and matched controls but found worse performance in the parents of SZ probands compared with controls.

3.3.2. Eye tracking abnormalities in bipolar disorder

In contrast to the depth of data in SZ, SPEM has not been extensively studied in BD. Early studies note abnormalities in BD patients but also noted that abnormal findings could be secondary to lithium (Flechtner et al., 1992; Gooding et al., 1993; Holzman et al., 1991; Levy et al., 1985). Subsequently, SPEM abnormalities were observed in patients with affective disorders even when these patients were in relative remission and not on treatment. Rosenberg et al. (1997) demonstrated that relatives of BD patients show SPEM abnormality, as measured by closed-loop gain, similar to the relatives of SZ probands in the NY High-Risk sample. Kathmann et al. (2003) also more recently demonstrated reduced gain in relatives of both SZ and affective patients. Studies in BD probands, and particularly their families, are few. However, Sweeney et al. (1998) demonstrated deficits in affective disorders that parallel those seen in SZ. There is a need for further investigation in this area.

3.3.3. Heritability of eye tracking characteristics

Both smooth pursuit and saccade generation during pursuit are under at least partial genetic control, accounting for 40–60% of the variance in a sample of monozygotic and dizygotic twins (Katsanis et al., 2000). In a more recent study comparing the heritability of predictive pursuit measure with a traditional closed-loop gain measure the physiologically specific predictive measures were associated with better heritability estimates than traditional ones (Hong et al., 2006). In addition, linkage findings to locus on chromosome 6p21 in SZ families have been reported in two independent samples (Arolt et al., 1999; Matthysse et al., 2004). While data support predictive pursuit measures as a good measure of heritability, a meta-analysis by Levy et al. (1994) indicated that data did not support the antisaccade task as useful in identifying genetic vulnerability in unaffected relatives of SZ patients.

3.3.4. Effects of medications on eye tracking

Friedman et al. (1992) found that treatment with and length of time on clozapine worsened overall oculo-motor performance in 13 SZ and schizoaffective patient. Sweeney et al. (1994) reported that patients treated with antipsychotic medications had similar but generally more severe deficits than medication-naïve patients in eye tracking and that short-term treatment with neuroleptics improved attention-related features of performance such as anticipatory saccades. Hutton et al. (2001) showed that velocity gain is impaired early in the course of SZ, even in medication-naïve patients, and that oculo-motor performance negatively impacts illness chronicity. Several additional studies suggest that increased antisaccade error rates are not drug artifacts (Crawford et al., 1995; Hutton et al., 1998a, 2001). Nicotine improves SPEM performance in SZ. Administration of nicotine shows increased smooth pursuit gain and decreased antisaccade errors in all volunteers and suggests improved attention as the mediating factor (Depatie et al., 2002). Independent groups found nicotine improved performance in SZ: Sherr et al. (2002) found increased eye acceleration in patients but not controls; Olincy et al. (2003) found improved smooth pursuit in both patients and controls. However neither study found an effect of nicotine on saccadic tasks. Tanabe et al. (2006) also showed an improvement of SPEM with nicotine. SZ volunteers with the anticholinergic drug procyclidine showed a decreased SPEM, as indicated by reduced gain and increased anticipatory saccades during pursuit (Ettinger et al., 2003). Low does of the NMDA receptor agonist ketamine induces SZ-like errors in oculo-motor performance in healthy controls (Avila et al., 2002), consistent with the theory that decreased cortico-limbic NMDA receptor function may be associated with SZ.

3.3.5. Association between eye tracking and other intermediate phenotypes

Rosse et al. (1993) found a positive correlation between antisaccade errors and perseverative errors on the Wisconsin Card Sorting Test (WCST). A study by Tien et al. (1996) found the same pattern of deficits in SZ, as well as in BD patients. Schulze et al. (2006) did not find a correlation between eye movement abnormalities and brain morphology abnormalities in a sample of 70 patients with SZ and schizoaffective disorder.

3.3.6. Commentary: eye tracking in schizophrenia/bipolar disorder psychosis

Similar oculo-motor abnormalities (impaired smooth pursuit eye movement and saccadic disinhibition) have been reported in probands with SZ and psychotic BD. More significantly, recent studies demonstrated that biological relatives of BD probands show SPEM alterations similar to the relatives of SZ probands. These data support the proposal that the eye tracking intermediate phenotypes may overlap across the two psychotic disorders, and may rely on shared psychosis liability genes. Specifically, independent samples linked SPEM abnormalities to chromosome 6p21 in SZ families, although no data available yet on specific candidate genes in BD.

3.4. Structural imaging intermediate phenotypes

Within the diversity of structural brain characteristics (e.g., global vs. regional, tissue type), there are specific measures, empirically derived, which fit the description of an intermediate phenotype and can potentially be useful in genetic studies (Keshavan et al., 2007). The most replicated structural measures in psychosis are whole brain volume with grey matter, white matter, and ventricular space quantified separately. Parcelation of the brain into anatomic sub-regions allows the independent quantification of regions like prefrontal cortex (PFC), anterior cingulate cortex (ACC), medial temporal lobe (MTL) structures, basal ganglia, thalamus and other relevant regions. Regions can be assessed using region-of-interest (ROI) methodologies or voxel-based morphometry (VBM), the latter being an unbiased process for comparing group structural analyses. Structural imaging alterations have been associated with both SZ and BD psychoses, with some overlap as review below.

3.4.1. Structural abnormalities in schizophrenia

Individuals with SZ have reductions in cerebral volume in some but not all brain regions (Keshavan et al., 2007). Whole brain volume, particularly grey matter, is decreased and ventricular volume is increased (Steen et al., 2006; Ward et al., 1996; Woodruff et al., 1995; Wright et al., 2000) in persons with SZ. Regionally, the PFC shows a reduction in grey matter density, as do areas of the MTL, particularly the hippocampal formation and superior temporal gyrus (STG) (Honea et al., 2005; Lawrie and Abukmeil, 1998; McCarley et al., 2008; Nelson et al., 1998; Pearlson et al., 1997). Orbital frontal cortex and thalamus have also been implicated in the volume reduction (Buchanan et al., 1998; Konick and Friedman, 2001) in SZ probands. Some analyses suggest that disproportionate volume loss occurs in hetero-modal association cortex (Schlaepfer et al., 1994). Studies with VBM tend to confirm the findings from the initial region of interest ROI analyses (Honea et al., 2005). The question of whether volume loss is progressive or not is less clearly answered, with some studies finding no change over time (Steen et al., 2006; Whitworth et al., 2005), while others finding progressive volume loss (DeLisi et al., 2004; Ho et al., 2003; Hulshoff Pol et al., 2002). Kasai et al. (2003) reported that while his population of individuals with SZ showed progressive loss of grey matter in the STG, subjects with affective psychosis did not. Moreover, the data showing similar volumetric changes in first-break SZ individuals, who are medication-naïve, suggest that these structural changes are, at least partially, independent of antipsychotic medication treatment (Keshavan et al., 2005; Kuroki et al., 2006; Steen et al., 2006). Unaffected family members of persons with SZ, nonetheless, show alterations in brain structural measures, albeit milder than the SZ persons themselves. MTL volumes are reduced (Boos et al., 2007; Keshavan et al., 1997; Lawrie et al., 1999), PFC grey matter volume is smaller (Diwadkar et al., 2006; Job et al., 2003), and thalamus volume is reduced (McDonald et al., 2005; McIntosh et al., 2005), the latter possibly also related to psychotic BD. Cortical thickness and surface area is affected in SZ families, with reduction in ACC thickness as well as lower surface area in the right ACC and in the STG (Goghari et al., 2007).

3.4.2. Structural abnormalities in bipolar disorder

The literature describing imaging alterations in BD, especially its psychotic variant, is more modest than that in SZ. Nonetheless, observations have been made. While a reduction in whole brain volume or grey matter volume is not generally reported in BD (Altshuler et al., 1991; McDonald et al., 2004; Pearlson et al., 1997; Pearlson, 1999; Strakowski et al., 2000; Zipursky et al., 1997), volume increases in amygdala may occur (Altshuler et al., 1998; McDonald et al., 2005). Early literature reported increases in the lateral ventricular size, changes in the MTL volume and subcortical white matter hyperintensities in BD (Altshuler et al., 1998; Dupont et al., 1995; Swayze et al., 1990). Reports differ on whether individuals with BD have smaller or larger MTL volumes (Altshuler et al., 1991; Pearlson, 1999; Swayze et al., 1990, 1992). Subsequent data suggest that the various subgroups within BD may have different structural brain characteristics, depending on whether or not they have psychosis as an overlapping feature with SZ (Potash, 2006; Strasser et al., 2005); psychotic BD shows alterations in brain volume that tend to parallel those found in SZ. Similarly, Salokangas et al. (2002) showed that psychotic but not non-psychotic depressed individuals showed an increase in ventricular volume. Adolescents with BD show alterations in brain structure not dissimilar to those reported for bipolar adults (Friedman et al., 1999; Pearlson, 1999). Some studies, comparing individuals across the psychosis syndrome, find no differences from normal in BD grey matter, despite seeing the usual changes described above in SZ, while white matter deficits in BD were evident throughout the neocortex and brain stem, with similar regions of white matter loss in the SZ and BD groups (Harvey et al., 1994; McDonald et al., 2005). Studies which define the clinical correlates of structural abnormalities in psychotic BD have not yet been published. Moreover, the potential effects of mood stabilizers and antidepressants on volumes are virtually unstudied. Finally, the robust association of structural alterations with other intermediate phenotypes in psychotic BD has not been attempted; even the identification of these robust phenotypes has lagged behind that seen in SZ.

3.4.3. Heritability of the structural alterations

Two factors, genetic predisposition and environmental events, influence brain volume measures. Evidence for heritability of structural imaging traits is greatest for intracranial volume (Pfefferbaum et al., 2000) and total brain volume (Bartley et al., 1997), suggesting these measures for use as volumetric intermediate phenotypes; while the lowest heritability characterizes ventricular volume (Baare et al., 2001a,b; Rijsdijk et al., 2005; Sullivan et al., 2001), suggesting its vulnerability to environmental factors. With respect to volumes of anatomic sub-regions, some ROIs have highly heritable volumes, including the ACC, the MTL, the STG, and cerebellum (Wright et al., 2000) whereas the asymmetries in total hemispheric volume give no evidence of being heritable. The size and shape of the corpus callosum is highly heritable (Oppenheim et al., 1989; Pfefferbaum et al., 2000). Hippocampal size appears to have rather low heritability as does ventricular size, indicating a stronger environmental influence and plasticity on these structures. Reduction in the volume of the dorso-lateral PFC is related in a dose sensitive way to the genetic propensity for SZ (Cannon et al., 2000). Moreover, Baare et al. (2001a) found a lower whole brain volume in the well monozygous twin of a group of SZ probands, whereas the normal controls, had a higher brain volume than either the ill or non-ill twin.

3.4.4. Clinical correlates of the structural brain abnormalities

Volume reductions in the MTL have been associated with cognitive dysfunction; while, STG volume correlates with positive symptoms (Antonova et al., 2004; Lawrie et al., 2004; Pearlson et al., 1997). Severity of delusions has been associated with increased volume in entorhinal cortex (Prasad et al., 2004a) and volume reduction in the MTL and parahippocampal gyrus (Prasad et al., 2004b). Negative symptoms were associated with the orbitofrontal volume increase (Lacerda et al., 2007). Reductions in the MTL also correlate with memory dysfunction in SZ probands (Antonova et al., 2004; Lawrie et al., 2004; Pearlson et al., 1997) and their biological relatives (O’Driscoll et al., 2001; Seidman et al., 2002a). Moreover, in monozygotic twins discordant for SZ, the ill twin shows a greater volume loss in the MTL than the non-ill twin (Cannon et al., 2002; van Erp et al., 2004), but both have reduced volumes from healthy controls. All of these associations are weak and have involved low subject numbers, but indicate potential regions and circuits involved in mediating symptoms of psychosis.

3.4.5. Effects of psychotropic medications on the structural abnormalities

Typical structural alterations have been observed in first-break SZ and in psychosis prodrome individuals, suggesting that neither chronic medication nor illness is entirely responsible for the volume change (Keshavan et al., 2005; Kuroki et al., 2006) Two different meta-analyses of volumetric alterations in SZ probands (Steen et al., 2006; Vita et al., 2006) showed reduced whole brain and MTL volume in individuals at their first-psychotic break. However, several pieces of evidence suggest that antipsychotic medications can have some influence on brain volume. Medication effects have been shown in the basal ganglia. Specifically, first generation antipsychotic drugs are associated with increased basal ganglia volume and greater concentrations of GM in the basal ganglia in SZ samples (Gaser et al., 1999; Kubicki et al., 2002; Wilke et al., 2001). When it was studied in experimental animals, the medication effect was associated with a medication-induced enlargement of the bouton in the neuronal terminal fields and expanded mitochondria, consistent with the increase in metabolic rate produced by antipsychotic medication in basal ganglia (Kung and Roberts, 1999). In another analysis of MTL and PFC brain volume, Lieberman et al. (2005) compared two groups of first-break individuals with SZ, one treated with a first generation medication (haloperidol) while the other was treated with a second generation drug (olanzapine); the first generation medication group showed greater volume reductions in the PFC and the temporal horn than the second generation drug group, suggesting that antipsychotics may have some impact on volume changes and that these may differ across drugs and regions. In BD, Lithium was shown to increase grey matter density in diffuse cortical regions with greatest differences found in bilateral cingulate and para-limbic cortices, brain regions critical for attentional, motivational, and emotional modulation (Bearden et al., 2007).

3.4.6. Commentary: structural brain abnormalities in schizophrenia/bipolar disorder psychosis

There are specific volumetric brain abnormalities which appear to be characteristics of psychosis as a dimension. While grey matter loss in prefrontal, temporal and inferior parietal cortices is characteristic of SZ, no differences in grey matter volume are evident in BD. However, loss of white matter is evident throughout the neocortex and brain stem in SZ and BD with similar regions affected in the two illnesses, although an interpretation of the results for the neurophysiology of psychosis is unclear. Moreover, in BD brain structure intermediate phenotypes vary, depending on the presence or absence of lifetime psychosis. Individuals with psychotic BD show alterations in ventricular and cortical volumes similar to those in SZ. These typical structural alterations are observed in medication-naïve first-break psychotic individuals, in psychosis prodrome, and most importantly, in unaffected relatives of psychotic probands. This suggests a stable trait-like nature and familiarity of the volumetric measures and makes them promising candidates for future genetic research in psychosis.

In the future, it will be possible to base intermediate phenotypes analyses in psychosis on functional as well as structural imaging characteristics. The demonstration of an association between fMRI activation patterns to fearful stimuli and COMT genetic variants (Drabant et al., 2006) suggests that it will be possible to use multiple brain imaging markers as intermediate phenotypes. In the area of psychosis, it would be useful to develop common tasks that will activate regions thought to play an important role in psychosis, like the PFC, MTL, and STG.

3.5. Neurological soft signs as intermediate phenotypes

3.5.1. Neurological soft signs (NSS) in schizophrenia

Subtle neurological abnormalities are consistently found at higher rates in individuals with SZ (Chen et al., 1995; Heinrichs and Buchanan, 1988; Tsuang et al., 1999; Weinberger and Wyatt, 1982). These abnormalities are conventionally understood as maldevelopmental manifestations of the illness, reflecting non-localizing neurological symptoms (Chen et al., 1995; Heinrichs and Buchanan, 1988), and collectively referred to as neurological “soft signs”. NSS include abnormalities in several functions including motor/coordination (e.g., gait, balance abnormalities, disdiadochokinesia); sequencing of complex motor acts (e.g., abnormal “fist-edge-palm” probe and rhythmic tapping); sensory integration (e.g., impaired left-right orientation, audio-visual integration, extinction); and disinhibition (e.g., saccadic eye movements, abnormal “go/no-go” test) (see Chan and Gottesman (2008) for description of the symptoms and assessment instruments). Previous studies have reported NSS prevalence rates ranging from 50% to 73% in SZ patients, compared to 5% in the general population (Bombin et al., 2005; Chan et al., 2009; Heinrichs and Buchanan, 1988). In addition, NSS are seen with increased frequency in unaffected biological relatives of SZ probands (Compton et al., 2007; Ismail et al., 1998; Mechri et al., 2009; Yazici et al., 2002) and in high-risk samples (Erlenmeyer-Kimling et al., 2000; Mittal et al., 2007). These signs exhibit relative stability across the disease stages (Chen et al., 1996), are not secondary to psychotropic medications (Heinrichs and Buchanan, 1988) and do not improve with treatment (Arango et al., 2000; Cox and Ludwig, 1979; Dazzan et al., 2004; Sevincok and Topaloglu, 2006). Classic studies by Tsuang et al. proposed that NSS are expressions of vulnerability for SZ (Tsuang et al., 1991; Tsuang and Faraone, 1999). More recently, the soft signs have been considered as putative intermediate phenotypes of the illness: they are stable trait-like manifestations that are present from early in the illness; seen at a higher frequency in SZ probands and their unaffected relatives; presumably heritable; and can be reliably measured (see Chan and Gottesman (2008) for a comprehensive review).

3.5.2. Neurological soft signs in bipolar disorder

In contrast to extensive literature addressing NSS in SZ, very little is known about this intermediate phenotype in BD. An early report by Nasrallah et al. (1983) acknowledged manifestation of NSS in patients with acute mania. Negash et al. (2004) reported higher prevalence of motor coordination, sequencing of complex motor acts and sensory integration signs in symptomatic and euthymic probands with BD, type I in a large Ethiopian sample, confirmed by similar findings in euthymic BD patients in Asian sample (Goswami et al., 2006). Although specific brain regions mediating NSS in BD remain unknown, the signs documented most frequently in individuals with BD, such as suck reflex, grasp reflex and the glabellar tap, are regarded as evidence of frontal lobe dysfunction (Goswami et al., 2006). Severety of NSS directly correlated with expression of positive and negative symptoms in a sample of new-onset BD (Whitty et al., 2006), although no such associations emerged in other studies (Negash et al., 2004). In addition, poorer executive function and social function directly correlated with severity of NSS, whereas family history of mood disorders had no effect on prevalence of NSS (Goswami et al., 2006). Similarly to SZ, the NSS intermediate phenotype in BD features stable trait-like characteristics manifesting in childhood (Dickstein et al., 2005), persisting at age of first mood or psychotic episode (Whitty et al., 2006), and remaining stable across the longitudinal course of the illness (Goswami et al., 2007). To our knowledge, no reports on NSS in psychotic variant of BD or in biological relatives of BD probands are available.

3.5.3. Heritability and genetic correlates of NSS

Extensive evidence of familial co-segregation of NSS in SZ suggests a role for genetic factors in their etiology. However, the few existing studies of the heritability of NSS in SZ have provided very modest evidence for the heritability (Egan et al., 2001a; Hyde et al., 2007; Sanders et al., 2006), with only audio-visual integration abnormalities found to be heritable in SZ pedigrees (Sanders et al., 2006). Studies of risk genes have linked COMT Val(158)Met polymorphism to motor coordination and complex motor sequencing abnormalities in SZ probands with deficit symptoms, but not in non-deficit SZ probands (Galderisi et al., 2005). In addition, R.Y. Chen et al. (2001) found a trend association between a T(102)C polymorphism of the 5HT2a receptor gene and the motor coordination NSS in a large Chinese sample of individuals with SZ, although more recent meta-analysis that included European and East Asian populations did not support this association (Abdolmaleky et al., 2004). No studies to date have reported heritability estimates and genetic associations of NSS in BD.

3.5.4. Clinical correlates of NSS

NSS in SZ appear to be one of the earliest features of deviant neurodevelopment, consistently documented in children who later develop SZ, well beforeovert manifestations of the clinical symptoms (Isohanni et al., 2006; Leask et al., 2002). The majority of reports have shown relative stability of NSS throughout longitudinal course of the illness, from SZ prodrom (Erlenmeyer-Kimling et al., 2000; Lawrie et al., 2001) and first-psychotic break (Dazzan and Murray, 2002; Keshavan et al., 2003; Sanders et al., 1994) cases to chronic SZ (Tsuang and Faraone, 1999; Weinberger and Wyatt, 1982), with only minor progressive deterioration in aging patients (Chen et al., 1995; Heinrichs and Buchanan, 1988). However, some studies suggested that sensory integration and sequencing of complex motor acts could vary with severity of disorganization and deficit symptoms (Arango et al., 2000); whereas reports on association between NSS and psychotic symptoms are inconsistent (Arango et al., 2000; Bachmann et al., 2005; Whitty et al., 2003). More severe NSS were associated with lower educational level and older age at SZ onset in a first-break sample (Chen et al., 2005), however expression of NSS did not predict the illness outcomes, such as frequency of relapses or occupational functioning. Studies in BD revealed inconsistent findings, with some studies reporting associations between NSS and clinical symptoms (Whitty et al., 2006), and others reporting no such associations (Negash et al., 2004).

3.5.5. Effects of medications on NSS

Studies consistently reported no effect of psychotropic treatment on NSS in SZ as evident from first-break medication-naïve (Browne et al., 2000; Dazzan and Murray, 2002; Keshavan et al., 2003; Sanders et al., 1994) and chronic SZ samples (Arango et al., 2000; Cox and Ludwig, 1979; Dazzan et al., 2004; Heinrichs and Buchanan, 1988; Sevincok and Topaloglu, 2006). In contrast, no studies specifically examined the effect of mood stabilizers and other medications on NSS in BD.

3.5.6. Association between NSS and other intermediate phenotypes

Extensive evidence suggests an association between NSS and neurocognitive deficits in SZ, possibly relying on common neural substrates (Arango et al., 1999; Cuesta et al., 1996; Flashman et al., 1996; Wong et al., 1997). In particular, motor coordination signs were specifically associated with abnormalities in attention (E.Y. Chen et al., 2001; Cuesta et al., 1996; Wong et al., 1997) and visual-spatial memory (Arango et al., 1999; Cuesta et al., 1996). The sequencing of complex motor acts was correlated with executive functioning (Smith et al., 1999); and sensory integration signs were related to executive functioning, general intelligence level and verbal performance (Chan and Chen, 2004b). These findings suggest that motor/coordination components of NSS may involve common neural substrates of higher cognitive functioning rather than reflect simple motor control (Chan and Gottesman, 2008). In BD, one study reported a direct association between impaired executive function measured by Trials B test and severity of NSS (Goswami et al., 2006) in euthymic BD probands.

Few studies have reported associations between NSS and oculo-motor abnormalities in SZ samples. In particular, Ross et al. (1998) linked poor sensory integration to abnormalities in smooth pursuit eye movement in chronic SZ subjects. In addition, disinhibition signs were foundto be associated with increased blink rate in SZ (Chan and Chen, 2004a). This may suggest that sensory integration and disinhibition components of NSS and oculo-motor disorder in SZ may be various manifestations of common, underlying pathophysiological processes.

Recent studies have shown associations between NSS and brain structure intermediate phenotypes, with earlier reports on enlargement of cerebral ventricles (Weinberger and Wyatt, 1982); smaller whole brain volume and smaller temporal horn volume (Rubin et al., 1994), although there are inconsistent across the studies (Kolakowska et al., 1985). More recent reports based on structural MRI voxel-based morphometry analysis found associations between motor coordination and sensory integration NSS and reduced grey matter volume in subcortical structures, including putamen, globus pallidus and thalamus in new-onset SZ sample (Dazzan et al., 2004). In addition, this study reported an association between sensory integration signs and grey matter reduction in frontal and temporal cortices. Later analysis from the same group found similar associations between NSS and cortical grey matter reductions in healthy volunteers (Dazzan et al., 2006).

3.5.7. Commentary: NSS in schizophrenia/bipolar disorder psychosis

NSS appear to be promising intermediate phenotypes of SZ as these objective markers of the illness are stable, trait-like, independent of clinical phase, psychosis duration or treatment, that show clear familial association, and are presumably heritable. Although reports on genetic underpinnings of this intermediate phenotype in SZ are preliminary, linkages to COMT and 5HT2a receptor genes have been suggested. In addition, associations between NSS, neurocognitive deficits, eye movement abnormalities and brain structure alterations may suggest that these intermediate phenotypes of SZ may be various manifestations of interrelated pathophysiological mechanisms of the illness that may rely on common neural substrates. Reports on NSS in BD are currently at the preliminary stage. A few available reports have shown that NSS in symptomatic and euthymic BD are not dissimilar to the signs observed in SZ. Growing evidence suggests that NSS do not exhibit diagnosis-specific characteristics in either their prevalence or subtypes and that SZ and BD are indistinguishable in terms of subtle neurological dysfunction. Therefore, it possible to speculate that NSS may be one of the overlapping intermediate phenotypes in SZ and BD, reflecting shared neuro-developmental and genetic abnormalities, although future studies are required to explore this assumption.

4. Discussion

Categorization of psychotic illnesses remains an ongoing controversy. Although Kraepelinian subtyping of psychosis, based on symptom characteristics, course of illnesses and overall outcome, was historically beneficial and allowed clinicians to better systematize psychiatric illnesses, this dichotomous approach introduced some obstacles in developing biologically based conceptualization of psychosis. Recent genetic and intermediate phenotype studies do not support dichotomous categorization of psychosis. Instead, there is a growing body of literature suggesting that SZ and BD represent a clinical continuum with partially overlapping symptom dimensions, neurophysiology, genetics and treatment responses (Fig. 1). Recent whole genome studies have identified a number of “hotspots” that overlap in SZ and BD (1q32, 10p11–15, 13q32, 18p11.2 and 22q11–13) and, perhaps, influence susceptibility to psychosis, independent of categorical diagnosis. Although the majority of studies which have attempted to search for individual candidate genes have been statistically underpowered, recent findings suggest that the identified candidate genes show stronger associations with symptom dimensions, such as psychosis (NRG1, DISC1 and Dysbindin) or mood symptoms (G72/G30, BDNF), across the SZ-mood disorder continuum, but not with the diagnoses themselves. A growing number of reports suggest that psychosis may be conceptualized as a distinct clinical phenotype with, at least partially, a unique neurophysiology and genetic background. Hypothetically, genes or sets of genes, interacting with environmental factors, may predetermine vulnerability to psychosis. Depending on additional syndrome-specific genetic combinations and environmental interactions, psychosis may co-manifest with different clinical phenotypes, e.g., mood symptoms, cognitive dysfunction, or negative symptoms, composing categorical diagnoses. Although intriguing, this conceptualization is preliminary and should be treated with caution due to difficulties in establishing unequivocal evidence for genetic associations in complex, clinically and genetically heterogeneous psychiatric illnesses.

Fig. 1.

Overlapping and unique candidate genes and intermediate phenotypes in schizophrenia—bipolar disorder continuum.

↑, increased; ↓, decreased; SZ, schizophrenia; P-BD, psychotic bipolar disorder; nonP-BD, non-psychotic bipolar disorder; SZP, probands with schizophrenia; SZR, biological relatives of probands with schizophrenia; BDP, probands with bipolar I disorder; COMT, catechol-O-methyltransferase; RGS4, regulator of G protein signaling 4; DISC1, disrupted in schizophrenia 1; NRG1, neuregulin 1; Dysbindin, distrobrevin-binding protein 1; BDNF, brain derived neurotrophic factor; GABRB1,2, GABA A receptor, beta 1,2; GRM7, glutamate receptor, metabotropic 7; SYN3, synapsin III; NSS, neurological “soft” signs; GM, grey matter; WM, white matter; DLPFC, dorso-lateral prefrontal cortex; MTL, medial temporal lobe; IPC, inferior parietal cortex; CG, cinulate gyrus; Thal, thalamus; Cer, cerebellum; V, volume; VV, ventricular volume.

As the prevalence of psychosis in the general population is modest and each candidate gene is likely to confer a low risk on its own, it is difficult to reach the necessary statistical power to identify firm genetic associations with psychosis. One strategy, mainly supported by molecular geneticists, is to collect very large samples including several thousand psychosis cases and controls. The caveat to this approach, in addition to apparent financial and time demands, is the difficulty in identifying the right population to study due to a low reliability of psychiatric diagnosis. A sample of several thousand of SZ and BD cases will undoubtedly have significant clinical heterogeneity, potentially severe enough to muddle/mask specific genetic effects. A complimentary strategy is to implement intermediate phenotypes as more homogenous and biologically related units of the illness in place of the categorical diagnosis in genetic studies. As intermediate phenotypes are thought to be more intimately related to causal biological pathways of the illness than the diagnosis itself, this strategy may help to identify candidate gene effects in significantly smaller samples composed based on similar intermediate phenotypes manifestation independent of the clinical constructs. Recent reports have shown that various intermediate phenotypes are inherited independently and may reflect effects of individual genes or small sets of genes, therefore promise to provide more direct clues in search for psychosis vulnerability genes. In addition, using samples of probands and their biological relatives, manifesting similar intermediate phenotypes, may significantly augment the search for causal genes as these intermediate phenotypes are familial and therefore heritable.

A large body of data has been developed around SZ intermediate phenotypes. These intermediate phenotypes in SZ appear stable, trait-like characteristics which are largely independent of clinical manifestation, course of the illness and medication effects. Importantly, these deficits have been consistently found in a high proportion of unaffected biological relatives of SZ probands, suggesting that they co-segregate in families and are heritable. These characteristics of SZ intermediate phenotypes suggest that they are indeed genetically predetermined, although particular genes underlying these neurophysiologic markers have not been reliably identified. Much less is known about intermediate phenotypes in BD. The majority of available studies have been conducted in acute mania or in mixed BD populations including patients with and without a history of psychosis. Only a few small reports have been conducted in euthymic bipolar patients and in biological relatives of BD probands. The putative intermediate phenotypes, which have recently begun to be explored, largely come from the SZ research field. In fact, the overall strategy has been to test the intermediate phenotypes established in SZ, in subpopulations of BD patients. This approach is largely based on a hypothetical assumption that SZ and BD may have overlapping genetic and neurophysiologic components. Recent reports suggest that, among the variety of intermediate phenotypes, there are specific elements which are shared between the two diagnostic groups. Examples reviewed above include similar neurocognitive deficits in the areas of executive function, sustained focused attention, working memory, and verbal declarative memory; neurological “soft” signs; smooth pursuit eye movement abnormalities and saccadic disinhibition; reduced PPI, altered P50 sensory gating, delayed P300 latency; and several anatomic brain characteristics, such as increased ventricular volume, decreased MTL volume and loss of white matter in neocortex and brain stem.

There may be a continuum of neurocognitive, neurophysiological and brain structural intermediate phenotypes along which SZ patients consistently manifest the most severe disturbances, and BD, on average, show a similar but less severe pattern of dysfunction. However, phenomenological heterogeneity in the BD population raises the question of whether BD patients show an attenuated profile of the intermediate phenotypes found in SZ. Perhaps, there are subpopulations of bipolar patients among which some resemble patients with SZ on intermediate phenotypes. Among the variety of clinical dimensions of SZ and BD, psychosis appears to be a critical component which predicts common intermediate phenotypes. Independent reports show that individuals with psychotic BD have typical “SZ-like” deficits, whereas patients with non-psychotic BD demonstrate less significant or even normal neurophysiological functions. Specifically, SZ-like PPI sensorimotor deficits have been observed in adults with acute psychotic mania, but not in non-psychotic euthymic BD children. P50 sensory gating abnormalities are uniquely observed only in BD patients with a lifetime history of psychosis. Similar oculo-motor abnormalities (impaired SPEM and saccadic disinhibition) have been reported in probands with SZ and psychotic BD. Even more importantly, a few existing studies in biological relatives of probands with psychotic BD showed significantly delayed P300 latency, as well as SPEM alterations similar to those in relatives of SZ probands. These results, although preliminary, suggest that there are components of intermediate phenotypes that may represent unique biological markers of psychosis shared between SZ and BD and are largely independent of diagnosis. If so, future search for intermediate phenotypes and underlying genetics should focus on dimensions, like psychosis, mood symptoms, negative symptoms, impulsivity, cyclisity, etc., instead of targeting categorical diagnosis. This strategy may also help to identify novel neurophysiological markers which uniquely characterize subtypes of the two illnesses, e.g., intermediate phenotypes and underlying genetics for subtypes of SZ with severe deficit symptoms vs. BD with rapid cycling or seasonal pattern.

Studies have already reported a heritability component to intermediate phenotypes. Studies which have genotyped intermediate phenotypes, although few in number, have implicated several genetic loci. For example, P50 sensory gating deficit has been linked to alpha 7 nicotinic receptor locus in 15q13–14 and D22s315 locus in 22q; PPI deficits to NRG1, COMT and PRODH; altered P300 was associated with Met/Met COMT genotype and NRG1; abnormal SPEM was linked to 6p21. From this perspective, intermediate phenotypes may be valuable biological predictors of the illness in genetically vulnerable individuals (e.g., offsprings of individual with SZ). Potentially, intermediate phenotypes may be used as objective biological risk markers of psychosis in “at-risk individuals”, similar to blood glucose level in offsprings of diabetic patients or screening colonoscopy in vulnerable individuals with a history of familial colon cancer. Screening measures of this kind are routine in medicine and are obligatory components of medical diagnoses. In psychiatry, where diagnosis is exclusively based on empirical clinical criteria, well defined and specific intermediate phenotypes may particularly contribute to reliability and specificity of diagnosis.

Despite these promising perspectives in the field of psychosis genetics and intermediate phenotypes, challenges remain. The methodologies used to measure intermediate phenotypes are relatively complex and can be variable across laboratories. As a result, there is often difficulty in comparing findings from study to study. In addition, the physiological meaning and functional interpretation of the intermediate phenotypes are still not well understood. Intermediate phenotypes remain measures of psychiatric illnesses which are hypothetically related to their pathogenesis, but the mechanisms of these relations are unclear. Moreover, while hypothesized, it still has to be established that intermediate phenotypes have a tighter link to genes than the disease itself. Intermediate phenotypes, identified so far, have shown only modest heritability estimates. This may be due to the relative complexity of the existing intermediate phenotypes, probably, the best illustration being with the imaging intermediate phenotypes. Further refinement of intermediate phenotypes and related methodologies is needed to produce more specific markers for implication in linkage and molecular genetic studies. Finally, while a rich body of work has been done in order to characterize SZ genetics and intermediate phenotypes, significant gaps remain in this research field in BD. Testing established in SZ intermediate phenotypes in various subpopulations of BD patients (e.g., psychotic vs. non-psychotic, BD, type I vs. BD, type II), as well as searching for unique BD intermediate phenotypes underlying characteristic symptom dimensions like impulsivity, cyclicity, seasonal and diurnal variation, are in emergent need. Finally, although existing research supports the conceptualization of SZ and BD as a “psychosis continuum”, the extent to which genetics and neurophysiological deficits overlap in the two illnesses is unclear. Future studies conducted in a comparative way with large samples of SZ and BD families, using a full range of intermediate phenotypes acquired with identical methodologies, are needed.