Molecular Genetics of the Psychosis Phenotype

Abstract

Objective

Relative to recent successes in elucidating the genetic mechanisms associated with complex diseases including macular degeneration, Type II diabetes, heart disease and cancer, molecular genetic approaches to psychiatric illness have met with more limited success. While factors such as small allelic effects, allelic heterogeneity and variation in population sub-structure have received considerable attention in attempt to explain the paucity of significant results in psychiatric genetics, significantly less focus has been directed towards phenotypic factors.

Method

Data derived from molecular genetic studies of the psychosis phenotype in patients with a range of psychiatric illnesses are reviewed.

Results

Available data suggest that genes do not respect the boundaries of the current diagnostic system but may confer risk for symptom-based phenotypic variation that traverses those boundaries.

Conclusions

Molecular genetic studies offer convincing evidence for a relation between genetic variation and symptom-based phenotypic variation within psychiatric illness. These data may provide novel insights into the pathophysiology of schizophrenia and other related disorders. The exploration of relationships between genetic variation and symptom variation that traverses traditional diagnostic boundaries may ultimately lead to more refined classification systems that more closely reflect the genetic etiology of psychiatric illness.

Despite its great utility for both research and practice in psychiatry, the conceptual basis of the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM) remains controversial. The primary basis for this controversy has been the reliance on a categorical approach in which different diagnostic entities share common phenomenological features. This overlap in features associated with diagnostic categories is perhaps most prominent in regards to symptoms of psychosis. For example, psychotic phenomena including positive (hallucinations, delusions), negative (apathy, avolition, and asociality), and disorganized (disorganized speech or behavior) symptoms are the cornerstones of schizophrenia (SZ) but, as illustrated in Figure 1, are often observed across a broad spectrum of diagnoses outside of the nominal “psychotic disorders” category as defined by the current edition of the DSM. Because psychotic symptoms are not pathognomonic for a specific diagnostic category, classification of an individual experiencing these symptoms is based solely on aggregate clinical phenomena. Thus, an individual who presents with auditory hallucinations and delusional beliefs may be diagnosed with SZ, schizoaffective disorder (SAD), bipolar disorder (BPD) or major depressive disorder (MDD) depending upon the presence, course and severity of concomitant affective symptoms. Due to the limitations of the categorical approach, it is anticipated that the Fifth Edition of the DSM planned for publication in 2013, will include a dimensional approach for key aspects of psychopathology. The examination of such dimensional phenotypes would likely facilitate the identification of some of the molecular genetic mechanisms underlying these complex, syndromal diseases ^{1, 2, 3, 4, 5}.

Figure 1.

Symptoms of psychosis are observed across multiple diagnostic categories. Psychosis is shown in the context of both affective and negative symptoms as usually manifested in the primary psychotic disorders (schizophrenia and schizoaffective disorder) and mood disorders (bipolar disorder and major depressive disorder). Also shown is the considerable degree of psychosis that is observed outside of these two major diagnostic categories. Such symptoms may be observed in other disorders and in non-clinical samples.

Heritability of psychosis

Initial support for a dimensional approach to elucidating the molecular genetic mechanisms associated with psychotic symptoms is derived from family studies demonstrating a partial overlap between the psychotic and affective disorders. Specifically, relatives of probands with SZ and BPD show some overlap in risk to the alternate disorder as well as an increased rate of both SAD and MDD relative to controls ^{6, 7}. This overlap is even more pronounced when comparing psychotic affective illness and SZ. For example, the reanalysis of the Iowa 500 sample using DSM-III criteria found a significantly increased prevalence of psychotic affective illness in relatives of patients with SZ compared with controls ⁸. Similar findings were reported in the Roscommon family sample ⁹; affectively ill relatives of SZ probands were more likely to have psychotic affective illness than relatives of controls, and probands with psychotic affective illness were more likely to have relatives with SZ. Similarly, Erlenmeyer-Kimling and colleagues ¹⁰ found a significantly elevated prevalence of affective psychoses in the offspring of SZ probands compared with the offspring of control probands. Finally, Laursen et al. ⁷ demonstrated an increased risk of both BPD and SZ in relatives of patients with SAD. Taken together, these epidemiological data suggest considerable overlap in the heritability of psychiatric illness characterized by the presence of psychosis.

Family-based studies have also demonstrated considerable overlap in the heritability of psychotic symptom dimensions within diagnostic categories. For example symptom factors including positive, negative and affective symptoms ¹¹, thought withdrawal, thought insertion, thought broadcasting and delusions of control ¹² and disorganization ^{13, 14, 15} have been found to be significantly correlated in sibling pairs (including twins) concordant for psychotic illness. Moreover, a meta-analysis of family-based studies of BPD demonstrated statistically significant familial aggregation for the psychosis phenotype ⁵. Finally, subthreshold psychotic symptoms such as paranoid ideation, altered perceptual experiences, odd or magical thinking and social isolation are often observed in unaffected relatives of people with psychotic illness with family-based studies consistently indicating a moderate heritability of psychosis-related phenotypes even when diagnostic status is discordant ¹⁶. The convergence of these findings provides strong support for the notion that the psychosis dimension has a heritable component.

Despite the convincing evidence suggesting familial aggregation and heritability of psychosis phenotypes, relatively limited work has been conducted to identify the molecular underpinnings of this shared familial risk. In fact, the vast majority of research in psychiatric genetics has been predicated upon the assumption that psychiatric diagnoses are distinct disease entities with distinct etiologies. The overlap in diagnostic categories containing features of psychosis, however, suggests that psychiatric diagnoses are not distinct disease entities. It is therefore not surprising that considerable overlap in the molecular mechanisms associated with psychiatric disorders, particularly among the psychotic and affective disorders, has been repeatedly supported by linkage studies ⁵, candidate gene studies ¹⁷ and more recently, by genomewide association studies (GWAS) ¹⁸.

Evidence derived from genetic linkage studies

The linkage evidence in support of shared genetic risk factors for psychosis comes primarily from linkage studies of BPD and SZ. Initial evidence for overlap in the genetics of these disorders came from a large meta-analysis of SZ and BPD that indicated strong combined linkage evidence at several loci including 1p13.3-q23.3, 2q22.1-q23.3, 3p25.3-p22.1, 5q23.2-q34, 6pterp22.3, 6p22.3-p21.1, 8p22-p21.1, 11q22.3-q24.1, 14pter-q13.1, 20p12.3-p11, and 22pter-q12.3 ¹⁹. Based on independent findings from multiple studies in both disorders other chromosomal regions including 1q42, 6q22, and 15q14 have also been shown to overlap in SZ and BPD ²⁰. Follow up studies testing the hypothesis that BPD pedigrees enriched for psychotic symptoms would show increased evidence of linkage to chromosomal regions of prior overlap between BPD and SZ, demonstrated that there were at least four overlapping chromosomal regions that could harbor shared susceptibility genes. These included regions on 13q31, 22q12, 10p12-14 and 18p11.2 ⁶. The psychotic BPD subtype was further studied in a genome-wide linkage analysis of 40 extended BP pedigrees with psychotic features; significant evidence was reported for linkage at 9q31 and 8p21, with suggestive evidence at 13q32, all of which had been previously implicated in SZ ²¹. Further, it should be noted that the region on chromosome 13q has also been linked to MDD ²². Unfortunately, of the few linkage studies that have been conducted on extended major depressive disorder pedigrees, none have focused on the psychosis phenotype within the broader diagnostic category. Finally, Fanous et al ²³ used a genome scan correlation method to test for a genetic correlation between SZ and schizotypy The results of these analyses suggested that a subset of schizophrenia susceptibility genes also affects schizotypy in non-psychotic relatives.

Evidence derived from candidate gene studies

Consistent with the aforementioned linkage evidence, data from candidate gene studies also suggest considerable overlap in genetic susceptibility across diagnostic categories ¹⁷ and implicate several well known candidate genes including: DTNBP1, DISC1, NRG1, DAOA, COMT, RGS4 and others in shared risk for the psychotic illnesses across traditional DSM boundaries. Moreover, several of these candidate genes have now been linked to specific clinical profiles within, and in some cases across, diagnostic categories.

D-amino acid oxidase activator (DAOA/ G72;G30)

Chumakov and colleagues ²⁴ first reported associations between SZ and markers in DAOA/G72 and G30, overlapping genes transcribed in opposite directions on chromosome 13q34, in a relatively large European cohort. Associations to SZ in this region are supported by at least two meta-analyses ^{25, 26}, although there is no consensus concerning the specific risk alleles or haplotypes across studies. Schumacher and colleagues ²⁷ reported an overlap in the association between SNPs within DAOA and risk for both SZ and BPD. Specifically, these authors assessed a sample of 599 patients (299 schizophrenic, 300 bipolar) and 300 controls and reported an association between SZ and several single nucleotide polymorphisms (SNPs) in a four-marker haplotype, with the most significant SNP, as well as the haplotype, also showing nominal association to BPD. Follow-up studies of DAOA have consistently replicated the association between DAOA and BPD and it is now considered one of the best supported candidate gene loci for BPD ¹⁷. Finally, Rietschel and colleagues ²⁸ assessed the relation between DAOA and major depression in 500 patients and 1,030 population-based controls and reported a significant association at the same haplotype identified by Schumacher and colleagues. The convergence of these data suggest that variation at DAOA may be more closely linked to specific aspects of psychiatric illness that cross the diagnostic boundaries of the DSM-IV than to a specific diagnostic category.

Data also suggests that DAOA may be associated with specific phenotypes of psychiatric illness rather than a specific diagnostic group. Specifically, Shulze and colleagues ²⁹ found that persecutory delusions were the only significant explanatory variable for a previously identified risk genotype shared in SZ and BPD. Moreover, in a study of SZ and BPD, Williams and colleagues ³⁰ found that variation at DAOA was more significantly associated with major mood episodes across these disorders rather than categorical diagnoses. Finally, Corvin et al ³¹ assessed the relationship between a previously identified risk variant in DAOA and symptom factors derived from Positive and Negative Symptom Scale (PANSS) in a sample of patients with SZ and SAD. These authors found that carriers of the DAOA risk variant scored significantly higher on a depression and anxiety factor than non-carriers (p = 0.01). Again, these data suggest that DAOA may influence phenotypic variation that traverses the traditional DSM-IV categories.

Neuregulin 1 (NRG1)

NRG1 was first implicated in SZ after a systematic study of 8p21-22 revealed an association between a haplotype at the 5' end of NRG1 and SZ ³². Additional support for this association was later provided by findings in 2 separate case-control samples ^{33, 34}. Moreover, a review of the NRG1 literature ³⁵ concluded that despite some differences in the associated haplotypes, the evidence for an association between NRG1 and SZ is quite strong. Although NRG1 has not yet been extensively studied in bipolar disorder, at few studies ^{36, 37} found significant evidence for association between a SZ risk haplotype within NRG1 and BPD. Moreover, in the Green et al ³⁶ study the strength of the association was substantially increased in BPD cases with predominantly mood-incongruent psychotic features, and SZ patients who had experienced mania. Goes et al ³⁸ also reported findings suggesting that NRG1 may be specifically associated with the psychotic subset of BPD probands. Thus, these data suggest that NRG1 may play a role in susceptibility to both SZ and BPD and further suggest that it may exert a specific effect on phenotypic variation that crosses diagnostic boundaries.

Dysbindin (DTNBP1)

DTNBP1 was first identified by Straub and colleagues ³⁹ as a susceptibility gene for SCZ in a follow-up analysis of their initial linkage findings on chromosome 6p. Using a family based association analysis including 270 families derived from the Irish Study of High Density Schizophrenia Families several SNPs within DTNBP1, alone or in combination, increased the risk for SZ. To date, several replication studies in different populations of European descent have been reported, and a comprehensive meta-analysis suggests that DTNBP1 is one of the most robust association signals in the literature ⁴⁰. Although the data linking DTNBP1 variation to BPD are limited, Raybould and colleagues ⁴¹ found modestly significant evidence for association in a subset of bipolar cases with predominantly psychotic episodes. Moreover, Domshke et al ⁴² recently found modest evidence for an association between variants in DTNBP1 and psychotic MDD. Collectively, these findings suggests that variation in DTNBP1 may confer risk to the psychotic syndrome rather than a specific DSM-IV diagnosis.

Available data also suggests that DTNBP1 may be more closely related to a dimension of negative symptoms than a specific diagnostic category. Negative symptoms are a particularly attractive phenotype for genetic investigation, as they tend to be more stable across the duration of illness, may present prior to the onset of illness, and are often unaffected by antipsychotic drug treatment ⁴³. Moreover, heritability studies of clinical symptomatology have indicated that affected relative pairs with schizophrenia may share similar severity of negative symptoms more often than expected by chance ¹¹, indicating a substantial genetic component. Data from several groups suggest that negative symptoms may be influenced by DTNBP1 genotype. For example, Fanous and colleagues ⁴⁴ assessed 755 subjects with psychotic illness in the Irish Study of High Density Schizophrenia Families with the Operational Criteria Checklist for Psychiatric Illness (OPCRIT) ⁴⁵ with factor analyses revealing that an 8-locus schizophrenia risk haplotype was over-transmitted to subjects with greater negative symptom factor scores. These findings have been replicated by multiple groups using measures of negative symptoms including lifetime severity ^{46, 47} longitudinal course ⁴⁸ and cross-sectional severity ^{49, 50, 51}. Taken together, these independent studies suggest that DTNBP1 variants may influence negative symptom dimensions rather than a specific diagnostic category.

Disrupted in Schizophrenia 1 (DISC1):

The gene encoding DISC1 is located at the breakpoint of a balanced translocation between chromosomes 1 and 11, t(1;11)(q42;q14) that was originally identified in an extended psychiatric pedigree and was shown to co-segregate with multiple major psychiatric disorders including SZ, SAD, BPD and MDD ⁵³. Genome scans ^{53, 54, 55} and candidate gene analyses ^{56, 57} have consistently linked DISC1 with risk for several psychiatric disorders including a report of multiple haplotypes within DISC1 associated with both affective and psychotic disorders ⁵⁸. The considerable overlap in findings relating DISC1 to a broad spectrum of affective and psychotic disorders suggests that variation in DISC1 may confer risks that traverse diagnostic boundaries.

Initial support for the relation between variation in DISC1 and specific clinical profiles was provided by Hennah and colleagues ⁵⁶. These researchers assessed 458 Finnish schizophrenia families and identified a single common, under-transmitted haplotype (HEP3) that significantly increased risk for the development of SZ. Additional analyses using more refined phenotypes that included factor scores derived the OPCRIT, however, indicated that the risk haplotype was more significantly associated with a specific clinical profile characterized by high loading on factor scores that included delusions and hallucinations, manic symptoms, depressive symptoms and negative symptoms. The strongest of these findings indicated association of DISC1 HEP3 to the delusions and hallucination factor and within this factor, the strongest association to HEP 3 was persecutory delusions (p < 0.00001). Follow up work in at least one additional patient sample has replicated these results suggesting that a known DISC1 SZ risk haplotype is more closely linked to lifetime measure of delusions ⁵⁹. Perhaps more convincing however, is the finding that DISC1 may associate to dimensions of psychosis in a general population sample. Specifically, Tomppo et al ⁶⁰ found that in a large, unselected birth cohort, variation at DISC1 was strongly associated with social anhedonia, a cardinal symptom of SZ. Collectively, these data suggest that DISC1 risk genotypes may be more closely related to dimensions of psychosis rather than a specific diagnostic category.

Catechol-O-methyltransferase (COMT)

COMT is an enzyme involved in the degradation of the neurotransmitter dopamine and has received considerable attention in molecular genetic studies of psychiatric populations. The gene encoding COMT is located in the region on chromosome 22q11 commonly deleted in velo-cardio-facial/DiGeorge syndrome (VCFS/DGS) whose phenotypic spectrum includes severe psychiatric disease that has been described as both SZ and BPD ⁶¹. To date, a plethora of candidate gene studies have found evidence for a link between variation in COMT and psychiatric illness and in contrast to other candidate gene studies, the SNPs and haplotypes implicated in the different diseases are strikingly similar. For example, Shifman and colleagues identified a single haplotype within COMT that increased risk for both SZ ⁶² and BPD ⁶³. These data suggest that it is likely that the association between COMT genotype and psychiatric illness may be better characterized by assessing the effect of COMT genetic variation on the overlapping clinical profiles of psychiatric illness.

To date, much of the work relating variation in COMT to symptom dimensions within psychiatric samples has focused on its relation to affective symptoms ^{64, 65, 66}. However, data derived from non-clinical samples suggest that variation at COMT may be associated with dimensions of psychosis. Stefanis and colleagues ⁶⁷ first reported an association of the Val158Met allele of COMT to measures of schizotypy in a sample of 543 males derived from a non-clinical population. These authors reported a significant effect of the high-activity Val allele load on negative and disorganized dimensions of schizotypy as measured by the Schizotypal Personality Questionnaire (SPQ) ⁶⁸. These data have been replicated by the same group in much larger sample comprised of 2,130 young males ⁶⁹ as well as independent samples ^{70, 71}. Moreover, Van Winkel and colleagues ⁷² recently tested for an association between variants of COMT and psychosis as measured by the Brief Psychiatric Rating Scale (BPRS) in unaffected relatives of SZ patients. These authors reported that sharing of the met allele of COMT between affected and unaffected relative pairs was a significant predictor of an association between the affected and unaffected's scores on the BPRS. These data suggest that genetic variation may be, at least in populations with a genetic predisposition to psychosis, related to the pathophysiology of psychosis.

Other candidate genes

The evidence reviewed above suggests that delineation of the role of specific genes on symptom dimensions, rather than diagnostic entities, may lead to more refined approaches and may suggest novel treatment targets for specific domains of illness. It should also be noted that in addition to the overlapping effects of the aforementioned risk genotypes, a number of studies have also demonstrated associations between symptom domains and variants within genes not associated with diagnostic categories. These include relations between negative symptoms and BDNF, DAT1 ⁷³ and HkCa3 ⁷⁴; positive symptoms and DRD4 ⁷⁵, DRD2 ⁷⁶, CCK-A ⁷⁷ and 5-HTT ⁷⁸; and disorganization with DRD2 ⁷⁶. Thus, specific genetic variants may act to modify ⁷⁹ the clinical presentation of illness without increasing the risk for the illness.

Evidence derived from linkage using quantitative traits for symptom dimensions

An alternative approach to candidate gene studies is the use of genetic linkage approaches that have the capacity to detect novel loci located within relatively large chromosomal regions. To date, a very limited number of studies have used this approach. Results of these studies, however, have identified several regions on chromosomes 6 ^{80, 81}; 8 ^{82, 81}, 9, 12 ⁸¹ and 20 ⁸¹, 23 that are associated with dimensions of psychosis in SZ suggesting that several regions of the genome may harbor genes that influence the clinical features rather than diagnosis. Although some of the regions suggestive of linkage to these quantitative traits, including several loci on chromosomes 6 and 8, had previously been linked to SZ, the regions identified on chromosomes 9, 12 and 20 had not been previously linked to the disease. Thus, these data suggest that these regions on the genome may harbor genes that influence the clinical features of schizophrenia without increasing the risk for the illness and follow-up analyses have supported these findings. Specifically, our group ⁸³ identified an association between positive psychotic symptoms is patients with SZ and a SNP in the gene encoding for the origin recognition complex (ORC3L) located within the linkage regions previously associated with positive symptoms ⁸⁰. Moreover, we also identified and association between disorganization in patients with SZ and several SNPs in the gene encoding for brain specific angiogenesis inhibitor (BAI3) located within the linkage region previously associated with disorganization ⁸¹. Although preliminary, these data point towards specific candidate genes located within previously implicated linkage peaks for psychotic symptoms and provide additional support for the role of genetic variation in dimensional measures of psychosis.

Summary and future directions

Accumulating evidence supports the role of genetic factors in variation across the psychosis phenotype and may suggest an overlap in the pathophysiological mechanisms of psychotic illness. Such findings challenge the current concepts of disease classification and demonstrate the potential utility of a symptom-based, dimensional approach. Data derived from large-scale, collaborative genome-wide association studies (GWAS) in psychiatry, albeit quite limited, provide strong support for such an approach. Indeed, many of the genetic variants that have been identified in GWAS support what had previously been observed in both linkage and candidate gene studies; they suggest considerable overlap between diagnostic categories. For example, Williams et al ⁸⁴ recently reported that of the eight robust associations that have been identified in GWAS studies for SZ or BPD, six of them show strong trans-disorder effects. Indeed, the largest psychiatric GWAS to date ⁸⁵ recently reported that when SZ and BPD are analyzed jointly, the number of genome-wide significant loci increases with and enhancement of signals at loci previously shown to confer risk to both disorders. These data are consistent with data derived from the International Schizophrenia Consortium (ISC) that showed that common polygenic variation explained approximately 1/3 of the variation in liability to SZ and the same polygenic variation was substantially shared with BPD ⁸⁶. Follow-up analyses using the same polygenic model in an independent sample demonstrated that the observed overlap may relate more closely to the sharing of a ‘schizophrenia-like’ phenotype ⁸⁷. Given the clinical heterogeneity and molecular complexity of psychiatric illness it is unlikely that any single genetic paradigm will account for a large degree of the variance to be explained. However, the exploration of relationships between genetic variation and symptom variation that traverses traditional diagnostic boundaries may ultimately lead to more refined classification systems that more closely reflect the genetic etiology of psychiatric illness.

Clinical Implications.

• The psychosis phenotype traverses traditional diagnostic boundaries

• Studies seeking to assess the relationship between genetic variation and the psychosis phenotype may provide insight into the pathophysiological mechanisms of psychiatric illness.

Limitations

• Despite its potential utility, there is a very limited amount of work that has been conducted to parse the specific effects of genetic variants on phenotypic factors that are shared amongst diagnostic groups.

• Large-scale studies using genome-wide association, next-generation sequencing and other emerging analytic tools targeted at specific phenotypes are needed.