Evidence for shared susceptibility in bipolar disorder and schizophrenia

INTRODUCTION

This paper will present a limited review of family and molecular genetic linkage studies of bipolar disorder (BPD) and schizophrenia (SZ). This review emphasizes the evidence that BPD and SZ share some genetic susceptibility. The evidence for this hypothesis derives from family studies and molecular linkage studies.

EPIDEMIOLOGIC AND CLINICAL SIMILARITIES

BPD and SZ are common, chronic illnesses that share several epidemiologic characteristics [Nurnberger and Berrettini, 1998]. For example, when reasonably narrow diagnostic criteria are used, BPD and SZ each have lifetime risks of ∼1% across the world's populations, occurring at this rate in several continents.

BPD and SZ are common, chronic illnesses that have in common several epidemiologic characteristics. For example, when reasonably narrow diagnostic criteria are used, BPD and SZ each have lifetime risks of ∼1% across the world's populations, occurring at this rate in several continents.

Although BPD and SZ are common in young adulthood (onset of illness typically occurs between ages 15 and 25), these disorders are unusual in prepubertal children. Additionally, it is unusual for either disorder to arise de novo after age 50. Thus, BPD and SZ have similar age-at-onset distributions. BPD and SZ describe psychotic disorders that often assume episodic courses of illness (with partial to complete remissions and clear exacerbations). Although severe forms of BPD (in which the victim is ill almost continuously) are described, relatively chronic, unremitting SZ is a classic form of illness. With rare exceptions, BPD and SZ disorders are lifelong conditions: once DSM-IV criteria for BPD or SZ are met, the disorder persists through life. Spontaneous and lifelong (permanent) remissions in either diagnostic category are very unusual. BPD and SZ are characterized by increased risk for suicide, and both syndromes affect men and women equally [for review, see Tsuang and Own, 2002].

While there are clinical distinctions between these two nosological groups, there are no pathognomonic signs or symptoms. Although BPD is characterized by repeated episodes of both depression and mania, no such opposite poles of SZ exist. However, to some extent the negative symptoms of SZ (avolition, poverty of thought, asociality, etc.) do share clinical characteristics of depression (e.g., social withdrawal, psychomotor retardation). To some extent, the positive symptoms (delusions, hallucinations) share characteristics of mania, a syndrome frequently characterized by delusions and hallucinations.

Until recently, treatments for BPD and SZ were relatively separate. Mood stabilizers, such as lithium and anticonvulsants, and antidepressants, do not show substantial efficacy in SZ. Atypical antipsychotic medicines represent the first choice treatment for SZ, as they improve both positive and negative symptoms. Recently, olanzepine showed efficacy among BPD patients in preventing recurrences of both mania and depression [Tohen et al., 2003]. These data suggest that atypical antipsychotics may be mood stabilizers. In light of the evidence for overlap in susceptibility (see below), perhaps it should be expected that some medications for one category of these disorders might be beneficial for individuals with the other category of illness.

Familial aggregation has been demonstrated repeatedly for SZ and BPD (see below). Twin and adoption studies of BP and SZ disorders are consistent with substantial heritability: estimates of heritability for these disorders are quite similar: ∼70% for SZ disorders and BP disorders

Twin and adoption studies of BP and SZ disorders are consistent with substantial heritability: estimates of heritability for these disorders are quite similar: ∼70% for SZ disorders and BP disorder.

[for review, see Berrettini, 2002]. Considerable familial aggregation is reflected in the multiple BPD family studies conducted in the last several decades [Heltzer and Winokur, 1974; James and Chapman, 1975; Johnson and Leeman, 1977; Angst et al., 1980; Gershon et al., 1982; Winokur et al., 1982, 1995; Baron et al., 1983; Weissman et al., 1984; Maier et al., 1993]. Given this genetic background, it was widely expected that linkage techniques, which have been so successful in identifying genes for Mendelian disorders, would reveal the genes that increase risk for BPD. BPD susceptibility gene identification has been a slow, difficult process, as with other common complex traits (such as asthma, hypertension, and diabetes). This is attributable to the small effect sizes of BPD susceptibility genes and genetic heterogeneity. Recent identification from confirmed linkage regions of susceptibility genes in non-insulin-dependent diabetes mellitus [Horikawa et al., 2000], inflammatory bowel disease [Hugot et al., 2001; Rioux et al., 2001], and SZ [Steffansson et al., 2002; Straub et al., 2002] suggests that identification of BPD susceptibility genes can be expected in the near future.

FAMILY STUDIES IN SZ AND BPD: POTENTIAL OVERLAP IN SUSCEPTIBILITY

The hypothesis that BPD and SZ might share some of the same susceptibility factors derives support from family studies.

The hypothesis that BPD and SZ might share some of the same susceptibility factors derives support from family studies.

In family studies of SZ, an increased risk for schizoaffective and unipolar (UP) disorders among the first-degree relatives of schizophrenic probands is reported [Gershon et al., 1988; Kendler et al., 1993; Maier et al., 1993, 2002; Somnath et al., 2002]. Kendler et al. [1993] found increased risk for psychotic affective illness among the relatives of individuals with SZ, compared to relatives of controls.

Similarly, increased risks for schizoaffective and UP disorders are found among the first-degree relatives of BPD probands, compared to first-degree relatives of controls [Gershon et al., 1982; Winokur et al., 1982; Weissman et al., 1984]. Thus, there is overlap in risk for schizoaffective disorders and UP disorders among relatives of BPD or SZ probands. However, there is no increased risk for BPD among first-degree relatives of SZ probands [Baron et al., 1985; Coryell and Zimmerman, 1988; Maj et al., 1991; Kendler et al., 1993; Maier et al., 1993], nor is there increased risk for SZ among first-degree relatives of BPD probands [Heltzer and Winokur, 1974; James and Chapman, 1975; Johnson and Leeman, 1977; Angst et al., 1980; Gershon et al., 1982; Winokur et al., 1982, 1995; Baron et al., 1983; Weissman et al., 1984; Maier et al., 1993]. This leaves an ambiguity. In the most rigorous sense, the impression that BPD and SZ cannot share susceptibility is found in the fact that there is no increased risk for the one disorder among relatives of the other disorder. However, the family study data risk for schizoaffective disorder and recurrent UP disorders are consistent with partial overlap in susceptibility for SZ and BPD, in that relatives of probands with these disorders are at increased risk for schizoaffective and recurrent UP disorders.

MOLECULAR LINKAGE STUDIES IN SZ AND BPD: POTENTIAL OVERLAP IN SUSCEPTIBILITY AT 18p11.2

Berrettini et al. [1994, 1997] and Detera-Wadleigh et al. [1999] reported evidence for a BPD susceptibility locus on 18p11 using affected sibling pair (ASP) and affected pedigree member (APM) methods (P = 10⁻⁴–10⁻⁶). Independent evidence (0.01 ≤ P ≤ 0.0001) for confirmation of this finding was reported by Stine et al. [1995], Nothen et al. [1999], Bennett et al. [2002], and Turecki et al. [1999]. As part of Genetic Analysis Workshop 10, independent BPD chromosome 18 linkage data sets, including ∼1,200 samples, were assembled for meta-analyses [Goldin et al., 1997]. An ASP (N = 382 sibling pairs) meta-analysis yielded P = 2.8 × 10⁻⁸ at marker D18S37 [Lin and Bale, 1997].

In light of the family studies suggesting partial overlap in susceptibility for BPD and SZ (see above), it is of interest to determine whether confirmed BPD loci might overlap with reports of SZ susceptibility loci. Schwab et al. [1998] employed ∼20 chromosome 18 markers in a linkage study of 59 multiplex German and Israeli SZ pedigrees, in which there were 24 affective disorder cases (two were BP). When these data were analyzed in two-point parametric methods, using a broad affection status model (including affective disorders), the maximum LOD score was 3.1 at D18S53. A multipoint nonparametric analysis revealed P = 0.002 at D18S53. Evidence for SZ linkage disequilibrium at an 18p11.2 microsatellite was also noted by Schwab et al. [1998]. The data of Schwab et al. [1998] were most positive when a broad affection status model was employed, including SZ, schizoaffective disorders, UP disorders, and BPDs.

This is the only SZ linkage report to identify the 18p11 genomic region. However, most SZ linkage reports do not include affective disorders in the affection status model. Thus, there may not be comparable methods in most SZ linkage studies. Given the results of Schwab et al. [1998], it is reasonable to consider the 18p11.2 region as one of potential susceptibility to both BPD and SZ.

A promising 18p11.2 candidate gene is an inositol monophosphatase gene (IMPA2), which is an enzyme of the phosphoinositol triphosphate second messenger signaling cascade [Yoshikawa et al., 2001]. Single nucleotide polymorphisms (SNPs) in IMPA2 are associated with SZ in a Japanese sample [Yoshikawa et al., 2001].

MOLECULAR LINKAGE STUDIES IN SZ AND BPD: POTENTIAL OVERLAP IN SUSCEPTIBILITY AT 13q32

One genomic region of potential overlap in genetic susceptibility for SZ and BPD is 13q32. Lin and Bale [1997] observed a LOD score of 2.58 (P = ∼0.001) at 13q32 markers (D13S122 and D13S128) in a linkage study of SZ. In a genome scan of 54 SZ families, Blouin et al. [1998] report a P value of 0.00002 (LOD = 3.6) at the 13q32 marker, D13S174. Subsequently Brzustowicz et al. [1999] confirmed these reports in 21 Canadian SZ families, with a maximal LOD score of 3.92 at the 13q marker D13S793. Thus, there are at least three independent reports, with substantial statistical significance, consistent with a 13q32 SZ susceptibility locus.

Detera-Wadleigh et al. [1999] described linkage (P = 0.00003) to 13q32 markers (D13S1271 and D13S779) in 22 BP kindred of European ancestry. One may be concerned that the kindred studied by Detera-Wadleigh et al. [1999] were misclassified. However, these kindred reveal evidence for linkage to 18p11.2 and 21q21 [Detera-Wadleigh et al., 1999], confirmed BP susceptibility loci. Kelsoe et al. [2001] reported linkage of BP disorder to 13q32 markers, with LOD = 2.4 at D13S154. Thus, in the 13q32 region, a confirmed SZ susceptibility locus, there are statistically impressive reports of linkage at this locus in BP disorder.

A promising candidate gene in this region is G72 [Chumakov et al., 2002], a gene of uncertain function that interacts with D-amino acid decarboxylase. SNPs in the G72 gene are associated with SZ [Chumakov et al., 2002] and BPD [Hattori et al., 2003].

MOLECULAR LINKAGE STUDIES IN SZ AND BPD: POTENTIAL OVERLAP IN SUSCEPTIBILITY AT 22q11-13

The velocardiofacial syndrome (VCFS) presents in childhood with variable clinical manifestations, including cardiac anomalies, typical face, learning disabilities, and, in ∼30% of cases, psychosis. The form of the psychosis is affective according to some authorities [Papolos et al., 1996; Carlson et al., 1997], although others describe it as SZ-like [Pulver et al., 1994a]. The VCFS is caused by microdeletions in 22q11. Although the exact boundaries of the critical region remain uncertain, most deletions are ∼3 Mb in size. Stemming from an initial report by Pulver et al. [1994b], a substantial multicenter effort to confirm a SZ susceptibility locus in this region resulted in evidence for linkage, at D22S278 [Gill et al., 1996].

Lachman et al. [1996] first reported evidence for BPD linkage in the VCFS region. Subsequent weakly positive reports include Detera-Wadleigh et al. [1999] and Edenberg et al. [1997]. Kelsoe et al. [2001] report a LOD score of 3.8 at D22S278, while at this same microsatellite locus, Mujaheed et al. [2000] reported evidence for linkage disequilibrium in BPD Arab Palestinian kindred. Thus, this 22q11 region may be a locus for susceptibility to SZ and BPD.

Several promising candidate 22q11 genes have been described as explanations for the 22q11 linkage results in BPD and SZ.

Several promising candidate 22q11 genes have been described as explanations for the 22q11 linkage results in BPD and SZ.

Proline dehydrogenase (PRODH2) variants are in linkage disequilibrium with SZ in some populations [Liu et al., 2002]. A second candidate is a G-protein-coupled receptor kinase (GRK3), a gene that is markedly upregulated in rat brain during chronic amphetamine exposure [Niculescu et al., 2000]. A third promising candidate is catechol-O-methyltransferase (COMT), which has been associated with cognitive difficulties in SZ [Egan et al., 2001] and with SZ itself [Shifman et al., 2002].

MOLECULAR LINKAGE STUDIES IN SZ AND BPD: POTENTIAL OVERLAP IN SUSCEPTIBILITY AT 8p22

There have been several reports of SZ susceptibility mapped to 8p22-24. In 54 extended SZ kindred, Blouin et al. [1998] reported evidence for a SZ locus at 8p22: the heterogeneity LOD score was 4.5, and nonparametric analysis yielded P = 0.0001. Brzustowicz et al. [1999], in 21 extended Canadian SZ pedigrees with 8p markers, described a maximum multipoint LOD score of 2.1 at D8S136. Levinson et al. [1996], in a multicenter collaborative effort, reported independent results (that did not include the pedigrees of Blouin et al. [1998]), yielding P = 0.00018 in this same region of 8p22. Gurling et al. [1999], in a study of 13 extended European SZ kindred, reported a LOD of 3.6 for 8p22 markers. Thus, these reports constitute a confirmed SZ linkage. Recently, Stefansson et al. [2002] described evidence from both mouse and human studies that neuregulin 1 (NRG1) is an 8p susceptibility gene for SZ. They found linkage disequilibrium with NRG1 haplotypes in Icelandic individuals with SZ.

Recently Ophoff et al. [2002] described a linkage disequilibrium signal at 8p22 in distantly related BPD persons from a population isolate in the Central Valley of Costa Rica. Greater-than-expected sharing of a 5 cM three-marker haplotype at D8S503 was observed. The significance level reported was 0.000057. Thus, there is a statistically impressive report of BPD susceptibility mapped to this same region. The implicated region does not contain the NRG1 gene, however.

MOLECULAR LINKAGE STUDIES IN SZ AND BPD: POTENTIAL OVERLAP IN SUSCEPTIBILITY AT 10p14

A fifth region of potential BPD/SZ susceptibility is found at 10p14. Faraone et al. [1998], Straub et al. [1998], and Schwab et al. [2000] reported evidence for linkage of SZ to 10p14 markers. Faraone et al. [1998] reported P = 0.0004 for marker D10S1423 and P = 0.0006 for D10S582, in a study of 43 American SZ kindred of European ancestry. Straub et al. [1998], in a study of Irish SZ kindred, reported P = 0.006 for this region in a multipoint analysis. For marker D10S582, Schwab et al. [1998] reported P = 0.0058 for German SZ kindred. These three groups of investigators studied independent sets of kindred that were of general European ancestry.

Foroud et al. [2000] studied BP kindred from the NIMH Genetics Initiative. They found LOD = 2.5 (P = 0.001) for marker D10S1423. Thus, the 10p14 region may represent another genomic region at which there is shared susceptibility for BPD and SZ.

META-ANALYSES

Badner and Gershon [2002] analyzed complete genome scans for BPD and SZ. There were 11 BPD genome scans (∼1,250 affecteds) and 18 SZ genome scans (∼1,900 affecteds). The most promising regions of the genome for BPD were 13q32 and 22q11, while the most promising SZ regions were 8p24, 13q32, and 22q11. All of these regions have been implicated in shared BPD/SZ susceptibility, as noted above.

A rank-sum type of meta-analysis [Lewis et al., 2002] was applied to all SZ linkage data by Levinson et al. [2002]. They concluded that 8p and 22q were among the most promising regions. A similar analysis was applied to BPD linkage data [Segurado et al., 2003]. These authors concluded that there was no single region that reached genome-wide significance, perhaps due to the relatively small numbers of affected genotyped, when compared to the available SZ data.

SUMMARY

Family and linkage studies are consistent with the concept that SZ and BPD share some genetic susceptibility. Multiple regions of the genome, including 18p11, 13q32, 22q11, 10p14, and 8p22, represent areas with potential BPD/SZ shared genetic susceptibility.

Family and linkage studies are consistent with the concept that SZ and BPD share some genetic susceptibility. Multiple regions of the genome, including 18p11, 13q32, 22q11, 10p14, and 8p22, represent areas with potential BPD/SZ shared genetic susceptibility.

One must remember that no single locus has been identified in the majority of linkage scans for either disorder. Thus, one must regard the hypothesis presented here as only a theory, to be tested through linkage disequilibrium studies of candidate genes from these linkage regions.

As susceptibility genes in these regions are identified, through the application of linkage disequilibrium mapping methods to large sample sizes, it may be necessary to develop a new, genetically based nosology, in which this overlap is more accurately represented. This process of BPD/SZ susceptibility gene identification has begun, tentatively, with the 13q32 region, where studies indicate that haplotypes in the G72 gene increase risk for both SZ [Chumakov et al., 2002] and BPD [Hattori et al., 2003].