First family-based test for association of neuregulin with bipolar affective disorder

The neuregulin gene (NRG1) has generated a great amount of interest in recent years as a highly replicated susceptibility gene for schizophrenia.1 As a result of the genetic overlap between schizophrenia and bipolar affective disorder (BPAD) that has been revealed by linkage studies2 and the location of NRG1 within one such overlapping linkage region (8p21), variants within this gene were tested for association with BPAD in a case–control sample of English/Welsh origin.3 A three-marker haplotype comprised of the alleles from the core makers of the schizophrenia-associated haplotype, HAPICE, exhibited a marginal association with BPAD, an effect which became highly significant when the analysis was restricted to cases with mood incongruent psychotic episodes. Here, we report the results of the first family-based association analysis of NRG1 with BPAD. The three core markers from HAPICE (SNP8NRG221533, 478B14-878, 420M9-1395) and two additional markers (SNP8NRG243177 and D8S1810) were genotyped, using standard methods, in a collection of 92 BPAD type I trios from the Irish population. All families were interviewed using the Schedule for Affective Disorders and Schizophrenia (Lifetime version: SADS-LB) and diagnoses met DSM-IV criteria. Association analyses were performed using the TDT,4 ETDT5 and Transmit6 programmes. From the single marker association analysis, there was no evidence of association of NRG1 with BPAD in our sample (Table 1). D8S1810 did exhibit a significant over transmission of allele 5 to bipolar probands (uncorrected P=0.04) but the global P-value for this marker was not significant (Table 1: P=0.212). Interestingly, this marker was selected based on its presence within an Irish schizophrenia-associated haplotype, HapBIRE.7 A range of two to three marker haplotypes, including HAPICE and HapBIRE, also failed to exhibit any evidence of association with BPAD (Table 1). A haplotype comprised of alleles from the three microsatellite markers (M3-M4-M5) did exhibit a marginal association (Table 1, haplotype 1, 3, 5, uncorrected P=0.04) which was not significant at the global haplotype level (Table 1: P=0.31). We hypothesized that if NRG1 is principally a gene for schizophrenia, a stratified analysis restricted to trios with probands that have experienced at least one psychotic episode may reveal an association with BPAD, similar to the findings of Green et al.3 However, restriction of the association analysis to the subset of 59 trios with psychosis did not reveal any evidence for association of NRG1 with BPAD (data not shown). To conclude, we have failed to replicate the findings of Green et al.3 in a family-based collection which is not subject to the population substructure effects of case–control association studies. These findings suggest that NRG1 is not a susceptibility gene for BPAD in the Irish population. Interestingly, despite a plethora of positive associations of NRG1 with schizophrenia in diverse populations, the two Irish schizophrenia investigations have also been negative.8, 9 Although a two-marker NRG1 haplotype (HapBIRE) was initially reported to be associated with schizophrenia in an Irish case–control sample,7 the association was no longer evident after further investigation within an extended sample.9 We cannot rule out the possibility that our collection had insufficient power to replicate the association of NRG1 with BPAD. Indeed, there are moderate differences in the allele frequencies for SNPs surrounding the core haplotype in European populations, which would impact the power of replication studies, particularly for rarer alleles/haplotypes.10 However, geographical variation is unlikely to be an issue in this case as the estimated frequency of the core haplotype in our samples (7.7%) was very similar to that reported by Green et al.3 (7.8% in controls). To elucidate whether NRG1 is truly a susceptibility gene for BPAD, additional studies in family-based collections are required.