Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
SIR – The pericentromeric region of chromosome 18, specifically 18p11.2, is described as schizophrenia susceptibility locus1,2 (the maximum LOD score to date is 3.1 by Schwab et al1), in addition to being a strong candidate region for bipolar disorder.3 We have previously cloned two novel brain-derived transcripts from this region: the gene for a second human myo-inositol monophosphatase (IMPA2)4 and a gene of unknown function, C18orf1.5 Our prior genetic analysis of IMPA2 generated evidence for its association with Japanese schizophrenia in a case–control context.6 Owing to an inherent risk for producing false-positive results in case–control designs, we performed a family-based linkage disequilibrium (LD) study as the first step towards identifying relevant genetic loci around 18p11.2, and then followed this up using independent case–control samples. The ethics committee of RIKEN approved the present study, and written informed consent was obtained from all participants.
First, we genotyped 80 independent trios, each composed of schizophrenic offspring (based on DSM-IV) and their parents, using 15 markers that included seven microsatellites and eight single-nucleotide polymorphisms (SNPs) (Table 1, except for 6409T>C). These markers, chosen at even intervals as long as possible, were selected from the 14 Mb region spanning 18p11.22 and the proximal q-arm, with the exception of the sparsely represented pericentromeric region and two markers each from the GNAL and IMPA2 genes (the average marker density on 18p is 1/280 kb). This marker spacing is valid for an initial genetic scan, given the recently reported LD distance of 0.5–2 cM retained in the Japanese population.7 We analyzed genotype data using the extended transmission disequilibrium test (ETDT),8 which calculates the allele wise TDT statistics that determine the preferential transmission of specific alleles, and the genotype-wise TDT statistic that evaluates the deviation of allele transmission from each parental genotype. We simulated the empirical significance levels of the ETDT results using the MCETDT program, (http://www.mds.qmw.ac.uk/statgen/dcurtis/software.html). P values indicating significantly distorted transmission were obtained from markers D18S852 and D18S40 (Table 1). Since D18S852 consists of a GCT triplet repeat present within the 3′-untranslated region of the C18orf1 gene, we tested another polymorphism, 6409T>C, which we had identified from the same region.5 This SNP also showed a modest but significant association in both allele- and genotype-wise TDT statistics (Table 1). However, all three markers lost significance if a Bonferroni correction for multiple testing was applied. To compute the statistical power obtained from the present family-based association study, we used the TDT Power Calculator program.9 Our sample size had a power of 0.821 to detect significant association, based on a model assuming a dominant trait with an allele frequency of 0.2, a penetrance of 0.4 and a phenocopy risk of 0.4%.10
Kikuchi, M., Yamada, K., Toyota, T. et al. Two-step association analyses of the chromosome 18p11.2 region in schizophrenia detect a locus encompassing C18orf1.
Mol Psychiatry8, 467–469 (2003). https://doi.org/10.1038/sj.mp.4001280