Tanahashi Toshihito, Osabe Dai, Nomura Kyoko, Shinohara Shuichi, Kato Hitoshi, Ichiishi Eiichiro, Nakamura Naoto, Yoshikawa Toshikazu, Takata Yoichiro, Miyamoto Tatsuro, Shiota Hiroshi, Keshavarz Parvaneh, Yamaguchi Yuka, Kunika Kiyoshi, Moritani Maki, Inoue Hiroshi, Itakura Mitsuo
Division of Genetic Information, Institute for Genome Research, The University of Tokushima, 3-18-15, Tokushima, Japan.
Hum Genet. 2006 Nov;120(4):527-42. doi: 10.1007/s00439-006-0231-0. Epub 2006 Sep 6.
Several linkage studies have predicted that human chromosome 20q is closely related to type 2 diabetes, but there is no clear evidence that certain variant(s) or gene(s) have strong effects on the disease within this region. To examine disease susceptibility variant in Japanese, verified SNPs from the databases, with a minor allele frequency larger than 0.15, were selected at 10-kb intervals across a 19.31-Mb region (20q11.21-13.13), which contained 291 genes, including hepatocyte nuclear factor 4alpha (HNF4alpha). As a result, a total of 1,147 SNPs were genotyped with TaqMan assay using 1,818 Japanese samples. By searching for HNF4alpha as a representative disease-susceptible gene, no variants of HNF4alpha were strongly associated with disease. To identify other genetic variant related with disease, we designed an extensive two-stage association study (725 first and 1,093 second test samples). Although SNP1146 (rs220076) was selected as a landmark within the 19.31 Mb region, the magnitude of the nominal P value (P = 0.0023) was rather weak. Subsequently, a haplotype-based association study showed that two common haplotypes were weakly associated with disease. All of these tests resulted in non-significance after adjusting for Bonferroni's correction and the false discovery rate to control for the impact of multiple testing. Contrary to the initial expectations, we could not conclude that certain SNPs had a major effect on this promising locus within the framework presented here. As a way to extend our observations, we emphasize the importance of a subsequent association study including replication and/or meta-analysis in multiple populations.
多项连锁研究预测人类20号染色体与2型糖尿病密切相关，但尚无明确证据表明该区域内的某些变异或基因对该疾病有强烈影响。为了研究日本人的疾病易感性变异，我们从数据库中选择了次要等位基因频率大于0.15的已验证单核苷酸多态性（SNP），以10 kb的间隔跨越一个19.31 Mb的区域（20q11.21 - 13.13），该区域包含291个基因，包括肝细胞核因子4α（HNF4α）。结果，使用1818份日本样本通过TaqMan分析对总共1147个SNP进行了基因分型。通过将HNF4α作为代表性疾病易感基因进行搜索，未发现HNF4α的变异与疾病有强烈关联。为了鉴定与疾病相关的其他遗传变异，我们设计了一项广泛的两阶段关联研究（725个首次测试样本和1093个第二次测试样本）。尽管SNP1146（rs220076）被选为19.31 Mb区域内的一个标记，但名义P值（P = 0.0023）的幅度相当小。随后，基于单倍型的关联研究表明，两种常见单倍型与疾病的关联较弱。在对Bonferroni校正和错误发现率进行调整以控制多重检验的影响后，所有这些测试均未得出显著结果。与最初的预期相反，我们无法得出某些SNP对本文提出的这个有前景的基因座有主要影响的结论。作为扩展我们观察结果的一种方式，我们强调后续关联研究（包括在多个群体中进行重复研究和/或荟萃分析）的重要性。
