Akinori Miyashita, Asako Koike, Gyungah Jun, Li-San Wang, Satoshi Takahashi, Etsuro Matsubara, Takeshi Kawarabayashi, Mikio Shoji, Naoki Tomita, Hiroyuki Arai, Takashi Asada, Yasuo Harigaya, Masaki Ikeda, Masakuni Amari, Haruo Hanyu, Susumu Higuchi, Takeshi Ikeuchi, Masatoyo Nishizawa, Masaichi Suga, Yasuhiro Kawase, Hiroyasu Akatsu, Kenji Kosaka, Takayuki Yamamoto, Masaki Imagawa, Tsuyoshi Hamaguchi, Masahito Yamada, Takashi Moriaha, Masatoshi Takeda, Takeo Takao, Kenji Nakata, Yoshikatsu Fujisawa, Ken Sasaki, Ken Watanabe, Kenji Nakashima, Katsuya Urakami, Terumi Ooya, Mitsuo Takahashi, Takefumi Yuzuriha, Kayoko Serikawa, Seishi Yoshimoto, Ryuji Nakagawa, Jong-Won Kim, Chang-Seok Ki, Hong-Hee Won, Duk L. Na, Sang Won Seo, Inhee Mook-Jung, Peter St. George-Hyslop, Richard Mayeux, Jonathan L. Haines, Margaret A. Pericak-Vance, Makiko Yoshida, Nao Nishida, Katsushi Tokunaga, Ken Yamamoto, Shoji Tsuji, Ichiro Kanazawa, Yasuo Ihara, Gerard D. Schellenberg, Lindsay A. Farrer, Ryozo Kuwano

To discover susceptibility genes of late-onset Alzheimer's disease (LOAD), we conducted a 3-stage genome-wide association study (GWAS) using three populations: Japanese from the Japanese Genetic Consortium for Alzheimer Disease (JGSCAD), Koreans, and Caucasians from the Alzheimer Disease Genetic Consortium (ADGC). In Stage 1, we evaluated data for 5,877,918 genotyped and imputed SNPs in Japanese cases (n = 1,008) and controls (n = 1,016). Genome-wide significance was observed with 12 SNPs in the APOE region. Seven SNPs from other distinct regions with p-values <2×10(-5) were genotyped in a second Japanese sample (885 cases, 985 controls), and evidence of association was confirmed for one SORL1 SNP (rs3781834, P = 7.33×10(-7) in the combined sample). Subsequent analysis combining results for several SORL1 SNPs in the Japanese, Korean (339 cases, 1,129 controls) and Caucasians (11,840 AD cases, 10,931 controls) revealed genome wide significance with rs11218343 (P = 1.77×10(-9)) and rs3781834 (P = 1.04×10(-8)). SNPs in previously established AD loci in Caucasians showed strong evidence of association in Japanese including rs3851179 near PICALM (P = 1.71×10(-5)) and rs744373 near BIN1 (P = 1.39×10(-4)). The associated allele for each of these SNPs was the same as in Caucasians. These data demonstrate for the first time genome-wide significance of LOAD with SORL1 and confirm the role of other known loci for LOAD in Japanese. Our study highlights the importance of examining associations in multiple ethnic populations.