The Promise and Limitations of Genome-wide Association Studies

AS OF MAY 30, 2012, THE CATALOG OF PUBLISHED GEnome-wide association studies (GWAS) lists an impressive 1269 GWAS, covering a broad spectrum of conditions including Alzheimer disease, breast cancer, and human immunodeficiency virus susceptibility. The catalog also contains studies on common traits such as height and freckles, as well as responses to drugs for various medical conditions. It is difficult to discuss GWAS without sounding megalomaniacal. Considerably more than 1000 published GWAS, replication studies, and metaanalyses have been conducted in an unprecedented global research effort in only 7 years. Most GWAS have included hundreds or even thousands of patients and controls, have hundreds of thousands of participants worldwide, and although genotyping costs have plummeted in recent years, hundreds of millions of research dollars have been spent on GWAS since 2005. A 2007 fact sheet released by the National Human Genome Research Institute, in the early days of GWAS, raised expectations that personalized medicine, including individual risk prediction, disease prevention, and specific treatment, was just around the corner. “With the first GWAS published in 2005, . . . health professionals will be able to use such tools to provide patients with individualized information about their risks of developing certain diseases . . . to tailor prevention programs to each person’s unique genetic makeup . . . to select the treatments most likely to be effective and least likely to cause adverse reactions. . . .” Has the promise of GWAS been realized 5 years later? Although there is no simple answer to this question, it is helpful to consider 3 important and closely intertwined features of GWAS, ie, sample size, characterization of probands (samples), and effect size.