|
Previous genetic studies on schizophrenia has been failed to show any confirmatory results, even though it is clear that genetic factors are working substantially on developing schizophrenia. In this review, the reasons of difficulties to reach consistent findings was searched at the aspect of phenotype definition and statistical methods as well as alternative methods possibly to overcome these difficulties. The common statistical problems are multiple testing, population stratification, improper study design or analytic methods, and small sample size. The problems associated with phenotype are vagueness of schizophrenic symptoms and no objective laboratory methods to diagnose schizophrenia. Alternative methods possibly to solve these problems are suggested. For statistical difficulties, pvalue or LOD score corrected for multiple testing, simulation studies, family based control association study, large scale association studies using dense SNPs, nonparametric analysis, collaborative studies, metaanalysis are suggested. For phenotype difficulties, more objective standardized international instrument such as DIGS, clinical subtype, symptom itself as a phenotype, various trials to define diagnostic borderline, quantitative biological endophenotype, and animal model are considered. Recent rapid progress of molecular genetic technology and rapidly accumulating genetic information and development of more proper study design and statistical analysis for schizophrenia will make it possible to find susceptible genes for schizophrenia in nearest future.
|