Lobach Iryna, Kim Inyoung, Alekseyenko Alexander, Lobach Siarhei, Zhang Li
Department of Epidemiology and Biostatistics, University of California, San Francisco, San Francisco, CA, United States.
Department of Statistics, Virginia Tech University, Blacksburg, VA, United States.
Front Genet. 2019 Oct 9;10:886. doi: 10.3389/fgene.2019.00886. eCollection 2019.
Case-control genetic association studies are often used to examine the role of the genetic basis in complex diseases, such as cancer and neurodegenerative diseases. The role of the genetic basis might vary by nongenetic (environmental) measures, what is traditionally defined as gene-environment interactions (G×E). A commonly overlooked complication is that the set of clinically diagnosed cases might be contaminated by a subset with a pathologic state that presents with the same symptoms as the pathologic state of interest. The genetic basis of the pathologic state of interest might differ from that of the nuisance pathologic state. Often, frequencies of the pathologically defined states within the clinically diagnosed set of cases vary by the environment. We derive a simple and general approximation to bias in G×E parameter estimates when the presence of the nuisance pathologic state is ignored. We then perform extensive simulation studies to show that ignoring the presence of the nuisance pathologic state can result in substantial bias in G×E estimates and that the approximation we derived is reasonably accurate in finite samples. We demonstrate the applicability of the proposed approximation in a study of Alzheimer's disease.
病例对照基因关联研究常用于检验遗传基础在复杂疾病（如癌症和神经退行性疾病）中的作用。遗传基础的作用可能因非遗传（环境）因素而有所不同，这就是传统上定义的基因-环境相互作用（G×E）。一个常被忽视的复杂情况是，临床诊断的病例组可能被一部分具有与感兴趣的病理状态相同症状的病理状态所污染。感兴趣的病理状态的遗传基础可能与干扰性病理状态的遗传基础不同。通常，临床诊断病例组中病理定义状态的频率会因环境而异。当忽略干扰性病理状态的存在时，我们推导出了一个简单通用的近似值，用于估计G×E参数中的偏差。然后，我们进行了广泛的模拟研究，以表明忽略干扰性病理状态的存在会导致G×E估计值出现实质性偏差，并且我们推导的近似值在有限样本中相当准确。我们在一项阿尔茨海默病研究中证明了所提出近似值的适用性。
