New methods for testing sufficient-cause gene-environment interactions were proposed by researchers at National Taiwan University (NTU). The study was published in Am J Epidemiol 2015;182(1):9-16.
To detect gene-environment interactions, a logistic regression model is typically fitted to a set of case-control data, and the focus is on testing of the cross-product terms (gene × environment) in the model. But this is only for the sake of ‘multiplicative interactions’, which unfortunately, even if they are statistically significant cannot be interpreted as genes and environmental factors being interacting mechanistically in causing the disease. Based on the sufficient-component-cause model, Professor Wen-Chung Lee of Institute of Epidemiology and Preventive Medicine in the College of Public Health put forward a general approach to testing for mechanistic gene-environment interactions in case-control studies. The proposed tests can be tailored to detect a particular type of interaction with greater sensitivity. These tests include testing for autosomal dominant, autosomal recessive, and gene-dosage interactions. The tests can also detect trend interactions (e.g., a larger gene-environment interaction with a higher level of environmental exposure) and threshold interactions (e.g., gene-environment interaction occurs only when environmental exposure reaches a certain threshold level).
Professor Lee also developed easy-to-use SAS code for implementing the methods in Web Materials accompanying the published paper.