Different interaction concepts to detect gene-environment interactions in the context of upper aero-digestive tract cancer

Abstract

Introduction: Tobacco smoking and consumption of alcoholic beverages are established causes of cancers of the upper aero-digestive tract (UADT) and reduced intake of vegetables and fruits are likely causes. Little is known, however, about the role of gene-environment interactions (GEIs) respecting nutritional factors. Material and Methods: The study is embedded in the Alcohol-Related Cancers and Genetic Susceptibility in Europe (ARCAGE) study, a multi-centre case-control study conducted 2002–2005 in 10 European countries (Lagiou, 2009). Dietary habits were assessed using a semi-quantitative food frequency questionnaire. The analysis dataset included 1,395 controls and 1,379 cases. We used approximately 50% of the data for an exploratory screening procedure to generate a list of several potential gene-nutrition interactions. The remaining 50% build an independent dataset, which is used to estimate interaction effects between genes and nutrition variables. Logistic regression models are applied to estimate various interaction effects that are based on odds ratios (OR) like RERI (relative excess risk due to interaction), AP (attributable proportion due to interaction), the synergy index (S) and the multiplicative synergy index (SM). Interaction measures of combined gene-nutrition information will be compared with interaction effects of stratified data. Results: We observe inverse interaction effects between fruit consumption and MTHFR (folate metabolism). Homozygous minor allele carriers (AA genotype) show a significantly increased OR with less than daily fruit consumption (OR=3.8, CI=2.2–3.8) compared to non-carriers with more than daily fruit consumption. RERI=1.99 (CI=-0.1–4.0) indicates that the excess risk due to the exposure combination of the AA-genotype and „daily fruit consumption“ is 1.99 times higher compared to participants being exposed to neither of these two factors. The synergy index shows a significant synergistic departure from additivity (S=3.3, CI=1.2–8.9). Discussion: We demonstrate that different interaction measures lead to different interpretations of „interaction“. This will be discussed in the context of GEIs that could modify the risk of UADT cancers.