A Novel Genetic Score Approach Using Instruments to Investigate Interactions between Pathways and Environment: Application to Air Pollution

PLOS ONE, April 2014

24755831

Marie-Abele Bind, Brent Coull, Helen Suh, Robert Wright, Andrea Baccarelli, Pantel Vokonas, Joel Schwartz

Air pollution has been associated with increased systemic inflammation markers. We developed a new pathway analysis approach to investigate whether gene variants within relevant pathways (oxidative stress, endothelial function, and metal processing) modified the association between particulate air pollution and fibrinogen, C-reactive protein (CRP), intercellular adhesion molecule-1 (ICAM-1), and vascular cell adhesion molecule-1 (VCAM-1). Our study population consisted of 822 elderly participants of the Normative Aging Study (1999-2011). To investigate the role of biological mechanisms and to reduce the number of comparisons in the analysis, we created pathway-specific scores using gene variants related to each pathway. To select the most appropriate gene variants, we used the least absolute shrinkage and selection operator (Lasso) to relate independent outcomes representative of each pathway (8-hydroxydeoxyguanosine for oxidative stress, augmentation index for endothelial function, and patella lead for metal processing) to gene variants. A high genetic score corresponds to a higher allelic risk profile. We fit mixed-effects models to examine modification by the genetic score of the weekly air pollution association with the outcome. Among participants with higher genetic scores within the oxidative stress pathway, we observed significant associations between particle number and fibrinogen, while we did not find any association among participants with lower scores (p(interaction) = 0.04). Compared to individuals with low genetic scores of metal processing gene variants, participants with higher scores had greater effects of particle number on fibrinogen (p(interaction) = 0.12), CRP (p(interaction) = 0.02), and ICAM-1 (pinteraction = 0.08). This two-stage penalization method is easy to implement and can be used for large-scale genetic applications.