LAUSR

Assessing gene‐environment interactions requires careful modeling of how the environmental exposure affects the outcome. For interactions with time‐ varying exposures, such as air pollution, one needs to account for cumulative effects. We propose, and validate in simulations, flexible modeling of interactions between genetic factors and time‐varying exposures with cumulative effects. Weighed cumulative exposure (WCE) models cumulative effects through weighed mean of past exposure intensities. The weight function w(t), that quantifies the relative importance of exposures at different times, is estimated with cubic B‐splines. Interactions with genotype or sex, are assessed by comparing fit to data of three alternative WCE models. Model 1 assumes no interactions that is, common w(t) for all subgroups. Model 2 assumes the same shapes of subgroup‐specific w(t)'s but different effect strengths. Model 3 assumes past exposure effects cumulate differently across the subgroups, implying different shapes of w(t). Likelihood ratio tests help identify the model most consistent with the data. Simulation results validate the proposed models and tests. We illustrate real‐life advantages of the WCE modeling by re‐assessing interactions between sex and low‐dose radiation in cancer. Flexible cumulative effects modeling may yield novel insights regarding interactions between genotype and time‐varying exposures. [1] McAllister et al, AJE 2019:753‐61 [2] Cruz‐Fuentes et al, Brain and Behavior 2014: 290‐297 [3] Danieli & Abrahamowicz, SMMR 2019:248‐262 [4] Danieli et al, AJE 2019:1552‐1562. 2 | Racial differences in methylation pathway‐structured predictive models and breast cancer survival Tomi Akinyemiju*, Abby Zhang, April Deveaux, Lauren Wilson, Stella Aslibekyan Duke University, Department of Population Health Sciences, Durham, North Carolina, USA; Duke University, College of Arts and Sciences, Durham, North Carolina, USA; University of Alabama at Birmingham, Department of Epidemiology, Birmingham, Alabama, USA