Maternal smoking during pregnancy is the most important modifiable risk factor for reduced birth size and reduces infant birth weight in a dose related manner (US Dept. of Health and Human Services, 2004). The precise biological routes by which exposure to smoking during pregnancy directly or indirectly influences fetal growth remain unknown. Few unreplicated candidate gene studies have suggested that the offspring's genotype can influence the sensitivity to the effects of maternal smoking (Infante-Rivard, et al., 2006; Price, et al., 2010). Genome-wide interaction studies are warranted and only meta-analysis across large cohorts can provide the sample size necessary for adequate statistical power. It will be vital to include as many cohorts as possible.

Heterogeneity in the exposure variable compromises meta-analytic studies of genotype-environment interaction (Palla et al., 2010). Therefore, we first focus on maternal smoking as a binary exposure as it can be measured reliably and consistently across studies. Dosage and timing are likely to be important modifiers and may be less well recorded. Thus, we will perform secondary analyses using these variables in cohorts where such information is available.

We propose a genome-wide meta-analysis of interactions between child genotype and maternal smoking on singleton infant birth weight adjusted for gestational age. The investigation will focus on term deliveries. Gestational age and gender are desirable covariates in the analysis.