Preterm birth, which affects approximately 12% of US births (1), has significant medical and societal implications, including increased risk of neonatal death and both acute and chronic medical and neurocognitive disease (2). Although the earliest preterm births have the most severe outcomes (3), late preterm births comprise the majority (74.1%) (4) of preterm births and also are associated with increased risk of death and respiratory difficulties (5). Having had a previous preterm birth is one of the strongest maternal risk factors for a subsequent preterm birth. In addition to environmental exposures that might persist between pregnancies, this risk could reflect innate susceptibility factors from maternal genes (6–8). Although several candidate pathways have been identified (9), inflammatory pathways have been a particular focus because pregnancy is a state of altered immune function (10–13). Pregnancy appears to involve a global shift toward defensive immunity (14–17). In particular, T cells and natural killer (NK) cells decrease in both number and function over the course of pregnancy (14), and both cell types have been linked to initiation of labor before term (18, 19). Candidate genes involved in inflammation have been examined in several diverse populations (20–38), although many of the studies have had small case groups, and few have addressed population stratification. Given differences in allelic frequencies in genes related to inflammatory cytokines (39), as well as differences in the risk of preterm birth (26–29, 40) among whites and African Americans, genetic ancestry is of particular importance. Existing studies with larger case groups (28–32, 41) have had few single-nucleotide polymorphisms (SNPs) per gene and often have not included regulatory regions that flank genes. Using the Pregnancy, Infection, and Nutrition Cohort, we expanded candidate genes to include lesser-studied genes in the critical T cell and NK cell pathways and deepened the per-gene SNP coverage while addressing issues of population stratification. We additionally improved on previous studies by incorporating novel statistical methodology to identify gene-based associations, and we used inverse probability of selection modeling to account for any differences between the genotyped and parent cohort. In this article, we describe our findings for polymorphisms in 30 candidate genes and the risk of preterm birth.
