RESUMO
OBJECTIVE: To establish if low maternal selenium (Se) was associated with sPTB in women with recurrent sPTB and identify genetic link with maternal Se levels. DESIGN: Nested case-control study. SETTING: Tertiary Maternity Hospital. POPULATION: Plasma and whole blood from pregnant women with history of early sPTB/PPROM < 34+0 and European ancestry were obtained at 20 weeks (range 15-24 weeks). 'Cases' were recurrent PTB/PPROM < 34+0 weeks and term (≥37+0) deliveries were classified as 'high-risk controls.' Women with previous term births and index birth ≥ 39 weeks were 'low-risk controls'. METHODS: Maternal plasma Se measured by ICP-MS was used as a continuous phenotype in a GWAS analysis. Se was added to a logistic regression model using PTB predictor variables. MAIN OUTCOME MEASURES: Maternal Se concentration, recurrent early sPTB/PPROM. RESULTS: 53/177 high-risk women had a recurrent sPTB/PPROM < 34+0weeks and were 2.7 times more likely to have a Se level < 83.3 ppm at 20weeks of pregnancy compared with low-risk term controls (n = 179), (RR 2.7, 95%CI 1.5-4.8; p = .001). One SNP from a non-coding region (FOXN3 intron variant, rs55793422) reached genome-wide significance level (p = 3.73E-08). Targeted analysis of Se gene variant did not show difference between preterm and term births. (χ2 test, OR = 0.95; 95%CI = 0.59-1.56; p = 0.82). When Se levels were added to a clinical prediction model, only an additional 5% of cases (n = 3) and 0.6% (n = 1) of controls were correctly identified. CONCLUSIONS: Low plasma Se is associated with sPTB risk but is not sufficiently predictive at individual patient level. We did not find a genetic association between maternal Se levels and Se-related genes.