Interrogating the causal effects of maternal circulating CRP on gestational duration and birth weight

Abstract

AbstractBackgroundC-reactive protein (CRP) is an acute-phase marker of inflammation. Previous epidemiological studies have associated elevated maternal CRP levels during pregnancy with preterm birth risk and lower birth weight. However, the causal relationships behind these observed associations are not clear.Methods and FindingsWe utilized phenotype and genotype data on 10,734 mother/child pairs of European ancestry collected from six birth cohorts. We performed two-sample multivariate Mendelian Randomization (MR) using maternal non-transmitted alleles to interrogate the causal effect of maternal CRP on gestational duration and birth weight. Based on the non-transmitted alleles of 55 CRP associated single nucleotide polymorphisms (SNPs), a unit increase in log-transformed maternal CRP level was associated with a reduction of 1.58 days (95% CI: 0.23, 2.94, P = 0.022) in gestational duration and a reduction of 76.3g (95% CI: 18.2, 134.4, P = 0.010) in birth weight. The magnitude of the association between genetically increased CRP and birth weight was reduced after adjusting for gestational age. Utilizing the effect estimates of CRP-associated SNPs on birth weight from the UK Biobank genome-wide association (GWA) summary results, our results suggested one unit of genetically increased maternal CRP was associated with a decrease of 50.1g (95% CI: 24.7, 75.5, P = 9.6E-5) of birth weight (unadjusted by gestational age). We validated the results using a single genetic variant (rs2794520) near the CRP gene, which is less prone to horizontal pleiotropy bias.ConclusionsIncreased maternal CRP levels appear to causally reduce gestational duration and birth weight. The effect of maternal CRP on birth weight is partially due to its effect on gestational duration. The estimated causal effect sizes are consistent with previous epidemiological reports and this triangulation increases confidence in these results being causal.