Genetic Variations and Risk of Placental Abruption

Tsegaselassie Workalemahu | 2018

Advisor: Daniel A. Enquobahrie

Research Area(s): Genetic Epidemiology, Maternal & Child Health



Placental abruption (PA) is a premature separation of an implanted placenta due to the rupture of the maternal vessels in the decidua basalis prior to delivery of the fetus. PA complicates approximately 1 in 100 pregnancies leading to significant maternal and perinatal morbidity and mortality worldwide. The etiology of PA is not fully known. Suggested pathophysiologic mechanisms that lead to PA include uteroplacental ischemia, underperfusion, chronic hypoxia, infarctions and thrombosis. Further, abnormal mitochondrial biogenesis (MB) and oxidative phosphorylation (OP), important in placental physiology, contribute to PA risk. Emerging evidence from genome-wide (GWAS) and candidate gene association studies, including several by our group, support potential roles of genetic variations, characterized by single nucleotide polymorphisms (SNPs), in PA. However, these studies were small in size and results have been inconsistent across studies. Further, prior investigations did not address maternal-fetal genetic interactions and imprinting effects in PA risk, potential contributors for the missing heritability of PA. The motivation of this dissertation research was to address these limitations.


The research was conducted using data and genomic DNA samples collected from participants of the previously described Peruvian Abruptio Placentae Epidemiology (PAPE) study and the Placental Abruption Genetic Epidemiology (PAGE) study (REF), case-control studies of PA conducted in Lima, Peru. PAPE participants were genotyped using the Illumina Cardio-Metabochip. PAGE participants were genotyped using the Illumina HumanCore-24 BeadChip platform. Genotypes were imputed using the 1000 genomes reference panel, and >4.9 million SNPs that passed quality control were available. The project had three specific aims. Aim #1 was to conduct a GWAS on PAGE participants (507 PA cases and 1,090 controls) and a GWAS meta-analysis in PAPE and PAGE participants (959 PA cases and 1,553 controls) using population stratification-adjusted logistic regression models and fixed-effects meta-analyses using inverse variance weighting, respectively. Aim #2 was to conduct a replication analysis in the PAGE study population examining associations of previously reported (in the PAPE study) weighted genetic risk scores (wGRS) of 11 SNPs in nine MB/OP genes with risk of PA using multivariable-adjusted logistic regression models. Aim #3 was to investigate (in the combined PAPE and PAGE study) maternal-fetal genetic interaction on PA risk for 78 independent (linkage disequilibrium <0.80) SNPs in MB/OP genes using multinomial models, and imprinting (parent-of-origin effect) effect on PA risk for 2713 independent SNPs in 73 imprinted genes using a likelihood ratio test.


We found 174 independent loci suggestively associated with PA in the PAGE GWAS (P-value<5e-5) including rs4148646 and rs2074311 in ABCC8, rs7249210, rs7250184, rs7249100 and rs10401828 in ZNF28, rs11133659 in CTNND2, and rs2074314 and rs35271178 near KCNJ11. Similarly, we found 119 independent loci suggestively associated with PA in the GWAS meta-analysis, including rs76258369 near IRX1, and rs7094759 and rs12264492 in ADAM12. Functional analyses of these genes revealed trophoblast-like cell interaction, as well as networks involved in endocrine system disorders, cardiovascular disease and cellular function. PAGE participants in the 25-50th (score:12.6-13.8), 50-75th (score:13.9-15.0) and 75-100th (score ≥15.1) wGRS percentiles had 1.45-fold (95% confidence interval [CI]:1.04, 2.02), 1.42-fold (95% CI: 1.02-1.98), and 1.75-fold (95% CI: 1.27, 2.42) higher odds of PA, respectively, compared to those in the lowest wGRS percentile (score <12.6; P-for-trend <0.001). We observed maternal-fetal interaction effects for rs12530904 (log-likelihood=-1874.6; P-value=1.2e-04) in CAMK2B, and rs73136795 (log-likelihood=-1644.5; P-value=1.9e-04) in PPARG, both MB genes. We identified 311 SNPs in 35 imprinted genes (including KCNQ1, NPM, and, ATP10A) with parent-of-origin effects on PA risk (with P-value<1.8e-5). Among these, top hits included rs8036892 (P-value=2.3e-15) in ATP10A, rs80203467 (P-value=6.7e-15) and rs12589854 (P-value=1.4e-14) in MEG8, and rs138281088 in SLC22A2 (P-value=1.7e-13).


Using the largest GWAS of PA, to date, we identified several candidate and novel genetic loci and related functions that may play a role in PA risk. We also replicated previous findings of genetic variants in MB/OP that influence PA risk. Lastly, we identified novel maternal-fetal MB gene interactions and imprinting effects of SNPs in imprinted regions in relation to PA risk. Elucidating genetic factors that underlie pathophysiological mechanisms of PA may facilitate prevention and early diagnostic efforts.