Placental protein levels in maternal serum are associated with adverse pregnancy outcomes in nulliparous patients

Abstract

Background: The Eunice Kennedy Shriver National Institute of Child Health and Human Development Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be was established to investigate the underlying causes and pathophysiological pathways associated with adverse pregnancy outcomes in nulliparous gravidas.

Objective: This study aimed to study placental physiology and identify novel biomarkers concerning adverse pregnancy outcomes, including preterm birth (medically indicated and spontaneous), preeclampsia, small-for-gestational-age neonates, and stillbirth. We measured levels of placental proteins in the maternal circulation in the first 2 trimesters of pregnancy.

Study design: Maternal serum samples were collected at 2 study visits (6-13 weeks and 16-21 weeks), and levels of 9 analytes were measured. The analytes we measured were vascular endothelial growth factor, placental growth factor, endoglin, soluble fms-like tyrosine kinase-1, A disintegrin and metalloproteinase domain-containing protein 12, pregnancy-associated plasma protein A, free beta-human chorionic gonadotropin, inhibin A, and alpha-fetoprotein. The primary outcome was preterm birth between 20 0/7 and 36 6/7 weeks of gestation. The secondary outcomes were spontaneous preterm births, medically indicated preterm births, preeclampsia, small-for-gestational-age neonates, and stillbirth.

Results: A total of 10,038 eligible gravidas were enrolled in the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be cohort, from which a nested case-control study was performed comparing 800 cases with preterm birth (466 spontaneous preterm births, 330 medically indicated preterm births, and 4 unclassified preterm births), 568 with preeclampsia, 406 with small-for-gestational-age birth, and 49 with stillbirth with 911 controls who delivered at term without complications. Although levels of each analyte generally differed between cases and controls at 1 or 2 visits, the odds ratios revealed a <2-fold difference between cases and controls in all comparisons. Receiver operating characteristic curves, generated to determine the relationship between analyte levels and preterm birth and the other adverse pregnancy outcomes, resulted in areas under the receiver operating characteristic curves that were relatively low (range, 0.50-0.64) for each analyte. Logistic regression modeling demonstrated that areas under the receiver operating characteristic curves for predicting adverse pregnancy outcomes were greater using baseline clinical characteristics and combinations of analytes than baseline characteristics alone, but areas under the receiver operating characteristic curves remained relatively low for each outcome (range, 0.65-0.78).

Conclusion: We have found significant associations between maternal serum levels of analytes evaluated early in pregnancy and subsequent adverse pregnancy outcomes in nulliparous gravidas. However, the test characteristics for these analytes do not support their use as clinical biomarkers to predict adverse pregnancy outcomes, either alone or in combination with maternal clinical characteristics.

Keywords: A disintegrin and metalloproteinase domain-containing protein 12; endoglin; placental growth factor; preeclampsia; pregnancy-associated plasma protein A; preterm birth; small for gestational age; soluble fms-like tyrosine kinase-1; stillbirth; vascular endothelial growth factor.

Figure 1.. Flow chart of nuMoM2b participants who were included in the nested case-control study.

Levels of analytes in maternal serum samples were compared between term delivery controls (green box) and cases with adverse pregnancy outcomes (APO, gray boxes). SGA = small for gestational age; SPTB = spontaneous preterm birth; PTB = preterm birth; preeclampsia + GHTN/PTB = preeclampsia cases (all gestational ages) plus preterm births complicated by gestational hypertension

Figure 2.. Areas under receiver operating characteristic curves (95% confidence intervals) for prediction of adverse pregnancy outcome using baseline characteristics and selected placental analytes by study visit.

Analyses are based on logistic regression models for cases of adverse pregnancy outcome versus controls from a random sub-cohort. Baseline characteristics included: age, race/ethnicity, BMI, smoking status, pregestational diabetes, and chronic hypertension. A lasso cross-validation method was used for selection of placental analytes for inclusion in the logistic regression models, allowing for interaction terms with gestational age, and offset for baseline characteristics. The placental analytes were transformed prior to implementation of the cross-validation. Power transformations were used to normalize the data, followed by regression on gestational age to generate studentized residuals.