Adiponectin and Glucocorticoids Modulate Risk for Preterm Birth: The Healthy Start Study

Abstract

Context: Adiponectin is a potent uterine tocolytic that decreases with gestational age, suggesting it could be a maternal metabolic quiescence factor. Maternal stress can influence preterm birth risk, and adiponectin levels may be stress responsive.

Objective: We characterized associations between adiponectin and glucocorticoids with preterm birth and modeled their predictive utility. We hypothesized maternal plasma adiponectin and cortisol are inversely related and lower adiponectin and higher cortisol associate with preterm birth.

Methods: We performed a nested case-control study using biobanked fasting maternal plasma. We included low-risk singleton pregnancies, and matched 1:3 (16 preterm, 46 term). We quantified high molecular weight (HMW), low molecular weight (LMW), and total adiponectin using an enzyme-linked immunosorbent assay. We validated a high-performance liquid chromatography-tandem mass spectrometry serum assay for use in plasma, to simultaneously measure cortisol, cortisone, and 5 related steroid hormones. We used linear/logistic regression to compare group means and machine learning for predictive modeling.

Results: The preterm group had lower mean LMW adiponectin (3.07 μg/mL vs 3.81 μg/mL at 15 weeks (w) 0 days (d), P = .045) and higher mean cortisone (34.4 ng/mL vs 29.0 ng/mL at 15w0d, P = .031). The preterm group had lower cortisol to cortisone and lower LMW adiponectin to cortisol ratios. We found HMW adiponectin, cortisol to cortisone ratio, cortisone, maternal height, age, and prepregnancy body mass index most strongly predicted preterm birth (area under the receiver operator curve = 0.8167). In secondary analyses, we assessed biomarker associations with maternal self-reported psychosocial stress. Lower perceived stress was associated with a steeper change in cortisone in the term group.

Conclusion: Overall, metabolic and stress biomarkers are associated with preterm birth in this healthy cohort. We identify a possible mechanistic link between maternal stress and metabolism for pregnancy maintenance.

Keywords: Cohen's Perceived Stress Scale; Edinburgh Perinatal/Postnatal Depression Scale; adipokine; neuroactive steroid measurement; neuroendocrine; steroid hormone measurement.

Figure 1.

Subject selection flow diagram. Subject selection based on Lifetime Epidemiology of Adiposity and Diabetes (LEAD) Center's Healthy Start Study Pre-Birth Cohort enrollment and case-control assignments.

Figure 2.

Estimated mean analyte concentrations in preterm and term groups. Asterisk indicates difference was statistically significant between groups.

Figure 3.

Preterm birth prediction model. (A) Predictor variables in order of relative importance. (B) Impact of SHAP values on prediction model. A positive SHAP value indicates increasing odds of preterm delivery (to the right of the center vertical line). Negative SHAP value indicates decreasing odds of preterm delivery (to the left of the center vertical line). (C) ROC (receiver operator curve) plot. ROC for the xgboost model based on the testing partition of the dataset.

Figure 4.

Change in self-reported stress scores and cortisone concentrations between first and second blood sample. The term group is in the box on the left and the preterm group is in the box on the right.