Can Placental Corticotropin-Releasing Hormone Inform Timing of Antenatal Corticosteroid Administration?

Abstract

Context

Antenatal corticosteroids are commonly administered to pregnant women at risk for delivering between 23 and 34 gestational weeks; they provide crucial benefits to fetal lung maturation and reduce risk for neonatal morbidity and mortality. Corticosteroids are maximally efficacious for lung maturation when administered within 2 to 7 days of delivery. Accurately identifying the timing of preterm delivery is thus critical to ensure that antenatal corticosteroids are administered within a week of delivery and to avoid unnecessary administration to women who will deliver at term. A plausible biomarker for predicting time of delivery is placental corticotropin-releasing hormone (pCRH).

Objective

To assess whether pCRH concentrations predict time to delivery and specifically which women will deliver within a week of treatment.

Design

pCRH concentrations were evaluated before administration of the corticosteroid betamethasone, and timing of delivery was recorded.

Participants

A total of 121 women with singleton pregnancies who were prescribed betamethasone.

Results

Elevated pCRH concentrations were associated with a shorter time from treatment to delivery. Receiver-operating characteristic curves revealed that pCRH may improve the precision of predicting preterm delivery.

Conclusions

In the current sample, pCRH concentrations predicted the likelihood of delivering within 1 week of corticosteroid treatment. Current findings suggest that pCRH may be a diagnostic indicator of impending preterm delivery. Increasing the precision in predicting time to delivery could inform when to administer antenatal corticosteroids, thus maximizing benefits and reducing the likelihood of exposing fetuses who will be delivered at term.

Preterm birth (i.e., delivery before 37 weeks' gestation) is one of the leading causes of perinatal morbidity and mortality, as 12% to 13% of women deliver preterm each year within the United States (1). Infants born preterm are at heightened risk for respiratory distress syndrome and other birth complications. The administration of corticosteroids to women at risk for preterm delivery has been the frontline of treatment to promote fetal lung maturation and ultimately decrease morbidity and mortality among preterm infants (2). According to guidelines established by the American College of Obstetricians and Gynecologists (ACOG), corticosteroid treatment is efficacious for fetal lung maturation when administered antenatally to women who will deliver between 24 0/7 and 33 6/7 weeks (2, 3); recent data indicate a benefit through 36 6/7 weeks (4). The benefits of antenatal corticosteroids are maximized when delivery occurs 2 to 7 days after the initial dose, with respiratory benefits dissipating after 7 days (3, 5). Corticosteroid treatment is also associated with reductions in neonatal mortality when delivery occurs within 24 hours of administration (5); thus, treatment is recommended when women are at risk for delivering within 1 week (2). Treatment of infants born preterm has clear benefits, including reduced risk for respiratory distress syndrome, intraventricular hemorrhage, morbidity, and mortality (3, 6). However, in addition to these benefits, antenatal corticosteroids may have long-term adverse developmental consequences (7, 8), especially for infants who were at risk for preterm birth but proceeded to be delivered at term. Although negative effects of antenatal corticosteroids have not been observed in all studies (2, 8–10), administration of corticosteroids has been associated with fetal growth restriction, altered brain development, increased emotional reactivity, and persisting differences in stress physiology and hypothalamic-pituitary-adrenal axis dysregulation in human (7, 8, 11–16) and animal (17–26) studies.

Accurately identifying the risk for impending preterm delivery (i.e., delivery within a week) is critical for two reasons: first, to ensure that the timing of antenatal corticosteroid administration maximizes benefit to the fetus; and second, to avoid administering corticosteroids to women who will go on to deliver at term. As argued by Kamath-Rayne et al. (27), improvement in prediction of the timing of preterm birth is needed so that antenatal corticosteroids can be administered within the ideal time frame of 1 week before delivery. We propose that placental CRH (pCRH), a peptide postulated to lie on the pathway to preterm birth, could inform the timing of antenatal corticosteroid treatment (28).

CRH, a 41–amino acid neuropeptide, is synthesized primarily in the paraventricular nucleus of the hypothalamus and has a central role in regulating pituitary and adrenal function and the physiological response to stress (29–31). During pregnancy, CRH is also synthesized and released from the placenta (pCRH) and increases dramatically throughout pregnancy, rising 25-fold through 31 weeks of gestation (31–33). pCRH is identical in structure and activity to hypothalamic CRH. Concentrations of CRH in maternal circulation are almost exclusively of placental origin because the quantity of hypothalamic CRH in maternal circulation is minute and rapidly degrades (34, 35). In the mother, pCRH binds to receptors in the pituitary and the adrenal glands, promoting the production of cortisol and dehydroepiandrosterone sulfate and further release of pCRH from the placenta. pCRH also binds to receptors in the myometrium, stimulating contractile and relaxatory mechanisms (29, 34, 36, 37). In addition to effects on the mother, pCRH is released into the fetal compartment, binding to receptors in the pituitary and fetal zone of the adrenal, promoting the synthesis of cortisol and maturation of fetal lungs. Maturation of the fetal lungs is associated with increases in surfactant protein A, which may stimulate myometrial contraction. The increase of pCRH, especially over the latter part of gestation, may therefore act upon various mechanisms to promote the onset of labor. pCRH has thus been implicated in the pathways underlying the timing of parturition (28, 38).

Prior work has established that pCRH predicts preterm birth in both low-risk pregnancies (28, 34, 38, 39) and pregnancies at high risk for preterm delivery (40). However, only two known studies have explored whether pCRH can predict delivery within days of assessment (e.g., 24 or 48 hours) (41, 42). Korebrits et al. (41) reported that elevated pCRH levels predict delivery within 24 hours; Hill et al. (42) reported that after 28 gestational weeks, elevated pCRH predicts delivery within 48 hours. Research, however, is needed to evaluate whether pCRH concentrations assessed before antenatal corticosteroid administration can inform the timing of treatment within the optimal window (i.e., whether delivery will occur within a week).

The current study evaluates whether pCRH concentrations before corticosteroid treatment can identify women at risk for impending preterm delivery (i.e., delivery within a week). We assess the specificity and sensitivity of pCRH in predicting timing of delivery among pregnant women prescribed betamethasone, the most commonly used antenatal corticosteroid (43).

Participants and Methods

Study overview

We recruited women at risk for preterm delivery who were prescribed betamethasone between 23 0/7 and 34 0/7 gestational weeks. pCRH levels were evaluated before betamethasone administration, and date of delivery was determined via medical record abstraction. The institutional review board reviewed and approved the study protocols. Each mother provided written and informed consent before study enrollment.

Participants

Participants included 121 women with singleton pregnancies at risk for preterm delivery who were prescribed betamethasone, from the University of California Irvine Medical Center. Exclusion criteria included a known hypothalamic-pituitary-adrenal axis or endocrine disorder, self-reported maternal substance use, prior corticosteroid exposure during this pregnancy, in vitro fertilization, and major fetal or chromosomal anomalies. During the recruitment period (from 2011 to 2014), 43.8% of women treated with betamethasone failed to meet eligibility criteria and were not recruited. Among the eligible women who were approached for participation 18.8% declined for various reasons, the most common of which were study burden, feeling overwhelmed, and pregnancy-related worries.

The decision to administer betamethasone rested entirely on the clinical judgment of the attending obstetrician and was made independent of study participation. As per standardized hospital protocol, the attending obstetrician evaluated preterm labor at <34 weeks, administered temporizing measures, such as parenteral magnesium sulfate or oral nifedipine; and determined whether betamethasone treatment was needed. If deemed necessary, betamethasone was administered in two doses (12 mg intramuscularly 24 hours apart). Consistent with standard of care, the first dose of betamethasone was given to participants between 23 and 34 weeks’ gestation (mean, 28.4; SD, 0.3 weeks). On average, delivery occurred 42 days after the first dose of betamethasone administration (SD, 31.0 days; range, 0 to 104 days).

Assessment of pCRH

Maternal blood samples were collected for pCRH on average 36 minutes before betamethasone treatment (range, 1 minute to 4 hours and 16 minutes). A total of 50 μL of protease inhibitor, Aprotinin (ThermoFisher Scientific), was added to the plasma tubes and refrigerated. After 30 minutes, tubes were centrifuged, processed, and frozen at −70°C. Plasma pCRH assays were completed by Dr. Roger Smith's laboratory at the University of Newcastle, Australia. The pCRH concentrations (pg/mL) were analyzed by radioimmunoassay as previously described (29). Extraction recovery was 82.5%. The data were not corrected for extraction recoveries. Inter- and intra-assay coefficients of variance were 9.8% and 7.6%, respectively. Data reduction for the radioimmunoassay was performed with a computer-assisted logistics program.

Analyses were done with both raw pCRH data and log-transformed data because of skewness. Statistical significance was maintained whether untransformed or transformed data were used. Raw data are thus presented here in nanomoles per liter (nmol/L) to improve interpretability.

Sociodemographic and medical characteristics

Sociodemographic characteristics, including marital and cohabitation status, education, and household income, were determined at the time of study entry by maternal interview. Family income was assessed based on an income-to-needs ratio, calculated by dividing total household income by the poverty threshold at the time of assessment, specified by the U.S. Census Bureau.

Medical records were reviewed to determine obstetric and neonatal medical characteristics. Body mass index (BMI, expressed as kg/m²) was calculated based on maternal height and weight in the third trimester, or within 1 month of delivery if the mother delivered before the third trimester. Primary reason for betamethasone treatment was grouped into three categories: inflammation, placental issues, and utero-placental insufficiency. The inflammation group included women with preterm labor, short cervix, preterm premature rupture of membranes, chorioamnionitis, or a positive fetal fibronectin test result. The placental issues group included women with placenta previa, abruption, accreta, or oligohydramnios. The utero-placental insufficiency group included women with pre-eclampsia, chronic hypertension, gestational hypertension, pregnancy-induced hypertension, intrauterine growth restriction, or abnormal end-diastolic flow results from umbilical artery Doppler velocimetry.

Data analyses

As expected, pCRH concentrations were positively correlated with gestational age (GA) at time of assessment (r = 0.65; P < 0.001) (37). Thus, all analyses with pCRH covaried with GA at time of assessment. The following variables were evaluated as potential covariates and were included in the models if they were significantly associated (P < 0.10) with both predictor (pCRH) and outcome (time from betamethasone administration to delivery) after adjustment for GA at assessment: maternal age, cohabitation, ethnicity, education level, income-to-needs ratio, living below the federal poverty line, pregnancy history, and late-pregnancy BMI. None of these additional variables met the criteria for inclusion as a covariate.

We first used partial correlations by using IBM SPSS Statistics for Windows, version 23 (IBM Corp., Armonk, NY) to evaluate whether pCRH concentrations were associated with time from treatment to delivery and GA at birth. Linear regression was performed by using SPSS software as well to evaluate whether the reason for betamethasone administration accounted for study findings (i.e., reason for betamethasone was included as a covariate in a linear regression model with pCRH concentrations as the predictor and time from treatment to delivery as the outcome variable) or moderated the association between pCRH and time from treatment to delivery (i.e., an interaction term was created between reason for betamethasone treatment and pCRH concentrations, which was then added to the linear regression model). Next, Cox proportional hazard models were fitted to investigate whether pCRH concentrations predicted risk for delivering over time. Cox proportional hazard models were performed by using SAS software for Unix (SAS Institute Inc., Cary, NC).

Guidelines from ACOG during the study period recommend that corticosteroids be administered to pregnant women who are between 24 0/7 and 33 6/7 weeks of gestation and are at risk for delivering within a week (2). Thus, the sample was divided into two groups: women who delivered within a week of treatment (n = 25) and those who delivered after a week (n = 96). Computed receiver-operating characteristic (ROC) curves assessed the sensitivity and specificity of pCRH concentrations in predicting whether women would deliver within a week. These models were performed by using SAS software. The Youden index [sensitivity + (specificity − 1)] was computed to obtain a cutoff score that optimizes differentiating ability when equal weight is given to sensitivity and specificity. Because optimizing sensitivity is a priority for clinical practice, we also identified an alternative cutoff that weighed sensitivity over specificity.

Results

Sample characteristics

Demographic characteristics of study participants are summarized in Table 1. Women were, on average, 30 years of age at recruitment (SD, 6.6; range, 18 to 46 years) and were primarily Latina (63.6%). The majority were cohabitating with a partner (81%). The range of the income-to-needs ratio was 0 to 33.19. Approximately one in three participants were living below the federal poverty line (income-to-needs ratio < 1.00).

Table 1.

Maternal Demographic and Medical Characteristics

Characteristic	Participants (n = 121)
Age at enrollment, y	30.2 (6.6) [18.6, 46.9]
Parity (% primiparous)	29 (24.0)
Ethnicity
Non-Latina white	27 (22.3)
Latina	77 (63.6)
Other	17 (14.0)
Gestational age at assessment (wk)	28.4 (3.2) [23.0, 33.9]
BMI, kg/m2a	31.3 (0.6) [19.1, 46.5]
Married or cohabitating	98 (81.0)
Duration of education, y	13.37 (2.89) [2.00, 21.00]
Income-to-needs ratiob
<1	36 (32.4)
1–2	33 (29.7)
2–4	17 (15.3)
>4	25 (22.5)
Indication for betamethasone treatment
Placental abnormalities	27 (22.3)
Inflammation	77 (63.6)
Utero-placental insufficiency	17 (14.0)
Obstetric risk factors
Preeclampsia	11 (9.1)
HELLP syndrome	1 (0.8)
Preterm premature rupture of membranes	27 (22.3)
Gestational diabetes	14 (11.6)
Diabetes	6 (5.0)
pCRH, pmol/L	69.9 (102.0) [0.25, 600.0]

Continuous variables are expressed as mean (SD) [range]; categorical variables are expressed as n, %. HELLP, hemolysis, elevated liver enzymes, and low platelet count.

^a

Missing for four participants.

^b

Missing for 10 participants.

Time from treatment to delivery

All 121 women were prescribed betamethasone at the independent discretion of the attending physician and in adherence with current ACOG guidelines. Risk status was thus determined entirely by the physician and not in relation to the current study protocol. Of these women, 25 (20.7%) delivered within a week of treatment. Additionally, 69 (57%) women went on to deliver preterm and 52 (43%) delivered at term (Table 2).

Table 2.

Neonatal Medical Characteristics

Characteristic	Participants (n = 121)
Gestational age at birth, wk	34.3 (4.7) [24.3, 40.7]
Preterm (<37 wk)	69 (57.0)
Birth weight, g	2336.20 (1008.33) [570, 4480]
Apgar score at 5 mina	8.00 (1, 10)
Sex (% male)	69 (57.0)

Continuous variables are expressed as mean (SD) [range]; categorical variables are expressed as n, %.

^a

Apgar score is presented as median (range).

pCRH

Women with elevated pCRH concentrations were more likely to delivery earlier (r = −0.35; P < 0.001) and were more likely to have a shorter time between treatment and delivery (r = −0.35; P < 0.001) [see Supplemental Figs. 1A and 1B in the online data repository for correlation plots (44)]. Further, pCRH concentrations were higher among women who subsequently delivered preterm (mean, 86.87 pmol/L; SD, 127.34 pmol/L) than among women who delivered full term (mean, 47.28 pmol/L; SD, 43.90 pmol/L; F = 9.51; P = 0.003). The pCRH concentrations were also higher among women who delivered within the week (mean, 170.89 pmol/L; SD, 178.45 pmol/L) than in those who did not (mean, 44.86 pmol/L; SD, 46.70 pmol/L; F = 42.13; P < 0.001). For the subset of women who remained pregnant for at least a week after treatment, pCRH showed the expected increase with advancing gestation [t(89) = −6.194; P < 0.001]. However, pCRH levels were highly correlated across this 1-week period (r = 0.971; P < 0.001).

Reason for betamethasone administration (i.e., inflammation, placental issues, utero-placental insufficiency) did not account for study findings; pCRH was still a significant predictor of time from treatment to delivery after covarying reason for betamethasone administration [t (117) = −1.32; P = 0.189]. Further, reason for betamethasone administration did not moderate the association between pCRH concentrations and time from treatment to delivery (β = −0.335; P > 0.250).

Cox hazard modeling revealed that higher concentrations of pCRH were associated with earlier delivery (β = 0.0058; SEM, 0.001; P < 0.0001; 95% CI, 0.0036 to 0.0079). Data were analyzed continuously but for illustrative purposes are presented as the top and bottom quartiles of pCRH concentrations in Fig. 1. For every increase of 10 pmol/L of pCRH, the risk for delivery (i.e., delivery within a week) was 5.8% higher.

Figure 1.

pCRH concentrations before betamethasone treatment predict time to birth. The Kaplan-Meier plot shows the probability of remaining pregnant, based on pCRH concentrations before betamethasone administration. Although all analyses were run continuously, for illustrative purposes we present the top and bottom quartiles of pCRH concentrations. Women with the highest pCRH concentrations (top quartile, 81.9 pmol/L) were more likely to deliver sooner, whereas women with the lowest pCRH concentrations (bottom quartile, 15.0 pmol/L) were more likely to remain pregnant.

ROC curves reveal that pCRH concentrations may improve the precision of predicting preterm delivery among a sample of women who received betamethasone under current standard of care (area under the curve, 0.79; 95% CI, 0.68 to 0.90; P = 0.001) (Fig. 2A). The ROC model was used to determine optimal pCRH level cutoffs for these data in order to predict which women were likely to deliver within a week. When specificity was weighted equally with sensitivity (maximizing the Youden index), the optimal cutoff score for pCRH for this sample with this assay was 72.8 pmol/L (Youden index = 0.52). At this cutoff, 72% of women in this sample who delivered within 1 week would have been correctly identified (sensitivity), whereas 20% of women who delivered after 1 week would have been falsely identified (1 − specificity). See Supplemental Fig. 2A in the online data repository for Kaplan-Meier plots showing the probability of remaining pregnant for women with pCRH concentrations above and below the Youden index cutoff (44).

Figure 2.

(A) ROC curve and sensitivity and specificity at (B) each possible cutoff point for pCRH at time of betamethasone treatment. Sensitivity indicates the probability that a case is correctly identified (true-positive), and 1 − specificity is the probability that a woman who will not deliver within 7 d is falsely identified (false-positive). AUC, area under the ROC curve.

However, providing betamethasone to women who will deliver within a week (sensitivity) is a priority for clinical care, and available tools to predict preterm birth are limited. Therefore, to prioritize sensitivity while maintaining a reasonable degree of specificity, we identified an alternative cutoff value of 15.5 pmol/L for this sample. At this cutoff, 96% of women who delivered within 1 week would have been correctly identified as delivering within this window (sensitivity), whereas 69% of women who delivered after 1 week would have been falsely identified (1 − specificity). Additionally, if a pCRH cutoff of 15.5 pmol/L were applied to the current study sample, 30 women who delivered more than a week after betamethasone administration would have avoided unnecessary exposure, 14 of whom delivered at term gestation. See Figure 2B in the online data repository for Kaplan-Meier plots showing the probability of remaining pregnant for women with pCRH concentrations above and below the alternative cutoff (44).

Discussion

Our study findings suggest that pCRH concentrations before betamethasone treatment predict impending preterm delivery. Specifically, among women admitted to the hospital for threatened preterm labor and given corticosteroids, elevated pCRH concentrations prior to treatment improved discrimination between women who would deliver within a week and those who would not. The critical importance of this issue is highlighted by the fact that 43% of women in our sample delivered at term gestation. Because of challenges predicting timing of delivery, there has been a call for new tools and biomarkers to identify optimal timing of antenatal corticosteroid administration (27). The present findings indicate that pCRH could be an additional tool that contributes to the determination of risk for impending preterm delivery and thus informs timing of corticosteroid administration.

Decades of research have established that antenatal corticosteroids are a lifesaving treatment of preterm infants (2, 5, 45). Antenatal corticosteroids provide multiple benefits to fetal health and development by stimulating fetal maturation and reducing risk for neonatal morbidity and mortality (2, 5, 45). The benefits of antenatal corticosteroids are greatest when the infant is delivered within 7 days following treatment (3, 5). However, exposure to corticosteroids may be associated with elevated risk for neurodevelopmental delays (7, 8, 11–16), especially in cases where the infant was at risk for preterm delivery but was delivered at term.

Because of the challenges of predicting length of gestation and preterm birth (27), further investigation into biological measures predicting the onset of parturition is needed. Pretreatment levels of pCRH show promise in addressing this clinical need because they predict preterm birth and play an integral role in the causal pathways of parturition (38). Specifically, our findings suggest that pCRH may improve our ability to identify women who will go on to deliver within a week, so that infants born preterm receive maximum benefit from corticosteroid treatment and infants born term avoid unnecessary exposure. In the current sample, a cutoff of 15.5 pmol/L correctly identified 96% of women who delivered within a week, with a rate of misclassification of 69%. Notably, based on current standard of care, all women within this sample were administered corticosteroids, although 79% did not deliver within a week of treatment. Using the pCRH cutoff of 15.5 pmol/L, 31% of currently treated women would have been appropriately identified as not benefiting from treatment at this time. Findings highlight the potential clinical utility of pCRH in identifying if and when offspring may benefit from treatment.

The possible utility of pCRH as a clinical indicator of impending preterm delivery is supported by evidence that pCRH is part of the causal pathway controlling length of gestation (28). pCRH concentrations increase dramatically over the course of pregnancy (38). pCRH binds to receptors, promoting myometrial relaxation; however, as the pregnancy progresses, these receptors become less effective and ultimately activate contractile pathways (29). Because of the robust association between pCRH and timing of delivery, pCRH has been characterized as part of a “placental clock” controlling the onset of parturition (30, 38).

To our knowledge, only two studies have used pCRH to prospectively predict delivery within a set time frame (e.g., 24 and 48 hours) (41, 42). These studies suggest that pCRH after 28 weeks’ gestation predicts delivery within 24 to 48 hours (41, 42). The present results extend these findings by suggesting that pCRH is a useful biomarker that can aid in the decision of whether to administer corticosteroids by identifying women who will deliver in less than a week.

Although findings suggest that pCRH could serve as a diagnostic indicator of impending preterm delivery, several limitations should be addressed. First, validation studies with larger samples sizes and across varying populations (e.g., in low-risk pregnancies) are needed to replicate the findings of this single-site investigation and to evaluate clinical utility. Second, in our sample, the indications for corticosteroid administration were heterogeneous, and we did not find that the reason for treatment moderated the relation between pCRH and time from treatment to delivery. A strength of this approach is that it increases generalizability of our findings to a more medically diverse population of women; however, future studies should carefully evaluate these and other clinical risk factors in study design and interpretation of findings. Third, the current study recruited women with singleton pregnancies. The relevance to multifetal pregnancies thus remains unknown. Fourth, it is also unclear if pCRH could be combined with other clinical indicators of preterm birth, including biomarkers such as placental α microglobulin-1, fetal fibronectin, and transvaginal cervical length measurements, to optimize the ability to predict gestational length. Exploration of the possible advantages of using multiple indicators of risk for preterm birth could further improve accuracy in predicting preterm delivery.

The findings of this study could have important implications for obstetric practice. If these results are replicated, work could be done to develop a cost-effective and clinically feasible pCRH screen and evaluate its clinical utility in obstetric practice. ELISAs have been developed for assessing CRH, and future work could explore the feasibility of this method as a faster and more cost-effective approach (46). pCRH taps into the causal pathways of delivery, which makes it a promising biomarker for predicting impending preterm delivery. More effective clinical tools to predict the timing of parturition could guide when to administer corticosteroids and increase the likelihood of providing treatment within the optimal time frame of 1 week before preterm delivery. This would maximize benefits to preterm neonates, improving lung maturation and reducing rates of mortality and respiratory distress syndrome. Further, more precise prediction of timing of delivery would reduce the likelihood of administering antenatal corticosteroids to women who will go on to deliver at term, and thus whose offspring will not benefit from treatment. A current clinical goal is to optimize the sensitivity and specificity of clinical indicators of impending preterm delivery (2, 27). pCRH may be a promising biomarker to inform the timing of antenatal corticosteroid treatment, which would maximize the likelihood of delivering treatment within the optimal window.

Abbreviations:

Abbreviations:
 
• ACOG

  American College of Obstetricians and Gynecologists

 
• BMI

  body mass index

 
• GA

  gestational age

 
• pCRH

  placental corticotropin-releasing hormone

 
• ROC

  receiver-operating characteristic

Acknowledgments

The assistance of Megan Faulkner, Kendra Leak, and Natalie Hernandez for their participation in data collection is gratefully acknowledged. The authors thank the families who participated in this project.

Financial Support: This research was supported by the National Institutes of Health Grants R01 HD065823 and P50MH 096889 (to E.P.D.).

Disclosure Summary: The authors have nothing to disclose.

Author notes

D.A.S. and L.A.G. contributed equally to this study and share first authorship.