Ethics statement

The study was approved by the Institutional Review Boards at each of the clinical recruiting sites and the data coordinating center (Brown University; Emory University; University of Texas Medical Branch at Galveston; University of Texas Health Sciences Center at San Antonio; University of Utah Health Sciences Center; RTI International). All mothers participating in the study gave written informed consent.

Study design

Women with singleton stillbirths and live births enrolled in the Stillbirth Collaborative Research Network (SCRN) who consented to placental examination were analyzed. The SCRN was a population-based case-control study conducted in five geographic areas in the United States defined by state and county lines: the state of Rhode Island and specific counties in Massachusetts, Georgia, Texas, and Utah. Women were recruited between March 2006 and September 2008 at the time of delivery in 59 community and academic, urban and rural hospitals in the 5 areas with a cumulative average of 80,000 deliveries per year. The hospitals were selected to ensure access to at least 90% of all pregnancies ending in either stillbirth or live birth within each geographic area based on estimates from vital statistics data. All women whose pregnancies resulted in stillbirth and a representative sample of women with live births, oversampled for women delivering live-born infants at <32 weeks’ gestation and those of African-American descent delivering live-born infants at ≥32 weeks’ gestation, were approached for enrollment. Oversampling of live births <32 weeks’ gestation was necessary due to the difference in gestational age distribution between stillbirths and live births, while oversampling of African-American women delivering live births ≥32 weeks’ gestation was undertaken to achieve a ratio of approximately twice the number of live births to stillbirths, similar to the ratio seen among white non-Hispanic women and Hispanic women, out of concern for the increased burden of stillbirth in this population and the potential interaction of race with some hypothesized stillbirth risk factors. Terminations of pregnancy were excluded. Details of the study design have been reported previously [9].

Stillbirth

Fetal death was defined by Apgar scores of 0 at 1 and 5 minutes, and no signs of life by direct observation, at 20 weeks 0 days of pregnancy or later. This analysis was limited to stillbirths and live births at ≥ 24 weeks 0 days gestation. Births < 24 weeks gestation were excluded to avoid confounding due to variable clinical management of live births at periviable gestational ages. Guidelines for determining grades of maceration were defined in the postmortem examination protocol for intact fetuses as: Grade 0—No maceration, Grade I—Desquamation involving ≤1% of total body surface and brown-red discoloration of umbilical cord stump, Grade II—Desquamation of face, abdomen, or back involving >1% and ≤5% total body surface, Grade III—Desquamation involving >5% of body surface, Grade IV—Total brown skin discoloration, Grade V—Mummification; and for fragmented fetuses as: Grade 0—No maceration, Grade I-II—Tissue appears red/pink and fresh, Grade III—Tissue appears red/pink and mixed with brown, Grade IV—Tissue appears brown/gray, Grade IV—Tissue appears gray.

Placental abnormalities

The study included a standardized placental examination protocol [10]. This protocol included digital imaging, macroscopic examination, collection of frozen and ambient temperature samples of the cord, membranes and the placental disc, and microscopic examination of sections collected according to a specific sampling protocol. A minimum of five full-thickness placental tissue samples were obtained, one at the umbilical cord insertion and four others determined by random numbers that specified the axes and spacing of sampling. This strategy was used to avoid systematic differences in sampling locations and to ensure dispersion of the tissue sampling sites. All perinatal pathologists participated in training sessions to assure uniform execution of the protocol. The details of placental examination, sample and data collection were described previously [10]. General descriptions of the placental abnormalities investigated are given in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptions of the placental findings.

https://doi.org/10.1371/journal.pone.0182874.t001

Fetal growth abnormalities

Gestational age (GA) at death (for stillbirths) or delivery (for live births) was estimated using an algorithm developed by the SCRN investigators [11]. Preterm birth was defined as GA at delivery or death < 37 weeks 0 days. Birth weight percentiles were determined using birth weight, the SCRN estimate of GA at death (stillbirths) or delivery (live births), and an expected weight for GA based on ultrasound norms [12]. Ultrasound norms were used for both live births and stillbirths because they were derived using ultrasound measurements from ongoing uncomplicated pregnancies and have been shown to identify more babies as abnormally grown compared to population norms which are derived from birth weights of all live births, including preterm and/or with growth abnormalities, and underestimate the proportion of babies with abnormal growth [8, 13].

Small for gestational age (SGA) was defined as birth weight less than the 10^th percentile, average for gestational age (AGA) as birth weight between the 10^th and 90^th percentiles, and large for gestational age (LGA) as birth weight above the 90^th percentile.

Statistical analysis

Analysis was restricted to singletons with an adequate placental examination. Fragmented placentas and placentas from mummified stillbirths were excluded. Mummified stillbirths were defined as fragmented babies who had Grade IV-V maceration and intact babies who had Grade V maceration. In addition, stillbirths and live births with missing birth weight or with GA < 24 weeks by SCRN algorithm were excluded.

Data were weighted to take into account the study design and other aspects of the sampling plus differential availability of an adequate placental examination. Weighted analyses were conducted using SUDAAN version 11.0.1 software. Proportions of stillbirths and live births with selected maternal and infant characteristics and with each placental finding were compared with statistical significance determined by the Wald chi-square test. Comparisons of the continuous measures placental weight and ratio birth weight to placental weight were based on weighted ranks and conducted using regression models that included the stillbirth indicator (stillbirth/live birth) only. Next, tests of association between each placental finding and birth weight percentile were conducted for stillbirths and live births separately with statistical significance determined by the adjusted Wald F test. Regression models used for comparisons of continuous measures based on weighted ranks included the percentile group indicator only.

Subsequently, tests for whether the association between each placental finding and birth weight percentile differed for stillbirths and live births were conducted using a logistic or linear regression model with a placental finding as the dependent variable and the percentile group indicator, stillbirth indicator, and effects for the interaction as independent variables, with the p-value associated with the interaction between percentile group and stillbirth indicator reported. Where a placental finding was not noted for any stillbirths or live births in a birth weight percentile category (or conversely when noted for all in a category), the observation in that percentile group with the smallest analysis weight was recoded as having the finding (or not having the finding) for the purpose of providing an upper bound on the p-value for the interaction in the presence of these zero cells. An asterisk next to the p-value indicates it was derived in this manner. All proportions reported, as well as all other p-values, were based on the data without recoding zero cells. In each case where the interaction p-value was approximated by recoding zero cells, the revised proportions were checked to verify that the recoding did not introduce more extreme results than were present in the original data. Examples are provided in S1 Table.

In the primary analyses, comparisons of placental weight and the ratio birth weight/placental weight were additionally conducted with adjustment for GA. Linear regression models that included GA as a continuous variable and models that included GA as a categorical variable (groups as noted in table footnotes) were compared with similar inference. P-values from models that included categorical GA are reported in brackets beside unadjusted p-values in the tables.

Mechanistic insights

Placental findings associated with stillbirth but not fetal growth abnormalities (group 1) may affect placental function acutely and severely. Such a sudden and severe impairment of placental function could lead to stillbirth without sufficient time necessary to affect fetal growth or through mechanisms unrelated to fetal growth. Acute chorioamnionitis of the placental membranes and chorionic plate, fetal chorionic plate vascular degenerative changes, perivillous or intervillous fibrin or fibrinoid depositions, and fetal vascular thrombi in the chorionic plate were findings of this type. It is unlikely that the five placental findings associated with stillbirth but not fetal growth could in some cases represent post-mortem changes. The changes the placental tissue undergoes in fetal death are well established and are unlikely to be confused with the underlying pathology. The only lesion that can continue for a limited time and to a very limited degree is acute chorioamnionitis of the placental membranes. There can be a mild polymorphonuclear response in the membranes after fetal death, especially in a case of severe amniotic fluid infection syndrome but that can be differentiated by other findings.

The second group of placental findings, consisting of velamentous umbilical cord insertion, terminal villous hypoplasia, and parenchymal infarction, was associated with fetal growth abnormalities among preterm stillbirths only (group 2). This suggests that when these findings impair placental function chronically and severely leading to fetal growth abnormalities stillbirth may result. Milder forms might not substantially affect fetal growth and may result in live birth rather than stillbirth.

A single umbilical artery was associated with fetal growth abnormalities in term live births only (group 3). This pattern of association suggests that single umbilical artery may impair placental function chronically and mildly leading to fetal growth impairment that becomes obvious late in pregnancy and without causing stillbirth.

Terminal villous hypoplasia and parenchymal infarction appear to have different effects in preterm and term pregnancies. In preterm pregnancies significant associations with fetal growth abnormalities were found in stillbirths only (group 2), while in term pregnancies significant associations with fetal growth abnormalities were found among live births only (group 3). This may suggest that a more severe form of terminal villous hypoplasia and parenchymal infarction can lead to fetal growth abnormalities culminating in stillbirth of the fetus before reaching term, while a less severe form of these placental findings leads to fetal growth abnormalities, but does not lead to stillbirth and results in fetal growth abnormalities reaching term. However, it should be noted that the relatively small (weighted) numbers of preterm live births and of term stillbirths in the fetal growth categories likely limited our ability to detect significant associations in these groups.

Similar relationships with fetal growth in stillbirths and live births (group 4) were observed for placental weight in both preterm stillbirths and live births and for intraparenchymal thrombus in term stillbirths and live births. The similar patterns of association suggest chronic impairment of placental function leading to fetal growth abnormalities in both stillbirths and live births. In pregnancies ending in stillbirth, this pattern suggests a second hit phenomenon in consequence of which the stillbirth occurs. The first hit may result in fetal growth abnormalities without resulting in stillbirth such that some placental findings are associated with fetal growth abnormalities in both live births and stillbirths. In some of the pregnancies with such induced growth abnormalities, a second insult may result in a stillbirth.

The last group includes placental findings with a different patterm of fetal growth in stillbirths and live births (group 5). This suggests that these placental findings impair placental function chronically affecting fetal growth. In pregnancies ending in stillbirth the placental findings may be more severe resulting in growth abnormalities incompatible with live birth. In live births the placental findings may be milder resulting in growth abnormalities but without stillbirth. In preterm pregnancies, the placental findings of this type were terminal villous immaturity, avascular villi, placental edema and retroplacental hematoma and in term pregnancies retroplacental hematoma, placental weight, and birth weight to placental weight ratio. In term pregnancies, placental weight was similar between SGA and AGA stillbirths and live births, but was substantially higher in LGA live births than in LGA stillbirths. The birth weight to placental weight ratio in term pregnancies was similar in all fetal growth groups among live births, but was substantially smaller in SGA stillbirths and larger in LGA stillbirths. Term SGA stillbirths had mean birth weight to placental weight ratio of 6.8, substantially lower than AGA and LGA term stillbirths (7.7 and 9.3, respectively). Thus birth weight was approximately 6, 7, and 9 times larger than placenta weight in the term stillbirth groups. Such an observation suggests absolute placental insufficiency in SGA stillbirths with placental size and function insufficient to support fetal growth and then life. In LGA stillbirths, the high birth weight to placental weight ratio indicates a relative placental insufficiency with placental size and function large but insufficient to support an even larger fetus [14].

Stillbirths and live births with congenital anomalies were included in the analysis as well as births resulting from pregnancies with complications. Although the proportion of babies with congenital anomalies was higher in stillbirths than in live births, as were the proportions born to mothers with pregnancy complications including chronic hypertension, gestational hypertension/pre-eclampsia, and pre-gestational diabetes, we have previously shown that the pattern of association observed between fetal growth and stillbirth in the complete cohort was similar to that in the subset excluding births with congenital anomalies and the subset without maternal diabetes or hypertension/pre-eclampsia [8]. Further, since pregnancy complications are very likely mediating factors in the relationship between fetal growth abnormalities and placental findings, it would not be appropriate to adjust for them in the multivariable analysis. Macerated stillbirths were also included. Although maceration may have made evaluation of birth weight more difficult, we previously found that the association between fetal growth and risk of stillbirth observed in the entire study population was also observed in the subset with optimal estimates of fetal age that included non-macerated stillbirths with a short interval of less than 7 days between the time they were last reported alive and first identified as demised [8]. Subset analyses of the association between fetal growth and placental findings were not undertaken here due to small sample sizes in some birth weight percentile categories and the corresponding risk of false negative findings. Although preterm pregnancies and their placentas, including AGA pregnancies, are not normal by definition, their inclusion allowed us to account for the effects of preterm gestational age while examining associations between placental findings and fetal growth abnormalities.

Clinical implications

These distinct patterns of association between placental abnormalities, fetal growth, and stillbirth suggest not only different mechanisms of placental function impairment, but also different clinical management strategies. Some of the placental abnormalities, such as velamentous cord insertion, single umbilical artery, and placental volume can be detected during pregnancy using ultrasound [15, 16]. Placental volume correlates well with placental weight and is a good proxy for placental weight [17]. Other placental abnormalities, which are associated with stillbirth are at this time not detectable prenatally, but provide an important target of investigation within initiatives such as The Human Placenta Project [18]. The Human Placenta Project is a collaborative effort undertaken by the Eunice Kennedy Shriver National Institute of Child Health and Development “to understand the role of the placenta in health and disease”. It “aims to develop new tools to study the organ in real time to learn how it develops and functions throughout pregnancy. That knowledge may one day prevent and treat a range of common pregnancy complications, while providing insights into other areas of science and medicine as well.” https://www.nichd.nih.gov/research/HPP/Pages/default.aspx

The group of placental findings associated with stillbirth but not fetal growth abnormalities (group 1) appears to impair placental function in stillbirths severely and acutely and thus their diagnosis is at this time unlikely to affect clinical management in preterm pregnancies. This is because preterm delivery, which would prevent stillbirth, is associated with significant risk of neonatal mortality and morbidity that strongly depends on gestational age at delivery [19, 20]. In preterm pregnancies with these placental abnormalities, it would be difficult to decide when to intervene with delivery. However, presence of these placental abnormalities in term pregnancies could be a reasonable indication for delivery, if validated in future studies, since at term the risks of neonatal mortality and morbidity are very low and are likely outweighed by the risk of stillbirth [20–22].

In pregnancies with placental findings associated with fetal growth abnormalities in stillbirths only (group 2), and with placental findings associated with fetal growth abnormalities in both stillbirths and live births (group 4 and 5), the placental function appears to be affected chronically and severely in pregnancies ending in stillbirth. When growth abnormalities occur with these placental abnormalities, the risk of stillbirth may be substantially increased. Further studies are needed to quantify the risk of stillbirth in pregnancies with these placental findings and growth abnormalities to determine the gestational age at which delivery is associated with a net benefit of preventing stillbirth while minimizing the risk of neonatal mortality and morbidity. The net benefit in preterm pregnancies depends strongly on gestational age and the associated neonatal mortality and morbidity risk. However, at term, the risk of stillbirth associated with concurrent presence of these placental abnormalities and fetal growth abnormalities would likely outweigh the very low risk of neonatal mortality and morbidity [20–22]. Similarily, intervention in pregnancies with placental findings associated with fetal growth abnormalities in live births only (group 3) would require knowledge of the risk of adverse pregnancy outcomes, especially neonatal mortality and morbidity, associated with those findings. Also needed are studies of novel methods of non-invasive evaluation of placental structure and function, such as the ones conducted within The Human Placental Project, which would allow prenatal identification of these placental findings [18].

The clinical applicability at this stage is limited and likely restricted to birth weight to placental weight ratio and a few other placental findings identifiable by ultrasound examination. Recent developments in the estimation of placental weight hold the promise of clinical application but will have to be evaluated for clinical effectiveness.

Strengths

There are several strengths of this study. The SCRN was a population-based study, which attempted to capture all stillbirths and a representative sample of live births with geographic, racial and ethnic diversity. The population-based design strengthens the external validity of the study findings and thus their generalizability. This is also one of the largest studies of stillbirths with over 500 fetal deaths and well characterized phenotype. Another strength of the study is the use of an algorithm developed to estimate gestational age of stillbirths at the time of death [11]. This allowed a more accurate estimation of gestational age of stillbirth than one based on age at delivery, which is critical in precise estimation of fetal growth and its abnormalities [8]. Another strength of the study is a standardized placental examination with central training for pathologists [10]. Standardization of the placental examination increases its objectivity and decreases the variability of the placental findings.

Limitations

Despite the large number of stillbirths in the study, the numbers in individual fetal growth categories stratified into preterm and term pregnancies were relatively small, particularly those for term stillbirths. This may have resulted in insufficient power to detect more subtle differences in the patterns of association between placental abnormalities and fetal growth and stillbirth. Thus, interpretation of placental findings in the groups with smaller numbers of subjects requires caution. While estimation of GA in stillbirth based on time of death rather than time of birth significantly improves the accuracy of assessment of fetal growth [8], the exact time of death is not known and thus estimation of GA is not absolutely precise and is subject to some misclassification. Nevertheless, our prior study demonstrated that the algorithm used to estimate time of death in stillbirths performs well in estimation of GA and fetal growth in stillbirths [8, 11]. Pathologists could not be blinded to the origin of the placenta from stillbirths or live births due to the need to perform both clinical and research placental examinations. Chronic chorioamnionitis was not included in this analysis, because it is clinically significant only if it is associated with diffuse chronic villitis (of unknown etiology). There were very few cases of chronic chorioamnionitis with diffuse chronic villitis among the stillbirths in the cohort preventing analysis of the association with fetal growth. For these reasons, we did not examine chronic chorioamnionitis.

The clinical management implications are limited by the fact that many of the placental findings cannot be detected prenatally at this time. However, there is a large initiative underway within The Human Placenta Project to develop novel methods of prenatal evaluation of placental structure and function [18]. It is thus likely that more placental abnormalities associated with the risk of stillbirth and fetal growth will become amenable to diagnosis prenatally.

Summary

Placental function supports fetal growth and development and its impairment may result in fetal growth abnormalities and stillbirth. We identified placental abnormalities which impair placental function as demonstrated by fetal growth abnormalities and / or stillbirth. All placental abnormalities associated with fetal growth aberrations, except for terminal villous hypoplasia, were also associated with stillbirth. This observation indicates that these placental abnormalities are important for placental function.

Implications

The patterns of association between placental abnormalities, fetal growth, and stillbirth provide insights into the mechanism of placental function impairment and stillbirth. They also suggest implications for clinical care, especially for the placental findings amenable to prenatal diagnosis using ultrasound and associated with stillbirth in pregnancies at term. Clinical implications are especially clear for birth weight to placental weight ratio. The birth weight to placental weight ratio could be estimated prenatally using estimated fetal weight to placental volume ratio. Pregnancies with the abnormal ratio, at increased risk of stillbirth at term, may benefit from early term delivery, because of very low risk of neonatal mortality and morbidity at term.

Financial disclosures

This work including the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review and approval of the manuscript was supported by grant funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development: U10-HD045953 Brown University, Rhode Island; U10-HD045925 Emory University, Georgia; U10-HD045952 University of Texas Medical Branch at Galveston, Texas; U10-HDO45955 University of Texas Health Sciences Center at San Antonio, Texas; U10-HD045944 University of Utah Health Sciences Center, Utah; and U01-HD045954 RTI International, RTP. Program officers from the funding agency (Reddy and Willinger) were members of the Steering Committee of the study, and contributed to the study design, management, interpretation of the data, as well as preparation, review and approval of the manuscript. Otherwise the funders did not have any involvement in the design of the study; the collection, analysis, and interpretation of the data; the writing of the article; or the decision to submit the article for publication.

The Stillbirth Collaborative Research Network is solely responsible for the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. The following institutions and researchers comprise the Stillbirth Collaborative Research Network: University of Texas Health Science Center at San Antonio–Donald J. Dudley, Deborah Conway, Karen Aufdemorte, Angela Rodriguez, Monica Pina; University of Utah School of Medicine–Robert M. Silver, Michael W. Varner, Kristi Nelson; Emory University School of Medicine and the Rollins School of Public Health–Carol J. Rowland Hogue, Barbara Stoll, Janice Daniels Tinsley, Bahig Shehata, Carlos Abramowsky; Brown University–Donald Coustan, Halit Pinar, Marshall Carpenter, Susan Kubaska; University of Texas Medical Branch at Galveston–George R. Saade, Radek Bukowski, Jennifer Lee Rollins, Hal Hawkins, Elena Sbrana; RTI International–Corette B. Parker, Matthew A. Koch, Vanessa R. Thorsten, Holly Franklin, Pinliang Chen; Pregnancy and Perinatalogy Branch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health–Marian Willinger, Uma M. Reddy; Columbia University School of Medicine–Robert Goldenberg.

Additional contributions

We would like to acknowledge the members of the NICHD Scientific Advisory and Safety Monitoring Board for their review of the study protocol, materials, and progress, as well as all of the other physicians, study coordinators, research nurses, and patients who participated in the Stillbirth Collaborative Research Network.