Abstract
Introduction:
Adverse pregnancy outcomes such as preterm delivery and preeclampsia are associated with a higher maternal risk for subsequent cardiovascular disease (CVD) and all-cause mortality. While such pregnancy conditions are related to abnormal placentation, little research has investigated whether pathologic placental measures could serve as a risk factor for future CVD mortality in mothers.
Methods:
Longitudinal study of 33,336 women from the Collaborative Perinatal Project (CPP; 1959–1966) linked to mortality information through December 2016. Pathologists took extensive morphological and histopathological measures. Apart from assessing associations with morphological features, we derived an overall composite score and specific inflammation-related, hemorrhage-related, and hypoxia-related pathologic placenta index scores. Cox regression estimated hazard ratios (HR) and 95% confidence intervals (CI) for mortality adjusting for covariates.
Results:
Thirty-nine percent of women died with mean (standard deviation, SD) time to death of 39 (12) years. Mean (SD) placental weight and birthweight were 436g (98) and 3156g (566), respectively. Placenta-to-birthweight ratio was associated with all-cause mortality (adjusted HR 1.03: 1.01, 1.05 per SD in ratio). In cause-specific analyses, it was significantly associated with respiratory (HR 1.06), dementia (HR: 1.10) and liver (HR 1.04) related deaths. CVD, cancer, diabetes and kidney related deaths also tended to increase, whereas infection related deaths did not (HR 0.94; 0.83, 1.06). Placental measures of thickness, diameters, and histopathological measures grouped by inflammatory, hemorrhagic, or hypoxic etiology were not associated with mortality.
Discussion:
Placental weight in relation to birthweight was associated with long-term maternal mortality but other histopathologic or morphologic features were not.
Keywords: placental weight, placental diameter, placental histopathology, birthweight, mortality
Introduction
Pregnancy requires a complex interplay between placental and maternal cardiovascular systems.[1] Normal maternal cardiac adaptations in pregnancy involve increases in cardiac output and decreases in vascular resistance to promote nutrient and gas exchange for the fetus.[2] Meanwhile, the placenta produces hormones such as estrogens, progesterone, and placental growth factor to mediate these cardiac adaptations.[1] Longitudinal studies have pointed to the importance of maternal hypertensive disorders during pregnancy as risk factors for elevated risk of cardiovascular disease (CVD) and mortality decades after delivery.[3, 4] An umbrella review of 32 published systematic reviews and meta-analyses found that reproductive outcomes in women, including those with placental origins (i.e., preeclampsia, placental abruption) are associated with 1.5–2.0 times the risk for later CVD.[5] In 2014, the American Heart Association (AHA) recognized preeclampsia as a CVD risk factor. Whether these associations are due to underlying maternal risk unmasked during the cardiac adaptations required of normal pregnancy or are themselves causes of future risk remain unknown. As preeclampsia is tied to abnormal placentation [6], and preeclampsia is also tied to maternal CVD risk, we hypothesize that placental gross characteristics and histology may serve as a biomarker for future risk of CVD risk in mothers. Such placental abnormalities may mark suboptimal vascular capacity that then signals higher risk of CVD later in life.
Placental gross characteristics of weight, diameter, and thickness may also serve as crude, global markers of abnormal placental development during pregnancy. Furthermore, placental characteristics have been linked to adverse offspring health [7], but less work has focused on how placental characteristics affect subsequent maternal health. Yet, there is a growing body of literature that indicates placental features are related to maternal hypertensive disorders [8, 9] and CVD [10–12]. These studies are limited by short follow-up time, and a focus on diagnoses related to adverse placentation such as preeclampsia or preterm delivery. Since maternal mortality (particularly from CVD) may be related to indices of hypoxia and inflammation as well, the utility of placental characteristics attributed to these conditions and their ability in determining future mortality risk also deserves examination. Thus, the objective of this study was to investigate placental gross characteristics and histology among a diverse cohort of pregnant women with long-term mortality information five decades after delivery. In addition, we used machine learning techniques to examine the most informative individual placental measures for maternal mortality.
Methods
CPP Mortality Linkage Study
The Collaborative Perinatal Project (CPP) was a pregnancy cohort that recruited 48,197 women who had 59,391 pregnancies between 1959 to 1966 at 12 prenatal care centers across the United States. [13] Supplemental Figure 1 shows the flow diagram of those included in the current analysis. About 18% of women had more than one birth in the CPP and therefore, multiple placentas could have been evaluated. To be consistent in including only one set of observations across participants, analyses were limited to the last singleton pregnancy, among women with at least placental weight information and data sufficient for mortality linkage to be completed. Additional deliveries were excluded due to extremes in gestational age (i.e., <28 or >44 weeks) leaving 33,366 women for the present analysis.
The CPP Mortality Linkage Study was established to ascertain vital status through linkage to the National Death Index (NDI) and the Social Security Death Master File (SSDMF).[14] The NDI captures all U.S. deaths beginning from 1979 and was probabilistically matched to CPP participant information. The SSDMF has tracked deaths since CPP inception. The primary outcome of all-cause mortality was determined by deaths identified in the NDI or SSDMF as of December 31, 2016. Women were otherwise assumed alive. We previously demonstrated that there was reasonable agreement between these linkages and mortality status.[14]
Secondary outcomes are cause-specific mortality. Cause of death is provided in the NDI records as International Classification of Diseases (ICD) 9 or 10 codes and were classified as CVD (including coronary heart and cerebrovascular), cancer, diabetes, respiratory, infection, dementia, kidney, or liver mortality. Specifically, the corresponding codes were: CVD (ICD-10 codes I00-I78, I80-I99), cancer (C00–C97), diabetes (E10-E14), respiratory disease (pneumonia, influenza, chronic obstructive pulmonary disease, and allied conditions, J09-J18, J40-J47), infections (A00-B99), dementia (F01, F03, G30, G31.0, G31.83), kidney disease (N00-N07, N17-N19, N25-N27), chronic liver disease (K70, K73-K74), and other causes. All other types of deaths were grouped together and included in the analysis but results are not presented. The IRB at the National Institutes of Health approved the linkage study (protocol #14CHN196-A).
Placenta characteristics
The CPP had 103 placenta measures taken from specimens of over 80% of deliveries according to a strict protocol.[15] Placentas were collected and sealed in a plastic bag and delivered to the placental pathologists for examination. Almost all pathologists (97%) were blinded to maternal clinical condition at the time of exam.[16] Gross and/or microscopic measures were taken of the maternal surface, fetal surface, cut surface, umbilical cord, membranes, decidua, terminal villi, and intervillous space. Briefly, placental weight (g) was measured after removal of the cord, membranes, and visible blood clots. [15] Thickness (cm) was measured by piercing the center of the placenta tissue with a marked knitting needle. As the placenta is typically elliptical, two measures of diameters were taken. A ratio of the largest to smallest diameter was derived from the original measures only if the placenta did not have an unusual shape (i.e. non-round/oval) in order to remove impossible values (n=122). Placental weight, largest to smallest diameter, thickness and placenta-to-birthweight ratio were assessed separately.
Previously, measures were selected by experienced placental pathologists who created a score to classify placentas using morphologic and histologic information.[16] An overall composite index using 81 of the 103 placental measures that were deemed to have clinical significance was created along with 3 specific sub-categories for inflammation-related, hemorrhage-related and hypoxia-related measures which have been previously associated with childhood morbidities and development.[16, 17] These 3 indices were derived from 26, 16, and 47 gross/histologic measures. For instance, 1 point was given for calcification on the maternal surface and 2 points if throughout and so forth on categorical measures. Continuous histologic measures, such as the size of hemorrhages, were given 1 point if above the median and 0 if below. The placenta-to-birthweight ratio was given 1 point if above the 90th percentile for gestational week. Thus, characteristics were coded and summed into the 3 indices and the overall composite index consisted of all measures. The complete list of measures with their coding was previously published. [16]
Covariate information
Information on maternal age, race (White, Black, or other), socioeconomic status (based on education, income, employment [18]), smoking, pre-pregnancy weight, height, and gravidity were based on maternal report by in-person interview at study entry. The study also collected information on study site, year of pregnancy, gestational age (based on last menstrual period) and birthweight. Pre-pregnancy body mass index (BMI) was calculated as weight (kg) over height squared (m2). Hypertensive disorders were classified according to modern clinical definition by the American College of Obstetricians and Gynecologists Task Force on Hypertension in Pregnancy. [19] Chronic hypertension was classified if reported on the past medical history or obstetric diagnostic summary, or if blood pressure <20 weeks gestation was elevated (i.e., at least two readings of systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg). Gestational hypertension was classified if blood pressure ≥20 weeks gestation was elevated. Preeclampsia/eclampsia was classified in the presence of gestational hypertension with any of the following: proteinuria (≥20 weeks gestation dipstick readings of 1+ or recorded on obstetric summary), a headache within one week of delivery unresponsive to medication, visual disturbance within one week of delivery, pulmonary edema, or eclampsia documented on the delivery report or obstetric diagnostic summary. Superimposed preeclampsia/eclampsia was classified as preeclampsia/eclampsia in the presence of chronic hypertension. Prior medical conditions were classified as cardiovascular (e.g., hypertension, rheumatic fever, CVD), respiratory (e.g., asthma, thoracic surgery, tuberculosis), renal (e.g., glomerulonephritis, surgery), neurological/psychiatric (e.g., neuromuscular disease, alcoholism, psychosis, addiction, convulsive disorder, mental illness), cancer, and diabetes. Each class contributed to one point on the prior history score which ranged from 0 to 4.
Statistical analysis
Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CIs) for all-cause mortality with respect to individual placental characteristics (i.e., placental weight, ratio of largest to shortest diameter, thickness) and four placental indices (i.e., overall composite and specific inflammation-related, hemorrhage-related, and hypoxia-related). The index scores were dichotomized at the 90th percentile for analysis. Time to death was calculated from date of delivery to date of death or censored at Dec 31, 2016, if no death record was linked to the participant. Risk of cause-specific mortality was calculated using a competing risks model. While we retained all pregnancies with placental weight information, information on selected variables of histology or morphology may lead to missing index scores. To retain the women in the analysis, missing index information and covariate data were imputed in 25 datasets. Cause of mortality, which was not collected in SSMDF, was additionally imputed for cause-specific mortality models. An unadjusted model was first run and then compared against subsequent models adjusting first for sociodemographic factors associated with both placental weight [20] and mortality and subsequently adding preeclampsia, gestational age and birthweight. In sensitivity analyses, women with diabetes (n=514), or CVD conditions (n=3971) such as chronic hypertension prior to pregnancy were removed from the analyses described above (n=28,944, due to overlap in 75 women having both diabetes and other CVD conditions and 12 women missing information). Sensitivity analysis models retained counts of other prior chronic conditions but with a ceiling of 3 or more due to few numbers after removing diabetes and CVD.
As placental indices were not previously derived for maternal long-term outcomes or mortality, a machine learning approach was also applied to interrogate whether specific individual measures could be identified as associated with maternal long-term mortality. Elastic net, a machine learning technique for variable selection, was used to identify morphological and histopathological measures related to all-cause mortality among women. Analysis was first restricted to 30,424 (91%) Black or White women due to the strong impact of race/ethnicity on mortality and the heterogeneity of the other race/ethnic group (i.e., specified in 1960s with options of “Oriental”, “Puerto Rican”, “Other”). First, variable selection was conducted using the glmnet R package to perform elastic net on 85 placenta variables. These 85 consisted of the 80 measures used in the indices [16] (excluding the placental weight-to-birthweight ratio ≥90th percentile as these were entered individually as continuous measures) and placental weight, birthweight, large diameter, small diameter and thickness with Cox regression. A total of 1000 simulations were performed, and variables were retained if selected at least 700 times. Placental weight and birthweight were kept in all selections. The selected variables were then included in a Cox regression model to estimate HRs and 95% confidence intervals for all-cause mortality adjusting for age (continuous), year of pregnancy (continuous), race (Black/White), socioeconomic status (categorical), smoking (yes/no), preeclampsia (yes/no), gravidity (categorical), birthweight (continuous), placental weight (continuous), placental weight-to-birthweight ratio (continuous), length of gestation (continuous), number of medical conditions prior to pregnancy (categorical) and study site (categorical). This last model requiring all placenta measures and covariate information to be available included 23,830 women. In sensitivity analyses, women with prior diabetes or CVD were removed and elastic net was re-run (n=20,551).
All analyses were conducted in R v4.02 and SAS 9.4.
Results
Among the 33,366 women included, 48% were White, 43% were Black and 9% were of another race. (Table 1).). Average maternal age was 24.6 years and pre-pregnancy BMI 23.3 kg/m2. A third of women reported prior medical conditions and 14% had a preterm delivery. Placenta weight and birthweight averaged 436 grams and 3156 grams, respectively. Among 33,366 women, 13,124 (39%) died, with average (SD) time of follow-up until death of 39 (12) years.
Table 1.
Characteristics of women with placenta data in the Collaborative Perinatal Project (CPP) Mortality Linkage Study (n*=33,366)
| Characteristic | N (%) or mean (SD) |
|---|---|
| N | 33,366 |
| Age at delivery (years) | 24.60 (6.15) |
| Pre-pregnancy BMI (kg/m2) | 23.34 (4.42) |
| Race | |
| White | 15,965 (47.8) |
| Black | 14,459 (43.3) |
| Other | 2942 (8.9) |
| SES categories | |
| 1 (low) | 2198 (6.7) |
| 2 | 9750 (29.9) |
| 3 | 10,226 (31.3) |
| 4 | 6763 (20.7) |
| 5 (high) | 3708 (11.4) |
| Smoking status | |
| Non-smoker | 17,361 (52.9) |
| <1 pack/day | 9981 (30.4) |
| ≥ 1 pack/day | 5487 (16.7) |
| Gravidity | |
| 0 | 8977 (35.5) |
| 1 | 7069 (28.0) |
| 2 | 5284 (20.9) |
| 3 | 3788 (15.0) |
| 4 or more | 149 (0.6) |
| Number of medical conditions** | |
| 0 | 21,740 (65.2) |
| 1 | 8966 (26.9) |
| 2 | 2186 (6.6) |
| 3 | 407 (1.2) |
| 4 | 62 (0.2) |
| Preeclampsia/Eclampsia | 356 (1.1) |
| Male infant sex | 16,181 (49.1) |
| Gestational age at delivery (weeks) | 39.06 (2.77) |
| Preterm delivery (<37 weeks) | 4666 (14.0) |
| Birthweight (grams) | 3156 (566) |
| Placental weight (grams) | 436 (98) |
| Placental to birthweight ratio | 0.14 (0.3) |
| Placental thickness (mm) | 21.7 (4.8) |
| Largest to smallest diameter (ratio) | 1.17 (0.15) |
| Placental indices | Mean (SD); [range] |
| Overall composite index | 10.68 (3.80); [3–36] |
| Inflammation related index | 1.49 (1.67); [0–19] |
| Hemorrhage related index | 0.43 (0.85); [0–9] |
| Hypoxia related index | 8.72 (2.75); [1–25] |
Numbers of women with missing data: 712 on SES, 2849 on BMI, 537 on smoking status, 64 on parity, 12 on family history of diabetes, 413 on infant sex, 5 on medical history and pre-gestational diabetes status
Number of medical conditions related to CVD (n=3971), diabetes (n=514) and respiratory (n=2590), renal (n=3022), or neurological diseases (n=3412) prior to pregnancy, 5 missing
Placental weight and its ratio to birthweight were associated with increased all-cause maternal mortality (Table 2). Multiple models were used to account for covariates, with a final model additionally adjusting for preeclampsia to better disentangle placental pathology from those events. Associations remained after additional adjustment for covariates including preeclampsia (HR 1.05; 95% CI: 1.02–1.07) for placental weight and for its ratio with birthweight (HR 1.03; 1.01, 1.05). The largest to smallest placenta diameter (1.01; 1.00, 1.03) and placental thickness were not (1.00; 0.99, 1.02). We also examined whether placenta characteristics were associated with cause-specific mortality (Supplemental Table 1). The placenta-to-birthweight ratio was associated with increased risk of respiratory (HR 1.06; 1.00, 1.12), dementia (HR 1.10; 1.02, 1.19) and liver (1.04; 1.00, 1.08) related deaths after adjustment for covariates including preeclampsia. However, hazard ratios for all other causes of death, except for those infection related, were also elevated (e.g., CVD HR 1.03; 0.99, 1.06). Placental weight adjusted for birthweight had similar results with cause-specific deaths except for an association remaining with diabetes (HR 1.14; 1.03, 1.26), which attenuated after removing women with pre-pregnancy diabetes or CVD in sensitivity analyses (HR 1.09; 0.96, 1.24). (Supplemental Tables 1) Associations with dementia and liver disease were also attenuated but associations with respiratory causes of death remained.
Table 2.
Placental characteristics and risk of maternal all-cause mortality, CPP Mortality Linkage Study
| Placental Weight (per SD) | 1 | 1.04 (1.02,1.06) | 1.03 (1.01,1.05) |
| 2 | 0.99 (0.98,1.01) | 0.99 (0.97,1.01) | |
| 3 | 1.05 (1.02,1.07) | 1.05 (1.02,1.07) | |
| 4 | 1.05 (1.02,1.07) | 1.05 (1.02,1.07) | |
| Placenta to birthweight ratio (per SD) | 1 | 1.07 (1.04,1.09) | 1.06 (1.04,1.09) |
| 2 | 1.04 (1.01,1.07) | 1.04 (1.02,1.06) | |
| 3* | 1.03 (1.01,1.05) | 1.03 (1.01,1.05) | |
| 4* | 1.03 (1.01,1.05) | 1.03 (1.01,1.05) | |
| Largest to smallest diameter (per SD) | 1 | 1.03 (1.01,1.05) | 1.02 (1.00,1.04) |
| 2 | 1.01 (1.00,1.03) | 1.01 (0.99,1.03) | |
| 3 | 1.01 (1.00,1.03) | 1.01 (0.99,1.03) | |
| 4 | 1.01 (1.00,1.03) | 1.01 (0.99,1.03) | |
| Thickness (per SD) | 1 | 0.96 (0.94,0.98) | 0.95 (0.94,0.97) |
| 2 | 0.99 (0.98,1.01) | 1.00 (0.98,1.02) | |
| 3 | 1.00 (0.99,1.02) | 1.01 (0.99,1.03) | |
| 4 | 1.00 (0.99,1.02) | 1.01 (0.99,1.03) | |
| Placental Indices ** | |||
| Overall composite | 1 | 1.07 (1.02,1.12) | 1.05 (1.00,1.11) |
| 2 | 1.04 (0.99,1.10) | 1.04 (0.98,1.10) | |
| 3 | 1.03 (0.98,1.08) | 1.02 (0.97,1.08) | |
| 4 | 1.03 (0.98,1.08) | 1.02 (0.97,1.08) | |
| Inflammation related | 1 | 1.15 (1.10,1.20) | 1.16 (1.11,1.22) |
| 2 | 1.03 (0.99,1.08) | 1.05 (1.00,1.11) | |
| 3 | 1.01 (0.97,1.06) | 1.03 (0.98,1.09) | |
| 4 | 1.01 (0.97,1.06) | 1.03 (0.98,1.09) | |
| Hemorrhage related | 1 | 1.00 (0.94,1.06) | 0.98 (0.93,1.05) |
| 2 | 1.00 (0.94,1.06) | 1.00 (0.94,1.07) | |
| 3 | 0.98 (0.93,1.04) | 0.99 (0.92,1.05) | |
| 4 | 0.98 (0.93,1.04) | 0.99 (0.92,1.05) | |
| Hypoxia related | 1 | 0.98 (0.93,1.03) | 0.97 (0.92,1.02) |
| 2 | 1.03 (0.98,1.08) | 1.03 (0.97,1.08) | |
| 3 | 1.03 (0.98,1.08) | 1.02 (0.97,1.08) | |
| 4 | 1.03 (0.98,1.08) | 1.02 (0.97,1.08) |
Model 1: unadjusted; Model 2: age, race, composite socioeconomic status, pre-pregnancy BMI, smoking, gravidity, number of prior medical diseases, site, year of pregnancy; Model 3: model 2 + gestational age, birth weight; Model 4: model 3 + preeclampsia (*Models 3 and 4 removed additional birthweight adjustment for placenta-to-birthweight ratio models);
dichotomized at 90th percentile;
Sensitivity analysis excluding women with diabetes or CVD conditions (including hypertension) prior to pregnancy.
Bolding denotes p<0.05.
An overall composite index grouping 81 morphological and histopathological measures into 3 specific sub-categories for inflammation-related, hemorrhage-related and hypoxia-related measures were derived. The overall composite and inflammation-related placenta indices were initially associated with increased risk in crude models, but not after adjustment for confounding (Table 2). With respect to specific causes of death, the inflammation-related placental index was associated with increased risk of mortality from cancer, cardiovascular, diabetes, respiratory, and dementia related deaths prior to, but not after, adjustment for covariates (Supplemental Table 2).
Associations between placental histologic features and maternal long-term mortality were identified using an agnostic machine learning approach in a subgroup of 23,830 Black/White women with complete information on placental characteristics and covariates (Supplemental Figure 1). In the first step, of the 85 placental variables, 51 were retained as having been kept in 70% of iterations ran with elastic net as significantly associated with total mortality. A Cox regression model including all placenta characteristics and baseline characteristics, identified 6 placental factors associated with long-term maternal mortality (p<0.05) (Supplemental Table 3). Placental weight (HR 1.05; 1.01, 1.08), largest diameter (1.03; 1.00, 1.06), presence of cystic change in cytotrophoblast of columns (1.07; 1.01, 1.14), number of micro infarcts in the terminal villi (1.09; 1.01, 1.17), and number of pink/red marginal cut surface infarcts (1.18; 1.02, 1.36) were associated with increased mortality. Presence of pathological edema was associated with decreased total mortality (0.78; 0.67, 0.92). Sensitivity analysis excluding women who had diabetes or CVD conditions including hypertension prior to pregnancy did not meaningfully change results except the association with number of pink/red marginal cut surface infarcts was attenuated (HR 1.13; 0.95, 1.35).
Discussion
Increasing evidence suggests that pregnancy is a stress test for women, unmasking underlying chronic disease risks that may not be apparent until later in life. In this novel investigation of placental characteristics with long-term maternal mortality, gross characteristics such as placental weight and diameter were associated with increased mortality risk 50 years after delivery. These associations remained even after adjustment for preeclampsia, a known risk factor for long-term CVD and renal disease mortality. Previously derived placental indices grouped by inflammation-related, hemorrhage related, and hypoxia-related characteristics, and that have been found to be related to childhood outcomes [16, 17], were not associated with subsequent maternal mortality. However, several individual histologic/morphologic features identified through elastic net were significantly related to all-cause mortality, underscoring the placental development as a potential important marker for subsequent maternal mortality.
Placental gross measures
The finding that increased placental weight in proportion to birthweight was associated with increased risk of subsequent maternal all-cause mortality may indicate the level of adaptability for women to the biological requirements of normal pregnancy that then is related to longevity. The average human placenta has a diameter of 22cm, thickness of 2.5cm and weight of 470g. [21] Placental weight may be altered by timing of cord clamping and when the placenta is weighed in relation to delivery. Hence, there may be error around such measurements, but given a standard protocol was available in the CPP to weigh the placenta before storage, the errors should be minimal. Similarly, placental weight is a function of how wide and thick the placenta grows. Here, we attempted to tease these aspects apart. While lateral growth is an indicator of surface area, thickness is a function of how dense the branches of the villous tree are for nutrient exchange.[21] While we identified associations for all-cause mortality with placental weight, ratio of the diameters and thickness were not associated with mortality after adjustment for covariates. Despite a rigorous protocol, diameters and thickness may have been more subject to measurement error and driven associations towards the null.
Placental weight may be measured with less inter-rater error than other features and with the availability of data in large registries, remains what previous studies have mostly examined. Nevertheless, placental weight as a ratio of birthweight is useful as a crude measure of placental efficiency. The modeling of the two in relation to each other has been discussed.[22, 23] Taking the ratio of the two measures may potentially misclassify placentas as being efficient at lower weight due to a mathematical property of not having a true intercept of zero. [22] Hence, we modeled placental weight using the ratio and we adjusted for birthweight to see if there were important differences in findings. We found both models of assessing placental weight to be associated with long-term mortality and did not pursue further models that would have required further assumptions or were difficult to interpret.
Our findings also pointed to the ratio of placental weight to birthweight being associated with several causes of death (i.e., diabetes, respiratory, dementia and liver) after adjustment for preeclampsia and other risk factors. Findings remained even with adjustment for gestational age. All other causes of death, except for infection, were also increased in risk with higher placental weight but estimates were imprecise. As such, a clear association with CVD causes of death was not observed despite our original hypothesis. On the other hand, an association with respiratory disease related deaths was observed for placental weight. Drastic changes occur in respiratory processes during pregnancy.[24] Inability to adapt to changes in respiratory requirements during pregnancy may have resulted in lower than optimal placental to birthweight ratios and became reflected in longer term mortality related to this cause of death. Dementia deaths, which includes Alzheimer’s disease, were also elevated (although this association was attenuated by exclusion for prior diabetes and CVD). Three studies previously found preeclampsia to be associated with Alzheimer’s development in women.[25] Underlying common genetic causes also may explain the association.[26] Again, given that placental weight is a crude measure, it is difficult to pinpoint etiology and biologically relevant pathways. Moreover, all causes except infections, were increased despite some statistical imprecision.
Placenta histology
Using a machine learning approach, we identified few associations between all-cause mortality and placental characteristics. While we found positive associations for two types of infarcts, caution should be taken for the interpretation of the results as a recent analysis of the CPP infarct data (which in its completeness includes 12 different metrics) showed high variability across study sites in their prevalence and in their associations with clinical outcomes including birthweight and placental weight. [27] The authors recommended digitizing data in the future and using technology to reliably identify features and increase the validity of these measures as a useful pathological construct. We suspect that the low reliability of individual measures may be true of the other microscopic features we observed which identified putative as well as protective associations. There also remains an incomplete understanding of the process of calcification in the placenta.[21] Nevertheless, similar to other studies evaluating placental calcification at delivery or by ultrasound during pregnancy on other clinical outcomes [21], we identified no associations with mortality risk. Calcification has been shown to increase over the course of pregnancy, with highest levels in term deliveries and some evidence of seasonal variations. Calcification therefore may only indicate placental age, with abnormally high calcification only meaningful in preterm deliveries. Grade and quantity of calcification has also not been shown to be clinically meaningful.[21] Hence, our finding that calcification is not associated with mortality risk aligns with previous findings.
Strengths & Limitations
Strengths of the study include the extensive placenta data captured in a large racially diverse cohort with long-term mortality data. This analysis is novel in examining mortality associations using both an approach founded on the biological basis of similarities (i.e., the placenta indices) and an agnostic, machine learning approach to identify placenta characteristics. However, there are some notable limitations. The NDI was established in 1979 which left a gap of approximately twenty years in which no cause-specific mortality data was available and the SSDMF might not fully capture all deaths, due to its reliance on social security number information which was incomplete among CPP participants.[14] Cause specific mortality was imputed for deaths classified by SSDMF based on placental information and covariates which might be incomplete predicators. Given the young age of women between 1959–1979, deaths should have been fewer during this period of missing NDI data but the possibility that any early deaths may have differentially occurred in women who had poor placental characteristics cannot be ruled out. Although many placental characteristics were captured, the methods 50 years ago for defining histology may not align with current practice. [28] Other historical context should also be noted. For example, rheumatic fever, which is currently rare, accounted for 4% of prior CVD medical history because the women grew up during the time penicillin was being introduced.[29] Inaccuracy of gestational age without ultrasound confirmation may have caused some residual confounding but it would have biased any associations towards the null assuming it was random error.
Conclusion
In this large racially diverse cohort sampled from the U.S. population, lower placental weight accounting for birthweight is associated with long-term mortality of mothers up to 50 years after pregnancy. Attempts to identify the biological processes underlying these associations were inconclusive, with no grouping of the gross and histological characteristics by placental indices involved with inflammation or hypoxia or hemorrhage standing out as predictors of mortality. Placental weight remains a crude but reliable measure that hints at potential multi-factorial inability to optimally adapt to pregnancy requirements that have an impact on long-term mortality.
Supplementary Material
Highlights.
High placental weight to birthweight is related to mothers’ long-term mortality.
All causes of mortality (aside from infection-related deaths) were elevated.
Deaths from respiratory, dementia, and liver causes were particularly elevated.
Histopathological measures grouped by etiologies were not related to mortality.
Funding:
Supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD; contract HHSN275200800002I/27500013)
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data Availability:
The original CPP data regarding placenta characteristics and covariates are available by request through the NICHD DIPHR Biospecimen Repository Access and Data Sharing (BRADS) repository. The mortality data are available on request from the corresponding author [EY]. The mortality data are not on a public database due to ethical/privacy restrictions (i.e., releasing information that could compromise participant privacy).
References
- 1.Yagel S, Cohen SM, Goldman-Wohl D. An integrated model of preeclampsia: a multifaceted syndrome of the maternal cardiovascular-placental-fetal array. Am J Obstet Gynecol. 2020. 10.1016/j.ajog.2020.10.023 [DOI] [PubMed] [Google Scholar]
- 2.Mulder EG, De Haas S, Mohseni Z, Schartmann N, Abo Hasson F, Alsadah F, et al. Cardiac output and peripheral vascular resistance during normotensive and hypertensive pregnancy - a systematic review and meta-analysis. BJOG. 2021. 10.1111/1471-0528.16678 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mannisto T, Mendola P, Vaarasmaki M, Jarvelin MR, Hartikainen AL, Pouta A, et al. Elevated blood pressure in pregnancy and subsequent chronic disease risk. Circulation. 2013;127(6):681–90. 10.1161/CIRCULATIONAHA.112.128751 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bellamy L, Casas JP, Hingorani AD, Williams DJ. Pre-eclampsia and risk of cardiovascular disease and cancer in later life: systematic review and meta-analysis. BMJ. 2007;335(7627):974. 10.1136/bmj.39335.385301.BE [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Okoth K, Chandan JS, Marshall T, Thangaratinam S, Thomas GN, Nirantharakumar K, et al. Association between the reproductive health of young women and cardiovascular disease in later life: umbrella review. BMJ. 2020;371:m3502. 10.1136/bmj.m3502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Aplin JD, Myers JE, Timms K, Westwood M. Tracking placental development in health and disease. Nat Rev Endocrinol. 2020;16(9):479–94. 10.1038/s41574-020-0372-6 [DOI] [PubMed] [Google Scholar]
- 7.Thornburg KL, Marshall N. The placenta is the center of the chronic disease universe. Am J Obstet Gynecol. 2015;213(4 Suppl):S14–20. 10.1016/j.ajog.2015.08.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Holzman CB, Senagore P, Xu J, Dunietz GL, Strutz KL, Tian Y, et al. Maternal risk of hypertension 7–15 years after pregnancy: clues from the placenta. BJOG. 2021;128(5):827–36. 10.1111/1471-0528.16498 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Staff AC, Dechend R, Pijnenborg R. Learning from the placenta: acute atherosis and vascular remodeling in preeclampsia-novel aspects for atherosclerosis and future cardiovascular health. Hypertension. 2010;56(6):1026–34. 10.1161/HYPERTENSIONAHA.110.157743 [DOI] [PubMed] [Google Scholar]
- 10.Ray JG, Vermeulen MJ, Schull MJ, Redelmeier DA. Cardiovascular health after maternal placental syndromes (CHAMPS): population-based retrospective cohort study. Lancet. 2005;366(9499):1797–803. 10.1016/S0140-6736(05)67726-4 [DOI] [PubMed] [Google Scholar]
- 11.Catov JM, Muldoon MF, Reis SE, Ness RB, Nguyen LN, Yamal JM, et al. Preterm birth with placental evidence of malperfusion is associated with cardiovascular risk factors after pregnancy: a prospective cohort study. BJOG. 2018;125(8):1009–17. 10.1111/1471-0528.15040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Cain MA, Salemi JL, Tanner JP, Kirby RS, Salihu HM, Louis JM. Pregnancy as a window to future health: maternal placental syndromes and short-term cardiovascular outcomes. Am J Obstet Gynecol. 2016;215(4):484 e1–e14. 10.1016/j.ajog.2016.05.047 [DOI] [PubMed] [Google Scholar]
- 13.Niswander KR, Gordon M, National Institute of Neurological Diseases and Stroke. The women and their pregnancies; the Collaborative Perinatal Study of the National Institute of Neurological Diseases and Stroke. Philadelphia,: Saunders; 1972. [Google Scholar]
- 14.Pollack AZ, Hinkle SN, Liu D, Yeung EH, Grantz KL, Mumford SL, et al. Vital Status Ascertainment for a Historic Diverse Cohort of U.S. Women. Epidemiology. 2020;31(2):310–6. 10.1097/EDE.0000000000001134 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Benirschke K Examination of the Placenta. Obstet Gynecol. 1961;18(3):309. [Google Scholar]
- 16.Ye J, Chen Y, Zhu J, Chen C, Zhu X, Feng L, et al. Aspirin use during pregnancy and hypoxia-related placental pathology. Pregnancy Hypertens. 2018;14:177–88. 10.1016/j.preghy.2018.07.009 [DOI] [PubMed] [Google Scholar]
- 17.Chen Y, Zou L, Zhao Y, Wu T, Ye J, Zhang H, et al. Creating a placental inflammatory composite index that has a high prognostic relevance to child morbidity. J Obstet Gynaecol Res. 2017;43(7):1169–79. 10.1111/jog.13328 [DOI] [PubMed] [Google Scholar]
- 18.Myrianthopoulos NC, French KS. An application of the U.S. Bureau of the Census socioeconomic index to a large, diversified patient population. Soc Sci Med. 1968;2(3):283–99. 10.1016/0037-7856(68)90004-8 [DOI] [PubMed] [Google Scholar]
- 19.ACOG Practice Bulletin No. 202: Gestational Hypertension and Preeclampsia. Obstet Gynecol. 2019;133(1):1. 10.1097/AOG.0000000000003018 [DOI] [PubMed] [Google Scholar]
- 20.Salafia CM, Zhang J, Charles AK, Bresnahan M, Shrout P, Sun W, et al. Placental characteristics and birthweight. Paediatr Perinat Epidemiol. 2008;22(3):229–39. 10.1111/j.1365-3016.2008.00935.x [DOI] [PubMed] [Google Scholar]
- 21.Benirschke K, Burton GJ, Baergen RN. Pathology of the Human Placenta. 6 ed. New York: Springer; 2012. [Google Scholar]
- 22.Christians JK, Grynspan D, Greenwood SL, Dilworth MR. The problem with using the birthweight:placental weight ratio as a measure of placental efficiency. Placenta. 2018;68:52–8. 10.1016/j.placenta.2018.06.311 [DOI] [PubMed] [Google Scholar]
- 23.Salafia CM, Misra DP, Yampolsky M, Charles AK, Miller RK. Allometric metabolic scaling and fetal and placental weight. Placenta. 2009;30(4):355–60. 10.1016/j.placenta.2009.01.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.LoMauro A, Aliverti A. Respiratory physiology of pregnancy: Physiology masterclass. Breathe (Sheff). 2015;11(4):297–301. 10.1183/20734735.008615 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Basit S, Wohlfahrt J, Boyd HA. Pre-eclampsia and risk of dementia later in life: nationwide cohort study. BMJ. 2018;363:k4109. 10.1136/bmj.k4109 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.van Dijk M, van Bezu J, Poutsma A, Veerhuis R, Rozemuller AJ, Scheper W, et al. The preeclampsia gene STOX1 controls a conserved pathway in placenta and brain upregulated in late-onset Alzheimer’s disease. J Alzheimers Dis. 2010;19(2):673–9. 10.3233/JAD-2010-1265 [DOI] [PubMed] [Google Scholar]
- 27.Misra DP, McNally S, Chen S, Salafia CM. Placental infarcts in the collaborative perinatal project: Variable associations infer variable constructs. Placenta. 2020;99:1–7. 10.1016/j.placenta.2020.06.004 [DOI] [PubMed] [Google Scholar]
- 28.Christians JK, Huicochea Munoz MF. Pregnancy complications recur independently of maternal vascular malperfusion lesions. PLoS One. 2020;15(2):e0228664. 10.1371/journal.pone.0228664 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Massell BF, Chute CG, Walker AM, Kurland GS. Penicillin and the marked decrease in morbidity and mortality from rheumatic fever in the United States. N Engl J Med. 1988;318(5):280–6. 10.1056/NEJM198802043180504 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The original CPP data regarding placenta characteristics and covariates are available by request through the NICHD DIPHR Biospecimen Repository Access and Data Sharing (BRADS) repository. The mortality data are available on request from the corresponding author [EY]. The mortality data are not on a public database due to ethical/privacy restrictions (i.e., releasing information that could compromise participant privacy).