Early prediction of preeclampsia and small-for-gestational-age via multi-marker model in Chinese pregnancies: a prospective screening study

Constructing small for gestational age prediction models: A retrospective machine learning study

OBJECTIVE

To develop machine learning prediction models for small for gestational age with baseline characteristics and biochemical tests of various pregnancy stages individually and collectively and compare predictive performance.

STUDY DESIGN

This retrospective study included singleton pregnancies with infants born between May 2018 and March 2023. Small for gestational age was defined as a birth weight below the 10th percentile according to the Intergrowth-21st fetal growth standards. The pregnancy data were categorized into four datasets at different gestational time points (14 and 28 weeks and admission). The LightGBM framework was utilized to assess the variable importance by employing a five-fold cross-validation. RandomizedSearchCV and sequential feature selection were applied to estimate the optimal number of features. Seven machine learning algorithms were used to develop prediction models, with an 8:2 ratio for training and testing. The model performance was evaluated using receiver operating characteristic curve analysis and sensitivity at a false positive rate of 10 %.

RESULTS

We included data of 4,394 women with singleton pregnancies, including 148 (3.4%) small for gestational age infants. Women delivering small for gestational age infants exhibited significantly shorter stature and lower fundal height and abdominal circumference at admission. Maternal height, age, and pre-pregnancy weight consistently ranked among the top 20 features in prediction models with any dataset. The models incorporated variables of admission stage have strong predictive performance with the area under the curves exceeding 0.8. The prediction model developed with variables of admission stage yielded the best performance, achieving an area under the curve of 0.85 and a sensitivity of 73% at the false positive rate of 10%.

CONCLUSIONS

By machine learning, various pregnancy stages' prediction models for small for gestational age showed good predictive performance, and the predictive value of variables at each pregnancy stage was fully explored. The prediction model with the best performance was established with variables of admission stage and emphasized the significance of prenatal physical examinations.

BMI modifies the effect of pregnancy complications on risk of small- or large-for-gestational-age newborns

BACKGROUND

Maternal physical condition (reflected by maternal body mass index (BMI) at delivery) and pregnancy complications influence neonatal health outcomes. High BMI during pregnancy increases various health problems' risks, but studies about the synthesized effect of these factors on fetal growth, are scarce.

METHODS

The retrospective cohort study was conducted in Zhejiang Province, China from 1 January 2019 to 31 December 2021. The associations between complications and small-for-gestational-age (SGA) and large-for-gestational-age (LGA) were measured by the Fine-Gray model and subgroup analysis. Effect modification and interaction analyses were conducted to explore BMI's modification effect and complications' interaction.

RESULTS

Several complications increased the risk for SGA and LGA, some significance varied in different subgroups. There was a positive effect modification of gestational diabetes mellitus (GDM) across BMI strata on LGA (relative excess risk due to interaction (RERI) [95% CI] = 0.57 [0.09,1.04]). Several pairwise complications' interactions were synergistic (e.g., pregestational diabetes and intraamniotic infection for SGA (ratio of ORs [95% CI] = 8.50 [1.74,41.37]), pregestational diabetes and assisted reproductive technology (ART) for LGA (ratio of ORs [95% CI] = 2.71 [1.11,6.62])), one was antagonistic (placental problems and ART for LGA (ratio of ORs [95% CI] = 0.58 [0.35,0.96])).

CONCLUSIONS

High-BMI positively modified the risk of GDM on LGA. Many interactions existed when two specific pregnancy complications occurred simultaneously.

IMPACT

This is the largest retrospective study covering more than 10 pregnancy complications to date in this aspect. High-BMI (BMI > 28 kg/m2) positively modifies the risk of GDM on LGA. Many pregnancy complications influence the risk of SGA and LGA, with several interactions that may create a "syndrome" effect. Pregnant women with different BMIs should consider the additional risks caused by pregnancy complications for their heterogeneous effects on abnormal fetal growth. Measures should be taken to prevent the occurrence of other exposure factors in the "syndrome". This study may aid in developing a new strategy for improving neonatal outcomes.

Epigenetic phenotype of plasma cell-free DNA in the prediction of early-onset preeclampsia

BACKGROUND

In the current study, we sought to characterise the methylation haplotypes and nucleosome positioning patterns of placental DNA and plasma cell-free DNA of pregnant women with early-onset preeclampsia using whole genome bisulphite sequencing (WGBS) and methylation capture bisulphite sequencing (MCBS) and further develop and examine the diagnostic performance of a generalised linear model (GLM) by incorporating the epigenetic features for early-onset preeclampsia.

METHODS

This case-control study recruited pregnant women aged at least 18 years who delivered their babies at our Hospital. In addition, non-pregnant women with no previous history of diseases were included. Placental samples of the villous parenchyma were taken at the time of delivery and venous blood was drawn from pregnant women during non-invasive prenatal testing at 12-15 weeks of pregnancy and nonpregnant women during the physical check-up. WGBS and MCBS were carried out of extracted genomic DNA. Then, we established the GLM by incorporating preeclampsia-specific methylation haplotypes and nucleosome positioning patterns and examined the diagnostic performance of the model by receiver operating characteristic (ROC) curve analysis.

RESULTS

The study included 135 pregnant women and 50 non-pregnant women. Our high-depth MCBS revealed notably different DNA methylation and nucleosome positioning patterns between women with and without preeclampsia. Preeclampsia-specific hypermethylated sites were found predominantly in the promoter regions and particularly enriched in CTCF on the X chromosome. Totally, 2379 preeclampsia-specific methylation haplotypes were found across the entire genome. ROC analysis showed that the area under the ROC curve (AUC) was 0.938 (95%CI 0.877, 1.000). At a GLM cut-off of 0.341, the AUC was the maximum, with a sensitivity of 95.6% and a specificity of 89.7%.

CONCLUSION

Pregnant women with early-onset preeclampsia exhibit DNA methylation and nucleosome positioning patterns in placental and plasma DNA.

Role of First Trimester Screening Biochemical Markers to Predict Hypertensive Pregnancy Disorders and SGA Neonates-A Narrative Review

Early recognition of high-risk pregnancies through biochemical markers may promote antenatal surveillance, resulting in improved pregnancy outcomes. The goal of this study is to evaluate the possibilities of using biochemical markers during the first trimester of pregnancy in the prediction of hypertensive pregnancy disorders (HPD) and the delivery of small-for-gestational-age (SGA) neonates. A comprehensive search was conducted on key databases, including PubMed, Scopus, and Web of Science, for articles relating to the use of biochemical markers in the prediction of HPD and SGA. The findings show that changes in the levels of biomarkers in the early pregnancy phases could be an important indicator of adverse pregnancy outcomes. The literature shows that low PAPP-A (pregnancy-associated plasma protein A) and PlGF (placental growth factor) levels, low alkaline phosphatase (AP), higher sFlt-1 (soluble fms-like Tyrosine Kinase-1) levels, higher AFP (alfa fetoprotein) levels, and elevated levels of inflammatory markers such as β-HGC (free beta human chorionic gonadotropin), interferon-gamma (INF-γ), and tumor necrosis factor-α (TNF-α) may be associated with risks including the onset of HPD, fetal growth restriction (FGR), and delivery of SGA neonates. Comparatively, PAPP-A and PlGF appear to be the most important biochemical markers for the prediction of SGA and HPD.

New immune horizons in therapeutics and diagnostic approaches to Preeclampsia

Hypertensive disorders of pregnancy (HDP) are one of the commonest maladies, affecting 5%-10% of pregnancies worldwide. The American College of Obstetricians and Gynecologists (ACOG) identifies four categories of HDP, namely gestational hypertension (GH), Preeclampsia (PE), chronic hypertension (CH), and CH with superimposed PE. PE is a multisystem, heterogeneous disorder that encompasses 2%-8% of all pregnancy-related complications, contributing to about 9% to 26% of maternal deaths in low-income countries and 16% in high-income countries. These translate to 50 000 maternal deaths and over 500 000 fetal deaths worldwide, therefore demanding high priority in understanding clinical presentation, screening, diagnostic criteria, and effective management. PE is accompanied by uteroplacental insufficiency leading to vascular and metabolic changes, vasoconstriction, and end-organ ischemia. PE is diagnosed after 20 weeks of pregnancy in women who were previously normotensive or hypertensive. Besides shallow trophoblast invasion and inadequate remodeling of uterine arteries, dysregulation of the nonimmune system has been the focal point in PE. This results from aberrant immune system activation and imbalanced differentiation of T cells. Further, a failure of tolerance toward the semi-allogenic fetus results due to altered distribution of Tregs such as CD4+FoxP3+ or CD4+CD25+CD127(low) FoxP3+ cells, thereby creating a cytotoxic environment by suboptimal production of immunosuppressive cytokines like IL-10, IL-4, and IL-13. Also, intracellular production of complement protein C5a may result in decreased FoxP3+ regulatory T cells. With immune system dysfunction as a major driver in PE pathogenesis, it is logical that therapeutic targeting of components of the immune system with pharmacologic agents like anti-inflammatory and immune-modulating molecules are either being used or under clinical trial. Cholesterol synthesis inhibitors like Pravastatin may improve placental perfusion in PE, while Eculizumab (monoclonal antibody inhibiting C5) and small molecular inhibitor of C5a, Zilucoplan are under investigation. Monoclonal antibody against IL-17(Secukinumab) has been proposed to alter the Th imbalance in PE. Autologous Treg therapy and immune checkpoint inhibitors like anti-CTLA-4 are emerging as new candidates in immune horizons for PE management in the future.

Reviewing Accuracy of First Trimester Screening for Preeclampsia Using Maternal Factors and Biomarkers

Preeclampsia is a common and important complication of pregnancy, one with potentially significant morbidity and even mortality to both mother and baby. Identifying those at high risk of developing the condition is helpful as there is evidence that the incidence of preeclampsia can be reduced with low dose aspirin taken in pregnancy. Accurately predicting the risk of preeclampsia allows for more targeted aspirin prophylaxis and a greater opportunity for early detection of maternal and/or fetal complications associated with impaired placentation through a schedule of enhanced antenatal surveillance. Traditional preeclampsia prediction models use maternal characteristics and risk factors and have been shown to be of low predictive value. Multiparametric screening tests combine patient characteristics with serum biomarkers and ultrasound Doppler indices and have been shown to be more effective at detecting those at high risk of preeclampsia - more specifically, early-onset preeclampsia (onset of preeclampsia <34 weeks' gestation). Multiparametric screening has now been validated in different populations. The true cost effectiveness of a multiparametric screening model for preeclampsia screening is not yet fully known and will vary depending on the clinical setting. Despite the growing body of evidence for its improved detection rates, first trimester preeclampsia screening using multiparametric models is not widely implemented and is not part of the recommendations for antenatal screening from most international bodies. The International Federation of Gynecology and Obstetrics has advised universal preeclampsia screening using maternal risk factors and biomarkers and has strongly encouraged its promotion worldwide. Various barriers to implementation must be considered such as the immediate cost of equipment and training, the need for audit and quality control, and the expected benefit to the population. Low to middle income settings may require a pragmatic approach to the implementation of multiparametric screening given limited resources.

PREDICTION OF GESTATIONAL DIABETES AND HYPERTENSION BASED ON PREGNANCY EXAMINATION DATA

Gestational diabetes mellitus and hypertension are two common pregnancy complications, which seriously threaten the life safety of pregnant women and adversely affect the growth and development of the fetus. Therefore, it is of great significance to detect and prevent hypertension and diabetes at an early stage of pregnancy. Each pregnant woman will undergo multiple tests at different gestational weeks. This progress produces lots of pregnancy examination data. These data can reflect the dynamic changes of pregnant women’s health indicators during pregnancy. This study aims to establish gestational diabetes and hypertension prediction model with a machine learning method based on real pregnancy examination data from the hospital. We use Logistic Regression, XGBoost, LightGBM, and Neural Network Model based on LSTM to do the prediction, respectively, and compare the performance. We check the prediction accuracy at different stages of pregnancy. We found that with pregnancy examination data at all gestational weeks, the predictive AUCs for diabetes and hypertension can reach 0.92 and 0.87, respectively. At 16th gestational week, the AUCs are 0.68 for diabetes and 0.70 for hypertension. We extract the checking items which are most important and get a simplified model with a modest reduction in predictive accuracy. This study demonstrates that based on several routine pregnancy examination items we can establish a machine learning model to detect and predict gestational diabetes and hypertension. This can be used as a diagnostic aid and is conducive to early prevention and treatment.

Optimal blood pressure target to prevent severe hypertension in pregnancy: A systematic review and meta-analysis

Severe hypertension in pregnancy is a hypertensive crisis that requires urgent and intensive care due to its high maternal and fetal mortality. However, there is still a conflict of opinion on the recommendations of antihypertensive therapy. This study aimed to identify the optimal blood pressure (BP) levels to prevent severe hypertension in pregnant women with nonsevere hypertension. Ovid MEDLINE and the Cochrane Library were searched, and only randomized controlled trials (RCTs) were included if they compared the effects of antihypertensive drugs and placebo/no treatment or more intensive and less intensive BP-lowering treatments in nonsevere hypertensive pregnant patients. A random effects model meta-analysis was performed to estimate the pooled risk ratio (RR) for the outcomes. Forty RCTs with 6355 patients were included in the study. BP-lowering treatment significantly prevented severe hypertension (RR, 0.46; 95% CI, 0.37-0.56), preeclampsia (RR, 0.82; 95% CI, 0.69-0.98), severe preeclampsia (RR, 0.38; 95% CI, 0.17-0.84), placental abruption (RR, 0.52; 95% CI, 0.32-0.86), and preterm birth (< 37 weeks; RR, 0.81; 95% CI, 0.71-0.93), while the risk of small for gestational age infants was increased (RR, 1.25; 95% CI, 1.02-1.54). An achieved systolic blood pressure (SBP) of < 130 mmHg reduced the risk of severe hypertension to nearly one-third compared with an SBP of ≥ 140 mmHg, with a significant interaction of the BP levels achieved with BP-lowering therapy. There was no significant interaction between the subtypes of hypertensive disorders of pregnancy and BP-lowering treatment, except for placental abruption. BP-lowering treatment aimed at an SBP < 130 mmHg and accompanied by the careful monitoring of fetal growth might be recommended to prevent severe hypertension.

A risk model that combines MAP, PlGF, and PAPP-A in the first trimester of pregnancy to predict hypertensive disorders of pregnancy

Whether the first trimester maternal mean arterial pressure (MAP), placental growth factor (PlGF), and pregnancy-associated plasma protein A (PAPP-A) can predict hypertensive disorders of pregnancy (HDP) is unclear. We conducted a retrospective case-control study with the total population of 539 gravidas, of these 447 had normal pregnancy, 27 had gestational hypertension (GH), 36 had preeclampsia (PE), and 29 had preeclampsia with severe features (SPE). Prediction for HDP was determined by the area under curve (AUC). Compared to the healthy group, the multiple of the median (MoM) for MAP was increased in the study groups, while PlGF and PAPP-A were decreased. When the cutoff values for MAP, PlGF, and PAPP-A were 1.069, 0.769, and 0.673 MoM, respectively, the sensitivities for predicting HDP were 0.517, 0.446, and 0.500 and the specificities were 0.744, 0.826, and 0.769, respectively. To predict GH, the highest AUC was 0.755 (95% CI: 0.655-0.856, p < 0.001) based on MAP, PlGF, and PAPP-A. The combined PlGF and PAPP-A had the highest AUC (0.683 [95% CI: 0.584-0.782, p < 0.001] and 0.755 [95% CI: 0.682-0.829, p < 0.001]) for prediction of PE and SPE. We found that MAP, serum levels of PlGF, and PAPP-A in the first trimester pregnancy are markers that predict HDP in the third trimester. The combination of markers is far superior to single markers alone. To improve the diagnostic value, specific cutoff values should be applied to GH, PE, SPE in each condition.

Systematic review of prediction models for gestational hypertension and preeclampsia

INTRODUCTION

Prediction models for gestational hypertension and preeclampsia have been developed with data and assumptions from developed countries. Their suitability and application for low resource settings have not been tested. This review aimed to identify and assess the methodological quality of prediction models for gestational hypertension and pre-eclampsia with reference to their application in low resource settings.

METHODS

Using combinations of keywords for gestational hypertension, preeclampsia and prediction models seven databases were searched to identify prediction models developed with maternal data obtained before 20 weeks of pregnancy and including at least three predictors (Prospero registration CRD 42017078786). Prediction model characteristics and performance measures were extracted using the CHARMS, STROBE and TRIPOD checklists. The National Institute of Health quality assessment tools for observational cohort and cross-sectional studies were used for study quality appraisal.

RESULTS

We retrieved 8,309 articles out of which 40 articles were eligible for review. Seventy-seven percent of all the prediction models combined biomarkers with maternal clinical characteristics. Biomarkers used as predictors in most models were pregnancy associated plasma protein-A (PAPP-A) and placental growth factor (PlGF). Only five studies were conducted in a low-and middle income country.

CONCLUSIONS

Most of the studies evaluated did not completely follow the CHARMS, TRIPOD and STROBE guidelines in prediction model development and reporting. Adherence to these guidelines will improve prediction modelling studies and subsequent application of prediction models in clinical practice. Prediction models using maternal characteristics, with good discrimination and calibration, should be externally validated for use in low and middle income countries where biomarker assays are not routinely available.

Early prediction of preeclampsia via machine learning

BACKGROUND

Early prediction of preeclampsia is challenging because of poorly understood causes, various risk factors, and likely multiple pathogenic phenotypes of preeclampsia. Statistical learning methods are well-equipped to deal with a large number of variables, such as patients' clinical and laboratory data, and to select the most informative features automatically.

OBJECTIVE

Our objective was to use statistical learning methods to analyze all available clinical and laboratory data that were obtained during routine prenatal visits in early pregnancy and to use them to develop a prediction model for preeclampsia.

STUDY DESIGN

This was a retrospective cohort study that used data from 16,370 births at Lucile Packard Children Hospital at Stanford, CA, from April 2014 to January 2018. Two statistical learning algorithms were used to build a predictive model: (1) elastic net and (2) gradient boosting algorithm. Models for all preeclampsia and early-onset preeclampsia (<34 weeks gestation) were fitted with the use of patient data that were available at <16 weeks gestational age. The 67 variables that were considered in the models included maternal characteristics, medical history, routine prenatal laboratory results, and medication intake. The area under the receiver operator curve, true-positive rate, and false-positive rate were assessed via cross-validation.

RESULTS

Using the elastic net algorithm, we developed a prediction model that contained a subset of the most informative features from all variables. The obtained prediction model for preeclampsia yielded an area under the curve of 0.79 (95% confidence interval, 0.75-0.83), sensitivity of 45.2%, and false-positive rate of 8.1%. The prediction model for early-onset preeclampsia achieved an area under the curve of 0.89 (95% confidence interval, 0.84-0.95), true-positive rate of 72.3%, and false-positive rate of 8.8%.

CONCLUSION

Statistical learning methods in a retrospective cohort study automatically identified a set of significant features for prediction and yielded high prediction performance for preeclampsia risk from routine early pregnancy information.

High Neutrophil-to-Lymphocyte Ratio Is an Early Predictor of Bronchopulmonary Dysplasia

Background and Objective: Bronchopulmonary dysplasia (BPD) is a common complication in preterm infants; predicting the degree of BPD at an early life stage is difficult. Inflammation is a crucial risk factor for BPD pathogenesis, and the neutrophil-to-lymphocyte ratio (NLR) is a potential systemic inflammatory biomarker. We aimed to assess the predictive value of the NLR for BPD. Methods: We carried out a retrospective, single-center, observational study of neonates with gestational ages (GAs) <32 weeks and assessed the association between the NLR and BPD. Results: The study population included 296 preterm infants with BPD (n = 144) or without BPD (n = 152). Among the infants, 75 (25.3%) had mild BPD, 37 (12.5%) had moderate BPD, and 32 (10.8%) had severe BPD. The BPD group had a higher NLR at birth and at 72 h than the non-BPD group. The NLR cutoff value at 72 h for the prediction of BPD was 3.035 (sensitivity = 0.519, specificity = 0.964), and the area under the curve (AUC) was 0.714. The NLR cutoff value at 72 h for predicting severe BPD was 3.105 (sensitivity = 0.607, specificity = 0.819), with an AUC of 0.756. At the NLR cutoff value at 72 for the prediction of BPD, the AUCs were 0.640 and 0.970 in the preterm infants with EOS and congenital pneumonia, respectively. Conclusions: The NLR is an inexpensive, accessible and convenient tool; an increase in the NLR at 72 h could be an early predictor of BPD, especially severe BPD. Additionally, the NLR at 72 h could be a predictor of BPD in preterm infants with intrauterine infections.