Prediction of pre-eclampsia-related complications in women with suspected or confirmed pre-eclampsia: development and internal validation of clinical prediction model

Predicting intensive care need in women with preeclampsia using machine learning - a pilot study

BACKGROUND

Predicting severe preeclampsia with need for intensive care is challenging. To better predict high-risk pregnancies to prevent adverse outcomes such as eclampsia is still an unmet need worldwide. In this study we aimed to develop a prediction model for severe outcomes using routine biomarkers and clinical characteristics.

METHODS

We used machine learning models based on data from an intensive care cohort with severe preeclampsia (n=41) and a cohort of preeclampsia controls (n=40) with the objective to find patterns for severe disease not detectable with traditional logistic regression models.

RESULTS

The best model was generated by including the laboratory parameters aspartate aminotransferase (ASAT), uric acid and body mass index (BMI) with a cross-validation accuracy of 0.88 and an area under the curve (AUC) of 0.91. Our model was internally validated on a test-set where the accuracy was lower, 0.82, with an AUC of 0.85.

CONCLUSION

The clinical routine blood parameters ASAT and uric acid as well as BMI, were the parameters most indicative of severe disease. Aspartate aminotransferase reflects liver involvement, uric acid might be involved in several steps of the pathophysiologic process of preeclampsia, and obesity is a well-known risk factor for development of both severe and non-severe preeclampsia likely involving inflammatory pathways..[Figure: see text].

Angiogenic imbalance in pre-eclampsia and fetal growth restriction: enhanced soluble fms-like tyrosine kinase-1 binding or diminished production of placental growth factor?

OBJECTIVES

To assess levels of total placental growth factor (PlGF), soluble fms-like tyrosine kinase-1 (sFlt-1) and free PlGF in women with pre-eclampsia (PE) with or without a small-for-gestational-age (SGA) neonate in order to establish whether low free PlGF levels associated with PE and SGA are due to enhanced sFlt-1 binding or decreased PlGF production.

METHODS

This was a secondary analysis of a prospective multicenter cohort study involving 407 pregnancies with suspected or confirmed PE, in which total PlGF levels were calculated from measured sFlt-1 and free PlGF levels. The control group included women who were suspected to have PE at a certain point in pregnancy but did not develop PE. The analysis was stratified according to whether PE was early- or late-onset (gestational age < 34 weeks vs ≥ 34 weeks) and according to the presence of SGA at birth, which was used as a proxy of fetal growth restriction in the absence of Doppler ultrasound and biometric data.

RESULTS

In early-onset PE, both women with and those without SGA had lower free (19 and 45 pg/mL) and total (44 and 100 pg/mL) PlGF levels compared with women without PE (free and total PlGF, 300 and 381 pg/mL, respectively). SGA alone did not affect free and total PlGF in this condition (free and total PlGF, 264 and 352 pg/mL, respectively). Observations in women with late-onset PE were similar, although the changes were more modest. Both SGA (gestational age < 34 weeks) and PE were individually associated with increased sFlt-1 and, in women with both PE and SGA, the upregulation of sFlt-1 occurred in a synergistic manner, thus resulting in the highest sFlt-1/free PlGF ratio in this group. This occurred in both early- and late-onset PE.

CONCLUSIONS

Particularly in pregnancies with early-onset PE and SGA, diminished PlGF production is an important cause of low free PlGF levels. Under such conditions, sFlt-1 lowering is unlikely to restore the angiogenic balance. © 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Comparison of the Soluble fms-Like Tyrosine Kinase 1/Placental Growth Factor Ratio Alone versus a Multi-Marker Regression Model for the Prediction of Preeclampsia-Related Adverse Outcomes after 34 Weeks of Gestation

INTRODUCTION

The objective of this retrospective study was to compare the predictive performance of the soluble fms-like tyrosine kinase 1 (sFlt-1)/placental growth factor (PlGF) ratio alone or in a multi-marker regression model for preeclampsia-related maternal and/or fetal adverse outcomes in women >34 weeks of gestation.

METHODS

We analyzed the data collected from 655 women with suspected preeclampsia. Adverse outcomes were predicted by multivariable and univariable logistic regression models. The outcome of patients was evaluated within 14 days after presentation with signs and symptoms of preeclampsia or diagnosed preeclampsia.

RESULTS

The full model integrating available, standard clinical information and the sFlt-1/PlGF ratio had the best predictive performance for adverse outcomes with an AUC of 72.6%, which corresponds to a sensitivity of 73.3% and specificity of 66.0%. The positive predictive value of the full model was 51.4%, and the negative predictive value was 83.5%. 24.5% of patients, who did not experience adverse outcomes but were classified as high risk by sFlt-1/PlGF ratio (≥38), were correctly classified by the regression model. The sFlt-1/PlGF ratio alone had a significantly lower AUC of 65.6%.

CONCLUSIONS

Integrating angiogenic biomarkers in a regression model improved the prediction of preeclampsia-related adverse outcomes in women at risk after 34 weeks of gestation.

Clinical utility of sFlt-1 and PlGF in screening, prediction, diagnosis and monitoring of pre-eclampsia and fetal growth restriction

Pre-eclampsia (PE) is characterized by placental and maternal endothelial dysfunction, and associated with fetal growth restriction (FGR), placental abruption, preterm delivery and stillbirth. The angiogenic factors soluble fms-like tyrosine kinase-1 (sFlt-1) and placental growth factor (PlGF) are altered in pregnancies complicated by placenta-related disorders. In this Review, we summarize the existing knowledge, examining the performance of maternal PlGF, sFlt-1 and the sFlt-1/PlGF ratio for screening PE, predicting development of PE in the short term, diagnosing PE, monitoring established PE and predicting other placenta-related disorders in singleton pregnancy. We also discuss the performance of PlGF and the sFlt-1/PlGF ratio for predicting PE in twin pregnancy. For first-trimester screening in singleton pregnancy, a more accurate way of identifying high-risk women than current practice is to combine maternal PlGF levels with clinical risk factors and ultrasound markers. Later in pregnancy, the sFlt-1/PlGF ratio has advantages over PlGF because it has a higher pooled sensitivity and specificity for diagnosing and monitoring PE. It has clinical value because it can rule out the development of PE in the 1-4-week period after the test. Once a diagnosis of PE is established, repeat measurement of sFlt-1 and PlGF can help monitor progression of the condition and may inform clinical decision-making regarding the optimal time for delivery. The sFlt-1/PlGF ratio is useful for predicting FGR and preterm delivery, but the association between stillbirth and the angiogenic factors is unclear. The sFlt-1/PlGF ratio can be used to predict PE in twin pregnancy, although different sFlt-1/PlGF ratio cut-offs from those for singleton pregnancy should be applied for optimal performance. In summary, PlGF, sFlt-1 and the sFlt-1/PlGF ratio are useful for screening, diagnosing, predicting and monitoring placenta-related disorders in singleton and twin pregnancy. We propose that tests for these angiogenic factors are integrated more fully into clinical practice.© 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Prediction of pre-eclampsia complicated by fetal growth restriction and its perinatal outcome based on an artificial neural network model

Objective: Pre-eclampsia (PE) complicated by fetal growth restriction (FGR) increases both perinatal mortality and the incidence of preterm birth and neonatal asphyxia. Because ultrasound measurements are bone markers, soft tissues, such as fetal fat and muscle, are ignored, and the selection of section surface and the influence of fetal position can lead to estimation errors. The early detection of FGR is not easy, resulting in a relative delay in intervention. It is assumed that FGR complicated with PE can be predicted by laboratory and clinical indicators. The present study adopts an artificial neural network (ANN) to assess the effect and predictive value of changes in maternal peripheral blood parameters and clinical indicators on the perinatal outcomes in patients with PE complicated by FGR.

Methods: This study used a retrospective case-control approach. The correlation between maternal peripheral blood parameters and perinatal outcomes in pregnant patients with PE complicated by FGR was retrospectively analyzed, and an ANN was constructed to assess the value of the changes in maternal blood parameters in predicting the occurrence of PE complicated by FGR and adverse perinatal outcomes.

Results: A total of 15 factors—maternal age, pre-pregnancy body mass index, inflammatory markers (neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio), coagulation parameters (prothrombin time and thrombin time), lipid parameters (high-density lipoprotein, low-density lipoprotein, and triglyceride counts), platelet parameters (mean platelet volume and plateletcrit), uric acid, lactate dehydrogenase, and total bile acids—were correlated with PE complicated by FGR. A total of six ANNs were constructed with the adoption of these parameters. The accuracy, sensitivity, and specificity of predicting the occurrence of the following diseases and adverse outcomes were respectively as follows: 84.3%, 97.7%, and 78% for PE complicated by FGR; 76.3%, 97.3%, and 68% for provider-initiated preterm births,; 81.9%, 97.2%, and 51% for predicting the severity of FGR; 80.3%, 92.9%, and 79% for premature rupture of membranes; 80.1%, 92.3%, and 79% for postpartum hemorrhage; and 77.6%, 92.3%, and 76% for fetal distress.

Conclusion: An ANN model based on maternal peripheral blood parameters has a good predictive value for the occurrence of PE complicated by FGR and its adverse perinatal outcomes, such as the severity of FGR and preterm births in these patients.

Dynamic gestational week prediction model for pre-eclampsia based on ID3 algorithm

Pre-eclampsia (PE) is a type of hypertensive disorder during pregnancy, which is a serious threat to the life of mother and fetus. It is a placenta-derived disease that results in placental damage and necrosis due to systemic small vessel spasms that cause pathological changes such as ischemia and hypoxia and oxidative stress, which leads to fetal and maternal damage. In this study, four types of risk factors, namely, clinical epidemiology, hemodynamics, basic biochemistry, and biomarkers, were used for the initial selection of model parameters related to PE, and factors that were easily available and clinically recognized as being associated with a higher risk of PE were selected based on hospital medical record data. The model parameters were then further analyzed and screened in two subgroups: early-onset pre-eclampsia (EOPE) and late-onset pre-eclampsia (LOPE). Dynamic gestational week prediction model for PE using decision tree ID3 algorithm in machine learning. Performance of the model was: macro average (precision = 76%, recall = 73%, F1-score = 75%), weighted average (precision = 88%, recall = 89%, F1-score = 89%) and overall accuracy is 86%. In this study, the addition of the dynamic timeline parameter "gestational week" made the model more convenient for clinical application and achieved effective PE subgroup prediction.

Development and evaluation of a nomogram for adverse outcomes of preeclampsia in Chinese pregnant women

Objective

Preeclampsia, the main cause of maternal and perinatal deaths, is associated with several maternal complications and adverse perinatal outcomes. Some prediction models are uesd to evaluate adverse pregnancy outcomes. However, some of the current prediction models are mainly carried out in developed countries, and many problems are still exist. We, thus, developed and validated a nomogram to predict the risk of adverse pregnancy outcomes of preeclampsia in Chinese pregnant women.

Methods

The clinical data of 720 pregnant women with preeclampsia in seven medical institutions in Chongqing from January 1, 2010, to December 31, 2020, were analyzed retrospectively. The patients were divided into two groups: 180 cases (25%) with adverse outcomes and 540 cases (75%) without adverse outcomes. The indicators were identified via univariate analysis. Logistic regression analysis was used to establish the prediction model, which was displayed by a nomogram. The performance of the nomogram was evaluated in terms of the area under the receiver operating characteristic (ROC) curve, calibration, and clinical utility.

Results

Univariate analysis showed that 24 indicators were significantly different (P < 0.05). Logistic regression analysis showed that gestational age, 24 h urine protein qualitative, and TT were significantly different (P < 0.05). The area under the ROC curve was 0.781 (95% CI 0.737–0.825) in training set and 0.777 (95% CI 0.689–0.865) in test set. The calibration curve of the nomogram showed good agreement between prediction and observation. The analysis of the clinical decision curve showed that the nomogram is of practical significance.

Conclusion

Our study identified gestational age, 24 h urine protein qualitative, and TT as risk factors for adverse outcomes of preeclampsia in pregnant women, and constructed a nomogram that can easily predict and evaluate the risk of adverse pregnancy outcomes in women with preeclampsia.

Two-Dimensional Ultrasound and Triplane Tissue Doppler Ultrasound of Patients with Severe Preeclampsia

This study was to investigate the cardiac function characteristics under two-dimensional ultrasound and triplane tissue Doppler imaging (TDI) of patients with severe preeclampsia (SPE). 28 SPE patients with singleton pregnancy from January 2018 to December 2020 were included in the SPE group. 25 healthy nonpregnant women of reproductive age were taken as the control group (Ctrl group), and 26 normal pregnant women with singleton pregnancy were selected as the normal group (Norm group); all the research objects underwent ultrasonography. The morphological and functional indexes of left and right ventricles were compared among the cases in different groups. The results showed that the left ventricular end-diastolic period diameter (LVEDd), left ventricular relative wall thickness (LV-RWT), left ventricular mass index (LVMi), left anterior descending (LAd), left ventricular $E/e$ and $e/a$ values, right ventricular diameter (RV-D), right ventricular anterior wall thickness (RVAW), $a$ value, right atrial septum (RA-S), pulmonary artery systolic pressure (PASP), left ventricular end-systolic period diameter (LVEds), interventricular septal thickness (IVSd), posterior wall thickness (PWd), end-diastolic period volume (EVD), end-systolic period volume (ESV), relative wall thickness (RWT), sphericity index (SpI), left atrium volume index (LAVi), and $E/e$ value of patients in the SPE group were higher than those in the Ctrl group and the Norm group ( $P<0.05$ ). The mitral annular plane systolic excursion (MAPSE), $s$ value, tricuspid annual plane systolic excursion (TAPSE), ratio of early diastolic blood flow velocity to late diastolic blood flow velocity ( $E/A$ ), ratio of peak early diastolic velocity to peak late diastolic velocity ( $e/a$ ), peak early diastolic velocity ( $e$ ), and ejection fraction (EF) of the SPE group were lower than those of the Ctrl group and the Norm group ( $P<0.05$ ). The ratio of mitral valve early diastolic blood flow velocity to peak early diastolic velocity ( $E/e$ ) of the Norm group was higher than that of the Ctrl group ( $P<0.05$ ). In two-dimensional ultrasound of the SPE group, the maximum difference in time from the start to the peak of systole (Ts) of the right ventricle between the basal and middle segments of the lateral wall and that of interventricular septum (RV-Ts-max) was $31.56\pm 0.39\%$ . The maximum difference in time to peak of early diastole (Te) under the same condition (RV-Te-max) was $47.16\pm 0.19\%$ . Left ventricular LV-Ts-max and LV-Te-max were $9.83\pm 0.80\%$ and $8.37\pm 0.68\%$ , respectively, in triplane TDI, which were considerably higher than those in the Ctrl and Norm groups ( $P<0.05$ ). It suggested that two-dimensional ultrasound and triplane TDI could reflect the ventricular morphology as well as diastolic and systolic function injury in patients, which offered a reference basis for the diagnosis of SPE.

Performance of sFlt-1/PIGF Ratio for the Prediction of Perinatal Outcome in Obese Pre-Eclamptic Women

Obese women are at high risk of developing pre-eclampsia (PE). As an altered angiogenic profile is characteristic for PE, measurement of soluble fms-like tyrosine kinase-1 (sFlt-1)/placental growth factor (PIGF) ratio in the maternal serum can be helpful for PE diagnosis, as well as for adverse perinatal outcome (APO) prediction. There is growing evidence that obesity might influence the level of sFlt-1/PIGF and, therefore, the aim of the study was the evaluation of sFlt-1/PIGF as an APO predictor in obese women with PE. Pre-eclamptic women who had an sFlt-1/PIGF measurement at the time of diagnosis were retrospectively included. Women were classified according to their pre-pregnancy body mass index (BMI) as normal weight (BMI < 25 kg/m2), overweight (BMI > 25−29.9 kg/m2) or obese (BMI ≥ 30 kg/m2). APO was defined as the occurrence of one of the following outcomes: Small for gestational age, defined as a birthweight < 3rd centile, neonatal mortality, neonatal seizures, admission to neonatal unit required (NICU) or respiratory support. A total of 141 women were included. Of them, 28 (20%) patients were obese. ROC (receiver operating characteristic) analysis revealed a high predictive value for sFlt-1/PIGF and APO across the whole study cohort (AUC = 0.880, 95% CI: 0.826−0.936; p < 0.001). However, the subgroup of obese women showed a significantly lower level of sFlt-1 and, therefore, the performance of sFlt-1/PIGF as APO predictor was poorer compared to normal or overweight PE women (AUC = 0.754, 95% CI: 0.552−0.956, p = 0.025). In contrast to normal or overweight women, a ratio of sFlt-1/PIGF < 38 could not rule out APO in women with obesity.

Establishment of a nomogram model for predicting adverse outcomes in advanced-age pregnant women with preterm preeclampsia

Aim

To establish a model for predicting adverse outcomes in advanced-age pregnant women with preterm preeclampsia in China.

Methods

We retrospectively collected the medical records of 896 pregnant women with preterm preeclampsia who were older than 35 years and delivered at the Affiliated Hospital of Qingdao University from June 2018 to December 2020. The pregnant women were divided into an adverse outcome group and a non-adverse outcome group according to the occurrence of adverse outcomes. The data were divided into a training set and a verification set at a ratio of 8:2. A nomogram model was developed according to a binary logistic regression model created to predict the adverse outcomes in advanced-age pregnant women with preterm preeclampsia. ROC curves and their AUCs were used to evaluate the predictive ability of the model. The model was internally verified by using 1000 bootstrap samples, and a calibration diagram was drawn.

Results

Binary logistic regression analysis showed that platelet count (PLT), uric acid (UA), blood urea nitrogen (BUN), prothrombin time (PT), and lactate dehydrogenase (LDH) were the factors that independently influenced adverse outcomes (P < 0.05). The AUCs of the internal and external verification of the model were 0.788 (95% CI: 0.737 ~ 0.764) and 0.742 (95% CI: 0.565 ~ 0.847), respectively. The calibration curve was close to the diagonal.

Conclusions

The model we constructed can accurately predict the risk of adverse outcomes of pregnant women of advanced age with preterm preeclampsia, providing corresponding guidance and serving as a basis for preventing adverse outcomes and improving clinical treatment and maternal and infant prognosis.

A machine-learning-based algorithm improves prediction of preeclampsia-associated adverse outcomes

BACKGROUND

Preeclampsia presents a highly prevalent burden on pregnant women with an estimated incidence of 2% to 5%. Preeclampsia increases the maternal risk of death 20-fold and is one of the main causes of perinatal morbidity and mortality. Novel biomarkers, such as soluble fms-like tyrosine kinase-1 and placental growth factor in addition to a wide span of conventional clinical data (medical history, physical symptoms, laboratory parameters, etc.), present an excellent basis for the application of early-detection machine-learning models.

OBJECTIVE

This study aimed to develop, train, and test an automated machine-learning model for the prediction of adverse outcomes in patients with suspected preeclampsia.

STUDY DESIGN

Our real-world dataset of 1647 (2472 samples) women was retrospectively recruited from women who presented to the Department of Obstetrics at the Charité - Universitätsmedizin Berlin, Berlin, Germany, between July 2010 and March 2019. After standardization and data cleaning, we calculated additional features regarding the biomarkers soluble fms-like tyrosine kinase-1 and placental growth factor and sonography data (umbilical artery pulsatility index, middle cerebral artery pulsatility index, mean uterine artery pulsatility index), resulting in a total of 114 features. The target metric was the occurrence of adverse outcomes throughout the remaining pregnancy and 2 weeks after delivery. We trained 2 different models, a gradient-boosted tree and a random forest classifier. Hyperparameter training was performed using a grid search approach. All results were evaluated via a 10 × 10-fold cross-validation regimen.

RESULTS

We obtained metrics for the 2 naive machine-learning models. A gradient-boosted tree model was performed with a positive predictive value of 88%±6%, a negative predictive value of 89%±3%, a sensitivity of 66%±5%, a specificity of 97%±2%, an overall accuracy of 89%±3%, an area under the receiver operating characteristic curve of 0.82±0.03, an F1 score of 0.76±0.04, and a threat score of 0.61±0.05. The random forest classifier returned an equal positive predictive value (88%±6%) and specificity (97%±1%) while performing slightly inferior on the other available metrics. Applying differential cutoffs instead of a naive cutoff for positive prediction at ≥0.5 for the classifier's results yielded additional increases in performance.

CONCLUSION

Machine-learning techniques were a valid approach to improve the prediction of adverse outcomes in pregnant women at high risk of preeclampsia vs current clinical standard techniques. Furthermore, we presented an automated system that did not rely on manual tuning or adjustments.