Validation and development of models using clinical, biochemical and ultrasound markers for predicting pre-eclampsia: an individual participant data meta-analysis.

BACKGROUND: Pre-eclampsia is a leading cause of maternal and perinatal mortality and morbidity. Early identification of women at risk is needed to plan management. OBJECTIVES: To assess the performance of existing pre-eclampsia prediction models and to develop and validate models for pre-eclampsia using individual participant data meta-analysis. We also estimated the prognostic value of individual markers. DESIGN: This was an individual participant data meta-analysis of cohort studies. SETTING: Source data from secondary and tertiary care. PREDICTORS: We identified predictors from systematic reviews, and prioritised for importance in an international survey. PRIMARY OUTCOMES: Early-onset (delivery at < 34 weeks' gestation), late-onset (delivery at ≥ 34 weeks' gestation) and any-onset pre-eclampsia. ANALYSIS: We externally validated existing prediction models in UK cohorts and reported their performance in terms of discrimination and calibration. We developed and validated 12 new models based on clinical characteristics, clinical characteristics and biochemical markers, and clinical characteristics and ultrasound markers in the first and second trimesters. We summarised the data set-specific performance of each model using a random-effects meta-analysis. Discrimination was considered promising for C-statistics of ≥ 0.7, and calibration was considered good if the slope was near 1 and calibration-in-the-large was near 0. Heterogeneity was quantified using I2 and τ2. A decision curve analysis was undertaken to determine the clinical utility (net benefit) of the models. We reported the unadjusted prognostic value of individual predictors for pre-eclampsia as odds ratios with 95% confidence and prediction intervals. RESULTS: The International Prediction of Pregnancy Complications network comprised 78 studies (3,570,993 singleton pregnancies) identified from systematic reviews of tests to predict pre-eclampsia. Twenty-four of the 131 published prediction models could be validated in 11 UK cohorts. Summary C-statistics were between 0.6 and 0.7 for most models, and calibration was generally poor owing to large between-study heterogeneity, suggesting model overfitting. The clinical utility of the models varied between showing net harm to showing minimal or no net benefit. The average discrimination for IPPIC models ranged between 0.68 and 0.83. This was highest for the second-trimester clinical characteristics and biochemical markers model to predict early-onset pre-eclampsia, and lowest for the first-trimester clinical characteristics models to predict any pre-eclampsia. Calibration performance was heterogeneous across studies. Net benefit was observed for International Prediction of Pregnancy Complications first and second-trimester clinical characteristics and clinical characteristics and biochemical markers models predicting any pre-eclampsia, when validated in singleton nulliparous women managed in the UK NHS. History of hypertension, parity, smoking, mode of conception, placental growth factor and uterine artery pulsatility index had the strongest unadjusted associations with pre-eclampsia. LIMITATIONS: Variations in study population characteristics, type of predictors reported, too few events in some validation cohorts and the type of measurements contributed to heterogeneity in performance of the International Prediction of Pregnancy Complications models. Some published models were not validated because model predictors were unavailable in the individual participant data. CONCLUSION: For models that could be validated, predictive performance was generally poor across data sets. Although the International Prediction of Pregnancy Complications models show good predictive performance on average, and in the singleton nulliparous population, heterogeneity in calibration performance is likely across settings. FUTURE WORK: Recalibration of model parameters within populations may improve calibration performance. Additional strong predictors need to be identified to improve model performance and consistency. Validation, including examination of calibration heterogeneity, is required for the models we could not validate. STUDY REGISTRATION: This study is registered as PROSPERO CRD42015029349. FUNDING: This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 24, No. 72. See the NIHR Journals Library website for further project information.

WHAT IS THE PROBLEM?: Pre-eclampsia, a condition in pregnancy that results in raised blood pressure and protein in the urine, is a major cause of complications for the mother and baby. WHAT IS NEEDED?: A way of accurately identifying women at high risk of pre-eclampsia to allow clinicians to start preventative interventions such as administering aspirin or frequently monitoring women during pregnancy. WHERE ARE THE RESEARCH GAPS?: Although over 100 tools (models) have been reported worldwide to predict pre-eclampsia, to date their performance in women managed in the UK NHS is unknown. WHAT DID WE PLAN TO DO?: We planned to comprehensively identify all published models that predict the risk of pre-eclampsia occurring at any time during pregnancy and to assess if this prediction is accurate in the UK population. If the existing models did not perform satisfactorily, we aimed to develop new prediction models. WHAT DID WE FIND?: We formed the International Prediction of Pregnancy Complications network, which provided data from a large number of studies (78 studies, 25 countries, 125 researchers, 3,570,993 singleton pregnancies). We were able to assess the performance of 24 out of the 131 models published to predict pre-eclampsia in 11 UK data sets. The models did not accurately predict the risk of pre-eclampsia across all UK data sets, and their performance varied within individual data sets. We developed new prediction models that showed promising performance on average across all data sets, but their ability to correctly identify women who develop pre-eclampsia varied between populations. The models were more clinically useful when used in the care of first-time mothers pregnant with one child, compared to a strategy of treating them all as if they were at high-risk of pre-eclampsia. WHAT DOES THIS MEAN?: Before using the International Prediction of Pregnancy Complications models in various populations, they need to be adjusted for characteristics of the particular population and the setting of application.

External validation of prognostic models predicting pre-eclampsia: individual participant data meta-analysis.

BACKGROUND: Pre-eclampsia is a leading cause of maternal and perinatal mortality and morbidity. Early identification of women at risk during pregnancy is required to plan management. Although there are many published prediction models for pre-eclampsia, few have been validated in external data. Our objective was to externally validate published prediction models for pre-eclampsia using individual participant data (IPD) from UK studies, to evaluate whether any of the models can accurately predict the condition when used within the UK healthcare setting. METHODS: IPD from 11 UK cohort studies (217,415 pregnant women) within the International Prediction of Pregnancy Complications (IPPIC) pre-eclampsia network contributed to external validation of published prediction models, identified by systematic review. Cohorts that measured all predictor variables in at least one of the identified models and reported pre-eclampsia as an outcome were included for validation. We reported the model predictive performance as discrimination (C-statistic), calibration (calibration plots, calibration slope, calibration-in-the-large), and net benefit. Performance measures were estimated separately in each available study and then, where possible, combined across studies in a random-effects meta-analysis. RESULTS: Of 131 published models, 67 provided the full model equation and 24 could be validated in 11 UK cohorts. Most of the models showed modest discrimination with summary C-statistics between 0.6 and 0.7. The calibration of the predicted compared to observed risk was generally poor for most models with observed calibration slopes less than 1, indicating that predictions were generally too extreme, although confidence intervals were wide. There was large between-study heterogeneity in each model's calibration-in-the-large, suggesting poor calibration of the predicted overall risk across populations. In a subset of models, the net benefit of using the models to inform clinical decisions appeared small and limited to probability thresholds between 5 and 7%. CONCLUSIONS: The evaluated models had modest predictive performance, with key limitations such as poor calibration (likely due to overfitting in the original development datasets), substantial heterogeneity, and small net benefit across settings. The evidence to support the use of these prediction models for pre-eclampsia in clinical decision-making is limited. Any models that we could not validate should be examined in terms of their predictive performance, net benefit, and heterogeneity across multiple UK settings before consideration for use in practice. TRIAL REGISTRATION: PROSPERO ID: CRD42015029349 .

Early prediction of gestational diabetes: a practical model combining clinical and biochemical markers.

BACKGROUND: Gestational diabetes (GDM) is usually diagnosed late in pregnancy, precluding early preventive interventions. This study aims to develop a predictive model based on clinical factors and selected biochemical markers for the early risk assessment of GDM. METHODS: Based on a prospective cohort of 7929 pregnant women from the Quebec City metropolitan area, a nested case-control study was performed including 264 women who developed GDM. Each woman who developed GDM was matched with two women with normal glycemic profile. Risk prediction models for GDM and GDM requiring insulin therapy were developed using multivariable logistic regression analyses, based on clinical characteristics and the measurement of three clinically validated biomarkers: glycated hemoglobin (HbA1c), sex hormone binding globulin (SHBG) and high-sensitivity C-reactive protein (hsCRP) measured between 14 and 17 weeks of gestation. RESULTS: HbA1c and hsCRP were higher and SHBG was lower in women who developed GDM (p<0.001). The selected model for the prediction of GDM, based on HbA1c, SHBG, BMI, past history of GDM, family history of diabetes and soft drink intake before pregnancy yielded an area under the ROC curve (AUC) of 0.79 (0.75-0.83). For the prediction of GDM requiring insulin therapy, the selected model including the same six variables yielded an AUC of 0.88 (0.84-0.92) and a sensitivity of 68.9% at a false-positive rate of 10%. CONCLUSIONS: A simple model based on clinical characteristics and biomarkers available early in pregnancy could allow the identification of women at risk of developing GDM, especially GDM requiring insulin therapy.

Soluble Fms-like tyrosine kinase-1 to placental growth factor ratio in mid-pregnancy as a predictor of preterm preeclampsia in asymptomatic pregnant women.

BACKGROUND: This study aims to evaluate the performance of the soluble Fms-like tyrosine kinase-1/placental growth factor (sFlt-1/PlGF) ratio to predict early-onset, preterm and severe preeclampsia at mid-pregnancy in asymptomatic women. METHODS: Based on a prospective cohort of 7929 pregnant women from the Quebec City metropolitan area, a nested case-control study was performed including 111 women who developed preeclampsia and 69 women with gestational hypertension matched with 338 normotensive women. Serum sFlt-1 and PlGF were measured between 20 and 32 weeks of gestation. The performance of the sFlt-1/PlGF ratio, expressed as raw values and multiples of the median (MoM) for the prediction of early-onset, preterm and severe preeclampsia was evaluated. RESULTS: Women who developed preeclampsia had significantly higher MoM sFlt-1/PlGF ratio (p<0.001). In the early-onset preeclampsia group, the median of the MoM distribution was 24.0 and the area under the receiver operating characteristic curve (AUC) was 0.977, giving sensitivities of 77.8% and 88.9% at false-positive rates of 5% and 10%. Positive predictive values (PPV) were 2.5% and 1.5%, respectively. In a subset between 26 and 32 weeks of gestation, at a threshold of 30, the sFlt-1/PlGF ratio yielded 100% specificity and identified, respectively, 85.7% and 65.2% of women who developed early-onset and preterm preeclampsia. CONCLUSIONS: The sFlt-1/PlGF ratio has the potential to predict early-onset and preterm preeclampsia at mid-pregnancy in asymptomatic women. However, care must be paid to the prevalence of early-onset preeclampsia in the population since low prevalence reduces PPV and may hamper clinical utility.

Validation of early risk-prediction models for gestational diabetes based on clinical characteristics.

AIMS: Gestational diabetes (GDM) is generally diagnosed late in pregnancy, precluding early preventive interventions. This study aims to validate, in a large Caucasian population of pregnant women, models based on clinical characteristics proposed in the literature to identify, early in pregnancy, those at high risk of developing GDM in order to facilitate follow up and prevention. METHODS: This is a cohort study including 7929 pregnant women recruited prospectively at their first prenatal visit. Clinical information was obtained by a self-administered questionnaire and extraction of data from the medical records. The performance of four proposed clinical risk-prediction models was evaluated for identifying women who developed GDM and those who required insulin therapy. RESULTS: The four models yielded areas under the receiver operating characteristic curve (AUC) between 0.668 and 0.756 for the identification of women who developed GDM, a performance similar to those obtained in the original studies. The best performing model, based on ethnicity, body-mass index, family history of diabetes and past history of GDM, resulted in sensitivity, specificity and AUC of 73% (66-79), 81% (80-82) and 0.824 (0.793-0.855), respectively, for the identification of GDM cases requiring insulin therapy. CONCLUSIONS: External validation of four risk-prediction models based exclusively on clinical characteristics yielded a performance similar to those observed in the original studies. In our cohort, the strategy seems particularly promising for the early prediction of GDM requiring insulin therapy. Addition of recently proposed biochemical markers to such models has the potential to reach a performance justifying clinical utilization.

Absence of association between serum folate and preeclampsia in women exposed to food fortification.

OBJECTIVE: To evaluate serum folate concentration early in pregnancy and any association with hypertensive disorders of pregnancy in a population exposed to folic acid supplementation and food fortification. METHODS: This is a nested case-control study based on a prospective cohort of 7,929 pregnant women recruited in the Quebec City metropolitan area, including 214 participants who developed a hypertensive disorder of pregnancy and 428 normotensive participants in the control group matched for parity, multiple pregnancy, smoking status, gestational, and maternal age at inclusion, and duration of blood sample storage. Serum folate levels were measured at a mean of 14 weeks of gestation. RESULTS: More than 98% of the participants took folic acid or multivitamins before the end of the first trimester. Mean serum folate levels were accordingly high and there were no differences between women who further developed a hypertensive disorder of pregnancy compared with women in the control group (60.1 nmol/L compared with 57.9 nmol/L; P=.51). The proportion of participants with serum folate below the 10th percentile (less than 22.3 nmol/L) of age-matched women in our outpatient population was similar between groups (P=.66) and no participant had levels generally defined as folate deficiency (less than 10 nmol/L). CONCLUSION: In a general cohort of pregnant women benefiting from a national policy of folic acid food fortification combined with a high adherence to folic acid supplementation, serum folate levels are high and do not differ between women who develop a hypertensive disorder of pregnancy and women who remain normotensive. Further supplementation with higher doses is unlikely to be beneficial in such populations. LEVEL OF EVIDENCE: II.

Candidate biochemical markers for screening of pre-eclampsia in early pregnancy.

Pre-eclampsia (PE) and other hypertensive disorders of pregnancy (HDP) are a leading cause of adverse outcomes. Their pathophysiology remains elusive, hampering the development of efficient prevention. The onset of HDP and PE and the severity of their clinical manifestations are heterogeneous. The advent of preventive measures, such as low-dose aspirin that targets high-risk women, emphasizes the need of better prediction. Until recently, only environmental information and maternal risk factors were considered, with equivocal predictive value. No validated screening procedures were available to identify at-risk women despite the emergence of Doppler ultrasonography parameters for the uterine artery (e.g., pulsatility index and bilateral notching) and pathophysiological biochemical markers (e.g., angiogenesis, inflammation, and endothelial dysfunction). Owing to its heterogeneity and lack of specific, sensitive markers among those studied so far (>200), PE is unlikely to be detected early by a single predictive parameter. Systematic reviews have concluded that no single test fulfilling World Health Organization criteria for biomarker selection can diagnose/predict a disease. However, by combining antenatal risk factors, clinical parameters, as well as biophysical and biochemical markers into multivariate algorithms, the risk of PE can be estimated with performance levels that could reach clinical utility. Performance characteristics of selected algorithms will be presented and discussed with respect to transferability to different geographic and healthcare environments.

Early occurrence of metabolic syndrome after hypertension in pregnancy.

OBJECTIVE: The objective of the present study was to evaluate the cardiovascular risk profile and the prevalence of metabolic syndrome among women with a history of pregnancy-induced hypertension (PIH). METHODS: From a cohort of 3,799 nulliparous women prospectively recruited between 1989 and 1997, we performed an observational study on 168 case-control pairs 7.8 years after delivery. Participants were scheduled for a visit with a research nurse to evaluate their cardiovascular risk profile using a questionnaire, anthropometric measurements and blood specimen analysis. RESULTS: One hundred sixty-eight women with prior PIH (105 with gestational hypertension and 63 with preeclampsia) and 168 controls matched for age and year of index delivery were evaluated. The women with PIH (34.6 +/- 4.4 years) were more obese and had higher systolic (115 mm Hg versus 108 mm Hg) and diastolic (75 mm Hg versus 70 mm Hg) blood pressures (P < .001) than the 168 controls (35.1 +/- 4.5 years). They had lower high-density lipoprotein cholesterol level (1.30 mmol/L versus 1.42 mmol/L; P < .001), increased fasting blood glucose concentration (5.2 mmol/L versus 5.0 mmol/L; P = .002), and higher insulin levels (119 versus 91 pmol/L; P < .001). The prevalence of the metabolic syndrome was higher in the PIH group (unadjusted odds ratio = 4.9; 95% confidence interval 2.1-10.9) compared with controls, even after adjustment for confounders (adjusted odds ratio = 3.6; 95% confidence interval 1.4 -9.0). CONCLUSION: In white women in their mid-30s, the prevalence of the metabolic syndrome is 3- to 5-fold increased in those with a history of PIH in their first pregnancy. This emphasizes the importance of long-term follow-up assessment for cardiovascular risk factors in these women.

Trophoblastic remodeling in normal and preeclamptic pregnancies: implication of cytokines.

OBJECTIVE: To summarize the recent knowledge on the implications of placenta and cytokines in normal and preeclamptic pregnancies. DATA SOURCES: A literature search was conducted of applicable articles related to interactions between trophoblast and cytokines in generating preeclampsia. CONCLUSIONS: The initiating event in preeclampsia has been postulated to be the reduced uteroplacental perfusion as a result of abnormal extravillous cytotrophoblast invasion and remodeling of the uterine spiral arteries. Focal ischemia and hypoxia, deportation of hypoxemic trophoblast cells and abnormal expression of various placental biologic molecules, particularly the cytokines, are thought to lead to widespread dysfunction of the maternal vascular endothelium resulting in overproduction of endothelin and thromboxane, enhanced vascular sensitivity to angiotensin II, and reduced secretion of vasodilators such as nitric oxide and prostacyclin. These alterations, in turn, cause hypertension, proteinuria and edema, and pathologies in many organ systems (kidney, lung, liver, brain).

Pathophysiology and maternal biologic markers of preeclampsia.

Preeclampsia-increased blood pressure and proteinuria appearing after the twentieth week of pregnancy--is a major cause of materal and neonatal morbidity, leading to iatrogenic prematurity. Several lines of evidence suggest that the disorder is owing to diminished invasion of spiral arteries by trophoblastic cells, followed by reduced perfusion of the fetoplacental unit and oxidative stress. These alterations, in the presence of maternal predisposition, lead to endothelial dysfunction and occurrence of the clinical syndrome of preeclampsia (multisystemic lesions). Although the pathophysiology of preeclampsia is still unknown, progress has been made during the past 10 yr, and the early identification of at-risk women with the use of biochemical; ultrasonographic; and, more recently, genetic susceptibility markers has been the subject of intense research. In the present review, markers of maternal predisposition, placental implantation, oxidative stress, vasomotor regulation, and endothelial dysfunction are investigated as candidate markers in the early prediction of preeclampsia. Unfortunately, at the present time, no marker has been proven to have a clinically useful predictive performance in the general pregnant population, and, therefore, more research in that area is warranted.