Prediction of pre-eclampsia and its subtypes in high-risk cohort: hyperglycosylated human chorionic gonadotropin in multivariate models

Abstract

BackgroundThe proportion of hyperglycosylated human chorionic gonadotropin (hCG-h) to total human chorionic gonadotropin (%hCG-h) during the first trimester is a promising biomarker for prediction of early-onset pre-eclampsia. We wanted to evaluate the performance of clinical risk factors, mean arterial pressure (MAP), %hCG-h, hCGβ, pregnancy-associated plasma protein A (PAPP-A), placental growth factor (PlGF) and mean pulsatility index of the uterine artery (Uta-PI) in the first trimester in predicting pre-eclampsia (PE) and its subtypes early-onset, late-onset, severe and non-severe PE in a high-risk cohort.MethodsWe studied a subcohort of 257 high-risk women in the prospectively collected Prediction and Prevention of Pre-eclampsia and Intrauterine Growth Restriction (PREDO) cohort. Multivariate logistic regression was used to construct the prediction models. The first model included background variables and MAP. Additionally, biomarkers were included in the second model and mean Uta-PI was included in the third model. All variables that improved the model fit were included at each step. The area under the curve (AUC) was determined for all models.ResultsWe found that lower levels of serum PlGF concentration were associated with early-onset PE, whereas lower %hCG-h was associated with the late-onset PE. Serum PlGF was lower and hCGβ higher in severe PE, while %hCG-h and serum PAPP-A were lower in non-severe PE. By using multivariate regression analyses the best prediction for all PE was achieved with the third model: AUC was 0.66, and sensitivity 36% at 90% specificity. Third model also gave the highest prediction accuracy for late-onset, severe and non-severe PE: AUC 0.66 with 32% sensitivity, AUC 0.65, 24% sensitivity and AUC 0.60, 22% sensitivity at 90% specificity, respectively. The best prediction for early-onset PE was achieved using the second model: AUC 0.68 and 20% sensitivity at 90% specificity.ConclusionsAlthough the multivariate models did not meet the requirements to be clinically useful screening tools, our results indicate that the biomarker profile in women with risk factors for PE is different according to the subtype of PE. The heterogeneous nature of PE results in difficulty to find new, clinically useful biomarkers for prediction of PE in early pregnancy in high-risk cohorts.Trial registrationInternational Standard Randomised Controlled Trial number ISRCTN14030412, Date of registration 6/09/2007, retrospectively registered.Electronic supplementary materialThe online version of this article (10.1186/s12884-018-1908-9) contains supplementary material, which is available to authorized users.

Comparison of Multivariable Logistic Regression and Other Machine Learning Algorithms for Prognostic Prediction Studies in Pregnancy Care: Systematic Review and Meta-Analysis

“…As applicability was already included in the eligibility assessment before the qualitative analysis, eligible studies were not found to use data sets from either primary care or hospital settings, such as from a house-to-house survey or a screening program. Most used data sets were from hospital settings, whereas only a few of those were from primary care settings in the LR (6/77, 8%) [65,69,73,77,78,87], non-LR (6/50, 12%) [119,122,132,135,148,153], or both algorithms (1/15, 7%) [162]. A detailed description of this is also given in Multimedia Appendix 1.…”

“…The characteristics of study populations showed that pregnant women and fetuses or newborns were the populations of most studies developed using LR and non-LR models, respectively. Among pregnant women, the LR algorithm was mostly applied to develop predictions for outcome categories of obstetric labor (13/77, 17%) [36,46,47,54,57,62,64,70,83,86,91,97,103], pregnancy-induced hypertension (12/77, 16%) [30,31,43,48,55,65,66,68,76,81,93,105], and gestational diabetes (7/77, 9%) [33,45,49,84,94,100,104]. Among fetus or newborn populations, non-LR algorithms were mostly applied to develop predictions for outcome categories of premature birth (12/50, 24%) [111,112,115,116,118,119,121,122,125,130,141,143] and fetal distress (9/50, 18%) [113,…”

“…The same algorithm was applied to a prediction model from a non-LR low ROB study in pre-eclampsia [147]. For random effects modeling, this model also significantly outperformed those from 4 LR studies (1.2, 95% CI 0.72-1.67) [31,48,65,76].…”

“…Meanwhile, the LR counterparts used maternal demographics or lifestyle [31,65,76], medical histories [31,65,76], obstetric histories [31,65,76], family histories [31,76], clinical or obstetric examinations [31,65], ultrasound measures [65], routine laboratory tests [76], and biomarkers [48,65]. These predictors were obtained before pregnancy [31], at 11 to 13 weeks of gestation [65], and at <20 weeks of gestation [48]. LR models were developed to predict pre-eclampsia of any severity and timing [31,48,65,76].…”

Placenta-Specific Genes, Their Regulation During Villous Trophoblast Differentiation and Dysregulation in Preterm Preeclampsia

“…Preeclampsia, a life-threatening obstetrical syndrome related to placental dysfunction [32,[57][58][59][60][61][62][63][64][65], is associated with dysregulated expression or function of many PPE proteins (e.g., CGB, LEP), transcription factors (e.g., GCM1, ESRRG), and signaling and kinase pathways (e.g., ERK1/2, p38), especially when it develops preterm [19,57,[66][67][68][69][70][71][72][73][74][75]. Based on the observations that the volume [76] and cellular architecture [67][68][69][77][78][79] of villous trophoblasts are severely impacted in preterm preeclampsia, it has been proposed that either the entire cytotrophoblast differentiation program [80], or particularly syncytialization [81][82][83][84][85][86][87], are affected in preterm preeclampsia, although neither have been proven at the systems level yet.…”