Preeclampsia Predicament: Study Forecasts Risks from Gestational Day One

Recently published study focused on the challenges of predicting preeclampsia (PE) within the first 16 weeks of gestation due to its complex nature, poorly understood causes, and multiple pathogenic phenotypes. PE is characterized by hypertension after 20 weeks of gestation and related complications affecting maternal and fetal health. Despite advancements in understanding PE, predicting it early remains difficult due to various risk factors and different phenotypes of the condition.

Multi-Omics Approach in Predictive Modeling

The research employed a multi-omics approach, analyzing proteomic and metabolomic data alongside clinical and laboratory information to develop predictive models for EPE and LPE. By utilizing machine learning algorithms like the Boruta algorithm and random forest models, the study identified sets of metabolites and proteins predictive of PE. Notably, certain metabolites such as L-Malic acid and proteins like Superoxide dismutase 3 were found to be associated with PE. Different combinations of clinical factors, omics biomarkers, and laboratory test results were integrated into the predictive models to enhance accuracy. The models proved effective in distinguishing EPE and LPE patients from healthy controls, showcasing superior performance compared to models solely based on clinical factors or single omics data. The addition of laboratory test variables further improved prediction accuracies early in pregnancy.

Identification of Potential Biomarkers for PE

The study highlighted metabolites related to pathways like arginine biosynthesis and proteins involved in immune responses and oxygen transport as potential biomarkers for PE. Integrating clinical, omics, and laboratory data offered novel approaches to predict PE in early pregnancy, providing valuable insights for timely intervention and management. Although the study demonstrated promising results, certain limitations like a small sample size and focusing on a specific cohort were acknowledged. Generalizability to other populations and the interpretation of identified biomarkers require further investigation. Future studies will aim to validate the developed models with larger cohorts and explore the functional implications and molecular mechanisms underlying the identified biomarkers for better understanding and management of PE.

Key Points

- Preeclampsia (PE) is a complex condition characterized by hypertension and related complications affecting maternal and fetal health after 20 weeks of gestation.

- Despite advancements in understanding PE, early prediction remains challenging due to its multifactorial nature, poorly understood causes, and different pathogenic phenotypes.

- A multi-omics approach, combining proteomic and metabolomic data with clinical and laboratory information, was used to develop predictive models for early-onset (EPE) and late-onset (LPE) preeclampsia.

- Machine learning algorithms such as the Boruta algorithm and random forest models identified specific metabolites (e.g., L-Malic acid) and proteins (e.g., Superoxide dismutase 3) associated with PE. - Integrated predictive models that included clinical factors, omics biomarkers, and laboratory test results showed superior performance in distinguishing EPE and LPE patients from healthy controls compared to models based solely on clinical factors or single omics data.

- Metabolites related to pathways like arginine biosynthesis and proteins involved in immune responses and oxygen transport were highlighted as potential biomarkers for PE, offering novel approaches for early prediction and management, though further validation with larger cohorts is needed for generalizability and functional implications.

Reference –

Qiang Zhao et al. (2025). Early Prediction Of Preeclampsia From Clinical, Multi-Omics And Laboratory Data Using Random Forest Model. *BMC Pregnancy And Childbirth*, 25. https://doi.org/10.1186/s12884-025-07582-4.