Predictive Model for Preeclampsia Combining sFlt-1, PlGF, NT-proBNP, and Uric Acid as Biomarkers

Area of Science:

• Obstetrics and Gynecology
• Clinical Chemistry
• Biomedical Data Science

Background:

• Preeclampsia (PE) diagnosis is challenging, with current angiogenic factor markers showing high false-positive rates for short-term prediction.
• Elevated N-terminal pro-brain natriuretic peptide (NT-proBNP) and uric acid levels are observed in pregnancies with PE.

Purpose of the Study:

• To validate a machine-learning model (MLM) for predicting preeclampsia (PE) in patients with clinical suspicion.
• To compare the MLM's predictive performance against the established sFlt-1/PlGF ratio.

Main Methods:

• A multicentric cohort study included 597 women with suspected PE (24-37 weeks gestation).
• The MLM incorporated gestational age, chronic hypertension, sFlt-1, PlGF, NT-proBNP, and uric acid.
• Performance metrics including positive predictive value (PPV), specificity, and area under the curve (AUC) were evaluated.

Main Results:

• The MLM demonstrated a higher PPV (83.1%) for PE within 6 weeks compared to the sFlt-1/PlGF ratio (72.8%).
• The MLM achieved superior specificity (94.9% vs. 91%) and a significantly greater AUC (0.941 vs. 0.901).
• For predicting preterm PE within 1 week, the MLM's AUC (0.954) was significantly higher than the ratio alone (0.914).

Conclusions:

• A machine-learning model combining angiogenic factors, NT-proBNP, and uric acid offers improved prediction of preterm preeclampsia.
• This MLM shows potential for enhancing clinical precision in diagnosing PE.
• The model outperforms the sFlt-1/PlGF ratio alone in predicting PE, particularly for preterm cases.