Rethinking Risk Prediction in Preeclampsia: From Biomarkers to Mechanistic Phenotypes and Longitudinal Models

Preeclampsia remains a major cause of maternal and perinatal morbidity and mortality worldwide, yet progress in biomarker discovery and predictive modeling has translated only modestly into clinically meaningful risk stratification. Over the past two decades, numerous biomarkers and predictors reflecting placental-angiogenic dysfunction, maternal cardiovascular maladaptation, and inflammatory-metabolic stress have been proposed, alongside increasingly sophisticated statistical and machine learning approaches. However, many predictive strategies continue to treat preeclampsia as a single disease entity and rely on static thresholds applied at isolated gestational time points. Accumulating biological and clinical evidence instead suggests that preeclampsia represents a heterogeneous syndrome composed of partially overlapping mechanistic phenotypes whose relative contributions vary across pregnancy and across individuals. In this narrative review, we argue that further progress in prediction is likely to depend less on the identification of additional biomarkers and more on how biological heterogeneity and temporal dynamics are integrated into predictive frameworks. We synthesize current evidence supporting multimarker approaches, phenotype-informed frameworks, and longitudinal risk trajectories that conceptualize prediction as a dynamic process rather than a binary classification task. We also examine the complementary roles of classical statistical models and machine learning, emphasizing that calibration, external validation, interpretability, transportability, and clinical usability are essential, alongside discrimination, for successful clinical implementation. Finally, we outline key research priorities for the next generation of predictive studies, including mechanistically grounded phenotyping, dynamic risk updating across gestation, rigorous evaluation across diverse populations, and explicit linkage of risk stratification to preventive interventions and clinical decision-making. Together, these directions support a shift toward an integrative, longitudinal, and clinically anchored approach to preeclampsia prediction.