Performance of a panel of maternal serum markers in predicting preeclampsia at 11-15 weeks' gestation

Objective We evaluated whether a discriminant model of prediction based on quantitative distribution of a panel of biomolecules in maternal serum can discriminate normal pregnancies from those who will develop preeclampsia (PE) prior to onset of clinical symptoms at 11-15 weeks' gestation. Methods Case control study encompassing 56 women destined to develop PE cases matched I : 3 for gestational age with 168 controls. After multiple of median (MoM) conversion of all available markers, comprising total Activin A (t-activin A), P-selectin, and vascular endothelial growth factor receptor (VEGFR) the combined likelihood ratios generated for each marker were used to calculate, for each patient enrolled in the study, the odds of being affected given a positive results (OAPR) of developing PE. For all the analyses performed, the type 11 error was < 20% with a type I error fixed at 5%. Results Data were expressed in MoM of controls. P-selectin was identified as the marker with the best discriminant ability between controls and PE, followed by (t-activin A). No significant differences in VEGFR were observed between cases and controls. By using a 3% prevalence of PE (or, about 1 :33) we found that the median OAPR of developing PE for the 56 cases was 1 :9 or 10% (1 : 1-1 :417). The median OAPR of PE for controls was I : 40 or 2.5% (range, I : 6-1: 4205). Detection rate of the statistical model, with a 5% false-positive rate was 59%. Conclusion This analysis revealed that maternal serum markers assessed at the first and second trimester of pregnancy in asymptomatic patients can improve the early detection of cases at higher risk of developing PE. Copyright (c) 2007 John Wiley & Sons, Ltd.