Clinical biomarker for predicting preeclampsia in women with abnormal lipid profile: Statistical pattern classification approach

Abstract

Maternal dyslipidemia in preeclampsia is well established. Serum lipid levels as potential predictors of preeclampsia are yet to be investigated. Discriminant analysis and k-means clustering were used to predict preeclampsia (PE) based on the contribution of lipid parameters. Serum total cholesterol (TC), high density lipoprotein (HDL-C), low density lipoprotein (LDL-C) and triglycerides (TG) were measured in venous blood samples of women with PE (Group A; n=62) and normotensive pregnant women (Group B; n=54). Very low density lipoprotein (VLDL) was calculated as 1/5 of TG. Discriminant analysis was used to identify the clinical markers amongst these parameters. k-means clustering was used to validate the parameters identified. TC, LDL-C, TG and VLDL levels were significantly higher and HDL-C significantly lower in Group A when compared with Group B. Amongst these, TG, VLDL and TC emerged as the ideal set of clinical markers in discriminating Group A and Group B with an overall classification accuracy of 87.9%, 87.9% and 86.1%, respectively. The clusters centers indicating mean values of TG, TC and VLDL were significantly higher in Group A as compared to Group B. Discriminant analysis was used to identify the most useful set of clinical markers amongst all the lipid parameters. Serum TG, VLDL and TC levels predicted PE with maximum accuracy, which was further verified by k-means clustering.