Fig. 4

A ROC analysis to identify the best predictor of preeclampsia (PE) at 22–24 weeks using RNAs from EVs MT-TA, MT-ND2, MT-CYB, CLU, PF4 and PPBP (35 PE, 35 normal pregnancy (NP)) and MT-TA^PBMC from leukocytes (peripheral blood mononuclear cells, PBMC) (38 PE, 215 NP). B Expression of mitochondrial tRNA MT-TA in leukocytes (PBMC) and plasma between women who subsequently developed PE and those that did not (NP) at weeks 22–24 and 36–38, adjusted for age and BMI. C Scatter plot showing correlation between MT-TA in leukocytes (PBMC) and EVs at weeks 22–24 and 36–38 between women who developed PE and those that did not (NP). r=correlation coefficient, *p<*, ** p<0.01, ***p<0.001. D Logistic regression analysis of predictors of PE at 22–24 weeks. The top part shows MT-TA^PBMC and established risk predictors of PE in univariate analysis. The bottom part shows MT-TA^PBMC when adjusted for each of the established PE risk predictors. MT-TA^PBMC , MAP, Uterine artery PI (artUP) and sFlt-1/PlGF ratio were dichotomized according to Youden’s index: MT-TA^PBMC cut-off: 0.13, likelihood ratio (LHR): 3.2; MAP cut-off: 85.8, LHR: 2,1; artUP cut-off: 0.93, LHR: 1.9; sFlt-1/PlGF ratio cut-off: 9.3, LHR: 1.7