Table 3.

ROC analysis of the sFlt-1/PlGF and peptide biomarker candidates in preeclampsia prediction.

	sFlt-1/PlGF	ITIH4-2026.9	FGA-1033.4	ITIH4-2051.1	Combineda
Discovery cohort (n = 115)
AUC (95% CI)	0.637 (0.525–0.750)	0.946 (0.905–0.988)	0.939 (0.898–0.980)	0.820 (0.744–0.896)	0.985 (0.965–1.000)
Sensitivity, %	79.3	90.0	100.0	93.3	96.7
Specificity, %	43.5	92.9	74.1	70.6	94.1
Positive predictive value, %	32.4	81.8	57.7	52.8	85.3
Negative predictive value, %	86.0	96.3	100.0	96.8	98.8
p-value of AUCs comparison	<0.001	0.036b	0.026c	<0.001d	–
Validation cohort (n = 80)
AUC (95% CI)	0.733 (0.617–0.850)	0.866 (0.771–0.961)	0.896 (0.795–0.997)	0.734 (0.598–0.871)	0.923 (0.845–1.000)
Sensitivity, %	95.0	75.0	80.0	50.0	95.0
Specificity, %	48.3	91.7	95.0	90.0	88.3
Positive predictive value, %	38.0	75.0	84.2	62.5	73.1
Negative predictive value, %	96.7	91.7	93.4	84.4	98.1
p-value of AUCs comparison	0.004	0.040b	0.185c	0.002d	–

^aThe three-peptide combined panel in ROC analysis;

^bAUCs comparison between ITIH4-2026.9 and three-peptide panel;

^cAUCs comparison between FGA-1033.4 and three-peptide panel;

^dAUCs comparison between ITIH4-2051.1 and three-peptide panel. ROC, receiver operating characteristic; AUC, area under curve; CI, confidence interval.