Table 3. Prediction models for the pre-set clinical use scenarios.
| Primary Objective: Prediction of increased risk of preterm preeclampsia at PPV > = 0.05 | ||||||
|---|---|---|---|---|---|---|
| Model | Method | AUROC | PPV | Sensitivity | Specificity | |
| PlGF (ref) | N/A | 0.73 (0.61–0.85) | 0.059 (0.033–0.096) | 0.48 (0.26–0.70) | 0.89 (0.86–0.93) | |
| DLG | N/A | 0.70 (0.59–0.82) | 0.054 (0.027–0.091) | 0.39 (0.22–0.61) | 0.90 (0.87–0.93) | |
| PlGF + DLG (ref) | PLS-DA | 0.79 (0.69–0.89) | 0.056 (0.034–0.082) | 0.57* (0.39–0.74) | 0.86 (0.82–0.90) | |
| PlGF + DLG + 1-HGP | PLS-DA | 0.78 (0.69–0.88) | 0.052 (0.035–0.075) | 0.65*° (0.43–0.83) | 0.83 (0.79–0.87) | |
| PlGF || MAP || BMI (ref) | RP | N/A | 0.054 (0.036–0.078) | 0.61*° (0.43–0.78) | 0.85 (0.81–0.89) | |
| PlGF || DLG (ref) | RP | N/A | 0.061 (0.044–0.085) | 0.74*°§ (0.57–0.91) | 0.84 (0.80–0.88) | |
| PlGF || (DLG + 1-HGP) | RP & PLS-DA | N/A | 0.052 (0.039−0.069) | 0.78*°†§ (0.61−0.96) | 0.80 (0.75−0.84) | |
| PlGF || DLG || 1-HGP | RP | N/A | 0.060 (0.043–0.084) | 0.74 *°†§ (0.57–0.91) | 0.84 (0.79–0.87) | |
| Secondary Objective: Prediction of decreased risk of any preeclampsia at NPV> = 0.99 | ||||||
| Model | Method | AUROC | NPV | Specificity | Sensitivity | |
| PlGF (ref) | N/A | 0.60 (0.53–0.67) | 1.00 (1.00–1.00) | 0.02 (0.01–0.03) | 1.00 (1.00–1.00) | |
| DLG | N/A | 0.61 (0.55–0.67) | 1.00 (0.95–1.00) | 0.01 (0.00–1.00) | 1.00 (0.00–1.00) | |
| PlGF + DLG (ref) | PLS-DA | 0.64 (0.57–0.70) | 0.99 (0.97–1.00) | 0.06* (0.04–0.09) | 0.99 (0.97–1.00) | |
| PlGF + DLG + 1-HGP | PLS-DA | 0.66 (0.60–0.72) | 1.00 (1.00–1.00) | 0.03 (0.01–0.04) | 1.00 (1.00–1.00) | |
| PlGF || MAP || BMI (ref) | RP | N/A | 0.99 (0.98–1.00) | 0.32*°† (0.27–0.37) | 0.94 (0.89–0.98) | |
| PlGF || DLG (ref) | RP | N/A | 0.99 (0.98–1.00) | 0.13*° (0.10–0.17) | 0.98 (0.95–1.00) | |
| PlGF || (DLG + 1-HGP) | RP & PLS-DA | N/A | 0.99 (0.98−1.00) | 0.22*°† (0.18−0.26) | 0.96 (0.92−0.99) | |
| PlGF || DLG || 1-HGP | RP | N/A | 0.99 (0.98–1.00) | 0.37*°†§ (0.32–0.42) | 0.93 (0.88–0.98) | |
(ref) reference models for superiority testing
* significantly higher compared to PlGF
° significantly higher compared to PlGF+DLG
† significantly higher compared to PlGF||DLG
§ significantly higher compared to PlGF||MAP||BMI (T test, p<0.05, Bonferroni multiple testing correction). CI Confidence interval; DLG = Dilinoleoyl-glycerol; 1-HGP = 1-heptadecanoyl-2-hydroxy-sn-glycero-3 phosphocholine; RP = recursive partitioning; PLS-DA = partial least squares discriminant analysis. N/A = not applicable.