Table 2.
ROC analyses to discriminate Aβ PET-positive from Aβ PET-negative individuals.
| Cohort | Predictor | AUC (%) | 95% CI | Cut-off | Specificity (%) | Sensitivity (%) | PPV* | NPV** |
|---|---|---|---|---|---|---|---|---|
| A: Discriminatory performance of plasma biomarkers: ROC curves analysis | ||||||||
| Entire cohort (n =123†) | BMs combined via Lasso regression (dropped pTau181 and Aβ40, and pTau181/Aβ42 ratio) | 85.9 | 79–92.8 | 1.5 | 77.4 | 82.9 | 83 | 77 |
| Plasma GFAP | 78.0 | 69.5–86.4 | 90.3 | 50.9 | 94.3 | 72 | 87 | |
| Plasma pTau231 | 68.9 | 59.3–78.5 | 16.1 | 77.4 | 60.0 | 78 | 59 | |
| Plasma pTau181/Aβ42 | 74.3 | 65.2–83.4 | 1.6 | 71.7 | 75.7 | 78 | 69 | |
| Plasma pTau181 | 67.8 | 58.1–77.6 | 8.6 | 47.2 | 85.7 | 68 | 71 | |
| Plasma Aβ42 | 65.1 | 55.2–75.1 | 6.4 | 77.4 | 54.3 | 76 | 56 | |
| Plasma Aβ42/40 | 62.0 | 51.6–72.4 | 0.1 | 47.2 | 78.6 | 66 | 62 | |
| Plasma Aβ40 | 59.6 | 49.1–70 | 117.9 | 58.5 | 70.0 | 69 | 60 | |
| Plasma NFL | 51.3 | 40.6–62.1 | 14.3 | 30.2 | 82.9 | 61 | 57 | |
| B: Discriminatory performance of plasma biomarkers: ROC curves analysis | ||||||||
| Subset of individuals with MCI (n =74††) | BMs combined via Lasso regression (dropped Aβ40) | 93.9 | 87.2–100 | 1.5 | 81.5 | 100.0 | 90 | 100 |
| Plasma GFAP | 84.1 | 75–93.2 | 90.3 | 59.3 | 95.7 | 80 | 89 | |
| Plasma pTau231 | 70.6 | 58.1–83.1 | 15.4 | 81.5 | 57.4 | 84 | 52 | |
| Plasma pTau181/Aβ42 | 63.3 | 49–77.5 | 1.6 | 63.0 | 70.2 | 77 | 55 | |
| Plasma pTau181 | 64.7 | 50.8–78.6 | 10.9 | 74.1 | 59.6 | 80 | 51 | |
| Plasma Aβ42 | 57.1 | 42.6–71.5 | 6.5 | 70.4 | 53.2 | 76 | 46 | |
| Plasma Aβ42/40 | 55.2 | 40.2–70.3 | 0.1 | 40.7 | 80.9 | 70 | 55 | |
| Plasma Aβ40 | 55.2 | 40.6–69.7 | 116.8 | 59.3 | 61.7 | 72 | 47 | |
| Plasma NFL | 58.6 | 43.7–73.4 | 20.3 | 70.4 | 55.3 | 76 | 48 | |
| C: Discriminatory performance of plasma biomarkers: ROC curves analysis | ||||||||
| Subset of individuals with follow-up (n = 18) | Lasso regression (kept only NfL-> see NfL) | – | – | – | – | – | – | – |
| Plasma GFAP | 65.4 | 37.3–93.5 | 142.0 | 77.8 | 66.7 | 75 | 70 | |
| Plasma pTau231 | 59.3 | 31.1–87.4 | 17.0 | 44.4 | 77.8 | 58 | 67 | |
| Plasma pTau181/Aβ42 | 67.9 | 39.3–96.5 | 1.9 | 55.6 | 100.0 | 69 | 100 | |
| Plasma pTau181 | 72.8 | 47.9–97.8 | 12.5 | 44.4 | 100.0 | 64 | 100 | |
| Plasma Aβ42 | 49.4 | 20.1–78.7 | 7.2 | 44.4 | 77.8 | 58 | 67 | |
| Plasma Aβ42/40 | 54.3 | 25.5–83.1 | 0.0 | 22.2 | 100.0 | 56 | 100 | |
| Plasma Aβ40 | 66.7 | 38.8–94.5 | 128.4 | 44.4 | 100.0 | 64 | 100 | |
| Plasma NFL | 76.5 | 53.6–99.5 | 25.1 | 44.4 | 100.0 | 64 | 100 | |
*PPV = positive predictive value. **NPV = negative predictive value. †pTau231 missing in 3 subjects. ††pTau231 missing in 3 subjects.
Sensitivity = probability of detecting true positive (100% if all true positives (TP) are detected and no false negatives (FN) are present).
Specificity = probability of detecting true negative (100% if all true negatives (TN) are present and no false positive (FP) are present).
Positive Predictive Value (PPV) = probability of true positive test (100% if only true positives (TP) are detected and no false positives (FP) are present).
Negative Predictive Value (NPV) = probability of true negative test (100% if only true negatives (TN) are detected and no false negatives (FN) are present).