TABLE 4.
Prediction of Transient AKI by Biomarker Composites
| Marker | Sensitivity | Specificity | PPV | NPV | + LR | –LR |
|---|---|---|---|---|---|---|
| ΔSCr 0%-24% | 42 (25-61) | 59 (39-77) | 54 (33-74) | 47 (30-65) | 1.0 (0.6-1.9) | 1.0 (0.6-1.5) |
| ΔSCr 25%-49% | 24 (11-42) | 83 (64-94) | 62 (32-86) | 49 (34-64) | 1.4 (0.5-3.8) | 0.9 (0.7-1.2) |
| ΔSCr ≥50% | 15 (5-32) | 69 (49-85) | 36 (13-65) | 42 (28-57) | 0.5 (0.2-1.3) | 1.2 (0.9-1.6) |
| uNGAL– pCysC– |
3(0-16) | 93 (77-99) | 33 (5-88) | 46 (33-59) | 0.4 (0-4.6) | 1.0 (0.9-1.2) |
| uNGAL– pCysC+ |
42 (25-61) | 100 (88-100) | 100 (77-100) | 60 (45-74) | NC | 0.6 (0.4-0.8) |
| UNGAL+ pCysC– |
24 (11-42) | 76 (56-90) | 53 (27-79) | 47 (32-62) | 1.0 (0.4-2.4) | 1.0 (0.8-1.3) |
| UNGAL+ pCysC+ |
30 (16-49) | 31 (15-51) | 33 (17-53) | 28 (14-47) | 0.4 (0.3-0.8) | 2.3 (1.3-4.0) |
Sensitivity, specificity, PPV, and NPV are % (95% CI). For patients who developed AKI (62 of 345), biomarker composites of uNGAL and pCysC in relation to their respective cutoff values for positivity ≥200 ng/mg and 0.8 mg/l (+) demonstrate the ability to use composite biomarker panels to identify transient injury. Prediction was significantly superior to that afforded by any ΔSCr from pre-operative to first post-operative value. uNGAL is normalized to urinary creatinine.
Abbreviations as in Table 2.