Table 2.
The diagnostic efficiency of models in differentiating breast cancer and controls
| Training set | Test set | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AUC (95% CI) |
SN (%) | SP (%) | PPV (%) | NPV (%) | +LR | -LR | AUC (95% CI) |
SN (%) | SP (%) | PPV (%) | NPV (%) | +LR | -LR | |
| Breast cancer VS. Controls | ||||||||||||||
| CPT1Aa | 0.892 (0.872–0.920) | 75.4 | 86.1 | 81.8 | 77.1 | 5.42 | 0.29 | 0.904 (0.869–0.939) | 81.4 | 81.7 | 84.2 | 76.0 | 4.45 | 0.23 |
| CA153 | 0.553 (0.506–0.599) | 52.8 | 61.4 | 56.1 | 56.8 | 1.37 | 0.77 | 0.548 (0.475–0.620) | 62.8 | 55.0 | 41.7 | 68.2 | 1.40 | 0.68 |
| CEA | 0.608 (0.562–0.654) | 69.1 | 49.3 | 62.5 | 51.2 | 1.36 | 0.63 | 0.560 (0.488–0.631) | 55.0 | 62.5 | 50.0 | 63.6 | 1.47 | 0.72 |
| CA125 | 0.524 (0.477–0.571) | 65.1 | 42.5 | 32.1 | 73.4 | 1.13 | 0.82 | 0.547 (0.475–0.618) | 39.5 | 73.3 | 45.0 | 56.6 | 1.48 | 0.83 |
| CA153 + CEA + CA125 | 0.605 (0.559–0.651) | 88.7 | 27.9 | 60.4 | 48.2 | 1.23 | 0.41 | 0.581 (0.510–0.652) | 82.9 | 35.0 | 54.2 | 58.9 | 1.28 | 0.49 |
| CPT1A + CA153 + CEA + CA125 | 0.904 (0.881–0.927) | 76.4 | 86.1 | 83.9 | 77.1 | 5.50 | 0.27 | 0.902 (0.865–0.938) | 76.0 | 90.8 | 88.3 | 77.5 | 8.26 | 0.26 |
| Breast cancer VS. Benign breast disease | ||||||||||||||
| CPT1A | 0.807 (0.759–0.855) | 74.6 | 72.9 | 75.7 | 68.3 | 2.75 | 0.35 | 0.781 (0.717–0.846) | 68.7 | 75.9 | 44.8 | 85.5 | 2.85 | 0.41 |
| TNBC VS. Controls | ||||||||||||||
| CPT1A | 0.948 (0.919–0.978) | 87.7 | 88.2 | 98.2 | 70.2 | 7.43 | 0.14 | 0.945 (0.891–0.998) | 94.1 | 81.4 | 99.6 | 52.9 | 5.06 | 0.07 |
Abbreviations: +LR positive likelihood ratio, −LR negative likelihood ratio, PPV positive predictive value, NPV negative predictive value
aThe diagnostic cut-off value was 26.08 ng/mL