Table 4.
Diagnostic accuracy measurements in all HCCs and early HCCs, respectively, according to ROC curve analysis of the trifucosylation degree of AGP, AFP, INR, and marker panels.
| All HCC vs. Cirrhosis | |||||
|---|---|---|---|---|---|
| Model | AUC | 95% CI | Cutoff* | Sens* | Spec* |
| 1 * AFP | 0.820 | (0.747, 0.884) | 2.722 | 0.72 | 0.79 |
| 1 * INR | 0.612 | (0.523, 0.708) | 1.138 | 0.39 | 0.47 |
| 1 * triFc | 0.750 | (0.659, 0.828) | 0.013 | 0.64 | 0.76 |
| 0.7296 * AFP +12.6855 * triFc | 0.832 | (0.764, 0.897) | 2.121 | 0.76 | 0.81 |
| 0.9622 * AFP −5.967 * INR | 0.837 | (0.764, 0.901) | −4.147 | 0.86 | 0.71 |
| −5.7676 * INR +39.8998 * triFc | 0.780 | (0.695, 0.859) | −5.653 | 0.77 | 0.77 |
| 0.817 * AFP −6.8425 * INR +17.8902 * triFc | 0.853 | (0.781, 0.913) | −4.960 | 0.80 | 0.79 |
| Early HCC vs. Cirrhosis | |||||
| 1 * AFP | 0.777 | (0.680, 0.865) | 2.433 | 0.77 | 0.67 |
| 1 * INR | 0.598 | (0.496, 0.709) | 1.138 | 0.40 | 0.47 |
| 1 * triFc | 0.695 | (0.590, 0.786) | 0.013 | 0.53 | 0.76 |
| 0.6951 * AFP +0.3118 * triFc | 0.780 | (0.688, 0.870) | 1.696 | 0.77 | 0.67 |
| 0.8585 * AFP −5.9504 * INR | 0.793 | (0.698, 0.876) | −4.297 | 0.80 | 0.71 |
| −3.9602 * INR +26.5163 * triFc | 0.699 | (0.591, 0.807) | −3.846 | 0.65 | 0.79 |
| 0.8387 * AFP −6.0249 * INR +2.8025 * triFc | 0.795 | (0.691, 0.887) | −4.297 | 0.74 | 0.77 |
*The cutoff and its corresponding sensitivity (sens) and specificity (spec) are determined by the point with shortest distance to the point when sensitivity = 1 and specificity = 1. +Marker values were taken log2 transformation when fitting the models.