Table 3. Diagnostic accuracy of plasma CTSL, CTSB, or both for the assessment of liver cirrhosis.
| AUC (95%CI) | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Cutoff (ng/ml) | |
|---|---|---|---|---|---|---|
| Control vs. patients with liver cirrhosis (Child–Pugh stage A+B+C) | ||||||
| CTSL | 0.91 (0.85–0.98) | 80.4 | 78.6 | 93.2 | 52.3 | >0.27 |
| CTSB | 0.89 (0.81–0.98) | 76.6 | 80 | 93 | 47.8 | >287.4 |
| CTSL+CTSB | 0.95 (0.91–1.00) | 87.2 | 85.7 | 95.3 | 66.7 | — |
| APRI | 0.94 (0.89–1.00) | 87.1 | 85 | 61.3 | 42.3 | >0.23 |
| FIB-4 | 0.92 (0.85–0.99) | 88.2 | 85.7 | 95.7 | 66.7 | >1.07 |
| Control vs. patients with Child class A liver cirrhosis | ||||||
| CTSL | 0.82 (0.68–0.96) | 80.0 | 71.4 | 80 | 71.4 | >0.15 |
| CTSB | 0.83 (0.69–0.97) | 79.0 | 73.3 | 80.2 | 71.6 | >235.74 |
| CTSL+CTSB | 0.90 (0.80–1.00) | 78.6 | 78.6 | 83 | 73 | — |
| APRI | 0.94 (0.87–1.00) | 85.0 | 92.9 | 94.4 | 81.3 | >0.23 |
| FIB-4 | 0.93 (0.84–1.00) | 85.0 | 85.7 | 89.5 | 80.0 | >1.17 |
| Controls vs. patients with Child class B+C liver cirrhosis | ||||||
| CTSL | 0.97 (0.93–1.00) | 90.3 | 85.7 | 93.1 | 85.7 | >0.28 |
| CTSB | 0.93 (0.84–1.00) | 85.7 | 86.7 | 93 | 85.2 | >317.17 |
| CTSL+CTSB | 0.99 (0.98–1.00) | 92.7 | 92.7 | 96.2 | 86.7 | — |
| APRI | 0.94 (0.88–1.00) | 87.1 | 92.3 | 96.4 | 76.5 | >0.23 |
| FIB-4 | 0.91 (0.83–0.99) | 87.1 | 85.7 | 93.1 | 75.0 | >1.07 |
| Patients with Child class A vs. B+C liver cirrhosis | ||||||
| CTSL | 0.64 (0.48–0.80) | 58.1 | 55.0 | 66.7 | 45.8 | >0.53 |
| CTSB | 0.72 (0.61–0.89) | 73.9 | 73.7 | 82.1 | 65.2 | >430.6 |
| CTSL+CTSB | 0.68 (0.52–0.84) | 60.7 | 63.2 | 70.8 | 47.8 | — |
| APRI | 0.45 (0.28–0.63) | 61.3 | 45.0 | 63.3 | 42.8 | >0.35 |
| FIB-4 | 0.52 (0.35–0.69) | 64.5 | 50.0 | 66.7 | 47.6 | >2.6 |
APRI, aspartate aminotransferase/platelet ratio index; AUC, area under the curve; CTSB, cathepsin B; CTSL, cathepsin L; FIB-4, fibrosis index; NPV, negative predictive value; PPV, negative predictive value.
Estimated threshold for CTSL+CTSB is determined by logistic regression analysis.
AUC were calculated using receiver–operating characteristic curves. Cutoff values were determined by maximizing the sum of sensitivity and specificity.