Table 4.
Sensitivity and specificity of the 6 candidate microRNAs of 104 patients with UICC stage II–III colorectal cancer according to real-time quantitative PCR
| microRNAs | AUCa (95 % CI) | Cut-offb value | Sensitivity (%) | Specificity (%) | Accuracyc (%) |
|---|---|---|---|---|---|
| miR-7 | 0.647 (0.543–0.739) | −0.37 | 53.1 | 80.0 | 66.7 |
| miR-93 | 0.752 (0.658–0.832) | 1.57 | 72.0 | 74.1 | 73.1 |
| miR-141 | 0.617 (0.516–0.711) | 0.59 | 76.9 | 52.6 | 72.8 |
| miR-195 | 0.615 (0.514–0.710) | 0.93 | 96.0 | 26.4 | 56.3 |
| miR-494 | 0.634 (0.534–0.727) | 3.05 | 42.9 | 85.2 | 61.2 |
| let-7b | 0.671 (0.567–0.764) | 2.09 | 55.1 | 76.1 | 61.1 |
| 6-miR panel | 0.834 (0.740–0.905) | 2.00 | 76.6 | 71.4 | 77.4 |
| Combined panel | 0.948 (0.881–0.984) | 4.00 | 89.4 | 88.9 | 89.1 |
aArea under the ROC curve
bThe optimal cut-off value for each microRNA was calculated by analyzing receiver-operating characteristic (ROC) curves, with the relative expression level of miR being normalized to U6b
cAccuracy is defined as the proportion of samples correctly classified into early relapsed (true positives) or nonearly relapsed (true negatives) groups