Table 2.
Performance characteristics of tissue-specific microsatellite panels for MSI diagnosis in validation set.
| Cancer Type (TCGA Code) | Smallest panel achieving sensitivity and specificity of ≥95% |
Smallest panel achieving AUROC of 0.9 |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Number of markers | Minimum unstable markers for MSI-H classification (%) | Area under the curve | Sensitivity | Specificity | Number of markers | Minimum unstable markers for MSI-H classification (%) | Area under the curve | Sensitivity | Specificity | |
| COAD | 1 | 50.0% | 0.98 | 100.0% | 95.8% | 1 | 50.0% | 0.98 | 100.0% | 95.8% |
| ESCA | 1 | 50.0% | 1.00 | 100.0% | 100.0% | 1 | 50.0% | 1.00 | 100.0% | 100.0% |
| GBM | 1 | 50.0% | 1.00 | 100.0% | 100.0% | 1 | 50.0% | 1.00 | 100.0% | 100.0% |
| LUAD | 1 | 50.0% | 0.99 | 100.0% | 97.1% | 1 | 50.0% | 0.99 | 100.0% | 97.1% |
| LUSC | 1 | 50.0% | 0.99 | 100.0% | 97.1% | 1 | 50.0% | 0.99 | 100.0% | 97.1% |
| READ | 2 | 50.0% | 1.00 | 100.0% | 100.0% | 2 | 50.0% | 1.00 | 100.0% | 100.0% |
| STAD | 2 | 25.0% | 1.00 | 100.0% | 100.0% | 1 | 50.0% | 0.92 | 84.2% | 100.0% |
| LGG | 5 | 12.5% | 1.00 | 100.0% | 100.0% | 5 | 12.5% | 1.00 | 100.0% | 100.0% |
| PRAD | 5 | 16.7% | 1.00 | 100.0% | 100.0% | 5 | 16.7% | 1.00 | 100.0% | 100.0% |
| CESC | 7 | 45.8% | 1.00 | 100.0% | 100.0% | 7 | 45.8% | 1.00 | 100.0% | 100.0% |
| DLBC | 7 | 25.0% | 1.00 | 100.0% | 100.0% | 7 | 25.0% | 1.00 | 100.0% | 100.0% |
| UCEC | 20 | 12.9% | 0.98 | 97.0% | 97.9% | 2 | 25.0% | 0.95 | 90.3% | 97.2% |
| KIRC | 65 | 7.5% | 0.99 | 100.0% | 98.5% | 55 | 4.6% | 0.93 | 100.0% | 86.6% |
| BRCA | 800 | 3.6% | 1.00 | 100.0% | 99.5% | 50 | 4.3% | 0.92 | 100.0% | 81.1% |
| OVa | 2000 | 2.0% | 0.63 | 100.0% | 47.8% | 2000 | 2.0% | 0.63 | 100.0% | 47.8% |
a
Did not achieve cutoff for test-characteristic threshold with the maximum number of loci tested.