Table 2. Performance of BRCA carrier prediction models at different thresholds.
| Threshold (%) |
Model |
BRCA vs Non-BRCA (n=2,426) |
BRCA1 vs Non-BRCA1 (n=2,426) |
BRCA2 vs Non-BRCA2 (n=2,426) |
|||
|---|---|---|---|---|---|---|---|
| Sensitivity (%) |
Specificity (%) |
Sensitivity (%) |
Specificity (%) |
Sensitivity (%) |
Specificity (%) |
||
| 4.0 * | ARiCa | 71 | 71 | 83 | 70 | 66 | 70 |
| 8.0 * | PENNII | 63 | 70 | 83 | 70 | 51 | 69 |
| 2.2 * | BOADICEA | 66 | 67 | 83 | 67 | 56 | 67 |
| 4.0 * | KOHCal | 66 | 68 | 83 | 67 | 56 | 67 |
| 10.0 ** | ARiCa | 34 | 94 | 58 | 93 | 19 | 93 |
| PENNII | 60 | 74 | 78 | 73 | 49 | 73 | |
| BOADICEA | 22 | 95 | 44 | 95 | 8 | 95 | |
| KOHCal | 44 | 88 | 64 | 87 | 32 | 87 | |
| 20.0 ** | ARiCa | 15 | 99 | 25 | 98 | 8 | 98 |
| PENNII | 15 | 99 | 25 | 98 | 8 | 98 | |
| BOADICEA | 7 | 99 | 14 | 99 | 3 | 99 | |
| KOHCal | 22 | 97 | 42 | 97 | 10 | 96 | |
Sample: 2426 MyBrCa and SGBCC breast cancer patients with pedigree available in single imputed validation set.
*
Optimal threshold.
**
Conventional threshold.