Table 2.
Summary of Resampling Results in Training Dataset Using PCA Classification
| Ct standardized | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| MGMT-LDHA | MGMT | LDHA | |||||||
| Estimate | Training | Test | All | Training | Test | All | Training | Test | All |
| Sensitivity | 99.9 | 91.4 | 97.5 | 85.8 | 85.5 | 85.7 | 42.8 | 42.9 | 42.9 |
| Specificity | 87.3 | 86.0 | 86.9 | 77.7 | 78.0 | 77.8 | 44.4 | 44.5 | 44.4 |
| Accuracy | 91.0 | 87.4 | 89.8 | 80.1 | 79.9 | 80.0 | 44.0 | 44.1 | 44.0 |
| % IDH1 mut GPG | 28.6 | 13.9 | 21.8 | 33.2 | 9.9 | 20.0 | 66.6 | 5.1 | 15.4 |
| % IDH1 mut PPG | 0.0 | 0.02 | 0.006 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| Fold-change PPG/GPG MGMT | 11.0 | 6.2 | 7.8 | 19.3 | 7.2 | 10.6 | 23.3 | 0.6 | 1.6 |
| % GPG<mean MGMT | 85.7 | 78.9 | 82.5 | 100 | 100 | 100 | 100 | 53.8 | 61.5 |
| % PPG<mean MGMT | 22.1 | 20.6 | 21.6 | 7.2 | 5.8 | 6.7 | 0.0 | 46.1 | 25.0 |
| Ct quantiles normalized+standardized | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| MGMT-LDHA | MGMT | LDHA | |||||||
| Estimate | Training | Test | All | Training | Test | All | Training | Test | All |
| Sensitivity | 100.0 | 99.9 | 99.9 | 100.0 | 100.0 | 100.0 | 39.7 | 40.4 | 40.0 |
| Specificity | 85.2 | 81.1 | 83.8 | 80.0 | 80.1 | 80.0 | 50.0 | 50.1 | 50.0 |
| Accuracy | 88.0 | 85.3 | 87.0 | 83.7 | 84.5 | 84.0 | 48.1 | 48.0 | 48.0 |
| % IDH1 mut GPG | 40.0 | 15.2 | 24.4 | 40.1 | 13.3 | 22.2 | 90.0 | 5.5 | 16.7 |
| % IDH1 mut PPG | 0.0 | 0.01 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| Fold-change PPG/GPG MGMT | 16.4 | 7.8 | 9.6 | 18.8 | 9.4 | 11.2 | 24.5 | 0.6 | 1.7 |
| % GPG<mean MGMT | 100.0 | 85.0 | 90.2 | 100.0 | 100.0 | 100.0 | 90.0 | 56.2 | 66.7 |
| % PPG<mean MGMT | 21.8 | 21.3 | 21.6 | 12.5 | 12.6 | 12.5 | 0.0 | 47.6 | 23.1 |
This table depicts the average classification and molecular features across 10,000 iterations. Only those samples having survival data were considered (n=25). Two thirds of cases were used as training set (n=16), and the remaining ones were used as a test set (n=9). The information in this table is equivalent to that provided in Table 1, but in this case the LDA equations were fitted only using the standardized RT-PCR expression values from MGMT and LDHA genes. On the top panel, values were only standardized, while on the bottom one, the Ct values were first normalized by the quantiles method and the same standardization as the one described above was undertaken afterwards.