microRNA and mRNA class predictors for subsets of AML with (cyto)genetic abnormalities identified using prediction analysis of microarrays (PAM)
| (Cyto)genetic subset . | Training set, N = 143 . | Validation set, N = 72 . | Class predictor No. microRNAs . | ||||
|---|---|---|---|---|---|---|---|
| No. of cases . | True prediction . | False positive . | No. of cases . | True prediction . | False positive . | ||
| microRNA class predictor | |||||||
| t(15;17) | 5 | 5 | 3 | 1 | 1 | 4 | 7 |
| t(8;21) | 8 | 7 | 8 | 4 | 3 | 3 | 10 |
| Inv(16) | 6 | 5 | 10 | 9 | 4 | 2 | 72 |
| NPM1 | 47 | 46 | 21 | 21 | 20 | 6 | 10 |
| CEBPA | 17 | 2 | 3 | 5 | 4 | 1 | 28 |
| FLT3-ITD | 46 | 33 | 22 | 16 | 14 | 5 | 48 |
| . | No. of cases . | True prediction . | False positive . | No. of cases . | True prediction . | False positive . | No. probe sets . |
| mRNA class predictor | |||||||
| t(15;17) | 5 | 4 | 0 | 1 | 1 | 1 | 8 |
| t(8;21) | 8 | 8 | 0 | 4 | 4 | 0 | 2 |
| Inv(16) | 6 | 6 | 0 | 9 | 9 | 1 | 2 |
| NPM1 | 47 | 23 | 11 | 21 | 20 | 2 | 7 |
| CEBPA | 17 | 13 | 4 | 5 | 3 | 2 | 20 |
| FLT3-ITD | 46 | 31 | 7 | 16 | 13 | 2 | 36 |
| (Cyto)genetic subset . | Training set, N = 143 . | Validation set, N = 72 . | Class predictor No. microRNAs . | ||||
|---|---|---|---|---|---|---|---|
| No. of cases . | True prediction . | False positive . | No. of cases . | True prediction . | False positive . | ||
| microRNA class predictor | |||||||
| t(15;17) | 5 | 5 | 3 | 1 | 1 | 4 | 7 |
| t(8;21) | 8 | 7 | 8 | 4 | 3 | 3 | 10 |
| Inv(16) | 6 | 5 | 10 | 9 | 4 | 2 | 72 |
| NPM1 | 47 | 46 | 21 | 21 | 20 | 6 | 10 |
| CEBPA | 17 | 2 | 3 | 5 | 4 | 1 | 28 |
| FLT3-ITD | 46 | 33 | 22 | 16 | 14 | 5 | 48 |
| . | No. of cases . | True prediction . | False positive . | No. of cases . | True prediction . | False positive . | No. probe sets . |
| mRNA class predictor | |||||||
| t(15;17) | 5 | 4 | 0 | 1 | 1 | 1 | 8 |
| t(8;21) | 8 | 8 | 0 | 4 | 4 | 0 | 2 |
| Inv(16) | 6 | 6 | 0 | 9 | 9 | 1 | 2 |
| NPM1 | 47 | 23 | 11 | 21 | 20 | 2 | 7 |
| CEBPA | 17 | 13 | 4 | 5 | 3 | 2 | 20 |
| FLT3-ITD | 46 | 31 | 7 | 16 | 13 | 2 | 36 |
PAM was performed to define microRNA and mRNA class predictors containing a minimal number of microRNAs and probe sets (mRNAs), respectively, that could predict whether a case belonged to a particular (cyto)genetic subset (first column). The cohort was randomly segregated into a training set and a validation set. The number of cases in the training set (column 2) and the validation set (column 5) are given. The 10-fold method of cross-validation, applied on the training set was used to compute truly predicted cases and errors. The cases that were truly predicted (column 3) or were false positive (column 4) in the training set are indicated. The minimal number of microRNAs or probe sets (mRNAs) that were identified in the training set was subsequently tested on the validation set. The number of cases correctly predicted (ie, true prediction) in the validation set (column 6) or were false positively predicted (column 7) are shown. The characteristics of the microRNAs and probe sets (mRNAs) are presented in Table S3.