Table 2.
Post hoc OS analyses: results of multivariate Cox models
| Covariates | HR | 95% CI | P |
|---|---|---|---|
| Cox model 1, adjusted for baseline covariates* | |||
| Treatment (azacitidine vs CCR) | 0.80 | 0.66-0.99 | .0355 |
| Cytogenetic risk status (intermediate vs poor) | 0.42 | 0.34-0.52 | <.0001 |
| ECOG performance status (0-1 vs 2) | 0.62 | 0.48-0.79 | .0001 |
| BM blasts (%)† | 1.01 | 1.01-1.01 | <.0001 |
| Preselected treatment | |||
| BSC vs IC | 1.88 | 1.32-2.69 | .0005 |
| LDAC vs IC | 1.14 | 0.84-1.53 | .3990 |
| Geographic area | |||
| Asia vs Western Europe | 1.19 | 0.86-1.64 | .2915 |
| Eastern Europe vs Western Europe | 1.73 | 1.30-2.31 | .0002 |
| North America/Australia vs Western Europe | 1.16 | 0.88-1.52 | .3036 |
| Age (years)† | 1.02 | 1.00-1.04 | .0219 |
| AML classification (AML with myelodysplasia-related changes vs all other AML types) | 0.83 | 0.67-1.03 | .0973 |
| Cox model 2, adjusted for subsequent treatment | |||
| Treatment (AZA vs CCR) | 0.75 | 0.59-0.94 | .0130 |
| Subsequent therapy (yes vs no) | 1.62 | 1.16-2.25 | .0043 |
| Treatment × subsequent therapy interaction | 2.19 | 1.40-3.42 | .0006 |
| Cox model 3, adjusted for baseline covariates and subsequent treatment | |||
| Treatment (azacitidine vs CCR) | 0.69 | 0.54-0.88 | .0027 |
| Subsequent therapy (yes vs no) | 1.67 | 1.19-2.34 | .0028 |
| Treatment × subsequent therapy interaction | 2.75 | 1.72-4.38 | <.0001 |
| Cytogenetic risk status (intermediate vs poor) | 0.41 | 0.33-0.51 | <.0001 |
| ECOG performance status (0-1 vs 2) | 0.60 | 0.46-0.76 | <.0001 |
| BM blasts (%)† | 1.01 | 1.01-1.02 | <.0001 |
| Preselected treatment | |||
| BSC vs IC | 2.37 | 1.64-3.42 | <.0001 |
| LDAC vs IC | 1.24 | 0.92-1.67 | .1667 |
| Geographic area | |||
| Asia vs Western Europe | 1.15 | 0.84-1.59 | .3804 |
| Eastern Europe vs Western Europe | 1.84 | 1.38-2.45 | <.0001 |
| North America/Australia vs Western Europe | 1.11 | 0.84-1.47 | .4669 |
| Age (years)† | 1.02 | 1.00-1.04 | .0741 |
| AML classification (AML with myelodysplasia-related changes vs all other AML types) | 0.81 | 0.65-1.00 | .0554 |
Covariates included in the model were selected by using a stepwise variable selection procedure in which covariates were selected incrementally into a Cox model with the significance level for inclusion set to P ≤ .25. After each addition of a covariate to the model, the contribution of each covariate present in the model at that step was evaluated; significance level for retention in the model was set to P ≤ .15. Patients with a missing value for any covariate in the final model (n = 10 [2.0%]) were excluded from analysis. Additional covariates that were considered for inclusion but that did not meet the criterion for selection into the model were gender (male, female), WBC (continuous, log-transformed), and prior history of MDS (yes or no).
Continuous variables.