Cox analysis: effect of inclusion of imatinib and use of alloHSCT on outcome
| . | . | Preimatinib vs any imatinib . | Late vs early imatinib . | Pre- vs late vs early imatinib (trend) . |
|---|---|---|---|---|
| . | . | . | HR (95% CI), P value . | . |
| OS | Univariate Cox model | 0.66 (0.53-0.83), P = .0004 | 0.77 (0.53-1.12), P = .2 | 0.76 (0.66-0.88), P = .0002 |
| Multivariate Cox model* | 0.57 (0.45-0.73), P < .0001 | 0.72 (0.49-1.05), P = .09 | 0.69 (0.59-0.80), P < .0001 | |
| Multivariate Cox allowing for alloHSCT | 0.69 (0.47-1.01), P = .06 | 0.78 (0.41-1.48), P = .4 | 0.77 (0.60-1.00), P = .05 | |
| EFS | Univariate Cox model | 0.64 (0.51-0.80), P < .0001 | 0.76 (0.53-1.08), P = .1 | 0.74 (0.65-0.86), P < .0001 |
| Multivariate Cox model* | 0.58 (0.46-0.73), P < .0001 | 0.67 (0.46-0.98), P = .04 | 0.69 (0.60-0.80), P < .0001 | |
| Multivariate Cox allowing for alloHSCT | 0.64 (0.44-0.93), P = .02 | 0.62 (0.33-1.18), P = .1 | 0.73 (0.57-0.93), P = .01 | |
| RFS | Univariate Cox model | 0.58 (0.43-0.78), P = .003 | 0.68 (0.42-1.09), P = .1 | 0.69 (0.57-0.84), P = .0001 |
| Multivariate Cox model* | 0.50 (0.37-0.68), P < .0001 | 0.57 (0.35-0.95), P = .03 | 0.63 (0.52-0.76), P < .0001 | |
| Multivariate Cox allowing for alloHSCT | 0.68 (0.45-1.02), P = .06 | 0.45 (0.22-0.94), P = .03 | 0.74 (0.56-0.96), P = .03 | |
| Survival free from death in remission | Univariate Cox model | 0.83 (0.55-1.27), P = .4 | 0.85 (0.45-1.61), P = .6 | 0.87 (0.67-1.14), P = .3 |
| Multivariate Cox model* | 0.76 (0.49-1.17), P = .2 | 0.76 (0.40-1.46), P = .4 | 0.82 (0.63-1.07), P = .1 | |
| Multivariate Cox allowing for alloHSCT | 0.68 (0.24-1.97), P = .5 | 2.04 (0.36-11.5), P = .4 | 0.89 (0.47-1.68), P = .7 |
| . | . | Preimatinib vs any imatinib . | Late vs early imatinib . | Pre- vs late vs early imatinib (trend) . |
|---|---|---|---|---|
| . | . | . | HR (95% CI), P value . | . |
| OS | Univariate Cox model | 0.66 (0.53-0.83), P = .0004 | 0.77 (0.53-1.12), P = .2 | 0.76 (0.66-0.88), P = .0002 |
| Multivariate Cox model* | 0.57 (0.45-0.73), P < .0001 | 0.72 (0.49-1.05), P = .09 | 0.69 (0.59-0.80), P < .0001 | |
| Multivariate Cox allowing for alloHSCT | 0.69 (0.47-1.01), P = .06 | 0.78 (0.41-1.48), P = .4 | 0.77 (0.60-1.00), P = .05 | |
| EFS | Univariate Cox model | 0.64 (0.51-0.80), P < .0001 | 0.76 (0.53-1.08), P = .1 | 0.74 (0.65-0.86), P < .0001 |
| Multivariate Cox model* | 0.58 (0.46-0.73), P < .0001 | 0.67 (0.46-0.98), P = .04 | 0.69 (0.60-0.80), P < .0001 | |
| Multivariate Cox allowing for alloHSCT | 0.64 (0.44-0.93), P = .02 | 0.62 (0.33-1.18), P = .1 | 0.73 (0.57-0.93), P = .01 | |
| RFS | Univariate Cox model | 0.58 (0.43-0.78), P = .003 | 0.68 (0.42-1.09), P = .1 | 0.69 (0.57-0.84), P = .0001 |
| Multivariate Cox model* | 0.50 (0.37-0.68), P < .0001 | 0.57 (0.35-0.95), P = .03 | 0.63 (0.52-0.76), P < .0001 | |
| Multivariate Cox allowing for alloHSCT | 0.68 (0.45-1.02), P = .06 | 0.45 (0.22-0.94), P = .03 | 0.74 (0.56-0.96), P = .03 | |
| Survival free from death in remission | Univariate Cox model | 0.83 (0.55-1.27), P = .4 | 0.85 (0.45-1.61), P = .6 | 0.87 (0.67-1.14), P = .3 |
| Multivariate Cox model* | 0.76 (0.49-1.17), P = .2 | 0.76 (0.40-1.46), P = .4 | 0.82 (0.63-1.07), P = .1 | |
| Multivariate Cox allowing for alloHSCT | 0.68 (0.24-1.97), P = .5 | 2.04 (0.36-11.5), P = .4 | 0.89 (0.47-1.68), P = .7 |
Adjusting for gender, age at diagnosis, initial WBC, CNS disease at presentation, and an interaction between WBC and CNS disease at presentation and in “allowing for alloHSCT,” treatment (MUD/sibling alloHSCT vs chemotherapy). Preimatinib vs late imatinib vs early imatinib (treated as an ordinal variable). HR, hazard ratio. HR <1 indicates better outcome with increasing values of the variable (ie, outcome improves with cohort: any imatinib better than preimatinib; early better than late imatinib; early imatinib better than late imatinib better than preimatinib cohort).