Initial and late prognostic factors to predict survival in adult acute lymphoblastic leukaemia

Factors able to predict overall survival in adult patients with acute lymphoblastic leukaemia were assessed according to the period since initiation of the treatment using a Cox proportional hazards model. This period covers successively an initial period during the induction treatment and a consolidation period during the postinduction treatment. From 1994 to 2002, 922 patients with acute lymphoblastic leukaemia (excluding French-American-British L3 subtype) were enrolled in a multicentre protocol and followed, with a mean follow up of 58 months. A multivariate time-segmented analysis was performed on 658 patients. Analyses of the initial (before 100 d) and the late phases were realised after stratification on the type of induction treatment and on the different treatment strategies respectively. Age was the sole factor that influenced survival during the initial phase (hazard ratio 1.48 per 10-yr increase; P < 0.01). Factors that predicted survival during the late phase were age (hazard ratio 1.12, P = 0.02), white blood cells count (hazard ratio 1.01 per 10(10) cells/L increase; P < 0.05), lactic dehydrogenase level (hazard ratio 1.001 for 10 IU/L increase; P < 0.01) and t(9;22) karyotype or miscellaneous others vs. normal karyotype (hazard ratios 1.40; P < 0.01 and 1.06; P = 0.04 respectively). This analysis suggests that predictive factors may be split into tolerance factors and haematological factors. Determination of such factors is crucial to adapt postremission therapeutic strategies in acute lymphoblastic leukaemia.