Recent developments in management of older patients with acute myeloid leukemia

Abstract

Patients with acute myeloid leukemia (AML) are currently primarily classified as ‘younger’ or ‘older’. The latter are typically regarded as those who are 60 or above. Inherent in the above approach is the view that age is the principal predictor of outcome in AML. The two principal determinants of outcome are whether the patient incurs treatment-related mortality (TRM) or whether, absent TRM, the patient’s AML is resistant to treatment, with resistance manifested as failure to enter complete remission (CR) or relapse from CR. A problem with this single-minded focus on age less or above 60 is that it is intuitively obvious that a 71-year-old patient, for example, can be fitter than a patient 15 years younger. Indeed data indicate that the effect of age in AML is numerical (continuous) [Thall and Estey, 2001]. Thus, other things being equal, there is more difference in outcome between a patient age 68 and a patient age 61 (both ‘older’) than between the latter patient and a patient age 59 (‘younger’). Even more important the effect of age is modified by the effect of other factors. Walter and colleagues studied TRM and resistance in 1127 patients (median age 57) treated on Southwest Oncology (SWOG) trials from 1986 to 2009 and 1604 patients (median age 61) treated on M.D. Anderson Cancer Center protocols during the period 2000–2008 [Walter et al. 2010b]. They first found that the weekly risk of death declined sharply once 3–5 weeks had elapsed from the start of remission induction therapy. This was true in various age groups and prompted the authors to define TRM as death occurring within the first 30 days of such therapy, with patients having this outcome comprising a qualitatively distinct group. The authors then used the area under the receiver operator characteristic curve (AUC) to quantify the effects of covariates for prediction of TRM and resistance (no TRM but patient does not enter CR or relapses within 1 year of CR), where an AUC of 1 indicates that a covariate is perfect at prediction while an AUC of 0.5 indicates no prediction (i.e. it is no better than flipping a coin). Age alone had an AUC of 0.67 for prediction of TRM while performance status alone had an AUC of 0.72. The inclusion of various additional covariates resulted in an AUC of 0.86. Elimination of age from this model lowered the AUC only to 0.85, suggesting that age is primarily a surrogate for these other covariates. Prediction of resistance was more difficult than prediction of TRM (AUC with our best model, which included de novo or secondary AML, leukemia cell cytogenetics and FLT3 internal tandem duplication [ITD] status, was only 0.70). Here again, however, age alone was inadequate (AUC 0.64), and elimination of age from the best model lowered the AUC only to 0.67. These data suggest that using age as the sole criterion to assign patients to treatment, as is commonly done, is analogous to using hair color (gray or not) to decide a person’s age.