Declining Rates of Treatment-Related Mortality in Patients with Newly Diagnosed Acute Myeloid Leukemia (AML) Given “Intensive” Induction Regimens: A Report From the Southwest Oncology Group (SWOG) and MD Anderson Cancer Center (MDA)

Recent emphasis has been placed on administration of induction regimens less intense than standard 3+7 for patients with newly diagnosed AML. A primary goal is to reduce the presumed average treatment related mortality (TRM) rate of 10% occurring within the first 28 days after start of 3+7 or higher intensity therapies; TRM rates have been > 30% in patients who are older and/or have poor performance status (PS]. (Walter et al. JCO 2012). This practice presupposes that TRM rates with higher intensity induction regimens are static, a notion seemingly difficult to reconcile with advances in supportive care (e.g. newer anti-aspergillosis drugs) that have sharply reduced rates of non-relapse mortality after allogeneic hematopoietic cell transplant (Gooley et al. NEJM 2010).

We thus addressed rates of TRM from 1991–2009 in 1,409 patients given 3+7 induction regimens on SWOG protocols (cytarabine dose 100 mg/m2 daily × 7) and 1,933 patients given induction regimens containing higher cytarabine doses (at least 1.0 g/m2 daily × 4–5 days) at MDA, variably combined with idarubicin, fludarabine or other agents. Multivariate analyses were used to account for confounding factors.

TRM rates declined both in SWOG and at MDA.

However this reduction must account for the declining ages of patient given 3+7 or more intense induction (p<0.001 in both SWOG and at MDA) and their improved PS (p<0.001 SWOG and MDA);the considerably younger nature of SWOG patients during 2006–2009 reflects the switch to less intense induction regimens for many older patients; such regimens were not included in this analysis. Additionally other covariates associated with TRM (more blood blasts, lower platelets, secondary AML) by Walter et al. (JCO 2012)were unevenly distributed in the various time periods(for example no secondary AML in SWOG 2006–2009). Multivariate logistic regression was thus performed to account for the effect of age, PS, and these other covariates in the reduction in TRM. After such accounting, odds ratios (ORs) for TRM at MDA were (relative to 1995) 0.89, 0.7, and 0.36 for 1996–2000,2001–2005, and 2006–2009 respectively with the null hypothesis of no change over time rejected at p = 0.006. For SWOG, not including secondary AML as a covariate ORs were (relative to 1991– 1995) 0.75,0.78, and 0.42 for 1996–2000,2001–2005,and 2006–2009 respectively; again the hypothesis of no change with time was rejected (p = 0.037). There were no interactions between reduced TRM and age, WBC or performance status suggesting the reduction in TRM was a general phenomenon.

There has been a reduction over time in TRM after “intensive” induction possibly due to better supportive care. Although various selection biases cannot be excluded, this decline is not due to younger age or better performance status and needs to be considered when choosing AML induction therapy.

No relevant conflicts of interest to declare.