A Network Meta-analysis of Randomized Controlled Trials of Induction Treatments in Acute Myeloid Leukemia in the Elderly

Background: The optimal induction treatment of acute myeloid leukemia (AML) in the clinically and biologically heterogeneous group of elderly patients is not well-defined since direct comparisons between treatments is limited. Objective: The aim of this study was to estimate the relative effectiveness of induction treatments in AML elderly patients. Methods: A network of multiple treatments meta-analysis was performed by combining direct and indirect evidence from published randomized controlled trials (RCTs). The complete remission (CR) was considered as the outcome of interest. We systematically searched PubMed, EMBASE, and Cochrane Library until September 30, 2010, to identify all RCTs published in English that compared diverse induction chemotherapies in elderly AML patients. Our search strategy focused on RCTs that included AML patients > 60 years, considering, however, that age stratification was varied between countries and over time and studies were not necessarily designed to compare elderly patients. Regimens were grouped a priori into 42 different types of induction treatment. Prognostic parameters (age, performance status, unfavorable cytogenetics, antecedent malignancy), AML outcomes (median disease-free, overall survival, CR, induction deaths), and myelotoxicity parameters (median duration until neutrophil recovery [> 1.0 x 10(9)/L], platelet recovery [> 100 x 10(9)/L], hospitalization duration) were evaluated for each induction type. In combining direct and indirect evidence, we calculated the odds ratio (OR) for each therapy relative to the most commonly used combination of standard-dose daunorubicin (30-60 mg/m(2) for 3 days) and standard-dose cytarabine (100 mg/m(2) for 7-10 days) that was set as the reference induction treatment. Results: We identified 65 RCTs (15,110 patients) that described 64 direct comparisons of induction treatments, but only 14 of them showed significant differences in CR after random effects meta-analyses. Median age of included patients was 68, 18.0% had secondary AML, 21.1 A had poor performance status, 26.7% displayed unfavorable cytogenetics, and 49.3% finally achieved CR. No significant differences were observed in the recorded induction toxicity parameters among the treatment arms. Through the network meta-analysis, the addition of all-trans retinoic acids or lomustine to the combination of idarubicin plus cytarabine showed significantly higher CR rate (OR = 1.93 [1.06-3.49] and 1.76 [1.08-2.88], respectively), whereas no treatment, clofarabine, daunorubicin plus topotecan, and the 2 different schedules of gemtuzumab ozogamicin (at 1, 3, and 5 days and at 1 and 8 days) showed a significantly lower CR rate (OR = 0.01 [0.001-0.19], 0.15 [0.04-0.58], 0.03 [0.002-0.64], 0.06 [0.01-0.51], and 0.05 [0.01-0.32], respectively) than that of the reference induction. Median overall survival showed no difference between treatments (P = 0.150) but was significantly increased during the last 30 years (P < 0.001), presumably reflecting advances in AML management. Conclusions: Compared with the reference induction, significant differences were found for specific induction regimens, although most compared regimens appeared to have similar efficacy profiles. These results, however, should be interpreted with caution because the network was dominated by indirect comparisons. Data from large RCTs that make direct comparisons between treatments are needed to detect the optimal induction regimen for elderly AML patients. (Clin Ther. 2011;33:254-279) (C) 2011 Elsevier HS Journals, Inc. All rights reserved