INTRODUCTION: Acute myeloid leukemia (AML) is the most common form of acute leukemia. Induction therapy usually consists of a cytarabine-anthracycline combination, such as daunorubicin and cytarabine (DA), a regimen which has been in use for over 30 years. There have been few advances in treatment and thus there are few options for induction therapies that prolong remission. Gemtuzumab ozogamicin (GO) was approved in the United States in 2000 as a treatment for relapsed AML. Although withdrawn in 2010, data from recent randomized controlled trials (RCTs) showed a lower dose of GO (3 mg/m2) in combination with DA led to improved event-free survival (EFS), relapse-free survival (RFS), and overall survival (OS) in the newly-diagnosed setting.

OBJECTIVE: Our aim was to conduct a systematic literature review (SLR) and network meta-analysis (NMA) evaluating the clinical efficacy and safety of GO + DA and other induction therapies for newly diagnosed ( de novo or secondary) AML.

METHODS: A systematic search was performed in MEDLINE (via PubMed), Embase, the Cochrane Central Register of Controlled Trials, and the Database of Abstracts of Reviews of Effects for RCTs published in English between January 2000 and September 2015 in accordance with published guidance (PRISMA, National Institute for Health and Care Excellence [NICE] and Cochrane) and a pre-defined protocol. Scientific conferences proceedings were also searched. A feasibility assessment evaluated data availability, clinical heterogeneity, and risk of bias to identify the set of studies suitable for the NMA. Fixed- and random-effects (FE; RE) Bayesian NMAs were conducted to compare the treatment effects of GO vs. its comparators following methodology recommended by the NICE Decision Support Unit for synthesis of RCT data. Base-case analysis included all trials reporting data on analyzable outcomes: OS, RFS, relapse rate, resistant disease, complete remission (CR), overall response rate (ORR), induction death, bleeding events (all grades) and grade 3+ adverse events (AE) such as infections, skin effects, and hepatic events. Two subgroup analyses were conducted: a) an analysis that pooled results from all GO arms and compared them to all non-GO arms (GO vs. non-GO) and b) an analysis limited to studies with a 100% de novo population. Results were considered significant if the 95% credible interval (CrI) did not cross 1 (roughly equivalent to a Bayesian pairwise probability p≥97.5%).

RESULTS: Seventeen trials (24 publications; N=14,218) evaluating 25 therapies (including GO + DA) with moderate-to-high risk of bias (using Cochrane criteria) and sufficiently similar study and population characteristics provided data for the NMA. Three GO trials were relevant to the base-case and subgroup analyses. Results of the FE and RE models were similar. FE results are reported as most outcome comparisons included data from no more than two trials.

In the base-case analyses, GO + DA led to significantly greater OS vs. DA alone (hazard ratio [HR] 0.69, 95% CrI [0.49, 0.97]), idarubicin + cytarabine (IA) (0.68 [0.47, 0.99]), and mitoxantrone + etoposide (ME) (0.52 [0.34, 0.80]), as well as significantly greater RFS vs. all evaluated comparators. Similar survival benefits were also observed in both subgroup analyses. Resistant disease occurred less frequently with GO + DA than ME (0.36 [0.16, 0.79]) in the base-case analysis and vs. mitoxantrone + cytarabine (0.65 [0.44, 0.96]) and ME (0.56 [0.34, 0.91]) in the GO vs non-GO subgroup analysis.

There was no difference between GO + DA and comparators for all other efficacy outcomes. Survival benefits apparent with GO + DA were contrasted with a higher rate of bleeding events compared with DA alone (odds ratio [OR] 3.33 [1.10, 12.33]), bevacizumab (7.93 [1.99, 36.49]), and IA (4.18 [1.17, 17.87]), as well as a higher rate of hepatic events compared with DA + fludarabine (DAFL) (OR 3.30 [1.14, 10.19]). There was no evidence of differences for skin effects and infections.

CONCLUSIONS: In this NMA, GO-based regimens, in particular GO+DA, led to improved OS and RFS among patients with AML when compared to 24 other regimens. These survival benefits were balanced by a greater likelihood of a bleeding event compared with DA, bevacizumab, or IA or a hepatic event compared with DAFL. The "GO vs. non-GO" and 100% de novo subgroup analyses results were consistent with the base-case, for most endpoints, including survival benefits.


Ashaye: Pfizer Inc: Other: Contracted by Pfizer Inc in connection with the conduct of this study, and drafting of this manuscript.; Evidera Inc.: Employment. Khankhel: Pfizer Inc: Other: Contracted by Pfizer Inc in connection with the conduct of this study, and drafting of this manuscript.; Evidera Inc.: Employment. Xu: Evidera Inc.: Employment; Pfizer Inc: Other: Contracted by Pfizer Inc in connection with the conduct of this study, and drafting of this manuscript.. Fahrbach: Pfizer Inc: Other: Evidera employees work with a variety of companies and are precluded from receiving payment directly from these organizations. Evidera received funding from Pfizer Inc. to participate in the study and develop this manuscript.; Evidera Inc.: Employment. Mokgokong: Pfizer Ltd: Employment; Pfizer Inc: Equity Ownership. Orme: Pfizer Inc: Other: Contracted by Pfizer Inc to perform consultative services.; ICERA Consulting Ltd: Employment. Lang: Pfizer Inc: Equity Ownership; Pfizer Ltd: Employment. Cappelleri: Pfizer Inc: Employment, Equity Ownership. Mamolo: Pfizer Inc: Employment, Equity Ownership.

Author notes


Asterisk with author names denotes non-ASH members.