The comparative effectiveness of glasdegib in combination with low-dose cytarabine versus azacitidine by bone marrow blasts counts among patients with newly-diagnosed acute myeloid leukemia who are ineligible for intensive chemotherapy.

Abstract

e19512 Background: Acute myeloid leukemia (AML) is an orphan disease with one of the lowest five-year survival rates among myeloid malignancies. Recently, a randomized, open label study among previously untreated, chemotherapy-ineligible AML patients demonstrated improved overall survival (OS) among patients treated with glasdegib (GLAS) + LDAC compared with LDAC alone. Two trials of AZA vs. conventional care regimens report data by bone marrow blast (BMB) counts: one with 20-30% and the other with >30%. In the absence of head-to-head comparisons, this study aims to perform the indirect treatment comparison between GLAS+LDAC and AZA by BMB counts. Methods: As there were potential imbalances between the GLAS and AZA trials and within the AZA trial arms in the baseline characteristics (e.g. poor cytogenetics% and de novo%), simulated treatment comparisons (STCs) for GLAS+LDAC vs. LDAC were performed to derive robust estimation by adjusting for the imbalances in the baseline effect modifiers. Afterwards, the classical network meta-analysis (NMA) was conducted. To derive the hazard ratio (HR) of GLAS+LDAC vs. AZA, three NMAs were conducted in each BMB group. Each NMA used a different HR of GLAS+LDAC vs. LDAC: 1) an unadjusted HR (classical NMA), 2) an STC adjusted HR adjusting for potential imbalances between the trials, and 3) an STC adjusted HR additionally accounting for potential imbalances between arms within the AZA trial. Results: In the 20-30% BMB group (N = 30), the OS HRs of GLAS+LDAC vs. AZA resulting from the three respective NMAs were as follows: 1) 0.46 [95% confidence interval: 0.10-2.14], 2) 0.31 [0.06-1.69], and 3) 0.36 [0.06-2.15]. In the > 30% BMB group (N = 80), the HRs were 1) 0.69 [0.39-1.20], 2) 0.48 [0.23-0.97], and 3) 0.48 [0.24-1.00]. All the HRs suggest that patients with GLAS+LDAC have a survival advantage over patients with AZA. Conclusions: Both the classical NMAs and the NMAs based on the STC adjusted HRs correcting for the potential imbalances at baseline suggest that GLAS+LDAC may be preferred over AZA as a treatment option for previously untreated chemotherapy-ineligible AML patients regardless of BMB counts. [Table: see text]