Outcome of azacitidine therapy in acute myeloid leukemia is not improved by concurrent vorinostat therapy but is predicted by a diagnostic molecular signature

Purpose: Azacitidine (AZA) is a novel therapeutic option in older patients with acute myeloid leukemia (AML), but its rational utilization is compromised by the fact that neither the determinants of clinical response nor its mechanism of action are defined. Co-administration of histone deacetylase inhibitors, such as vorinostat (VOR), is reported to improve the clinical activity of AZA, but this has not been prospectively studied in patients with AML. Experimental Design: We compared outcomes in 259 adults with AML (n ¼ 217) and MDS (n ¼ 42) randomized to receive either AZA monotherapy (75 mg/m² 7 days every 28 days) or AZA combined with VOR 300 mg twice a day on days 3 to 9 orally. Next-generation sequencing was performed in 250 patients on 41 genes commonly mutated in AML. Serial immunophenotyping of progenitor cells was performed in 47 patients. Results: Co-administration of VOR did not increase the overall response rate (P ¼ 0.84) or overall survival (OS; P ¼ 0.32). Specifically, no benefit was identified in either de novo or relapsed AML. Mutations in the genes CDKN2A (P ¼ 0.0001), IDH1 (P ¼ 0.004), and TP53 (P ¼ 0.003) were associated with reduced OS. Lymphoid multipotential progenitor populations were greatly expanded at diagnosis and although reduced in size in responding patients remained detectable throughout treatment. Conclusions: This study demonstrates no benefit of concurrent administration of VOR with AZA but identifies a mutational signature predictive of outcome after AZA-based therapy. The correlation between heterozygous loss of function CDKN2A mutations and decreased OS implicates induction of cell-cycle arrest as a mechanism by which AZA exerts its clinical activity.