Differentiation Response to Quizartinib, a Potent 2^nd Generation FLT3 Inhibitor, in Relapsed FLT3 -Mutant AML Patients Correlates with Specific DNA Methylation Signature Prior to Therapy

Activating mutations in Fms-like tyrosine kinase 3 receptor ( FLT3 ) gene occur in 30% of AML patients, and are correlated with a high risk for early relapse. The second generation FLT3 inhibitor AC-220 (quizartinib), has shown significant responses in patients with relapsed and refractory FLT3 mutant AML. For those patients who respond, two different types of pathologic responses have been observed: a cytotoxic response or a differentiation response. Given these two different responses to therapy, we hypothesized that pre-existing epigenetic programs may be responsible for determining response type, and that DNA methylation analysis of AML blasts before and after treatment may help better understand the molecular underpinnings of the varied roles of FLT3 in AML pathogenesis. For this purpose, we analyzed a cohort of 13 FLT3 -ITD mutant AML patients relapsed after standard chemotherapy and treated with quizartinib as part of the Phase II ACE trial (NCT00989261). Patients were evaluated prior to therapy and after at time of maximal response and classified as having either cytotoxic (6 patients) or differentiation (7 patients) responses as previously described. Bone marrow or peripheral blood specimens were collected from 13 patients at baseline, prior to the administration of quizartinib, and 10 matched, post-treatment specimens were available for 9 patients. DNA was extracted from mononuclear cell fractions and analyzed using multiplexed Enhanced Reduced Representation Bisulfite Sequencing (mERRBS), which captures quantitative DNA methylation status at ~3 million CpG sites across the human genome. Supervised analysis between the different groups was performed using a beta binomial model as implemented in MethylSig, with significant differentially methylated regions (DMRs) being called with methylation differences ≥25% and false discovery rates (FDR) < 0.1. Notably, paired analysis of pre- and post-treatment samples revealed very little impact of quizartinib on the DNA methylation landscape of patients with cytotoxic responses, with only 21 DMRs detected. Similarly, those who exhibited a differentiation response, showed moderate methylation changes with 84 DMRs enriched at introns compared to whole coverage of mERRBS (Background [BG] 32% vs DMRs 56%; exact binomial test [EBT] p-value 5.053e-06). Thus, FLT3 inhibition did not appear to modify DNA methylation in these patients. However, and importantly, a direct comparison at baseline, prior to the administration of quizartinib, between patients who developed either differentiation or cytotoxic response, revealed robust epigenetic differences between these two groups of patients, with 580 DMRs. The majority of these (414 DMRs, 71%) were hypermethylated in patients with cytotoxic response. These DMRs were strongly depleted from promoter regions (BG 24% vs DMRs 9%; EBT p-value 2.2x10E-16) while they were enriched at intergenic regions (BG 35% vs DMRs 49%; EBT p-value = 1.064e-10). Given the preferential location of DMRs at these distal intergenic regions, we analyzed their relationship to enhancer elements. A total of 46.2% of DMRs (n=268) overlapped with ENCODE-defined enhancers (EBT p-value 3.57x10E-13 compared to 31.8% of BG). Pathway annotation of DMRs to the nearest genes revealed that they were associated with ion transmembrane transporter activity and extracellular matrix binding. In addition, amongst the DMRs were several annotated to HOX genes, including HOXB3 , HOXA7 and HOXD13 genes, which were hypomethylated in patients with a differentiation response. In summary, our analyses reveal that pre-existing epigenetic programs are encoded in FLT3 -mutant AML cells that correlate with the type of response they will undergo upon quizartinib treatment. The precise consequences of these epigenetic differences at the transcriptional level are still being explored and will be discussed at the meeting. We hypothesize that the pathologic role of FLT3 in AML may be modified by the epigenetic state of the transformed cells.