Exome Sequencing Informs Mechanisms of Clinical Resistance to the FLT3-D835 Inhibitor Crenolanib

Introduction: Variation of the D835 residue of FLT3 occurs in ~5% of de novo AML cases and is a prominent feature of drug resistance in the setting of FLT3-ITD. Crenolanib treatment of AML patients exhibiting FLT3 D835 and/or FLT3-ITD positive tumors has yielded significant clinical responses. However, these clinical responses to single-agent crenolanib were transient, so we performed whole exome sequencing to determine the mechanisms of crenolanib resistance. Mutation of the FLT3 F691 "gatekeeper" residue is a known resistance mechanism to crenolanib and other FLT3 inhibitors in vitro. Hence, secondary mutation of FLT3 at the gatekeeper or other sites was hypothesized to be a predominant mechanism of resistance. We report that a minority of patients acquired FLT3 gatekeeper or other secondary FLT3 mutations, however, the majority of cases relapsed in the absence of additional FLT3 mutations indicating alternative mechanisms of drug resistance.

Methods: Genomic DNA was isolated from bone marrow aspirates from 44 AML patients treated with crenolanib. We performed whole exome sequencing using Illumina Nextera exome capture and paired end sequencing on an Illumina 2500 HiSeq. Paired specimens (pre/post-treatment) were available for 24 of these patients, allowing for analysis of acquired mutations in the context of crenolanib therapy.

Results: Although secondary mutation of FLT3 at the gatekeeper residue, FLT3 F691, was predicted to be a prominent mechanism of resistance, we did not observe FLT3 gatekeeper mutations in most patients. In total, only 3/44 (7%) of cases exhibited a FLT3 gatekeeper mutation despite good coverage of the FLT3 locus (average of 134 fold read depth; range of 77-177 fold average read depth across all 44 cases). In addition, we observed secondary FLT3 mutations at alternative residues, D200N, K429E, and L601F, in 3 other patients. Hence, only 6/44 (14%) of patients exhibited evidence of relapse due to secondary mutation of FLT3, and the majority of patients (38/44; 86%) presented with crenolanib resistance in the absence of secondary FLT3 mutations. Exome analysis revealed that many of these cases acquired mutations involving transcriptional regulators, suggesting alternate pathways of escape as the predominant mechanism to crenolanib therapy.

Conclusions: Crenolanib is broadly effective against FLT3-ITD and FLT3-D835 mutant AML. Despite predictions of FLT3 gatekeeper mutations (or other secondary FLT3 mutations) as a primary mechanism of resistance, we only observe these events in a minority of patients. Instead, we observe a prominent signal of acquired mutations in transcriptional regulators, suggesting a more elaborate genetic/epi-genetic mechanism of resistance to crenolanib.