Recurrent Mutations in CCND3 Confer Clinical Resistance to FLT3 Inhibitors

Background. Activating mutations in FLT3 occur in ~30% of adult acute myeloid leukemia (AML) cases, including internal tandem duplication (ITD) mutations (~25%) and point mutations in the tyrosine kinase domain (KD). In recent years, multiple selective and potent FLT3 inhibitors such as quizartinib, PLX3397, crenolanib and ASP2215 have demonstrated encouraging preliminary clinical activity in FLT3 mutant AML. Both quizartinib (Schiller et al, ASCO 2014) and ASP2215 (Levis et al, ASCO 2015) have reported composite complete remission rates (CRc) as high as ~50% in phase II trials. Despite these high response rates, the majority of patients relapse after initial response (acquired resistance) and a significant proportion of patients also fail to respond at all (primary resistance). On-target resistance due to secondary KD mutations in FLT3 is a common cause of acquired resistance to the clinically active FLT3 inhibitors quizartinib (Smith et al. Nature 2012) and PLX3397 (Smith et al, Cancer Discovery 2015). However, the causes of primary clinical resistance to FLT3 inhibitors have not been characterized. We performed targeted capture based sequencing of sorted pre-treatment blasts from 8 responding (R) and 21 non-responding (NR) patients treated on the phase I/II trial of PLX3397 in FLT3-ITD+ AML. We deeply sequenced the coding regions of 585 genes known to be mutated in hematologic malignancies/solid tumors and identified genes mutated only in NR patients (compared to genes mutated in both R and NR patients).

Results. The number of mutations detected in genes other than FLT3 ranged from 2-18 per sample. There was no increase in the number of mutations found in patients with pre-existing hematologic conditions or in NR patients. Surprisingly, one of the most frequently mutated genes observed exclusively in NR patients was CCND3, the gene encoding cyclin D3, which has rarely been reported to be mutated in AML, though it is mutated in 38% of sporadic Burkitt's lymphoma (BL). A total of 4 individual mutations in CCND3 (Q276*, Q280fs, R271fs, and T283A) were identified in 3/21 NR patients (one patient had both Q276* and Q280fs). No CCND3 mutations were found in R patients. The identified mutations were the same mutations commonly found in BL, known to result in a more stable isoform of cyclin D3 and retain sensitivity to CDK4/6 inhibitors (Schmitz et al., Nature 2012). Expression of the Q276* and T283A mutations in FLT3-ITD+ MV4;11 cells conferred resistance to apoptosis induced by several FLT3 inhibitors (PLX3397, AC220 and crenolanib). However, inhibition of CDK4/6 activity in CCND3 mutant MV4;11 cells by either the CDK4/6 inhibitor palbociclib or the combined FLT3-CDK4/6 inhibitor AMG925 (FLX925) was unable to restore sensitivity to FLT3 inhibition. Moreover, CCND3 mutant MV4;11 cells demonstrated no increase in Rb phosphorylation, suggesting resistance to FLT3 inhibitors facilitated by CCND3 mutations is not predicated on CDK4/6 activation of Rb-dependent E2F-mediated transcription.

Conclusions. We have identified recurrent mutations in CCND3, a gene not previously known to be commonly mutated in AML, as a novel cause of clinical primary resistance to FLT3 inhibitors in AML. This represents the first report of a specific non-FLT3 dependent mechanism of clinical resistance to FLT3 inhibitors. Unlike in BL, the functional effects of CCND3 mutations in FLT3-ITD+ AML do not appear to be mediated via CDK4/6 activity and is therefore unresponsive to CDK4/6 inhibition. The molecular mechanism(s) of CCND3-mediated resistance to FLT3 inhibitors are currently being investigated.