Comprehensive Molecular Profiling of FLT3-Mutated Acute Myeloid Leukemia (AML) Patients Treated within the Ratify Trial (Alliance C10603)

Background: Recently, the oral multitargeted small molecule FLT3 inhibitor midostaurin (M) was approved for treatment of FLT3-mutated AML in combination with standard chemotherapy. In the international RATIFY (NCT00651261) trial, addition of M led to superior overall and event-free survival compared to placebo, thus defining a new standard of care in this AML subset (Stone RM et al. NEJM 2017). Although not powered for subgroup analyses, M showed consistent effects across all FLT3 mutation strata [tyrosine kinase domain (TKD); internal tandem duplication (ITD) with low (0.05-0.7; ITD^low) or high (>0.7; ITD^high) allelic ratio] suggesting significant off-target activity beyond FLT3 inhibition.

Aim: We aimed to comprehensively profile the mutational landscape of FLT3 mutated (FLT3^mut) AML in a large, well characterized cohort of patients (pts) treated within the RATIFY trial using a high-throughput targeted sequencing (HTS) approach.

Methods: HTS was performed on the entire coding region of 262 genes involved in hematologic malignancies including 20 genes that encode kinases targeted by M (M kinome, MK). Pretreatment peripheral blood (PB; 14%) or bone marrow (BM; 86%) specimens were available from 475 (66%) of 717 FLT3^mut AML RATIFY pts. Libraries were prepared using SureSelectXT custom solutions (Agilent). Paired-end sequencing was carried out on a HiSeq platform (Illumina). FLT3 mutation (mut) status was available for all pts [TKD: 24%; ITD: 76% (ITD^low: 45%; ITD^high:31%)], and cytogenetic data for 454 pts (96%).

Results: An average sequencing depth of 978x was obtained for all pts. In sum, 1815 mut (missense: 49%; indels: 40%; nonsense: 7%; other: 3%) were identified with a mean of 3.8 mut per pt (FLT3 strata; TKD: 4; ITD^low: 4; ITD^high: 3.6).Overall, recurrent mut (>5% of all pts) were found in NPM1 (61%), DNMT3A (39%), WT1 (21%), TET2 (12%), RUNX1 (11%), NRAS (11%), PTPN11 (9%), ASXL1 (8%), IDH1 (8%), IDH2 (7%; R140 only), and SMC1A (6%). In contrast, TP53 (1%) and biallelic CEPBA (1%) mut were rare events. This was also true for aberrations of the MK (7% in total) with KIT (2%), MAP3K11 (1%), and NTRK3 (1%) being most frequently mutated. When stratified according to FLT3^mut type, mut in NRAS (24% vs 7%, p<.0001), SMC1A (10% vs 4%, p=.02), and KIT (4% vs 1%, p=.02) occurred significantly more often in TKD than ITD groups, respectively, whereas WT1 (13% vs 24%, p=.018) was more frequently co-mutated in the ITD group. In general, pts in the TKD group had significantly more mut in genes encoding for cohesin (TKD: 29% vs ITD: 16%, p=.004) and signaling (TKD: 40% vs ITD: 24%, p=.001) proteins compared to ITD pts, who had significantly more mut in transcription genes (TKD: 37% vs ITD: 48%, p=.03).

Based on the mut and cytogenetic data, we next sought to assign all FLT3^mut pts to the 11 recently defined molecular AML classes (Papaemmanuil E et al. NEJM 2016). The majority fell into two classes, namely the NPM1 (N; 62%) and the chromatin-spliceosome (CS; 15%) classes, underscoring the significance of FLT3^mut as the driver in these particular genomic classes. Other class-defining lesions were rare or absent in this cohort [inv(16): 2%; t(8;21): 2%; t(11q23;x): 2%; t(6;9): 1%, TP53-aneuploidy: 1%; CEBPA^biallelic: 1%; IDH2^R172: 0%]. In 14% of all pts categorization was not possible (no or >1 class-defining lesion), emphasizing the need for further refinement of this classification. When focusing on these two groups, N and CS had comparable FLT3^mut patterns (TKD: 24% vs 21%; ITD^low: 44% vs 45%; ITD^high: 32% vs 33%), whereas N more frequently correlated with a normal karyotype (N: 91% vs CS: 63%). With respect to clinical characteristics, no differences between N and CS in terms of age, white blood cells, platelets, PB and BM blasts, as well as history of MDS were observed.

Conclusion: In this comprehensive sequencing approach, we could further delineate the molecular pattern of FLT3^mut AML. Here, FLT3-ITD and -TKD groups exhibited remarkable differences in cooperating pathways, highlighting distinct biologic features in the leukemogenesis of FLT3^mut AML. Overall, mut of MK genes were rare events, not fully explaining the complexity of M off-target effects. Understanding the underlying disease mechanism will potentially provide useful information on prognosis and prediction of response to M. Further analyses including correlation with clinical outcome are ongoing.

Support: U10CA180821, U10CA180861, U10CA180882, U24CA196171