Clonal Evolution of FLT3 -ITD Positive AML in Patients Treated with Midostaurin in Combination with Chemotherapy within the Ratify (CALGB 10603) and AMLSG 16-10 Trials

Background: Internal tandem duplications (ITD) in the receptor tyrosine kinase FLT3 gene are found in about 22% of acute myeloid leukemia (AML) patients (pts) and confer a poor prognosis depending on the mutation (mut) load. Recently, the multi-targeted tyrosine kinase inhibitor (TKI) midostaurin (M) was reported to improve outcome in FLT3 -ITD positive (FLT3 -ITD^pos) and FLT3 -TKD^mutAML pts in combination with intensive chemotherapy [Stone et al., NEJM 2017; RATIFY (CALGB 10603) trial]. However, a significant proportion of pts relapsed after initial response due to yet unknown mechanisms.

Aims: To study the clonal evolution in paired samples obtained at diagnosis (D) and relapse (R) by whole exome sequencing (WES) in FLT3 -ITD^pos AML pts enrolled by the AMLSG and treated in the RATIFY (NCT00651261) or AMLSG 16-10 (NCT01477606) trial.

Methods: WES was performed in 25 pts with FLT3 -ITD^pos AML at D and R using the Nextera Rapid Capture Exome kit (Illumina) for library preparation followed by sequencing on an Illumina HiSeq platform. NPM1 mut and FLT3 -ITD/TKD mut were determined according to standard protocols; FLT3 -ITD was analyzed with respect to the allelic mut to wild type ratio (AR), nucleotide (nt) length, insertion site [juxtamembrane domain (JMD), TKD1] and number of the ITD clones. Nine pts were treated within the RATIFY trial which used intensive chemotherapy during induction (I) and consolidation (C) combined with either M or placebo (PBO) followed by a 1-year maintenance therapy with M or PBO (not yet unblinded for this analysis). Sixteen pts were treated in the AMLSG 16-10 trial which included intensive chemotherapy during I and C plus M followed by 1-year maintenance therapy with M; 4/16 pts in the AMLSG 16-10 trial received allogeneic hematopoietic cell transplantation in 1^st complete remission.

Results: The median AR of FLT3 -ITD was [0.4 (0.1-18.9)] and [0.7 (0.07-18.9)] at D and R, respectively. FLT3 -ITD was undetectable at R in 10 (40%) pts; a change of the FLT3 -ITD clone at R was found in 9 (36%) pts, with a switch of the insertion site in 7 (28%) pts. There was no difference of the median ITD length between D [42 (range, 18-198) nt] and R [42 (range, 18-93) nt] or the distribution of the ITD insertion sites in the JMD/TKD1 region at D and R. Using WES (average coverage 122 x), a total of 795 mut were identified at D and R. The average number of mut per sample was 16 (range, 7-23) at D and 16 (range, 5-29) at R. A total of 378 (48%) mut were found at both time points; 213 (27%) mut were present only at D and 204 (25%) mut only at R. The percentage of constant mut at D and R among all pts ranged from 9% to 85% (median 48%). Recurrently mut genes at D were NPM1 (n=12), DNMT3A (n=10), WT1 (n=6), TET2 (n=6), IDH1/2 (n=4), NRAS (n=4), FLT3 -TKD^D835Y (n=3), ASXL1 (n=2), CEBPA (n=2), RUNX1 (n=2) and NF1 (n=2). Of those, mut in epigenetic modifying genes (DNMT3A, TET2, IDH1/2, ASXL1) persisted in 16/17 pts (98%) at R and NPM1 mut in 10/12 pts (83%) at R. Mut in hematopoietic transcription factors (RUNX1,CEBPA, WT1) were stable at R in most pts (9/10; 90%) and acquired in 4 pts (WT1 n=3; RUNX1 n=1) at R. Signaling related gene mut showed a high variability between D and R: FLT3 -TKD^D835Y mut were lost in all pts (3/3) at R; NF1 and NRAS mut were lost in 3/6 pts (50%) while 1 pt gained a NRAS mut at R. A FLT3^N676K resistance mut was only found in 1 pt with a variant allele frequency of 4.3% at R. Pathway analysis revealed that mut related to the spliceosome complex were enriched at D, whereas mut involved in TP53 signaling were present at D and R and mut related to mitogen-activated protein kinase signaling were enriched at D and R but varied between both time points.

Conclusions: FLT3 -ITD was lost in 40% of pts at R. M associated FLT3 resistance mut were not recurrently found at R suggesting FLT3 independent mechanisms driving resistance to treatment with chemotherapy and M. The stability of mut present at both D and R showed a wide range (9% - 85%) among all pts and varied between functional groups: Mut in epigenetic modifying genes that commonly occur early in the clonal evolution were most stable between D and R, whereas signaling, spliceosome and transcription related gene mutations were lost or acquired at R, indicating heterogeneous mechanisms of clonal evolution. Larger cohorts of pts are currently under investigation to further characterize mut patterns underlying resistance to therapy with M in FLT3 -ITD^pos AML.

KD and LB contributed equally to this work.