Relapse Kinetics in Acute Myeloid Leukemis Cases Encompassing More Than One Common Mutation: Towards a Unified Model

BACKGROUND: Molecular relapse (MR) of acute myeloid leukemia is defined as reversion to minimal residual disease (MRD) positivity and is now generally recognized to inevitably be followed by clinical relapse. The action to take upon a MR has still to be defined. The favorable use of Azacytidine in the case of relapse of NPM1+ AML or after allogeneic transplant has been reported (Sockel et al, Haematologica 2011, Platzbecker et al, Leukemia 2012), but most clinicians only exploit the pre-relapse period to optimize logistics for donor searches. We advocate for the relation between leukemic clone relapse speed and treatment effect onset to be an integral part of the decision process in all relapsing patients. To this end, we delineated relapse kinetics in CBFB-MYH11, RUNX1-RUNX1T1 and MLL translocated AML (Ommen et al, BLOOD 2010, B J Haematol 2014). Moreover, kinetics of AML carrying a balanced translocation were also dependent on secondary mutations, e.g. FLT3-ITD. In non-translocated AML, mutations generally coexist, and relapse speed determination in these patients will require the integrated analysis of contributions of individual mutations.

AIM: To delineate relapse speed in non-translocated AML on the basis of the separate and simultaneous presence of seven common mutations: FLT3-ITD, FLT3 D835, NPM1, CEBPA, cKIT, WT1 and IDH1.

PATIENTS AND METHODS: 31 relapses detected prior to clinical relapse using Wilms tumor gene (WT1) over-expression were included. Mutational status in the FLT3-ITD, FLT3 D835, NPM1, CEBPA, cKIT, WT1 and IDH1 loci was determined using fragment analysis based techniques.

RESULTS: At diagnosis, mutations were found in 10/25 (FLT3-ITD), 1/24 (FLT3 D835), 9/23 (NPM1), 1/23 (CEBPA), 0/23 (cKIT), 4/23 (WT1), and 5/21 relapses (IDH1). Due to the paucity of FLT3 D835+, CEBPA+ and cKIT+ cases, these were excluded from further analysis. As the NPM1 mutation is uniformly conserved at relapse, we only re-analyzed relapses for FLT3-ITD and WT1 and IDH1mutations. We found discrepancy of mutation status in 19% of FLT3-ITD (one gain, four losses), in 12% of WT1 (two gains, one loss) and in 8% of IDH1 (one loss, one gain) relapses. Including the patients in whom only relapse material was available, mutation was seen in 8/31 (FLT3-ITD), 9/31 (NPM1), 5/31 (WT1) and 5/31 (IDH1) relapses. We next determined relapse speed in peripheral blood based on WT1 expression at MR (WT1 expression relative to diagnosis level, median 0.0096 range 0.00042-0.24) and clinical relapse (WT1 expression, median 0.37 range 0.0086-12.1). Median time from molecular to clinical relapse was 55 days (range 13-135 days). We then constructed a linear regression model featuring mutational status in the four loci as covariates, first testing all possible mutation combinations, for evidence of a multiplicative effect of harboring more than one mutation. As none was detected, we next used the model to determine the contribution of individual mutations to pre-relapse leukemic growth. Compared to unmutated cases (leukemic clone doubling time 13 days, log change/month 0.7), we found clones harboring solely the WT1 mutation or FLT3-ITD to be significantly faster (leukemic clone doubling time 7 (P=0.04) and 8 (P=0.05) days, respectively, log chance/month 1.35 and 1.18, respectively) and a trend towards slower leukemic growth in NPM1+ clones (leukemic clone doubling time 23 days (P=0.2), log change/month 0.4). By contrast no effect of the presence of the IDH1 mutation was seen (leukemic clone doubling time 13 days, log change/month 0.69 (P=0.95)). The described additive effect of harboring several mutations allowed us to calculate relapse speed in these cases, i. e. for the very common NPM1+/FLT3-ITD+ combination, which was found to have a leukemic doubling time of 10 days (0.87 log/month). Using this model, after due confirmation and possibly expansion of the number of mutations included, we are now able to individualize assessment of AML relapsing clones, even those displaying a more complex interplay of genetic aberrations, thereby enhancing the decision-making process following a MR.

CONCLUSION: We have determined the respective relapse speeds of the most common genetic aberrations and combinations hereof in non-translocated AML. These data provide important information for researchers and clinicians evaluating MRD detection studies and deciding the optimal response to a recurrence of leukemia in AML patients.