The association of NPM1mut and FLT3-ITD in de novo acute myeloid leukemia (AML) with intermediate-risk cytogenetics has different prognostic impact depending on the FLT3 allelic burden. Previous studies published by our cooperative group showed that patients with de novo AML of intermediate-risk cytogenetics with NPM1mut and FLT3-ITD low ratio (<0.5, FLT3low) at diagnosis presented an overall survival and relapse rate similar to those with NPM1mut and FLT3wt. Therefore, in the CETLAM-2012 protocol, patients with FLT3low NPM1mut AML are not considered for allogenic hematopoietic stem cell transplant (allo-HSCT) in first complete remission (CR1). Recent studies suggest that the co-occurrence of DNMT3A mutation in FLT3-ITD NPM1mut AML patients confers a worse prognosis regardless of FLT3-ITD ratio.

We analysed our data to determine whether these findings were confirmed in our cohort, specifically in the low FLT3-ITD ratio patients, since this could have therapeutic implications.

Methods and patients

A total of 163 patients with de novo AML, intermediate-risk cytogenetics and NPM1mut were analysed (median age 53 years (18-72); male:female 72:91 (0.79)). Eighty patients (49%) harboured an FLT3-ITD, with a high allelic ratio in 42 of 76 patients with available ITD/wt ratio (55%). They were included in the AML-2003 (n=49) and AML-2012 (n=114) CETLAM protocols. Proportion of patients undergoing alloHSCT in CR1 is detailed in table 1.

Bone marrow samples from diagnosis were studied for DNMT3A mutations as previously described. The definition of complete remission (CR), overall survival (OS), leukemia-free survival (LFS) and risk of relapse (RR) followed recommended ELN criteria. The Kaplan-Meier method was used to estimate the distribution of LFS and OS, for RR cumulative incidence was used.


Out of the 163 patients with AML of intermediate risk cytogenetics and NPM1mut, 78 presented DNMT3A mutations (48%). Of these, 62 (79%) presented mutations in codon R882 or corresponded to DNA insertions/deletions while 16 (21%) harboured missense mutations. Presence of DNMT3A mutation did not associate with FLT3-ITD (ITD/85 DNMT3Awt vs ITD/78 DNMT3Amut, p=0.394). In the entire cohort, 5-year OS, LFS and RR were 58±4.5%, 59±4.6% and 27±13.9%. FLT3-ITD ratio confirmed its prognostic impact when analysing FLT3wt (n=83) vs FLT3low (n=34) vs FLT3high (n=42) patients (5-year OS of 68±6% vs 62±8.7% vs 37±8.6%; p=0.002; and 5-year RR of 18±9.4% vs 27±16.1% vs 41±23.2%; p=0.023). On the contrary, DNMT3Amut did not exert any effect on overall outcome (5-yr OS DNMT3Awt vs DNMT3Amut 61±6.2% vs 55±6.2%; p=0.234)

When DNTM3A mutational status was considered, the impact of FLT3-ITD on outcome was mitigated in wild-type DNMT3A population. Thus, we found that DNMT3Awt patients presented no statistical differences in OS according to FLT3 mutational status or ratio: FLT3wt (n=46) vs FLT3-ITD (n=39) was 67±8.5% vs 57±8.2%; p=0.122, whereas FLT3wt (n=46) vs FLT3low (n=18) vs. FLT3high (n=19) was 67±8.5% vs. 66±11.5% vs 46±11.8%; p=0.088 (image 1A).This was also seen in relation to LFS and RR according to FLT3 ratio: 5-yr LFS of FLT3wt vs FLT3low vs FLT3high was 72±7.9% vs 61±12.6% vs 51±13.4%; p=0.244 and 5-year RR of the same groups: 19±8.8% vs 26±12.5% vs 27±21.9%; p=0.724 (image 2A).

In the DNMT3Amut group, patients with FLT3-ITD (n=41) presented shorter OS than those with FLT3wt (n=37) with an OS of 37±10.7% vs 69±7.8%; p=0.028. When FLT3 ratio was considered, FLT3wt (n=37) vs FLT3low (n=16) vs FLT3high (n=23) showed an OS of 69±7.8% vs. 58±13.2% vs 27±13.1%; p=0.038 (image 1B).

Similar results were seen in LFS according to FLT3 ratio (FLT3wt (n=29) vs FLT3low (n=16) vs FLT3high (n=20) 71±8.6% vs 53±12.9% vs 18±13.8%; p=0.012).

Finally, we observed significant differences in the 5-year RR when considering DNMT3Amut patients in relation to FLT3 ratio (FLT3wt vs FLT3low vs FLT3high 18±10.6% vs 27±20% vs 54±28.8%; p=0.021)(image 2B).


In this study, patients with NPM1mut and FLT3-ITDlow presented a similar outcome to patients with NPM1mut and FLT3wt regardless of DNMT3A mutational status. These results support the modification of alloHCST policy in CR1 in CETLAM-2012, which do not consider alloHSCT for patients with FLT3low. On the other hand, concurrence of DNMT3A mutation may have an added negative effect in patients with NPM1mut and FLT3-ITDhigh, which should be further confirmed in larger studies.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.