Introduction The DNMT3A gene is recurrently mutated in normal karyotype acute myeloid leukemia (NK-AML) and the most common alteration (65% of cases) is located at residue R882 of methyltransferase domain. DNMT3A mutations often occur in association with other mutations such as FLT3 and NPM1 and while it is now generally accepted that AML patients with NPM1 or FLT3 -ITD mutations have favourable and poor outcome respectively, the impact of DNMT3A mutations on survival remains controversial. Recent studies reported that DNMT3A -mutated cells are detectable in patients with AML in long-lasting complete remission, questioning whether quantification of DNMT3A levels could be used to minimal residual disease (MRD). To address this issue, we carried out sequential monitoring studies of DNMT3AR882H levels in AML patients and compared our results with those obtained with multiparametric flow-cytometry (MPFC) and NPM1 molecular status and expression levels.

Methods The mutational analysis of DNMT3AR882, NPM1 and FLT3 (ITD/TKD) was assessed at diagnosis in 556 patients with de novo AML (median age: 49 years, range 16-89 years) enrolled in clinical trials conducted by the Gruppo Italiano Malattie EMatologiche dell'Adulto (GIMEMA) during the period 2012-2015. DNMT3AR882H molecular status was also studied in 100 patients with chronic myeloid leukemia (CML) and in 56 elderly healthy individuals. The expression levels of DNMT3AR882H and NPM1 Type_A mutation (NPM1mutA) were longitudinally analysed by qRT-PCR in 21 AML patients. For patients achieving complete morphologic remission (CR), DNMT3AR882Hlevels were compared with those of NPM1mutA and with MRD assessed by multiparametric flow cytometry (MPFC). W e then compared the levels of DNMT3AR882H and of MPFC-MRD in follow-up BM-samples from AML patients in CR , classified as MRDneg or MRDpos, according to threshold established based on previous work from our group (Buccisano et. al. Blood 2010). Finally, we evaluated the kinetics of DNMT3AR882H, NPM1mutA and FLT3 -ITD in 9 FLT3 -ITD+ve patients, using a FLT3 -ITD patient-specific qRT-PCR, and in 6 FLT3 -wt patients.

Results Expression levels of DNMT3AR882H was also studied in sorted cell subpopulations from AML and CML patients and in healthy individuals. Distinct from NPM1mutA, DNMT3AR882H expression levels did not correlate with leukemic blast proportion assessed by MPFC (Figure 1) and decreased in AML samples only after induction treatment (Figure 2), while it was constantly detectable during remission and at relapse (Figure 3). Within FLT3 -ITD+ve AML, the kinetics of DNMT3AR882H paralleled that of NPM1mutA in 3 of 9 patients only. DNMT3AR882H expression levels were constant, independent from NPM1mutA in all 6 FLT3 -wt patients. Figure 4 shows 4 exemplary cases, with stable DNMT3AR882H levels during disease course in three of them. In CML patients, the DNMT3AR882H transcript was found only during follow-up in 4/100 patients, while it was undetectable at the time of CML diagnosis. DNMT3AR882H transcript levels were constantly low in all available CML samples. As previously reported, clonal hematopoiesis of indetermined potential (CHIP) was identified in 3/56 elderly healthy individuals. Similarly low expression levels of DNMT3AR882H were detected in blast and T-lymphocyte samples obtained from AML cases at CR and relapse phases, in CML patients and healthy individuals (Figure 5 and 6). There were no differences in DNMT3AR882H expression between MRDpos and MRDneg follow-up samples. DNMT3AR882H transcripts levels in MRDpos follow-up samples were significantly lower than those of samples harvested at diagnosis (Figure 6). These data show that DNMT3AR882H mutations probably arise in very early hematopoietic precursors in the context of CHIP. Interestingly, as reported in figure 6, DNMT3AR882H expression levels were similar in samples from CML patients, healthy individuals and AML follow-up samples.

Conclusions In summary, the results of our study indicate that DNMT3AR882H quantification is unreliable as a tool for AML monitoring. In fact, together with its detection in CML and healthy subjects, the presence of DNMT3AR882H at constant levels detected in AML during active disease or CR clearly point to a lack of specificity of this abnormality for the leukemic AML blast population.


Cicconi: TEVA: Speakers Bureau. Lo Coco: Lundbeck: Honoraria, Speakers Bureau; Novartis: Honoraria, Speakers Bureau; TEVA: Honoraria, Speakers Bureau.

Author notes


Asterisk with author names denotes non-ASH members.