Persistence of DNMT 3A does not influence clinical outcome in acute myeloid leukaemia

The advent of next-generation sequencing (NGS) has led to the way to the identification of a plethora of additional mutations in acute myeloid leukaemia (AML). These mutations, genetic and epigenetic alterations, have all been reported to positively or negatively associate with AML outcome. Some of the mutations identified prior to NGS have been extensively studied, such as the internal tandem duplication (ITD) of the FLT3 gene, mutations of the NPM1 or CEPBA genes and have been included in the World Health Organization classification of AML as defining specific subcategories of patients with normal karyotypes (Arber et al, 2016). The DNA (cytosine-5)-methyltransferase 3 alpha gene (DNMT3A) was identified by NGS and was shown to be mutated in approximately 20–35% of patients (Ley et al, 2010); this places it as one of the most frequently mutated genes in AML (Thol et al, 2011). The majority of mutations occur in the methyltransferase domain and involve the R882 codon. The value of determining minimal residual disease (MRD) levels has been clearly demonstrated in blood cancers notably with the regular monitoring of the BCR-ABL1 levels in chronic myeloid leukaemia (CML) (Hanfstein et al, 2012) and PML-RARA in acute promyelocytic leukaemia (APL) (Grimwade et al, 2015). In each of these diseases, the failure to reduce the leukaemic clone burden below the level of conventional polymerase chain reaction (PCR) detection or an increasing level of the fusion genes has been shown to be associated with the impending onset of disease resistance or clonal evolution. These signs could be used to alter therapeutic options to improve the patient’s outcome trajectory. The relevance of the prognostic significance of persistent DNMT3A mutations has been examined in four studies prior to the one reported in this issue. Two have reported that the presence of DNMT3A mutations is associated with a poor clinical outcome (Hou et al, 2012; Klco et al, 2015); whilst the other two reported no difference in outcome (Ploen et al, 2014; Debarri et al, 2015). There are also differences in the size of the sample population, detection methodologies and the type of mutations analysed. To clarify this situation, in this issue of bjh, Bhatnagar et al (2016) have reported a study on a larger cohort of AML patients who had a dominant clone with DNMT3A R882 mutation at diagnosis. All the patients received intensive induction treatment within Cancer and Leukemia Group B (CALGB) trial protocols. Additional samples were then obtained when the patients had achieved a complete remission (CR) and both the diagnostic and CR samples were analysed using a targeted NGS approach covering 35 genes. For the determination of positivity of DNMT3A mutations, a detection level cut-off of >3% was used. It should be noted that this study only included patients with a R882 mutation in DNMT3A and the influence of other mutations was not assessed. At diagnosis, all the patients had at least one mutation in addition to the DNMT3A, with the level of DNMT3A mutation between 40–53%. However, the interesting aspect was observed after patients had obtained CR where two cohorts were identified: in approximately 24% of patients (Cohort 1) no mutations were found; whilst in Cohort 2 (76%) a DNMT3A mutation persisted beyond CR with levels ranging from 3% to 45%. Surprisingly, there was no significant difference between the two cohorts with regard to overall survival (OS) or disease-free survival (DFS), irrespective of the time that the CR was taken post morphological confirmation of CR or if the sample was within 30 days of CR being confirmed. This is contrast to expectation from studies on CML and APL where the clearance or non-detection of a mutation, albeit a fusion gene rather than a point mutation, is highly significant in terms of predicting outcome. However, the persistence of DNMT3A R882 mutation post-CR appears not to have any impact on outcome. Within Cohort 2, DNMT3A mutations alone were detected in around 3/4 (~73%) of the patients (Cohort 2a), with the remaining 12 patients (~27%) having a DNMT3A mutation and one additional mutation (Cohort 2b) mainly TET2, TP53 or ASXL1. Although no significant difference was seen for OS between Cohorts 2a and 2b, the authors Correspondence: Professor Ken Mills, CCRCB, Queen’s University Belfast, 97 Lisburn Road, Belfast BT9 7AE, UK. E-mail: k.mills@qub.ac.uk editorial comment