DNMT3A is a Powerful Follow-up Marker in NPM1 mutated AML

Background:NPM1 mutated (mut) FLT3-ITD negative acute myeloid leukemia (AML) is a distinct prognostically favorable subtype of AML. Robust data is available demonstrating that monitoring therapy response using NPM1mut-specific real time PCR is an important tool to early detect relapses and provides important information to guide therapy. Since next generation sequencing techniques have become available further gene mutations were detected that accompany NPM1mut in AML. Of these DNMT3Amut were the most frequent and stable ones (Krönke et al., Blood, 2013).

Aim: 1) Analyse the stability of DNMT3Amut in paired diagnostic and relapsed samples. 2) Evaluate whether monitoring of DNMT3Amut provides additional information to monitoring of NPM1mut.

Patients and Methods: Samples were selected from a cohort of 359 NPM1mut de novo AML cases with an available DNMT3Amut status.First, toevaluate the stability of DNMT3Amut paired diagnostic and relapse samples of 103 patients were analyzed (44 males, 59 females; median age 60 years, range: 26-82 years). Median time to relapse was 11 months (range: 3-68 months). NPM1mut status was assessed at diagnosis with a LightCycler melting curve analysis assay. Non type A mutations were further characterized by Sanger sequencing. Second, all diagnostic and follow-up samples (n=1,813) were quantified by real time PCR specific for the individual NPM1mut. Analysis for DNMT3Amut was performed using either the 454 technology (454 Life Sciences, Branford, CT) or the MiSeq instrument (Illumina, San Diego, CA). Deep DNMT3A sequencing of remission samples was performed using the 454 technology.

Results: Out of 103 paired samples 61 (59.2%) carried a DNMT3Amut at diagnosis. 57/61 (93.4%) patients stably retained the mutation at relapse, in 4 (6.6%) the DNMT3Amut was lost. On the other hand 2 of 42 (4.8%) cases with DNMT3A wildtype at diagnosis gained the mutation at relapse. Thus, DNMT3Amut status was shown to be relatively stable (97/103; 94.1%) and thus qualifies as a promising target for follow-up controls.

For comparison of DNMT3Amut and NPM1mut status during follow-up 54 patients that were NPM1/DNMT3A double mutated at diagnosis were selected according to the availability of at least one sample in first remission with an NPM1mut level <0.01%. These samples were reanalyzed by deep sequencing for the respective DNMT3A amplicons that had identified mutations at diagnosis. Two of these 54 cases (3.7%) showed morphologic relapse but NPM1mut was negative at relapse (sensitivity of 10^-7). In one of these two cases at diagnosis NPM1mut and TET2mut were observed while at relapse IDH1mut and RUNX1mut were present. However, at both time points the DNMT3Amut was identified. The second case lost NPM1mut and CEBPAmut and retained the DNMT3Amut and TET2mut. Thus in these 2 cases the DNMT3Amut can be regarded as the common ancestor. 1 case retained the NPM1mut at relapse but lost the DNMT3Amut. Of note, in 32/54 (59.3%) cases the DNMT3Amut persisted in the remission samples (NPM1mut low level <0.01% or negative) with high DNMT3Amut loads (median: 20%, range: 2-59%) that was only slightly below the load at diagnosis (median: 45%, range: 38-58%).

To analyze the clinical importance of these persisting DNMT3Amut survival analysis was performed. Median overall survival for patients with persisting DNMT3Amut (n=32) was 69 vs 96 months in those who lost also the DNMT3Amut in remission (n=22, p=0.053). Median event free survival was 38 vs. 96 months (p=0.031). Thus the DNMT3A mutational status in remission of NPM1mut AML is a further important parameter for prognostication.

The mechanisms underlying this observation are obscure. As NPM1mut disappeared in remission and DNMT3A was retained and with the exception of 2 cases all others (n=18) relapsed with an NPM1mut (even the same type as at diagnosis) two mechanisms may be discussed: 1) Persisting DNMT3Amut cells predispose by a yet unknown mechanisms to the development of a secondary NPM1mut or 2) a residual DNMT3Amut/NPM1mut very low level survivor is able to overgrow the DNMT3Amut sole mutated clone in remission and cause relapse.

Conclusions: 1) DNMT3Amut persist in remission of NPM1mut AML in the majority of cases (59.2 %). 2) DNMT3Amut analysis in remission of NPM1mut AML is an important parameter for prognostication. 3) Clones with DNMT3Amut as the sole mutation may have a normal phenotype and thus DNMT3Amut may even be regarded as premalignant mutation.