A pre-leukaemic reservoir

Early events in the development of leukaemias are not well defined. For example, it is unclear whether cells that are non-tumorigenic but contain clinically relevant alterations exist and, if so, whether these cells remain present through the course of the disease.

Credit: PHOTODISC

While using deep targeted sequencing of genes that are known to be mutated in leukaemia to analyse intratumour heterogeneity in patients with acute myeloid leukaemia (AML), John Dick and colleagues discovered that not only leukaemic samples but also normal T cells from the same patients carried mutations in DNA (cytosine-5-)-methyltransferase 3 alpha (DNMT3A^mut). Mutation of another known AML gene, nucleophosmin (NPM1), and the FLT3 internal tandem duplication (FLT3-ITD), often co-occur with DNMT3A^mut. Although coincident NPM1 mutations (NPM1c) and FLT3-ITD were present at similar allele frequencies to DNMT3A^mut in leukaemic blasts, they did not co-occur in T cells, which indicates that DNMT3A^mut arises early and occurs in cells that can give rise to T cells. Examination of several non-leukaemic, haematopoietic cell populations showed that DNMT3A^mut occurs without NPM1c in many lineages, including haematopoietic stem cells/multipotent progenitors (HSCs/MPPs). In HSCs/MPPs, the allele frequency of DNMT3A^mut was substantially higher than estimated for the clonal contribution from a normal HSC, which suggests that clonal expansion of these cells occurs.

pre-leukaemic HSCs have a competitive advantage over normal HSCs

Analysis of samples from five patients with AML, which were taken at diagnosis, remission and relapse, indicated that many haematopoietic cell populations at remission and relapse contained DNMT3A^mut but not NPM1c, and in one patient with long-term remission, an increase in the DNMT3A^mut allele frequency was observed over time. This suggests that cells carrying DNMT3A^mut can survive chemotherapy and can expand, potentially creating a reservoir of cells that are primed for relapse. To determine whether cells with DNMT3A^mut can function as HSCs, xenograft repopulation assays in immunocompromised mice were carried out with HSCs/MPPs that were derived from two different patients (who had DNMT3A^mut allele frequencies of 30% and 20% in these cells). Many of the mice had multilineage engraftment, with a large proportion of cells of different lineages carrying DNMT3A^mut but not NPM1c, which indicates that the DNMT3A^mut-carrying cells are indeed functional HSCs. Furthermore, the allele frequency of DNMT3A^mut in the mice increased over time, which suggests that these pre-leukaemic HSCs have a competitive advantage over normal HSCs.

A similar situation might occur in leukaemias with mutations in isocitrate dehydrogenase 1 (IDH1) or IDH2. The authors looked back at previous xenograft repopulation data from 25 patient AML samples that had generated multilineage grafts. Of these, 10 were from patients with DNMT3A^mut, and 12 had IDH1 or IDH2 mutations. Analysis of purified HSC and progenitor populations from six further patients with IDH1 or IDH2 mutations indicated that two of these patients had mutant IDH2 without NPM1c, which suggests that IDH2 mutations might also generate pre-leukaemic cells.

Several interesting insights arise from these results. In addition to shedding light on the order of mutational events in AML and showing that pre-leukaemic HSCs are present at diagnosis, the data suggest that these cells are resistant to chemotherapy and could contribute to relapse. Therefore, the presence of these cells might need to be included in the definition of minimal residual disease, and the eradication of these pre-leukaemic HSCs might be necessary to prevent relapse.