Dr Chris Pallister, Managing Director of SciMed Consulting, offers his interpretation of two recent publications that provide an insight into how gene sequencing could be used to identify individuals at higher risk of developing AML, and could also predict which post-remission AML patients are most likely to relapse.
The role of non-random somatic mutations present at diagnosis in the pathogenesis of AML is relatively well characterised. However, the increasing ease and rapidity with which large scale gene sequencing can be performed is taking us in new directions that may have profound implications for the way AML is diagnosed and treated.
Although these findings require confirmation in larger studies, this study caught my eye because it raises the possibility of identifying people at risk of developing AML several years before overt disease. It is only a small step from this to imagine a future where AML screening programmes are routine and interventional studies that aim to prevent disease progression may be possible!
The somatic mutational landscape at AML diagnosis is well-characterised, but much less is known about the pre-leukaemic mutational landscape of AML and how this relates to the the risks and kinetics of AML development. Desai et al. used serial samples obtained from the Women’s Health Initiative study to explore this question. 1, 2 They identified 212 women healthy at baseline who went on to develop AML (median time to AML 9.6 years) and the same number of age-matched controls that did not develop AML and performed deep sequencing on stored serial blood samples to characterize the changing pre-malignant genetic landscape.
They found that subjects with higher mutational complexity at baseline had a higher risk of developing AML and that mutations of TP53, IDH1/2, spliceosome genes (SRSF2, SF3B1 and U2AF1), TET2 or DNMT3A were particularly associated with an increased risk. Intriguing observations included that every subject with TP53 or IDH1/2 mutations developed AML, but that for DNMT3A and TET2 mutations the size of the allelic fraction appeared to be important. More rapid progression to AML appeared to be associated with the presence of baseline TP53 mutations, DNMT3A co-mutation with spliceosome genes and RUNX1 mutations.
As we have seen, AML results from progressive acquisition of somatic mutations in key driver genes and prior to frank AML development, pre-leukaemic clones may be identifiable. 1 It is well-known that pre-leukaemic clones can persist after remission induction treatment, but the precise prognostic significance of this has been uncertain.
Rothenberg-Thurley et al. used targeted sequencing to identify mutations in 68 AML-associated genes in paired pre-treatment and post-remission samples in 126 AML patients from the AMLCG-2008 study (NTC01382147). 3 All patients had at least one AML driver mutation at diagnosis, most commonly NPM1 (57 patients), DNMT3A (43 patients), and FLT3 (ITD 40 patients, TKD 20 patients) and had achieved CR or CRi after induction chemotherapy.
Persistence of ≥ 1 clonal mutation was found in 50 patients (40%), most often involving DNMT3A (65%), SRSF2 (64%), TET2 (55%), and ASXL1 (46%) genes. Clonal persistence was significantly associated with older age, inferior relapse-free survival and overall survival. Interestingly, alloHSCT abrogated the increased risk of relapse associated with these persistent pre-leukemic clones.
The key findings from this study is that persistence of pre-leukaemic clones following attainment of first-remission CR/CRi is an independent predictor of AML relapse. AlloHSCT appears to correct this increased risk. Future studies are needed to characterise the predictive value of clonal persistence and whether this finding could guide post-remission treatment decisions.