Identifying prognostic constellations of driver mutations in acute myeloid leukaemia and myelodysplastic syndromes
Dr Charles Mullighan - St. Jude Children's Research Hospital, Memphis, USA
The late breaking abstract that we presented is a collaboration between St Jude Children’s Research Hospital and the Munich Leukaemia Laboratory. The goal of the study was to use integrated whole genome sequencing and whole transcriptome sequencing to better understand the genomic basis of acute myeloid leukaemia and myeloid dysplasia.
We know that these diseases are driven by genetic mutations, there has been a lot of work over many years to finding and identifying many of the recurrent mutations in AML and MDS but most centres and laboratories now use focussed genotyping of gene panels, limited numbers of genes, usually only looking at sequence mutations. This has led to the position where many cases are still not classified, we don’t have a full understanding of how the two diseases relate to each other at the mutational level and we don’t have a full understanding of how the different mutations or combinations of mutations influence treatment response and outcome.
The study used a cohort of over 1,000 patients of either AML or MDS that were newly diagnosed and all were subjected to comprehensive whole genome sequencing and transcriptome sequencing in a seven day timeframe in a clinical environment showing that this approach can be used in a clinically feasible way. There were a number of findings from the study so we confirmed many of the mutations and disease subtypes that have been previously identified by others and used in classifications from the World Health Organisation and the European Leukaemia Network. However, we identified a number of new targets and mutations that were not previously identified. We identified using gene expression subgroups of cases that didn’t necessarily have just one genetic change, be it a translocation or mutation, but different groups of mutations that acted together or alternately to define the same biology. We also identified targets of overexpression that are potentially very interesting and targets for therapy.
So a couple of examples of these themes were we could show that cases with NPM, or nucleophosmin, mutations, which is well recognised in AML, could be further sub-classified by the presence of mutations such as in genes like FLT3 and cohesion. Those subgroups were associated with very different treatment outcomes.
Another finding was that there were a group of cases that had a very high level of expression of a gene called MN1 which had previously been seen in a subset of leukaemias but here we found them in a different group of leukaemias with very high expression. This was associated with high levels of expression of the homeobox cluster of genes which has been seen in myeloid leukaemias to be a potential target for therapy. So this suggests, although more work is needed, that this has identified another group of cases that might have a new therapeutic approach.
We confirmed and extended some other previous observations such as genetic changes in RUNX1 and p53 as frequent, recurrent lesions associated with very poor outcome. An interesting theme here, though, was that these were two examples of genes where we showed that the genes are not just targeted by sequence mutations but also structural changes such as deletions and rearrangements which are important but not detected by the usual gene panel sequencing approaches. So, again, it gave further emphasis that this approach of using a comprehensive genomic approach can give us significantly more information than currently widely used diagnostic testing approaches.
There are several clinical implications, some are more immediately translatable, some are more for future work and validation. As mentioned, we’ve identified groups of cases that were previously unclassified or were left in groups of AML that were classified only by morphology which doesn’t really give us much insight into the biology or driving pathogenesis. So now we’re able to carve out some of those patients and say they fall into a distinct group with a distinct outcome. So that gives the diagnostician and the clinician more concrete information about this person’s risk of failing or succeeding with conventional therapy.
As mentioned, there are some very strong pointers from some of these findings that there are patients that might have a completely new vulnerability to therapy that’s not previously been recognised. Again, these are often patients that were treated as just not otherwise classified group of patients. So now this gives additional insight.
The final important thing is the feasibility of this approach. There has been a lot of interest in using whole genome sequencing and transcription sequencing. It’s widely used in other malignancies such as lymphoblastic leukaemia and lymphoma. It’s also very widely used as a research tool but people have been perhaps reluctant to pursue it as a diagnostic test, thinking that it’s too complex or would take too long to get results back. This study shows that that’s not the case, that with the right infrastructure that this is a tractable pipeline that can be performed with high quality and returned within a feasible timeframe for clinical management.
The study is ongoing and so so far in the presentation that will occur at ASH this year we will show outcome associations where we look at conventional approaches to associate one group or one feature or a distinct set of features with overall survival. There’s more work to be done. There is a lot of interest in the field in developing more sophisticated biomarkers using a risk prediction algorithm. That’s work that is ongoing at the moment to find the right combination or combinations of genetic markers that can most precisely identify the exact risk of a patient of succeeding with therapy or failing.
This is part of a larger study that is being spearheaded by Torsten Haferlach and his group at the Munich Leukaemia Laboratory that sequenced over 4,500 cases of patients with hematologic malignancies. Many of these studies are being presented at the ASH annual meeting this year.