Genomic characterisation and personalised risk stratification of acute myeloid leukaemia

Acute myeloid leukaemia (AML) is an aggressive blood cancer with median survival of approximately 3yrs. We are presenting data from a screen of 111 cancer genes in 1540 patients of three clinical trials conducted by the German-Austrian AML Study Group. Combined with conventional cytogenetics these data provide an unprecedented comprehensive characterisation of the molecular changes underlying AML.

A probabilistic clustering approach shows that there are at least 11 genomic subtypes, each characterised by particular constellations of genetic and cytogenetic changes. Associating genomic and clinical data with outcome shows that approximately 2/3 of the explained differences in survival are related to aggregated genomics, which are build up by many small contributions from individual mutations.

Novel approaches for modelling the different stages and endpoints during treatment show that patient fate can be accurately predicted based on data at diagnosis. In particular, these approaches allow for modelling the impact of haematopoietic stem cell transplants either in first complete remission or after relapse on a per patient basis, providing data to underpin clinical decisions.

We round up our analysis by an assessment of the expected gains of our approach by either broadening the panel of sequencing or extending the number of samples, as suggested by current initiatives in the U.S. and UK.