Mutational Landscape of Relapsed Core-Binding Factor Acute Myeloid Leukemia (CBF-AML)

Background: Acute myeloid leukemias (AML) with rearrangements of core-binding factor (CBF) complex genes (CBF-AML), comprising t(8;21) and inv(16) subgroups, are considered as diseases with favorable outcome. Nevertheless, CBF-AML relapse rates remain high, with ~40% of patients (pts) relapsing after standard intensive chemotherapy.

Aim: To dissect the biology of relapse in CBF-AML, we performed whole exome sequencing (WES) in a large cohort of 101 cases at the time of diagnosis and for 47 cases also at the time of relapse.

Methods: All pts were treated either with standard chemotherapy or with standard chemotherapy and kinase inhibitor dasatinib within clinical trials of the German-Austrian AML Study Group (AMLSG). Using the Nextera Rapid Capture Exome kit (Illumina) we performed WES of paired diagnostic (dx), remission and relapse samples of 47 pts, namely 21 pts with t(8;21) and 26 pts with inv(16). RNAseq was performed in 18 of these pts using the Ribo Zero RNA-sequencing kit (Illumina). To better define genomic signatures related to CBF-AML relapse, we included WES data previously published by our group (Faber et al. Nat Genet 2016). This set comprised dx samples of 8 t(8;21) and 10 inv(16) pts who relapsed as well as a control group of 20 t(8;21) and 16 inv(16) CBF-AML pts, who did not experience relapse.

Results: For the new cohort, WES sequencing of 47 pts was performed with a mean coverage of 127-fold. In t(8;21), we identified a median of 3.5 mutations exclusively present at dx (range: 0-8), 11.6 mutations persistent from dx to relapse (range: 4-19), and 4.0 mutations gained at relapse (range: 2-7). For the inv(16) subgroup a median of 2.0 mutations were dx specific (0-7), 6.0 mutations persisted during tumor evolution (3-26) and 2.5 were gained at relapse (0-9). As previously reported, the spectrum of genes affected by mutations showed little overlap between t(8;21) and inv(16), except for commonly affected ‘signaling’ genes such as KIT, RAS, FLT3 and epigenetic players such as TET2. In total, in t(8;21) we identified 94 relapse-specific mutations or mutations displaying a strong increase in variant allele frequency (VAF) at relapse, and 63 of such relapse-specific alterations in inv(16) pts. In addition to the previously reported RUNX1 and cohesin complex gene mutations showing an increase in VAF at relapse, we found recurrent novel relapse-specific mutations in LAMC3, which occurred exclusively in the t(8;21) subgroup affecting 9% of pts. In inv(16), recurrent mutations in the tumor suppressor gene WT1 occurred in 12% of pts, either acquired at relapse or already present at dx as a minor subclone. Remarkably, mutations in relapsed t(8;21) pts often affected genes involved in PI3K-AKT and in cell cycle regulation pathways. In the inv(16) relapse group, in addition to dysregulation of the MAPK signaling pathway, we found several non-recurrent mutations in genes involved in ribosomal RNA metabolism, like in PRNAD1.

Conclusion: Our WES sequencing results already provide first insights into the molecular composition and mechanisms underlying relapse in CBF-AML which often affect pathways linked to proliferation, such as PI3K-AKT and MAPK signaling. While we are currently validating additional hits, updated results will be provided at the ASH meeting, which will also address combinatorial mutation patterns underlying chemotherapy resistance in t(8;21) and inv(16) positive AML.