Introduction. The core binding factor rearrangements, inv(16)(p13;1q22), t(16;16)(p13;q22) and t(8;21)(q22;q22), are among the most frequent cytogenetic abnormalities found in children with acute myeloid leukemia (AML). They are considered mutations associated with good responsiveness to chemotherapy and favorable patient outcome. Despite this, in the AIEOP LAM2002 trial, we found that t(8;21) RUNX1-RUNX1T1- rearranged patients had an unexpectedly high cumulative incidence of relapse (30%), which was comparable to that of patients considered at higher risk of treatment failure in light of recurrent molecular/cytogenetic alterations or levels of minimal residual disease at the end of induction therapy. Thus, we aimed at re-evaluating at both genetic and epigenetic level samples of t(8;21) AML cases collected at diagnosis addressing the hypothesis whether the methylation process may influence response to treatment and contribute to further stratify patients carrying the same genetic lesion.

Methods and results. We performed high throughput profiling of DNA methylation (Reduced Representation Bisulfite Sequencing), gene (HTA 2.0 Affymetrix) and protein expression (Reverse phase protein array) on samples collected at diagnosis of 34 t(8;21)(q22;q22)/RUNX1-RUNX1T1 children enrolled in the AIEOP-AML 2002/01 trial who either did or did not experience leukemia recurrence. We found that the DNA methylation profile of t(8;21)-rearranged patients who relapsed was peculiarly different from that of patients maintaining a state of complete remission. We identified a new epigenetic signature of 337 differentially methylated regions(DMRs) separating relapsed (n=6) from non-relapsed patients (n=6). We investigated DMRs association to preferential Transcription Factor Binding Sites (ENCODE) and found that the CCCTC-binding factor (CTCF) is the most represented, it demarcates genomic regions called Topologically Associating Domains, known to be sensitive to CpG methylation with main consequences over gene transcription. We identified 652 differentially methylated CTCF binding sites between patients who either did or did not experience relapse, and looked at gene expression correlation matrixes (transcripts distant within 500 kb from differentially methylated CTCF were considered as a module), revealing that patients who relapsed lost gene expression correlation with respect to non-relapsed children. These results support the role of CTCF in modulating gene expression in relapsed patients. We then performed gene expression profile (GEP) to investigate the expression levels of DMRs neighbor gene transcripts. We found 77 DMRs-controlling genes being targets of RUNX1-RUNX1T1 chimera, and clustering analysis using their expression showed a discrete ability in identifying t(8;21) rearranged patients with respect to their risk of relapse. Through GEP analysis, we identified 18 genes differentially expressed between the two groups (fold change >|2|; unpaired t-test, p<0.05, 87 genes fold change >|1.5|), and Ras Homolog Family Member (RHOB) being the most overexpressed in relapsed patients. We integrated DNA methylation and gene expression platforms with STRING database defining an active network of pathways of cell adhesion and cell-cell signaling, where RHOB emerged as the key player of this network, validated in patient's samples by RPPA. We hypothesized that RHOB is the candidate gene promoting relapse. We moved to in vitro validation and modeled the t(8;21)-rearranged cells to grow in suspension or in adhesion, this latter experimental condition mimicking the bone marrow niche. We showed that RHOB overexpression enhanced cell adhesion properties, by strongly enhancing stress fibers formation, and protected blasts from the cytotoxic action of etoposide. On the contrary, RHOB silencing abolished these features, thus reverting the cell chemotherapy resistance by interrupting RHOB-phospho-COFILIN axis.

Conclusion. This study identifies a novel role of RHOB in influencing the risk of leukemia relapse, indicating that the RHOB-mediated interaction of leukemia cells with the BM niche may support leukemia growth and disease progression. To the best of our knowledge, this is the first study that establishes a dynamic interplay of epigenetic heterogeneity within the same genetic in pediatric AML.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.