Abstract

Although CEBPA mutations (CEBPA+) have been reported to predict favorable outcome in CN-AML, their prognostic value has not been evaluated in the context of such established prognostic molecular markers in CN-AML as the combination of FLT3-ITD and NPM1 mutational status and BAALC and ERG expression. 169 adults aged <60 years (yrs) with untreated, de novo CN-AML, enrolled on CALGB protocols 9621 or 19808 that included autologous stem cell transplantation for consolidation, were analyzed for CEBPA+ by DNA PCR amplification/direct sequencing. Testing for BAALC and ERG expression, FLT3-ITD, FLT3-TKD, MLL-PTD and NPM1 mutations (NPM1+) was performed centrally in pretreatment marrow or blood samples. Unexpectedly, CEBPA+ patients (pts) were more likely to have FLT3-ITD/NPM1 high-risk molecular features [ie, FLT3-ITD+ and/or NPM1 wild-type (NPM1wt)] than low-risk molecular features (FLT3-ITD-/NPM1+; 26 v 3 pts, respectively; P=.001). Thus, we focused subsequent analyses on FLT3-ITD/NPM1 high-risk pts (n=109) that included 90% of the CEBPA+ pts. In this group, a microarray gene-expression signature of 2,342 probes, 59% of which were downregulated in CEBPA+ pts, separated CEBPA+ and CEBPA wild-type (CEBPAwt) pts [false discovery rate (FDR)=.01]. Among the 20 most downregulated probes in CEBPA+ pts, 9 corresponded to Homeobox genes (HOXA3, A5, A9, A10, B2, B3, MEIS1). Also downregulated in CEBPA+ pts were other Homeobox genes (HOXA1, A2, A4, A6, A7, B4, B5, B6), FLT3, RUNX1 and RAS superfamily members, while CEBPA and GATA1 were upregulated. Additionally, a 13-probe microRNA (miRNA) expression signature distinguished CEBPA+ from CEBPAwt pts (FDR=.11). This signature shared features with a previously reported miRNA-signature predictive of clinical outcome in FLT3-ITD/NPM1 high-risk CN-AML (Radmacher et al. JCO 2007;25:359s). Eight miRNA probes for miRNA 181 family members were upregulated in CEBPA+ pts; an association between miRNA 181 family upregulation and good outcome was a major feature of the previously reported outcome miRNA signature. A probe for miRNA 194, whose downregulation was associated with good outcome in the prior outcome signature, was also downregulated in CEBPA+ pts. Consistent with these findings, CEBPA+ status predicted better outcome in the FLT3-ITD/NPM1 high-risk group. CEBPA+ pts had a better event-free survival (EFS) than CEBPAwt pts (P<.0001), with estimated 3-yr EFS rates of 57% and 17%, respectively. In a multivariable analysis, CEBPA+ independently predicted longer EFS (P=.0004; hazard ratio=0.30; 95%CI=0.15–0.58), after adjusting for ERG expression (P=.03). In summary, we report that CN-AML pts with CEBPA+ mostly have FLT3/NPM1 high-risk molecular features, and that the FLT3/NPM1 high-risk group can be subdivided based on the presence or absence of CEBPA+ into 2 subsets characterized by strong gene- and miRNA-expression signatures and different outcome. It is likely that testing for CEBPA+ at diagnosis will improve molecular risk-based classification of de novo CN-AML and aid in risk-adapted treatment stratification. Gene- and miRNA-expression profiling may provide insights into disease biology leading to development of novel therapies.

Author notes

Disclosure: No relevant conflicts of interest to declare.