Analysis of the ETS Family Member Genes ERG, ETS2, ETS1 and FLI1 in Acute Myeloid Leukemia (AML) Patients with Normal Cytogenetics: Expression Levels and Impact On Clinical Outcome. A Study of the AMLSG.

The V-ets erythroblastosis virus E26 (ETS) oncogene family is one of the largest families of transcription factors. ETS transcription factors are characterized by two major functional domains, a transcription domain and an evolutionarily highly conserved DNA-binding domain, also known as ETS domain that mediates binding to purine-rich DNA sequences. Most ETS family proteins are nuclear targets for activation of the Ras-MAP kinase signalling pathway. Therefore, they play a significant role in regulating cellular functions such as cell growth, apoptosis, development and differentiation. ETS transcription factors have been implicated in leukemia by chromosomal rearrangement, and more commonly by gene amplification and/or overexpression. Moreover, overexpression of ERG was shown to be an adverse predictor for clinical outcome in AML with normal cytogenetics (CN). In our recent study on complex karyotype AML, array-CGH (comparative genomic hybridization) analysis identified small genomic amplifications affecting ERG/ETS2 in 21q22 and ETS1/FLI1 in 11q23 in about 10% of the cases. Correlation with global gene expression profiling showed that ERG and ETS2 as well as ETS1 and FLI1 were overexpressed in these cases.

Aims: To evaluate expression levels of ERG, ETS2, ETS1 and FLI1 in a large cohort of younger (16 to 60 years of age) adult CN-AML patients (pts) and their impact on clinical outcome.

The expression of ERG, ETS2, ETS1 and FLI1 was determined by quantitative real-time reverse transcriptase polymerase chain reaction (qPCR) assay in 343 CN-AML pts who were entered on 3 AMLSG treatment protocols (AMLHD93, AML HD98-A, AMLSG 07-04). ERG, ETS2, ETS1, and FLI1 were dichotomized into two major groups according to their expression levels. The upper quartile was chosen as the cut point and the set of patients with gene expression above were defined as Q4 group. Univariable as well as multivariable regression models were used to evaluate the influence of ERG, ETS2, ETS1 and FLI1 on induction success, event-free, relapse-free and overall survival. Multivariable analyses were stratified for AMLSG treatment protocols.

Partial correlation analysis revealed positive correlations of expression levels between ETS2 and ERG (ρ=0.45) being the strongest, followed by ERG and FLI1 (ρ=0.4), as well as ETS1 and FLI1 (ρ=0.31). Correlation of ERG, ETS2, ETS1 and FLI1 with white blood count (WBC) revealed a significant association between high gene expression (Q4) and elevated WBC (ERG, p=0.004; ETS2, p=0.002, FLI1 p<0.001), whereas high expression of ETS1 was associated with a significantly lower WBC (p<0.001). Univariable as well as multivariable analyses on induction success revealed high ETS2 as an unfavourable marker (OR, 0.29, p=0.01). In univariable analysis, there was a significantly inferior relapse-free survival (RFS) and overall survival (OS) for high ERG (p=.01; p=.06, respectively) and high ETS2 (p=.002; p=.03, respectively) that was even more pronounced when both ERG Q4 and ETS2 Q4 (ERG Q4 ∩ ETS2 Q4) (p<0.001; p=.001, respectively) were included as one variable and compared with the rest. In multivariable analysis for the endpoints event-free survival (EFS), RFS and OS, a significant effect was found for RFS for ERG Q4 ∩ ETS2 Q4 (p=.002); the only significant variables that consistently appeared in the model were NPM1^mut, FLT3-ITD^pos and WBC. In subgroup analysis for the genotypes CEBPA^mut, NPM1^mut/FLT3-ITD^neg, and all others (NPM1^mut/FLT3-ITD^pos, NPM1^wt/FLT3-ITD^pos, NPM1^wt/FLT3-ITD^neg) according to the hierarchical model, ERG Q4 was associated with an inferior EFS (p=.04) and OS (p=.03) in the favorable CEBPA^mut genotype and became even more significant for the variable ERG Q4 ∩ ETS2 Q4 (EFS, p=.007, RFS, p=.002; OS, p=.06, respectively). For the NPM1^mut/FLT3-ITD^neg subgroup, again ERG Q4 ∩ ETS2 Q4 was associated with an adverse RFS (p=.04), but not with OS (p=0.07).

In our study on a large cohort of homogenously treated CN-AML patients, ERG and ETS2 expression were highly correlated. Overexpression of both genes had a significant impact on clinical outcome of CN-AML patients. Moreover, adverse effects of high ERG and high ETS2 expression on prognosis were also shown for the genetic AML subgroups CEBPA^mut and NPM1^mut/FLT3-ITD^neg.

No relevant conflicts of interest to declare.