Around 45% of adults with AML have NC and are included in an intermediate-risk group. However, their 5-year (yr) overall survival (OS) rates vary between 24 and 42%, likely due to the prognostic impact of submicroscopic genetic alterations, e.g., mutations in FLT3, CEBPA, MLL and NPM genes and overexpression of BAALC. We recently showed that ERG, an ETS-Related Gene mapped to 21q22, is often overexpressed in AML with unfavorable complex karyotypes, and in a subset of NC AML (

PNAS
2004
;
101
:
3915
), suggesting that ERG overexpression contributes to an aggressive phenotype in AML. Here we analyzed ERG expression levels by real-time RT-PCR in pretreatment blood from 84 adults with NC AML, aged <60 yrs, treated on CALGB 9621 and characterized for BAALC expression (
Blood
2003
;
102
:
1613
). Patients (pts) were divided into quartiles according to ERG levels and dichotomized into 2 groups: the uppermost quartile of ERG expression values (Q4) and a group comprising 3 lower quartiles (Q1-3), as relapse risk was significantly different for Q4 compared with Q1 (P=.024), Q2 (P=.002) and Q3 (P=.009). The complete remission rates were similar for the 2 groups (76% vs. 83%; P=.532). With a median follow-up of 5.7 yrs, Q4 pts had a worse cumulative incidence of relapse (CIR; P<.001) and OS (P=.011) than Q1-3 pts. For Q4 pts, the estimated 5-yr CIR and OS rates were, respectively, 81% and 19% compared with 33% and 51% for Q1-3 pts. In multivariable models, high ERG expression (Q4) adversely impacted CIR (P<.001), whereas an interaction between ERG and BAALC expression (P=.013) was observed for OS, where Q4 predicted shorter survival only in low BAALC expressers (P=.002; Table 1).

Table 1.

Multivariable analysis for pts divided into Q4 and Q1-3 groups according to ERG expression

EndpointVariableHazard ratio (95% CI)P
CIR ERG expression (Q4 vs. Q1-3) 3.71 (1.88 to 7.31) <.001 
 Present vs. absent MLL PTD 2.70 (1.12 to 6.52) .027 
OS Interaction ERG of and BAALC  .013 
     - Pts with low BAALC expression, ERG expression (Q4 vs. Q1-3) 5.40 (1.87 to 15.64) .002 
     - Pts with high BAALC expression, ERG expression (Q4 vs. Q1-3) 1.04 (0.50 to 2.16) .922 
 Log[WBC] 1.35 (1.07 to 1.70) .012 
EndpointVariableHazard ratio (95% CI)P
CIR ERG expression (Q4 vs. Q1-3) 3.71 (1.88 to 7.31) <.001 
 Present vs. absent MLL PTD 2.70 (1.12 to 6.52) .027 
OS Interaction ERG of and BAALC  .013 
     - Pts with low BAALC expression, ERG expression (Q4 vs. Q1-3) 5.40 (1.87 to 15.64) .002 
     - Pts with high BAALC expression, ERG expression (Q4 vs. Q1-3) 1.04 (0.50 to 2.16) .922 
 Log[WBC] 1.35 (1.07 to 1.70) .012 

When ERG expression was evaluated in the context of pts with known FLT3 internal tandem duplication (ITD) status, including those with the very unfavorable FLT3ITD/-genotype, i.e., lacking the FLT3 wild-type allele, a multivariable analysis showed that higher risk of relapse and death was independently predicted by both high ERG expression values (i.e., Q4) (P=.023 and P=.002, respectively) and FLT3 ITD mutations (P≤ .001 and P=.002, respectively). We also used Affymetrix U133 plus 2.0 GeneChips to identify genes differentially expressed (P≤ .001) between Q4 and Q1-3 pts. Q4 pts displayed a signature characterized by overexpression of 63 genes and 49 expressed-sequenced tags. Fourteen genes, including general (GTF2H2) and lineage-specific (BCL11A, HEMGN) transcription regulators and genes involved in cell proliferation (RAB10, GAS5) and apoptosis (IKIP, DAPK), had at least a two-fold difference in expression levels between the Q4 and Q1-3 groups. In conclusion, we show for the first time that ERG overexpression in NC AML constitutes an adverse prognostic factor and is associated with a distinct gene-expression signature.

Author notes

Corresponding author

Sign in via your Institution