Few Genes Expression Predicts Outcome in Adult Acute Myeloid Leukemia (AML) with Normal Karyotype.

AML with normal karyotype constitutes a heterogeneous group considered of intermediate cytogenetic risk. Several gene mutations (CEBPAm, FLT3/ITD, MLL/PTD, NPMm) have recently helped to precise the prognostic in this AML subset. At the patient level, target gene expression profiles might be more accurate than gene mutations themselves to correctly assess prognosis. In this retrospective study including 120 AML patients with normal karyotype, we thus evaluated the global contribution of the expression of 7 genes, all reported to be individually associated with patient outcome: BAALC, EVI1, FLT3, HOXA9, TGIF1, TM4SF1, and TRIM16. Gene expression quantification was performed by real-time PCR using TBP as housekeeping gene. FLT3/ITD was present in 38/120 patients (32%), CEBPA mutation in 20/120 (17%) and MLL/PTD in 3/120 (3%). The recent NPM mutation (NPMm) was found in 50/106 screened patients (47%). We first performed an unsupervised hierarchical clustering analysis that delineated 4 subgroups of patients: 1) one group (n=13) with low HOXA9 expression and favorable prognosis, mostly composed of patients with CEBPAm (66%); 2) a second group (n=18) with high EVI1 expression, low rate of recurrent gene mutations, and poor outcome; 3–4) two groups (n=29 and 60) with high frequency of NPM mutations (70 and 56%) and intermediate outcome. Overall survival (OS) was significantly different in these 4 subgroups (p=.03). As no significant expression correlations were found between these 7 genes, we next considered each gene to contribute with an identical weight to patient outcome. Univariate analysis showed that all genes but BAALC were significantly associated with either OS and/or event-free survival (EFS). We built a molecular risk index based on best prediction cut-off analyses for the 6 remaining genes. This 3-class index (favorable risk, n=22; intermediate risk, n=57; high risk, n=41) was highly predictive of complete remission rate (100%, 91%, and 76%; p=.01), OS (6y-OS: 77%, 37%, and 25%; p=.0001) and EFS (6y-EFS: 62%, 29%, and 13%; p<.0001). Age and FAB subtypes did not significantly differ in each subclass. White blood cell count (WBC) was higher in the high-risk class (median WBC: 17, 31 and 55 G/L; p=.13). CEBPA mutations were significantly more frequent in the favorable-risk class (55%, 7%, and 10%; p<.0001). FLT3/ITD and NPMm were significantly more frequent in the intermediate- and high-risk class (FLT3/ITD: 14%, 30%, and 44%, p=.05; NPMm: 18%, 49%, 61%, p=.005). In multivariate analysis considering this molecular risk index, WBC, age, CEBPAm, and FLT3/ITD as covariates, the molecular risk index and age were the only variables significantly associated with OS (p=.0008 and p=.03, respectively) and EFS (p=.0005 and p=.04 respectively). In conclusion, quantification of selected gene expression may be useful and more powerful than gene mutation detection to predict the outcome of patients with AML and normal karyotype.