Acute myeloid leukemia (AML) is characterized by an impaired differentiation and abnormal growth of myeloid-progenitor cells, resulting in an accumulation of malignant blast cells in the bone marrow. It represents 1% of all malignancies with overall incidence of 4.5 per 100 000 adults. About 10% of AML patients carry a chromosomal translocation t(8;21) that leads to the production of an AML1/ETO fusion protein. We have previously shown that mice lacking the transcription factor Gfi1 (Growth factor independent 1) show an accumulation of aberrant myelo-monocytic cells reminiscent of myeloproliferative disorders in patients and that Gfi1 is required to maintain the self-renewal and differentiation capability of hematopoietic stem cells. We show here that a variant allele of the gene encoding the transcriptional repressor GFI1, (GFI136N), is associated with AML in three independent Caucasians patient cohorts with an odds-ratio of 2.89 (p=2 x 10–5). The GFI136N variant occurred in 469 AML patients with an allele frequency of 0.064 compared to 0.023 in 503 healthy controls. Patients homozygous for the common form of GFI1 (GFI136S) and patients either heterozygous or homozygous for GFI136N did not differ with regard to the occurrence of AML with specific FAB subtypes, cytogenetic findings or Nucleophosmin 1 mutations. The variant GFI136N did not alter disease progression of AML overall, but correlated with a worse prognosis for patients carrying the t(8;21) chromosomal translocation. Here, the 5 year survival rate was 64% for GFI136S homozygous patients compared to 40% for patients carrying at least one allele of GFI136N. The sub-nuclear localization of GFI136S or GFI136N was different when transiently expressed in 3T3 cells. GFI136S was located mainly at the nuclear border, whereas GFI136S was distributed in a dotted pattern in the nucleus. In addition, we found that GFI136S, but not the variant GFI136N protein co-localized with AML1/ETO in the nucleus. Both GFI136S and GFI136N were able to repress transcription in established reporter gene assays. However, whereas AML/ETO was able to inhibit this activity of GFI136S, the variant GFI136N form was refractory to this inhibition by AML1/ETO, suggesting that the AML1/ETO-GFI1 interaction may be one of the factors responsible for a favorable outcome of t(8;21) positive AML. Our study indicates that the GFI136N variant allele may serve as a novel genetic marker to measure genetic predisposition for the development of AML and provides a first indication that this specific genetic variation at the Gfi1 locus affects the clinical outcome of t(8;21) positive AML patients.

Author notes

Disclosure: No relevant conflicts of interest to declare.