Acute myelogenous leukaemia (AML) is the most common form of leukaemia, but as there are so many subgroups of AML, it is a challenge to select the most effective course of therapy. In the 20 January issue of New England Journal of Medicine, Brunangelo Falini et al. report that the intracellular location of the nucleophosmin (NPM) protein can be used to characterize a large subgroup of patients with AML, and to predict response to therapy.

In normal cells, NPM shuttles between the nucleus and cytoplasm, and is most frequently found in the nucleolus. It is thought to have a tumour-suppressor function, and is a molecular chaperone that might regulate the ARF–p53 pathway. However, it also acts oncogenically in some leukaemias and lymphomas, as the NPM gene is a partner in cancer-associated translocations. In this study, the authors investigated the subcellular location of NPM in 591 specimens from patients with AML, with a view to discovering whether this might affect its tumour-suppressor function.

They found that in 208 of the samples (35%), NPM was located in the cytoplasm (NPMc+). This pattern was observed across all of the different leukaemia FAB (French–American–British) subtypes tested, usually associated with CD34 negativity, but was only seen in cells from patients with primary AML. In secondary AML cells, NPM localization was exclusively nuclear. Significantly, 85% of patients with NPMc+ had normal karyotypes, compared with 27% of patients with NPMc−. So, cytoplasmic NPM characterizes a distinct subgroup of patients with AML who have a normal karyotype — a group that was previously difficult to stratify.

Does the location of the NPM protein affect the response to therapy? The association was investigated in 126 patients with normal karyo-types for whom NPM immuno-staining and clinical information were known. A multivariate logistic-regression model showed that a lower white-cell count and cytoplasmic localization of NPM are independent prognostic factors for a complete remission.

The authors next investigated whether NPM was mutated in the cases in which the protein was cytoplasmic. The characteristic fusion proteins were not found, but sequencing revealed mutations in exon 12 in all but one case of NPMc+ disease. Although these mutations were diverse, they all resulted in frameshifts that altered the carboxy-terminal region of the protein. Transient expression of these mutant proteins in NIH-3T3 cells resulted in their cytoplasmic localization, whereas a tagged form of the wild-type protein remained in the nucleolus. These mutations might therefore directly cause the altered distribution.