Molecular and alternative methods for diagnosis of acute myeloid leukemia with mutated NPM1: flexibility may help

A cute myeloid leukemia (AML) is a molecularly and clinically heterogeneous disease.1 Nucleophosmin (NPM1) gene mutations resulting in cytoplasmic delocalization of nucleophosmin (NPMc+)2 are the most common genetic alteration in AML, being detected in about one-third of cases. Because of its unique molecular, genotypic, immunophenotypic and prognostic features3–6 (Table 1), AML with mutated NPM1 was included as a separate provisional entity in the 2008 World Health Organization (WHO) classification of myeloid neoplasms, under the heading of “AML with recurrent genetic abnormalities”. 7 In the 2001 WHO classification, this category included only AML with t(15;17), t(8;21), inv(16), and MLL rearrangements and in 2008 it was expanded to include AMLs carrying t(6;9), inv(3) or t(3;3), AML (megakaryoblastic) with t(1;22) and two provisional entities, i.e. AML with mutated NPM1 and AML with mutated CEBPA. Thus, in the 2008 WHO classification, the category of “AML with recurrent genetic abnormalities” covers about 60% of AML.

The identification of the specific genetic alteration underlying each of the AML subtypes listed in the category of “AML with recurrent genetic abnormalities” is of critical importance since it helps to assign patients to different prognostic groups, thus influencing therapy. For example, in AML patients under 60 years of age, NPM1 mutations consistently predict favorable prognosis,6 when no concomitant FLT3-ITD mutation is present, whilst in AML patients 70 years old or over NPM1 mutations appear to be the only factor influencing prognosis in multivariate analysis.8

The search for genetic alterations using molecular methods would ideally be the gold standard for diagnosis. Unfortunately, not all centers, especially in developing countries, are equipped for molecular studies and this may hamper the worldwide use of the WHO classification. These problems could potentially be solved by developing simple surrogates for molecular studies. Here, we briefly review the molecular methods and their alternatives that are currently available for diagnosing AML with mutated NPM1.

Detection of NPM1 mutations by molecular techniques

The normal NPM1 gene configuration and the first six NPM1 mutations we identified in AML,2 which lead to common structural changes of the NPM1 protein C-terminus, are shown in Figure 1. Over the past five years, several qualitative and quantitative molecular assays for identifying NPM1 mutations have been developed and tested in a large number of AML patients.

• Qualitative assays for NPM1 mutations: these highly sensitive and specific assays for detecting NPM1 mutations9–15 are best applied to RNA or DNA extracted from fresh bone marrow or peripheral blood leukemic cells13 but plasma16 and paraffin-embedded samples17 are also suitable. More than 50 molecular variants of NPM1 mutations have been identified to date,18 mostly involving exon-12, and occasionally other exons.19 NPM1 mutations occur in about 30% of adult AML4 (50–60% of AML with normal karyotype). NPM1 mutation A (a duplication of TCTG at position 956 to 959 of the reference sequence) accounts for 75–80% of cases.2 Mutation B and D account for about 10% and 5% of cases, respectively; other mutations are very rare. NPM1 mutations are less frequent in childhood (about 8% of pediatric AML)20,21 and have been never found in children under three years of age.21 Pediatric and adult AML with mutated NPM1 appear to differ not only in frequency but also in the type of mutation as, unlike adults, the majority of children carry non-type A mutations.22 Identifying the specific type of mutation by molecular techniques is essential when PCR (Polymerase Chain Reaction)-based quantitative studies are planned (see below).
• Monitoring of minimal residual disease. Since NPM1 mutations are frequent and very stable over the course of disease,23 they are an optimal marker for monitoring minimal residual disease (MDR)24 in approximately 30% of adult AML. Indeed, the clinical value of PCR-based quantitative assessment of NPM1 mutant copies in predicting relapse and prognosis of AML with mutated NPM1 was demonstrated in several studies.25–28 Notably, Schnittger et al.28 reported the best clinical outcome was associated with the greatest reduction in the number of NPM1 mutant copies (<0.01 NPM1/ABL ratio). Monitoring of NPM1 mutant copies every 4–6 months is advisable.29

Table 1.Main features of acute myeloid leukemia with mutated NPM1 (NPMc+ AML).

Detection of NPM1 mutated proteins by Western blot

The Western blot assay uses antibodies that recognize NPM1 mutants but not wild-type NPM1 protein in lysates from AML samples.30 These antibodies identify the specific band (37kDa) of mutated NPM1 protein only in NPM1-mutated AML cases and recognize over 95% of NPM1 mutations.30

Morphology and immunophenotype of NPM1-mutated acute myeloid leukemia

Advanced molecular or biochemical techniques are not always available or easy to apply and, therefore, a “realistic classification” based on morphological appearance was, in the past, suggested as a compromise solution for some AML subtypes.31 Since some typical FAB categories such as M3, M2 with increased eosinophils (M2eo), or M4 with increased eosinophils (M4eo), closely correlated with the presence of t(15;17), t(8;21), and inv(16), respectively, recognition of these morphological features was proposed as a surrogate for cytogenetic studies. The morphology-based approach is, however, limited by imperfect morphological-genetic correlations32 since inv(16) or t(16;16) correlate with M4eo morphology in only some cases, whilst AML with t(8;21) sometimes shows an M1 or M4 morphology. Morphology is also a poor predictor of NPM1 mutations because, although often associated with M4 and M5 morphology,2,33 AML with mutated NPM1 encompasses all other FAB categories except M3, M4eo and M7.2 Interestingly, a similar broad morphological spectrum is also seen in AML with MLL rearrangements.34

Immunophenotype combined with morphology may further increase the ability to identify specific AML genetic entities. Examples include positivity for CD19 and PAX5 in AML with t(8;21),35 low expression of HLA/DR and CD34 in acute promyelocytic leukemia (APL),7 expression of the condroitin sulfate molecule NG2 omolog (encoded by CSGP4)36 in AML with MLL rearrangements, and the consistent CD34 negativity in NPM1-mutated AML,2,4 but again the correlation immunophenotype/genotype is not complete.

Thus, in most AML cases, morphological/immunophenotypic studies alone cannot reliably predict genetic lesions, and other surrogates for molecular investigations need to be found.

Immunohistochemical detection of cytoplasmic nucleophosmin

In the 2008 WHO classification, AML with mutated NPM1 is also indicated with the synonym of NPMc+ AML (NPM cytoplasmic positive AML).7 In fact, one simple, low cost, highly-specific alternative approach to diagnosis is immunohistochemical detection of cytoplasmic nucleophosmin.37,38 This assay is fully predictive of NPM1 mutations since all molecular variants of NPM1 mutations (including those affecting exons other than 12) result in aberrant export of NPM1 mutant from the nucleus to the cytoplasm of leukemic cells.37,38 Immunohistochemistry is usually performed with monoclonal antibodies that recognize wild-type and mutated NPM1 proteins.38 Cytoplasmic nucleophosmin is optimally detected in paraffin sections from B5-fixed/EDTA-decalcified bone marrow trephines2,38,39 (Figure 2). Partial concordance between sub-cellular expression of nucleophosmin and NPM1 gene status was seen in one study40 where samples were fixed in formalin and decalcified in formic acid. It is at present unclear whether the technical problem is due to formalin fixation and/or formic acid decalcification.

Figure 1.Configuration of the normal NPM1 gene. NPM1 mutations in AML occur almost exclusively at exon-12 (asterisk). (A – F) The first 6 discovered mutations.2 Mutation A is the most frequent (75–80% of cases). All NPM1 mutations (about 50 so far identified) result in common changes at the C-terminus end (asterisk) of the NPM1 protein, i.e. changes of tryptophans (288 and 290) and insertion of a new nuclear export signal (NES) motif. These changes cause aberrant cytoplasmic accumulation of NPM1 mutants which is easily detectable by immunohistochemistry (see Figure 2). This figure is provided by the Author (BF) and was originally reported in the 2008 WHO classification. 7

Searching for cytoplasmic nucleophosmin is reminiscent of identifying APL with t(15,17) by means of the PG-M3 (anti-PML) monoclonal antibody.41 In fact, the rationale for both tests is similar. The NPM1 mutation and the PML/RAR-alpha fusion gene both cause ectopic subcellular relocalization of the respective proteins: nucleophosmin is found in cytoplasm (instead of nucleolus) and PML is observed as nuclear microspeckles (instead of nuclear bodies). In a recent study, Rego et al.42 combined morphology with the PML immunofluorescence test (PG-M3 monoclonal antibody) to investigate 102 APL patients from developing countries (Brazil, Mexico and Uruguay). Notably, this approach resulted in a more accurate and rapid diagnosis which led in turn to the immediate start of chemotherapy plus ATRA. Consequently, overall survival markedly increased from the unsatisfactory 50% in historical controls to about 80%. This study provides an example model of the big impact of using a simple test as surrogate for molecular investigations in the therapy of AML.

Figure 2.Detection of cytoplasmic nucleophosmin by immunohistochemistry (bone marrow trephines, paraffin sections). (A) AML with mutated NPM1. Marrow infiltration by myeloid blasts admixed with megakaryocytes (double arrows) and erythroid precursors (single arrow). T indicates a bone trabecula (hematoxylin-eosin; x1,000). (B) Myeloid blasts, megakaryocytes (double arrows) and erythroid precursors show aberrant cytoplasmic expression of nucleophosmin (multilineage involvement) (APAAP technique; x1,000). T indicates a bone trabecula. (C) Myeloid blasts, as well as megakaryocytes (double arrows) and erythroid precursors (single arrow) belonging to the leukemic clone, are CD34-negative (APAAP technique; x1,000). (B and C) Hematoxylin counterstaining.

Immunohistochemical detection of cytoplasmic nucleophosmin may also be useful in developed countries as simple front-line screening for NPM1 mutations.37,38 The expected 30% of adults with aberrant cytoplasmic nucleophosmin could then be referred to more specialized centers for confirmation and identification of mutation type by molecular techniques. Moreover, immunohistochemistry is critical for diagnosis of AML cases presenting with “dry tap” or as myeloid sarcoma.17

As bone marrow trephines are not routinely performed in AML patients in all hematologic centers, the ability to predict NPM1 mutations from cytoplasmic expression of nucleophosmin in smears or cytospins would be particularly useful. In the present issue of Haematologica, Mattsson et al.43 found no significant correlation between sub-cellular expression of nucleophosmin and NPM1 gene status in their immunocytochemical study of smears and cytospins from 60 AMLs (31 NPM1-mutated; 29 NPM1 wild-type). Why cytoplasmic nucleophosmin is detected in fixed paraffin-embedded material but not cytological samples remains unclear. A possible explanation is that preparation and/or fixation of smear and cytospins leads to nucleophosmin diffusion across cell compartments and even out of the cells, thus preventing accurate tracking of the protein. In contrast, fixation and paraffin-embedding may optimally stabilize cell membranes (especially nuclear membrane) thus allowing accurate visualization of sub-cellular distribution of nucleophosmin.

Mattsson et al.’s43 claim that cytoplasmic nucleophosmin in paraffin sections may represent non-specific staining appears unfounded because there is much strong evidence indicating that in NPM1-mutated AML immunohistochemistry on paraffin sections depicts the real status of nucleophosmin sub-cellular distribution. In fact, aberrant cytoplasmic expression of nucleophosmin in B5 fixed/EDTA decalcified bone marrow trephines is fully predictive of NPM1 mutations.37,39 Aberrant nuclear export of nucleophosmin is consistent with the molecular alterations at the C-terminus of all NPM1 mutant proteins (Figure 1) including those generated by NPM1 mutations involving exons other than exon 12.38 Moreover, antibodies that specifically recognize mutant but not wild-type NPM1 protein, consistently label the cytoplasm of NPM1-mutated AML cells at both immunohistochemistry44 and Western blotting.30 Finally, laser confocal microscopy of transfected cells clearly showed that fluorescent-tagged wild-type and mutated NPM1 proteins localized respectively in the nucleolar and cytoplasmic cell compartments. 4,38

Whatever the reason for the failure to detect cytoplasmic nucleophosmin in smears or cytospins, the comprehensive study by Mattsson et al.43 clearly indicates this method cannot be at present recommended as surrogate marker for NPM1 mutations. One promising approach based on the use of intracellular flow cytometry for rapid, specific detection of nucleophosmin in the cytoplasm of leukemic cells has been recently proposed as an alternative to bone marrow trephines in the diagnosis of NPM1-mutated AML.45 Because of its simplicity, this test may emerge in the future as a valuable surrogate to molecular studies for initial screening of AML samples. Another method that can be used in alternative to bone marrow trephines for detection of cytoplasmic nucleophosmin is immunostaining of sections cut from paraffin-embedded pellets of peripheral blood leukemic cells (B. Falini, unpublished observation, 2009).

Surrogates for molecular studies are expected to be particularly useful in older patients where NPM1 mutations appear to play a prognostic role independently of the FLT3 gene status.8 For younger patients, whose favorable prognosis is associated with the NPM1-mutated/FLT3-ITD negative genotype,4,6 we must search for rapid, inexpensive assays which can also allow us to assess the FLT3 gene status, which is still only evaluable by molecular methods.

Conclusions

Because of its distinctive features, AML with mutated NPM1 (NPMc+ AML) was included as a provisional entity in the 2008 WHO classification of myeloid neoplasms. Over the past five years several methods have been developed to diagnose this new entity. NPM1 mutations can be detected by molecular techniques or surrogates such as immunohistochemistry, Western blotting and possibly flow cytometry. These methods are complementary rather than competitive and offer a flexible approach to diagnosis which is essential if the WHO Classification is to be implemented, as intended, worldwide.

Footnotes

• Brunangelo Falini is Professor of Hematology at the University of Perugia, Perugia, Italy. Maria Paola Martelli is a Researcher in Hematology at the University of Perugia, Perugia, Italy. Stefano A Pileri is Professor of Pathology and Director of the Hematopathology Unit at Bologna University, Bologna, Italy. Cristina Mecucci is Associated Professor of Hematology at the University of Perugia, Perugia, Italy.
• Drs. Falini and Mecucci applied for a patent on the clinical use of NPM1 mutants. No other potential conflicts of interests relevant to this article were reported.
• Supported by the Associazione Italiana per la Ricerca sul Cancro (A.I.R.C.) and Fondazione Cassa di Risparmio di Perugia (Grants n. 2007.0099.020 and 2008.020.058). The authors would like to thank Claudia Tibido’ for her excellent secretarial assistance and Dr. Geraldine Anne Boyd for her help in editing this paper. The authors apologize to those whose papers could not be cited owing to space limitations.
• ( Related Original Article on page 670)