Acute myeloid leukaemia (AML) carrying a mutated NPM1 gene and aberrant dislocation of nucleophosmin (NPM1) into the cytoplasm1 accounts for about one-third of adult AML and shows distinctive biological and clinical features.2, 3, 4 AML with normal karyotype and mutated NPM1, in the absence of FLT3-ITD, exhibits a favourable prognosis.2 Moreover, NPM1 mutations may serve as a marker for monitoring minimal residual disease.2 Thus, the search for NPM1 mutations has now become part of the initial diagnostic work-up in patients with AML.
Approximately 40 different molecular variants of NPM1 mutations,2 the most frequent genetic alteration in AML, have been identified so far (Supplementary Table 1). Mutation A is the most common type (75–80% cases)1 and mutations B and D account for about 10 and 5% of NPM1-mutated AML, respectively; other mutations are very rare.2 Although several sensitive molecular techniques detect NPM1 mutations in DNA or RNA,2 a demand has arisen for simpler, less expensive diagnostic assays, which are suitable for centres that are not equipped for molecular screening. Availability of such tests would contribute to expand the use of a genetic-based WHO classification of AML.
Immunohistochemical detection of cytoplasmic NPM is a very simple, rapid, cheap, sensitive, specific assay that can be used as a surrogate for NPM1 mutational analysis.5, 6 It is applicable to paraffin sections from bone marrow biopsies5 or extramedullary tissues7 but cannot be used on cytological samples (smears or cytospins).1, 5 This is a potential limitation as not all haematological centres perform bone marrow biopsy as a frontline diagnostic procedure in AML patients. The present study assessed the value of western blot in identifying NPM1 mutants in cytological AML samples, using rabbit polyclonal antibodies that react with mutated, but not wild-type, NPM1 proteins.8, 9
We applied western blot analysis to leukaemic samples from 213 AML patients at first diagnosis (Table 1). A total of 135 consecutive patients were analyzed prospectively using freshly collected cells (see Supplementary Materials), whereas 78 patients were studied retrospectively using liquid nitrogen snap-frozen dry pellets of bone marrow or peripheral blood mononuclear cells that had been stored for the past 3 years. The inclusion criterion was the availability of good quality protein extracts, as defined by the absence of protein degradation when membranes were probed with an anti-NPM wild-type specific antibody (anti-NPM, Clone FC-61991; Invitrogen, Carlsbad, CA, USA). Western blot analysis was performed according to the standard procedure on whole-cell lysate from bone marrow or peripheral blood samples (Supplementary Materials), using two rabbit polyclonal antibodies (Sil-A and Sil-C)8, 9 raised against synthetic peptides corresponding to the C-terminal portion of the NPM1 mutant A (Figure 1a and Supplementary Materials). The specific wild-type NPM monoclonal antibody was used as the positive control. Results of the western blot were compared blindly with the results of immunohistochemical analysis of NPM1 subcellular expression, which was available in all patients. The presence of mutated NPM1 was confirmed by molecular analysis in the 42 cases investigated.
In 106/213 AML patients, western blot analysis identified a specific band of 37 kDa molecular weight corresponding to a mutated NPM1 protein (Figure 1b). Immunohistochemistry confirmed the results of western blot in all cases, showing aberrant cytoplasmic expression of NPM1, which is fully predictive of NPM1 mutations5, 6 (Table 1). Although band intensity was strong in almost all cases bands were less intense in one patient with low bone marrow leukaemic infiltration (about 25%) and in some cases with FAB-M5b morphology. The latter finding is in keeping with our previous immunohistochemical observation that NPM1 is downregulated during maturation from the leukaemic monoblast to pro-monocytes.1 No band was detected in 107/213 AML samples. On immunohistochemical analysis, all these cases showed nucleus-restricted expression of NPM1, which is fully predictive of NPM1 gene in germ line configuration (Table 1).
In all 213 cases, western blot analysis was performed on Ficoll-isolated mononuclear cells (Figure 1b). To determine whether the procedure could be simplified, we performed western blot analysis of lysates of cells recovered from 1–2 drops of bone marrow or peripheral blood upon red cell lysis, in 67/213 cases. In all 67 cases, western blot predicted the NPM1 gene mutational status (29 positive and 38 negative) with similar efficiency as in isolated cells (Figure 1c). Results of all the studies were available within 24 h.
As NPM1 mutations generate similar new sequences at the NPM1 C-terminus (Supplementary Table 1), we determined whether, in western blot analysis, Sil-A and Sil-C, rabbit polyclonal antibodies raised against peptide sequences of the most common NPM1 mutant A (286-DLCLAVEEVSLRK), identified leukaemic mutants other than the type-A mutant. Indeed, a strong positive signal identified NPM1 mutation B (Figure 2a, pt. 163), which generates a mutant with a slightly different C-terminal protein sequence (286-DLCMAVEEVSLRK) (Table 2). Thus, as our western blot-based approach identified mutations A and B, it may be expected to identify rare mutants carrying identical protein sequences to mutant A (mutants D, Om, 4, 7, G† and H†) or to mutant B (mutant J†) (Table 2, Supplementary Table 1). Indeed, mutation D was identified in one patient (Figure 2a, pt. 76; Table 2).
Although these mutations already account for over 95% of NPM1 mutations, we investigated whether western blot identified the other NPM1 mutant sequences (Supplementary Table 1) and analyzed cellular lysates of NIH-3T3 or Phoenix cell lines that had been transiently transfected with NPM1 gene constructs encoding NPM1 mutants E (286-DLWQSLAQVSLRK), G (286-DLWQCFAQVSLRK), L (286-DLSRAVEEVSLRK) and P (286-DLCTFLEEVSLRK) as green fluorescent protein-fusion proteins (Figure 2b), using cells transfected with NPM1 mutant A as control. Sil-A and Sil-C antibodies identified all these mutants, although the western blot signal was less intense for mutations E and G (Figure 2b), possibly because the acquired C-terminal sequences were markedly different from NPM1 mutant A. As expected, western blot did not identify the very rare exon-11 NPM1 mutation10, 11 (NPM_mutVi2; Figure 2b), as it generates a truncated mutant form of NPM1 lacking the epitope sequence recognized by Sil-A and Sil-C antibodies. Thus, with the exception of the NPM1 exon-11 mutant, western blot may be presumed to identify all NPM1 mutations.
In conclusion, our study provides evidence that western blot with specific antibodies is a highly flexible, simple and rapid assay for identifying NPM1 mutants in AML samples except for exon-11 NPM1 mutant, which appears to be of little clinical relevance as only two AML patients with this mutation have been reported worldwide.10, 11 Although objections could be raised against the use of polyclonal antibodies for diagnostic purposes, the results of our western blot analysis were highly reproducible with batches of reagents derived from different rabbits that had been immunized with the peptide corresponding to mutation A. Generation of monoclonal antibodies against NPM1 mutants may further facilitate large-scale production and use.
We expect that adding western blotting to the armamentarium of current procedures for detecting NPM1 mutations (mutational analysis and immunohistochemistry) will encourage widespread use of a genetic-based WHO classification of AML, which is an important step, as the detection of NPM1 and FLT3 mutation status is becoming more and more crucial in therapeutic decisions.12 These techniques should be considered complementary rather than competitive as they offer a flexible approach to diagnosis and provide each centre with the option of whatever single or combined approach most suitable.
Using one of the available molecular techniques13 to search for NPM1 mutations in all AML patients at first presentation is probably the most informative approach and is recommended, when possible. Disadvantages include sequencing of all AML cases, costs and need for sophisticated equipment and expert laboratory staff. In haematological centres that are not equipped for advanced molecular analyses, western blot and/or immunohistochemistry could, because of their simplicity and low costs, serve as front-line large-scale screening assays to identify AML patients carrying NPM1 mutations. Choice of either technique will depend on available material, considering that immunohistochemistry can be performed only on paraffin-embedded sections,1 whereas western blot is more suitable for analysis of cytological samples. Cases that result positive on western blot analysis, or express aberrant cytoplasmic NPM at immunohistochemistry, could subsequently be referred to more specialized laboratories to identify the type of NPM1 mutation in order to design primers for monitoring minimal residual disease. Finally, anti-NPM mutant antibodies might be useful for research purposes, for example, to isolate NPM1 mutant-specific protein complexes for proteomic analysis.
This work was supported by Associazione Italiana per la Ricerca sul Cancro (A.I.R.C.) and Fondazione Monte dei Paschi di Siena. We thank Roberta Pacini and Manola Carini for performing immunohistochemical staining. We also thank Dr Geraldine Boyd, for editing the manuscript, and Mrs Claudia Tibidò, for secretarial assistance. Brunangelo Falini applied for a patent on the clinical use of NPM1 mutants.