| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Published online before print April 10, 2008
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Commentary |
From the Department of Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
Abstract
This Commentary highlights two articles that focus on molecular techniques to identify mutations in nucleophosmin (NPM1), which is the most frequently mutated gene in cytogenetically normal acute myeloid leukemia (CN-AML)
Contemporary laboratory evaluation and classification of acute myeloid leukemia (AML) is multimodal, using a diverse array of tools such as morphology, cytochemistry, flow cytometric immunophenotyping, and genetic testing. Genetic testing incorporates both conventional cytogenetic karyotyping and molecular analysis, with the latter including fluorescence in situ hybridization- and polymerase chain reaction (PCR)-based approaches. Of all of these parameters, genetic features are the most pertinent with regard to the biology, classification, prognosis and, ultimately, therapy of AML.1, 2, 3, 4 The importance of genetic changes was, for the first time, codified in the 2001 World Health Organization classification of AML and will assume an even greater role in the update, slated for publication later this year.
In the 2001 classification, four recurrent cytogenetic abnormalities have been selected to define specific diagnostic entities. Those AMLs that harbor gene fusions resulting from t(8;21), t(15;17), or inv(16) have a favorable prognosis, whereas tumors with 11q23 translocations tend to be more aggressive, with a poor or occasionally intermediate prognosis, depending on the chromosomal partner. Not only does the karyotype define prognosis, but it determines specific treatment options as well.5 Thus, the decision to use bone marrow transplant as a first line treatment following induction of remission is guided, in large part, by the karyotype of the AML, and the use of targeted therapies, such as all-trans retinoic acid, is essentially restricted to those AMLs with t(15;17). Together, the aforementioned karyotypically detectable lesions are seen in approximately 30% of adult AMLs.2 Approximately 30% of AMLs may contain other extremely diverse, but far less frequent, non-random chromosomal abnormalities.2 Indeed, over 150 recurrent, and sometimes prognostically relevant, lesions have been described in AML, yet most have not currently been formally incorporated into classification schemes.6 However, three additional genetically defined AML categories to be included in the new (2008) World Health Organization classification are AMLs with t(6;9), inv(3) and t(1;22).
Although these numerous recurrent karyotypic aberrations are both biologically and clinically significant, approximately 40% of AMLs in adults (and 
25% of pediatric cases) do not harbor genetic alterations that can be detected by conventional cytogenetics.2
Variability in the frequency of these so-called cytogenetically normal (CN) AMLs as reported in the literature is likely due to a number of factors. These include age (more CN AMLs are evident with increasing age), methodology (direct preparation versus 24- or 48-hour culture, with the latter sometimes providing a higher yield of clonal abnormalities), and specimen source (in 5% of cases, bone marrow may be superior to peripheral blood in detecting clonal abnormalities). A fourth factor to consider is biological, in that a minor proportion of cases harboring one of the four common translocations/genetic fusions alluded to above may be karyotypically cryptic. Either reverse transcription (RT)-PCR or fluorescence in situ hybridization can be used to uncover these infrequent events, and given the clinical importance of proper diagnosis, they should not be misclassified as CN-AML.
In large clinical outcome analyses, CN-AMLs, collectively the most common group of AMLs, are assigned to an intermediate prognosis category.2 However, CN-AMLs are clearly not homogeneous, and numerous insights into the diversity present in this group have been unraveled recently.7 Indeed, distinct subsets of CN-AMLs have been recognized to possess genetic or epigenetic alterations in specific genes, as well as (sometimes related) expression profiles, dictating their biological and clinical characteristics. While the entire spectrum of genetic alterations has certainly not been defined, many recurrent genetic abnormalities have been identified and have been demonstrated to be important in both tumorigenesis and the clinical outcome of CN-AMLs. For example, internal tandem duplication (ITD) within FLT3, which leads to ligand-independent activation of this receptor tyrosine kinase, has a significantly adverse effect on clinical outcome.8 Similarly, partial tandem duplications of MLL (also disrupted in 11q23 translocations, noted above) are associated with a poor prognosis. Conversely, mutations in CEBPA correlate with favorable prognosis. Yet other biologically and clinically relevant genetic lesions are being uncovered, including mutations of RUNX1, KIT, N-RAS, and K-RAS, as well as overexpression of BAALC, ERG, MN1, and EVI1.
Although enriched in the group of CN-AMLs, the above lesions are not found exclusively in these cases, and a number of genetic aberrations have been noted to occur simultaneously, with intriguing biological and clinical associations. In brief, most of the aforementioned translocations disrupt transcription factors (such as RUNX1 and RARA in the t(8;21) and t(15;17), respectively), in a dominant-negative fashion, and lead to a block in normal myeloid differentiation (so-called class 2 mutations).9 Experimentally, and also likely clinically, such defects are necessary but not sufficient for the AML phenotype. Only when combined with an additional mutation, typically one having a positive effect on proliferation (a so-called class 1 mutation), does the AML manifest, at least experimentally. In patients, the combination of the two types of mutations typically has prognostically adverse connotations. Thus, prognosis for patients with t(8;21) and an associated KIT mutation, or with t(15;17) and a FLT3-ITD mutation, is typically worse than for patients with the translocations alone.10, 11
To date, the most frequently identified mutated gene in CN-AML is nucleophosmin (NPM1).12
Mutations in this gene, which are typically small insertions (usually of 4 bp, sometimes up to 11 bp) in the coding region of the terminal exon (exon 12), occur in 50 to 60% of CN-AMLs, equivalent to 20 to 25% of all AMLs.12
The nucleophosmin gene product is an 37-kDa protein that actively shuttles between the nucleolus, nucleoplasm, and cytoplasm; however, as its name suggests, it is predominantly found in the nucleolus. The nucleophosmin protein was initially suspected to be involved in promoting cell growth, in part through its mediation of ribosomal biogenesis. Consistent with this hypothesis, NPM1 is overexpressed in a variety of human tumors. Studies have demonstrated that NPM1 might also promote growth inhibition by its functional interactions with the tumor suppressors p14ARF and p53.13
Mutations in NPM1 found in CN-AML alter tryptophan residues required for proper nucleolar localization and create a putative nuclear export signal at the C terminus of the protein. Consequently, the mutant nucleophosmin protein is predominantly localized to the cytoplasm and, through dimerization, causes the mislocalization of the wild-type protein as well.14
This leads to the mislocalization and destabilization of p14ARF and to the inhibition of p53 activity.13
Thus cytoplasmic nucleophosmin (NPMc) may be oncogenic, a hypothesis that is supported by tissue culture studies.13
In addition to this interesting pathobiology, the protein mislocalization may be exploited diagnostically since cytoplasmic nucleophosmin can be detected immunohistochemically, and some reports suggest an essentially 100% correlation between immunohistochemical (NPMc) and molecular (NPM1 mutations) findings. Despite this concordance, the terms NPMc and NPM1 mutation signify fundamentally distinct phenomena—with distinct methods of detection—and should probably not, therefore, be used interchangeably.
Given the high frequency of NPM1 mutations in CN-AMLs, multiple studies assessing the relationship between these mutations and survival have been performed. One study showed a weak, but statistically significant increase in disease-free and overall survival in patients with an AML that possessed an NPM1 mutation.15 This benefit, however, is affected by the FLT3 status. FLT3 ITDs are enriched in AMLs with NPM1 mutations, in that they are seen twice as often in this group as compared with AMLs with wild-type NPM1.16 Numerous studies have demonstrated that patients who lack an NPM1 mutation and possess an FLT3-ITD have a much worse prognosis, as gauged by both disease-free survival and overall survival, than do patients who possess a NPM1 mutation without a concomitant FLT3-ITD.15, 17, 18 Overall survival of these latter patients approaches that of patients with AMLs that harbor karyotypes correlated with a favorable prognosis, such as t(8;21), t(15;17), or inv(16) and for whom bone marrow transplant may not have survival benefit. In fact, CN-AML patients with mutant NPM1 and wild-type FLT3 may not benefit from bone marrow transplant. It has been proposed that combining the status of these two "dueling" mutations allows for stratification into three prognostic groups.18, 19 Accordingly, patients may be assigned to good (FLT3-ITD–/NPM1+), intermediate (FLT3-ITD–/NPM1– or FLT3-ITD+/NPM1+), and poor (FLT3-ITD+/NPM1–) categories.
Detection of a mutant NPM1 allele may also be of clinical benefit in following the treatment course of patients with CN-AML. Longitudinal studies have demonstrated that NPM1 mutations are stable throughout the course of the disease, presumably because they often occur as an early or initiating event in the genesis of the AML and are required for tumor maintenance.15, 20 Thus, an NPM1 mutation, if present, represents a marker that can be used to monitor for minimal residual disease (MRD). It has further been demonstrated that quantitative assessment of these mutations correlates well with responsiveness to initial induction chemotherapy.21 Monitoring patients by quantitative assessment of NPM1 mutations can predict overt clinical relapse. Accordingly, these data suggest that NPM1 mutation detection may have clinical and treatment ramifications not only at initial diagnosis but also subsequently for monitoring remission and predicting relapse.
Nascent molecular discoveries and their demonstrated associated clinical relevance are characteristically and predictably followed by a flurry of publications detailing methodologies that might be used in a molecular diagnostic laboratory. These laboratories are then charged with determining the most appropriate assay to adopt. Such is the case for NPM1, and in this regard, two papers in this issue of The Journal of Molecular Diagnostics are of note. The study by Szankasi and colleagues22 details a non-quantitative, genomic DNA-based PCR assay for detecting NPM1 mutations. This assay can be used on paraffin-embedded tissue and can detect mutations when the leukemic cells represent 5% of the population. Since the assay is performed using genomic DNA as a template, intronic primers are used to avoid the amplification of the known pseudogenes. Moreover, since small-length nucleotide insertions distinguish wild-type and mutant alleles, a polymerase with editing capabilities is used. By contrast, the methodology described by Ottone et al23 begins with an RNA template, employs an RT-PCR protocol with and without semi-nesting, and uses allele-specific oligonucleotide (ASO) primers. The assay only detects the most common NPM1 A mutation, which represents 75 to 80% of mutated cases and consists of a tetranucleotide TCTG insertion in exon 12. Although the latter protocol does not detect the full range of NPM1 mutations, it can detect mutant clones that represent as little as 0.001% of the population. The greater lability of the RNA template in this assay is a potential shortcoming; however, since the molecular assays to detect translocations in AML require RNA, this ought not to be a major issue for molecular diagnostic laboratories.
Although both articles in this issue present viable methods for detection of NPM1 mutations, a comparison of the two methodologies highlights recurring issues surrounding both mutational testing in general and NPM1 in particular. The genomic DNA-based method detects all known NPM1 mutations, yet the diagnostic utility of the test is limited to cases in which leukemic cell proportion represents at least 5% of the assayed population. While this relatively low analytic sensitivity is not a limitation at diagnosis, since by definition there will be >20% blasts, it diminishes the utility of the assay for MRD detection. The RNA-ASO-based method has superior analytic sensitivity. However, this advantage comes at a price, in that diagnostic sensitivity is restricted to detection of one specific mutation and bypasses other NPM1 mutations that are seen in 20 to 25% of positive cases. The trade-off evident in these two different methodologies is thus quite striking and is practically inherent in the detection of multiple mutant alleles: a gain in the range of mutation detection is associated with a sacrifice in the sensitivity of detecting a specific allele. Such factors are clearly germane to deciding which assay to institute in a molecular diagnostic laboratory, which in turn is dictated by the specific clinical questions that are being asked. If the primary goal of NPM1 testing is to stratify patients at diagnosis, a qualitative PCR of NPM1 exon 12 from genomic DNA followed by separation of products by capillary electrophoresis (similar to the assay described by Szankasi et al22
) may be instituted. Alternatively, if the major goal of testing is MRD detection in a subset of NPM1-mutated AMLs, an ASO-PCR or RT-RCR assay (similar to that described by Ottone et al23
), is preferred. If both allele detection at diagnosis and MRD detection are desirable, a combination of methods is required. In this scenario, both assays could be performed at diagnosis to classify patients, and MRD testing would only be performed if the allele-specific assay were positive. Alternatively, individual patient/mutation-specific ASO assays could be developed, although this may be prohibitively expensive. The methods described in the reports in this issue of the JMD, and highlighted in this commentary, are among numerous other assays that have been reported for NPM1 mutation detection. Table 1
outlines some of the key features of various different assays, which may help determine which assay is most applicable to a particular laboratory. Ultimately, the decision as to which assay to use is likely predicated on a number of factors, including i) what is the clinical question and ii) what material is available.
|
As noted, the clinical impact of NPM1 mutations is affected by the mutational status of the FLT3 gene. Indeed, NPM1 mutation analysis is of little or no value in the absence of FLT3 testing and the two should always be performed together. Thus, clinical laboratories establishing an NPM1 assay may wish to attempt to multiplex this assay with an FLT3-ITD assay. Detection by capillary electrophoreses of FLT3-ITD mutations and NPM1 mutations in a single assay has been reported.29 We have previously used a multiplex approach, with gel-based detection, to identify recurrent translocations in acute leukemia.30 However, multiplex mutational testing may be particularly well suited to the use of a Luminex-based systems using X-MAP technology, as has been adopted at our institution to identify these recurrent translocations. Theoretically, if PCR and hybridization conditions were similar to those used for detection of recurrent translocations, identification of both recurrent karyotypic abnormalities and these intragenic mutations could be distilled into a single (RNA-based) assay. Additional clinically relevant mutations, such as those seen in CEBPA, KIT, and others, may also be included to increase significantly and comprehensively the prognostic yield. Indeed, using this bead-based technology, the number of discrete mutations that can be detected in a single assay is vast. Thus, the ultimate goal of detecting all clinically relevant mutations for AML—the "mutationome"—in a single reaction is theoretically feasible.
Although the two articles in this issue of JMD, as well as multiple other studies, focusing on NPM1 have used molecular techniques to identify mutations, the cytoplasmic mislocalization of the mutant protein allows for the use of immunohistochemistry to determine NPM1 status.12 Using immunohistochemistry to detect underlying mutations or dysregulated gene expression is certainly not without precedent in hematopathology, with the detection of BCL2 in follicular lymphoma (in contrast to follicular hyperplasia), CyclinD1 in most mantle cell lymphoma, and ALK in a subset of anaplastic large cell lymphoma being central to diagnosis of these lymphomas. Accordingly, in laboratories that lack molecular diagnostic capabilities but can perform adequate immunohistochemistry, this method may be a viable alternative to direct analysis of the NPM1 gene. However, interpretation of NPM1 localization may be difficult, especially in blasts that lack abundant cytoplasm. Moreover, NPM1 immunohistochemistry is unlikely to be useful for MRD detection, and NPM1 localization is only relevant with known FLT3 mutational status, which is only obtained through molecular characterization. Thus, it appears currently appropriate to recommend nucleic acid-based assays rather than immunohistochemical approaches for the detection of NPM1 mutations.
Despite NPM1 being only one of many genes that are frequently mutated in AML, the high occurrence of mutation in CN-AML, the correlation with prognosis, the temporal stability of the mutation, and the possible impact the mutation may have on treatment options strongly argue that testing for NPM1 mutations should be routinely performed in molecular laboratories. However, the existence of multiple NPM1 mutant alleles complicates testing strategies, which currently ensures that a single testing methodology is not ideal for all clinical situations. As such, to institute the appropriate test or tests, laboratory directors must not only be familiar with the benefits and limitations of the various approaches (the duality of methodologies), but more importantly, must also be cognizant of specific and contextual (dueling mutations) clinical questions that are being asked, and what the clinical consequences of the answers are going to be.
Footnotes
Address reprint requests to Adam Bagg, MD, Director, Hematology, Department of Pathology and Laboratory Medicine, University of Pennsylvania, 7.103 Founders Pavilion, 3400 Spruce Street, Philadelphia, PA 19104-4283. E-mail: adambagg{at}mail.med.upenn.edu
See related articles on pages 212 and 236
Supported, in part, by a SCOR grant from the Leukemia and Lymphoma Society of America (to A.B.).
Accepted for publication February 29, 2008.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
