CLINICAL TRIALS AND OBSERVATIONS
Mutations in nucleophosmin (NPM1) in acute myeloid leukemia (AML):
association with other gene abnormalities and previously established gene
expression signatures and their favorable prognostic significance
Roel G. W. Verhaak, Chantal S. Goudswaard, Wim van Putten, Maarten A. Bijl, Mathijs A. Sanders, Wendy Hugens, Andre´ G. Uitterlinden,
Claudia A. J. Erpelinck, Ruud Delwel, Bob Lo¨wenberg, and Peter J. M. Valk
Mutations in nucleophosmin NPM1 are the
most frequent acquired molecular abnor-
malities in acute myeloid leukemia (AML).
We determined the NPM1mutation status in
a clinically and molecularly well-character-
ized patient cohort of 275 patientswith newly
diagnosed AML by denaturing high-perfor-
mance liquid chromatography (dHPLC). We
show that NPM1mutations are significantly
underrepresented in patients younger than
35 years. NPM1 mutations positively corre-
late with AML with high white blood cell
counts, normal karyotypes, and fms-like ty-
rosine kinase-3 gene (FLT3) internal tandem
duplication (ITD) mutations. NPM1 muta-
tions associate inversely with the occur-
rence of CCAAT/enhancer-binding protein-

(CEBPA) andNRASmutations. With respect
to gene expression profiling, we show that
AML cases with an NPM1 mutation cluster
in specific subtypes of AML with previously
established gene expression signatures, are
highly associated with a homeobox gene–
specific expression signature, and can be
predicted with high accuracy. We demon-
strate that patientswith intermediate cytoge-
netic risk AML without FLT3 ITD mutations
but with NPM1 mutations have a signifi-
cantly better overall survival (OS) and event-
free survival (EFS) than those withoutNPM1
mutations. Finally, in multivariable analysis
NPM1mutations express independent favor-
able prognostic value with regard to OS,
EFS, and disease-free survival (DFS). (Blood.
2005;106:3747-3754)
© 2005 by TheAmerican Society of Hematology
Introduction
Acute myeloid leukemia (AML) is a heterogeneous disease with
diverse genetic abnormalities and variable responsiveness to therapy.
Cytogenetic analyses and molecular analyses are currently used to
risk-stratify AML. For instance, the translocations inv(16), t(8;21),
and t(15;17) herald a favorable prognosis, whereas certain other
cytogenetic aberrations indicate leukemia with intermediate or high
risk of relapse.
1-5
Nevertheless, the classification of AML on the
basis of karyotyping is still far from satisfactory. In recent years
extended molecular analyses have yielded novel molecular markers
important for proper diagnostics of AML. The internal tandem
duplication (ITD) in the fms-like tyrosine kinase-3 gene (FLT3),
6,7
partial tandem duplication (PTD)
8,9
of the mixed lineage leukemia
gene (MLL), and increased expression of the transcription factor
ecotropic virus integration site 1 (EVI1)
10
are indicative of poor
prognosis. In contrast, mutations in the transcription factor CCAAT/
enhancer-binding protein-
(CEBPA) have been associated with a
favorable response to therapy.
11,12
A recent study showed mutations
in exon 12 of the gene encoding nucleophosmin NPM1 in
approximately 35% of cases of de novo AML.
13
Mutations in
NPM1 were found to be mutually exclusive with certain common
recurrent chromosomal aberrations and are predominantly seen in
AMLwith normal karyotypes and FLT3 ITD mutations.
NPM1 is predominantly localized in the nucleolus and is thought to
function as a molecular chaperone of proteins, facilitating the transport
of ribosomal proteins through the nuclear membrane.
14-16
Disruption of
NPM1, either by chromosomal translocation or by mutation, results in
the cytoplasmic dislocation of NPM1. The high frequency of NPM1
mutations in AML with normal karyotypes and the observation that
cytoplasmic NPM1 cannot exert its normal functions as binding partner
and transporter protein lead to the notion thatNPM1mutation may be an
early event in leukemogenesis.
An important role for NPM1 in leukemias and lymphomas has
been proposed previously. NPM1 has been found to be part of
several fusion proteins that are formed as a result of chromosomal
translocation and in which only the NPM1 N-terminal region is
conserved. A t(2;5)(p23;q35) chromosomal translocation occurs in
approximately 8% of non-Hodgkin lymphomas in children and
young adults and results in the chimeric fusion of NPM1 to ALK.
17
In rare cases of acute promyelocytic leukemia (APL), characterized
by chromosomal translocations that disrupt the gene encoding the
retinoic acid receptor (RARA), fusion of NPM1 to RARA was
shown.
18
A t(3;5)(q25.1;q34) chromosomal translocation, infre-
quently seen in myelodysplastic syndrome andAML, gives rise to a
fusion transcript of NPM1 andMLF1.
19
Gene expression profiling is a powerful way to comprehen-
sively classify individuals with AML and to further resolve the
heterogeneous nature ofAML.
20
Using this technique, new prognos-
tically relevant AML subtypes have been identified, while the
From the Departments of Hematology, Statistics, and Internal Medicine,
Erasmus University Medical Center, Rotterdam, The Netherlands.
Submitted June 1, 2005; accepted July 19, 2005. Prepublished online as Blood
First Edition Paper, August 18, 2005; DOI 10.1182/blood-2005-05-2168.
Supported by grants from the Dutch Cancer Society (Koningin Wilhelmina
Fonds) and the Erasmus University Medical Center (Revolving Fund).
The online version of the article contains a data supplement.
An Inside Blood analysis of this article appears in the front of this issue.
Reprints: Peter J. M. Valk, Erasmus University Medical Center Rotterdam,
Department of Hematology, Ee1391a, Dr Molewaterplein 50, 3015 GE
Rotterdam Z-H, The Netherlands; e-mail: p.valk@erasmusmc.nl.
The publication costs of this article were defrayed in part by page charge
payment. Therefore, and solely to indicate this fact, this article is hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. section 1734.
? 2005 by The American Society of Hematology
3747BLOOD, 1 DECEMBER 2005


VOLUME 106, NUMBER 12

presence of recurrent chromosomal abnormalities such as inv(16),
t(15;17), and t(8;21) as well as other molecular aberrations (eg, C-
and N-terminal mutations in CEBPA) could be predicted with high
accuracy by unique expression patterns.
21-23
In a recent study, novel
subtypes of AML have also been defined based on gene expression
profiling; however, the common molecular abnormalities in these
AML subtypes are largely unknown.
21
Because NPM1 is mutated
in approximately one third of AML patients, this molecular
abnormality may drive the clustering of these AML subtypes. The
effect of mutant NPM1 has been studied using gene expression
profiling and revealed a distinctive signature for NPM1 mutations.
24
Among players in this signature were several homeodomain-containing
family members of homeobox (HOX) transcription factors and CD34,
both observations being indicative of hematopoietic development.
24
However, it is currently not known whether NPM1 mutations are
predictable on the basis of a gene expression signature.
Cytoplasmic NPM1 has been positively associated with remis-
sion rate
13
; however, the relation of mutant NPM1 with survival
outcome parameters remains to be elucidated.
We have studied a well-characterized cohort of 275 cases of de
novo AML for the presence of a NPM1 mutations to (1) validate
denaturing high-performance liquid chromatography (dHPLC) as a
rapid approach to determine NPM1 mutations; (2) investigate the
relation of NPM1 mutations with regard to clinical parameters,
cytogenetics, and various molecular abnormalities; (3) determine
the relation of NPM1 mutations in subtypes of AML, recently
identified by gene expression profiling
21
; (4) deriveNPM1mutation–
specific and predictive gene expression signatures; and (5) deter-
mine the prognostic value of mutated NPM1.
Patients, materials, and methods
Patients and cell samples
Patients had a diagnosis of primary AML confirmed by cytologic examina-
tion of blood and bone marrow (median age, 44 years; range, 15-78 years);
median bone marrow blast count, 65% (range, 0% [for APL] to 98%);
median white blood cell (WBC) count, 32  10
9
/L (range, 0.3  10
9
/L to
263  10
9
/L). All patients had been treated according to the Dutch-Belgian
Hemato-Oncology Cooperative Group (HOVON) protocols.
25-27
After
informed consent, bone marrow aspirates or peripheral blood samples were
taken at diagnosis. Blasts and mononuclear cells were purified by Ficoll-
Hypaque (Nygaard, Oslo, Norway) centrifugation and cryopreserved. The
AML samples contained 80% to 100% blast cells after thawing, regardless
of the blast count at diagnosis.
PCR, WAVE, and sequence analyses
RNA isolation and cDNA synthesis were performed as described.
21,28
Complementary DNA prepared from 50 ng RNA was used for all
polymerase chain reaction (PCR) amplifications. NPM1 mutations in exon
12 were determined by cDNA amplification using the primers NPM1-FOR
5-CTTCCGGATGACTGACCAAGAG-3 and primer NPM1-REV 5-
CCTGGACAACATTTATCAAACACG-3 (25 mM deoxyribonucleoside
triphosphate [dNTP], 15 pmol primers, 2 mM MgCl
2
, Taq polymerase, and
10  buffer [Invitrogen Life Technologies, Breda, The Netherlands]).
Cycling conditions for NPM1mutation detection were as follows: 1 cycle, 5
minutes at 94°C; 30 cycles, 1 minute at 94°C, 1 minute at 58°C, and 1
minute at 72°C; and 1 cycle, 7 minutes at 72°C. PCR products were
subsequently subjected to dHPLC using a Transgenomics (Omaha, NE)
WAVE dHPLC system.
29
Samples were run at 56°C and 58°C. The exact
NPM1 mutant sequence was confirmed for all samples showing an
abnormal dHPLC profile. PCR products were purified using theMultiscreen-
PCR 96-well system (Millipore, Bedford, MA) followed by direct sequenc-
ing with NPM1-REV using an ABI-PRISM3100 genetic analyzer (Applied
Biosystems, Foster City, CA).
Sequence analyses for mutations in FLT3 (ITD and tyrosine kinase
domain [TKD] mutation), N-RAS, K-RAS, and CEBPA were performed as
described previously.
12,30,31
Gene expression profiling and unsupervised cluster analyses
A total of 285 AML cases were analyzed using Affymetrix HGU133A
GeneChips (Affymetrix, Santa Clara, CA).
21
Unsupervised cluster analysis
on the basis of the gene expression profiles of the 285 cases of AML was
performed using the correlation view tool (version 3.6) of OmniViz
(Maynard, MA).
21
The Pearson correlation values calculated in OmniViz
were subsequently imported into the MicroArray Data Explorer (MADEx),
which was developed in our laboratory. MADEx was used to visualize the
relations between the OmniViz unsupervised clustering results and other
parameters, such as clinical and molecular characteristics of the AML
patients (Figure 1). MADEx is a database system that stores, mines, and
visualizes microarray data in a secure and scalable manner.
Significance analysis of microarrays (SAM)
All supervised analyses were performed using significance analysis of
microarrays (SAM; version 1.21).
32
A threshold was set for a minimum
change in expression of at least 1.5-fold. A false discovery rate (FDR) of
0.01 was used to select the differentially expressed genes.
Prediction analysis of microarrays (PAM)
All supervised class prediction analyses were performed by applying
prediction analysis of microarrays (PAM; version 2.0).
33
The gene signature
was selected based on the smallest prediction error in the training set and
was subsequently tested using the test set. The positive predictive value was
calculated as follows: true positives/(true positives false positives).
Statistical analyses of survival
Cytogenetic abnormalities were categorized in 3 cytogenetic groups for
statistical analyses. Patients with inv(16)/t(16;16), t(8;21), and t(15;17)
abnormalities were considered as being in the favorable-risk category. The
unfavorable-risk category was defined by the presence of 5/del(5q),
7del(7q), t(6;9), t(9;22), 3q26 abnormality, or complex karyotype (more
than 3 abnormalities). All other patients were classified as intermediate risk.
Statistical analyses were performed with Stata Statistical Software, Re-
lease 7.0 (Stata, College Station, TX). Actuarial probabilities of overall
survival (OS) (with failure defined as death due to any cause) and event-free
survival (EFS) (with failure defined as not achieving complete remission
[set at day 1], relapse, or death in first complete remission) were estimated
by the method of Kaplan and Meier. The Cox proportional hazards model
was applied to determine the association of NPM1 mutation with OS, EFS,
and disease-free survival (DFS) without and with adjustment for other
factors such as cytogenetic risk, age, WBC count, and FLT3 ITD. All tests
were 2-sided, and a P of less than .05 was considered statistically
significant.
Results
Different NPM1 variant mutations in AML
The presence of NPM1 mutations in 275 cases of primary AML
was rapidly and reliably detected by dHPLC WAVE. Nucleotide
sequencing was performed on those cases with an abnormal
dHPLC profile (Table 1). Each NPM1 mutation variant reveals a
specific dHPLCWAVE profile. Thus, each type of NPM1 mutation
could be predicted on the basis of a specific dHPLCWAVE profile.
In addition, 3 novel NPM1 mutant variants were identified
(NPM1 mutants I to K [Table 1]). These rare variants have
comparable 4 bp insertions, like NPM1 variant mutations A to D,
13
3748 VERHAAK et al BLOOD, 1 DECEMBER 2005


VOLUME 106, NUMBER 12