OUP user menu

FLT3 and NPM1 Mutations in Myelodysplastic Syndromes
Frequency and Potential Value for Predicting Progression to Acute Myeloid Leukemia

Ashish Bains MD, Rajyalakshmi Luthra PhD, L. Jeffrey Medeiros MD, Zhuang Zuo MD, PhD
DOI: http://dx.doi.org/10.1309/AJCPEI9XU8PYBCIO 62-69 First published online: 1 January 2011


We reviewed FLT3 and NPM1 mutation data in a large cohort of patients with myelodysplastic syndrome (MDS). The frequencies of FLT3 and NPM1 mutation were 2.0% and 4.4%, respectively, and mutations were restricted to cases of intermediate- and high-risk MDS. Cytogenetic abnormalities were identified in 46.9% of cases. FLT3 mutations were associated with a complex karyotype (P = .009), whereas NPM1 mutations were associated with a diploid karyotype (P < .001). FLT3 mutation (P < .001) was associated with progression to acute myeloid leukemia (AML), as were a higher bone marrow (BM) blast count (P < .001) and complex cytogenetics (P = .039). No patient with an NPM1 mutation alone had disease that progressed to AML. Cox proportional regression multivariate analysis indicated that FLT3 mutation, NPM1 mutation, complex cytogenetics, BM blast count, pancytopenia, and age were independent factors that correlated with progression-free survival. We conclude that FLT3 and NPM1 mutations are rare in MDS, but assessment of mutation status is potentially useful for predicting progression to AML.

Key Words:
  • Myelodysplastic syndromes
  • FMS-like tyrosine kinase
  • FLT3
  • Nucleophosmin
  • NPM1
  • Mutation

Myelodysplastic syndromes (MDSs) constitute a heterogeneous group of stem cell disorders characterized by ineffective hematopoiesis with an increased but variable risk of progression to acute myeloid leukemia (AML). Currently, the diagnosis of MDS is mainly based on clinical, morphologic, and cytogenetic features. Although cytogenetic data have been shown to have a major role in predicting prognosis, only 30% to 50% of patients with de novo MDS have cytogenetic abnormalities.1 An improved understanding of the molecular pathogenesis of MDS is needed for more accurate prediction of prognosis and the development of targeted therapy. FMS-like tyrosine kinase 3 (FLT3) and nucleophosmin (NPM1) are among the most commonly mutated genes in AML. The frequency of these mutations, their possible pathogenetic role in MDS, and their potential value for predicting progression to AML are less well studied.

FLT3, a tyrosine kinase receptor, has a key role in growth regulation of hematopoietic progenitor cells.2 Mutations in FLT3 have been detected in approximately 30% of AML cases.3,4 The most common mutations include an internal tandem duplication (FLT3-ITD) in the juxtamembrane domain and missense point mutations in the tyrosine kinase domain, most frequently at codon 835 (D835). FLT3-ITD has been observed twice as frequently in AML cases with NPM1 mutations compared with AML cases with wild-type NPM1.5 The frequency of FLT3 mutations has been reported to be much lower in MDS, approximately 3% to 6%, in studies of small case series.68

The NPM1 gene encodes for a multifunctional phosphoprotein located primarily in the nucleolus.15 NPM1 mutations are known to be common in AML and are commonly associated with a diploid karyotype. Others have reported NPM1 mutations in 46% to 62% of patients with de novo AML with a normal karyotype.5,9 NPM1 mutations in AML involve exon 12 and cause a frame shift in the translation and aberrant cytoplasmic dislocation of the protein.10 In a recent study, cytoplasmic localization of NPM1 corresponded with responsiveness to induction chemotherapy in patients with AML with a normal karyotype.11 However, the mechanism of NPM1 mutation in leukemogenesis is not clearly established. Some have speculated that NPM1 mutations occur in a small, undetectable subset of normal pluripotent hematopoietic stem cells.11,12 Relatively few data regarding the presence of NPM1 mutations are available for MDS cases. One recent small study in a Chinese population found NPM1 mutation in approximately 5% of patients with MDS.13

The goal of this study was to review our experience in assessing MDS patients for FLT3 and NPM1 mutations. Unique features of this study are the very large cohort of cases reported and the integration of mutation status with morphologic, cytogenetic, and other molecular findings and clinical outcome. Our findings suggest that FLT3 and NPM1 mutations occur predominantly in high-risk types of MDS and that FLT3 mutation is associated with a higher frequency of progression to AML. Routine screening for these mutations in patients with MDS seems to be useful for clinical stratification and risk assessment.

Materials and Methods

Case Selection

In this retrospective study, we searched the records of the Molecular Diagnostics Laboratory, Department of Hematopathology, The University of Texas M. D. Anderson Cancer Center, Houston, for cases with a diagnosis of MDS that had been tested for FLT3 or NPM1 mutations. At our institution, testing for FLT3 has been routinely performed on virtually all patients with AML and patients with MDS since 2002 as a part of routine clinical protocols. By contrast, routine testing of patients with AML and patients with MDS for NPM1 mutations was instituted much more recently, and NPM1 testing is virtually always ordered in combination with FLT3 testing. For this reason, we began this study by collecting all FLT3 mutation data in patients with MDS, and, from this group, we then searched for patients with MDS tested for NPM1 mutations. For most cases in this study group, molecular testing was performed on diagnostic bone marrow (BM) samples in untreated patients. A smaller subset of patients had received therapies before being referred to our institution when molecular testing on a BM specimen showing MDS was performed. Seven cases in this study had well-documented MDS, but the initial BM samples were not available for molecular testing. For these cases, molecular testing was performed on BM samples at the time of progression to AML. In each of these 7 cases, progression to AML occurred within 1 year of the initial diagnosis of MDS.

As part of the diagnostic workup, clinical data, the CBC, and morphologic findings in blood and BM were assessed. All cases were diagnosed and classified as MDS or MDS/myeloproliferative neoplasm (MPN) according to the current World Health Organization criteria.14 Cytogenetic and additional molecular test results were also collected and analyzed whenever available. This study was approved by the institutional review board and conducted in accordance with the Declaration of Helsinki.

Cytogenetic and Molecular Analysis

Conventional cytogenetic analysis was performed on BM aspiration specimens in most cases, as described previously.15 Karyotypes were reported using the International System for Human Cytogenetic Nomenclature.

For mutation testing, genomic DNA was extracted from fresh BM aspiration specimens for mutational analyses using the Autopure extractor (QIAGEN/Gentra, Valencia, CA). FLT3 (ITD and D835) and NPM1 (exon 12) mutations were screened using polymerase chain reaction (PCR) followed by capillary electrophoresis on an ABI Prism 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA), as previously described.16,17 For FLT3 codon 835 point mutation analysis, PCR products were digested with EcoRV before capillary electrophoresis. A subset of cases was also analyzed for other genetic mutations, including RAS (K- and N-), c-KIT, and JAK2. These genes were assessed by PCR followed by direct sequencing, Sanger sequencing, or pyrosequencing, according to previously described protocols.15,18 The sensitivity of our FLT3 and NPM1 tests is about 2.5%. The sensitivities for Sanger sequencing and pyrosequencing are about 20% and 10%, respectively.

Statistical Analysis

Categorical variables were compared between different groups using the χ2 test. The Mann-Whitney test was applied to compare numeric variables. Multivariate analysis was performed by using the Cox proportional regression model to examine the significance of clinical, cytogenetic, and molecular factors on disease progression. Progression-free survival was estimated by the Kaplan-Meier method, and differences between the presence and absence of factors were compared by using the log-rank test.


Study Group and Frequencies of FLT3 and NPM1 Mutations

Our database search identified 1,316 patients with MDS (n = 1,025) or MDS/MPN (n = 291) who were tested for FLT3 (ITD and D835) mutations. Table 1 summarizes the patient characteristics and other data for this study group. The cases of MDS were further subcategorized as de novo (n = 875) or therapy-related (t-MDS) (n = 150). Most cases of MDS/MPN were further classified as chronic myelomonocytic leukemia, but a subset was unclassifiable. A total of 26 patients (2.0%) had FLT3 mutations: 20 (1.5%) with FLT3-ITD and 6 (0.5%) with FLT3-D835. The frequency of FLT3 mutation in each group was as follows: 15 (1.7%) of 875 de novo MDS cases, 4 (2.7%) of 150 t-MDS cases, and 7 (2.4%) of 291 MDS/MPN cases. There was no statistical difference in the frequency of FLT3 mutations among these 3 groups (P = .381).

A total of 160 patients with FLT3 data were also tested for NPM1 mutation, and 7 (4.4%) had an NPM1 mutation. Table 2 lists the detailed clinical information for these patients. Mutations occurred in 3 (2.8%) of 107 de novo MDS cases, 1 (3%) of 32 t-MDS cases, and 3 (14%) of 21 MDS/MPN cases. The frequency of NPM1 mutation was similar in the de novo MDS and t-MDS groups (P = .286) but significantly higher in the MDS/MPN group (P = .023). There was no significant difference in the frequencies of FLT3 and NPM1 mutations according to sex (P = .129 and P = .172, respectively) or age (P = .256 and P = .672, respectively).

Correlation of FLT3 and NPM1 Mutations With Morphologic Features, Cytogenetics and Other Mutations

As shown in Table 1, within de novo MDS categories, FLT3 mutations were restricted to 2 morphologic categories: refractory cytopenia with multilineage dysplasia (RCMD) and refractory anemia with excess blasts (RAEB), 1 or 2 (P < .001 for both comparisons). The mutation frequency was significantly higher in RAEB (3.1%) than in RCMD (1.1%; P < .001), but the difference was not significant between RAEB-1 and RAEB-2 (P = .497). NPM1 mutations were observed exclusively in the RAEB category (3/58 [5%]; P < .001), with no significant difference between RAEB-1 and RAEB-2 (P = .467).

View this table:
Table 1
View this table:
Table 2

Table 3 summarizes the FLT3 and NPM1 mutation results and the correlation with cytogenetic data and the results of other gene mutations tested. Among the 26 FLT3 mutated cases, 15 (58%) had a normal karyotype. Of the FLT3-ITD and FLT3-D835 mutated cases, 60% (12/20) and 50% (3/6), respectively, were seen in association with normal karyotype, but overall there were no significant differences in the frequencies of mutated FLT3 (15/26 [58%]) and wild-type FLT3 (53.0%) in patients with normal karyotypes (P = .316). Among the 11 FLT3-mutated MDS cases with abnormal cytogenetics, 1 patient had del(5q), 3 patients had −7/del(7q), and 1 patient had trisomy 8. These abnormalities did not significantly correlate with FLT3 mutation status (P = .397, P = .322, and P = .712, respectively). Of 26 FLT3-mutated MDS cases, 5 (19%) had complex cytogenetics with a similar frequency between FLT3-ITD (4/20 [20%]) and FLT3-D835 (1/6 [17%]) cases (P = .317). The frequency of complex cytogenetics in FLT3-mutated cases was significantly higher than the 6.5% frequency in patients with wild-type FLT3 (P = .009), suggesting a possible association between FLT3 mutations and complex cytogenetics. NPM1 mutations, on the other hand, were significantly associated with normal cytogenetics, as all 7 MDS patients with NPM1 mutation had a normal karyotype (P < .001).

Of the 26 FLT3+ MDS cases, 21 were tested for RAS mutations and 3 (14%) were mutated, significantly higher than the 4.9% frequency in wild-type FLT3 MDS cases (P < .001). A single MDS case with an FLT3 mutation had a concomitant c-KIT mutation (9%), and this was significantly more frequent than in MDS cases with wild-type FLT3 (P < .001). No significant association between FLT3 and JAK2 mutations was found (P = .794). In the small group of cases with an NPM1 mutation, there were no cases with additional mutations in c-KIT, RAS, or JAK2. However, FLT3 mutations (3 cases with FLT3-ITD and 1 with FLT3-D835) were almost 4 times more frequent in NPM1-mutated MDS compared with wild-type NPM1 MDS (P < .001), suggesting a strong association between these 2 mutations.

Correlation of FLT3 and NPM1 Mutations With Disease Progression

To identify factors that contribute to progression of MDS to overt AML, we obtained follow-up data on 156 of 160 patients who were tested for FLT3 and NPM1 mutations. The median follow-up duration for this cohort was 18 months. The disease in 49 of 156 patients progressed to AML after a median interval of 9 months (range, 1–108 months). Table 4 shows the comparison of the morphologic category of MDS, cytogenetic data, and mutation status in these cases with or without progression to AML. The rate of progression was not significantly different for patients with de novo MDS, t-MDS, and MDS/MPN. However, within the de novo MDS group, RCMD and RAEB were associated with significantly higher rates of progression to AML (P < .001 for both).

View this table:
Table 3
View this table:
Table 4

FLT3 mutations, higher initial BM blast count, and complex cytogenetics were significantly associated with progression to AML (P < .001, P = .039, and P < .001, respectively). The rate of progression to AML also seemed to be higher in patients with FLT3-D835 compared with patients with FLT3-ITD (5/6 vs 11/20; P < .001). Complex cytogenetics was observed in 57 patients, of whom 23 had disease that progressed to AML, with a median duration of 5 months (range, 1–108 months). No significant difference in progression rate between patients with or without NPM1 mutations alone was identified (P = .133). However, as shown in Figure 1, mutations in FLT3 alone, FLT3 in combination with NPM1, and complex cytogenetics had a significant negative impact on progression-free survival (P = .010, P = .026, and P = .011, respectively).

Among the 22 MDS cases with mutated FLT3 and wild-type NPM1, 12 progressed to AML during a median follow-up of 8 months (range, 1–72 months); the progression rate was not significantly different from that for the entire cohort of FLT3-mutated MDS cases, regardless of the NPM1 status (P = .149). All 4 cases with both NPM1 and FLT3 mutations progressed to AML with a median interval of 12 months (range, 2–13 months; P < .001). Three patients had NPM1 mutation without FLT3 mutation; none had disease that progressed to AML after 7 to 14 months of follow-up (P < .001).

Figure 1

Progression-free survival in 156 patients with myelodysplastic syndrome according to FLT3 mutation status (A, P = .010), FLT3 and NPM1 combined mutation status (B, P = .026), and with or without complex cytogenetics (C, P = .011).

Normal cytogenetics did not correlate with progression to AML (P = .884). Among the cases with normal cytogenetics, 20 progressed to AML. The FLT3 and NPM1 mutation rates were significantly higher, at 45% (9/20) and 20% (4/20), respectively (P < .001 for both), in these progressed cases, and all patients with NPM1 mutations also had FLT3 mutations. Pancytopenia, although not significantly associated with progression (P = .104), was correlated with the presence of complex cytogenetics (P = .007). RAS mutation status had no significant impact on disease progression (P = .707). In addition, neither BM blast count nor pancytopenia was associated with FLT3 or NPM1 mutation (P = .183 and P = .763; and P = .526 and P = .443, respectively).

We applied the aforementioned parameters into a Cox proportional regression multivariate analysis model to assess their effects in disease progression. Table 5 shows the significant factors and their hazard ratios. Besides the factors identified in univariate analysis, advanced age appeared to have a negative impact on progression-free survival. After adjusting for age, pancytopenia also became marginally significant in reducing progression-free survival.

We also calculated the International Prognosis Scoring System (IPSS) score, a widely accepted system used to assess risk in patients with MDS. The IPSS is based on the number of cytopenias, BM blast count, and cytogenetic data and uses a point system. A high score correlates with higher risk for a given patient with MDS. A high IPSS score (P = .001), FLT3 (P = .017) or NPM1 (P = .037) mutation, and advanced age (P = .011) were significantly associated with progression to AML (Table 5).


To our knowledge, this is the largest cohort of patients with MDS studied for FLT3 and NPM1 mutations reported in the literature. Our findings show that FLT3 and NPM1 mutations occur in approximately 2% and 4% of cases of MDS, respectively. Although these frequencies are substantially lower than those reported in AML, the characteristic association between NPM1 and FLT3 mutations and diploid karyotype that has been observed in patients with AML held true in patients with MDS.11 FLT3 and NPM1 mutations were significant independent factors in predicting progression of MDS to AML, as were complex cytogenetics, BM blast count, pancytopenia, and age.

The frequencies of FLT3 and NPM1 mutation did not differ significantly between patients with de novo MDS and t-MDS, suggesting that these mutations are not specific for, or strongly associated with, the effects of prior therapies. Rather, we found that FLT3 mutations were restricted to the intermediate- and high-risk morphologic categories of MDS, in particular RCMD and RAEB, and NPM1 mutations were restricted to the RAEB group. Patients with low-risk MDS, including the categories of refractory cytopenia with unilineage dysplasia, refractory anemia with ringed sideroblasts, and MDS associated with isolated del(5q), did not have FLT3 or NPM1 mutations. These findings collectively reflect possible distinct mechanisms in the molecular development of these types of MDS and/or the dynamic evolution of the tumor genome with disease progression. The latter concept is further supported by the strong association between FLT3 mutations and complex cytogenetics and progression to AML observed in this study.

Cases of MDS/MPN had the highest frequency of NPM1 mutations, approximately 15%, implying that NPM1 mutation may be involved in the pathogenesis of these disorders and probably is associated with the myelodysplastic component of these neoplasms. In support of this suggestion, NPM1 mutation alone appears adequate to cause an MDS in a transgenic mouse model.19 In our study, in nearly half of the NPM1-mutated cases, there were no cytogenetic or other molecular abnormalities, further suggesting a key role for NPM1 mutation in the development of myelodysplasia.

View this table:
Table 5

NPM1 mutation alone, however, is not adequate to explain progression of MDS to AML. In our study, none of the cases of MDS with NPM1 mutations alone progressed to AML. In fact, many MDS cases with NPM1 mutation alone did not progress to AML during prolonged follow-up. When NPM1 mutations were concomitant with FLT3 mutations, however, every case progressed to AML during the follow-up period. Similar phenomena have been reported in AML, as an NPM1 mutation has been shown to predict a favorable prognostic outcome in AML with normal cytogenetics but only in the absence of FLT3 mutation.20 This further substantiates the idea that NPM1 mutation alone may not be sufficient for progression into overt AML but, in conjunction with secondary events in the form of other mutations, predisposes a progenitor/stem cell to malignant transformation. To date, the only cytogenetic and molecular abnormality that has been found to be associated with NPM1 mutation is FLT3 mutation. Our results indicate that this association may be even stronger in MDS than in AML and suggest the possibility that mutations of NPM1 may induce or facilitate the development of FLT3 mutations. However, the exact mechanism of this association is unexplained.

A model of cooperation between 2 classes of mutations has been proposed as a possible mechanism in AML leukemogenesis.21 A class II mutation, such as that involving NPM1, impairs hematopoietic differentiation, perhaps similar to MDS. A second class I mutation provides a proliferative and/or survival advantage needed for progression to AML.22 The findings in this study and in others in the literature have suggested that FLT3 mutation has a central role in progression of MDS to AML.7 Our study showed that FLT3 mutations were not only associated with NPM1 mutation in cases with a diploid karyotype, but were also strongly associated with complex cytogenetics, suggesting that mutations of this gene can be induced through multiple mechanisms. Similarly, we found that mutation in another class I gene, RAS, was also significantly associated with complex cytogenetics, although we could not find a statistically significant association with disease progression as suggested in another study.23 It is interesting that we found that mutations of 3 class I genes, FLT3, RAS, and c-KIT, correlated with each other, raising the possibility that these mutations resulted from similar mechanisms, possibly related to tumor genomic instability.

Cytogenetic data, BM blast count, and the number of hematopoietic cell lines showing cytopenia are components of the widely adopted IPSS for risk assessment in patients with MDS, useful for predicting survival and risk of progression to AML.14 The importance of these parameters was validated in this study cohort. However, MDS cases with a diploid karyotype also are known to progress to AML. Our data and the data of others show that MDS cases with diploid cytogenetics that progress to AML are more likely to harbor FLT3 mutations or concomitant NPM1 and FLT3 mutations, which most likely contribute to disease progression. These findings argue for the necessity of incorporating FLT3 and NPM1 mutation status into the assessment of prognosis in patients with MDS to increase the prediction accuracy and benefit a broader patient population.

Although FLT3 and NPM1 mutations occur at a low frequency in MDS, cases with these mutations share cytogenetic and mutational profiles similar to those in AML. These mutations can serve as potential molecular markers for predicting progression to AML in patients with MDS. Hence, we believe that evaluation of these mutations should be part of routine clinical practice for stratification and risk assessment for patients with MDS. These genes also represent potential targets for novel therapies.


  • * Dr Bains participated in this work as a visiting resident. He is presently with the Department of Pathology, Oklahoma University Health Sciences Center, Oklahoma City.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
View Abstract