Introduction

The prognosis of pediatric acute myeloid leukemia (AML) patients has improved markedly over the past decade, with an overall survival rate of about 60–70 % according to the results of various clinical trials; however, relapse remains a major cause of treatment failure, occurring in 30–40 % of patients in their first complete remission (CR) [1–6]. Several study groups have shown that chromosome abnormalities are independent and strong predictors of the outcome in both childhood and adult AML [1, 4, 7]. On the other hand, cytogenetically normal acute myeloid leukemia (CN-AML) is the largest cytogenetic subgroup of AML, representing approximately 40 % of pediatric AML patients [7]. Recently, CN-AML has been recognized as highly heterogeneous molecularly, since several abnormalities were discovered, including mutations in FLT3, NPM1, CEBPA, and MLL genes and aberrant expression of BAALC, ERG, and MN1 genes [8]. These alterations have been associated with the treatment outcome and serve as a basis for risk assessment in CN-AML [8, 9]. Discovering novel genetic markers may lead to an improvement in molecular risk stratification and allow a more accurate prediction of the response to therapy.

Wilms tumor 1 (WT1) is located at chromosome 11p13 [10] and encodes a transcription factor capable of activating or repressing gene transcription, depending on the cell type, WT1 protein isoform, and target gene [11]. Although initially considered a tumor suppressor gene [12], WT1 has also been demonstrated to act as an oncogene [11, 13–15]. Mutations of the WT1 gene have been reported in 10–22 % of cases of CN-AML in both adults [16–18] and children [19]. WT1 gene mutations cluster to exons 7 and 9, and are associated with induction failure and/or relapse in adults and children [16–22]. However, there have been few reports on WT1 gene mutation in pediatric AML patients. Thus, we performed mutational analysis of WT1 in pediatric AML patients who were treated on the Japanese Childhood Cooperative Study Group Protocol, AML99 [5], and demonstrated that WT1 mutations were related to a poor prognosis in patients older than 2 years with CN-AML excluding those with FLT3-ITD. Furthermore, we analyzed the association between WT1 mutations and other gene aberrations including RAS and KIT mutations, FLT3-ITD, FLT3-D835, and MLL-PTD.

Materials and methods

Patients

The diagnosis of AML was based on the FAB classification, and cytogenetic analysis was performed using a routine G-banding method. From January 2000 to December 2002, 318 patients were newly diagnosed with de novo AML. Of these, samples from 157 patients were available for molecular analysis, including 13 with FAB-M3 and 10 with Down syndrome (DS), who were treated on different treatment protocols [5, 23–25]. There were no significant differences between the 134 patients without FAB-M3 or DS and the 106 non-analyzed patients in terms of the age [median 6 years (range 0–15 years) vs. 6 years (range 0–15 years), respectively] and initial WBC count [median 24.8 × 109/L (range 1.65-621.0 × 109/L) vs. 13.8 × 109/L (range 1.0-489.0 × 109/L, P = 0.08), respectively]. Patients who were younger than 2 years or had an initial WBC count <100,000/μL were treated using the induction A regimen [etoposide (VP-16), cytarabine, and mitoxantrone (MIT), (ECM)]. Patients who were older than 2 years and had an initial WBC count > 100,000/μL were treated using the induction B regimen [VP-16, cytarabine, and idarubicin (IDA), (ECI)]. If patients achieved a complete remission (CR), they were classified into three risk groups (62 low, 57 intermediate, and 10 high) according to the results of cytogenetic analyses or the achievement of CR after the 2 initial courses of chemotherapy [5, 23–25]. AML patients with t(8;21)(q22;q22) (except for those with WBC counts >50,000/μL) or inv(16)(p13q22) were classified into the low-risk (LR) group. Patients with monosomy 7, 5q-, t(16;21), or Philadelphia (Ph) chromosome were classified into the high-risk (HR) group. Patients were treated with additional chemotherapy or allogeneic hematopoietic stem cell transplantation (allo-HSCT) in each risk group.

Informed consent was obtained from the patients or their parents, according to guidelines based on the tenets of the revised Helsinki protocol. The institutional review board of Gunma Children’s Medical Center approved this project.

Detection of WT1 mutations

Total RNA (4 μg) extracted from the bone marrow or peripheral blood samples at diagnosis was reverse transcribed to cDNA with a cDNA Synthesis Kit (GE Healthcare Japan Corporation, Tokyo, Japan). Mutations of exons 7 and 9 of the WT1 gene were directly sequenced using the following primers: exon 7 WT1-1s 5′-TACGAGAGCGATAACCACAC-3′; exon 7 WT1-4as 5′-GTCCTTGAAGTCACACTGGT-3′; exon 9 WT1-3s 5′-ACCAGTGTGACTTCAAGGAC-3′; exon 9 WT1-2as 5′-TCAAAGCGCCAGCTGGAGTT-3′.

Detection of FLT3-ITD, FLT3-D835, MLL-PTD, KIT, and RAS mutations

Mutation analysis of internal tandem duplication (ITD) within the JM domain and D835 mutation (D835Mt) within the TK2 domain of FLT3 was performed as previously reported [23, 25–27]. Mutation analysis of partial tandem duplication (PTD) of MLL was performed using the primer pair 6.1 (located in exon 9) and E3AS (located in exon 4), as previously reported [25]. Mutation analysis of the kinase domain, extracellular domain, and transmembrane domain of the KIT gene was performed with the reverse transcriptase-polymerase chain reaction (RT-PCR) followed by direct sequencing, as previously reported [23]. Mutation analysis of the RAS gene around hot spots at codons 12, 13, and 61 was performed as previously reported [28].

Detection of WT1 mRNA expression by quantitative RT-PCR (qRT-PCR)

WT1 expression at diagnosis was measured using the qRT-PCR system, as previously reported [29]. We determined the cut-off value of WT1 expression to be 2,500 copies/μg RNA, because the value for the 90th percentile of WT1 expression in normal bone marrow samples was 2,519 copies/μg RNA [29].

Statistical analysis

The χ 2 test was used to compare the frequencies of mutations. Fischer’s exact test was used when data were sparse. The survival distribution was assessed using the Kaplan–Meier method, and differences were compared using the log-rank test [30]. Overall survival (OS) was defined as the time from diagnosis until death owing to any cause or the last follow-up. Disease-free survival (DFS) was defined as the time from the date of complete remission until relapse or death; patients alive and relapse-free at last follow-up were censored. Multivariate analyses were performed to investigate whether WT1 mutation might serve as a prognostic factor in 130 AML patients, excluding those with FAB-M3 and Down syndrome. FLT3-ITD, FLT3-D835, MLL-PTD, KIT, RAS, and WT1 mutation were examined whether these alterations influenced about the 3-year OS and DFS. Karyotypic abnormalities were not included in analytic variables since they were apparently confounded with aforementioned genomic alterations. These statistical analyses were based on Dr. SPSS II for Windows (release 11.0.1J, SPSS; Japan, Inc.).

Results

WT1 mutations

Of the 157 analyzed pediatric AML patients, amplification of the WT1 gene by RT-PCR was not completed in 4 (2.5 %) cases. Therefore, the following analyses were conducted with the 153 evaluable cases excluding these four.

Ten (6.5 %) of the 153 cases had an activating mutation (Table 1). In 7 cases (70 %) with WT1 mutation, two or more mutations were detected in the WT1 gene (Table 1). There was no significant difference in the age, sex, WBC count at diagnosis, or frequency of extramedullary infiltration of leukemic cells between patients with and without WT1 mutations; however, the frequency of allo-HSCT was significantly higher in patients with than in those without WT1 mutation (70.0 vs 35.7 %, respectively, P = 0.03) (Table 2).

Table 1 Ten cases with WT1 mutation in this study
Full size table
Table 2 Clinical characteristics of 153 evaluable patients with and without WT1 mutations
Full size table

The incidence of mutations in WT1 was significantly higher in pediatric CN-AML (15.2 vs. 4.5 %, respectively, P = 0.04) (Table 3, in which DS patients were not included in karyotypic abnormalities).

Table 3 FAB classification, other genomic alterations, and karyotypic abnormalities of 153 evaluable patients with WT1 mutations
Full size table

Correlations between WT1 mutations and other gene aberrations

The incidence of mutations in KIT was significantly higher in patients with than in those without the WT1 mutation (30 vs. 5.6 %, respectively, P < 0.01). Moreover, the incidence of FLT3-D835 mutation was also significantly higher in patients with than in those without the WT1 mutation (20.0 vs. 4.2 %, respectively, P = 0.03). The distribution of FLT3-ITD, MLL-PTD, and mutations in NRAS and KRAS was not different from those without WT1 mutation (Table 3).

Correlation between WT1 mutation and WT1 mRNA expression

A higher WT1 expression (≥2,500 copies/μg RNA) was detected in 9 (90 %) of 10 cases with WT1 mutation (Table 1). On the other hand, a higher WT1 expression was detected in 113 (77 %) of 147 cases without WT1 mutation. The difference was not significant (P = 0.33).

Clinical outcome and prognostic significance of WT1 mutations

There were no differences in the 3-year OS and DFS between those with and without WT1 mutation in 130 evaluable AML patients, excluding those with FAB-M3 and DS (Fig. 1). The frequency of WT1 mutation was not different between patients with and without CR after induction therapy (6.7 vs. 20.0 %, respectively, P = 0.13). Among patients with a normal karyotype, WT1 mutation tended to be related to a poorer 3-year OS and DFS than those without WT1 mutation, although the differences were not significant (P = 0.38 and P = 0.45, respectively) (Fig. 2).

Fig. 1
figure 1

Probability of 3-year OS (a) and DFS (b) in 130 AML patients, excluding those with FAB-M3 and Down syndrome. Kaplan–Meier estimates for patients with and without WT1 mutation are shown

Full size image
Fig. 2
figure 2

Probability of 3-year OS in 34 patients (a) and DFS in 29 patients (b) with CN-AML, excluding those with FAB-M3 and Down syndrome. Kaplan–Meier estimates for patients with and without WT1 mutation are shown

Full size image

WT1 mutations were not randomly distributed over the different cytogenetic subgroups. The frequency of WT1 mutation in CN-AML was higher than in other cytogenetic subgroups (P = 0.04). This trend was similar to previous pediatric reports [19, 21, 22]. Moreover, the frequency of WT1 mutation in patients <3 years was lower than in patients aged 3 years or older; however, this difference was not significant (0 vs. 8.6 %, respectively, P = 0.06). In other pediatric reports, the frequency of WT1 mutation was significantly lower in patients <3 years old than in patients aged 3 years or older [19, 21, 22]. Furthermore, FLT3-ITD in AML was too strong a prognostic factor to assess whether or not WT1 mutation has a prognostic impact [25]. Thus, we analyzed the clinical impact of WT1 mutation in patients with CN-AML excluding those with FLT3-ITD and <3 years. In patients with a normal karyotype, aged 3 years or older, and showing no evidence of FLT3-ITD, WT1 mutation was related to a poorer prognosis based on the 3-year OS and DFS (P = 0.17 and P < 0.01, respectively) (Fig. 3). WT1 mutation was not a significant risk factor on 3-year OS and DFS by multivariate analyses (Tables 4, 5).

Fig. 3
figure 3

Probability of 3-year OS in 20 patients (a) and DFS in 19 patients (b) aged 3 years or older with CN-AML, excluding those with FLT3-ITD, FAB-M3, and Down syndrome. Kaplan–Meier estimates for patients with and without WT1 mutation are shown

Full size image
Table 4 Multivariate model of prognostic risk factors (overall survival, N = 130)
Full size table
Table 5 Multivariate model of prognostic risk factors (disease-free survival, N = 122)
Full size table

Discussion

Although several papers reported the differences in clinical outcome between patients with and without WT1 mutation, we could not identify any differences between them in this study. However, other studies demonstrated that these mutations had no independent effects on the outcome when the FLT3-ITD status was taken into account [18, 21, 22]. These results suggest an effect on the clinical outcome due to the different treatment regimens used in each study. For example, when comparing the treatment protocols, the cumulative doses of high-dose cytarabine given for consolidation treatment were markedly different. The current patients were treated on the Japanese Childhood AML Cooperative Study Group Protocol, AML99, in which the intensive use of cytarabine, including high-dose cytarabine, was considered to improve the outcome. Improvement of the clinical outcome of patients with WT1 mutation in this study might have decreased the differences in the 3-year OS and DFS between patients with and without WT1 mutation, although WT1 mutation was not a significant risk factor on 3-year OS and DFS by multivariate analyses. WT1 mutations have been reported to be an adverse prognostic factor in some studies because of the high frequency of coexisting FLT3-ITD [18, 21, 22]; however, only 2 cases had both WT1 mutation and FLT3-ITD in our study. Because of the small number of cases, prognostic analysis for each status of WT1 mutation and FLT3-ITD was not performed. On the other hand, WT1 mutations were associated with a poor prognosis in patients with CN-AML excluding those with FLT3-ITD and less than 3 years old. There has been no similar report of this result. Although further validation of the present results is required, WT1 mutation might be a prognostic factor in patients with CN-AML excluding those with FLT3-ITD and <3 years old.

The frequencies of WT1 mutations (6.5 % of total AML and 15.2 % of CN-AML) tended to be low compared with previous pediatric reports (8.2–11.7 and 14.3–22.3 %, respectively) [19, 21, 22], although the differences were not significant. This might be due to racial differences or the samples used. Previous reports used genomic DNA for analyzing WT1 mutation; however, we could use only cDNA. We could not amplify WT1 genes by RT-PCR in 4 cases (2.5 %). In a previous study, Hollink et al. [19] described 2 cases (1 %) with homozygous deletion out of 298 patients. If homozygous deletions occur in tumor suppressor genes, expressions will be lost and analyses using cDNA will be impossible. As a result, the rate of detecting mutations in these genes is generally higher in genomic DNA than in cDNA [31–33]. Therefore, genomic DNA is usually used for the analysis of these genes. Loss of WT1 amplification in our study might be partially explained by homozygous deletions of WT1 genes; however, the frequency of homozygous deletion based on a previous report [19] seemed to be low. Thus, its influence on our research might be limited.

When combined with three previous pediatric reports and our data, WT1 mutations were identified in 50 out of 286 (17.5 %) patients with CN-AML [19, 21, 22]. On the other hand, they were identified in 146 out of 1,283 (11.4 %) adult patients when combined with three large-scale reports [16–18]. The frequency of WT1 mutation in patients with CN-AML was significantly higher in pediatric compared with adult patients (P < 0.01). It was impossible to compare the frequencies of WT1 mutations in AML patients other than CN-AML because the analyses were usually performed focusing on patients with CN-AML in adult reports.

In 70 % of cases with WT1 mutations, two or more mutations were detected in the WT1 gene (Table 1). This frequency was higher than those in previous reports by Hollink et al. [19] (16/35, 46 %, P = 0.28), and Ho et al. [21] (15/70, 21 %, P < 0.01). It became clear that the existence of multiple mutations of WT1 genes was not rare. Meanwhile, 90 % of cases with WT1 mutations were accompanied by other mutations, including FLT3-ITD, MLL-PTD, and mutations of FLT3 D835, KIT, or RAS (Table 1). This frequency was higher than in patients without WT1 mutation (67/143, 47 %, P < 0.01). From these results, there is a possibility that mutations of WT1 and other genes collaborate and participate in the development of AML. The traditional model of molecular-genetic cooperativity in myeloid leukemogenesis states that “class II” events, which impair differentiation, must be coupled with “class I” events, which confer a proliferative advantage [34]. In our study, WT1 mutations showed significant overlap with class I mutation, such as FLT3-ITD, MLL-PTD, and mutations of FLT3 D835 and KIT, so the role of WT1 mutation in the stepwise evolution might be associated with the arrest of differentiation.

WT1 mRNA expression at diagnosis tended to be higher in patients with compared with those without WT1 mutation, although the difference was not significant, probably due to the low number of patients with WT1 mutation. Overexpression of wild-type WT1 is a common finding in AML [35–37], although WT1 mutations in AML appear to result in a loss of WT1 function. This contradiction, in which a single gene might function as both an oncogene as well as a tumor suppressor, may stem from the ability of the WT1 protein to function either as a transcriptional activator or repressor, depending on a multitude of factors [11]. There is still much to be learned about the biology of WT1 in AML.

In conclusion, WT1 mutations were the most common in patients with normal karyotype AML, and showed no correlation with the 3-year OS and DFS. However, these mutations were associated with a poor prognosis in patients with CN-AML excluding those with FLT3-ITD and <3 years old.