- Heiko Becker1,
- Kati Maharry1,2,
- Michael D. Radmacher1,2,
- Krzysztof Mrózek1,
- Klaus H. Metzeler1,
- Susan P. Whitman1,
- Sebastian Schwind1,
- Jessica Kohlschmidt1,2,
- Yue-Zhong Wu1,
- Bayard L. Powell3,
- Thomas H. Carter4,
- Jonathan E. Kolitz5,
- Meir Wetzler6,
- Andrew J. Carroll7,
- Maria R. Baer8,
- Joseph O. Moore9,
- Michael A. Caligiuri1,
- Richard A. Larson10,
- Guido Marcucci1⇓ and
- Clara D. Bloomfield1⇓
- 1Division of Hematology, Department of Internal Medicine, Comprehensive Cancer Center, The Ohio State University, Columbus, OH, USA
- 2The Cancer and Leukemia Group B Statistical Center, Duke University Medical Center, Durham, NC, USA
- 3Comprehensive Cancer Center, Wake Forest University, Winston-Salem, NC, USA
- 4University of Iowa, Iowa City, IA, USA
- 5North Shore University Hospital, Manhasset, NY, USA
- 6Roswell Park Cancer Institute, Buffalo, NY, USA
- 7University of Alabama at Birmingham, Birmingham, AL, USA
- 8University of Maryland, Baltimore, MD, USA
- 9Duke University Medical Center, Durham, NC, USA
- 10University of Chicago, Chicago, IL, USA
- Correspondence: Clara D. Bloomfield, MD, The Ohio State University Comprehensive Cancer Center, 1216 James Cancer Hospital, 300 West 10th Ave, Columbus, OH 43210 USA. E-mail: , Guido Marcucci, MD, The Ohio State University Comprehensive Cancer Center, 809C Biomedical Research Tower, 460 West 12th Ave, Columbus, OH 43210 USA. E-mail:
Background The alleles of the Wilms tumor 1 (WT1) polymorphism rs16754 harbor adenine (A) or guanine (G). Recently, rs16754 has been reported to affect the outcome of patients with cytogenetically normal acute myeloid leukemia. To validate this finding, we investigated pretreatment features and outcome associated with rs16754 in a large cohort of patients with cytogenetically normal acute myeloid leukemia.
Design and Methods Four-hundred and thirty-three intensively treated and molecularly characterized cytogenetically normal patients with de novo acute myeloid leukemia (18–83 years old) were analyzed for rs16754. To gain biological insights, we studied the gene- and microRNA-expression profiles for associations with rs16754.
Results Three-hundred and nine (71%) patients were homozygous for A (WT1AA), 112 (26%) were heterozygous (WT1AG) and 12 (3%) were homozygous for G (WT1GG). For comparison with previous studies, we grouped WT1AG and WT1GG patients and compared them with WT1AA patients divided into younger (<60 years) and older (≥60 years) adults. We found no independent prognostic impact of WT1AA. However, WT1GG patients, who were less often Caucasian than WT1AG (P=0.001) or WT1AA (P=0.008) patients, and had TET2 mutations more often than WT1AG (P=0.02) patients, had, among patients with FLT3-internal tandem duplication and/or NPM1 wild-type, better disease-free (P=0.02) and overall survival (P=0.04) than WT1AA and WT1AG patients combined. Unsupervised and supervised analyses of the gene- and microRNA-expression profiles suggested that there were no distinct expression patterns associated with any rs16754 genotype.
Conclusions We did not observe the previously reported adverse impact of WT1AA but found favorable outcomes associated with the homozygous WT1GG. Considering its low frequency, confirmatory studies are necessary. The biological significance of rs16754 remains questionable as no distinct expression profiles were associated with the genotypes.
Mutations in the Wilms tumor 1 gene (WT1) occur in approximately 10% of adults with cytogenetically normal acute myeloid leukemia (CN-AML), and have been reported to confer a worse outcome1–6 or to have no prognostic impact.7–8 The synonymous single nucleotide polymorphism (SNP) rs16754 (dbSNP Build ID: 131),9 located in exon 7 of the WT1 gene, which is a “hot spot” for WT1 mutations in AML, has also recently been reported to be associated with outcome.8,10–13 SNP rs16754 has two alleles which differ by harboring the nucleotide adenine (A) or guanine (G). They are present in a homozygous (denoted WT1AA or WT1GG, respectively) or heterozygous state (WT1AG).
Damm et al.8 reported that among younger adults (17–60 years) with CN-AML, those who were homozygous for the A allele (WT1AA) had a trend for a lower complete remission rate, and significantly shorter relapse-free survival and overall survival than patients harboring at least one G allele (WT1AG/WT1GG). This adverse prognostic impact of the WT1AA genotype on relapse-free and overall survival remained significant in multivariable analyses considering other prognostic markers, and was most pronounced in the high-risk subset of patients with FLT3-internal tandem duplication (FLT3-ITD) and/or NPM1 wild-type.8 In contrast, Renneville et al.10 described in a preliminary report that among older CN-AML patients (aged 50 to 70 years) those with WT1AA had better outcomes than the group of WT1AG/WT1GG patients. In pediatric AML, SNP rs16754 was not found to be of prognostic significance in two independent cohorts of CN-AML patients.11,12 However, WT1AG/WT1GG genotypes were identified as a favorable prognostic factor among children with low-risk disease, defined by Ho et al.11 as patients with t(8;21)(q22;22), inv(16)(p13;q22) or t(16;16)(p13;q22), together with those carrying CEBPA or NPM1 mutations. In a meeting abstract, Ma et al.13 reported on cytogenetically heterogeneous adult AML patients, describing that those homozygous for the G allele (WT1GG) had a better overall survival than the combined group of WT1AA and WT1AG patients.
Thus, to further clarify the clinical significance of the rs16754 polymorphism, we investigated the prognostic impact of its genotypes in a relatively large group of 433 intensively treated de novo CN-AML patients aged 18 to 83 years who were comprehensively characterized for other molecular markers. Moreover, to gain insights into the potential biological diversity related to the rs16754 polymorphism, we compared the gene- and, for the first time, the microRNA-expression profiles according to the rs16754 genotypes.
Design and Methods
Patients, treatment and cytogenetic analysis
We studied pretreatment bone marrow or blood with 20% or more blasts from 433 patients enrolled on cytarabine and daunorubicin-based Cancer and Leukemia Group B (CALGB) first-line treatment protocols as summarized in the Online Supplementary Data. Patients with preceding hematologic disorders or treatment-related AML were excluded. In accordance with the treatment protocols, no patient received allogeneic stem-cell transplantation in first complete remission. The median follow-up for patients alive was 6.5 years (range, 2.3–11.7 years).
Cytogenetic analyses at diagnosis were performed by CALGB-approved institutional cytogenetic laboratories, and the results confirmed by central karyotype review.14 The diagnosis of normal cytogenetics was made based on the analysis of 20 or more metaphases in bone marrow specimens cultured for 24–48 h.
All patients provided written informed consent, and all study protocols were in accordance with the Declaration of Helsinki and approved by Institutional Review Boards at each center.
Analysis of single nucleotide polymorphism rs16754 and other molecular markers
Tissue preparation is detailed in the Online Supplementary Data. Genomic DNA and total RNA extraction and quality control of the extracted nucleic acids were performed as reported elsewhere.15 WT1 exon 7 was amplified from genomic DNA in a polymerase chain reaction, and SNP rs16754 was assessed by direct sequencing. WT1 exon 7 and 9 mutations,2,5 FLT3-ITD,16,17 FLT3-tyrosine kinase domain (FLT3-TKD),18 NPM1,19,20 CEBPA,21 MLL-partial tandem duplication (MLL-PTD),22,23 IDH1,24 IDH224 and TET215 mutations, and mRNA-expression levels of BAALC25–27 and ERG28,29 were assessed centrally as previously reported.
Genome-wide expression analyses
Gene-expression profiling was performed using Affymetrix U133 plus 2.0 arrays (Affymetrix, Santa Clara, CA, USA), as previously reported.15 The microarray data are available at http://www.ebi.ac.uk/microarray-as/ae/ (accession n. E-TABM-1165, E-TABM-1166, and E-TABM-1167). Gene-expression profiles were compared among younger patients (<60 years) for WT1AA (n=64) versus WT1AG (n=30), and among older patients (≥60 years) for WT1AA (n=148) versus WT1AG (n=46), WT1GG (n=6) versus WT1AA, and WT1GG versus WT1AG. A univariable significance level of 0.001 was used to identify differentially expressed gene probe-sets. MicroRNA expression profiling was performed using custom microRNA arrays (OSU_CCC version 3.0 for younger and OSU_CCC version 4.0 for older patients), as previously reported.15,30 MicroRNA-expression profiles were compared among younger patients for WT1AA (n=62) versus WT1AG (n=29), and among older patients for WT1AA (n=134) versus WT1AG (n=43), WT1GG (n=5) versus WT1AA, and WT1GG versus WT1AG. A univariable significance level of 0.005 was used to identify differentially expressed microRNA probes. A global test of significance based on a permutation procedure was performed to determine whether or not the number of differentially expressed gene probe-sets or microRNA probes was more than expected by chance; if not, no signature is reported for the comparison. Unsupervised hierarchical cluster analyses were performed using average linkage and one minus the Pearson’s correlation coefficient between two gene (or microRNA) expression profiles as the distance metric. Analyses were performed using BRB-ArrayTools Version 4.1.0 Beta_2 Release developed by Dr. Richard Simon and the BRB-ArrayTools Development Team.
The clinical endpoints (complete remission, disease-free survival, overall survival) were defined according to published recommendations,31 and are detailed in the Online Supplementary Data. Patients with each of the three rs16754 genotypes (WT1AA, WT1AG or WT1GG) were compared for baseline demographic, clinical and molecular features using Fisher’s exact and Wilcoxon rank sum tests for categorical and continuous variables, respectively. Estimated probabilities of disease-free and overall survival were calculated using the Kaplan-Meier method, and the log-rank test evaluated differences between survival distributions. Analyses of clinical endpoints in the entire study cohort were adjusted for age group (<60 years versus ≥60 years), when comparing WT1GG patients versus a combined group of WT1AG and WT1AA patients, and WT1AA patients versus a combined group of WT1AG and WT1GG patients.
All analyses were performed by the CALGB Statistical Center.
Frequencies and pretreatment features of the rs16754 genotypes
Of the 433 de novo CN-AML patients, 309 (71%) had the genotype WT1AA, 112 (26%) WT1AG, and 12 (3%) WT1GG. The pretreatment clinical and molecular characteristics according to the rs16754 genotypes are presented in Table 1. WT1GG patients were significantly less often Caucasian compared with WT1AA (P=0.008) or WT1AG (P=0.001) patients. The non-Caucasian WT1GG patients comprised one Hispanic patient, one Native American and three Asians. In contrast, there were no Asians among the patients with the more common WT1AA or WT1AG genotype. This is consistent with the higher frequency of the WT1GG genotype reported in Asians.9 WT1GG patients had leukemic skin infiltrates more often than WT1AA patients (P=0.04) and, by trend, more often than WT1AG patients (P=0.06). They also harbored mutations in TET2 more frequently than WT1AG patients (P=0.02) and, as a trend, more frequently than WT1AA patients (P=0.09). In contrast, when considered as one group, mutations in the IDH1 and IDH2 genes tended to be less frequent in WT1GG patients than in WT1AG patients (P=0.055). Compared with WT1AA patients, WT1AG patients tended to harbor mutations more frequently in IDH1 (P=0.06) and WT1 (P=0.08).
Outcome according to the rs16754 genotypes
In the entire study population, we observed no significant differences in outcome among WT1AA, WT1AG and WT1GG patients (Table 2). Considering the prognostic impact of WT1 mutations in our cohort,2,5 we analyzed the outcome associated with the genotypes separately in the WT1 wild-type and WT1-mutated groups of patients. There were too few WT1GG patients with a WT1 mutation, so only the patients with the WT1AA and WT1AG genotypes were compared in the WT1-mutated group. In both the WT1 wild-type and WT1-mutated groups of patients, the outcomes according to the rs16754 genotypes did not differ significantly from each other (Online Supplementary Table S1).
The prognostic impact of rs16754 has been suggested to be more pronounced among CN-AML patients with FLT3-ITD and/or NPM1 wild-type.8 In this molecular subset of our cohort, WT1GG patients had a longer disease-free survival (P=0.04) and a trend to a longer overall survival (P=0.12) than WT1AA patients, and they tended to have longer disease-free survival (P=0.06) and overall survival (P=0.13) than WT1AG patients (Figure 1A,B; Online Supplementary Table S2). Since there was no indication that WT1AA and WT1AG patients differ in their outcomes we combined these patients into one group. Compared with this combined group of WT1AA and WT1AG patients, WT1GG patients had significantly better disease-free survival [P=0.02, HR=3.88 (1.23–12.22)] and overall survival [P=0.04; HR=2.3 (1.03–5.27)] in analyses which were adjusted for age group (<60 years versus ≥60 years) to control for differences in treatment intensity between the protocols for younger and older patients.
The rare occurrence of the WT1GG genotype (n=10 among patients with FLT3-ITD and/or NPM1 wild-type) precluded its evaluation in multivariable models considering established prognostic markers. Of the patients disease-free at 2 years, two had an isolated CEBPA double mutation, which has been previously described to be associated with favorable outcomes in CN-AML.32,33 The remaining three patients disease-free at 2 years had no clear favorable marker constellation besides low BAALC and/or ERG expression (Online Supplementary Table S3).
Outcome comparisons of the WT1AA versus the WT1AG/WT1GG genotypes in different age groups
Previous studies combined the patients having at least one G allele in rs16754 into one group (WT1AG/WT1GG) and compared their outcome with that of WT1AA patients.8,10–12 In age group-adjusted analyses of our entire cohort, we observed no significant differences between WT1AA and WT1AG/WT1GG patients with regard to complete remission rates (P=0.21), disease-free survival (P=0.64) and overall survival (P=0.19).
To allow the comparison of the outcome results in our population of patients with those previously reported for younger adults with CN-AML,8 we performed analyses restricted to younger patients. In contrast to the reported adverse impact of WT1AA,8 we observed that younger patients with WT1AA in our study cohort had higher complete remission rates (P=0.04), similar disease-free survival (P=0.75) and a trend to a longer overall survival (P=0.09) than patients with WT1AG/WT1GG (Online Supplementary Figure S1A,B; Online Supplementary Table S4). However, in multivariable analyses adjusting for other prognostic molecular markers, there were no differences in outcome between WT1AA and WT1AG/WT1GG patients (data not shown). As in a previous study,8 we also assessed the prognostic impact of rs16754 in younger patients with FLT3-ITD and/or NPM1 wild-type. We found no significant outcome differences in this molecular subset (Online Supplementary Table S4).
Next, we compared the outcomes of the WT1AA and WT1AG/WT1GG patients aged 60 years or older. There were no significant differences in outcome between WT1AA and WT1AG/WT1GG patients among these older patients (Online Supplementary Figure 1C,D; Online Supplementary Table S4).
Gene- and microRNA-expression profiling of the rs16754 genotypes
To explore whether the rs16754 genotypes are associated with distinct biological features, we tested whether there was an association of the different genotypes (WT1AA, WT1AG and WT1GG) with distinct gene- and microRNA-expression patterns. Only four of the 12 WT1GG patients were in the younger age group and none of them had material available for gene- or microRNA-expression profiling analyses. Thus, to diminish the impact of potential age-related expression differences that might contribute to the SNP rs16754-associated signatures, particularly those involving the WT1GG genotype, we analyzed the expression profiles for the two age groups separately. Consequently, the analyses involving WT1GG were performed only in the older patients.
Pair-wise comparisons between the patients according to their rs16754 genotypes revealed no significant gene- or microRNA-expression signature associated with the rs16754 genotypes in the younger or older patients. Moreover, in unsupervised cluster analyses of the gene- and microRNA-expression profiles, there were no evident patterns of clustering of the patients according to the rs16754 genotypes (Figure 2).
The conflicting reports concerning the prognostic significance of SNP rs167548,10–13 prompted us to evaluate the clinical impact of this polymorphism in a relatively large cohort of intensively treated de novo CN-AML patients who we had previously comprehensively characterized at the molecular level.15 To gain insights into the biological features of the polymorphism, we also examined the gene-and microRNA-expression profiles according to the rs16754 genotypes.
The frequencies and the race distribution of the rs16754 genotypes in our cohort of CN-AML patients were in accordance with those expected in a normal population (dbSNP Build ID: 131).9 This suggests that none of the rs16754 genotypes is associated with a predisposition to AML. Our results also suggest that the WT1GG genotype, in addition to being more frequent in non-Caucasians, might be associated with other distinct pretreatment characteristics, such as a higher frequency of leukemic skin infiltrates and TET2 mutations.
While we found no significant differences in outcome between WT1AA, WT1AG and WT1GG patients in the entire study cohort, we observed that WT1GG patients had a more favorable outcome than WT1AA or WT1AG patients within the subset of patients with FLT3-ITD and/or NPM1 wild-type. Unfortunately, too few patients had WT1GG so we could not evaluate whether its prognostic impact was independent of other molecular markers in multivariable models. Thus, it is at present uncertain whether the better outcome of these patients can be attributed to the presence of the WT1GG genotype. However, while no differences in outcome between WT1GG and WT1AG patients were found in a pediatric AML cohort,11 a favorable prognostic impact of the WT1GG genotype compared with WT1AG and WT1AA was described in a meeting abstract on a cohort of adult AML patients.13 Unfortunately, neither study included outcome analyses of the WT1GG genotype restricted to CN-AML or to the molecular high-risk subset of patients with FLT3-ITD and/or NPM1 wild-type. We observed a potentially favor-able impact of WT1GG in this subset of patients, and Damm et al.8 also found that the outcome differences according to rs16754 in their CN-AML cohort were more pronounced in this molecular subgroup. The report by Damm et al.8 of WT1AA patients having worse outcomes than patients harboring at least one G allele in rs16754 (WT1AG/WT1GG) also suggests that it is the presence of the G allele that contributes to a prognostically favorable phenotype.
In accordance with previous studies,8,10–12 we also conducted outcome analyses with patients harboring at least one G allele in rs16754 combined into one WT1AG/WT1GG group. Considering the entire study cohort, no significant differences in outcome were observed between WT1AA and WT1AG/WT1GG patients. In the younger age group, we observed higher complete remission rates and longer overall survival of WT1AA patients compared with the WT1AG/WT1GG group, but this favorable impact of WT1AA was not significant in multivariable analyses. The lack of prognostic impact of WT1AA in our study contrasts with the previously reported adverse outcome associated with this genotype in younger adults with CN-AML.8 In preliminary results from a cohort of older (age 50 to 70 years) CN-AML patients, Renneville et al.10 reported that WT1AA patients had a favorable outcome. However, when we restricted our analyses to patients aged 60 years or older, we observed no significant differences in outcome between WT1AA and WT1AG/WT1GG patients.
Similar to what has been suggested for mutations in WT1,7 treatment differences could account for the discrepancies among the studies investigating the impact of WT1 SNP rs16754. While no patient in our cohort received allogeneic stem cell transplantation in first complete remission, approximately 20% of the patients investigated by Damm et al.8 received such consolidation. The significance of rs16754 may also vary among cytogenetic groups, as Ho et al.11 observed in pediatric AML that WT1AG/WT1GG had no prognostic impact in CN-AML, but was associated with a favorable outcome in a subset of patients denoted as having low-risk disease, which comprised children with t(8;21), inv(16) or t(16;16), or those with CEBPA or NPM1 mutations. Moreover, Damm et al.34 reported that adult patients with core-binding factor AML and WT1AG/WT1GG genotype had a trend to a longer overall survival than patients with WT1AA.
Another potential factor influencing response to treatment is race. WT1GG is more frequent among Asians. Although we are not aware of studies demonstrating that adult Asian patients with CN-AML respond to treatment differently from Caucasians and Blacks, there are studies suggesting the existence of differences in treatment outcomes between Caucasians and Blacks with AML.35,36 Thus, the racial composition of study cohorts might affect the impact of rs16754 in different studies.
A few molecular markers, such as FLT3-ITD and mutations in the NPM1 and CEBPA genes,37 are well established as prognostic factors in CN-AML, and further promising candidates, e.g., mutations in the IDH1/IDH224 or TET215 genes, are under investigation. Based on the current data, the role of SNP rs16754 in predicting outcome of CN-AML patients appears to be minor when compared to that of the aforementioned gene mutations. Although in some studies specific SNP rs16754 genotypes have been shown to be associated with outcome, the results are inconsistent.8,10–13 Thus, further molecular studies should clarify the prognostic role of the rs16754 genotypes before testing for SNP rs16754 should be considered for inclusion in the work-up of patients with CN-AML.
SNP rs16754 attracted particular attention because of its localization in a “hot spot” for WT1 mutations in AML, but its biological effects have not been well characterized. A gene set enrichment analysis on nine younger CN-AML patients with the WT1AA and eight with the WT1AG genotype suggested the existence of biological differences according to the polymorphism.8 However, our comparison of gene-expression profiles among relatively large CN-AML cohorts of younger WT1AA (n=64) and WT1AG (n=30) patients and older WT1AA (n=148) and WT1AG (n=46) patients did not identify any signatures of genes differentially expressed between these genotypes. In addition, we compared the gene-expression profiles of the patients with the prognostically favorable WT1GG with those of WT1AA and WT1AG patients, but we did not find a significant gene-expression signature. Likewise, we could not identify any microRNA-expression signature associated with the SNP rs16754 genotypes. Moreover, in hierarchical cluster analyses, which group patients with similar gene- or microRNA-expression profiles, we observed no clustering of patients according to their rs16754 genotype. Our data, therefore, suggest that the rs16754 polymorphism does not lead to robust biological differences among malignant blasts with different genotypes.
In summary, unlike previous observations in younger adults with CN-AML,8 WT1AA patients did not have worse outcomes than patients of the WT1AG/WT1GG group in our CN-AML series. The inconsistent outcome results according to rs16754 among different studies may reflect differences in patients’ characteristics or treatments administered. In our cohort, the WT1GG genotype appeared to be associated with distinct clinical and molecular characteristics and potentially better outcomes compared with the WT1AG or WT1AA genotypes. Because of the relatively low frequency of the WT1GG genotype in our study population, large collaborative studies should be performed to further evaluate whether the rs16754 polymorphism adds prognostic information to established molecular markers in CN-AML. These studies should include both large Asian populations, since the WT1GG genotype is more frequent among Asians, and large populations of Caucasians and Blacks, which would allow evaluation of the clinical significance of WT1 SNP rs16754 within ethnically homogeneous cohorts.
The Cancer and Leukemia Group B institutions, principal investigators, and cytogeneticists participating in this study are listed in the Appendix. The authors would like to thank Donna Bucci of the Cancer and Leukemia Group B Leukemia Tissue Bank at The Ohio State University Comprehensive Cancer Center, Columbus, OH, USA for sample processing and storage services, Lisa J. Sterling and Colin G. Edwards, PhD for data management, and Deedra Nicolet for statistical analyses.
GM and CDB contributed equally to this work.
Funding: this work was supported in part by National Cancer Institute, Bethesda, MD grants CA101140, CA114725, CA140158, CA31946, CA33601, CA16058, CA77658 and CA129657, The Coleman Leukemia Research Foundation and the Deutsche Krebshilfe – Dr. Mildred Scheel Cancer Foundation (Heiko Becker).
The online version of this article has a Supplementary Appendix.
Authorship and Disclosures
The information provided by the authors about contributions from persons listed as authors and in acknowledgments is available with the full text of this paper at www.haematologica.org.
- Received February 7, 2011.
- Revision received May 12, 2011.
- Accepted June 6, 2011.
- Copyright© Ferrata Storti Foundation