Mutation Patterns of 16 Genes in Primary and Secondary Acute Myeloid Leukemia (AML) with Normal Cytogenetics

Acute myeloid leukemia patients with normal cytogenetics (CN-AML) account for almost half of AML cases. We aimed to study the frequency and relationship of a wide range of genes previously reported as mutated in AML (ASXL1, NPM1, FLT3, TET2, IDH1/2, RUNX1, DNMT3A, NRAS, JAK2, WT1, CBL, SF3B1, TP53, KRAS and MPL) in a series of 84 CN-AML cases. The most frequently mutated genes in primary cases were NPM1 (60.8%) and FLT3 (50.0%), and in secondary cases ASXL1 (48.5%) and TET2 (30.3%). We showed that 85% of CN-AML patients have mutations in at least one of ASXL1, NPM1, FLT3, TET2, IDH1/2 and/or RUNX1. Serial samples from 19 MDS/CMML cases that progressed to AML were analyzed for ASXL1/TET2/IDH1/2 mutations; seventeen cases presented mutations of at least one of these genes. However, there was no consistent pattern in mutation acquisition during disease progression. This report concerns the analysis of the largest number of gene mutations in CN-AML studied to date, and provides insight into the mutational profile of CN-AML.


Introduction
Acute myeloid leukemia (AML) is a heterogeneous disease in terms of karyotype and molecular abnormalities.The discovery of the classic karyotype abnormalities in AML such as the t(15;17) has been invaluable in enabling more accurate prognostic estimates, the development of specific therapies and the molecular monitoring of disease.However, approximately half of AML patients have no karyotype abnormality (CN-AML).This group of AML cases is presumably heterogeneous in all respects, and molecular monitoring is not possible unless there is an associated mutation.Recently it has been demonstrated that mutations of FLT3, NPM1 and CEBPA genes are preferentially found in CN-AML.[1] Nevertheless many cases do not possess such mutations and this imposes a severe limitation in understanding their specific pathophysiology and monitoring disease progression.We have chosen to study CN-AML with the aim of finding a restricted panel of genes which are mutated in the majority of cases.In a series of 84 CN-AML patients, we examined 16 genes with mutations that had previously been described in cases of CN-AML (Table S1).[2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17] The characterisation of cases by the presence or absence of mutations in these selected genes should allow a molecular dissection of cases of CN-AML into different biological and prognostic groups, as well as achieving the long sought after goal of molecular monitoring of CN-AML.

Patients
A total of 84 AML patients (mean age 64, range 16 to 86, 23 patients under 60; 52 male, 32 female) with no cytogenetic abnormalities were recruited for mutational analysis, including 51 primary cases (mean age 60, range 16 to 86, 20 patients under 60; 27 male, 24 female) and 33 cases secondary to either MDS (n = 24) or CMML (n = 9) (mean age 70, range 51 to 81, 3 patients under 60; 25 male, 8 female).The karyotype was investigated again at the time of transformation in 31 of the 33 secondary cases, and found to be normal.An additional 100 cases were investigated for mutations in ASXL1 from AML patients showing different karyotypic abnormalities.Some of the cases included in the present study (16 CN-AML and 51 cases with aberrant cytogenetics) have been previously analyzed for ASXL1 exon 12 mutations, and results reported elsewhere.[18]
An analysis of the mutations occurring in more than 10% of cases revealed statistically significant associations (Figure 1, Table S3).In agreement with previous reports, FLT3 and DNMT3A mutations were significantly associated with NPM1 mutations, [12] whereas patients with ASXL1 mutations had significantly lower incidence of NPM1 and DNMT3A mutations.[8,9] IDH1 and IDH2 mutations were mutually exclusive.With the exception of one patient, no cases with IDH1/2 mutation also had a TET2 mutation.IDH1 and IDH2 mutations were less frequent in TET2mutated than in TET2-wt patients, and this has been reported before.[19,20] Concurrence of IDH1/2 and ASXL1 mutations was also a relatively infrequent event in our patient cohort (Figure 1).This observation is in agreement with a report on a series of 63 AML secondary to MPN cases.[3]  ASXL1 mutations were significantly more frequent in secondary AML compared to de novo AML cases (primary cases: 2/51, 3.9%; secondary to MDS/CMML: 16/33, 48.5%, p,0.0001).We have previously reported a high prevalence of ASXL1 mutations in advanced MDS. [18] NPM1, FLT3, and DNMT3A mutations were significantly more common in primary CN-AML than in secondary AML cases (Table 1).NRAS, JAK2, SF3B1 and TP53 mutations were exclusively present in secondary AML samples (Table 1).Only 9.5% of the samples analyzed (8/84, 6 de novo and 2 post-MDS cases) showed no mutation in any of the genes tested.When considering only the ASXL1, NPM1, FLT3, TET2, IDH1/2 and RUNX1 gene analysis, 88% of de novo CN-AML included in this series presented at least one molecular marker.For secondary cases, 85% of patients carried mutations in at least one of these 7 genes.
Recent reports showed that DNMT3A mutations are associated with a poor outcome in AML, [21,22] and that the location of the mutations could have an impact in age-related risk classification.[23] It is worth noting that in our series, DNMT3A was not found as a sole mutation suggesting that additional aberrations are needed to sustain leukemogenic development.
Approximately 70% of CN-AML cases secondary to either MDS or CMML presented mutations in at least one of ASXL1, TET2, IDH1 or IDH2 genes.Therefore, we chose to assess the presence and chronology of ASXL1, TET2 and IDH1/2 mutational events, in order to investigate whether they could have a role in disease development or evolution.We studied 15 MDS and 4 CMML cases that progressed to AML, for which at least two samples at different time-points were available.Remarkably, with the exception of two patients all of them possessed at least one gene mutation at some stage of the disease.The majority showed the same mutations at early and later stages of the disease, except one patient who developed an IDH1 mutation at transformation, a second patient with a TET2 mutation who acquired an additional ASXL1 mutation at transformation, and another patient who developed a nonsense mutation of ASXL1 at AML stage, and showed rapid disease evolution (Table 2).On the basis of this study we therefore did not find any consistent patterns in mutation acquisition.A sensitive mutation analysis (such as allele-specific PCR) at early stages of AML in future studies could help clarify whether the mutations found in cases from later stage AML were already present as a minor pre-existing clone at the earlier stage, and if so, how it evolved as the disease progressed to AML.
In order to investigate whether the observed low incidence of ASXL1 mutations is a specific characteristic of karyotypically normal de novo cases, or is a common feature of other subtypes of primary AML, we screened an additional cohort of 100 primary AML, including the most common karyotypic subgroups.Overall, only 8 out of 100 cases showed nonsense or frameshift mutations (8%) (Table S4), confirming that ASXL1 mutations are less common in primary AML than in secondary AML.
This report concerns the analysis of the largest number of gene mutations in CN-AML studied to date.Our results show that 85% of CN-AML patients have mutations in one or more of 7 selected genes (ASXL1, NPM1, FLT3, TET2, IDH1/2 and RUNX1).This finding will facilitate further analysis of this important group of patients by enabling CN-AML patients to be subdivided into groups with common mutation patterns.Detailed studies of the CN-AML subgroups in regard to their hematological features, prognosis, disease progression and treatment response will now be facilitated.
All karyotypes were analyzed by conventional G-banding in at least 30 metaphases.Samples showing inv(16), t(8;21) or t(15;17) at karyotype were subjected to confirmation by molecular techniques.This study was approved by the ethics committees of the institutes involved: the John Radcliffe Hospital (Oxford 06/ Q1606/110), the Royal Bournemouth Hospital (Bournemouth 9991/03/E) and the University of Navarre (Pamplona IRB00006933); written informed consent was received from all patients.

Figure 1 .
Figure 1.Concurrence of mutations in 16 genes analyzed in CN-AML samples.Columns show results for each of the 84 analysed cases.Solid boxes indicate mutated cases.Grey boxes mark unavailable data.FLT3-ITD mutations are indicated with top-half solid boxes and FLT3-TKD with bottom-half solid boxes.Similarly, IDH2-R140Q mutations are shown with top-half solid boxes and IDH2-R172K with bottom-half solid boxes.doi:10.1371/journal.pone.0042334.g001

Table 1 .
Frequency of mutations in normal karyotype AML samples.

Table 2 .
Mutational analysis of serial samples from patients in transformation.