Molecular alterations of isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2) metabolic genes and additional genetic mutations in newly diagnosed acute myeloid leukemia patients

Background Isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2) metabolic genes encode cytosolic and mitochondrial enzymes that catalyze the conversion of isocitrate to α-ketoglutarate. Acquired somatic mutations of IDH1 and IDH2 have recently been reported in some types of brain tumors and a small proportion of acute myeloid leukemia (AML) cases. Methods Two-hundred and thirty newly diagnosed AML patients were analyzed for the presence of IDH1 and IDH2 heterozygous mutations by polymerase chain reaction-denaturing high performance liquid chromatography (PCR-DHPLC) followed by direct sequencing. Clinical and biological characteristics were analyzed and correlated to the IDH mutational status. Coexisting mutations such as FLT3, PML-RARA, RAS, AML1, and NPM1 mutations were additionally explored. Results The prevalence of IDH1 and IDH2 mutations was 8.7% (20/230) and 10.4% (24/230), respectively. Six missense mutations were identified among IDH1-mutated cases; p.R132H (n = 8), p.R132C (n = 6), p.R132S (n = 2), p.R132G (n = 2), p.R132L (n = 1), and p.I99M (n = 1). Two missense mutations were found in IDH2-mutated cases; p.R140Q (n = 20) and p.R172K (n = 4). No patients had dual IDH1 and IDH2 mutations. About 18% of AML with normal cytogenetics and 31% of acute promyelocytic leukemia had IDH mutations. Half of the IDH-mutated cohort had normal karyotype and the major FAB subtype was AML-M2. Interestingly, IDH1- and IDH2-mutated cases predominantly had NPM1 mutations (60-74%) as compared to the wild type (P < 0.001). Very few IDH-mutated cases had FLT3 and/or RAS abnormalities and none of them had AML1 mutations. Older age and higher median platelet counts were significantly associated with IDH2 mutations although the clinical impact of either IDH1 or IDH2 mutations on patients' overall survival could not be observed. Conclusion Overall, 19% of newly diagnosed AML patients had alterations of IDH genes. No patients concurrently carried both IDH1 and IDH2 mutations suggesting that these mutations were mutually exclusive. NPM1 mutation appears as a major coexisting genetic mutation in IDH-mutated patients. Our present data failed to support the prognostic relevance of IDH mutations although alterations of these metabolic genes potentially have an important role in leukemia development.


Background
Acute myeloid leukemia (AML) is a malignant hematologic disorder characterized by abnormal expansion of differentiation-defective myeloid cells [1]. Various chromosomal aberrations have been identified in AML patients and are uniquely associated with distinct clinical entities and prognostic relevance [2]. Although 40-50% of AML cases do not carry any detectable chromosomal abnormalities, a fraction of them are found to have mutations of genes that normally function in cell proliferation, differentiation, and survival such as FLT3, NPM1, RAS, WT1, and AML1 [3,4]. Moreover, through a rapid whole genome sequencing approach, it is now evident that at least half of the AML cases with normal karyotype have readily identifiable genomic abnormalities [5].
In the present study, we aimed to characterize IDH1 and IDH2 mutations in newly diagnosed AML patients and investigate their correlations to other parameters such as clinical and hematologic characteristics, cytogenetics and additional genetic mutations.

Leukemia samples
Leukemic samples from 230 newly diagnosed AML cases were consecutively recruited into the study. Clinical and biological characteristics were collected including clinical history, complete blood counts, peripheral blood (PB) smear, bone marrow (BM) studies, flow cytometric immunophenotyping, and chromosome analysis. Mononuclear cells (MNC) were isolated from the leukemic samples by Ficoll-Hypaque density-gradient centrifugation and subsequently used for molecular analysis. Twenty consented normal individuals were used as controls. Patients were treated according to the standard AML regimen which included idarubicin and cytarabine induction therapy followed by high-dose cytarabinebased consolidation phase. This study was approved by the Ethical Committee for Human Research, Faculty of Medicine Siriraj Hospital, Mahidol University.

Mutational analysis of IDH1 and IDH2
Genomic DNA was extracted using standard phenolchloroform method or Gentra Puregene Blood Kit (Qiagen, Hidden, Germany) according to the manufacturer's protocol. DNA amplicons harbouring exon 4 of IDH1 and IDH2 were amplified by polymerase chain reaction (PCR) using the primer pair; IDHIf (5'-AGCTC TATATGCCATCACTGC-3'), IDH1r (5'-AACATG-CAAAATCACATTATTGCC-3'), IDH2f(5'-AATTTTA GGACCCCCGTCTG-3'), and IDH2r (5'-CTGCAGA-GACAAGAGGATGG-3') [13]. PCR reactions were performed in a total volume of 20 μL containing 50 ng of genomic DNA, PCR master mixture consisting of 1x Phusion ® HF Buffer (F-520), 200 μM dNTPs, 0.5 μM of each primer, 0.02 U/μL Phusion ® DNA polymerase, and Milli-Q water. The PCR was carried out in a Perkins Elmer PCR2400 thermal cycler (Applied Biosystems, Foster City, CA) using the following steps: initial denaturation at 98°C for 30 seconds (sec), 35 cycles at 98°C for 10 sec, 60°C for 30 sec. and 72°C for 30 sec, and final extension at 72°C for 5 minutes (min). Both amplicons were screened for heterozygous mutations by denaturing high-performance liquid chromatography (DHPLC) on a WAVE 3500HT with DNASep ® HT cartridge technology (Transgenomic Inc, Omaha, NE, USA). The optimized condition and temperature were predicted by the Navigator™ software to determine chromatographic peak pattern. PCR crude sample was injected into DHPLC column and the optimal temperature for IDH1 was 58.5°C and IDH2 was 64°C. Each DHPLC chromatogram was compared to a wild-type reference. The sensitivity of our assay was determined by performing a dilution series containing a different percentage (%) of mutant and wild-type IDH concentrations. Abnormal DHPLC peaks could be clearly detected in 50%, 20%, 10%, 5%, and 3.33% dilutions. The mutational chromatograms were re-amplified in an independent PCR reaction and further subjected to direct sequencing. The sequences were compared to the wildtype IDH1 and IDH2 cDNA (GenBank Accession number, NM_005896.2 and NM_002168.2, respectively) [25].

Statistical analysis
The relationship between IDH mutations and various patient characteristics such as age, Hb count, WBC count, platelet count, and percentages of blasts was determined by the student t-test, equal variances not assumed for continuous variables. Categorical variables such as FAB classification, cytogenetics, and test. The Kaplan-Meier method and the log-rank test were utilized to estimate the distribution of OS [40]. For all analyses, a p-value of less than 0.05 was considered statistically significant. All reported p-values were 2-sided.
Clinical parameters and morphologic subtypes of patients with IDH1 or IDH2 mutations Twenty IDH1-mutated patients showed no significant differences in age, hemoglobin level, WBC counts, platelet counts or percentages of blasts as compared to the IDH1 wild-type group, although a trend towards more females was observed (15 females vs 5 males; P = .058). Twentyfour IDH2-mutated patients also showed no significant differences in sex, WBC counts or percentages of blasts but an older age (49.5-year vs. 43-year; P = .001), a higher platelet count (59 vs. 45 × 10 9 /L; P = .048), and a trend towards higher hemoglobin level (9.25 g/dL vs. 7.7 g/dL; P = .059) were observed in the IDH2-mutated group as compared to the wild-type group (Tables 2 and 3). Among IDH1-mutated cases, the majority was classified as AML with maturation (AML M2) (13/20 cases, 65%) followed by APL (5/20 cases, 25%). None of the cases with APL, acute monoblastic or monocytic leukemia, acute erythroid leukemia, and acute megakaryoblastic leukemia was found among IDH1-mutated group. Similarly, AML with maturation (AML M2) was the most common subtype among the IDH2-mutated group (11/24 cases, 46%) ( Table 3). The second most common subtypes were APL (6/24 cases, 25%) and acute myelomonocytic leukemia (6/24 cases, 25%). The OS of patients with or without IDH1 or IDH2 mutations did not differ among the entire series of AML patients and AML patients with normal cytogenetics (P = 0.200 and 0.272, respectively). No significant difference was demonstrated in subgroups of younger patients (age < 60) and patients with/without additional mutations as compared to the wild-type patients (P = 0.471 and 0.812, respectively).

Discussion
In the present study, we developed a screening DHPLC method followed by sequencing analysis to detect and confirm the presence of IDH mutations in newly diagnosed AML patients. Twenty cases of IDH1 mutations and 24 cases of IDH2 mutations were discovered among the entire newly diagnosed AML cohort. Previous reports from the Asia continent were available from two countries, i.e. Taiwan [18] and China [28,31,32] while the Western studies were from USA [5,15,20,22,29], Canada [19], France [41], Germany [21,23,25,27], the Netherlands [24], and UK [26,30] (Table 5). The overall frequency of IDH mutations appears to vary between 2-14% for IDH1 and 1-19% for IDH2 from most Western reports [5,15,19,27,29,30,41]. Worthy of note, the frequency of IDH1 mutations in our population of 8.4% was comparable to 8.5% in the first study reported by Mardis et al. in 2009 [5] although these figures were somewhat higher than those of the Chinese AML studies (5.5%, 5.6%, 6.3%, and 3.6%) [18,28,31,32]. The frequency of IDH2 mutations of 10.4% in our cases was also slightly higher than the only available IDH2 study from Asia (8.3%, 4/48) [28]. The frequency discrepancies among various studies may reflect the variable inclusion criteria of the study samples, the variable sensitivity of the detection assays, the selective inclusion or exclusion of certain IDH aberrations or the true racial differences.
IDH1 mutations consisting of six different amino acid exchanges at p.R132 (n = 19) and p.I99M (n = 1) were identified. Within the p.R132 group, arginine was replaced by histidine (R132H) in most cases (n = 8, 40%), followed by cysteine (R132C; n = 6, 30%), serine (R132S; n = 2, 10%), glycine (R132G; n = 2, 10%) and leucine (R132L, n = 1, 5%). This pattern was extremely different to the mutation pattern reported in glioma, where R132H was predominant observed in 88% of all cases while R132C present in only 4.5% [13]. To date, results from structural and functional assays by several multicenter trials suggested that IDH1 R132, which resides at the active site of enzyme substrate affinity, promotes oncogenesis in both glioma and AML [9,11,20,33]. In the p.I99M case, isoleucine was substituted by methionine which was recently identified as a novel missense mutation in the Chinese cohort by Zou et al [28]. The same study revealed that this evolutionary point mutation was also located in the substrate binding site of enzyme and may drive pathogenesis; however, the exact mechanism needs further investigation. In addition, we detected one silent polymorphism (IDH1 G105G ) in 3 cases (1.3%). Wagner et al. [21] previously reported that IDH1 G105G allele conferred an adverse prognostic impact to patients' survival. The identified IDH2 mutations involved two different types of amino acid substitution spanning exon 4 of the IDH2 gene at arginine 140 and arginine 172. Of note, the former arginine was replaced by glutamine (R140Q; n = 20, 83.3%) and the latter arginine was replaced by lysine (R172K; n = 4, 16.7%). Our study was similar to previous studies which revealed that more than 80% of the IDH2 mutations involved R140 [15]. R172 mutations were profoundly associated with biological insights and clinical outcome [15,20] while R140 has not been addressed to associate with any prognostic significance in AML [23]. Therefore, functional validation should be employed to define whether R140 plays a significant role in AML pathogenesis or is simply a genuine polymorphism.
To explore if other genetic mutations coexist in AML cases with IDH mutations, we performed mutation analysis of various different genes, i.e. FLT3, NPM1, NRAS and AML1. IDH1 mutations were found to be most frequently accompanied by NPM1 mutations (74% of the cases; P < 0.001). Previous studies also demonstrated that IDH1 mutation was significantly associated with NPM1 mutation, ranging from 12.5% to 67% as compared to the wild-type IDH1 cases [5,18,24,25,27,41]. Similarly, IDH2 mutations were significantly associated with NPM1 mutations (60% of the cases; P < 0.001) which were comparable to other reports [24][25][26]. No significant association was found with other molecular alterations including FLT3-ITD, FLT3-TKD, NRAS and AML1 although FLT3-ITD was also frequently found co-existing with IDH1 mutation in some studies [15,25]. Meanwhile, other authors also showed no significant correlation between either IDH1 or IDH2 mutation and FLT3-TKD, NRAS and AML1 mutation [5,18,24,27,30,41]. With respect to clinical and hematologic parameters, IDH1 mutated cases were frequently females rather than males (15 cases vs. 5 cases) which was similar to the German study by Schnittger et al. [27]. Interestingly, we observed that both IDH1 and IDH2 mutations were predominantly found in AML with maturation (AML-M2; n = 24/44) and acute promyelocytic leukemia (APL) (AML-M3; n = 11/44) which were different from AML-M1 as reported by others [5,27,30]. Interestingly, the frequency of IDH2 mutation coexisting in AML-M4 of 25% in our study was comparable with 27% in the finding reported by Thol et al. [23]. IDH1 or IDH2 mutations did not significantly impact survivals when the whole AML cohort or AML with normal karyotype analyzed (P = 0.200 and 0.272). We therefore further analyzed OS according to age and NPM1 status. Unfortunately, we could not find a significant difference between IDH1-and IDH2-mutated and wild-type cases (P = 0.471 and .812) either in the younger age group (< 60 years) or the NPM1-mutated genotype. Our study was consistent with some studies that revealed no impact of IDH mutations on the OS of AML cases although other studies suggested that IDH1 or IDH2 mutations conferred an adverse effect among AML with normal karyotype or AML with favorable genotype (NPM1 mutated/FLT3 wild type) [15,[25][26][27]41]. IDH1  mutation conferred a shorter disease-free survival and IDH2 R172 mutation contributed to a lower complete remission or a higher relapse risk compared to wild-type IDH patients [25,41]. Our study may be limited by a small number of cases with IDH alterations and a substantial recruitment of cases with aberrant karyotype [18,23]. The possible oncogenic role of IDH mutations that contribute to AML development has been postulated by available evidence [20,33,34]. By structural and functional analysis, IDH1 and IDH2 mutated cells gained the neomorphic enzymatic activity creating a condition with 2HG oncometabolite accumulation which promotes tumorigenesis through inhibiting a cancer-associated transcription factor such as hypoxia-induced factor (HIF) [19,20,34]. Moreover, inhibition of normal myeloid differentiation and induction of global DNA hypermethylation by mutated IDH potentially lead to leukemogenesis [33], suggesting that IDH genes and their altered enzymatic pathways may be a potential new target for future drug development for AML patients. Intriguingly, IDH1 and IDH2 mutations were also found in other myeloid disorders such as myeloproliferative neoplasms (MPN) and myelodysplastic syndrome (MDS) which have a propensity to AML development, although at a much lower frequency than AML [42,43]. It was thus speculated that IDH mutations were likely to be associated with disease transformation or progression rather than disease initiation [44][45][46].
In conclusion, IDH1 and IDH2 mutations occur in a minor subset of newly diagnosed AML patients with a strong association with normal karyotype, AML-M2 subtype, and NPM1 mutation. No significant correlation with other mutations such as FLT3, RAS, and AML1 could be demonstrated. Larger studies are needed to confirm the prognostic impact of IDH1 and IDH2 mutations in AML patients from various ethnic backgrounds. Our results, nevertheless, provide a relevant rationale to utilize these genomic alterations to better characterize AML patients in the future.