Molecular genetics of acute myeloid leukemia: clinical implications and opportunities for integrating genomics into clinical practice

Abstract

Advances in sequencing technologies have led to the discovery of a series of mutations in a sizeable proportion of patients with acute myeloid leukemia (AML) over the last 10 years. Clinical correlative studies are now beginning to decipher the clinical importance, prevalence and potential prognostic significance of these mutations in AML but few studies have assessed the clinical implications of these mutations in a comprehensive fashion. Nonetheless, mutations in DNMT3A, TET2, and ASXL1 are emerging as important adverse prognosticators in subsets of patients with AML independent of FLT3 mutations whereas mutations in IDH2 at residue 140 are potential predictors of improved outcome in AML. Further improvements in cost, throughput, and clinical validation of second-generation sequencing technologies may allow for clinical implementation of comprehensive genetic profiling in the clinical care of AML patients.

Keywords:

• AML
• TET2
• ASXL1
• IDH1/2
• DNMT3a

Introduction

Advances in the molecular characterization of myeloid malignancies, including acute myeloid leukemia (AML), has resulted in the discovery of a series of recurrent genetic abnormalities found in patients with AML. Prior to 2009, this included the discovery of gain-of-function alterations in FLT3, MLL, c-KIT, the RAS family of oncoproteins, as well as loss-of-function/dominant-negative alterations in NPM1, CEBPa, and TP53. Findings from concurrent clinical correlative analyses have resulted in the routine clinical application of molecular testing for mutations in FLT3, NPM1, and CEBPA for improved risk stratification of AML patients.2,3 Since 2009, however, an even larger number of recurrent molecular alterations have been discovered in a sizeable proportion of patients with AML, including mutations in TET2, IDH1/2, ASXL1, DNMT3A, and PHF6. In addition to furthering our understanding of the molecular basis for AML, discovery of these mutations may be clinically relevant as many of these mutations appear to hold prognostic importance which may aid in risk stratification and/or therapeutic decision-making. Additionally, mutations in several of these genes may specifically impact leukemic cells in such a manner that is therapeutically targetable.

The widening spectrum of clinically relevant molecular alterations in AML presents the potential for applying genomic technologies to clinical care of AML patients. In addition, rapid developments in new technologies for genomic analysis may allow for comprehensive genetic analysis of AML patients in real-time clinical practice. Here we discuss the relevance of recently discovered molecular genetic alterations in AML in the context of currently clinically utilized risk stratification of AML patients as well as the potential for implementation of detailed molecular genetic approaches to routine practice.

Molecular Genetics in Current Clinical Practice in AML: Cytogenetics and Mutations in FLT3, c-KIT, NPM1, and CEBPA

Currently, the maximal genetic characterization of AML patients performed in routine clinical care of AML patients consists of determination of gross structural chromosomal abnormalities in leukemic cells by karyotype/FISH and testing for the presence of internal tandem duplications in FLT3 (FLT3-ITD), tyrosine kinase domain mutations in FLT3 (FLT3 D835A), mutations in the extracellular (exon 8) or PTK2 domain (D816 mutations) of C-Kit and mutations in NPM1 and CEBPA. The karyotype of leukemic cells remains the strongest established predictive factor for response to induction therapy and survival and patients are roughly subdivided into one of three prognostic categories with favorable, intermediate, or adverse outcome based on their cytogenetic findings.4,5,6 Within each of these cytogenetic groupings of AML patients, further molecular analysis identifies additional subsets of patients with a heterogeneity of clinical outcomes based on the presence of additional mutations.

Currently, molecular genotyping for improved risk stratification within cytogenetic groups is based on the use of FLT3, NPM1, and CEBPA mutational testing in patients with intermediate-risk cytogenetics and use of KIT mutational testing in patients with the otherwise prognostically favorable core-binding factor translocations (topics reviewed extensively elsewhere).4 As mentioned earlier, however, a number of additional molecular genetic events have been more recently described to hold prognostic significance in AML patients overall as well as in cytogenetically-defined intermediate-risk AML patients.

New Mutations with Potential Prognostic Importance in AML Patients: Mutations in DNMT3A, IDH1, IDH2, TET2, ASXL1, and PHF6

Mutations in DNMT3A were initially discovered in 2010 based on whole genome sequencing of AML patients as well as array-based sequencing platforms. Since then, it has become clear that mutations in DNMT3A are present in 20–25% of AML patients, making mutations in DNMT3A the second most commonly mutated gene in AML patients overall (following mutations in FLT3). The initial correlative studies of DNMT3A mutations in AML patients strongly suggested that DNMT3A mutations confer an adverse prognosis in AML patients overall, independent of FLT3 mutational status. Ley et al.7 found that DNMT3A mutant patients had a median survival of 12·3 months compared with 41·1months in DNMT3A-wildtype counterparts (P<0·0001). Yan et al.8 specifically studied the effect of DNMT3A mutations on the outcome of 91 patients with M5 AML and found an abysmal effect of DNMT3A mutations on overall survival (OS) and time-to-treatment failure such that DNMT3A-mutant patients had a median survival of 7 months compared with 19·5 months in DNMT3A-wildtype patients (P = 0·004). Looking specifically at patients with cytogenetically-normal AML, both Ley et al.7 and Thol et al.9 found that DNMT3A mutations predicted shorter OS and lower complete response (CR) rate in AML, independent of FLT3 mutational status.

As with mutations in DNMT3A, mutations in isocitrate dehydrogenase 1 and 2 (IDH1/2) were also discovered by whole genome sequencing of AML patients. In contrast to mutations in DNMT3A, however, mutations in IDH2 at the R140 codon appear to confer improved outcome in AML patients overall.¹⁰10,11 All discovered IDH mutations reside in the active site of the enzyme and participate in isocitrate binding. They are missense alterations affecting arginine 132 (R132) in IDH1, and either the analogous arginine residue (R172), or the residue at arginine 140 (R140), in the IDH2 protein. Although each of these mutations is mutually exclusive of one another and all have been shown to result in production of the metabolite 2-hydroxyglutarate, it appears that each of these mutations has differing prognostic impact in AML and studies which cluster the various IDH1/2 alleles together may obscure the prognostic effect of IDH1/2 mutations in AML. As such, the largest studies correlating IDH1/2 alleles separately with outcome in AML have come from analysis of more than 1000 patients treated on two United Kingdom Medical Research Council Trials who were also tested for FLT3ITD/TKD, NPM1, and CEBPA mutations.10,12 In both studies, patients with either IDH1 or IDH2 mutations were significantly enriched with NPM1 mutations. Amongst the entire cohort of patients, they found that those patients with an IDH2R140Q mutation had an improved OS and decreased response rate (RR) compared with all other patients. This finding was even more striking amongst the subset of FLT3 wildtype/NPM1 mutant patients who had survival similar to the most favorable subsets of patients. In contrast, IDH2R172 mutations had a neutral effect on outcome and response to therapy while IDH1R132 mutations seemed to impart worsened outcome on FLT3 wildtype patients. The latter finding of an adverse effect of IDH1R132 mutations on FLT3 wildtype subsets of AML patients have also been reported by Paschka et al. and Abbas et al. as well.

Comprehensive genetic analysis of AML patients has revealed that mutations in TET2 and IDH1/2 are mutually exclusive in AML patients.13 This finding identified a novel complementation group and served as the basis for understanding some of the biological effects of IDH1/2 mutations in leukemia development. Although mutations in IDH1/2 may have differing effects on prognosis depending on the mutated allele, the strongest data on TET2 mutations in AML suggests that TET2 mutations are important adverse predictors of prognosis in subsets of AML patients with normal cytogenetics without the FLT3-ITD. This data comes from (1) a study by Chou et al.14 of 486 de novo AML patients treated with standard induction chemotherapy where TET2 mutations were clearly associated with shorter OS only in the subset of 171 CN AML patients with intermediate-risk cytogenetics and NPM1 wildtype/TET2 mutant genotype and (2) a study by Metzeler et al.15 of 427 CN AML patients who received cytarabine/daunorubicin-based first line therapy which found that TET2 mutations adversely affected survival in patients within the European Leukemia Net (ELN) favorable-risk category of CN AML (CN AML patients with mutated NPM1 and/or CEBPA without FLT3ITD mutations).

Shortly after the discovery of mutations in TET2, mutations in the Addition of Sex Combs Like 1 were discovered using similar SNP-array based studies to identify regions of microscropic deletion in the genome of patients with myeloid malignancies.16 Although, there is some controversy of whether a repeatedly reported variant in ASXL1 is a bona fide somatic mutation,17 it appears that ASXL1 mutations represent important markers of adverse overall survival in patients with myelodysplasia and AML. By studying samples from 398 AML patients in an Eastern Cooperative Group (ECOG) E1900 trial, we found that mutations in ASXL1, although rare in AML patients less than age 60, were associated with worsened overall survival in the overall cohort of AML patients as well as in the subset of cytogenetically-defined intermediate risk AML patients.11 More recently, Metzeler et al.18 identified that ASXL1 mutations are associated with an unfavorable CR rate (P = 0·03), disease-free survival (P<0·001), OS (P<0·001) and event-free survival (P<0·001) amongst ELN Favorable patients. Also, Pratcorona et al.19 studied the impact of ASXL1 mutations on 882 AML patients treated on a number of different HOVON protocols. Similar to the results from the CALGB study, this analysis likewise revealed that ASXL1 mutations are associated with worsened survival (median OS 15·9 vs 22·3 months; P = 0·019) and significantly lower CR rate (61 vs 79·6%; P = 0·004). The studies from CALGB and HOVON also found a significantly higher-rate of ASXL1 mutations in AML patients beyond the age of 60 years old.

In addition to noting that mutations in ASXL1 impart worsened overall survival in the entire subset of AML patients in the ECOG E1900 trial, mutations in plant homeodomain finger 6 (PHF6) were also identified to be associated with worsened overall survival amongst AML patients overall.11 PHF6 mutations were initially identified in ∼20% of patients with T-cell acute lymphoblastic lymphoma20 and have more recently been identified in 3–5% of de novo AML patients.21 Although not as common in AML as mutations in the aforementioned genes, the correlation with worsened OS in overall and CN AML patients from a uniformly treated patient cohort suggests that PHF6 mutations should be studied further in AML.

Conclusion: the Potential for Comprehensive Genetic Characterization of AML Patients in Clinical Practice

The identification of the currently known 5–10 recurrent molecular genetic alterations in AML patients with prognostic importance presents a significant challenge for implementing testing of all of these genetic alterations into clinical practice. Current genetic testing of AML clinical patient samples relies on characterization of metaphase karyotype, FISH, restriction enzyme digestion of PCR products, and capillary sequencing. However, these currently clinically utilized technologies will be inadequate for comprehensive genetic characterization of AML patients in the future and conventional Sanger sequencing will be overly costly and unwieldy for these purposes as well. Moreover, it is expected that genetic discovery efforts will continue to uncover additional clinically important genetic alterations in AML in the near future. Mass-spectrometric genotyping as well as high-resoluting melting genotyping have emerged as cost-effective and rapid means of genotyping patient samples for individual recurrent mutations at specific amino acid residues (e.g. mutations in N/K-Ras, IDH1, IDH2). However, these methodologies cannot be used to identify the full catalogue of mutations which might occur throughout the open-reading frame of a gene (e.g. mutations in TP53, TET2, ASXL1). Thus, it appears that so-called ‘second-generation’ sequencing technologies (e.g. Illumina and SOLiD) and/or the use of array-based sequencing platforms (the Roche NimbleGen and Agilent Capture Array being two examples) may be the best candidates for initiating comprehensive genetic profiling of patient samples in clinical practice. This will allow for detection of somatic mutations, structural rearrangements, and copy-number changes simultaneously. The current limitations which prevent implementation of such sequencing in clinical practice include the high cost, slow turnaround time, and lack of clinical validation. Efforts to limit the sequencing to panels of candidate target genes may improve all of these limitations however. Array-based sequencing using hybrid capture technology may be time-saving compared with other next-generation sequencing approaches and is cost-effective compared to PCR based methods. This technique involves hybridizing shotgun libraries of genomic DNA to target-specific sequences on a microarray. However, this method is limited by the need for expensive hardware, bioinformatic analysis, and a relatively large amount of DNA.

Despite the difficulties of implementing comprehensive genetic profiling of molecular alterations in the clinical care of AML patients, it is clear that a more detailed genetic profiling of AML patients may be useful in improving risk stratification and possibly in providing therapeutic information and disease monitoring in the future.