Acute myeloid leukemia (AML) is characterized by uncontrolled proliferation of malignant myeloid progenitor cells in the bone marrow that are arrested in differentiation. In AML, genetic aberrations often involve the same genes and play an important role in risk assessment and treatment of AML. In the WHO classification 2016 (Arber et al., Blood 2016), nine AML subtypes of clinical and prognostic importance are distinguished by distinct and practically mutually exclusive mutations, covering 50-60% of AML cases. By analysing an extended panel of genes, Papaemmanuil et al. (NEJM 2016) developed a purely genomic classification of AML. In this system, 11 groups are defined including 6 entities characterized by chromosomal translocations. Similar as in WHO 2016, these 6 entities each account for less than 5% of AML and are identified by metaphase cytogenetics. Of these 6 entities, 5 groups are defined by fusion genes and one group by inv(3)/t(3;3) leading to overexpression of EVI1. The 5 remaining classes include 4 entities with cytogenetically normal AML defined by mutations in NPM1 (27%), bi-allelic CEBPA (4%), genes regulating RNA splicing, chromatin or transcription (18%) and IDH2 R172 mutations (1%) and one entity characterized by mutations in TP53, a complex karyotype or specific aneuploidies (13%). Although the majority of patients can be classified by this new system, 15% of patients still lack class-defining lesions and expression levels of structurally normal genes, which can also have a decisive prognostic impact, are not considered. We propose that whole transcriptome messenger RNA sequencing provides a single and flexible platform to identify the diversity of genetic aberrations relevant for classification of AML.
A panel of hundred AML were analysed and HAMLET (Human AMLExpedited Transcriptomics) was developed as bioinformatics pipeline to detect fusion genes, small variants in thirteen genes, long tandem duplications in FLT3 and KMT2A and overexpression of EVI1. In HAMLET, a new algorithm based on soft clipped reads was developed to detect long tandem repeats in FLT3 and KMT2A. All genetic aberrations called by HAMLET were validated by diagnostic data and targeted re-sequencing.
The data showed that HAMLET correctly called all genetic aberrations relevant for current classification of AML with high sensitivity and specificity. Moreover, the new soft clipped approach that has been integrated in HAMLET proved to be useful not only to detect long tandem duplications in FLT3 and KMT2A, but also to determine the allelic ratio of mutant-to-wild type FLT3, which is predictive for overall survival. By filtering small variants for predicted importance according to large AML sequencing data sets (Jaiswal et al., NEJM 2017), we classified the 100 AML according to genomic classification and showed that 87 cases were classified in single entities, 4 cases in two subgroups and 9 cases had no class-defining lesions. Of the 9 cases without class-defining lesions, 8 cases had detectable driver mutations and one case had no detectable driver mutation. These numbers perfectly match percentages reported by Papaemmanuil et al. (NEJM 2016). Apart from genetic aberrations that are relevant for current classification of AML, HAMLET also identified additional abnormalities. Of particular interest is NUP98-NSD1 (Hollink et al., Blood 2011), a cryptic fusion gene that is missed by metaphase cytogenetics in three AML with no class-defining lesions, and EVI1 overexpression in 5 cases without inv(3)/t(3;3) including three KMT2A-rearranged AML with extremely poor prognosis (Groschel et al., JCO 2013).
HAMLET correctly called all genetic aberrations relevant for current classification of AML and provides a wealth of additional information with potential consequences for patient management. In conclusion, HAMLET is a comprehensive and reliable pipeline for RNA sequence analysis that may contribute to better risk assessment and personalized treatment of AML.
Borras:GenomeScan B.V.: Employment. Janssen:GenomeScan B.V.: Employment.
Asterisk with author names denotes non-ASH members.