REVIEWS Genomics in acute lymphoblastic leukaemia: insights and treatment implications Kathryn G. Roberts and Charles G. Mullighan Abstract | Acute lymphoblastic leukaemia (ALL) is the commonest childhood cancer and an important cause of morbidity from haematological malignancies in adults. In the past several years, we have witnessed major advances in the understanding of the genetic basis of ALL. Genome-wide profiling studies, including microarray analysis and genome sequencing, have helped identify multiple key cellular pathways that are frequently mutated in ALL such as lymphoid development, tumour suppression, cytokine receptors, kinase and Ras signalling, and chromatin remodeling. These studies have characterized new subtypes of ALL, notably Philadelphia chromosome-like ALL, which is a high-risk subtype characterized by a diverse range of alterations that activate cytokine receptors or tyrosine kinases amenable to inhibition with approved tyrosine kinase inhibitors. Genomic profiling has also enabled the identification of inherited genetic variants of ALL that influence the risk of leukaemia development, and characterization of the relationship between genetic variants, clonal heterogeneity and the risk of relapse. Many of these findings are of direct clinical relevance and ongoing studies implementing clinical sequencing in leukaemia diagnosis and management have great potential to improve the outcome of patients with high-risk ALL. Roberts, K. G. & Mullighan, C. G. Nat. Rev. Clin. Oncol. advance online publication 17 March 2015; doi:10.1038/nrclinonc.2015.38
Introduction
Department of Pathology, St Jude Children’s Research Hospital, 262 Danny Thomas Place, Mail Stop 342, Memphis, TN 38105, USA (K.G.R., C.G.M.). Correspondence to: C.G.M. charles.mullighan@ stjude.org
Acute lymphoblastic leukaemia (ALL) is the commonest childhood cancer. 1 Current treatment regimens result in 5-year event-free survival rates that exceed 85% in children (aged 1–21); however, disease relapse is associated with a poor outcome,2,3 and ALL remains the leading cause of cancer-related death in children and young adults (aged 21–39).1 Although ALL is less common in adults, treatment outcomes are significantly inferior to those in children.4 The reasons underlying this age-related decline in outcome are not completely understood, but include a reduced prevalence of genetic alterations associated with a favourable outcome, such as high hyperdiploidy, presence of the ETV6–RUNX1 gene fusion and a higher incidence of genetic alterations associated with poor outcome, such as the BCR–ABL1 fusion in adults compared to children.4 Nevertheless, when compared to childhood ALL, detailed information on the genetic basis of ALL in adults is lacking. Importantly, few therapeutic strategies are available that specifically target genes or pathways known to be mutated in ALL. Development of such targeted approaches is urgently needed as currently used multi agent chemotherapy is associated with substantial shortterm and long-term dose-limiting toxicities. Here, we review the current understanding of cytogenetic and molecular classification of ALL, with an emphasis on the latest insights into new entities of ALL with implications for improved clinical practice. Competing interests The authors declare no competing interests.
Genomic profiling of ALL
Extensive efforts have been made to comprehensively define the genetic basis of ALL, and to identify all lesions contributing to leukaemogenesis and treatment fail ure.5–18 These efforts began with use of microarray profiling of gene expression and analysis of DNA copy number alterations. In the past 5 years studies have included use of next-generation sequencing (NGS) techniques to identify all inherited and somatic genetic alterations.11,19–24 NGS encompasses a range of techniques that enable the sequencing of hundreds of thousands of nucleic acid molecules simultaneously. In order of complexity, these approaches include sequencing of gene panels, exome sequencing, transcriptome (expressed RNA) sequencing and whole-genome sequencing (WGS),25 which are complementary approaches, each having different advantages and disadvantages (Table 1).26,27 Compared to WGS, exome sequencing is a relatively inexpensive approach to identify protein-coding mutations, but has limited capability to identify structural genetic alterations such as rearrangements, deletions and insertions of DNA that are important in the pathogenesis of many tumours, including ALL. Exome sequencing is generally performed at a depth of 100-fold to 200-fold coverage of the haploid genome, which enables the detection of mutations present in tumour subclones, which is important in the study of relapse. Transcriptome (RNA) sequencing involves sequencing the expressed genome, and can be tailored to selectively study RNA transcripts that encode proteins (mRNAsequencing), all transcripts regardless of coding potential (RNA sequencing) or a variety of small and non-coding
NATURE REVIEWS | CLINICAL ONCOLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ Acute lymphoblastic leukaemia (ALL) is comprised of multiple subtypes with constellations of chromosomal rearrangements, deletions and gains of DNA, and sequence mutations that target common cellular pathways ■■ The prevalence of ALL subtypes varies significantly with age ■■ ALL is commonly a polyclonal disease and specific genetic alterations influence the risk of drug resistance, treatment failure and disease relapse ■■ Philadelphia (Ph)-chromosome-like ALL is common in children with high-risk ALL, and adults with ALL and is characterized by genetic alterations activating kinase signalling pathways, which are sensitive to tyrosine-kinase inhibitors ■■ Common inherited genetic variants in lymphoid transcription factors and tumour-suppressor genes influence the risk of developing ALL, and are associated with ALL subtype and ethnicity ■■ Rare mutations have been identified that drive the development of familial ALL
RNA transcripts. Non-coding RNAs are a diverse family of RNA transcripts that are generally thought to lack coding potential, but are increasingly believed to have important roles in cancer biology,28,29 such as epigenetic regulation, enhancer activity, regulation of intra-chromosomal and inter-chromosomal interactions, which are the target of extensive investigation.30–32 RNA sequencing is a highly informative approach and enables identification of chromosomal rearrangements that result in expression of chimeric fusion genes, digital gene-expression profiling and sequence mutation detection. Transcriptome sequencing also enables the identification of new transcripts and gene isoforms. WGS entails sequencing the entire genome of a sample. In cancer genomics WGS usually involves sequencing the tumour and matched non-tumour DNA of an individual. Sequencing matched non-tumour DNA is important so that tumour-acquired (somatic) mutations, germline mutations and common inherited variations can be accurately distinguished. WGS is the most comprehensive modality, but might not identify all
genetic alterations, owing to variation in sequence coverage and difficulties in sequencing complex and GC-rich regions of the genome (including gene promoters). Also, owing to the large size of the human genome and extent of sequencing required, the coverage provided by WGS is generally lower than exome sequencing, and WGS is less sensitive in the identification of subclonal variants. The ability to identify genetic alterations is dependent on the analysis algorithms used, and complex variants, such as insertion or deletion mutations and structural rearrangements, are challenging to identify. Moreover, careful investigation of non-coding structural and transcriptomic alterations is important to identify regulatory elements including non-coding RNAs, and alterations in enhancer regions, which are critical events in leukaemogenesis.28,33,34 The above-mentioned considerations are important not only for research efforts in ALL, but also for the implementation of diagnostic sequencing, as the ‘simpler’ modalities, such as exome sequencing, do not detect all clinically important genetic alterations, and transcriptome and/or WGS are often required for a definitive analysis. These clinically important alterations include identification of new fusions that are hallmarks of ALL subtypes (for example CRLF2 rearrangem ents and fusions in Philadelphia chromosome [Ph]-like ALL); alterations that predict poor outcome (for example IKZF1 deletions and mutations); alterations that might be targets for new therapy (for example kinase-activating fusions, deletions and mutations in Ph-like ALL and T-cell ALL [T-ALL]); and mutations that confer resist ance to specific agents (for example CREBBP deletions and mutations, and 5’-nucleotidase, cytosolic II [NT5C2] mutations). The insights obtained from genomic profiling of ALL can be grouped in several categories: identification
Table 1 | Next-generation sequencing methods Method
Description
Comments
Targeted sequencing
Sets of genes or hotspots of mutation are studied by PCR amplification or oligonucleotide baits that ‘capture’ regions of interest
Provides sensitive and deep coverage of targeted sites, particularly for sequence mutations The ability to detect rearrangements and copy number alterations is limited and dependent on the assay design and analysis approach
Exome sequencing
Regions of the genome corresponding to coding exons (and variably, additional regions including promoter regions and non-coding RNAs) are captured and sequenced
Widely used to identify sequence mutations Might identify DNA copy number alterations and rearrangements that occur in the region of the genes captured
Transcriptome sequencing
Transcribed regions of the genome are converted to DNA and sequenced
A powerful and flexible approach that can identify gene rearrangements (for example, chimeric fusion genes commonly identified in ALL), provide sensitive measurements of gene-expression profiling, analyse expression of mutations, alternative and novel gene isoform expression, and non-coding RNA expression
Wholegenome sequencing
The entire DNA of a sample is sequenced
Matched tumour and non-tumour DNA are usually sequenced and sequence data are mapped to a reference genome (genome ‘resequencing’) The most comprehensive approach to identify sequence mutations, structural genetic changes and rearrangements The tumour genome is compared to the matched non-tumour genome to identify somatic variants, and the non-tumour sample is compared to a reference genome to identify germline variants Coverage is usually greater than exome sequencing (typically 30–50-fold compared with 100–200-fold) and the ability to detect low-frequency, subclonal variants is lower Does not analyse gene expression
Abbreviation: ALL, acute lymphoblastic leukaemia.
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrclinonc © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 2 | Key subtypes of ALL Subtype
Prevalence (%)*
Comment
Hyperdiploidy with more than 50 chromosomes
20–30
Excellent prognosis
Hypodiploidy with less than 44 chromosomes
2–3
Poor prognosis, high frequency of Ras pathway and IKAROS gene family and TP53 mutations
t(12;21)(p13;q22) translocation encoding ETV6–RUNX1 fusion
15–25
Excellent prognosis
t(1;19)(q23;p13) translocation encoding TCF3–PBX1 fusion
2–6
Increased incidence in African-Americans; generally excellent prognosis; association with CNS relapse
t(9;22)(q34;q11.2) translocation encoding BCR–ABL1 fusion
2–4
Historically poor outcome, improved with addition of imatinib and/or dasatinib to intensive chemotherapy
Ph-like ALL
10–15
Multiple cytokine receptor and kinase-activating lesions; associated with IKZF1 alteration and very high leucocyte count; potentially amenable to tyrosine-kinase inhibitor therapy
t(4;11)(q21;q23) translocation encoding MLL–AF4 fusion
1–2
Common in infant ALL (especially
Introduction
Department of Pathology, St Jude Children’s Research Hospital, 262 Danny Thomas Place, Mail Stop 342, Memphis, TN 38105, USA (K.G.R., C.G.M.). Correspondence to: C.G.M. charles.mullighan@ stjude.org
Acute lymphoblastic leukaemia (ALL) is the commonest childhood cancer. 1 Current treatment regimens result in 5-year event-free survival rates that exceed 85% in children (aged 1–21); however, disease relapse is associated with a poor outcome,2,3 and ALL remains the leading cause of cancer-related death in children and young adults (aged 21–39).1 Although ALL is less common in adults, treatment outcomes are significantly inferior to those in children.4 The reasons underlying this age-related decline in outcome are not completely understood, but include a reduced prevalence of genetic alterations associated with a favourable outcome, such as high hyperdiploidy, presence of the ETV6–RUNX1 gene fusion and a higher incidence of genetic alterations associated with poor outcome, such as the BCR–ABL1 fusion in adults compared to children.4 Nevertheless, when compared to childhood ALL, detailed information on the genetic basis of ALL in adults is lacking. Importantly, few therapeutic strategies are available that specifically target genes or pathways known to be mutated in ALL. Development of such targeted approaches is urgently needed as currently used multi agent chemotherapy is associated with substantial shortterm and long-term dose-limiting toxicities. Here, we review the current understanding of cytogenetic and molecular classification of ALL, with an emphasis on the latest insights into new entities of ALL with implications for improved clinical practice. Competing interests The authors declare no competing interests.
Genomic profiling of ALL
Extensive efforts have been made to comprehensively define the genetic basis of ALL, and to identify all lesions contributing to leukaemogenesis and treatment fail ure.5–18 These efforts began with use of microarray profiling of gene expression and analysis of DNA copy number alterations. In the past 5 years studies have included use of next-generation sequencing (NGS) techniques to identify all inherited and somatic genetic alterations.11,19–24 NGS encompasses a range of techniques that enable the sequencing of hundreds of thousands of nucleic acid molecules simultaneously. In order of complexity, these approaches include sequencing of gene panels, exome sequencing, transcriptome (expressed RNA) sequencing and whole-genome sequencing (WGS),25 which are complementary approaches, each having different advantages and disadvantages (Table 1).26,27 Compared to WGS, exome sequencing is a relatively inexpensive approach to identify protein-coding mutations, but has limited capability to identify structural genetic alterations such as rearrangements, deletions and insertions of DNA that are important in the pathogenesis of many tumours, including ALL. Exome sequencing is generally performed at a depth of 100-fold to 200-fold coverage of the haploid genome, which enables the detection of mutations present in tumour subclones, which is important in the study of relapse. Transcriptome (RNA) sequencing involves sequencing the expressed genome, and can be tailored to selectively study RNA transcripts that encode proteins (mRNAsequencing), all transcripts regardless of coding potential (RNA sequencing) or a variety of small and non-coding
NATURE REVIEWS | CLINICAL ONCOLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ Acute lymphoblastic leukaemia (ALL) is comprised of multiple subtypes with constellations of chromosomal rearrangements, deletions and gains of DNA, and sequence mutations that target common cellular pathways ■■ The prevalence of ALL subtypes varies significantly with age ■■ ALL is commonly a polyclonal disease and specific genetic alterations influence the risk of drug resistance, treatment failure and disease relapse ■■ Philadelphia (Ph)-chromosome-like ALL is common in children with high-risk ALL, and adults with ALL and is characterized by genetic alterations activating kinase signalling pathways, which are sensitive to tyrosine-kinase inhibitors ■■ Common inherited genetic variants in lymphoid transcription factors and tumour-suppressor genes influence the risk of developing ALL, and are associated with ALL subtype and ethnicity ■■ Rare mutations have been identified that drive the development of familial ALL
RNA transcripts. Non-coding RNAs are a diverse family of RNA transcripts that are generally thought to lack coding potential, but are increasingly believed to have important roles in cancer biology,28,29 such as epigenetic regulation, enhancer activity, regulation of intra-chromosomal and inter-chromosomal interactions, which are the target of extensive investigation.30–32 RNA sequencing is a highly informative approach and enables identification of chromosomal rearrangements that result in expression of chimeric fusion genes, digital gene-expression profiling and sequence mutation detection. Transcriptome sequencing also enables the identification of new transcripts and gene isoforms. WGS entails sequencing the entire genome of a sample. In cancer genomics WGS usually involves sequencing the tumour and matched non-tumour DNA of an individual. Sequencing matched non-tumour DNA is important so that tumour-acquired (somatic) mutations, germline mutations and common inherited variations can be accurately distinguished. WGS is the most comprehensive modality, but might not identify all
genetic alterations, owing to variation in sequence coverage and difficulties in sequencing complex and GC-rich regions of the genome (including gene promoters). Also, owing to the large size of the human genome and extent of sequencing required, the coverage provided by WGS is generally lower than exome sequencing, and WGS is less sensitive in the identification of subclonal variants. The ability to identify genetic alterations is dependent on the analysis algorithms used, and complex variants, such as insertion or deletion mutations and structural rearrangements, are challenging to identify. Moreover, careful investigation of non-coding structural and transcriptomic alterations is important to identify regulatory elements including non-coding RNAs, and alterations in enhancer regions, which are critical events in leukaemogenesis.28,33,34 The above-mentioned considerations are important not only for research efforts in ALL, but also for the implementation of diagnostic sequencing, as the ‘simpler’ modalities, such as exome sequencing, do not detect all clinically important genetic alterations, and transcriptome and/or WGS are often required for a definitive analysis. These clinically important alterations include identification of new fusions that are hallmarks of ALL subtypes (for example CRLF2 rearrangem ents and fusions in Philadelphia chromosome [Ph]-like ALL); alterations that predict poor outcome (for example IKZF1 deletions and mutations); alterations that might be targets for new therapy (for example kinase-activating fusions, deletions and mutations in Ph-like ALL and T-cell ALL [T-ALL]); and mutations that confer resist ance to specific agents (for example CREBBP deletions and mutations, and 5’-nucleotidase, cytosolic II [NT5C2] mutations). The insights obtained from genomic profiling of ALL can be grouped in several categories: identification
Table 1 | Next-generation sequencing methods Method
Description
Comments
Targeted sequencing
Sets of genes or hotspots of mutation are studied by PCR amplification or oligonucleotide baits that ‘capture’ regions of interest
Provides sensitive and deep coverage of targeted sites, particularly for sequence mutations The ability to detect rearrangements and copy number alterations is limited and dependent on the assay design and analysis approach
Exome sequencing
Regions of the genome corresponding to coding exons (and variably, additional regions including promoter regions and non-coding RNAs) are captured and sequenced
Widely used to identify sequence mutations Might identify DNA copy number alterations and rearrangements that occur in the region of the genes captured
Transcriptome sequencing
Transcribed regions of the genome are converted to DNA and sequenced
A powerful and flexible approach that can identify gene rearrangements (for example, chimeric fusion genes commonly identified in ALL), provide sensitive measurements of gene-expression profiling, analyse expression of mutations, alternative and novel gene isoform expression, and non-coding RNA expression
Wholegenome sequencing
The entire DNA of a sample is sequenced
Matched tumour and non-tumour DNA are usually sequenced and sequence data are mapped to a reference genome (genome ‘resequencing’) The most comprehensive approach to identify sequence mutations, structural genetic changes and rearrangements The tumour genome is compared to the matched non-tumour genome to identify somatic variants, and the non-tumour sample is compared to a reference genome to identify germline variants Coverage is usually greater than exome sequencing (typically 30–50-fold compared with 100–200-fold) and the ability to detect low-frequency, subclonal variants is lower Does not analyse gene expression
Abbreviation: ALL, acute lymphoblastic leukaemia.
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrclinonc © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 2 | Key subtypes of ALL Subtype
Prevalence (%)*
Comment
Hyperdiploidy with more than 50 chromosomes
20–30
Excellent prognosis
Hypodiploidy with less than 44 chromosomes
2–3
Poor prognosis, high frequency of Ras pathway and IKAROS gene family and TP53 mutations
t(12;21)(p13;q22) translocation encoding ETV6–RUNX1 fusion
15–25
Excellent prognosis
t(1;19)(q23;p13) translocation encoding TCF3–PBX1 fusion
2–6
Increased incidence in African-Americans; generally excellent prognosis; association with CNS relapse
t(9;22)(q34;q11.2) translocation encoding BCR–ABL1 fusion
2–4
Historically poor outcome, improved with addition of imatinib and/or dasatinib to intensive chemotherapy
Ph-like ALL
10–15
Multiple cytokine receptor and kinase-activating lesions; associated with IKZF1 alteration and very high leucocyte count; potentially amenable to tyrosine-kinase inhibitor therapy
t(4;11)(q21;q23) translocation encoding MLL–AF4 fusion
1–2
Common in infant ALL (especially
