Transcriptional assessment of haematopoietic differentiation to risk-stratify acute lymphoblastic leukaemia
Lead Research Organisation:
Newcastle University
Department Name: Biosciences Institute
Abstract
B-cell acute lymphoblastic leukaemia (B-ALL) is a life-threatening blood cancer which can affect both the young and the elderly. Many children with B-ALL can now be cured, but the outlook for adults is less favourable. In treating B-ALL there are two main challenges:
1) Identifying which children can be cured with less treatment.
B-ALL treatment is long and challenging. It has psychological, social and educational consequences for children and families. It also increases the risk of health
problems in later life, such as heart disease, bone or joint problems and infertility.
2) Identifying which adults are least likely to be cured with standard treatments.
Options for these patients include using new combinations of drugs and therapies that harness the anti-cancer functions of the immune system.
To personalise treatment in this way, we need to accurately predict the risk of leukaemia relapsing (coming back after treatment). At present, we predict risk using the patient's age, white blood cell count at diagnosis and certain changes in the leukaemia DNA (the genetic instructions of the leukaemia cell). It may be possible to improve this prediction by adding measurements of RNA (an indication of which DNA instructions the leukaemia cell is currently using).
In recent work, I matched signals from leukaemia RNA to signals from the RNA of healthy maturing blood cells. This 'fingerprint of maturation' allowed me to separate the known subgroups of B-ALL, which are based on DNA changes. Subgroups with a higher risk of relapse had more 'immature' signals. However, signals varied between patients, and it isn't yet clear what this means.
In the following work, I aim to:
1) Ensure that the method for matching signals is sensitive to maturation and reliable across RNA datasets
2) Test whether immature signals correlate with relapse risk, by comparing signals in leukaemia cells from patients who relapsed versus patients who were cured
3) Compare the maturation extremes within subgroups of B-ALL to understand what changes in DNA structure, DNA readability, and RNA affect maturation
With these insights, I will establish whether determining the RNA 'fingerprint of maturation' could improve our predictions of relapse risk. If successful, this method could support personalised B-ALL treatment decisions. Further exploration of the data could help identify ways of changing B-ALL maturation to modify how the leukaemia cells behave and respond to treatment. All RNA data generated will be shared with the research community to maximise scientific advances in this disease.
1) Identifying which children can be cured with less treatment.
B-ALL treatment is long and challenging. It has psychological, social and educational consequences for children and families. It also increases the risk of health
problems in later life, such as heart disease, bone or joint problems and infertility.
2) Identifying which adults are least likely to be cured with standard treatments.
Options for these patients include using new combinations of drugs and therapies that harness the anti-cancer functions of the immune system.
To personalise treatment in this way, we need to accurately predict the risk of leukaemia relapsing (coming back after treatment). At present, we predict risk using the patient's age, white blood cell count at diagnosis and certain changes in the leukaemia DNA (the genetic instructions of the leukaemia cell). It may be possible to improve this prediction by adding measurements of RNA (an indication of which DNA instructions the leukaemia cell is currently using).
In recent work, I matched signals from leukaemia RNA to signals from the RNA of healthy maturing blood cells. This 'fingerprint of maturation' allowed me to separate the known subgroups of B-ALL, which are based on DNA changes. Subgroups with a higher risk of relapse had more 'immature' signals. However, signals varied between patients, and it isn't yet clear what this means.
In the following work, I aim to:
1) Ensure that the method for matching signals is sensitive to maturation and reliable across RNA datasets
2) Test whether immature signals correlate with relapse risk, by comparing signals in leukaemia cells from patients who relapsed versus patients who were cured
3) Compare the maturation extremes within subgroups of B-ALL to understand what changes in DNA structure, DNA readability, and RNA affect maturation
With these insights, I will establish whether determining the RNA 'fingerprint of maturation' could improve our predictions of relapse risk. If successful, this method could support personalised B-ALL treatment decisions. Further exploration of the data could help identify ways of changing B-ALL maturation to modify how the leukaemia cells behave and respond to treatment. All RNA data generated will be shared with the research community to maximise scientific advances in this disease.
Technical Summary
With improved risk stratification in B-cell acute lymphoblastic leukaemia (B-ALL), we could personalise treatment decisions to reduce treatment toxicity without compromising cure. Whole-transcriptome-based matching of leukaemia cells to normal blood cells yields a pattern of differentiation signals that segregates by genetic subgroup of B-ALL, with some high-risk genetic subgroups exhibiting immature differentiation signals. Differential signals vary within genetic subgroups but the reason for these differences is currently unknown.
Objectives
1) Optimising the approach and computing differentiation signals across the B-ALL age and genomic spectrum
Published bulk leukaemia RNA sequencing (RNAseq) data will be deconvoluted using a single cell reference. Deconvolution methods and references will be tested to ensure the method is sensitive to maturation differences and generates consistent results across datasets.
2) Determining whether specific differentiation signals are associated with adverse outcome
The distribution of differentiation signals will be compared in bulk RNAseq data generated from patients who relapsed versus controls who did not (diagnostic material from two common B-ALL subtypes banked in ALL trials).
3) Unravelling the genomic control of B-ALL differentiation by single cell multi-omic assessment
I will take differentiation extremes from both B-ALL subtypes, assessing differences in gene expression (single cell RNAseq), immunoglobulin gene rearrangement and chromatin accessibility (single cell ATACseq).
Next steps: If distinct differentiation signals correlate with relapse, a retrospective cohort study will determine whether these signals are an independent risk factor for relapse. Candidate drivers of B-ALL differentiation will be interrogated in a model system, to identify whether perturbation impacts lymphoblast behaviour and drug sensitivity. All data will be shared with the community to maximise advances in the field.
Objectives
1) Optimising the approach and computing differentiation signals across the B-ALL age and genomic spectrum
Published bulk leukaemia RNA sequencing (RNAseq) data will be deconvoluted using a single cell reference. Deconvolution methods and references will be tested to ensure the method is sensitive to maturation differences and generates consistent results across datasets.
2) Determining whether specific differentiation signals are associated with adverse outcome
The distribution of differentiation signals will be compared in bulk RNAseq data generated from patients who relapsed versus controls who did not (diagnostic material from two common B-ALL subtypes banked in ALL trials).
3) Unravelling the genomic control of B-ALL differentiation by single cell multi-omic assessment
I will take differentiation extremes from both B-ALL subtypes, assessing differences in gene expression (single cell RNAseq), immunoglobulin gene rearrangement and chromatin accessibility (single cell ATACseq).
Next steps: If distinct differentiation signals correlate with relapse, a retrospective cohort study will determine whether these signals are an independent risk factor for relapse. Candidate drivers of B-ALL differentiation will be interrogated in a model system, to identify whether perturbation impacts lymphoblast behaviour and drug sensitivity. All data will be shared with the community to maximise advances in the field.
People |
ORCID iD |
| Laura Jardine (Principal Investigator / Fellow) |