Our Research Projects

Clear cell sarcoma: a cancer of unmet need.

Clear cell sarcoma (CCS) is an ultra-rare soft tissue cancer and represents less than 1% of the approximately 3,500 new cases of sarcoma diagnosed in the UK every year. In the majority of cases, the genetic driver of the disease is an EWSR1::ATF1 translocation. Beyond this, there is only a limited understanding of the contribution other genetic, epigenetic and transcriptomic features to disease progression. Also, whilst significant advances have been made in the treatment of other metastatic cancers through chemotherapies and immunotherapies, clear cell sarcomas usually do not respond well to these treatments. There is an unmet need to repurpose existing drugs for treatment of CCS and to develop new therapies.

Current projects:

1. Genomic landscape

2. Understanding the evolutionary patterns of metastatic spread

3. Therapeutic interventions - development of protein degraders (PROTACS).

GENOME COMPLEXITY IN SARCOMAS

Compared to other many other cancers, most sarcomas are copy number and rearrangement driven rather than demonstrating a high single nucleotide substitution burden. They show striking copy number heterogeneity within and between specific subtypes of which there are more than 100 types. We’ve shown that undifferentiated pleomorphic sarcomas (UPS) harbour early TP53 or RB1 mutations, whose unstable genomes are stabilised by whole genome doubling (WGD) and show alternative lengthening of telomeres (ALT). This study was the nidus for the development of a new computational framework to inform on the mutational processes associated with aneuploidy in sarcoma termed copy number (CN) signatures. We showed that CN signatures are useful for studying cancer heterogeneity and evolution not only in UPS and malignant peripheral nerve sheath tumours (MPNST) but applying it to 33 other cancer types revealed new signatures associated with WGD, homologous recombination deficiency and chromothripsis which was prognostically significant in specific cancer types.This reference set of signatures is now publicly available through the COSMIC website.

Circos plot of a whole genome sequence analysis of undifferentiated pleomorphic sarcoma

Ploidy analysis - plotting a course from flow cytometry and whole genome sequencing to imaging analysis

Deciphering the Pathological Context of Whole Genome Doubling in Sarcoma

Whole genome sequencing (WGS) provides unprecedented resolution for identification of single nucleotide variants, but is considerably less reliable for resolving large copy number alterations such as whole genome doubling (WGD) which is an extreme form aneuploidy where a cells chromosomal content doubles. Osteosarcoma and undifferentiated pleomorphic sarcoma show high levels of WGD and have a poor prognosis. In view of the poor understanding of their biology and the recent insights into the pathological consequences of WGD and aneuploidy (poor survival and metastatic disease) it argues
for the development of tools to interpret accurately these features from WGS and pathological imaging.

Using state of the art genomics and artificial intelligence of digital pathology images we aim to deliver innovative genomic and digital pathology biomarkers that predict disease behaviour.

AI generated post-impressionist image of transposable elements in DNA

Determining The Role of Mobile Transposable Elements in Sarcoma Genome Complexity and Immune Response

Transposable elements (TEs) or "jumping genes" comprise a significant proportion of the human genome and serve as a "fossil record" of previous viral intergration. Whilst the majority of these genetic sequences have lost their ability to transpose, their still exist a few classes sucha as LINE-1 and Alu elements that can continue to do so. Their integration into regulatory regions or within genes can disrupt gene function and contribute to disease and genomic instability. Somatic TE insertions are associated with structural variation in some cancer types and their expression resulting in viral mimicry promotes the innate immune response.

There is emerging evidence of their role in sarcoma biology and in particular with the stimulation of the immune response.

Our aim is to map out the regions of TE insertion across multiple sarcoma subtypes and investigate their role in immuno-oncology using long-read sequencing technologies and disease models.

This project contributes to Neo-Sarc, an initiative to enhance the whole genome sequencing of genomically complex sarcomas, which are bone and soft tissue tumours driven by multiple and often large-scale mutations. The focus is investigating the potential of different next generation sequencing platforms on blood-based liquid biopsies, as a future alternative to tissue biopsies, given they are less invasive and may overcome intra-heterogeneity by not being site-specific.

The plasma samples analysed belong to multiple clinical cohorts, posing an invaluable opportunity to study multiple subtypes in a systematic way. The aim is to uncover novel multimodal biomarkers in circulating tumour DNA (ctDNA), such as copy number alterations, methylation patterns, and fragmentomics profiles. In doing so, these findings could be translated into clinical medicine to monitor longitudinal disease, predict treatment response, and detect early recurrence.

Studying the effect of radiotherapy on soft tissue sarcomas

Soft tissue sarcomas are a very diverse group of tumours, for which the main treatment is radiotherapy followed by surgery. However, the response to radiotherapy is mixed with some tumour types responding better than others. Very little is understood about the effect of radiotherapy on the genomes of sarcomas and there is currently no way to predict response to radiotherapy. Our research studies the effect of radiotherapy on the genomes of sarcomas by comparing the DNA sequencing data before and after treatment. Coupling this information with transcriptional information from RNAseq data, the aim is to better understand how sarcomas respond to radiotherapy.

Signatures as a biomarker for homologous recombination deficiency (HRD) in high grade serous ovarian cancer (HGSC)

Homologous recombination repair deficiency (HRD) is a frequent feature of high-grade serous ovarian cancer and is emerging as a predictive biomarker for optimising PARP inhibitor treatment. Current commercial HRD assays are imperfect and cost prohibitive. We have developed a next generation HRD assay using copy number signatures (CNS) generated from exome sequencing. This pilot project will ascertain whether our HRD CNS can reliably be obtained from formalin-fixed paraffin embedded (FFPE) tissue without the reliance on a ‘research grade’ fresh tumour samples. This will allow our cost-effective assay to be applied at scale to routine clinical samples, to accurately identify patients who could benefit from PARP inhibition therapy.

Copy number signatures

https://github.com/AlexandrovLab/SigProfilerExtractor

SigProfilerExtractor allows de novo extraction of mutational signatures from data generated in a matrix format. The tool identifies the number of operative mutational signatures, their activities in each sample, and the probability for each signature to cause a specific mutation type in a cancer sample. The tool makes use of SigProfilerMatrixGenerator and SigProfilerPlotting.

Immunogenomic landscape of soft tissue sarcomas

Molecular characterisation of undifferentiated soft tissue sarcomas

One such tumour is undifferentiated pleomorphic sarcoma (UPS), which is the current focus of the lab. In order to address the unmet need to develop a systematic molecular classification of UPS to stratify these patients for suitable clinical trials we undertake comprehensive genomic profiling complemented by other techniques such as methylation and gene expression analysis using patient samples.

Using deep learning and computer vision to distinguish lipomas from well-differentiated liposarcomas

parallel with the work above we have an exciting new collaboration with Dr Kevin Bryson (UCL Computer Science) in order to better understand the genotype-phenotype correlations and the cellular spatial dynamics of sarcomas. This is accomplished by applying conventional learning machine techniques as well as newer deep learning methods on digitised histological images to unravel the unique architectural and cytological nuances of these enigmatic tumours.