Understanding the genomics of treatment resistance in ovarian cancer

Fewer than 30% of women presenting with high-grade serous ovarian cancer (HGSOC) are alive five years after diagnosis. This aggressive disease is characterised by genomic instability, resulting in significant loss or gain of genetic material (copy-number variation), facilitating rapid tumour evolution. Platinum-based chemotherapy is designed to target these genetically unstable tumour cells, however, a high proportion of patients ultimately relapse. This resistance to treatment has been attributed to the significant tumour heterogeneity observed in HGSOC, which facilitates enhanced sampling of the tumour fitness landscape. Thus there is a pressing need to understand the mechanisms behind copy-number variation in HGSOC, understand how these changes convey treatment resistance, and develop methods that track these changes in response to therapy that takes into account tumour heterogeneity.

As a post-doc in the Brenton Lab at CRUK Cambridge Institute, University of Cambridge, and in collaboration with Prof Iain McNeish at the University of Glasgow, I work on computational methods for tracking copy-number changes in tumour samples collected longitudinally from HGSOC patients as part of the BritROC study.

Modelling tumour evolution using structural variants

Working together with members of the Markowetz Lab and remotely with my PhD student Marek Cmero at the University of Melbourne, we develop methods for identifying subclonal structural variants and inferring the evolutionary history of tumours. We are currently applying our methods to over 2,500 tumours as part of the PanCancer Analysis of Whole Genomes project.

Integrated genomics for lethal prostate cancer (2012 – 2014)

Publication 1 | Publication 2 | Publication 3 | Publication 4

Prostate cancer is the most diagnosed internal malignancy in the western world. While the majority of prostate cancers are non-lethal, there is currently no reliable approach to distinguish lethal from non-lethal prostate cancer at an early, curable stage. To help understand the molecular mechanisms driving lethal (metastatic) prostate cancer, I collaborated with A/Prof Chris HovensDr Niall Corcoran and Dr Matt Hong, to carry out molecular profiling of primary tumours and metastases using whole-genome sequencing, RNA-SEQ, Illumina 2.5M SNP Chip, and Illumina 450K methylation chip.

Working with Dr Clare Sloggett at the Victorian Life Sciences Computation Initiative we designed robust software pipelines for whole-genome sequence analysis and transcriptome analysis.  In conjunction with Dr Haroon Naeem we have developed filters for removing noisy probes from the Illumina 450K methylation array.

The results from these data allowed us to model the subclonal evolution of these tumours in collaboration with Dr David Wedge, Dr Peter Van Loo and Dr Ultan McDermott from the Cancer Genome Project at the Wellcome Trust Sanger Institute. Findings which we validated with the help of Dr Sebastian Lunke from Department of Pathology, University of Melbourne.

This project was supported in part by NHMRC grant: NHMRC APP1047581 Determining the origin of lethal metastases in multifocal primary prostate cancer. CIA A/Prof Chris Hovens. CIB Dr Niall Corcoran. CIC Prof Izhak Haviv. CID Dr Adam Kowalczyk. CIE A/Prof Andrew Lonie. CIF Dr Geoff Macintyre. CIG Prof David Neal. CIH Prof Anthony Costello. CII A/Prof John Pedersen.

Biomarker discovery (2012-2014)


Working in close collaboration with Dr Adam Kowalczyk, we developed a suite of machine learning algorithms for robust biomarker discovery, algorithm training, and testing. Applied these algorithms to three classification tasks:

  1. Classification of high-grade and low-grade prostate cancer from gene expression profiling of benign biopsy tissueMany prostate cancer patients have indolent disease amenable to active surveillance, but sampling error on biopsy means that up to a third may actually harbour higher grade disease. This study aims to determine if benign cores of prostate glands known to harbour high or low grade disease are distinguishable on a molecular level. In addition, we aim to test the viability of the gene signatures identified for clinical prediction of tumour grade from benign prostate tissue.
  2. Classification of high-grade and low-grade bladder cancer from miRNAs in urine. The long-term surveillance mandated for non muscle-invasive bladder patients is expensive and resource intensive. The objective of this study is to determine if microRNA (miR) profiling of urine can identify the presence of bladder cancer (CaB) in a diagnostic or surveillance setting and to compare it’s performance characteristics to that of cystoscopy.
  3. Classification of high-grade and low-grade prostate cancer using gene expression profiling of periprostatic fat. Extensive epidemiological evidence indicates that dysregulated adipose homeostasis contributes to the development and progression of various cancers. The objective of this study is to determine the role of gene expression changes in periprostatic fat in contributing to prostate cancer progression as well as their ability to predict the presence of high-grade and low-grade tumours.

Methods for identifying regulatory variants (2008-2011)

Web server (SNPs) | Web server (INDELs) | Publication

Single nucleotide polymorphisms and small insertions and deletions (INDELs) account for a significant amount the variation between human individuals. Genome-wide association studies and large scale genome sequencing projects such as the International Cancer Genome Consortium and The Cancer Genome Atlas frequently identify SNPs and small INDELs which are associated with disease. Many of these SNPs/INDELs may be the causal variation actually contributing towards disease susceptibility. SNPs and small INDELs, can impact normal genomic function in a variety of ways, including disrupting the binding of transcription factors (TFs). We are therefore developing a tools which aim to identify SNPS and small INDELs which impact on the binding of a TF to DNA.