The majority of women with ovarian cancer (OC) have advanced disease at diagnosis and 5-year survival rates of less than 25%. Women with stage I disease have significantly better 5-year survival rates of over 90%. Recent large studies using CA 125 and transvaginal ultrasound have failed to improve mortality in a screened population. There is therefore a pressing need for new diagnostic biomarkers in OC. The primary aim of my project, as a first step in developing a diagnostic circulating tumour DNA (ctDNA) biomarker for high grade serous ovarian cancer (HGSOC), was to investigate low-cost high-throughput next generation sequencing assays in plasma samples collected from women with newly diagnosed OC. The secondary aim was to apply these methods to other non-invasive samples including cervical liquid based cytology samples that might contribute to earlier diagnosis or screening for women with OC. ctDNA was detected in 30-49% of women with newly diagnosed OC from the UKOPS (n=54) and CTCR-OV04 (n=156) cohorts using targeted sequencing. Using the trimmed median absolute deviation (t-MAD) score, a quantitative measure of genome wide copy number aberration generated from shallow whole genome sequencing (sWGS) data, ctDNA was detected in 39-41% of the women with newly diagnosed disease. To improve sensitivity of ctDNA detection I developed an optimised method for targeted sequencing that has the potential to lower the limit of detection of ctDNA in HGSOC by 100 fold. I have also shown that the size profile of HGSOC ctDNA fragments is different to that of wildtype DNA fragments and shown that selecting for DNA fragments between 90-150 bp can increase rates of ctDNA detection in HGSOC. ctDNA detection increased to 53-67% of women with newly diagnosed OC using the size selected t-MAD score. I have evaluated the utility of cervical sampling for earlier diagnosis of OC by testing and optimising DNA extraction, library preparation and sequencing methods. I have detected tumour DNA in routine cervical cytology samples collected from women subsequently diagnosed with cervical and endometrial cancers. In summary I have developed methods for ctDNA detection in women with newly diagnosed HGSOC that can be applied and refined in larger prospective studies of women undergoing follow-up for treated HGSOC, women with symptoms suggestive of OC and women at high risk of OC.

Ye / Genetic and epigenetic alterations significantly contribute to development of human cancer. Genotyping tumour tissue in search for these actionable genetic and epigenetic changes has become routine practice in oncology. However, sampling tumour tissue has significant inherent limitations. It provides only a single snapshot in time, prone to selection bias due to intra-tumour heterogeneity, and cannot always be performed owing to its invasive nature. Circulating tumour DNA (ctDNA) based liquid biopsy provides an effective alternative to invasive tissue sampling and have emerged as a minimally invasive, real-time biomarker. Recent advancements in DNA sequencing technologies have revealed enormous potential of ctDNA to improve tumour detection and stratification. In this review, we critically appraise the role of ctDNA as a liquid biopsy for cancer and evaluate the role of circulating tumour DNA as a diagnostic, prognostic and predictive biomarker. We also highlight some technical challenges and constraints associated with circulating DNA analysis.

There is a need for informative biomarkers in localised urological cancers. At present, no method can accurately distinguish between indolent and aggressive prostate cancers, and men often require repeated biopsies. Patients with muscle invasive bladder cancer undergo neo-adjuvant chemotherapy (NAC) to improve survival. However many do not respond to NAC, delaying definitive treatment. Cell-free mutant DNA (mutDNA) analysis represents an opportunity for non-invasive monitoring of cancer through tumour genome analysis. MutDNA derived from plasma can monitor tumour burden. There is emerging evidence that mutDNA can identify mutations from multiple clones and is abundant in adjacent body fluids. This work explores the utility of plasma and urinary mutDNA in localised prostate and bladder cancers. This thesis describes the optimisation of urinary mutDNA analysis by assessing urinary DNA processing and extraction methods using healthy volunteer and bladder cancer patient urine samples. Primer panels were designed and validated to target frequently mutated regions in prostate and bladder cancers, as well as for analysis of patient-specific mutations. Sequencing-based methods and dPCR were employed to analyse clinical samples including plasma and urine, to detect and quantify mutDNA. Molecular and clinical data were integrated to explore potential areas of application of mutDNA analysis. For bladder cancer, mutDNA was analysed from liquid-biopsy samples including plasma, cell pellets from urine and urine supernatant from multiple time-points of 17 MIBC patients undergoing NAC. I showed that mutDNA was more frequently detected and was present at higher AFs in urine compared to plasma samples. Of potential clinical relevance, I showed that the presence of mutDNA after starting NAC was associated with disease recurrence. This original contribution to knowledge could offer patients an opportunity to expedite surgical resection in a timely manner, if corroborated in large-scale trials. For prostate cancer, a TP53 specific panel was applied to men with metastatic disease, to demonstrate that clones containing TP53 mutations, which are dominant in at the metastatic stage were present in historical prostatectomy samples taken when then patient was believed to have localised disease only. Furthermore, I showed that these TP53 mutations could be detected at the localised stage of disease. To investigate the ability of mutDNA detection private clonal mutations I developed a method for higher sensitivity analysis (MRD-Seq). This was applied to a clinical cohort of 2 men with multi-focal localised prostate cancer to demonstrate the though the overall levels of mutDNA is low, private clonal mutations may be detectable. Taken together, these original contributions to knowledge could allow for less invasive surveillance of men with low risk prostate cancer and warrants further investigation. In this thesis, I used a range of molecular methods were applied to small cohorts of clinical samples from patients with urological malignancies, in an exploratory analysis. The molecular data was analysed in conjunction with clinical information to draw hypotheses on the biology and natural history of these cancer, and to suggest possible utility of mutDNA analysis in their clinical management. Some of the findings suggest areas of potential utility, which merit further validation or investigation in larger cohorts or clinical studies.