Precision oncology refers to the use of sophisticated assays to tailor therapy to an individual patient. The feasibility of implementing such a program in a comprehensive cancer centre in Australia is unknown. The utility of precision oncology also depends on understanding how genomic profiles may evolve over time and differ between tumours in the same patient. A precision oncology program called SEGMENT was designed and implemented in a single centre. The program was popular, recruiting well over the study period. Timely acquisition of samples for sequencing was suboptimal from external pathology providers, and proved increasingly expensive during the study period. Delivery of results in a manner where they could be utilized by the patient was challenging in cases where patients were referred late in their natural history. A custom hybrid capture panel worked reliably. A total of 300 patients were recruited to the study, of which 288 had at least one sample received. Accounting for attrition, 214 patients or 71% went onto the main study. The spectrum of mutations and copy number alterations found in this study was similar to published cohorts. There were few differences between primary and metastatic lesions on average. Paired primary and metastatic samples however displayed discordance for both copy number and mutations. This was the case for actionable alterations in ESR1, ERBB2 and very rarely PIK3CA. Approximately 50% of patients had an actionable alteration. Of these 14% of the overall cohort received a therapy matched to their genomic profile. Five patients received matched therapy off trial and 26 received matched trial therapy. Three were zero responses in the off-trial group, and a response rate of 27% in the matched trial therapy group. To explore genomic heterogeneity in greater resolution, 4 patients with advanced breast cancer underwent rapid autopsies to collect large numbers of metastatic samples. Whole exome sequencing was performed on multiple lesions per patient which allowed inference of the subclonal structure. All patients displayed a monophyletic architecture, with truncal driver alterations giving rise to subclones with differing genomic profiles. One patient with a long disease free interval from primary to metastasis showed the acquisition of a new driver in the metastatic lesion. Driver alterations appeared to shape subsequent evolution.

Barrett’s oesophagus (BO) is a metaplastic condition in which the normal squamous epithelium of the oesophagus is replaced by a columnar gastro-intestinal like epithelium due to repeated gastro-intestinal reflux. BO is generally accepted to be a precursor condition with the potential to develop into oesophageal adenocarcinoma. Consensus for the cell of origin for Barrett’s oesophagus is still lacking. Different sources for the cell of origin have been proposed, one of which is the submucosal gland (SMG) and duct cells of the oesophagus. This hypothesis, however, has not been properly studied due to the lack of proper model systems. In this thesis, pig and human SMGs and ducts were characterised and compared to each other to assess the suitability of pig SMGs as a substitute for human SMGs using immunohistology and multiplex immunofluorescence staining. Pig epithelial SMG cells were also further characterised using single cell RNA sequencing. Furthermore, organoid culture systems were developed to functionally assess the potential of the progenitor cells found using histologic and transcriptomic characterisation. Pig and human SMGs show different distributions at the distal end of their respective oesophagi but are largely similar in cell type and progenitor cell marker expression as demonstrated by histologic characterisation. Both pig and human SMGs showed progenitor cell marker expression in their respective basal duct cells (CD49f and p75) and the myoepithelial cells (CD49f), suggesting that both the ductal and glandular compartments to contain their own respective pool of progenitor cells. The transcriptomic analysis of pig SMGs single cell RNA sequencing largely supports their role in maintaining oesophageal homeostasis, as previously is known for human. Furthermore, the pseudotime trajectory inference data support the notion of the basal duct cells to be progenitor which give rise luminal duct cells. Interestingly, the data suggest the myoepithelial cells to be progenitor cells that could give rise to both basal duct cells and the gland compartment cells. Sorting and organoid culturing of basal duct cells demonstrated their capacity to grow into squamous spheroids, supporting their role in contributing to normal oesophageal repair. This process could be inhibited by treatment with retinoic acid. Similarly, culture of gland compartment cells gave rise to spheroids with two distinct morphologies. Dense type spheroids showed a squamous morphology but also mucin production found in BO. The second type of spheroid showed a cystic morphology and similarly produces BO type mucin. Finally, viral induction of intestinalisation in submucosal basal duct cells, in the current culture system, did not show metaplastic changes similar to BO. In summary, pig SMGs show great similarity to human SMGs. The pig SMGs contain progenitor cells in their ductal and gland compartments. These progenitors can be purified and cultured in vitro. The current developed protocols could be used to test the hypothesis that the SMGs contain the cell of origin of BO.

Pancreatic cancer is one of the most aggressive malignancies with an overall 5-year survival rate of <7%. Pancreatic cancer is highly resistant to radiotherapy and chemotherapy, and surgery is not feasible in most patients. In this thesis, I developed a new form of treatment for pancreatic cancer, based on immunotherapy. Adoptive cell transfer (ACT) is a promising form of cancer immunotherapy, which involves the isolation and reinfusion of tumour specific T lymphocytes into patients. While ACT can eliminate substantial burdens of some leukaemia, the ultimate challenge remains the eradication of large solid tumours and metastases for most cancers, including pancreatic cancer. In this thesis, an enhanced ACT treatment strategy for pancreatic cancer was developed, which was termed ‘ACTIV: Adoptive Cell Transfer Incorporating Vaccination’. This treatment included dual-specific T cells that expressed a chimeric antigen receptor (CAR) specific for the tumour antigen Her2, and a TCR specific for the melanocyte protein (pMEL, gp100). These dual specific T cells were termed ‘CARaMEL T cells’. CARaMEL T cells were administered together with an injection of a recombinant vaccinia virus vaccine expressing gp100 (VV-gp100). We hypothesized that adoptively transferred CARaMEL T cells would proliferate mediated by their gp100 TCR, in response to the VV-gp100 vaccine, and kill Her2+ tumours through their anti-Her2 CAR. Functional assays performed in vitro indicated that murine CARaMEL T cells mediated antigen-specific cytokine secretion and killing abilities against pancreatic cancer cells, and demonstrated potent proliferative ability in response to gp100 antigen, confirming our hypothesis. In addition, I found that ACTIV therapy inhibited tumour growth and prolonged the survival of mice bearing Her2+ subcutaneous murine pancreatic tumour. However, tumours usually relapsed after ACTIV therapy administration. Therefore, I directed my study to augment the anti-tumour activity of ACTIV therapy by the administration of either a histone deacetylase inhibitor (Panobinostat) or an immune agonist monoclonal antibody specific for CD40. Panobinostat significantly suppressed the growth of pancreatic cancer cells in vitro through apoptosis and cell cycle arrest. Also, Panobinostat significantly increased the growth suppression of pancreatic cancer cells mediated by CARaMEL T cells. In addition, I found that the combination of ACTIV therapy and Panobinostat significantly reduced the tumour growth and prolonged the survival of mice bearing Her2+ subcutaneous murine pancreatic tumours. In addition, administration of an agonist CD40 monoclonal antibody with ACTIV therapy significantly reduced the tumour growth and prolonged survival of mice bearing subcutaneous Her2+ pancreatic tumours through a T-cell-dependent immune mechanism. Finally, I explored the clinical translational potential for ACTIV therapy through the generation of human CARaMEL T cells expressing both a Her2-specific CAR and a gp100-TCR. In vitro functional assays indicated that human CARaMEL T cells mediated powerful and antigen-specific killing and cytokine secretion against Her2, together with a strong proliferative ability in response to gp100 antigen. In addition, I found that the administration of both human CARaMEL T cells and an adenovirus vaccine expressing gp100 led to potent anti-tumour activity against subcutaneous human Her2+ pancreatic tumours in immunodeficient mice.

Cancer immunotherapies have shown astounding clinical results within the last decade, with complete eradication of advanced malignancies in certain cancer types, particularly melanoma and non-small cell lung cancer (NSCLC). These successes lead to clinical approval in multiple countries for checkpoint blockade and chimeric antigen receptor (CAR) T cell therapies, and the award of the 2018 Nobel Prize in Physiology or Medicine to James P. Allison and Tasuku Honjo for their research on checkpoint blockade molecules. Despite this, many patients receive minimal benefit and focus has shifted to understanding how to predict and enhance immunotherapy responses. It is now well established that the tumour microenvironment (TME) is a major limiting factor for immunotherapy efficacy. Studies in mice using genetically identical tumour lines implanted in different tissues have demonstrated that the location of tumour growth can directly impact the composition of the TME and response to anti-cancer therapies. Retrospective analysis of checkpoint blockade treated patients’ revealed tissue-specific patterns of response, where metastases in certain anatomical sites were more responsive than others. To date studies investigating the tissue-specific influence on immunotherapy responses in vivo have limited clinical relevance, and studies in patients are minimal. In this thesis, we investigated the influence of the tissue-specific TME on immunotherapy responses in vivo and assessed tissue-specific patterns in the TME of breast cancer metastases from patient samples. First, we investigated a murine breast cancer model comparing responses of primary tumours to tumours in the liver and lungs as common metastatic sites to two immunotherapies, anti-PD-1/anti-CTLA4 and trimAb (anti-4-1BB, anti-CD40, anti-DR5). We reported that the 67NR tumour line growing in the lungs was resistant to immunotherapy, whereas the same tumour line growing in the mammary fat pad (MFP, primary tumour site) or liver could be completely eradicated in a portion of mice. Our analysis revealed that the resistance of lung tumours was independent of the tumour cells, vasculature or drug delivery and that the immune TMEs of lung and MFP tumours were distinct. Specifically, we demonstrated that lung tumours had a more immunosuppressive TME, with increased myeloid derived suppressor cells (MDSCs), decreased T cells and decreased activation of T cells and natural killer (NK) cells. Furthermore, upon depletion of various immune subsets alongside therapeutic intervention we found that NK cell depletion had a significant impact on lung tumours, but not MFP tumours. Taken together our data suggests that tumours grown in different tissues sculpt different TMEs with varied levels of immunosuppression and require different immune cell subsets, and perhaps different immune stimulants, for optimal anti-tumour responses. Following on from this study, we next wanted to assess responses to immunotherapy in vivo in models where multiple tumours in different anatomical sites were present. The rationale of this model was to investigate a more accurate representation of advanced cancer, where tumours have metastasised to multiple locations throughout the body. We hypothesised that co-existing tumours in different sites with disparate TMEs could influence immunotherapy responses compared to tumours existing alone. Our results indicated that the presence of a concomitant MFP tumour enhanced responses of lung tumours to trimAb or anti-PD-1/anti-CTLA4 therapies compared with mice bearing only lung tumours. We observed a decrease in lung metastasis burden in mice with simultaneous MFP tumour growth even before therapy commencement, which likely contributed to enhanced therapy responses. Upon interrogation we found that CD8+ T cells were responsible for the decrease in lung tumour burden and that the lungs of mice with co-existing MPF tumours had more tumour reactive CD8+ T cells. From our results, we hypothesised that the presence of a tumour in a more immunogenic location, such as the MFP, promoted T cell priming within the tumour draining lymph node (TdLN) at this site and led to a systemic response against distal tumours, such as tumours within the lungs. Lastly, we aimed to identify tissue-specific patterns within metastases from human tumours. Herein, we utilised metastatic tumour samples collected as part of the cancer tissue collection after death (CASCADE) rapid autopsy program from three estrogen receptor positive (ER+) breast cancer patients and one triple negative breast cancer (TNBC) patient. We analysed the immune profiles of these samples by transcriptomic and immunohistochemical (IHC) analyses. Our data demonstrated that, although there were potential tissue-specific differences within the TME, the most significant trend delineated immunological differences between ER+ and TNBC patients. These results confirmed previous research describing a higher immune infiltrate in TNBC samples compared with ER+ samples. Our research highlights the potential of investigating metastatic tumour samples however, future studies with separation of disease subtypes and increased sample sizes are needed to truly investigate tissue-specific patterns within the TME. In summary, the data presented in this thesis highlights the importance in further defining tissue-specific response patterns and mechanisms in patients to optimise current and future immunotherapies. Our results indicate that an in depth understanding of the tissue-specific TME could reveal novel treatment options in tumours that are non-responsive to current immunotherapies.

Background Women who inherit a germline pathogenic variant in the BRCA1 or BRCA2 genes have a significantly elevated lifetime risk of breast and ovarian cancer. Women who are aware of their BRCA carrier status can mitigate their increased risk by undergoing intensive breast cancer screening from a young age for early detection, and risk-reducing surgery for prevention of breast and/or ovarian cancer. The effectiveness of these interventions is dependent on BRCA carriers taking up these risk management strategies at an appropriate time, considering factors such as their age, personal preferences, and life stage. The most effective approach to ensuring carriers adhere to risk management recommendations is unknown. This project evaluates the lifetime health outcomes and cost-effectiveness of long-term clinical management of BRCA carriers in the context of structured clinical programs, using real-world data. Methods This thesis describes the development and outcomes of a discrete-time state-based microsimulation model. The model, named miBRovaCare, simulates the gene-specific natural histories for breast and ovarian cancer in BRCA carriers. Cost-effectiveness and cost-utility analyses were performed to evaluate the lifetime outcomes of different approaches to clinical management of carriers from the perspective of the Australian public healthcare system. The comparator for the base case analysis was the natural history (no cancer risk management). The interventions included: (i) a structured familial cancer service with a multidisciplinary high-risk clinic, and (ii) a formal annual carrier review program. For the intervention arms, BRCA carriers could undergo annual breast imaging, risk-reducing bilateral or contralateral mastectomy, and risk-reducing bilateral salpingo-oophorectomy. Uptake of and adherence to these strategies (patient behaviour) was based on an analysis of 983 BRCA carriers seen through a clinic in Melbourne, Australia. Additional model inputs were obtained from a local hospital database, the literature, government reports, and expert opinion. Costs and health outcomes were discounted by 5%. Results Long-term management of BRCA carriers through a familial cancer service is likely to be cost-effective, with or without an annual review program. A familial cancer service was the preferred strategy if the willingness-to-pay was at least $29,000 per quality-adjusted life-year (QALY) for BRCA1 carriers and $57,000 per QALY for BRCA2 carriers. Inclusion of an annual review program for BRCA1 carriers had a 75% probability of being cost-effective at a willingness-to-pay threshold of $50,000 per QALY. For BRCA2 carriers, a familial cancer service with or without an annual review program had only a 37% probability of being cost-effective at a $50,000 per QALY willingness-to-pay threshold. Discounting of health outcomes had, by far, the greatest impact on cost-effectiveness outcomes. Conclusions This thesis describes a novel microsimulation model for optimising clinical management of BRCA carriers. BRCA carriers are likely to benefit from access to structured clinical programs and regular review, due to fewer cancer diagnoses, improved life expectancy and an increase in QALYs. Genetic testing for hereditary breast and ovarian cancer predisposition syndromes is steadily expanding, and may even be available at a population-based level in the near future. Maximising adherence to evidence-based risk management guidelines along with access to appropriate follow-up services will therefore be increasingly important. The model developed for this thesis can enable faster evaluation of emerging risk management strategies and behavioural interventions, and can be easily adapted to alternative settings and healthcare systems.

Abstract Colorectal cancer (CRC) is the third most common cause of cancer-related death worldwide. Approximately 5% of CRC is attributable to an inherited high-penetrance pathogenic variant in a known CRC-predisposing gene, often associated with a family history of CRC, a young onset of CRC and/or colonic polyposis. Identification of an inherited genetic variant predisposing to CRC in an individual with CRC enables predictive testing in blood relatives, and implementation of risk reduction strategies for individuals with the causative variant within the family. In individuals with familial or young onset CRCs/colonic polyposis in whom a causative germline variant in a known CRC-predisposing gene has not been identified, it is hypothesised that novel CRC-predisposing genes harbouring rare high risk germline variants remain to be discovered. Germline whole exome sequencing (WES) offered a novel and comprehensive approach to identifying potential pathogenic variants in candidate CRC-predisposing genes. This thesis used WES of germline DNA in individuals with unexplained CRC and a high a priori risk of an inherited predisposition to CRC, to prioritise candidate CRC-predisposing genes that were subsequently re-sequenced in a validation cohort of young-onset CRC cases. A discovery cohort was recruited comprising 54 individuals who met one or more of the following criteria: CRC and a family history of CRC (n=29), a diagnosis of CRC before 40 years of age (n=20), or a diagnosis of 20 or more bowel polyps (n=10). No known germline predisposition to CRC had been identified in these individuals through clinical testing. Rare (minor allele frequency of <0.5%) loss-of-function (nonsense, frameshift or essential splice site) variants or missense variants predicted to be deleterious using a Combined Annotation Dependent Depletion C Score of >10 were prioritised for analysis. Three individuals with pathogenic variants in known CRC-predisposing genes were identified and excluded from further analyses. In the remaining 51 individuals, four separate analysis approaches were used to prioritise candidate CRC-predisposing genes containing prioritised variants: 1) genes with variants in multiple families, 2) genes with a shared variant in multiple family members, 3) genes within known CRC pathways, and 4) genes intolerant of loss-of-function variants. Forty-five candidate CRC-predisposing genes were prioritised using these analyses. Segregation analysis of potentially pathogenic variants within a family (n=1) and loss-of-heterozygosity studies in tumour tissue (n=4) were performed but hampered by lack of availability of samples. Re-sequencing of the 45 candidate CRC-predisposing genes in a validation cohort of 333 individuals with unexplained young onset CRC identified seven genes (WNT10A, NEDD4, MS4A15, TNFRSF10B, LRRK2, INPP5E, UFL1) with a loss-of-function variant in 1 or more individuals. None of the seven genes harboured LoF variants in >1% of the validation cohort, indicating that a novel gene likely accounts for very few individuals with unexplained CRC and a high a priori risk of an inherited predisposition to CRC, pointing to a very high degree of genetic heterogeneity in such cases. Future studies will require very large numbers of cases and controls to confirm the association of these genes with CRC predisposition.

Cytotoxic lymphocytes are highly efficient killer cells of the immune system. They destroy cognate target cells by secreting highly toxic effector molecules, the pore-forming protein perforin and pro-apoptotic serine proteases granzymes, into the confines of the immune synapse. Despite both the lymphocyte and target cell plasma membrane being equally exposed to the perforin and granzymes, the lymphocytes invariably survive that encounter as they remain resistant to perforin pores. This project investigates the mechanisms behind this unique phenomenon.

Peritoneal metastases from colorectal cancer confer the worst survival in patients with metastatic colorectal cancer. Historical survival from peritoneal metastases was dismal, with the condition generally viewed with nihilism. The adoption of cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC) can offer selected patients with colorectal peritoneal metastases (CRPM) a favorable survival. However, its uptake is variable due to skepticism about its efficacy. Additionally, most patients have inoperable disease, and are treated with systemic chemotherapy, fraught with high rates of treatment failures. With limited advances in management of peritoneal disease, there is an urgent need to explore newer avenues of treatment. In this thesis, I firstly confirm that there remains skepticism among surgeons regarding the utilisation of CRS and HIPEC for CRPM. Most recognise the role of cytoreductive surgery, however there remains deficiencies in awareness and knowledge regarding its indications and efficacy. To explore the efficacy of surgery, I evaluated outcomes from CRS and HIPEC for all peritoneal surface malignancies at a statewide referral centre. CRS and HIPEC was safe, and offered favorable survival. With CRPM specifically, median survival was 32 months, with a relapse free survival of 13 months following CRS and HIPEC. Incomplete cytoreduction and mucinous histology were key factors influencing survival. The choice of mitomycin C or oxaliplatin as HIPEC agent did not influence survival. In a meta-analysis, apart from well known factors such as PCI, completeness of cytoreduction and lymph node involvement, the use of adjuvant chemotherapy, a rectal primary and grade III/IV morbidity were significant prognostic factors influencing survival in patients undergoing CRS and HIPEC. The immune landscape of CRPM, a previously unchartered area, was then explored as a means to exploring newer treatment avenues. CRPM do have an immune infiltrate, albeit largely stromal, with a prominence of T cells, with over a fifth expressing PD-1. In an in-vitro tumouroid-TIL co-culture platform, we demonstrated that T cells in CRPM are functional, and the use of checkpoint antibodies can significantly improve T cell cytotoxicity in selected patients, offering this platform to personalise use of checkpoint antibodies in patients with CRPM. Additionally, gene expression analysis revealed most CRPM to be part of the worst prognostic CMS 4 subtype, with up regulation of immunosuppressive pathways. Another avenue garnering interest is personalised medicine, wherein drugs can be assigned based on unique molecular features of each cancer. Lastly, I led a multicentre, prospective study wherein a novel in-vitro patient derived tumouroid platform was established, that can integrate functional drug testing with genomic profiling to identify suitable therapeutic options in a clinically timely manner. Additionally, our team showed that results from this in-vitro platform successfully mirror patient drug responses in-vivo, and can help identify novel therapeutic options in patients with treatment refractory disease with no genomic-guided biomarkers. While current treatment of CRPM does offer highly selected patients a favourable survival, much remains to be improved for the remainder. Exploring a tumouroid based precision model of care has true potential in offering novel therapeutic options to an otherwise poor prognostic cohort.

The overarching objective of this thesis was to investigate key clinical care questions along the ovarian cancer disease trajectory and formulate strategies to improve care provision. This thesis focused on three time points on the trajectory: i) cancer prevention; ii) tolerability of experimental treatment at disease recurrence; and iii) development of malignant bowel obstruction. For women at high risk of developing ovarian cancer, such as those with a germline BRCA1 or BRCA2 mutation (gBRCA1/2mut), risk-reducing salpingo-oophorectomy (RRSO) is the most effective preventative method and recommended by peak bodies. RRSO involves complete resection of the ovaries and fallopian tubes up to the insertion into the cornua of the uterus. The resection of the fallopian tubes is necessary as it is known that BRCA1- and BRCA2-associated gynaecologic cancers appear to originate in the fimbrial end of the fallopian tubes rather than the ovaries, even though they have been typically labelled as “serous ovarian cancer”. Previous research had raised concerns that the quality of RRSO surgery and associated pathology evaluation performed in Australasian women had been suboptimal prior to 2008. I therefore undertook an investigation of the quality of contemporary RRSO and compared it with data from the historical research, examining the clinical predictors of RRSO quality. Across 78 institutions in Australia and New Zealand, a total of 164 study participants had RRSO performed and the quality of RRSO, in particularly the related pathology examination, had improved significantly compared to the previous report. The proportions of women who received adequate surgery and pathology evaluation were 99% and 66% respectively, compared to 91% and 23%, respectively (P<0.001) in the previous study. Implementable clinical strategies to potentially further improve quality were identified and included i) to direct RRSO referral to a gynaecologic oncologist (rather than general gynaecologist or general surgeon) and ii) to ensure the reporting pathologist is aware that the intent of surgery is risk-reduction in a high-risk woman, in order to prompt adequate processing and review of the RRSO specimens for occult cancer and pre-malignant lesions. In women with gBRCA1/2mut who undergo RRSO, whether hysterectomy should be performed at the time of RRSO to prevent uterine cancer was controversial. I therefore investigated the incidence of uterine cancer for Australasian gBRCA1/2mut carriers. Of the 828 eligible women identified from a prospective multi-institutional follow up study (kConFab), there were five cases of uterine cancer. Compared to the expected incidence in the Australian population, the standardised incidence ratio (SIR) was 2.45 (95%CI: 0.80-5.72; P=0.11). All five cases were of endometrioid adenocarcinoma histology and importantly, none had serous histology. Of those cases, three had a history of breast cancer and exposure to tamoxifen, a known risk factor for uterine cancer. Therefore, this finding did not justify the need for routine hysterectomy at the time of RRSO to reduce uterine cancer incidence in Australasian gBRCA1/2mut carriers. Disease progression in women with advanced ovarian cancer typically involves the omentum and peritoneum which causes abdominal symptoms. The goal of systemic treatment is to improve symptoms and delay further disease progression. Therefore, it is crucial that we consider the burden of treatment-related toxicities particularly in evaluation of novel drugs. Phase I clinical trials typically would include patients with different solid tumour types and are primarily intended to assess the safety and tolerability of investigational agents. Using the National Cancer Institute Phase I database, I performed a retrospective analysis of the adverse events (AEs) reported in women with gynaecological cancers compared to patients with other cancer types. Patients enrolled in the 150 phase I trials identified were divided into three groups to allow comparison; i) females with gynaecological cancers (n=685); ii) females with non-gynaecological cancers (n=1698); and iii) males with cancers (n=1886). Of those females with gynaecological cancers, the majority had ovarian cancer (n=527, 77%) followed by cervical cancer (n=76, 11%) and uterine cancer (n=72, 11%). Overall, females with gynaecological cancers were reported to have a significantly greater number of AEs during treatment (mean, 17.1 vs 14.7 vs 13.5, P<0.001), despite being similar at baseline (mean 7.0 vs 7.4 vs 7.0, P=0.09). In terms of AE severity, the main difference was a higher prevalence of grade 2 AEs reported in women with gynaecological cancers (mean 4.6 vs 3.9 vs 3.5). The five most prevalent AEs in women with gynaecological cancers were nausea (n=617, 90%), fatigue (n=587, 86%), anaemia (n=381, 56%), anorexia (n=357, 52%) and vomiting (n=355, 52%). These findings highlighted the need to improve the management for low-grade AEs in particular abdominal-related AEs for women with gynaecological cancers being treated on clinical trials. The inclusion of specific supportive care protocols/strategies into clinical trial protocols will better address symptom burden and improve quality of life. Malignant bowel obstruction (MBO) is a common complication for women with recurrent ovarian cancer that causes protracted and debilitating symptoms. Recognising variation in clinical practice and the unmet need for evidence-based treatment, I conducted a literature review to summarise current treatment strategies for MBO in women with advanced gynaecological cancers from a multidisciplinary perspective. A pilot interprofessional MBO program was developed by the MBO working group (which I co-led with Dr Stephanie Lheureux) and implemented in June 2016 at Princess Margaret Cancer Centre to support women who had, or were at risk, of developing MBO. The integrated model of care consisted of: i) standardised MBO symptom triage tools; ii) establishment of MBO multidisciplinary case conferences; iii) consensus on MBO care algorithms for in-patients and out-patients; iv) development of patient education materials for MBO; and v) prompt access to allied health professionals and supported advanced care planning. To assess the impact of the interprofessional MBO program, I reviewed all consecutive patients presenting with MBO from April 2014 to March 2018 (i.e. before and after the implementation of the program) and compared their outcomes. Of the 169 patients included, the majority (n=124, 73%) had recurrent ovarian cancer. There were 106 patients admitted prior to implementation of the MBO program (baseline group) whilst 63 patients were managed under the MBO program (MBO program group). Overall, the MBO program group had a significantly shorter average accumulated hospital length of stay (LOSsum) by 9 days (13 vs 22 days, adjusted P=0.006). Furthermore, their median overall survival post MBO diagnosis was approximately 5 months longer compared to the baseline group (243 vs 99 days, P=0.002). This retrospective, single institution study suggests a beneficial impact towards improving the complex care of women with advanced gynaecological cancers who developed MBO. Following on from this retrospective study, I developed a prospective MBO study incorporating patient reported outcomes to validate these findings, which is currently recruiting (MAMBO study, N=61/150 NCT03260647). In conclusion, the body of academic work carried out in this thesis has addressed known clinical gaps in ovarian cancer care throughout the disease trajectory and generated specific care recommendations to guide risk-reducing surgery management, to improve symptom burden whilst undergoing cancer treatment, and to improve management of malignant bowel obstruction. Broadly, this research will help clinicians, peak bodies and health funders implement evidence based care and institutional and national policies to facilitate better care provision for patients with ovarian cancer.

The incidence of oesophageal adenocarcinoma (OAC) has risen rapidly over the last four decades and has a high overall mortality rate that has shown only incremental improvements over the same duration. OAC develops from the precursor intestinal metaplasia of the oesophageal mucosa known as Barrett’s oesophagus. However, limited knowledge of the molecular mechanisms driving disease progression makes effective clinical management of OAC challenging. One of the common genetic events associated with the progression from Barrett’s oesophagus to OAC is loss of the tumour suppressor, SMAD4 (mutated in 13% or loss of function in 34% of OAC cases). Herein, this thesis firstly investigated the effect of SMAD4 inactivation in Barrett’s carcinogenesis. Genetic knockdown or knockout of SMAD4 was sufficient to initiate tumourigenesis of dysplastic Barrett’s oesophagus cell line, CP-B, in vivo, establishing SMAD4 loss as a crucial event driving progression to OAC. Further, low coverage whole genome sequencing (LC-WGS) analysis revealed that tumourigenic SMAD4 knockdown/knockout CP-B cell lines exhibited distinctive and consistent copy number alterations (CNAs) compared to non-tumourigenic SMAD4 wild-type parental cells. Amongst the alterations we observed were loss of chromosome arm 14q, while amplified regions include chromosome arms 6q and 12p, consistent with common CNAs found in patient tumours. This high genomic instability, characterized by structural chromosomal rearrangements within the tumours following SMAD4 loss, implicates SMAD4 as a protector of genome integrity in OAC development and progression. Moreover, initial in vitro data shows that SMAD4 knockout in CP-B cell line, results in differential expression of transforming growth factor-beta (TGF-beta) pathway target genes (such as ACTA2, CRYAB, PTK2B, ATF3 and CDC6) and loss of cell cycle arrest in response to TGF-beta1 cytokine compared to SMAD4 wild-type parental cells. Furthermore, SMAD4 knockout negatively regulated transcript expression of the multifunctional tumour suppressor INK4/ARF locus, demonstrating the novel and complex network of SMAD4 tumour suppressive activity. This thesis also focused on deciphering the functional role of growth factor receptor bound protein 7 (GRB7) amplification and overexpression in OAC and its potential targeting. GRB7 gene is positioned within known 17q12 amplicon, together with HER2 gene encoding for human epidermal growth factor receptor 2 (HER2). GRB7 is an adaptor molecule that mediates networking of multiple cell surface receptors with downstream signalling pathways. GRB7 high expression was found to be associated with worse survival in OAC patient cohort. Further, genetic GRB7 knockdown (siRNA) inhibited cell proliferation and clonogenic survival and induced apoptosis in GRB7 amplified and overexpressing OAC cell lines (OE19 and Eso26), without altering proliferation of the cells with normal GRB7 expression. Furthermore, whilst HER2 amplification and overexpression was also observed in OE19 and Eso26 cells, they were not uniquely sensitive to trastuzumab (HER2 inhibitor), with Eso26 cells exhibiting resistance in vitro. Taken together, initial findings raise the possibility that GRB7 may be a better molecular therapeutic target than HER2 in OAC with the 17q12 amplicon. To address this possibility, OE19 and Eso26 cell line xenograft models with inducible expression of shRNA targeting GRB7 were used. Consistent with in vitro findings, HER2 expression did not predict sensitivity to trastuzumab, with Eso26 xenografts remaining refractory to trastuzumab treatment. Of high importance, mimicking GRB7 inhibition with inducible-shRNA significantly inhibited tumour growth in both OE19 and Eso26 xenografts. Thus, this part of the thesis demonstrates the functional role of GRB7 overexpression as an oncogenic driver independent of HER2. In summary, the identification of functional genetic drivers and a deeper understanding of the mechanisms that underlie tumour progression in Barrett’s carcinogenesis will lead to improved strategies for the clinical management of OAC patients. To this end, SMAD4 loss was sufficient for progression from dysplasia to OAC. Tumours driven by SMAD4 loss exhibit distinctive CNAs consistent with OAC and metastatic potential. In addition, GRB7 high expression predicts poor outcome in patients with OAC and as such, GRB7 represents an oncogenic driver that could be used as a therapeutic target. Crucially, this thesis provides in vitro and in vivo preclinical and molecular biology evidence for the potential therapeutic benefit of targeting GRB7 in cancer.

The Hippo pathway is an important regulator of organ growth during development where it ensures correct scaling of tissues and organs. It is well-conserved in animals and deregulation of the pathway has been found to drive development of a range of cancers. The Hippo pathway was first identified in the fruit fly, Drosophila melanogaster, and has been extensively studied in a number of different tissues. As well as controlling organ growth in Drosophila, the Hippo pathway has been repurposed to control the binary cell fate choice of the R8 photoreceptor cell in the Drosophila eye. The R8 photoreceptor cell is one of eight photoreceptor cell types in the Drosophila eye and is responsible for much of colour vision in Drosophila. There are two main subtypes of R8 cells – around 30% are the 'pale' (p) subtype and express Rhodopsin 5 (Rh5; responsive to blue light); the remaining 70% are the 'yellow' (y) subtype and express Rh6 (responsive to green light). The Hippo pathway acts in a bistable feedback loop to both specify and maintain R8 subtype specification. This bistable feedback loop is composed of the Hippo pathway kinase, Warts, the downstream transcriptional coactivator, Yorkie, and the Pleckstrin-homology domain containing protein, Melted. Yorkie and Melted both promote pR8 cell fate, while Warts represses pR8 cell fate and promotes yR8 cell fate. Although we have begun to study the role of the Hippo pathway in R8 cell fate choices, we still lack a clear understanding how the Hippo pathway functions in these cells. I sought to further our understanding of the Hippo pathway in R8 cells by answering three key questions: (1) Which Hippo pathway proteins control R8 cell fate? (2) What is the subcellular localisation of Hippo pathway components in R8 cells? (3) What are the target genes of the Hippo pathway in each R8 subtype? (1) I used genetic experiments to modulate the expression of Hippo pathway genes and found that Crumbs and Cka promote pR8 cell fate, while the apical spectrin cytoskeleton and the 14-3-3 proteins promote yR8 cell fate. Modulating other Hippo pathway regulators did not obviously alter the ratio of R8 subtypes, suggesting that not all Hippo pathway proteins function in R8 cell fate choices. (2) Using confocal and multiphoton microscopy, I showed that only Warts differed in expression or localisation between R8 subtypes, suggesting that regulation of Warts levels in R8 cells controls the Hippo pathway's role in R8 cell fate. I also showed that the upstream Hippo pathway proteins, Tao, Merlin and alpha-spectrin are relocalised between late pupal and adult photoreceptor cells, hinting that there may be differences in their mechanisms of action between these two stages. (3) Targeted DamID-seq showed that Yorkie and its binding partner Scalloped reside at over a thousand loci in each R8 subtype. Yorkie and Scalloped target genes included known target genes from proliferating Drosophila tissues, as well as genes that are specific to photoreceptor cells. I also investigated the role of a target gene in pR8 cells, ninaB, and found that it promoted pR8 cell fate in male Drosophila retinas. These results provide insight into the role of the Hippo pathway in R8 cells and highlight similarities and differences between the roles of the Hippo pathway in proliferating cells and in R8 cell fate specification.

The appropriate regulation of gene expression programs is essential for normal cellular function. In cancer, mutations derail normal developmental gene expression programs resulting in a malignant epigenetic state. Traditionally, therapies have attempted to kill cancer cells by targeting rapid proliferation or by directly disrupting the mutated protein. More recent therapies, such as BET inhibitors, attempt to directly target the oncogenic epigenetic state. All of these therapeutic approaches are frequently hampered by the emergence of drug resistance. In many cases, resistance is acquired through genetic changes. However, there is increasing evidence that drug resistance can also arise through non-genetic/epigenetic mechanisms. This has major implications for cancer treatment, as the processes that drive genetic and non-genetic resistance are completely different and epigenetic changes, unlike genetic changes, are potentially reversible. Our laboratory previously generated a model of non-genetic resistance to BET inhibitors in acute myeloid leukaemia (AML). Resistance coincided with the acquisition of a less differentiated stem cell like phenotype and was stable upon drug withdrawal. To discover targets that can overcome the resistant epigenetic state, I performed a focused CRISPR-Cas9 screen, which identified the enhancer regulator, LSD1, as key to maintaining the resistance phenotype. Treatment with an LSD1 inhibitor overcame resistance through time-dependent reprogramming of the resistant cells. This epigenetic reprogramming resulted in both differentiation and the formation of new enhancers around critical BET inhibitor target genes. Through functional genomics experiments, I demonstrated that re-sensitization of the resistant cells was driven by the new enhancer formation, rather than differentiation. Mechanistically, the enhancer remodeling is caused by upregulation of Irf8, which together with the pioneer factor, Pu.1, initiates the formation of new enhancers around BET inhibitor target genes, leading to a restored BET inhibitor transcriptional response. Motivated by the poor clinical responses and rapid acquisition of resistance to BET inhibitors in AML, I also sought to develop an assay that could identify new therapeutic targets that may be more effective. Transcription factors (TFs) are ideal targets, however they are difficult to disrupt directly. To circumvent this, I developed an assay that can identify what cofactors (which are often druggable) are required to drive the transcriptional activity of a given TF. This assay works by combining the Gal4 transactivation system with CRISPR-Cas9 screening. Preliminary screens using the transcription factors, VP64 and MYB, were able to identify previously validated cofactors for each TF. VP64 was dependent on MED25 and MYB was dependent on P300. A number of other potential specific cofactors were identified and will be validated in future work. This assay not only provides the potential to identify novel therapeutic targets, but will also provide insight into the poorly characterized interface between TFs and cofactors.