Adoptive cell therapy using chimeric antigen receptor (CAR) T cells has shown great success in haematological malignancies, leading to recent approval of two CD19-specific CAR T cell products by the Food and Drug Administration (FDA) for the treatment of B cell Acute Lymphoblastic Leukaemia (ALL) and non-Hodgkin lymphoma. However, despite this high success in CD19+ malignancies, clinical trials for solid cancers have yielded only limited efficacy to date. This discrepancy is thought to be due to, in part, the immunosuppressive tumour microenvironment and antigen heterogeneity in solid tumours. Therefore strategies to overcome these challenges are necessary for CAR T cells to be effective in solid malignancies. In this thesis, we have investigated two approaches aimed at augmenting CAR T cell therapy in a solid cancer setting. The first approach involved the activation of a costimulatory pathway known as CD137 (4-1BB). Targeting this pathway using an agonist α-4- 1BB antibody has been reported to induce robust co-stimulatory signals for diversifying and enhancing T cell responses, which led us to hypothesise that combining CAR T cells and α-4- 1BB antibody would result in improved therapeutic responses. Herein, we utilised a human- Her2 self-antigen mouse model to assess the therapeutic efficacy of Her2-specific CAR T cells against Her2-expressing solid tumours. We report that the use of an α-4-1BB antibody significantly increased CAR T cell anti-tumour responses against two established Her2+ solid tumours. Our analysis revealed an increase in expression of IFNγ and the proliferation marker Ki67 in intratumoural CAR T cells following treatment with α-4-1BB antibody. Strikingly, we also demonstrated that treatment with an α-4-1BB antibody greatly decreased host immunosuppressive cells at the tumour site, including regulatory T cells and myeloidderived suppressor cells, which correlated with an increased therapeutic response. Interestingly, further investigation in immunocompromised RAG-/- mice showed reduced synergistic effects following the combination therapy, suggesting a contribution of endogenous T cells in mediating these effects. Furthermore, we observed an increased frequency of dendritic cells (DCs) and the endogenous T cells exhibiting an effector CD44hi CD62Llow phenotype in tumour-draining lymph nodes following α-4-1BB antibody stimulation, further supporting the potential involvement of the endogenous immune cells. Taken together, these data indicate an important contribution of the endogenous immune system in enhancing overall CAR T cell efficacy, and thus formed the basis for our second approach that aimed to improve CAR T cell therapy by harnessing the endogenous DC population. Herein, we genetically modified Her2-specific CAR T cells to constitutively secrete a cytokine Flt3L, that is known to be critical for the differentiation of an important subset of DCs, the cross-presenting Batf3-dependent XCR1+ IRF8+ CD103+ DCs. We hypothesised that inducing the differentiation of these CD103+ DCs would increase the overall therapeutic efficacy of CAR T cells. We successfully generated Flt3L-secreting CAR T cells, and demonstrated that CAR T cell-derived Flt3L was able to induce expansion of CD103+ DCs and enhance their T cell stimulatory capacity. Adoptive transfer of Flt3L-secreting CAR T cells led to a significant enhancement of tumour growth inhibition, and subsequently enhanced mouse survival compared to the control CAR T cells. Similarly, using a TCR-based adoptive cellular therapy approach, namely the transfer of OT-I T cells against an OVAexpressing tumour, led to superior anti-tumour responses, particularly with the addition of poly I:C adjuvant that resulted in the most significant suppression of tumour growth. In summary, results presented herein demonstrate the promising potential of combining CAR T cells with immunomodulatory agents for enhancing the efficacy of CAR T cells against solid tumours. Our results indicate that in addition to directly augmenting CAR T cell activity, immune-modulating agents are able to enhance the therapeutic efficacy of CAR T cells by modulating the endogenous immune response. Our studies have established a strong rationale for combining CAR T cells and immunomodulatory agents in the treatment of solid cancer, and therefore warrant clinical investigation of these combination therapies.

Self-reported ‘taste’ problems are common in people receiving chemotherapy and have implications for nutritional and psychosocial domains. Taste refers to the perception derived when chemical molecules stimulate taste receptor fields in the oral cavity whereas flavour perception involves at least three independent sensory systems including taste, smell and texture. Conflation of the terms taste and flavour contributes to clinician confusion and reduces the opportunity to develop effective strategies to address taste problems in cancer patients. The research reported in this thesis aimed to characterise the extent to which taste function and food hedonics contribute to the eating and drinking experience in people receiving chemotherapy and to investigate how this is managed in the clinical setting. There was a mixed methods approach to this program of research that was comprised of three separate studies, in two phases, designed to: Phase 1, qualitative 1. Describe current practice surrounding taste function and food hedonics in the clinical oncology setting (Study 1); 2. Understand the experience and consequences of altered taste function and food hedonics for people receiving chemotherapy (Study 2); and Phase 2, quantitative 3. Describe the patterns of altered taste function and food hedonics across the chemotherapy treatment trajectory (Study 3). In the Phase 1 studies, patient and clinician interviews were used to explore a) clinician practice regarding the management of taste problems and b) patient and carer descriptions, experiences and consequences of taste changes. This qualitative phase informed the quantitative phase of the research: a longitudinal study of 52 women receiving chemotherapy for breast cancer that assessed taste function, appetite and food liking six times from before chemotherapy to 2 months after chemotherapy and investigated whether changes in these outcomes were associated with dietary intake, nutritional status or social dining activity. Phase 1 findings demonstrated that ‘taste’ problems refer to a raft of issues related to the wider aspects of flavour including changes to the sense of smell or touch, or to problems with appetite or food liking. Clinicians have limited capacity to distinguish between these scenarios. These changes shaped what patients chose to eat, drink, cook and purchase, and influenced how they dined and how they felt. A need for new approaches to classifying and describing flavour problems was identified and a requirement for better quality information and evidence with which to guide patients was indicated. In Phase 2, patterns of taste and hedonic changes were characterised in an unprecedented fashion across the treatment trajectory. Findings from Phase 2 analyses showed taste function and food hedonics were adversely influenced with greatest change closest to chemotherapy administration, followed by a gradual return to baseline measures. Problems resolved by 2 months after completion of chemotherapy. Change from baseline in ability to correctly identify all tastants was significant early in the third chemotherapy cycle (difference = 18.2%; 95% CI = 2.7, 32.9; p = 0.02) and final chemotherapy cycle (difference = 19.6%; 95% CI = 3.0, 35.1; p = 0.02). Decreased liking of sweet food (chocolate) was observed in the early (d = 0.77; p = 0.002) and middle stages of the third chemotherapy cycle (d = 0.70; p = 0.003) and early in the final chemotherapy cycle (d = 0.89; p = 0.001). Appetite was significantly decreased from baseline early in both the third and final chemotherapy cycles (d = 1.02; p < 0.001 respectively). Associations were found between taste and hedonic changes and dietary intake, nutritional status and social dining. Change in ability to correctly identify tastants was associated with reduced energy intake (r = 0.32; p = 0.005) early in the third chemotherapy cycle. At this assessment point, decreased liking of a sweet food item was also associated with reduced energy intake (r = 0.35; p = 0.001). Appetite loss was associated with reduced energy (r = 0.35; p = 0.001) and protein intake (r = 0.36; p = 0.001) early in a chemotherapy cycle, decrease in BMI over the study period (r = 0.36; p = 0.001), and change in overall nutritional status as assessed by PG-SGA score (r = 0.18; p = 0.09). Early in the final chemotherapy cycle, small-sized but non-significant associations were seen between taste change and social dining episodes (r = 0.22, p = 0.09) and between appetite loss and social dining episodes (r = 0.18, p = 0.16). Clinicians are ill equipped to support patients who report taste problems due to an absence of assessment tools or classification symptoms to identify problems described colloquially as ‘taste’. An emerging taxonomy of taste arising from this research goes some way to address the need for a classification system linking patient language to specific sensory or hedonic disturbances. Research findings will be used to guide the development of more specific pre-chemotherapy education material for patients.

Barrett’s oesophagus (BO), a metaplastic premalignant lesion, and gastroesophageal reflux disease (GORD) are two major risk factors for the development of oesophageal adenocarcinoma (OAC), an aggressive cancer the incidence of which is increasing faster than any other malignancy in the Western world. Proteases are known to be present in the refluxed contents of GORD patients and in inflamed BO and OAC tissue microenvironment. Yet, their role in these oesophageal pathologies is still poorly understood. Many (patho)-physiological responses induced by proteases can occur through their cell surface targets, the protease-activated receptors (PARs). Aberrant PAR expression and signalling have been reported in several malignancies. However, the tissue distribution patterns of PARs and their activating proteases as well as the effects of protease/PAR activation on oesophageal pathophysiology are largely unknown. Therefore, the general aim of this thesis was to investigate the role of PARs and proteases in BO and OAC pathogenesis. More specifically, this thesis aims to: 1) characterise the expression of PAR family members (PAR1-PAR4) in a panel of normal oesophageal mucosa, BO, and OAC tissues, epithelial cell lines, and tissue-derived fibroblasts; 2) characterise the expression and localisation of candidate PAR activating proteases - MMP1, trypsin-1, KLK5, and KLK6 in normal oesophageal mucosa, BO, and OAC tissues; and 3) investigate the signalling and functional effects of PAR1 and PAR2 activation in oesophageal epithelial cells. In Chapter 3, reverse transcription quantitative PCR (RT-qPCR) and immunohistochemical analyses revealed that PAR1 transcript and protein expression were significantly upregulated in BO and OAC tissues compared to normal oesophageal mucosa. PAR1 expression levels correlated significantly with disease progression. PAR3 and PAR4 mRNA overexpression were likewise detected; the former in BO and OAC tissues, and the later in tumour samples. PAR2 transcript and protein were found in all oesophageal tissue types with no statistically significant differences in expression levels. PAR1 and PAR2 expression within BO and OAC tissues were localised in various cell types but primarily in the epithelial cells. While BO and OAC tissues showed significant increase in MMP1 and trypsin-1 expression, comparable levels of KLK5 protein was noted in all tissue types (Chapter 4). KLK6 transcript was markedly downregulated in BO tissues. Strikingly, MMP1, trypsin-1, and KLK5 were frequently coexpressed with PAR1 and PAR2 in OAC and BO tissues. Analysis of the effects of PAR activation by the aforementioned proteases in a panel of representative epithelial cell lines revealed that PAR1- and PAR2-mediated signalling were dysregulated in OAC cells (Chapters 5 and 6). For instance, MMP1 but not thrombin induced Ca2+ signalling via PAR1 in OAC cells, whereas thrombin but not MMP1 evoked Ca2+ responses in normal and BO cells. In addition, while PAR2 activation triggered transient ERK1/2 signalling in normal and BO cells, ERK1/2 activation was sustained in OAC cells. Moreover, PAR1 and PAR2 stimulation activated PI3K/AKT signalling exclusively in OAC cells. Functionally, PAR1 and PAR2 activation promoted OAC cell invasion and chemotaxis, respectively. In all, these results highlight the PARs as potential players in OAC and BO pathogenesis and thus, potential therapeutic targets for these oesophageal pathologies.

Breast cancer is the second most common form of malignancy diagnosed in Australian women, imposing an enormous social and economic burden on society. If the cancer spreads to secondary locations, patient survival decreases dramatically. Therapeutic strategies for treatment of metastatic disease are desperately needed in breast cancer. Recently we have shown that when Myb is conditionally deleted from the mammary gland of MMTV-Neu and MMTV-PyMT mice, tumour formation is ablated. To provide further insight into the function of Myb in mammary gland development, cre-mediated conditional knock out of c-myb in the mouse mammary gland was examined. The conditional deletion of c-myb led to a reduction in branching and terminal end bud formation. These data indicate that if MYB could be inhibited for breast cancer therapy, there are potentially few side effects on the normal mammary gland. The ability to inhibit DNA binding transcription factors is a long sort after goal in oncology, as their importance in disease initiation and progression is well documented. To target MYB, we have developed a DNA vaccine. Preclinical studies have largely examined the MYB DNA vaccine in the context of colon cancer models using prophylactic vaccination. However, preliminary data indicate that it may also be effective in reducing the metastatic burden in preclinical breast cancer models. This thesis aims to further investigate the role of MYB during mammary gland development and provide insights into its involvement in tumourigenesis. Furthermore, the MYB DNA vaccine will be assessed for its effectiveness as a therapeutic treatment in clinically relevant surgical models of metastatic breast cancer, as well as further development as a therapeutic in the setting of colon cancer.

Amongst all head and neck squamous cell carcinomas (HNSCC), oral tongue carcinomas (OTSCC) have the worst prognosis for early stage disease. The current staging system is unable to consistently identify patients with high risk disease. In one of the largest comprehensively annotated OTSCC cohorts to date, this thesis examines a combination of literature identified candidate biomarkers, and also sought to determine if a novel prognostic molecular signature could be identified. Given the high frequency of reported CDKN2A alterations in HNSCC, a comprehensive analysis of the differential mechanisms of CDKN2A alteration was performed. Promoter methylation status, mutation status, copy number variation and protein expression were assessed. The majority of samples (95%) did not demonstrate p16 over-expression assessed by immunohistochemistry. Although disruption of CDKN2A was found to be a frequent event arising from a variety of mechanisms, no correlation between CDKN2A alteration and clinicopathological features was found. A panel of loci frequently reported to be hypermethylated in HNSCC was investigated within the OTSCC cohort, with three quantitative methodologies that assessed DNA methylation. In contrast to the literature, these loci were not commonly methylated. Findings were confirmed in an external cohort of HNSCC samples from The Cancer Genome Atlas (TCGA) that had methylation levels quantified with a fourth, orthogonal methodology. The use of non-quantitative methodology in the literature was the likely cause for the overestimation of significant methylation events, with this study highlighting the need for the cautious interpretation of this literature. Utilising a genome-scale wide methylation platform, a prognostic methylation signature was sought for the OTSCC cohort. Given the absence of consensus on data analysis, comprehensive bioinformatics analyses were performed utilising multiple contemporary R software library packages, to enable a thorough examination of the data with published algorithms used for pre-processing and downstream analysis. However, methylation assessed over greater than 450,000 CpG dinucleotides did not reveal a differentially methylated group of samples, and was not informative for clinicopathological variables. Furthermore, despite increasing the total number of samples to include the TCGA OTSCC dataset, a prognostic methylation signature was not identified in any cohort. Targeted mutational profiling of the cohort was also performed. A disproportionately large number of variant calls were identified on the initial processing of samples. Validation of a subset of variant calls with orthogonal methodology was able to confirm the presence of only 12/50 (24%) selected mutations. Despite pre-analytical quality assurance assessments, the replicated analysis of samples and the use of stringent filtering criteria, the presence of artefactual variant calls masked the identification of true mutations. The likely source of the large number of artefactual variant calls was from the PCR based amplification of artefact in DNA extracted from formalin-fixed, paraffin-embedded tissue. This thesis emphasises the importance and the impact of the choice of methodology on the successful identification of clinically relevant biomarkers. Within the limitations of current understanding and the size of the cohort examined, it also suggests that both CDKN2A alteration and DNA hypermethylation in isolation are not prognostically informative biomarkers for OTSCC. Further research is required into the prognostic value of other molecular alterations and the combined impact of simultaneous aberrations.

Over the last decade the management of colorectal cancer has changed significantly with the benefits of neoadjuvant therapies and new adjuvant treatments becoming apparent. Surgical strategies have also evolved with initial evidence that some patients can be successfully managed with local excision or omission of any surgery at all, resulting in a shift towards the individualisation of cancer management. The management of rectal cancer is based on primary staging assessment which relies on imaging techniques such as CT, MRI and ERUS. Recent advances in technology have improved the accuracy and widened the applications of these techniques. With the progress in medical and surgical treatments for rectal cancer, the optimal management of rectal cancer has become more complex. The evolving ability to tailor optimal treatment to the individual has created new roles for imaging such as prediction of response to treatment, restaging with assessment of response to treatment and prediction of prognosis. Consequently, prediction of response will become an important component of modern pre-operative assessment of rectal cancer to optimise individualisation of medical and surgical treatment. Beyond the established role of primary staging of malignancies, the role of conventional imaging techniques in re-staging following neoadjuvant treatment may be of increasing importance. Novel functional imaging techniques such as FDG-PET, DW (diffusion weighted) MRI are also emerging, the roles of which are yet to be determined. This thesis will examine the current status of bio-imaging and explore new imaging techniques in rectal cancer. At the Peter MacCallum Cancer Centre, we have been routinely performing staging and restaging imaging with CT, MRI and PET for the last 5 years which has resulted in a cohort of patients in whom these imaging techniques can be evaluated. This thesis also aims to evaluate a recent and evolving functional imaging technique- DW-MRI, in the prediction of response of rectal cancer to chemo-radiation.

Cancer is a cooperative process, involving mutations in multiple genes. Activation of a cancer-driving gene, the Ras small GTPase, via a mutation that locks Ras in the GTP-bound active form (RasV12), occurs in ~30% of human cancers. However, alone it is not sufficient for tumour formation. A loss of function screen previously performed in the vinegar fly, Drosophila melanogaster, identified 947 genes that potentiate RasV12-mediated tumourigenesis and metastasis (Zoranovic, et al. in prep.). This list has been narrowed down to 234 genes that 1) show increased tumourigenicity with RasV12 in vivo, 2) are in the top 100 genes down-regulated in human cancer, and 3) are known to regulate the cytoskeleton, polarity, adhesion or cell motility. This study has successfully confirmed involvement of autophagy-related genes Atg8a, Atg7 in regulating RasV12-mediated proliferation in the Drosophila eye epithelial tissue using the UAS/GAL4 system. The study identified the autophagy-related genes Atg1, Atg3, Atg4, Atg5, Atg6, Atg7, Atg8a, Atg12 and Atg101 that when knocked down cooperate with RasV12 and lead to increased tissue overgrowth in the Drosophila eye epithelium. Atg8a was chosen as the representative target gene to investigate this cooperation. It was observed that Atg8a cooperates with RasV12 through the Raf pathway. The role of p62 in this Ras-mediated cooperation with Atg8a was also examined and it was found that p62 levels increase in RasV12+ Atg8aRNAi expressing tissue in comparison with control. Investigations were also carried out to ascertain if knockdown of Atg genes cooperate with Ras through the JNK pathway. It was discovered that in the presence of oncogenic Ras, knock down of Atg8a increases the expression of the JNK target MMP1. The finding of this work could lead to use of this autophagy related genes as prognostic markers in Ras-driven oncogenesis and might reveal effective therapeutic targets to combat this deadly disease.

Natural killer (NK) cells and cytotoxic T lymphocytes (CTLs), collectively referred to as cytotoxic lymphocytes (CLs), are responsible for clearing virus infected and cancerous cells. The predominant manner by which CLs do this is through the delivery of the pore forming protein, perforin, and pro-apoptotic granzymes that synergise to induce apoptosis in a conjugated target cell. Humans who inherit bi-allelic inactivating perforin mutations develop the immunoregulatory disease familial haemophagocytic lymphohistiocytosis (FHL) and/or haematological malignancies, demonstrating the critical importance of expressing functional perforin for the maintenance of immune homeostasis and tumour immune surveillance. While most disease associated perforin mutations are rare, 8-9% of the Caucasian population are carriers of polymorphism A91V (rs35947132, 272C>T). It has been suggested that >50% of individuals homozygous for the A91V allele develop FHL and/or cancer and that individuals heterozygous for A91V have an increased susceptibility to ALL. Despite its frequency and disease association, it remained unknown whether heterozygous inheritance of the A91V allele impairs human CL cytotoxicity and, more broadly, whether perforin is rate limiting in CL cytotoxicity. Here, it has been demonstrated that NK cells from healthy humans heterozygous for the A91V allele show an almost 50% reduction in cytotoxicity compared to individuals homozygous for WT perforin. This reduction in function was due to A91V perforin protein being misfolded within human primary NK cells. Moreover, it was also observed that heterozygous perforin knockout mouse CTLs showed an ~50% reduction in cytotoxicity. Taken together, these data demonstrate that perforin is indeed rate limiting for CLs cytotoxicity and therefore, individuals heterozygous for defective perforin alleles have impaired CL function. Although FHL predominantly presents shortly after birth, a subset of patients present at an age greater than three years, owing to the expression of misfolded perforin variants. Previous studies have shown that when transiently expressed in CTLs, perforin variants associated with late onset disease failed to traffic within CTLs and the cells remained non-functional. Therefore, it was unknown how patient CTLs expressing these variants could avoid FHL in infancy, and maintain a level of immune homeostasis for many years, or even decades. Here, it is shown that perforin variants associated with late onset disease can fold correctly and traffic within CTLs, and thus provide a significant level of cytotoxic function. However, this function was found to be lost if CTLs were cultured at an increased temperature (39 ̊C). Taken together, these data suggest that the CTLs of late onset FHL patients may have sufficient cytotoxicity to delay FHL onset in infancy. However, prolonged fever and, potentially, a more rapid exhaustion of the limited pool of correctly folded perforin mutants may result in the loss of CTL function, leading to FHL and cancer later in life. Prior to its secretion from the CL, the evolutionarily conserved C-terminal residues of perforin are proteolytically cleaved. The functional significance of C- terminal processing has remained controversial. Here it is shown that perforin enriched from human NK cell with an intact and glycosylated C-terminus was not cytotoxic. However, removal of the C-terminal glycan from the protein was found to completely restore function. As full-length deglycosylated perforin has wild type activity, these data suggest that C-terminal cleavage of perforin is permissive for cytotoxic function due to removal of an inhibitory N-linked glycan moiety at the C-terminus of the protein. These findings position the protease(s) responsible for perforin cleavage as critical to CL function. In summary, the studies described in this thesis have added to the understanding to how perforin mutations affect CL cytotoxicity and described a critical final step in perforin maturation. Together, these advancements in perforin biology may contribute to the treatment of disease arising from perforin deficiency and also define new factors critical for CL function and human health.

This study focuses on how pericytes can contribute to the epidermal cell microenvironment in human skin, using a 3D organotypic co-culture model, I have shown that in the presence of pericytes, epidermal cells regenerate an epithelium that more closely resemble normal healthy skin with respect to basal cell morphology and various biochemical characteristics. Characterization of pericytes in skin ageing and their functionality in organotypic co-culture model suggests a microenvironmental modulatory role of pericytes in skin ageing and epithelial tissue repair.

Breast cancer is a highly prevalent disease that, like many cancers, lacks effective therapies aimed at treating and preventing metastasis. Harnessing the host immune system to recognise and eliminate malignant cells has recently emerged as an effective therapeutic strategy in many cancers. However, response rates to these approved immunotherapies remain modest in the absence of a more detailed understanding of tumour immunity. The type I interferons are a family of cytokines that have long been understood to enhance the immune response to cancers, though their clinical application has led to underwhelming results in numerous types of cancer. This thesis provides new evidence that proposes the re-visitation of cancer immunotherapies that stimulate the type I interferon pathway. We show that host-derived type interferon is critical for the suppression of breast cancer metastasis through natural- killer cell activation. Induction of a type I IFN response by administering agents that mimic a viral infection (poly(I:C), a double-stranded RNA analog) proved to be powerful anti-metastatic agents in multiple pre-clinical models of triple-negative breast cancer (TNBC). This was linked to widespread immune activation which conferred NK cells with enhanced cytotoxic function to eliminate disseminated tumour cells. The efficacy of this novel immunotherapeutic approach was also found to rely upon the treatment setting in which it was used. Evidence is presented that demonstrates administration prior to primary tumour removal (neo-adjuvant therapy) as the only effective therapeutic regimen. We propose that such immunotherapies are most effective at eliminating circulating and early disseminated cells rather than established metastatic lesions. This provides some explanation to the inefficacy of previous interferon trials that were conducted in patients with late-stage metastatic disease. It also calls into question whether other immunotherapies could be used earlier in cancer treatment to maximise the chances of a clinical response. Finally, we uncover that expression of IRF9, a key transcription factor in the type-I interferon signaling pathway, accurately predicts TNBC patient prognosis. Loss of IRF9 in a patient’s primary tumour predicted significantly poorer overall survival due to metastatic spread. As we show that tumour cells are not directly responsible for the poly(I:C)-induced interferon response, we propose that patients with IRF9-negative TNBC could benefit from neo-adjuvant interferon-based immunotherapy.

Metastasis is a complex process that is responsible for most human cancer deaths. Autopsy findings across multiple cancers suggest that metastatic dissemination from the primary site of disease to distant organs is not a random process; cancers can have different predilections to metastasize to particular organs. These observations raise the general question of whether determinants of metastasis, including organ-specificity, are driven by cancer cell-intrinsic or cell-extrinsic factors, or a combination of both. A fundamental question in cancer biology is whether metastasis capability in human tumours is acquired in a heritable manner. Work pertaining to this thesis tested whether organ-specificity during melanoma metastasis is a biologically heritable trait, using PDX modelling. Collectively, these studies suggest that while metastatic potential per se is a biologically heritable trait, organ-specificity of metastasis is not. To examine a candidate cell-intrinsic regulator of melanoma metastasis, desmoglein 2 (DSG2), expression and correlative studies were also performed on patient melanomas. DSG2, a member of the cadherin class of proteins that mediate cell-cell contacts, has been previously but inconsistently detected in human melanoma, however its role and function remain poorly defined. Herein, DSG2 expression was investigated as a potential prognostic factor in melanoma linked to the development of vasculogenic mimicry and poor patient outcomes. These studies confirm that DSG2 is expressed in primary and metastatic melanomas, but not in normal cutaneous melanocytes, and is associated with poor survival in melanoma. Although DSG2-positive melanomas consistently contained more features of VM than DSG2-negative, the differences were not statistically significant.

Gastric cancer (GC) is one of the most common malignancies and is frequently fatal. Epidemiological and biological evidence suggests GC is heterogeneous, with four subtypes of GC identified by the Cancer Genome Atlas (TCGA) in 2014, however currently it is treated as one disease. The aim of this thesis is to investigate how the clinical and biological inter-tumoural heterogeneity of GC affects survival. The clinical predictors of outcome were investigated for a cohort of curatively treated GC patients, revealing that age, stage of disease and Lauren histological subtype predict overall, cancer specific and relapse free survival. The molecular differences between different GC stages, and histological subtype (intestinal gastric cancer (IGC) compared to diffuse gastric cancer (DGC)) were then investigated with the aim of understanding the biological basis underlying survival, relapse and response to treatment. Key molecular pathways were identified as being differentially expressed between DGC and IGC, and having impact on survival. These included the transforming growth factor beta (TGF-β) signalling pathway with the expression of three bone morphogenic protein (BMP) antagonists from this pathway having associations with survival on multivariate analysis, and the mismatch repair (MMR) pathway with four genes from this pathway having impact on survival. BMP and activin membrane bound inhibitor (BAMBI), a gene from the TGF-β pathway was more highly expressed in IGC compared to DGC and was associated with poorer survival in IGC. Molecular analysis of BAMBI in GC cell lines revealed decreased proliferation and invasion, and G0/G1 cell cycle arrest. It was postulated that this change in proliferation and invasion was mediated via alterations in the balance of epithelial-to-mesenchymal transition (EMT) and mesenchymal-to-epithelial transition via a TGF-β mediated process. Further analysis revealed gene expression correlation between BMP antagonists and EMT associated genes, revealing highest correlation with gremlin 1 (GREM1). Higher expression of GREM1 was also associated with poorer survival in IGC. Another key pathway discovered as differentially expressed between DGC and IGC was the MMR pathway. Microsatellite instability was also identified as a molecular subtype of GC by the TCGA. The clinical and histological correlations of the MMR GC subtype were therefore examined. Patients with MMR deficient tumours had improved survival compared to MMR stable tumours, and a higher inflammatory score (IS). The simple IS used in this thesis could be a triaging tool to identify tumours to take forward for formal MMR testing. This thesis adds to the growing body of evidence that demonstrates GC is heterogeneous and elaborates on some of the molecular features that drive this inter-tumour heterogeneity. It identifies a number of genes, pathways and biological associations that impact survival, and potentially response to treatment. This work will help in the personalisation of medical therapy for a disease that can no longer be treated with a one-size-fits-all mentality for therapy. Future work is planned to expand the findings of this work, with the aim of translating these findings to the clinical setting, enabling better prognostication and more rational prescription of treatments for GC patients.