Pancreatic cancer is an aggressive disease with poor prognosis and a high fatality rate. Gemcitabine, the standard first-line chemotherapy for advanced disease, has negligible effects, necessitating the development of new therapies. We previously demonstrated that deletion of the anti-apoptotic gene E1B19K (AdΔ19K) in a replication-selective adenoviral mutant, caused synergistically-enhanced cell-killing when combined with low-dose DNA-damaging drugs in pancreatic cancer xenograft models. To delineate the cellular pathways targeted by the combination treatment we employed AdΔ19K and gemcitabine or irinotecan, with the goal of identifying cellular factors that are essential for the synergistic cell-killing. We hypothesised that AdΔ19K and DNA-damaging drugs act synergistically to deregulate cell-cycle mechanisms. Pancreatic cancer cell death induced by AdΔ19K and DNA-damaging drugs is apoptotic and time-dependent. AdΔ19K could not block DNA-damage responses (DDR) elicited by the drugs, despite virus-mediated degradation of the DDR factor Mre11. Mre11 siRNA-mediated knockdown augmented the synergistic cell death. Mitotic-index analysis in synchronised cells and immunofluorescence microscopy suggested that AdΔ19K promotes mitotic entry of gemcitabine-treated DNA-damaged cells. Moreover, AdΔ19K inhibited drug-induced accumulation of Claspin, a DDR protein whose degradation is required for checkpoint recovery. Treatment with AdΔ19K and gemcitabine accelerated Claspin degradation, and siRNA-mediated Claspin knockdown enhanced the synergistic cell death. Time-lapse microscopy in histoneH2B mCherry-expressing cells showed that AdΔ19K enhanced gemcitabine-induced mitotic catastrophe, characterised by prolonged mitosis, chromosome missegregation errors, cytokinesis failure and formation of multinucleated cells. Moreover, live-cell imaging revealed that the majority of cells treated with AdΔ19K and gemcitabine die before mitotic entry. 5 These findings suggest that E1B19K-deleted adenoviruses cannot prevent cell-cycle checkpoint responses elicited by DNA-damaging drugs, but enhance drug-induced cell death by downregulating DDR factors, such as Mre11 and Claspin. Additionally, the virus enhances mitotic catastrophe of DNA-damaged cells escaping cell-cycle checkpoints, eventually leading to increased apoptosis. Through these studies cellular pathways and factors involved in the synergistic cell killing were identified, that could be explored in the future to develop improved targeted therapies for pancreatic cancer.

Pancreatic cancer is often referred to as the “silent killer“ due to the asymptomatic nature of the disease in the early stages and the extremely poor prognosis overall. The average one-year survival rate for PDAC patients is 24% (American Cancer Society, facts and figures, 2010), decreasing to 5%-6% over 5 years (WHO report, Pancreatic cancer, 2010). Only 20% of patients are suitable for surgical resection at the time of diagnosis and treatment options available to PDAC patients have not improved significantly over the past few decades. Thus novel therapeutic approaches are essential to treat this disease. Our experimental, clinical and pre-clinical data suggest integrin αvβ6 may be a suitable target. Bioinformatics studies using the Pancreatic Expression Database revealed that the β6 gene (ITGB6) was highly up regulated in pancreatic ductal carcinoma (PDAC) compared with normal pancreas. Further analysis carried out showed that there was a significant correlation between ITGB6 expression at the mRNA level and survival in a cohort of 292 PDAC patients. Immunohistochemistry analysis on two separate patient cohorts (n=118 and n=147) showed that normal pancreas lacked αvβ6 expression whereas 91% of PDAC tissues expressed αvβ6 at the protein level. There was no significant correlation between αvβ6 expression and survival at the protein level in both cohorts of patients tested. Flow cytometry and Western blotting analyses on a panel of PDAC cell lines confirmed expression of αvβ6 in PDAC cell lines. This study investigated the functional role of αvβ6 in PDAC cell lines. Antibody mediated function blockade of αvβ6 significantly inhibited proliferation in a dose dependent manner, specifically in αvβ6 positive PDAC cell lines. A significant reduction in migration and invasion was also observed in a panel of αvβ6 positive PDAC cell lines when treated with an αvβ6 function-blocking antibody. αvβ6 targeted antibody mediated therapy in combination with gemcitabine significantly inhibited tumour growth in a physiologically relevant pre-clinical subcutaneous xenograft model of PDAC. These data reaffirms that αvβ6 is a potential novel therapeutic target and an αvβ6 specific function-blocking antibody can be used as a novel agent to treat pancreatic adenocarcinoma patients.

Activating mutations of the kras proto-oncogene are found in more than 90% of human pancreatic ductal adenocarcinoma (PDAC) and can result in increased activity of the NF-κB pathway, leading to constitutive production of proinflammatory cytokines such as TNF-α. Pancreatic cancer progression occurs through a series of pre-invasive lesions, pancreatic intraepithelial neoplasias (PanIN lesions), which progress into invasive carcinoma. The aim of this thesis is to understand the autocrine role of TNF-α produced by premalignant epithelial cells in pancreatic tumour progression. This cytokine has already been shown to be involved in the progression of cancer. The major hypothesis therefore tested was that TNF-α secreted by pre-malignant epithelial cells promotes the early stages of pancreatic carcinogenesis by sustaining an inflamed microenvironment. In the spontaneous kras+/LSL-G12D; pdx1-cre mouse model of pancreatic cancer, concomitant genetic deletion of the TNF-α/IKK2 pathway substantially delayed pancreatic cancer progression and resulted in downregulation of the classical Notch target genes hes1 and hey1. Cell lines from the different PanIN bearing mice were established and used to dissect the cooperation between TNF-α/IKK2 and Notch signalling during PanIN progression. Optimal expression of Notch target genes was induced upon TNF-α stimulation of the canonical NF-κB signalling pathway, in cooperation with basal Notch signals. Mechanistically, TNF-α stimulation resulted in phosphorylation of histone H3 at the hes1 promoter and this signal was lost upon ikk2 genetic deletion. HES1 suppressed the expression of pparg, which encodes for the anti-inflammatory nuclear receptor PPAR-γ. Thus, crosstalk between TNF-α/IKK2 and Notch sustained an intrinsic inflammatory profile of the transformed cells. The treatment of PanIN bearing mice with rosiglitazone, a PPAR-γ agonist, also delayed PanIN progression. A malignant cell-autonomous, low-grade inflammatory process was shown to operate from the very early stages of kras-driven pancreatic carcinogenesis, which may cooperate with the Notch signalling pathway to promote pancreatic cancer progression.

It is imperative that the cancer research community has the means with which to effectively locate, access, manage, analyse and interpret the plethora of data values being generated by novel technologies. This thesis addresses this unmet requirement by using pancreatic cancer and breast cancer as prototype malignancies to develop a generic integrative transcriptomic model. The analytical workflow was initially applied to publicly available pancreatic cancer data from multiple experimental types. The transcriptomic landscape of comparative groups was examined both in isolation and relative to each other. The main observations included (i) a clear separation of profiles based on experimental type, (ii) identification of three subgroups within normal tissue samples resected adjacent to pancreatic cancer, each showing disruptions to biofunctions previously associated with pancreatic cancer (iii) and that cell lines and xenograft models are not representative of changes occurring during pancreatic tumourigenesis. Previous studies examined transcriptomic profiles across 306 biological and experimental samples, including breast cancer. The plethora of clinical and survival data readily available for breast cancer, compared to the paucity of publicly available pancreatic cancer data, allowed for expansion of the pipeline’s infrastructure to include functionalities for cross-platform and survival analysis. Application of this enhanced pipeline to multiple cohorts of triple negative and basal-like breast cancers identified differential risk groups within these breast cancer subtypes. All of the main experimental findings of this thesis are being integrated with the Pancreatic Expression Database and the Breast Cancer Campaign Tissue Bank bioinformatics portal, which enhances the sharing capacity of this information and ensures its exposure to a wider audience.

Pancreatic ductal adenocarcinoma (PDAC) is characterised by an intense desmoplastic stromal response that can comprise 60 to 80% of tumour volume and has been implicated to be a factor in promoting tumour invasiveness and the poor prognosis associated with this cancer type. It is now well established that pancreatic stellate cells, which are vitamin A storing cells found in the periacinar spaces of the stroma in the normal gland, are primarily responsible for this desmoplastic reaction. Studying the interaction between stellate cells and cancer cells could provide for a better understanding of the disease process. During the evolution of PDAC, the stromal proportion increases from 4% in the normal gland to up to 80%. We hypothesised that there is an optimal proportion of stellate cells and cancer cells that modulates tumour behaviour and we attempted to dissect out this probable ‘tipping point’ for stromal composition upon cancer cell behaviour using a well-established in vitro organotypic culture model of pancreatic cancer. The cancer cell-stromal cell interaction led to extra-cellular matrix contraction and stiffening; and an increase in cancer cell number. The stromal stellate cells conferred a pro-survival and pro-invasive effect on cancer cells which was most pronounced at a stellate cell proportion of 0.66-0.83. The expression of key molecules involved in EMT and metastasis such as E-Cadherin and β-catenin showed a reduction and this was found to be most significant again at a stellate cell proportion of 0.66-0.83. Stellate cells altered the genetic profile of cancer cells leading to differential expression of genes involved in key cellular pathways such as cell-cycle and proliferation, cell movement and death, cell-cell signalling, and inflammatory response. qRT-PCR confirmed the differential expression of the top differentially expressed genes and protein validation by immunofluorescence staining using PIGR as a candidate molecule confirmed the experimental findings in human PDAC specimens. This study demonstrates that the progressive accumulation of desmoplastic stromal cells has a tumour progressive (pro-survival, pro-invasive) effect on cancer cells in addition to stiffening (contraction) of the extracellular matrix (maximum effect when the stromal cell proportion is 60-80%). This is mediated through a number of signalling cascades and molecular targets. Dampening this tumour-promoting interaction between cancer and stromal cells by ‘multi-targeting’ agents may allow traditional chemo- and/or radiotherapy to be effective.

Pancreatic ductal adenocarcinoma (PDAC) is a cancer of unmet need with a 5-year survival following diagnosis of 3% with limited surgical, radiotherapy and chemotherapy treatment options. Central to PDAC tumorigenesis is acquisition of an oncogenic Kras mutation to drive acinar-to-ductal metaplasia and progression to PDAC that is potentiated by NF-kB deregulation. However, PDAC requires the additional loss of tumour suppressors such as p53, SMAD4 or p16. ASPP family members ASPP2 and iASPP regulate both p53 and NF-kB, and are classified as a tumour suppressor and oncogene respectively. However, the precise roles of ASPP2 and iASPP in pancreatic cancer are unknown. In this thesis I demonstrate that ASPP2 suppresses metastasis and iASPP suppresses the pro-inflammatory tumour microenvironment. In a mouse model of PDAC development, ASPP2-deficiency does not alter metaplasia, PanIN progression or primary PDAC onset. However, median survival due to metastasis is significantly reduced in an ASPP2-deficient PDAC model. I demonstrate ASPP2-deficient PDAC can result in increased squamous differentiation defined histologically or via increased p63 expression. I propose ASPP2 is a key suppressor ΔNp63 and the squamous PDAC subtype in vivo. Conversely, iASPP is a putative oncogene and high expression in cancer associates with poor prognosis. However, in a mouse model of PDAC, loss of iASPP accelerates PDAC onset and metastasis. I demonstrate that iASPP is a functional tumour suppressor of a pro-inflammatory phenotype in response to oncogenic Kras and pancreatitis. I propose ASPP2- and iASPP-deficient mouse models of PDAC represent in vivo the squamous and immunogenic subtypes of PDAC respectively; and are relevant tools to study mechanisms of metastasis and inflammation-driven carcinogenesis.

Introduction: Pancreatic cancer is the 10th most common cause of cancer in the United Kingdom (UK) yet the 5th most common cause of cancer related death. Although excess adiposity, measured as body mass index (BMI), is a risk factor for the development of pancreatic cancer the increase in relative risk is modest. Animal models suggest that the intra-organ deposition of lipids may be more specific to disease risk than anthropometric measurements. There is therefore a need to develop non-invasive methods to quantify intra-pancreatic fat deposition as a potential biomarker for pancreatic cancer predisposition. Cancer Research UK (CRUK) sets out clear guidelines for biomarker discovery and development. Potential biomarkers must go through a process of discovery and assay development followed by qualification. Methods Three streams of research: (i) Stage-one of the PanORAMA project. Assessment of accuracy through comparison of CS-MR and MRS quantified intra-pancreatic fat with histologically quantified intra-pancreatic fat in 12 patients undergoing pancreatic surgery. (ii) Stage-two of the PanORAMA study. Assessment of precision (reproducibility) and comparison with other anthropometric markers of excess adiposity in healthy volunteers (n=15). Refinement of MRS protocols and repeated assessment of precision in healthy volunteers (n=10). (iii) The Breast Risk Reduction Intermittent Dietary Evaluation 2 (BRRIDE-2) trial. Comparison of the effects of Intermittent Energy Restriction (IER) with Daily Energy Restriction (DER) on intra-pancreatic and intra-hepatic fat stores and metabolic markers of disease risk (n=26). Results (i) CS-MR and MRS had agreement with histological assessment of intra-pancreatic fat, but correlations were only moderate to good (rho 0.672 and 0.781 respectively). (ii) CS-MR, and after refinement, MRS, have clinically acceptable precision. This study tested this principle in intra-pancreatic fat in healthy volunteers with a range of intra-pancreatic fat consistent with the literature on the healthy population. (iii) I found no differences in reduction in intra-hepatic or intra-pancreatic fat when comparing IER with DER. Overall, I found that significant reductions (mean: 6.5%) in both of these ectopic fat stores could be achieved with eight-weeks of dietary intervention. Discussion More recent hypotheses on the link between excess adiposity and cancer have focused on the importance of within organ local ectopic fat as an abnormal micro-environment favouring cancer development and progression. Importantly, this hypothesis explains the specificity of epidemiological associations between excess adiposity and cancer risk. The observations that within a given individual, in the presence of short-term weight reduction, there are differential changes in local within organ fats – hepatic fat and pancreatic fat – support the specificity hypothesis. This thesis has put us in position to scale-up and explore the importance of intra-organ fats using non-invasive imaging techniques.

Electronic medical record databases (e.g. the Clinical Practice Research Datalink, CPRD) are increasingly used in epidemiological research. The CPRD has two formats of data: coded, which is the sole format used in almost all research; and free-text (or ‘hidden’), which may contain much clinical information but is generally unavailable to researchers. This thesis examines the ramifications of omitting free-text records from research. Cases with bladder (n=4,915) or pancreatic (n=3,635) cancer were matched to controls (n=21,718, bladder; n=16,459, pancreas) on age, sex and GP practice. Coded and text-only records of attendance for haematuria, jaundice and abdominal pain in the year before cancer diagnosis were identified. The number of patients whose entire attendance record for a symptom/sign existed solely in the text was quantified. Associations between recording method (coded or text-only) and case/control status were estimated (χ2 test). For each symptom/sign, the positive predictive value (PPV, Bayes' Theorem) and odds ratio (OR, conditional logistic regression) for cancer were estimated before and after supplementation with text-only records. Text-only recording was considerable, with 7,951/20,958 (37%) of symptom records being in that format. For individual patients, text-only recording was more likely in controls (140/336=42%) than cases (556/3,147=18%) for visible haematuria in bladder cancer (χ2 test, p<0.001), and for jaundice (21/31=67% vs 463/1,565=30%, p<0.0001) and abdominal pain (323/1,126=29% vs 397/1,789=22%, p<0.001) in pancreatic cancer. Adding text records reduced PPVs of visible haematuria for bladder cancer from 4.0% (95% CI: 3.5–4.6%) to 2.9% (2.6–3.2%) and of jaundice for pancreatic cancer from 12.8% (7.3–21.6%) to 6.3% (4.5–8.7%). Coded records suggested that non-visible haematuria occurred in 127/4,915 (2.6%) cases, a figure below that generally used for study. Supplementation with text-only records increased this to 312/4,915 (6.4%), permitting the first estimation of its OR (28.0, 95% CI: 20.7–37.9, p<0.0001) and PPV (1.60%, 1.22–2.10%, p<0.0001) for bladder cancer. The results suggest that GPs make strong clinical judgements about the probable significance of symptoms – preferentially coding clinical features they consider significant to a diagnosis, while using text to record those that they think are not.