Automatic segmentation of tumours using Positron Emission Tomography (PET) was recommended for radiotherapy treatment (RT) planning of head and neck (H&N) cancer patients, and investigated in the scientific literature without reaching a consensus on the optimal process. This project aimed at evaluating the performance of PETCbased automatic segmentation (PETCAS) methods and developing an optimal PETC AS process to be used at Velindre Cancer Centre (VCC). For this purpose, ten algorithms were implemented to represent the most promising PETCAS approaches from a systematic review of the literature. The algorithms’ performance was evaluated on filled phantom inserts with variable size, geometry, tumour intensity and image noise. The impact of thick insert plastic walls on both image quantification and segmentation was thoroughly assessed. The PETCAS methods were further applied to realistic H&N tumours, modelled using a printed subresolution sandwich phantom developed and calibrated in house. Results showed that different PETCAS performed best for different types of target objects. An Advanced decision TreeCbased Learning Algorithm for Automatic Segmentation (ATLAAS) was therefore developed and validated for the selection of the optimal PETCAS approach according to the target object characteristics. Finally, a protocol was designed for the use of PETCAS within RT planning at VCC. The protocol was used retrospectively on a group of 10 oropharyngeal cancer patients, and the results highlighted the additional information brought by PET beyond anatomical imaging. In a prospective study on 10 additional patients, PETCAS replaced manual PET/CT delineation, and accounted for up to 33% of the modifications of manually drawn CT/MRI contours to derive the final planning contour. This study demonstrated the usefulness and reliability of the PETCAS method in RT planning, and led to modifying the clinical workflow for H&N patients at VCC. This work has the potential to be extended to other tumour sites and institutions.

Cell penetrating peptides (CPP) are short amino acid sequences with the potential to be used as vectors for delivering macromolecular therapeutics into cells. Five CPPs [R8, FFR8, (RXR)4, TP10 and PFV] were studied in primary human chronic lymphocytic leukaemia (CLL) cells using fluorescence-labelled CPPs. Uptake, sub-cellular localisation and toxicity were studied by confocal microscopy and flow cytometry. Two of the CPPs were selected, based on their cellular uptake and intracellular distribution characteristics, and used as delivery vectors for peptide-based NF-κB inhibitors. Four novel NF-κB inhibitory CPPs directed against p50 and p65 subunits were tested in primary CLL cells. Apoptosis was measured using AnnexinV/PI labelling and a caspase-3 activity assay by flow cytometry. Apoptosis was evident after one hour in cells treated with TP10-p50i and TP10-p65i and the LC50 of TP10-p50i and TP10-p65i was 6 μM and 10 μM respectively at 24 hours. This represents a ten-fold increase in toxicity when compared to the commercially available CPP NF-κB-inhibitors. Western blot analysis of NF-κB subunit translocation revealed NF-κB inhibition in some of the samples treated with TP10-p50i. However, the effects of the peptide varied from sample to sample. Studies using EMSA to measure NF-κB DNA binding revealed similar inconsistencies, even when CLL cells were stimulated with CD40L or CpG. Flow cytometic analysis of cell surface makers in CLL cells demonstrated that TP10-p50i did not alter the expression of CD69, a cell surface molecule regulated by NF-κB, indicating that the variations seen previously by EMSA and western blotting did not result from direct NF- κB inhibition. Although the exact mechanism of action of TP10-p50i was not determined, the cytotoxic effects observed with TP10-p50i are not likely to be related to a modulation of NF-κB activity.

Over 1 million women a year are diagnosed with Breast Cancer. The majority, approximately 70% express the oestrogen receptor (ER). ER positive breast cancer has historically been perceived as a ‘good cancer’, although many woman with ER+ breast cancer still succumb to their disease and globally breast cancer is the leading cause of female cancer deaths. The advent of gene expression profiling and the definition of the molecular instrinsic subtypes has defined at least two subtypes of ER positive breast cancers (luminal A and luminal B) that differ markedly in terms of biological behaviour, response to adjuvant therapies and most importantly patient outcome. The focus of this research is ER+ breast cancer and targeting patient therapy in this heterogeneous group. This work attempts to translate our understanding of the biology of the ER and cell signalling interactions to aid the correct identification of patients for both current therapy and more novel therapeutic approaches. Following a hypothesis generating pilot study examining whether the level of ER influences response to endocrine therapy, 557 formalin fixed paraffin embedded (FFPE) breast cancer specimens retrieved at time of definitive surgery from early breast cancer patients with available accurate 15 years follow up data were analysed to measure ER, Progesterone receptor (PgR), HER2 and Ki67 expression using immunohistochemistry. Tumour expression of ER, PgR and the combined endocrine receptor (CER), which considers the expression level of both hormone receptors and hypothesised to more accurately quantify endocrine responsiveness by acting as a surrogate marker of a functioning ER signalling pathway, were analysed. The results suggest that in this cohort of ER+ endocrine treated patients CER is a better predictor of endocrine response than either the ER of PgR independently. The CER was thereafter utilised as a surrogate marker of oestrogen receptor signalling pathway to develop a scoring system which included HER2 IHC expression and tumour histological grade, as surrogate markers of the 3 key pathways (ER signalling, HER2 signalling and proliferation). These were chosen as previous studies comparing various gene prognostic profiles indicate commonality in sampling groups of the genes representing their activation. The scoring system, named the Clinical Outcome Score (COS) was developed to represent a pragmatic equivalent of gene prognostic profiles utilising currently routinely measured tumour markers. We hypothesised that COS as an indicator of tumour biology may aid identification of risk in the very challenging group, ER+/HER2 negative patients with intermediate grade and low disease burden, and may help guide adjuvant therapy decisions particularly the indication for chemotherapy . In this exploratory analysis, the distribution of COS scores (2-10) followed a linear response with a notable separation between low scores (2-4) and high scores (5-10). Importantly, when analysed in combination with tumour burden, low COS may help identify patients with nearly 100% long term survival, however in all analysis high COS was associated with a highly significant poorer outcome in terms of early recurrence, late recurrence and 15 year breast cancer specific survival. This group of high risk ER+ breast cancer patients represent a real challenge (and concern) in the treatment of early breast cancer, as there is increasing evidence that ER+ tumours are relatively chemo-in senstive and the response to chemotherapy agents is limited. As a secondary analysis, within our cohort of ER+/ HER2- endocrine treated patients we retrospectively analysed the benefit of chemotherapy in patients with low and high COS scores and the results indicate lack of benefit in the cohort of patients diagnosed 1995-1998. Investigating novel therapeutic targets focusing on the subtypes of breast cancer, and tumour biology involved in endocrine resistance is now beginning to take precedence in breast cancer research. Two potential new therapeutic targets in ER+ breast cancer were studied. The first is the sodium iodide symporter, NIS, a transmembrane glycoprotein which has been exploited for the safe delivery of radio-iodide in the treatment of thyroid cancers for many years. NIS is expressed in many breast cancers, however most breast cancers expressing NIS lack functional uptake as demonstrated by scintography studies and in vivo animal work. In vitro results suggest that the ER is important in NIS regulation and function. In addition MAPK and PI3K-Akt signalling pathways may have a role in NIS regulation- both these pathways are often activated in ER+ breast cancer and known to have extensive crosstalk with the ER. Utilising ER+ and ER negative breast cancer cell lines we examined NIS function following gene delivery with a human NIS (hNIS) transfected plasmid and assessed function and expression of NIS following ER knockdown by siRNA. Our results suggest that the ER phenotype is important but not necessarily the ER per say. We examined NIS expression in a mixed ER+ and ER- cohort (n=50) of patient tumour samples using real time RT-PCR, and report high levels of NIS mRNA expression was limited to ER+ breast tumours. Prompting analysis of NIS expression, cellular location and correlations with cell signalling proteins in 300 ER+ breast cancers using IHC . Significant correlations were identified with key members of the PI3K-Akt and MAPK supporting their role in NIS regulation in vivo. Importantly, in both patient cohorts NIS was found to be significantly associated with poor outcome, and we hypothesis that this is an effect of enhanced growth factor signalling and activation of pathways in biologically more aggressive ER+ cancer (ER+/PgR-) may also regulate NIS and suggest future directions of research. Lastly, as a pilot study expression of Src kinase, a non receptor tyrosine kinase implicated in tamoxifen resistance and breast cancer virulence, was analysed by IHC in the ER+ breast cancer patient cohort. Interestingly nuclear Src kinase was found to be associated with improved outcome and hypothesise that Src Kinase expression in breast cancer may have varying roles in the different subtypes of breast cancer, an important consideration as Src Kinase inhibitors are currently in clinical trials. This pilot study formed a hypothesis that was subsequently examined in another student’s PhD thesis.

The BDCA antibodies allowed reliable measurement of dendritic cell (DC) subsets and B cell numbers in the blood of normal subjects, and patients with MM throughout the disease course. The numbers of blood myeloid DC (BmDC) and blood plasmacytoid DC (BpDC) are low throughout the course of the disease, and only improve for a short period of time following autologous HSCT. Thalidomide therapy of patients with relapsed disease was associated with an increase in BmDC1 and BpDC numbers. Monocytes, mobilised at the time of stem cell collection, were used to produce mature DC (matDC) from MM patients and normal donors (ND). The matDC produced from MM patients were of poorer quality as compared to those from ND, despite using combinations of GM/IL-4, GM/IL-13, X4 and MIMIC in the production process. The combinations that contained the X4 maturation cocktail produced the best quality matDC. The DC/T cell system is abnormal in MM patients. Despite this, it is possible to produce antigen loaded mature MoDC from MM patients. When combined with T cell pre-stimulation and IL-2 expansion, these DC are capable of inducing anti-MM cytotoxic T cells, which exhibit considerable anti-MM cytolytic activity. However, the DC from MM patients still display abnormal chemokine receptor expression, which may inhibit their capability to migrate to lymph nodes in-vivo in order to generate these cytotoxic T cell responses. These observations will aid in the optimisation of DC based immune therapies for MM, and suggest that a combined immunotherapy approach using pre-stimulated T cells, MM Ag primed DC and IL-2 may produce better clinical responses in MM patients.

Autophagy is an intracellular catabolic process that involves the sequestration of proteins and whole organelles into specialized cargo vesicles (autophagosomes) and their delivery to lysosomes with subsequent degradation. Autophagy is active at low levels at any time in virtually all cells and can be induced upon a variety of different stimuli. The core function of autophagy is the degradation and recycling of intracellular material. However, how this impacts on cellular survival likely depends on the biological context. The role of autophagy in cancer is very complex and incompletely understood. It is therefore very surprising that few studies exists that employ genetically modified mouse models of human cancer to examine the role of autophagy in this context. This is even more true when considering, that pharmacological inhibition of autophagy is currently being used in several clinical trials to treat cancer of various origins. The goal of this study was to examine the role of autophagy in a mouse model of pancreatic cancer. To achieve this several mouse strains were crossed: a) Pdx1-Cre LSLKRasG12D/wt mice that develop Pancreatic Ductal Adenocarcinoma (PDAC) similar to humans initiated by oncogenic Ras and b) Atg5flox/flox or Atg7flox/flox mice that permit Cre-induced deletion of either one of the essential autophagy regulating genes 5 and 7 (Atg5, Atg7). Offspring allowed us to examine the role of autophagy in pancreatic function. Loss of autophagy in the pancreas leads to exocrine and endocrine tissue destruction and reduces survival in approx. 60% of animals. The early death in autophagy-deficient mice can be delayed by additional deletion of p53; the mortality rate however remains unchanged. Moribund mice show a diabetic phenotype with elevated blood glucose and fructosamine levels. In the absence of oncogenic Ras autophagy deletion does not lead to cancer formation or occurrence of pre-malignant lesions in mice aged up to 700d. In mice that express oncogenic Ras in the pancreas (Pdx1-CreKRasG12D/wt) additional, genetic deletion of autophagy leads to accumulation of pre-malignant Pancreatic Intraepithelial Neoplasias (PanINs) that unlike their autophagy proficient counterparts never progress to cancer. In this genetic context autophagy therefore serves as a tumour promotor. In stark contrast in mice expressing oncogenic Ras and lacking both copies of p53 (Pdx1-KRasG12D/wt p53-/-) inhibition of autophagy, either genetically by deletion of Atg5, Atg7 or pharmacologically by chloroquine, tumour onset is accelerated. Therefore in a p53-deficient situation autophagy is now a tumour suppressor. Tumours that developed from a p53-proficient background have increased autophagy compared to tumours that developed from a p53-null background. Furthermore p53-/- Atg7-/- tumours have increased glycolysis in vitro and in vivo and enhanced intracellular metabolites of the anabolic Pentose Phosphate Pathway (PPP) compared to p53-/- Atg7+/+ tumours. In summary it is the p53 status that determines the role of autophagy in PDAC development. In tumours developing from a p53-proficient background loss of autophagy completely prevents cancer development; whereas in tumours arising from p53-deficient tissue loss of autophagy accelerates tumour formation.

This thesis examines factors influencing contemporary outcomes of patients managed by the South East Wales upper GI cancer network multidisciplinary team. The hypotheses tested were: PET/CT defined tumour characteristics influence outcomes of patients with oesophagogastric cancer; Centralisation of oesophagogastric cancer services improves outcomes significantly; HER2 overexpression is a poor prognostic indicator following oesophagogastric cancer resection; An involved circumferential resection margin (CRM) following oesophagectomy is an independent predictor of survival. PET/CT N stage was an independent and significant predictor of survival (p=0.022). SUVmax correlated positively and significantly with endoluminal ultrasound-defined tumour volume (Spearman’s rho=0.339, p=0.001). Centralisation increased the proportion of patients receiving potentially curative treatment by 78% (p<0.0001), reduced serious operative morbidity by 50% (p=0.062), shortened total length of hospital stay from 16 days to 13 days (p=0.024) and improved median and 1-year survival from 8.7 months and 39% to 10.8 months and 46.8% respectively (p=0.032). Centralisation was an independent and significant predictor of survival (p=0.03). HER2 overexpression and gene amplification was a predictor of poor prognosis in patients with curable oesophageal cancer (p=0.03). CRM involvement was also an indicator of poor prognosis in these patients (p<0.001). The College of American Pathologists’ criteria differentiate a higher risk group than Royal College of Pathologists’ criteria but overlook a patient group with similar poor outcomes (p<0.001).

This study describes the translation of an assay developed for use in cell culture models to into a method of measuring patterns of DNA damage from platinum agents in human blood samples. These adduct patterns could potentially be used in future studies for the stratification of patients for response and toxicity to oxaliplatin chemotherapy. Chapters 3 and 4 of this thesis describe the steps taken to translate our DIP--‐chip assay, a tool previously used in the study of DNA repair capacity in yeast and to measure induction of platinum-DNA adducts in cell culture models, into an assay capable of reproducibly analysing chemotherapy damage in human clinical samples. These results clearly demonstrate the protocol modifications required to use the assay on human blood samples, and show the reproducibly of the assay in detecting patterns of oxaliplatin induced DNA-adducts in clinical material. Chapter 5 describes the development of novel bioinformatic tools and analysis methods for interpreting DIP--‐chip DNA--‐adduct microarray outputs. The translation of a genomic--‐scale laboratory technology into a tool for patient stratification is a technical and bioinformatics challenge. The tools developed are a significant advance on previously available bioinformatic functions, and are essential for the application of this technique as a clinically useful assay. The final results section, chapter 6, documents the successful development of functional models to experimentally confirm links between single nucleotide polymorphisms in nucleotide excision repair genes with the development of oxaliplatin induced peripheral neuropathy (OIPN). This aspect of the study utilises new information, recently derived from experiments DNA-sequencing colorectal cancer patients, to develop a functional model of OIPN in Saccharomyces cerevisiae. This model is then used to demonstrate the impact of variations in DNA repair genes on the development of OIPN ‐ a relationship that highlights the significance of DNA repair to the development of oxaliplatin toxicity.

Introduction: It has been recently shown that the WISP proteins (Wnt-inducted secreted proteins), a group of intra- and extra-cellular regulatory proteins, have been implicated in the initiation and progression of variety types of tumours including colorectal and breast cancer. However, the role of WISP proteins in gastric cancer (GC) cells and clinical implication in gastric cancer has not yet been fully elucidated. Materials and methods: The expression of the WISP transcript and proteins in a cohort of GC patients was analysed using real-time quantitative PCR and immunohistochemistry, respectively. The expression of a panel of recognised EMT (epithelial-mesenchymal transition) markers were quantified (Q-PCR) in paired tumour and normal gastric tissues. WISP-2 knockdown sublines using anti-WISP-2 ribozyme transgenes were created in GC cell lines AGS and HGC27. Using the cell models and proteins extracted from gastric tissue samples, protein microarray was used to search for potential protein partners and signalling pathways involved with WISP-2. Subsequently, the biological functions, namely, cell growth, adhesion, migration and invasion, were studied. Potential mechanisms related with EMT, extracellular matrix and MMP (Matrix metalloproteinases) and signalling pathways were investigated. Results: Expression of WISP-2 was frequently detected in GC tissues. Levels of WISP-2, not WISP-1 and WISP-3, was significantly correlated with early TNM staging and differentiation status. High levels of WISP-2 were associated with a favourable clinical outcome and survival of the patients. We also found that WISP-2 expression inversely correlated with Twist and Slug in the paired gastric samples. Knockdown of WISP-2 expression increased the rate of proliferation, migration and invasion of GC cells and influenced expression of EMT biomarkers including Twist, Slug and Ecadherin. Using an antibody based protein microarray, ERK, JNK as well as AKT proteins were found to be co-precipiated with WISP-2 protein from human gastric tissue proteins. Furthermore, WISP-2 knockdown gastric cell lines also demonstrated a change in the ERK and JNK phophorylation. Mechanistically, WISP-2 suppressed GC cell metastasis through reversing epithelial-mesenchymal transition and suppressing the expression and activity of MMP-9 and MMP-2 via JNK and ERK. Cell motility analysis indicated that WISP-2 knockdown contributed to GC cells’ motility, an effect attenuated by PLC-γ and JNK small inhibitors. Conclusions: WISP-2 transcript and protein expressions are inversely linked to disease progression and linked to the survival of patients with gastric cancer. WISP-2 has a profound influence on the migration and adhesion of gastric cancer cells and is a powerful factor to reverse the EMT process in these cells. These effects of WISP-2 are via its involvement in the ERK and JNK pathways, which in turn modulate the MMP activities. Together, WISP-2 is an important regulator of the cellular function and an important factor in the progression of gastric cancer. It acts as a potential tumour suppressor in gastric cancer.

The transcription factor E2F1, a critical target of the tumour suppressor pRb, is deregulated in most human cancers. Oncogenes have been shown to deregulate E2F1 through inhibition of pRB and deregulation of E2F1 is an event that occurs in most human cancers. The essential role of E2F1 in apoptosis is well documented and deregulated E2F1 can enhance drug induced death. E2F1 is induced by various chemotherapeutic drugs and this induction, in addition with oncogenic stress, contributes to increased chemosensitivity. Cells expressing the adenovirus early region 1A (E1A) oncogene have been used as a tool to identify cellular regulatory pathways that modulate chemosensitivity. E1A sensitises cells to the induction of apoptosis by diverse stimuli, including many chemotherapeutic drugs. These E1A activities are mediated through binding the RB family proteins (pRb, p107 and p130) and via the E1A N-terminal domain that interacts with different cellular protein complexes including the p300/CBP transcriptional activator and p400/TRRAP chromatin-remodeling complex. The results presented here illustrate novel mechanisms of E2F1 induction both by oncogenes and chemotherapeutic drugs. Two minimal domains of E2F1 are described that are induced following DNA damage via mechanism(s) not previously identified. In addition, data are presented which show that E1A expression not only deregulates E2F1, but also elevates E2F1 levels. E1A is dependent on interaction with RB protein to induce E2F1 levels and this elevation contributes to cell death. Using previously described protein binding deficient truncations of E1A, we demonstrate that E1A binding to the p400/TRRAP protein complex is also critical for the induction of E2F1. E1A binding to p400/TRRAP was also critical in sensitizing these cells to drug induced apoptosis. Suppression of p400 using siRNA had similar affect on E2F1 induction and caused an increase in drug sensitivity indicating that E1A inhibits p400 function. These results contribute to the understanding of how activation of the E2F1 pathway may be targeted therapeutically to enhance chemotherapy-induced tumour cell death.

Loss of expression of the human DNA mismatch repair (MMR) gene, hMLH1, is seen in a number of tumour cell lines resistant to a variety of cytotoxic drugs. The aim of this study was to identify other proteins that interact with hMLH1 to attempt to further elucidate its role in MMR and the engagement of downstream damage response pathways. A yeast two-hybrid system, an in vivo system for detecting protein-protein interactions was utilised for this purpose. Fifteen known and five unknown genes were identified as encoding proteins interacting with hMLH1. These included three known hMLH1 binding proteins, hMLH3, hPMS1 and MED1. Amongst the other genes identified was the proto-oncogene c-MYC, a gene previously implicated in genetic instability and apoptosis. Using in vitro derived mutants of c-MYC, it has been shown that hMLH1 interacts with the leucine-zipper domain of c-MYC. The effect of elevated c-MYC expression on functional MMR was examined. An inducible c-MYC expression system, Rat-1 fibroblasts expressing c-MYCERTM, a fusion of c-MYC to the hormone binding domain of the oestrogen receptor was utilised. Elevated expression of c-MYC did not effect the mismatch specific binding complex activity in these cells as measured in EMSA experiments. However c-MYC overexpression utilising the Rat-1 cMYCERTM system was shown to result in a mutator phenotype in these cells. The results suggest there may be a link between the mutator phenotype, induced through overexpression of c-MYC, and loss of MMR. Overexpression of c-MYC, which is associated with many cancers, may result in the sequestration of hMLH1 preventing functional MMR. The interaction between hMLH1 and c-MYC is proposed to act in a DNA damage response pathway which is disrupted upon aberrant c-MYC expression.