Background: Phosphatase and tensin homolog (PTEN) is a multifunctional gene which acts as a tumour suppressor gene and is involved in DND damage response (DDR) mechanisms. PTEN has been found to be mutated in different types of human cancers including breast cancer. Ataxia-telangiectasia mutated (ATM) and RAD3- related (ATR) are involved in DDR and they have roles in cell cycle regulation and apoptosis. Previously ATM inhibition caused synthetic lethality in prostate and colorectal cancer cells with PTEN deficiency. In this study we hypothesize that PTEN plays key roles in breast carcinogenesis and that inhibition of ATR in the context of PTEN deficiency can provide a novel therapeutic approach through a synthetic lethality mechanism. Methods: In this study a large, well-characterised and molecularly annotated series of breast cancer (n=1954) was utilized to evaluate the clinicopathological and biological role of PTEN protein expression assessed using immunohistochemistry (IHC) and tissue microarrays technology. Several breast cancer cell lines (n=4) representing various molecular classes and PTEN status were studied in vitro using functional assays. Cellular consequences of ATR inhibitor (VE-821) treatment were investigated in a panel of PTEN-proficient including MCF7 and MDA-MB-231, and PTEN-deficient including BT-549 and MDA-MB-468 breast cancer cell lines. DNA repair expression profiling, MTS cell-proliferation assay, FACS for cell cycle, γH2AX and FITC-annexin V flow cytometry analysis were performed on PTEN-deficient and PTEN-proficient cells to study the functional consequences of PTEN deficiency on breast cancer cells. Results: PTEN was expressed in the nucleus and cytoplasm of malignant cells. The negative nuclear expression was detected in 62.6%, whilst negative/low cytoplasmic expression was found in 40.2% cases of breast cancer. The negative nuclear PTEN IHC expression was associated with features of aggressive behavior including higher grade, nuclear pleomorphism, higher mitotic index, larger tumour size, oestrogen receptor (ER) negativity, high risk Nottingham prognostic index (NPI≥3.4) and shorter breast cancer specific survival (BCSS) (pvalue < 0.05). Interestingly, in tumours with low nuclear PTEN, high ATR and/or high pChk1 (pCHK1Ser345) expression was also linked to poor BCSS (P-values < 0.05). Preclinical study demonstrated that PTEN-deficient breast cancer cells feature altered transcriptional expression of several genes involved in DNA repair pathways. Compared with ATR inhibitor (VE-821) in PTEN-proficient breast cancer cells, ATR inhibition in PTEN-deficient cells was associated with accumulation of double strand DNA breaks, cell cycle arrest at G2/M phases, and increased apoptosis. Conclusions: PTEN deficiency has prognostic significance in breast cancer, and selective targeting of ATR in PTEN-deficient cells with ATR inhibitor (VE-821) can serve as a potential avenue for development of personalized therapy for breast cancer patients.

Background: Polycystic ovarian syndrome (PCOS) is the most common endocrine, reproductive, metabolic and psychological disorder in women of childbearing age, affecting 6-10% of premenopausal women. Hyperandrogenism is the most important biochemical feature of the syndrome, which is responsible for the clinical features of PCOS, and is frequently associated with metabolic disturbance, such as insulin resistance, dyslipidaemia, glucose intolerance and hypertension, regardless of the presence of obesity. In several studies, attention has focused on androgen as the main factor for development of metabolic disturbance, which observed in women with PCOS. The relationship between adipose tissue and the pathophysiology of PCOS, in terms of development of hyperandrogenism and its relation to development of insulin resistance with compensatory hyperinsulinemia still not fully understood. Based on epidemiological studies, the association between circulating androgen levels and insulin resistance, as well as central obesity is a direct correlation. Although ovaries and adrenal glands are the main sources of androgens, adipose tissue is also one of the most important peripheral tissues involved in the production of androgens. Adipose tissue is not just an organ with energy storage; it also has endocrine, paracrine and autocrine functions, due to secretion of active peptides, known as adipokines, and hormones, such as androgens. In order to understand the role of adipose tissue in the development of hyperandrogenism, we hypothesised that “Androgen metabolic pathways leading to testosterone production in subcutaneous adipose tissue are altered in women with PCOS.” The excess adipose tissue androgen synthesis plays an important role in PCOS pathogenesis. Aims: The main aim of this study was to analyse and compare the expression level of the two key enzymes in androgens synthesis, 17-α-hydroxylase/17.20-lyase (CYP17A1) and 17-β-hydroxysteroid dehydrogenase type 5 (AKR1C3) in adipose tissue of women with and without PCOS. These are responsible for locally synthesised sex steroid hormones, mainly androgens; CYP17A1 is responsible in the production of the precursors of androgens and AKR1C3 is responsible in the conversion of inactive androgen (androstenedione) to its active form, testosterone. In addition, to understand the mechanism of androgen production this study employed isolated pre-adipocytes cultures and in vitro differentiated to mature adipocytes with close regulation of the impact of insulin and LH as the most two hormones which have effect in sex steroid synthesis. In doing so, we investigated androgen synthesis by activation and expression of the main steroidogenic enzymes CYP17A1 and AKR1C3 in adipocytes of non-PCOS and PCOS women. This enabled us to study the difference in the CYP17A1 mRNA and AKR1C3 mRNA expression levels, as well as the concentration of testosterone secretion across non-PCOS and PCOS cultures. This also indicated the probability of a pathway leading to localised synthesis of adipocytes and to investigate if there is a role for PI3-K signaling pathway in insulin regulation of testosterone synthesis in peripheral adipose tissue of women with and without PCOS. Methods: In order to achieve these aims, subcutaneous adipose tissue samples (SC) were surgically obtained during gynecological surgery from women with and without PCOS. All participants were of reproductive age (20-45) with a BMI of 20-35kg/m2. Total RNA was isolated from frozen adipose tissue samples of non-PCOS (n=8) and PCOS (n=8) after matching, using Trizol reagent method, followed by reverse transcription. Quantitative RT-PCR was performed to determine the expression of a panel of reference genes (GAPDH, ACTB, and LPR10), and target genes (CYP17A1 and AKR1C3). Data were analysed with GenEx and compared using ΔΔCt method. AKR1C3 protein expression in non-PCOS (n=3) and PCOS (n=3) was measured and compared by using western blot (WB) technique. Pre-adipocytes were isolated from fresh adipose tissue samples by enzyme digestion (collagenase) method and in vitro differentiated to mature adipocytes, which were cultured in FCS-free medium, Recombinant insulin +/-LH+/-PI3-k inhibitor (LY294002) was added to the cell culture at different concentrations, in preparation for investigating any change in the expression of steroidogenic enzymes (CYP17A1, AKR1C3) by RT-PCR, after extraction of total RNA The supernatant was collected for testosterone measurement before and after treatment using enzyme-linked immunosorbent assay (ELISA). Results: Of the reference genes testes, GABDH, ACTB, and LRP10 were shown to be consistently expressed across the PCOS and non-PCOS women. The mean± SEM relative expression level of AKR1C3 mRNA in PCOS adipose tissue was 15.1± 2.0, which was significantly (P=0.0003) greater than that (3.3±1.1) of non-PCOS women. However, the expression level of CYP17A1 mRNA was not significantly (p=0.56) different between the two groups. AKR1C3 protein expression level was less expressed in PCOS and there was no significant (P > 0.05) difference in the protein expression between two groups. CYP17A1, AKR1C3 and testosterone were significantly higher in PCOS (n=5) versus the non-PCOS (n=5) in treated and un-treated cultures. Insulin did not alter CYP17A1 or AKR1C3 mRNA expression in PCOS group. In the non-PCOS, AKR1C3 significantly increased with gradual increase in insulin concentrations, 1nM/l (P=0.001), 10 nM/l (P=0.004), and 100 nM/l (P=0.0003). Insulin up regulates AKR1C3 mRNA expression (no treatment (0), 1, 10,100) (0.96±0.21) (1.59±0.84) (2.39±1.23) and (7.42±0.85) respectively. LH± insulin did not alter the expression of either of the enzymes in PCOS Insulin increased testosterone concentration in non-PCOS but not in the PCOS, testosterone concentration in the supernatant of untreated cultured PCOS (n=5) adipocytes (mean± SEM, 129.3±2.5 pg /ml) was significantly higher (P < 0.0001) than that (33.7±4.6 pg /ml) of non-PCOS (n=5) adipocytes. Insulin addition in different concentrations (1nM/l, 10nM/l, 100nM/l) resulted in a significant increase in testosterone concentrations (94.1±7.1; 118.2±18.2, 200.0±7.3 pg/ml, respectively) in the supernatant of cultured non-PCOS adipocytes, but not in the PCOS adipocytes (118.1±1.8, 90.5±6.4, 89.3±7.6 pg/ml, respectively). The increase of testosterone levels in the non-PCOS adipocyte culture supernatant followed a dose dependent fashion. The magnitude of increase in testosterone concentrations in non-PCOS adipocyte culture supernatant was markedly increased when LH was added to insulin. Adding PI3-K inhibitor (LY294002, 10ng/ml) to insulin did not change the magnitude of insulin effects on testosterone concentrations in non-PCOS adipocyte culture supernatant. Conclusion: The data obtained suggests that adipose tissue has the ability to produce its own steroid hormone after detection of the main key enzymes of androgen biosynthesis, and it revealed a 5-fold increase in the expression level of AKR1C3 mRNA in subcutaneous adipose tissue of PCOS women. It is therefore possible to postulate that peripheral adipose tissue plays an important role as a source of excess androgen production in women with PCOS. This could potentially pave the way for the development of innovative therapeutic targets for the management of this very common syndrome. In addition, we show a markedly higher CYP17A1, AKR1C3 and testosterone in peripheral adipose tissue of PCOS vs. non-PCOS women. This supports the hypothesis that peripheral adipose tissue plays an important role in the pathogenesis of hyperandrogenaemia and PCOS. Insulin and LH seem to play a role in the increased androgen synthesis in adipose tissue of PCOS women, but not through the PI3-K signaling pathway. PI3-K signaling pathway does not seem to be involved in insulin regulation of testosterone synthesis in peripheral adipose tissue of women with or without PCOS.

The calpain system is associated with cancer chemotherapeutic response in both in vivo and in vitro studies. Previous immunohistochemistry (IHC) data conducted in our group indicated that high calpain-2 expression was associated with both the resistance to platinum-based adjuvant chemotherapy and worse patient outcome; moreover calpain-2 appeared as an independent prognostic factor in multivariate analysis. To test the hypothesis that conventional calpain subunits, especially calpain-2, are associated with the chemo-resistance of ovarian cancer cells to platinum-based chemotherapy (cisplatin and/or carboplatin), five ovarian cancer cell lines, with varying platinum-based chemotherapy sensitivities, were chosen as in vitro models: the platinum-sensitive A2780 cells and its resistant counterpart A2780-cis cells; the platinum-resistant SKOV3 cells; and the platinum-sensitive PEO1 cells and its platinum-resistant counterpart PEO4 cells. Western blotting was used to assess the expression of the conventional calpain subunits (i.e. calpain-1, -2 and -4) and calpastatin in this panel of cell lines. Calpain activity was regulated by inhibitor calpeptin and calpain-2 short hairpin RNA (shRNA) was used in attempt to specifically downregulate calpain-2 expression. Calpain activity was assessed by an activity assay using fluorogenic peptidase substrate t-BOC. The role of calpain in proliferation and resistance to platinum-based chemotherapy were examined in vitro using growth curves and colony formation. Moreover the study of calpain-4 expression was added into the current project, in addition to verifying the association between the expression of calpain-1, -2 and calpastatin and clinicopathologic variables (e.g. chemo-resistance and patient outcome) by standard immunohistochemistry (IHC) with a larger patient cohort (n=575). To test the hypothesis that conventional calpain subunits and calpastatin are associated with ovarian tumour metastasis, the effect of calpain inhibition (by calpain inhibitors) and activation (by calcium ionophore) on ovarian cancer cell migration was examined using haptotaxis (scratch wound) migration assay. Based on information from 2016 FASEB calpain conference, microtubule-associated protein 4 (MAP4) and spleen tyrosine kinase (Syk) appeared as potential calpain-related proteins associated with angiogenesis and epithelial-mesenchymal transition. Using IHC, their protein expression (i.e. MAP4 and Syk) was assessed and their associations with clinicopathological variables in ovarian cancer patient samples were studied; besides, their correlations with conventional calpain subunits and calpastatin, together with EMT (epithelial-to-mesenchymal transition)-associated proteins and angiogenesis-associated proteins (n=87, data provided by Dr S. Deen) were analysed. Significant variations of calpain system protein expression levels were observed between the different cell lines. Among the 5 cell lines, A2780 and A2780-cis cells (likely to be endometrioid carcinoma cell lines) expressed very low levels of the conventional calpain subunits and calpastatin; whilst PEO1 and PEO4 cells (high-grade serous carcinoma cell lines) expressed comparatively higher level of these proteins. Thus, different expression of the calpain system seemed to be associated with ovarian cancer histological subtypes, which was supported by the IHC study. No significant difference of the calpain system expression was detected and calpeptin caused a similar inhibition of cell proliferation between chemo-sensitive ovarian cancer cells and their resistant counterparts. Because SKOV3 and PEO4 cells expressed the highest levels of calpain-2, shRNA was used for specific knockdown of calpain 2 in these two cell lines, unfortunately numerous attempts proved unsuccessful. Although with the optimised concentration and treatment duration, calpeptin could inhibit approximately 30% of calpain activity, down-regulation of calpain activity via calpeptin could not sensitise ovarian cancer SKOV3, PEO1 and PEO4 cells to cisplatin and carboplatin. Hence, to revisit the question as to whether conventional calpains and calpastatin are associated with chemoresponse and patient survival, a larger cohort was used to validate the previous study. In the current study, the expression of the conventional calpain subunits and calpastatin were positively associated with each other. Calpain-2, -4 and calpastatin expression were associated with overall survival (OS) but none of them was an independent marker of OS in multivariate analysis. Neither the conventional calpain subunits nor calpastatin expression was found associated with resistance to platinum-based adjuvant chemotherapy. Since low calpain-1 expression was associated with low tumour stage, cellular processes that involved in cancer spread were studied. Again calpeptin was used for the inhibition of calpain activity, but no significant inhibition was observed on ovarian cancer cell migration. In contrast, upregulating calpain activity by A23187 using optimised concentration was found able to significantly inhibit the migration of SKOV3 cells. The recently found calpain-related proteins MAP4 and Syk were then included into the current study. Like calpain-1, neither MAP4 nor Syk expression was associated with patient outcome. Next, cases were grouped by clinicopathological variables for the examination of the association between the protein expression and survival; in such a case only high nuclear Syk expression was significantly associated with better patient outcome in certain subgroups. Low calpain-1 expression was associated with low tumour stage, so were MAP4 and Syk expression. MAP4, Syk and calpain-1 expression were significantly associated with tumour histological subtypes and their expression were significantly correlated with each other. Integrin α2β3 was moderately correlated with calpain-1, MAP4 and cytoplasmic DARC expression. In conclusion, although in both the previous and the current study, calpain-2 expression was adversely associated with OS of ovarian cancer patients, the current results did not support the initial hypothesis that calpain can sensitise ovarian cancer cells to cisplatin/carboplatin. The roles that calpain system played in cancer cell haptotactic migration appeared to vary with cell context. Calpain-1, MAP4 and Syk expression were significantly correlated with each other and were closely related to ovarian cancer spread.

Background and aims While current prognostic tools for breast cancer (BC) provide valuable information on behaviour and outcome, there are growing concerns that these parameters are not sufficient to reflect the degree of heterogeneity or to guide management decisions at individual patient level. Therefore, further refinement of the existing prognostic tools is needed. The advent of multi-parameter gene signatures has increased our hope of refining BC prognostic classification; however, their cost and restricted application to certain subgroups of BC limit their clinical usefulness. Molecular taxonomy of BC using intrinsic gene sets has not only improved our understanding of BC biology, but has also provided important prognostic information. Yet, integration of this molecular classification with the clinical parameters remains a challenging task. Our group has recently developed a prognostic tool that incorporates the molecular features of BC with the well-established prognostic morphological variables; the Nottingham Prognostic Index Plus (NPI+) that aimed at overcoming the limitations of using different molecular and clinicopathological prognostic parameters separately. However, NPI+ currently has limitations and needs further refinement to be applicable to BC management in routine practice. Therefore, this study aimed to investigate some relevant potential prognostic markers that can improve BC prognostic classification, in the context of combined molecular and morphological prognostic BC taxonomy, and to further refine the NPI+ to improve its prognostic value, while addressing some issues related to its components and performance. The study included four main objectives. The first was the integration of the proliferation biomarker Ki67 and addressing some technical issues related to its prognostic value and application in routine practice. The second was the evaluation of a high-throughput proteomic technique, reverse phase protein array (RPPA), for its potential use as a single-step quantification method of multiple proteins. The third aim proposed to determine the biological relevance of BC expressing low levels of oestrogen receptor (ER) using techniques at both transcriptome and protein levels. The fourth aim was to investigate the incorporation of a novel cancer stem cell (CSC) biomarker as a potential prognostic variable that can refine BC classification. Methodology In this study, five large primary invasive BC cohorts were investigated at both transcriptome and protein levels. Tissue microarray (TMA) and whole tissue section (WTS) were used. Molecular techniques used in this study included immunohistochemistry (IHC), western blot (WB), different protein and RNA extraction techniques, laser capture microdissection (LCM), RPPA, real time-polymerase chain reaction (RT-PCR) and RNAscope. Results Although Ki67 expression can be examined using both WTS and TMA, the assessment of Ki67 in whole sections is preferred and using multiple or larger TMA cores has to be explored as an alternative to WTS. When assessing Ki67 in TMAs in BC, a cut-point of 20% appears to be optimum in concordance with WTS and patients outcome. However, our results support the use of Ki67 as a continuous variable, particularly in the stratification of patients into prognostic groups, either using TMA or WTS assessment (Muftah et al., 2017). Using the MIB-1 clone with different optimisation conditions is associated with cytoplasmic/membranous reactivity. In this regard, it is recommended that different anti-Ki67 clones could be used for clearer staining. The results show that Ki67 can successfully replace mitotic frequency in the updated prognostic index, NPI+. Unlocking FFPE tissue lysates utilising RPPA is a reliable method for protein quantification. Data produced by this high-throughput technique could be used in concurrent analyses of protein profiles in a large number of clinical cases. Accordingly, RPPA could be consistently used in molecular classification of BC, such as the NPI+ (Negm et al., 2016). Adding Ki67 to the cluster using the RPPA technique improved molecular classification with reassignment of 16% of the unclassified patients. Investigating ER in BC at both levels (transcriptome and protein) shows that its expression is essentially bimodal (Muftah et al., 2016). Additionally, our results question the advantage of hormonal therapy in the low ER (< 10%) subgroup; where, in the study cohort, nearly half (42.2%) of the Core Needle Biopsy (CNB) cases showed WTS negative. There is strong agreement between the IHC and in situ RNAscope results, particularly in focal heterogeneous ER staining areas. This study provides scientific evidence that the actual ER cut-off seems to be 10%. Furthermore, this research supports the clinical importance of B-cell specific Moloney leukaemia virus insertion site-1 (Bmi-1) as a favourable prognostic biomarker in BC and its ability to refine the CD44/CD24 phenotypes as well as slow proliferating tumours into prognostically relevant subgroups. Conclusion This study presented data enabling the updating and evaluation of existing prognostic parameters and indices using promising biomarkers and high-throughput techniques, while combining molecular and clinical variables to stratify BC patients into relevant prognostic subgroups. Further investigation of the potential and refinement of the existing BC prognostic parameters is needed in order to allow more precise BC classifications that can predict patient outcomes and potential response to therapy.

There is an increase in the number of women with subfertility, who seek assisted reproductive technology (ART) treatment to overcome involuntary childlessness due to ovarian ageing, but which is commonly associated with poor response to ovarian stimulation protocols resulting in poor outcomes during ART. A few clinical studies have shown that DHEA, an androgen mainly secreted from the adrenal glands, can improve ovarian response and increase the chances of pregnancy after in vitro fertilisation (IVF) treatment in women affected by ovarian ageing. However, clinical evidence as well as knowledge regarding underlying mechanisms of DHEA on improving ovarian response is still limited. The research was divided into 2 arms; i) mechanistic experiments where sheep, a monoovulatory species with a long period of folliculogenesis, were used to investigate DHEA mechanisms and ii) a pilot randomised study called DITTO (DHEA intervention to treat ovarian ageing) to evaluate the efficacy of DHEA in clinical IVF practice. The first animal experiment (chapter 3) utilised an ovarian cortical ‘normograft’ model in which a single ovary (n=6) was taken out operatively. Half of the ovary was prepared for small pieces of ovarian cortical grafts and transferred back onto the cut- ovarian pedicle. The other half was fixed for use as a control. DHEA implants were administered to the animals and left for a 10-week period. The second operation was scheduled to harvest both the remaining ovary and cortical graft. The normograft procedure allows i) destruction of all growing follicular pool and synchronisation of primordial follicle initiation in the autograft without endocrinological disturbance, and ii) retention of the normal follicular hierarchy in the remaining ovary. DHEA stimulates primordial follicle initiation and preantral/early antral follicular development in the gonadotrophin responsive stages by improving granulosa cell proliferation and modulating local growth factors e.g. anti-Müllerian hormone (AMH). The second animal experiment (Chapter 4) was conducted partly to confirm the effects observed in the first experiment by comparing with the control animals. The animal number was double (12 mature ewes) and they were divided into two groups; animals treated with DHEA implants for 12 weeks and controls with placebo treatment. Moreover, the experiment was designed to compare, between treatment and control, ovarian response to controlled ovarian stimulation mimicking the human IVF cycle. The findings confirm that DHEA treatment increases the rate of follicle initiation as well as stimulating preantral follicular growth. Enhancing granulosa cell proliferation is one of the mechanisms but DHEA treatment also increases follicular response to both FSH and insulin-like growth factor 1 (IGF-1) by positive modulation of receptors expression. Higher follicular P4 production was observed in treated animals after the gonadotrophin stimulation when compared to the controls. A pilot double-blinded, placebo-controlled, randomised trial was performed over two years with 60 women undergoing IVF. Subjects were randomised to receive either 75 mg/day DHEA or a placebo for at least 12 weeks with both capsules having similar colour, size and appearance. Sixty women were recruited with poor ovarian reserve based on antral follicle count or anti-Mullerian hormone thresholds. Subsequently, all patients were scheduled to the long gonadotrophin releasing hormone (GnRH) agonist IVF protocol with hMG, starting dose at 300 IU/day. Ovarian response, live birth rates and molecular markers of oocyte quality were compared between the study and control groups on the ‘intention to treat’ basis. Unfortunately, the clinical trial could not demonstrate a significant improvement in terms of pregnancy outcomes and ovarian response markers in the treatment group when compared to the control. The numbers of oocytes obtained from the egg collection procedure were similar between the two groups. Also, there was no significant difference in all cumulus and granulosa mRNA expressions concerning oocyte developmental competence. In conclusion, there is currently no established clinical evidence supporting the use of DHEA as an adjuvant in IVF treatment. In contrast, in vivo animal studies demonstrate that DHEA treatment potentially provides stimulatory effects on folliculogenesis, thus indicating the importanceof androgens at all stages of follicular development. Further investigations are required to confirm these potential findings. If the beneficial effects of DHEA are confirmed, it can provide a safe and less costly intervention to fight against ovarian ageing in terms of infertility treatment and menopausal conditions.

Introduction Congenital uterine anomalies result from the abnormal embryological development of the paramesonephric (Müllerian) ducts. Debate surrounding their classification is ongoing. During this PhD project both the American Society for Reproductive Medicine and The European Society of Human Reproduction and Embryology (ESHRE) with the European Society for Gynaecological Endoscopy (ESGE) have published different classification systems. To adequately assess uterine morphology requires concurrent imaging of the external and internal contours of the uterus. Three-dimensional ultrasound can achieve this view and is more acceptable to women than more invasive tests. In view of the inconsistent classification systems and use of various diagnostic tests data regarding prevalence and effect on reproductive outcome must be interpreted cautiously. Nonetheless, it appears that uterine septa are more prevalent in women with poor reproductive outcomes and associated with infertility, miscarriage and preterm birth. The rationale for septal resection is to restore normal uterine anatomy with the intention of improving reproductive outcome. However all previous studies assessing the effect of septal resection are biased due to their observational design using women as their own controls. There have been no randomised controlled trials. Aims The aim was to investigate the hypothesis that a trial of septal resection for women with subfertility or miscarriage is feasible to conduct. Studies A non-invasive method for assessing uterine septa is a vital prerequisite for a trial, to allow women to be assessed properly and consented before an invasive operative hysteroscopy is done. Furthermore, a test is required to assess the efficacy of septal resection post-procedure as some women are left with a residual septum. So it was important to establish the diagnostic reliability of three-dimensional ultrasound to identify uterine septa. ESHRE recommended using three-dimensional ultrasound, but few data existed demonstrating its reliability. I conducted two reliability studies. The first found that the intra-rater reliability of experts in uterine anomalies was only moderate. This study was limited as it could not differentiate between the manipulation of uterine volumes, uterine measurements or application of classification systems. A second study assessed the intra- and inter-rater reliability of uterine measurements and then used algorithms to classify the volumes. This showed uterine measurements are highly reproducible with cavity indentation showing the highest intra- and inter-rater reliability (both ICC 0.98). Applying classification algorithms to these measurements showed almost perfect intra and inter-rater agreement. When designing a trial to definitively establish the efficacy of an intervention it is crucial to ensure it is adequately powered to detect a difference which is important to patients. So next I set out to calculate the minimum clinically important difference for women to consider septal resection. Using survey and interviews I calculated that the median minimum clinically important difference for women with a history of poor reproductive outcome was a change in future risk of miscarriage from 30% to 10%. The sample size required to detect this difference was 118, (59 in each arm) where _=0.05 and (1-_)=0.9. Between August 2014 and August 2017, 6035 women with subfertility and miscarriage were screened in three settings to identify those eligible to participate. Three-dimensional ultrasound was used as initial screening in a tertiary fertility clinic (n=2846) and recurrent miscarriage clinic (n=189). Around 3000 women underwent two-dimensional ultrasound in the early pregnancy unit, of which five had suspected uterine anomalies and were referred for three-dimensional ultrasound. A uterine septum was present in 15, of whom 13 were eligible to participate in the pilot septal resection trial. So identifying one eligible participant required approximately 464 women to be screened. Divided by patient population, to identify one eligible participant the number of women to be screened using three-dimensional ultrasound was 407 women with infertility, or 27 women with recurrent miscarriage. Approximately one septum was identified per 1500 women who had a two-dimensional scan in the early pregnancy unit. Subsequently I undertook three studies of the willingness of patients and clinicians to participate in the pilot trial. First I undertook a survey of clinicians. Sixty-seven clinicians responded and approximately half would randomise to a septal resection trial, but only 12 clinicians (18%) showed interest in becoming involved in such a trial. The next study investigated the willingness of eligible patients to participate. In Nottingham, 2 of 13 women were randomised to the study. A survey and interview with women who declined participation found that many had strong views on their treatment, either in favour of, or against resection and did not wish to leave it to chance. The third study assessed the willingness of other units to be involved in the trial. Twelve clinicians from eleven units expressed interest in becoming a trial site, but only one unit went on to randomise 4 further patients. Reasons for non-participation included lack of funding, exclusion from the NIHR Clinical Research Network Portfolio, inadequate access to three-dimensional ultrasound and local service configuration. During the study period six women were randomised. The final two chapters set out to establish if a septal resection trial is a research priority using two different methods. Firstly an economic evaluation and value-of-information analysis. Value of information is the amount a decision maker should be willing to pay for information prior to making a decision. As resources are finite, these methods can be used to direct future research efforts towards the most efficient studies. Septal resection, was compared with progesterone as a large study, the PROMISE trial, was recently funded. Using existing evidence, namely meta-analyses of biased studies, it predicted that there was no value in conducting further research into progesterone or septal resection as with existing evidence, they were both cost-effective interventions. Nonetheless, when more uncertainty was introduced based on expert opinion the value of information for both interventions increased. Using the most favourable assumptions for our present level of uncertainty, assuming the usual NHS value of £30,000 for a quality adjusted life year (QALY) the expected value over ten years to the NHS of the information from a progesterone trial was £1.6B compared with £80m for a septal resection trial. The main reason for this was that progesterone has a larger potential treatment population, all women with recurrent miscarriage, rather than just those with a uterine septum. Other trials are therefore a higher priority. Finally, I established the top research priorities for miscarriage. Miscarriage research uncertainties were identified and prioritised collaboratively by women and healthcare professionals. A priority setting partnership was conducted on the topic of miscarriage using methodology advocated by the James Lind Alliance, a subsidiary of NIHR. A total of 2402 unique questions were identified and refined to 58 summary questions. The top 10 research uncertainties were collaboratively agreed by women who have experienced miscarriage, those affected by miscarriage and healthcare professionals at a final workshop. The top question was “What are the effective interventions to prevent miscarriage, threatened miscarriage and recurrent miscarriage?”. This would include septal resection to prevent miscarriage, but would also be any other potential intervention. Discussion The high reliability of three-dimensional ultrasound to diagnose uterine septa was demonstrated. / This was dependent on taking uterine measurements, as subjective application of any classification system is less reliable. The low prevalence of septa and the lack of willingness from patients and clinicians to participate would make a trial prohibitively time consuming or expensive. Furthermore, a uterine septum is not the condition for which resection is performed. Instead resection is to prevent infertility, miscarriage or preterm birth, meaning that ideally three separate trials should be performed. I demonstrated that funding a trial is unlikely to be the best use of resources as the cost of conducting a trial is high compared with the value of information it could provide to decision makers. However, patients and clinicians still consider further research in this area a priority. Herein lies the dilemma of whether to pursue a trial, which will provide the highest quality evidence but is likely infeasible, or to consider alternative methodology, albeit with limitations, to help guide women and their clinicians who are considering septal resection. Collaboration with other international groups is the only way to advance this trial further. For example, the Dutch TRUST trial. TRUST has also had difficult recruiting, but has randomised 43 women over nine years. A collaborative approach between the two studies may lead to the completion of a trial. Nonetheless, TRUST is not adequately powered to detect the minimum clinically important difference and has less stringent criteria for diagnosis of a uterine septum. Conclusion Presently, a randomised controlled trial of hysteroscopic septal resection for women with a history of subfertility or miscarriage and a uterine septum is not feasible. The reasons for this are that sufficient women and clinicians are not willing to participate to achieve the required sample size. Furthermore the resource required to recruit enough participants would make the trial prohibitively expensive.

Background: Current evidence indicates that DNA damage response (DDR) is a highly complex process that involves various pathways working in an orchestrated and interwoven manner in response to different types of damage to DNA. Although specific defects of DDR remain to be deciphered in cancer as a general, there is certainly an undeniable relationship between a particular dysfunction of DDR and the phenotype of tumour [1, 2]. It has been demonstrated that familial forms of breast and ovarian cancer are characterised by defects in one of the main mechanisms of DDR homologous recombination (HR) as a result of germline loss-of-function mutations in one of HR modifying genes, such as BRCA1 and BRCA2 [1, 3, 4]. Defects of genes involved in other DDR pathways are also associated with specific types of cancers; for instance hereditary non polyposis colorectal cancer (HNPCC) is strongly associated with specific mutations in the DNA mismatch repair pathway. Several previous studies have demonstrated that impaired DDR play a fundamental role in the pathogenesis and behaviour of breast cancer (BC). However, characterisation of this complex process, the expression and co-expression of the key proteins involved in the various DDR pathways and their prognostic significance in BC remain to be defined. In BC, it is reported that genes involved in DNA double strand breaks (DSB) repair are the most important. Two main pathways are involved in the repair of DNA-DSB; HR and Non Homologous End Joining (NHEJ) [3]. The common characteristics of global DDR are multiple genes induction directly associated with sensing and repair of DNA, arrest of cell cycle, and cell division inhibition. As a result DDR process does not only include genes activation involved in damage sensing as well as repair but additionally genes involved in control of cell-cycle [5]. Despite the fact that DDR may possibly involve activation of several pathways (such as SUMOylation (SUMO)) [6, 7] and many genes are engaged in different overlapping mechanisms, each pathway is characterised by activation and expression of a unique set of genes. This could allow discovering the active or aberrant pathway in a given tumour [1, 4, 5]. This study explores the hypothesis that investigation of alterations in the different pathways of DNA-DSB, may contribute to the characteristics of BC. Therefore, the aim was to perform a comprehensive profiling of key proteins involved in the different DNA-DSB repair pathways in the different molecular classes of BC. This approach aims to address the inherent problems arising from the complexity of DDR mechanism in BC with the potential of discovering a key pathway that is active or inactive in specific forms of BC that can be helpful to identify DNA repair status in individual BC patients. Method: The study cohort comprises three BC groups: A) Large series of unselected primary sporadic operable invasive tumours (n=1904) in addition to B) 386 cases of oestrogen receptor (ER) negative tumours and C) a well-characterised series of BC from patients with known BRCA1 germline mutations (n=24). The proteins investigated in this study are known to participate in different DNA-DSB repair pathways including, DNA damage sensors (ATM and ATR), HR repair (BRCA1, BARD1, Rad51, γH2AX and SMC6L1), DNA damage checkpoint signalling protein (CHK1 and CHK2), NHEJ repair (KU70/KU80, and DNA-PK), and SUMO (PIAS1, PIAS4, and UBC9). Because subcellular localisation of DDR proteins may affect their function, two markers that have role in nuclear transport in the cell were examined (NPM and KPNA2). The expression of these proteins was assessed using the well-established immunohistochemical technique utilising tissue microarray technology. The expression of proteins was further evaluated in various cell lines; BRCA1 deficient HeLaSilenciX® cells, and control BRCA1 proficient HeLaSilenciX®, MDA-MB-436 (BRCA1 deficient), and MCF-7 (BRCA1 proficient and ER+) using Reverse Phase Protein Microarray (RPPA). Results: Both cytoplasmic and nuclear expression was observed for expression of Rad51, SMC6L1, BRCA1, BARD1; (HR markers), PIAS1, UBC9 (SUMO markers), γH2AX (DNA-DSB marker) and CHK1 (checkpoint signalling protein). In contrast, both NHEJ markers and most of the DNA damage sensors (ATM and ATR), CHK2 and PIAS4 were mainly expressed in the nucleus. Generally, tumours that showed positive cytoplasmic/negative nuclear expression such as CHK1, PIAS1, Rad51, and BRCA1, and positive nuclear NHEJ markers showed an association with a poor outcome and adverse prognostic characteristics including high histologic grade, high mitotic frequency, high nuclear pleomorphism and larger tumour size in addition to ER negativity, and triple negative breast cancer (TNBC). Conversely, nuclear+/cytoplasmic- expression showed an association the better outcome. Interestingly, ATM protein expression showed no association with the expression of the two NHEJ markers, whereas ATR showed an association with cytoplasmic expression of BRCA1 and BARD1 and was positively associated with NHEJ markers. In non-TNBC, tumours showing BRCA1-/KU70/KU80- phenotype had worse breast cancer specific survival (BCSS) than positive expression (P<0.0001), whereas in the TN cohort,complex of KU70/KU80-&DNA-PK+ had the worst BCSS (P=0.001), and both are independent prognostic markers for BC. KPNA2, but not NPM was highly associated with poor BCSS (P<0.0001). At least one of nucleocytoplasmic transport markers (NPM or KPNA2) was significantly associated with the subcellular localisation of the most of the markers that showed cytoplasmic expression including SMC6L1, γH2AX, BRCA1, BARD1, UBC9, PIAS1 ,Rad51 and CHK1. RPPA was used to investigate the protein expression in different cell lines, although the correlation between RPPA and IHC was not significant, the results of RPPA were consistent with that demonstrated by IHC further supporting the finding of the current study. Conclusion: This study highlight the complexity of DDR related proteins and the overlap between different pathways involved in DDR. The finding of this study may help in the classification of BC and therefore, targeting active pathways in the development of drugs would enhance better patients’ outcomes. Major prognostic and predictive variables can be very important in choosing suitable treatment plans, identifying the risk of recurrence and classifying patients for clinical trials. Our results show that the HR- repair marker Rad51, complex of HR and NHEJ repair markers (BRCA1&KU70/KU80) in non-TNBC, and a complex of NHEJ markers (KU70/KU80&DNA-PK) are all independent prognostic markers for BC. In addition to expression, subcellular localisation of DDR proteins appeared to be a major factor in their role. Particularly, HR repair markers (but not NHEJ) showed worse features of cytoplasmic location of expression, whereas nuclear expression was associated with more favourable features. Finally, the results of this study provide further evidence to support combined use of IHC with the parallel analytic capability of protein microarray RPPA to investigate protein alterations in human tumours.

There is evidence to suggest that some special types of breast cancer including tubular and lobular carcinoma and their putative precursor lesions including atypical ductal hyperplasia (ADH), low grade ductal carcinoma in-situ (DOS) and lobular neoplasia (LN) may consist in a family of interrelated lesions. Recently, an attention has been focused on the columnar cell lesion as an early non-obligate precursor lesion of breast cancer. In this study we examined this hypothesis by identifying the morphological and biological characteristics of these lesions. In addition, we used high resolution array comparative genomic hybridization to identify the molecular genetic profiles of invasive lobular and tubular carcinoma and their matched coexisting precursor lesions to investigate their relationship and to provide insight into some of the earliest events leading to invasive breast cancer. Moreover, by validating aspects of our immunohistochemical and in situ hybridization expression data with high throughput tissue microarrays, we identified potential oncogenes and tumour suppressor genes that could potentially drive the progression of BC and have clinicopathological implications on BC. Subsequently, diagnostic, predictive and genetic classifications of breast cancer and their putative precursor lesions were developed. In summary, our results suggest that 1) Tubular and lobular breast carcinoma arise as members of a low nuclear grade breast neoplasia (LNGBN) family, 2) CCLs are early non-obligate precursor components of the LNGBN family, 3) The common cell of origin of the LNGBN family may be the oestrogen receptor-alpha (ER α) positive luminal restricted progenitor cell (ER +/MUC1 + cells) of the terminal duct lobular unit that might acquire stochastic genetic and epigenetic changes that eventually lead to activation of the luminal "A" pathway, 4) Cyclin D1 and MDM4 are oncogenes that potentially lead to activation of the luminal pathway and progression of the LNGBN family, 5) An alteration of E-cadherin (CDH1) appears to be a secondary event resulting in the characteristic morphology of both in situ and invasive lobular lesions, 6) A biological grading system dependent on the balance between Bcl2 protein expression and mitotic figures could accurately reclassify patients with intermediately differentiated, small early stage or ER α negative breast cancers into two groups of low versus high risk of death and recurrence, and 7) The functional status of p53 transcriptional pathways can be assessed using immunohistochemistry protein expression of p53 downstream/regulator genes to accurately discriminate between low and high grade breast carcinoma and to assist routine clinical decision-making.

Breast carcinomas are often infiltrated by inflammatory cells, particularly macrophages and T-lymphocytes but the significance of these cells remains unclear. Potential roles include an immune response against the tumour, and pro-tumour effects such as the release of angiogenic factors or proteolytic enzymes. Breast cancer is a remarkably heterogeneous disease embracing a wide range of clinical patterns, stages of presentation, biological behaviour, prognostic characteristics and response to different types of treatment. International efforts have been made to improve survival rate by early diagnosis and multiple therapies. However, the limitations of the current therapeutic modalities have led to increasing enthusiasm for defining new prognostic tools and developing highly targeted therapies. The main aim of this thesis was to examine the characteristics of different components of innate and adaptive immune cells infiltrating breast carcinomas utilising a large cohorts of cases (n= 1902) including various histological and biological subtypes to determine the individual role for each component of breast cancer inflammatory cell infiltrate in disease progression. To this end, in situ immunohistochemical analyses were conducted on breast cancer tissue microarrays (TMAs) using monoclonal antibodies against T-cell, B-cell, macrophage, NK and dendritic cell markers. The interactions between the multiple cells in the tumour microenvironment were subsequently assessed using a powerful bioinformatics tool with the aim of classifying breast carcinomas into prognostic subgroups using multiple phenotypes of infiltrating immune/inflammatory cells. The results of this thesis have demonstrated that phenotype, in addition to the quantity, of tumour-infiltrating lymphocytes is a critical determinant for patient prognosis. In situ cytotoxic T lymphocytic infiltration was associated with a better prognosis independent of standard factors (HR= 0.58, 95% CI= 0.45 0.73, p< 0.001), suggesting that the cell mediated immune response is important in breast cancer. Furthermore, humoral immunity, investigated by the CD20 marker, had an independent prognostic role (HR= 0.75, 95% CI= 0.58-0.96, p= 0.025). The current study has also suggested that macrophage populations do not have a vital role in breast cancer patients’ survival but supported the previously described pro-tumour role played by macrophages in a large cohort. Furthermore, the present study has shown a scant number of dendritic cells infiltrating breast carcinomas. In general, infiltrating leucocytes have preferentially accumulated within the stromal elements away from tumour islands with less observed infiltration within tumour cell nests and this may be due to the higher concentration of chemoattractants in the stroma and/or inhibitory cytokines produced by tumour cells. Moreover, higher levels of inflammatory-cell infiltrate have correlated with higher tumour grade which could be explained by the higher levels of ‘dangerous signals’ in these tumours. The current study has determined a class of tumours that demonstrates higher levels of inflammatory cell infiltrate associated with a better prognosis, suggesting that the immune effects of the inflammation predominate over the pro-tumour effects, despite immunoediting. The present thesis has introduced into literature robust evidence which support the previous proposal to consider the immune reaction as the seventh hallmark of cancer. In conclusion, this thesis proposes an immune scoring based on the phenotype, count, and localisation of leucocytic infiltrates as a novel prognostic factor in breast cancer, in addition to currently available factors to aid in decisions regarding new modalities of immunotherapy. More studies are warranted to translate the histopathological and molecular investigation into clinical management of breast cancer.

Advances in the treatment of ovarian cancer have had a limited impact on prognosis over recent decades. Alternatives to the traditional surgical and chemotherapeutic approach are being sought. Many novel therapies relate to a greater understanding of the molecular changes which occur during carcinogenesis and the development of targeted therapies to exploit these abnormalities. The aim of this thesis was to investigate the prognostic significance of markers relating to tumour immunology, angiogenesis and apoptosis, through the use of Tissue Microarray Technology. 339 cases of ovarian cancers diagnosed between 1982 and 1997 were assessed. Tumours were analysed immunohistochemically for expression of components of the IFNy (IFNGR1, STAT1, p27, caspase 1), TRAIL (DR4 and DR5) and angiogenic (VEGF) pathways. Loss of expression of IFNGR1 was an independent predictor of poor prognosis, although STAT 1 was not. High levels of cytoplasmic and nuclear p27 expression were associated with a reduced survival; cytoplasmic was independently prognostic. Tumours with reduced levels of caspase 1 had improved survival. These results suggest that only patients expressing IFNGR1 may benefit from IFNy therapy and provides evidence of immunoediting in ovarian cancer. DR4 and DR5 did not predict prognosis suggesting that the TRAIL pathway may not be significant in ovarian cancer apoptosis with implications for TRAIL-related therapy. High levels of VEGF occurred infrequently, being an independent marker of poor prognosis. This may identify a group of patients who may preferentially benefit from anti-angiogenic therapy. The thesis illustrates that ovarian cancers are heterogeneous and variations in expression of protein markers can predict tumour behaviour and stratify for therapy. Future targeted therapies may be selected on the basis of an immunohistochemical profile which predicts the pathways that are still functioning. New therapies as they arise should be trialed and targeted to tumours expressing the appropriate molecular markers.