An investigation of the relationship between the components of tumour microenvironment, signal transduction pathways and outcome in breast cancer

Breast cancer, the most commonly diagnosed type of cancer in women, is a major cause of morbidity and mortality in the western world. Well-established risk factors of breast cancer are mostly related to women’s reproductive history, such as early menarche, late first pregnancy and late menopause. Survival rates have improved due to a combination of factors, including better health education, early detection with large-scale use of screening mammogram, improved surgical techniques, as well as widespread use of adjuvant therapy. At initial presentation, clinicopathological features of breast cancer such as age, nodal status, tumour size, tumour grade, and hormonal receptor status are considered to be the standard prognostic and predictive markers of patient survival, and are used to guide appropriate treatment strategies. Lymphovascular invasion (LBVI), including lymphatic (LVI) and blood (BVI) vessel invasion, has been reported to be prognostic and merit accurate evaluation, particularly in patients with node negative tumours who might benefit from adjuvant chemotherapy. There is a lack of standard assessment and agreement on distinguishing LVI from BVI despite the major challenges in the field. A systematic review of the literatures, examining methods of detection and the prognostic significance of LBVI, LVI and BVI, was carried out. The majority of studies used haematoxylin and eosin (H&E) and classical histochemistry to identify LVI and BVI. Only few recent studies used immunohistochemistry (IHC) staining of the endothelium lining lymphatic and blood vessels, and were able to show clear differences between LVI and BVI. The prognostic significance of LBVI and LVI was well-documented and strongly associated with aggressive features of breast tumours, while the prognostic value and the optimal detection method of BVI were unclear. Assessment and prognostic value of LBVI on H&E sections (LBVIH&E) was examined and compared to that of LVI and BVI detected using IHC with D2-40 for LVI (LVID2–40) and Factor VIII for BVI (BVIFVIII) in patients with breast cancer including node negative and triple negative patients (n=360). LBVIH&E, LVID2–40 and BVIFVIII were present in 102 (28%), 127 (35%) and 59 (16%) patients respectively. In node negative patients (206), LBVIH&E, LVID2–40 and BVIFVIII were present in 41 (20%), 53 (26%) and 21 (10%) respectively. In triple negative patients (102), LBVIH&E, LVID2–40 and BVIFVIII were present in 35 (29%), 36 (35%) and 14 (14%) respectively. LBVIH&E, LVID2–40 and BVIFVIII were all significantly associated with tumour recurrence in all cohorts. On multivariate survival analysis, only LVID2–40 and BVIFVIII were independent predictors of cancer specific survival (CSS) in the whole cohort (P=0.022 and P<0.001 respectively), node negative (P=0.008 and P=0.001 respectively) and triple negative patients (P=0.014 and P<0.001 respectively). Assessment of LVI and BVI by IHC, using D2-40 and Factor VIII, improves prediction of outcome in patients with node negative and triple negative breast cancer and was superior to the conventional detection method. Breast cancer is recognised as a complex molecular disease and histologically identical tumours may have highly variable outcomes, including different responses to therapy. Therefore, there is a compelling need for new prognostic and predictive markers helpful of selecting patients at risk and patients with aggressive diseases who might benefit from adjuvant and targeted therapy. It is increasingly recognised that the development and progression of human breast cancer is not only determined by genetically abnormal cells, but also dependent on complex interactions between malignant cells and the surrounding microenvironment. This has led to reconsider the features of tumour microenvironment as potential predictive and prognostic markers. Among these markers, tumour stroma percentage (TSP) and tumour budding, as well as local tumour inflammatory infiltrate have received recent attention. In particular, the local environment of cytokines, proteases, angiogenic and growth factors secreted by inflammatory cells and stromal fibroblasts has identified crucial roles in facilitating tumour growth, and metastasis of cancer cells through lymphatic and/or blood vessel invasion. This might help understand the underlying process promoting tumour invasion into these vessels. An increase in the proportion of tumour stroma and an increase in the dissociation of tumour cells have been associated with poorer survival in a number of solid tumours, including breast cancer. However, the interrelationship between these variables and other features of the tumour microenvironment in different subgroups of breast cancer are not clear. Also, whether their prognostic values are independent of other components of the tumour microenvironment have yet to be identified. Therefore, the relationship between TSP, clinicopathological characteristics and outcome in patients with invasive ductal breast cancer, in particular node negative and triple negative disease was examined in patients with invasive ductal breast cancer (n=361). The TSP was assessed on the haematoxylin and eosin-stained tissue sections. With a cut-off value of 50% TSP, patients with ≤50% stroma were classified as the low-TSP group and those with >50% stroma were classified as the high-TSP group. A total of 109 (30%) patients had high TSP. Patients with high TSP were old age (P=0.035), had involved lymph node (P=0.049), Her-2 positive tumours (P=0.029), low-grade peri-tumoural inflammatory infiltrate (P=0.034), low CD68+ macrophage infiltrate (P<0.001), low CD4+ (P=0.023) and low CD8+ T-lymphocytes infiltrate (P=0.017), tumour recurrence (P=0.015) and shorter CSS (P<0.001). In node negative patients (n=207), high TSP was associated with low CD68+ macrophage infiltrate (P=0.001), low CD4+ (P=0.040) and low CD8+ T-lymphocytes infiltrate (P=0.016) and shorter CSS (P=0.005). In triple negative patients (n=103), high TSP was associated with increased tumour size (P=0.017) high tumour grade (P=0.014), low CD8+ T-lymphocytes infiltrate (P=0.048) and shorter CSS (P=0.041). The 15-year cancer specific survival rate was 79% vs 21% in the low-TSP group vs high-TSP group. On multivariate survival analysis, a high TSP was associated with reduced CSS in the whole cohort (P=0.007), node negative patients (P=0.005) and those who received systemic adjuvant therapy (P=0.016), independent of other pathological characteristics including local host inflammatory responses. Therefore, a high TSP in invasive ductal breast cancer was associated with recurrence and poorer long-term survival. The inverse relation with the tumour inflammatory infiltrate highlights the importance of the amount of tumour stroma on immunological response in patients with invasive ductal breast cancer. Implementing this simple and reproducible parameter in routine pathological examination may help optimise risk stratification in patients with breast cancer. Similarly, the relationship between tumour budding, clinicopathological characteristics and outcome was examined in patients with invasive ductal breast cancer (n=474), using routine pathological sections. Tumour budding was associated with several adverse pathological characteristics, including positive lymph node (P=0.009), presence of LVI (P<0.001), and high TSP (P=0.001) and low-grade general peri-tumural inflammatory infiltrative (P=0.002). In node negative patients, a high tumour budding was associated with presence of LVI (P<0.001) and low-grade general peri-tumural inflammatory infiltrative (P=0.038). On multivariate survival analysis, tumour budding was associated with reduced CSS (P=0.001), independent of nodal status, tumour necrosis, CD8+ and CD138+ inflammatory cells infiltrate, LVI, BVI and TSP. Furthermore, tumour budding was independently associated with reduced CSS in node negative patients (P=0.004) and in those who have low TSP (P=0.003) and high-grade peri-tumoural inflammatory infiltrative (P=0.012). A high tumour budding was significantly associated with shorter CSS in luminal B and triple negative breast cancer subtypes (all P<0.001). Therefore, tumour budding was a significant predictor of poor survival in patients with invasive ductal breast cancer, independent of adverse pathological characteristics and components of tumour microenvironment. These results suggest that tumour budding may promote disease progression through a direct effect on local and distant invasion into lymph nodes and lymphatic vessels. Therefore, detection of tumour buds at the stroma invasive front might therefore represent a morphologic link between tumour progression, lymphatic invasion, spread of tumour cells to regional lymph nodes, and the establishment of metastatic dissemination. Given the potential importance of the tumour microenvironment, the characterisation of intracellular signalling pathways is important in the tumour microenvironment and is of considerable interest. One plausible signalling molecule that links tumour stroma, inflammatory cell infiltrate and tumour budding is the signal transducer and activator of transcription (STAT). The relationship between total and phosphorylated STAT1 (ph-STAT1), and total and ph-STAT3 tumour cell expression, components of tumour microenvironment and survival in patients with invasive ductal breast cancer was examined. IHC of total and ph-STAT1/STAT3 was performed on tissue microarray of 384 breast cancer specimens. Cellular STAT1 and cellular STAT3 expression at both cytoplasmic and nuclear locations were combined and identified as STAT1/STAT3 tumour cell expression. These results were then related to CSS and phenotypic features of the tumour and host. A high ph-STAT1 and a high ph-STAT3 tumour cell expression was associated with increased ER (P=0.001 and P<0.001 respectively) and PR (all P<0.05), reduced tumour grade (P=0.015 and P<0.001 respectively) and necrosis (all P=0.001). Ph-STAT1 was associated with increased general peri-tumoural inflammatory infiltrate (P=0.007) and ph-STAT3 was associated with lower CD4+ T-lymphocyte infiltrate (P=0.024). On multivariate survival analysis, including both ph-STAT1 and ph-STAT3 tumour cell expression, only high ph-STAT3 tumour cell expression was significantly associated with improved CSS (P=0.010) independent of other tumour and host-based factors. In patients with high necrosis grade, high ph-STAT3 tumour cell expression was independent predictor of improved CSS (P=0.021). Ph-STAT1 and ph-STAT3 were also significantly associated with improved cancer specific survival in luminal A and B subtypes. STAT1 and STAT3 tumour cell expression appeared to be an important determinant of favourable outcome in patients with invasive ductal breast cancer. The present results suggest that STATs may affect disease outcome through direct impact on tumour cells, and the surrounding microenvironment. The above observations of the present thesis point to the importance of the tumour microenvironment in promoting tumour budding, LVI and BVI. The observations from STATs work may suggest that an important driving mechanism for the above associations is the presence of tumour necrosis, probably secondary to hypoxia. Further work is needed to examine the interaction of other molecular pathways involved in the tumour microenvironment, such as HIF and NFkB in patients with invasive ductal breast cancer.

Soft tissue sarcoma: biology and therapeutic correlates

Soft tissue sarcomas (STS) comprise a heterogenenous group of greater than 50 malignancies of putative mesenchymal cell origin and as such they may arise in diverse tissue types in various anatomical locations throughout the whole body. Collectively they account for approximately 1% of all human malignancies yet have a spectrum of aggressive behaviours amongst their subtypes. They thus pose a particular challenge to manage and remain an under investigated group of cancers with no generally applicable new therapies in the past 40 years and an overall 5-year survival rate that remains stagnant at around 50%. From September 2000 to July 2006 I undertook a full time post-doctoral level research fellowship at the MD Anderson Cancer Center, Houston, Texas, USA in the department of Surgical Oncology to investigate the biology of soft tissue sarcoma and test novel anti- sarcoma adenovirus-based therapy in the preclinical nude rat model of isolated limb perfusion against human sarcoma xenografts. This work, in collaboration with colleagues as indicated herein, led to a number of publications in the scientific literature furthering our understanding of the malignant phenotype of sarcoma and reported preclinical studies with wild-type p53, in a replication deficient adenovirus vector, and oncolytic adenoviruses administered by isolated limb perfusion. Additional collaborative and pioneering preclinical studies reported the molecular imaging of sarcoma response to systemically delivered therapeutic phage RGD-4c AAVP. Doxorubicin chemotherapy is the single most active broadly applicable anti-sarcoma chemotherapeutic yet only has an approximate 30% overall response rate with additional breakthrough tumour progression and recurrence after initial chemo-responsiveness further problematic features in STS management. Doxorubicin is a substrate for the multi- drug resistance (mdr) gene product p-glycoprotein drug efflux pump and exerts its main mode of action by induction of DNA double-strand breaks during the S-phase of the cell cycle. Two papers in my thesis characterise different aspects of chemoresistance in sarcoma. The first shows that wild-type p53 suppresses Protein Kinase Calpha (PKCα) phosphorylation (and activation) of p-glycoprotein by transcriptional repression of PKCα through a Sp-1 transcription factor binding site in its -244/-234 promoter region. The second paper demonstrates that Rad51 (a central mediator of homologous recombination repair of double strand breaks) has elevated levels in sarcoma and particularly in the S- G2 phase of the cell cycle. Suppression of Rad51 with small interfering RNA in sarcoma cell culture led to doxorubicin chemosensitisation. Reintroduction of wild-type p53 into STS cell lines resulted in decreased Rad51 protein and mRNA expression via transcriptional repression of the Rad51 promoter through increased AP-2 binding. In light of poor response rates to chemotherapy, escape from local control portends a poor prognosis for patients with sarcoma. Two papers in my thesis characterise aspects of sarcoma angiogenesis, invasion and metastasis. Human sarcoma samples were found to have high levels of matrix metalloproteinase-9 (MMP-9) with expression levels that correlated with p53 mutational status. MMP-9 is known to degrade extracellular collagen, contribute to the control of the angiogenic switch necessary in primary tumour progression and facilitate invasion and metastasis. Reconstitution of wild-type p53 function led to decreased levels of MMP-9 protein and mRNA as well as zymography-assessed MMP-9 proteolytic activity and decreased tumour cell invasiveness. Reintroduction of wild-type p53 into human sarcoma xenografts in-vivo decreased tumour growth and MMP-9 protein expression. Wild-type p53 was found to suppress mmp-9 transcription via decreased binding of NF-κB to its -607/-595 mmp-9 promoter element. Studies on the role of the VEGF165 in sarcoma found that sarcoma cells stably transfected with VEGF165 formed more aggressive xenografted tumours with increased vascularity, growth rate, metastasis, and resistance to chemotherapy. Use of the anti-VEGFR2 antibody DC101 enhanced doxorubicin sensitivity at sub-conventional dosing, inhibited tumour growth, decreased development of metastases, and reduced tumour micro-vessel density while increasing the vessel maturation index. These effects were explained primarily through effects on endothelial cells (e.c.s), rather than the tumour cells per se, where DC101 induced e.c. sensitivity to doxorubicin and suppressed e.c. production of MMPs. The p53 tumour suppressor pathway is the most frequently mutated pathway in sarcoma. Recapitulation of wild-type p53 function in sarcoma exerts a number of anti-cancer outcomes such as growth arrest, resensitisation to chemotherapy, suppression of invasion, and attenuation of angiogenesis. Using a modified nude rat-human sarcoma xenograft model for isolated limb perfusion (ILP) delivery of wild-type p53 in a replication deficient adenovirus vector I showed that functionally competent wild-type p53 could be delivered to and detected in human leiomyosarcoma xenografts confirming preclinical feasibility - although not efficacious due to low transgene expression. Viral fibre modification to express the RGD tripeptide motif led to greater viral uptake by sarcoma cells in vitro (transductional targeting) and changing the transgene promoter to a response element active in cells with active telomerase expression restricted the transgene expression to the tumour intracellular environment (transcriptional targeting). Delivery of the fibre-modified, selectively replication proficient oncolytic adenovirus Ad.hTC.GFP/ E1a.RGD by ILP demonstrated a more robust, and tumour-restricted, transgene expression with evidence of anti-sarcoma effect confirmed microscopically. Collaborative studies using the fibre modified phage RGD-4C AAVP confirmed that systemic delivery specifically, efficiently, and repeatedly targets human sarcoma xenografts, binds to αv integrins in tumours, and demonstrates a durable, though heterogeneous, transgene expression of 1-4 weeks. Incorporation of the Herpes Simplex Virus thymidine kinase (HSVtk) transgene into RGD-4C AAVP permitted CT-PET spatial and temporal molecular imaging in vivo of transgene expression and allowed quantification of tumour metabolic activity both before and after interval administration of a systemic cytotoxic with predictable and measurable response to treatment before becoming apparent clinically. These papers further the medical and scientific community’s understanding of the biology of soft tissue sarcoma and report preclinical studies with novel and promising anti- sarcoma therapeutics.

The metabolic and epigenetic effects of the isocitrate dehydrogenase-1 mutation in glioma

Cancer cells have been noted to have an altered metabolic phenotype for over ninety years. In the presence of oxygen, differentiated cells predominately utilise the tricarboxylic acid (TCA) cycle and oxidative phosphorylation to efficiently produce energy and the metabolites necessary for protein and lipid synthesis. However, in hypoxia, this process is altered and cells switch to a higher rate of glycolysis and lactate production to maintain their energy and metabolic needs. In cancer cells, glycolysis is maintained at a high rate, even in the presence of oxygen; a term described as “aerobic glycolysis”. Tumour cells are rapidly dividing and have a much greater need for anabolism compared to normal differentiated cells. Rapid glucose metabolism enables faster ATP production as well as a greater redistribution of carbons to nucleotide, protein, and fatty acid synthesis, thus maximising cell growth. Recently, other metabolic changes, driven by mutations in genes related to the TCA cycle, indicate an alternative role for metabolism in cancer, the “oncometabolite”. This is where a particular metabolite builds up within the cell and contributes to the tumorigenic process. One of these genes is isocitrate dehydrogenase (IDH) IDH is an enzyme that forms part of the tricarboxylic acid (TCA) cycle and converts isocitrate to α-ketoglutarate (α-KG). It exists in three isoforms; IDH1, IDH2 and IDH3 with the former present in the cytoplasm and the latter two in the mitochondria. Point mutations have been identified in the IDH1 and IDH2 genes in glioma which result in a gain of function by converting α-KG to 2-hydroxyglutarate (2HG), an oncometabolite. 2HG acts as a competitive inhibitor of the α-KG dependent dioxygenases, a superfamily of enzymes that are involved in numerous cellular processes such as DNA and histone demethylation. It was hypothesised that the IDH1 mutation would result in other metabolic changes in the cell other than 2HG production, and could potentially identify pathways which could be targeted for therapeutic treatment. In addition, 2HG can act as a potential competitive inhibitor of α-KG dependent dioxygenases, so it was hypothesised that there would be an effect on histone methylation. This may alter gene expression and provide a mechanism for tumourogenesis and potentially identify further therapeutic targets. Metabolic analysis of clinical tumour samples identified changes associated with the IDH1 mutation, which included a reduction in α-KG and an increase in GABA, in addition to the increase in 2HG. This was replicated in several cell models, where 13C labelled metabolomics was also used to identify a possible increase in metabolic flux from glutamate to GABA, as well as from α-KG to 2HG. This may provide a mechanism whereby the cell can bypass the IDH1 mutation as GABA can be metabolised to succinate in the mitochondria by GABA transaminase via the GABA shunt. JMJ histone demethylases are a subset of the α-KG dependent dioxygenases, and are involved in removing methyl groups from histone tails. Changes in histone methylation are associated with changes in gene expression depending on the site and extent of chemical modification. To identify whether the increase in 2HG and fall in α-KG was associated with inhibition of histone demethylases a histone methylation screen was used. The IDH1 mutation was associated with an increase in methylation of H3K4, which is associated with gene activation. ChiP and RNA sequencing identified an increase in H3K4me3 at the transcription start site of the GABRB3 subunit, resulting in an increase in gene expression. The GABRB3 subunit forms part of the GABA-A receptor, a chloride channel, which on activation can reduce cell proliferation. The IDH1 mutation was associated with an increase in GABA and GABRB3 subunit of the GABA-A receptor. This raises the possibility of GABA transaminase as a potential therapeutic target. Inhibition of this enzyme could reduce GABA metabolism, potentially reducing any beneficial effect of the GABA shunt in IDH1 mutant tumours, and increasing activation of the GABA-A receptor by increasing the concentration of GABA in the brain. This in turn may reduce cell proliferation, and could be achieved by using Vigabatrin, a GABA transaminase inhibitor licensed for use in epilepsy.

The regulation of cancer cell invasive behaviour by Chloride Interacellular Channel 3 (CLIC3) and Transglutaminasae 2

A study into the role of secreted CLIC3 in tumour cell invasion. The initiation and progression of cancers is thought to be linked to their relationship with a population of activated fibroblasts, which are associated with tumours. I have used an organotypic approach, in which plugs of collagen I are preconditioned with fibroblastic cells, to characterise the mechanisms through which carcinoma-associated fibroblasts (CAFs) influence the invasive behaviour of tumour cells. I have found that immortalised cancer-associated fibroblasts (iCAFs) support increased invasiveness of cancer cells, and that this is associated with the ability of CAFs to increase the fibrillar collagen content of the extracellular matrix (ECM). To gain mechanistic insight into this phenomenon, an in-depth SILAC-based mass proteomic analysis was conducted, which allowed quantitative comparison of the proteomes of iCAFs and immortalised normal fibroblast (iNFs) controls. Chloride Intracellular Channel Protein 3 (CLIC3) was one of the most significantly upregulated components of the iCAF proteome. Knockdown of CLIC3 in iCAFs reduced the ability of these cells to remodel the ECM and to support tumour cell invasion through organotypic plugs. A series of experiments, including proteomic analysis of cell culture medium that had been preconditioned by iCAFs, indicated that CLIC3 itself was a component of the iCAF secretome that was responsible for the ability of iCAFs to drive tumour cell invasiveness. Moreover, addition of soluble recombinant CLIC3 (rCLIC3) was sufficient to drive the extension of invasive pseudopods in cancer cell lines, and to promote disruption of the basement membrane in a 3D in vitro model of the ductal carcinoma in situ (DCIS) to invasive ductal carcinoma (IDC) transition. My investigation into the mechanism through which extracellular CLIC3 drives tumour cell invasiveness led me to focus on the relationship between CLIC3 and the ECM modifying enzyme, transglutaminase-2 (TG2). Through this, I have found that TG2 physically associates with CLIC3 and that TG2 is necessary for CLIC3 to drive tumour cell invasiveness. These data identifying CLIC3 as a key pro-invasive factor, which is secreted by CAFs, provides an unprecedented mechanism through which the stroma may drive cancer progression.