High-grade serous ovarian carcinoma (HGSOC) is characterised by genomic instability, ubiquitous TP53 loss, widespread disease at diagnosis and the frequent emergence of platinum resistance. This thesis explores the use of high-throughput genomics technologies to understand if resistance could be explained by the model of tumour evolution. We performed SNP array analysis of a cell line model system of platinum resistance consisting of matched cell lines from three cases of HGSOC established before and after clinical resistance developed, the OVOl clinical study consisting of six matched pairs of tumours before and after three cycles of chemotherapy, and the OV03/0V04 study consisting of 18 cases sampled at multiple timepoints and from multiple metastatic sites. The results showed evidence of metastatic site dependent divergence. Moreover, mutually exclusive loss of heterozygosity patterns between presentation and relapse genomes, including all the cases in the cell line system and one of two OV03 cases for which relapse material was available, suggest that the relapse arises from a minor subclone of the presentation disease, while in the remaining case, the subclone with an NFJ homozygous deletion was enriched in the relapsed disease. I then asked which mutational process drives evolution. Using next-generation sequencing (NGS), I compared the structural variants between and within cases in the model system and in 6 cases of the OV03 cohort. From the genomic signatures in the cell lines, I demonstrated that a case with homologous recombination (HR) deficiency acquired numerous translocations and small deletions (median size of 13.4kb) , whereas another showed a novel tandem duplicator phenotype (median size of tandem duplications was 350kb). Mutator phenotypes in both cases arose early in progression and persisted, but the tumour with HR deficiency showed evidence of re-stabilising its ,"genome and lost platinum sensitivity after a revertant BRCA2 mutation restored its HR function. A subset of tumours from the Cancer Genome Atlas (TCGA) dataset suggested that these two phenotypes were mutually exclusive. Amongst the six OV03 cases, preliminary analysis suggests that one case showed an amplifier phenotype and three cases showed evidence of parallel evolution. Taken together, early onset of mutator phenotypes and parallel evolution may provide a mechanism by which resistance evolves. Further work should aim to identify the processes involved in tumour evolution in 'purified' populations such as cancer stem cells.

No description available.

Chromosome changes in the haematological malignancies, lymphomas and sarcomas are known to be important events in the evolution of these tumours as they can, for example, form fusion oncogenes or disrupt tumour suppressor genes. The recently described recurrent fusion genes in prostate and lung cancer proved to be iconic examples as they indicated that important gene fusions are found in the common epithelial cancers also. Breast cancers often display extensive structural and numerical chromosome aberration and have among the most complex karyotyes of all cancers. Genome rearrangements are potentially an important source of mutation in breast cancer but little is known about how they might contribute to this disease. My first aim was to carry out a structural survey of breast cancer cell line genomes in order to find genes that were disrupted by chromosome aberrations in 'typical' breast cancers. I investigated three breast cancer cell lines, HCC1187, VP229 and VP267 using data from array painting, SNP6 array CGH, molecular cytogenetics and massively parallel paired end sequencing. I then used these structural genomic maps to predict fusion transcripts and demonstrated expression of five fusion transcripts in HCC1187, three in VP229 and four inVP267. Even though chromosome aberrations disrupt and fuse many genes in individual breast cancers, a major unknown is the relative importance and timing of genome rearrangements compared to sequence-level mutation. For example, chromosome instability might arise early and be essential to tumour suppressor loss and fusion gene formation or be a late event contributing little to cancer development. To address this question, I considered the evolution of these highly rearranged breast cancer karyotypes. The VP229 and VP267 cell lines were derived from the same patient before and after therapy-resistant relapse, so any chromosome aberration found in both cell lines was probably found in the common in vivo ancestor of the two cell lines. A large majority of structural variants detected by massively parallel paired end sequencing, including three fusion transcripts, were found in both cell lines, and therefore, in the common ancestor. This probably means that the bulk of genome rearrangement pre-dated the relapse. For HCC1187, I classified most of its mutations as earlier or later according to whether they occurred before or after a landmark event in the evolution of the genome-endoreduplication (duplication of its entire genome). Genome rearrangements and sequence-level mutations were fairly evenly divided between earlier and later, implying that genetic instability was relatively constant throughout the evolution of the tumour. Surprisingly, the great majority of inactivating mutations and expressed gene fusions happened earlier. The non-random timing of these events suggests many were selected.

Despite the considerable expansion of knowledge in the development of pancreatic cancer, there has been little progress made in facilitating an early diagnosis of this disease and predicting an accurate response to treatment. We aim to translate this knowledge to clinical practice by using a prospective database of precursor cystic lesions in pancreas cancer, assessing the use of over-expressed genes in pancreatic juice as a surrogate marker of these pancreas cancer and finally, downstream of these changes at the genetic level, use metabolomic techniques to look for biomarkers in pancreas cancer in serum. In the first study, we investigate the natural history of pancreatic cystic neoplasms, specifically IPMNs, using a prospectively collected database to examine the profiles and outcomes of main duct IPMN, branch duct IPMN and cystic lesions measuring less than 3 cm in size. A total of 99 patients with suspected pancreatic cystic tumours were enrolled over 3 years. Median follow-up was 24 months (range 0 – 124). Cystic tumours comprised of 13 MD-IPMN, 40 BD-IPMN, 11 MCN and 8 adenocarcinomas among others. The complete cohort showed an overall risk of adenocarcinoma of 8%. Main duct IPMN showed a cumulative risk of 46% with evidence of progression of disease in a further 23%. The associated mortality in MD-IPMN was related to the underlying adenocarcinoma and was 38% in our group. The incidence of adenocarcinoma in branch duct IPMN was 11% with disease progression seen 13.8%. Evidence of extra-pancreatic malignancies was seen in 37.7% of patients with IPMN. In the second study, we explore the feasibility of gene expression profiling from RNA isolated from matched pancreatic juice and tumour tissue in patients with pancreatic cancer and pancreatic cystic tumours. RNA was isolated and Poly(A) PCR was used to globally amplify the RNA. RT-PCR was used to measure expression levels of 18 genes common to both pancreas cancer and pancreatic cystic tumours. Spearman’s rank correlation test was used to examine the relationship of gene expression between pancreatic juice and tissue. One gene out of eighteen, MSLN (p<0.008), showed significant correlation in the expression levels between paired pancreatic juice and tissue samples in pancreas cancer. In the cystic tumour group, only one gene MMP-7 (p<0.01), showed a significant correlation between paired juice and tissue samples. When the whole cohort was analysed for the false discovery rate, these genes did not exhibit statistically significant correlation between the samples. RNA analysis of pancreatic juice is feasible using the Poly(A) cDNA technique and correlation of gene expression is shown to exist, albeit with low sensitivity, indicating its potential use in clinical practice with small tissue and juice samples. In the final study, we performed a literature review on the use of metabolomics in pancreas cancer. We performed metabolic profiling of serum samples from selected cancer patients and noncancerous controls using UPHLC-MS to generate and compare the metabolic profiles in serum samples from a cohort of patients with pancreas cancer, ampullary cancer and endocrine cancer. Thirty nine serum samples (including 19 pancreatic cancers, 9 ampullary cancers and 5 endocrine cancers) and 21 matched HUSERMET controls were analysed using Ultra high performance liquid chromatography mass spectrometry (UHPLC-MS) in both positive and negative ESI modes. The output was generated as a data matrix of mass spectral features with related accurate m/z and retention time pairs. The data was then signal corrected and individual peaks were normalised and the resultant spectra were compared against a metabolite reference library and analysed using univariate and multivariate statistical tests. We found a disparity in the metabolite peaks between the cases and controls on PCA that did not permit the accurate interpretation of the data in the case study set compared to the control set. No obvious reason other than metabolite degradation during storage could account for this difference. PC-DFA analysis of metabolite peaks between pancreas cancer, ampullary cancer and endocrine cancer showed significant difference between endocrine cancers and the other two groups. Significant ESI positive metabolites included those involved in lipid pathways and metabolites involved in glucose metabolism. There is encouraging scope for studies using prospective controls to identify and develop metabolic biomarkers in pancreas cancer.

京都大学 / 新制・課程博士 / 博士(医学) / 甲第24507号 / 医博第4949号 / 新制||医||1064(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 村川 泰裕, 教授 金子 新, 教授 永井 純正 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

<p>The accumulation of high-throughput data from vast sources has drawn a lot attentions to develop methods for extracting meaningful information out of the massive data. More interesting questions arise from how to combine the disparate information, which goes beyond modeling sparsity and dimension reduction. This dissertation focuses on the innovations in the area of heterogeneous data integration.</p><p>Chapter 1 contextualizes this dissertation by introducing different aspects of meta-analysis and model frameworks for high-dimensional genomic data.</p><p>Chapter 2 introduces a novel technique, joint Bayesian sparse factor analysis model, to vertically integrate multi-dimensional genomic data from different platforms. </p><p>Chapter 3 extends the above model to a nonparametric Bayes formula. It directly infers number of factors from a model-based approach.</p><p>On the other hand, chapter 4 deals with horizontal integration of diverse gene expression data; the model infers pathway activities across various experimental conditions. </p><p>All the methods mentioned above are demonstrated in both simulation studies and real data applications in chapters 2-4.</p><p>Finally, chapter 5 summarizes the dissertation and discusses future directions.</p> / Dissertation

Intratumoural heterogeneity complicates the molecular interpretation of biopsies, as multiple distinct tumour genomes are sampled and analysed at once. Ignoring the presence of these populations can lead to erroneous conclusions, and so a correct analysis must account for the clonal structure of the sample. Several methods to reconstruct tumour clonality from sequencing data have been proposed, spanning methods that either do not consider phylogenetic constraints or posit a perfect phylogeny. Models of the first type are typically latent feature models that can describe the observed data flexibly, but whose results may not be reconcilable with a phylogeny. The second type, instead, generally comprises non-parametric mixture models, with strict assumptions on the tumour’s evolutionary process. The focus of this dissertation is on the development of a phylogenetic latent feature model that can bridge the advantages of these two approaches, allowing deviations from a perfect phylogeny. The work is recounted by three statistical models of increasing complexity. First, I present a non-parametric model based on the Indian Buffet Process prior, and highlight the need for phylogenetic constraints. Second, I develop a finite, phylogenetic extension of the previous model, and show that it can outperform competing methods. Third, I generalise the phylogenetic model to arbitrary copy-number states. Markov chain Monte Carlo algorithms are presented to perform inference. The models are tested on datasets that include synthetic data, controlled biological data, and clinical data. In particular, the copy-number generalisation is applied to longitudinal circulating tumour DNA samples. Liquid biopsies that leverage circulating tumour DNA require sensitive techniques in order to detect mutations at low allele fractions. One method that allows sensitive mutation calling is the amplicon sequencing strategy TAm-Seq. I present bioinformatic tools to improve both the development of TAm-Seq amplicon panels and the analysis of its sequencing data. Finally, an enhancement of this method is presented and shown to detect mutations de novo and in a multiplexed manner at allele fractions less than 0.1%.

Background: Despite multi-modality therapy, the majority of patients with metastatic or recurrent Ewing sarcoma (ES), the second most common pediatric bone malignancy, will die of their disease. ES tumors express aberrantly activated ETS transcription factors through translocations that fuse the EWS gene to ETS family genes FLI1 or ERG. The aberrant activation of ETS transcription factors promotes malignant transformation and proliferation. While, FLI1 or ERG cannot be readily targeted, there is an opportunity to deploy functional genomics screens, to develop novel therapeutic approaches by identifying targetable liabilities in EWS/FLI1 dependent tumors. Materials and Methods: We performed a near whole-genome pooled shRNA screen in a panel of five EWS/FLI1 dependent Ewing sarcoma cell lines and one EWS/ERG cell line to identify essential genes. Essential genes were defined as those genes whose loss resulted in reduced viability selectively in ES cells compared to non-Ewing cancer cell lines. Essential hits were subsequently validated with genomic knockdown and chemical inhibition in vitro, followed by validation of the on-target effect of chemical inhibition. Next, we determined the in vivo effects of small-molecule inhibition on survival and tumor growth in NOD scid gamma (NSG) mice with established subcutaneous ES xenografts. Results: Top hits in our screen that could be readily targeted by small-molecule inhibitors, and thus have potential for rapid clinical validation, were selected for further investigation. These hits included IKBKE, CCND1 and CDK4. IKBKΕ, a non-canonical IKK with an oncogenic role in breast cancer, was one of the top kinase hits in the screen. IKBKΕ shares significant homology to TBK1, another non-canonical IKK that is essential in k-RAS dependent lung cancer. We validated IKBKE through small-molecule inhibition of IKBKE/TBK1 and shRNA based knockdown. Ewing sarcoma cell lines are sensitive to low micromolar concentrations of two IKBKE/TBK1 inhibitors (CYT387 and MRT67307). Additionally, in a panel of ES cell lines, knockdown of IKBKE resulted in decreased growth and impaired colony formation. These observations, paired with impairment of NF-κB nuclear localization following CYT387 treatment suggests that non-canonical IKK mediated signaling may be essential in Ewing sarcoma. We further validated these results through inhibition of IKBKE/TBK1 in in vivo xenograft models treated with 100 mg/kg/day of CYT387. Treatment over the course of twenty-nine days resulted in a significant increase in survival (p-value = 0.0231) and a significant decrease (p-value = 0.036) in tumor size after fifteen days of treatment. CDK4 and CCND1 are highly expressed in Ewing sarcoma as compared to other tumor types. shRNA mediated knockdown of CDK4 and CCND1 resulted in impaired viability and anchorage independent growth. Furthermore, treatment of Ewing sarcoma cell lines with a highly selective CDK4/6 inhibitor, LEE011, resulted in decreased viability (IC50 range of 0.26-18.06 μM), potent G1 arrest in six of eight EWS/FLI1 containing Ewing sarcoma lines tested and apoptosis in a panel of four highly sensitive lines. Administration of 75 mg/kg/day and 250 mg/kg/day of LEE011 in NSG mice with Ewing xenografts resulted in significant impairment of tumor growth, (p-value <0.001 for both treatment arms), as compared to vehicle control. Conclusions: These studies suggest a role for the targeting of IKBKE and CDK 4/6 in Ewing sarcoma, findings with immediate clinical relevance for patients with this malignancy, because small-molecule inhibitors of these proteins have already entered clinical trial for other disease indications.

The non-coding space makes up around 98 % of the genome, but cancer-driving mutations have so far mostly been discovered in protein-coding regions. The majority of somatic non-coding mutations are neutral passenger mutations and identifying non-coding mutations with driving roles in cancer poses a challenge. In this work, evolutionary constraint was used to explore the non-coding space in human osteosarcoma to improve our understanding of how evolutionary constraint can be applied to identify non-coding driver mutations in cancer and describe the unknown role of non-coding mutations in osteosarcoma. Evolutionary constraint scores derived from an alignment of 33 mammals were used to extract non-coding mutations in functional elements from somatic variants of 38 osteosarcoma samples and genes with an enrichment of non-coding constraint mutations in their regulatory regions were identified. The investigation of those genes revealed that non-coding constraint mutations are likely involved in key osteosarcoma pathways. Furthermore, novel osteosarcoma genes and mechanisms were proposed based on the non-coding constraint mutation enrichment analysis. The regulatory potential of individual non-coding constraint mutations was evaluated based on regulatory annotations, functional evidence, transcription factor affinity predictions and electrophoretic mobility shift assays. We concluded that the analysis of non-coding constraint mutations is an efficient way to discover non-coding mutations with functional impact in osteosarcoma which likely play an important role in the disease.

No description available.