Sir Peter MacCallum Department of Oncology - Theses
Permanent URI for this collection
Search Results
1 - 4 of 4
-
ItemNo Preview AvailableDiscovery and validation of novel ovarian carcinoma predisposition genes( 2021)Epithelial ovarian carcinoma (EOC) has a significant hereditary component, over half of which cannot be explained by known hereditary breast and ovarian cancer (HBOC) genes (e.g. BRCA1 and BRCA2). Gene discovery studies to date have generally been limited to a small number of candidate genes in relatively few families and have failed to consistently identify any compelling new genes that may account for this missing heritability. The underlying hypothesis of this thesis is that the remaining unexplained EOC families are due to individually rare deleterious variants in numerous genes, each explaining a small proportion of families. To overcome the limitations of earlier targeted panel sequencing efforts, germline whole exome sequencing (WES) was performed on 516 likely familial high-grade serous ovarian cancer (HGSOC) patients with no pathogenic variants in BRCA1 or BRCA2 to discover novel predisposition genes. Forty-three candidate genes enriched for rare loss-of-function (LoF) variants were identified, along with LoF variants in several proposed EOC predisposition genes (e.g. ATM, PALB2). A high degree of genetic heterogeneity was observed, with no single gene harbouring LoF variants in more than 1% of cases. These candidate genes represent diverse functional pathways, with relatively few involved in DNA repair and only a small enrichment for genes involved in homologous recombination (HR) repair. This suggests that many of the remaining HGSOC families are explained by genes in pathways that have been previously under-explored. Since candidate gene variants were individually very rare, orthogonal approaches of tumour sequencing and segregation analysis were undertaken to validate these genes. WES and/or Sanger sequencing was performed on tumour DNA from 105 germline variant carriers, along with bisulphite sequencing of promoter CpG islands for selected genes, to identify evidence of biallelic inactivation and mutational signatures that might support a causative role for that gene. Two genes previously implicated as HGSOC predisposition genes, PALB2 and ATM, displayed biallelic inactivation in nearly every germline variant carrier tumour, associated with characteristic mutational signatures defined principally by the presence or absence of HR repair deficiency, respectively. Of the candidate genes, 19 out of 38 demonstrated biallelic inactivation in at least one tumour from affected carriers, but only three- LLGL2, SCYL3 and MIPOL1- displayed this result consistently in multiple samples, with the others showing loss of the variant allele or returning inconclusive results. Distinctive mutational signatures were found in the LLGL2 and SCYL3 tumours, similar to those for ATM and PALB2, respectively. In the limited number of segregation studies performed amongst six families, none of the tested germline variants consistently segregated with disease. In conclusion, these studies provided data supporting PALB2 and ATM as likely moderate-risk HGSOC predisposition genes, demonstrating the utility of this approach for validating novel familial cancer genes. Several candidate genes showed evidence to indicate a potential predisposing role, but the extreme genetic heterogeneity of unexplained familial HGSOC will necessitate larger studies to confirm these findings.
-
ItemApplications of massively parallel sequencing technology in the evaluation of haematological malignancies( 2021)Massively parallel sequencing (MPS) technology has revolutionised the genomic exploration of human disease. This is especially true in the case of cancer, which is primarily driven by the development of acquired genomic aberrations. The body of work described within this thesis represents a broad yet in-depth array of novel applications of MPS technology in the evaluation of haematological malignancies. This field is currently surging in relevance and clinical utility as the ongoing movement of MPS technology from the research to the routine diagnostic setting continues to facilitate the development of increasingly personalised medicine. High impact contributions have been made in a number of areas encompassing myeloid and lymphoid malignancies as well as haematological malignancies as a collective. Key achievements include: quantifying the risk of incidentally detecting germline variants of potential clinical significance during unpaired MPS testing of cancer samples; definitively proving that ASXL1 NM_015338.5:c.1934dup;p.Gly646Trpfs*12 is a true somatic alteration and developing an accurate and sensitive assay for its detection; exploring the pathogenesis of and mechanisms of resistance to histone deacetylase inhibitors in cutaneous T-cell lymphomas as well as defining the clinical features, outcomes and genomic landscape of transformed marginal zone lymphoma. This thesis represents a diverse portfolio of novel research with a strong translational focus. Despite the wide scope of the individual lines of inquiry described herein there is a common thread that binds the narrative together: the pursuit of innovative yet practical ways of utilising the powerful technology now available to improve the genomic characterisation of haematological malignancies and ultimately the lives of the patients and families they affect.
-
ItemAutomated discovery of interacting genomic events that impact cancer survival by using data mining and machine learning techniques( 2020)Rapid advancement in genomic technologies has driven down the cost of sequencing significantly. This efficiency has enabled large-scale cancer genomic studies to be conducted, generating a vast amount of data across different levels of omics variables. However, the tasks to extract new knowledge and information from this enormous volume of data present unique challenges. These analyses often require the application of specialised techniques for data mining, integration and interpretation to provide valuable insights. With the rise of machine learning adoption in recent decades, many advanced computational algorithms based on artificial intelligence techniques have also been proposed to analyse these genomics data. Although some of these applications have led to clinically relevant conclusions, many others are still relying on incomplete prior knowledge, or limited to only a selected number of features. These limitations raise the general question about the broader applicability of machine learning in the field of cancer genomics. This research addresses this question by assessing the application of machine learning techniques in the context of breast cancer genomics data. This assessment includes a comprehensive evaluation of computational methods for predicting cancer driver genes and the development of a novel deep learning approach for identifying breast cancer subtypes. The evaluation result of driver gene prediction algorithms suggests that the selection of the best method to be applied to a dataset will primarily be driven by the objectives of the study and the characteristics of the dataset. All of the evaluated approaches could identify well- studied genes, but not all of them performed as well on smaller datasets, subtype-specific cohorts, and in discovering novel genes. To examine the benefit of a more complex machine learning model, this thesis also presents a novel deep learning approach that integrates multi-omics data for predicting various breast cancer’ biomarkers and molecular subtypes. This method combines a semi-supervised autoencoder for dimensionality reduction, and a supervised multitask learning setup for the classifications. Taking an input of gene expression, somatic point mutation and copy number data, the algorithm predicts the ER-Status, HER2-Status and molecular subtypes of breast cancer samples. Further survival analysis of the outputs from this deep learning approach indicates that the predicted subtypes show a stronger correlation with patient prognosis compared to the original PAM50 label. While the outputs from machine learning algorithms still require further validation, the adoption of these complex computational methods in cancer genomics will become increasingly common. Collectively, the results from this thesis suggest that the machine learning analysis of ‘omics data hold great potential in automating the discovery of clinically- relevant molecular features.
-
ItemUnderstanding the clinical implications of the evolution of breast cancers from primary to metastatic disease using next generation sequencing( 2020)Precision oncology refers to the use of sophisticated assays to tailor therapy to an individual patient. The feasibility of implementing such a program in a comprehensive cancer centre in Australia is unknown. The utility of precision oncology also depends on understanding how genomic profiles may evolve over time and differ between tumours in the same patient. A precision oncology program called SEGMENT was designed and implemented in a single centre. The program was popular, recruiting well over the study period. Timely acquisition of samples for sequencing was suboptimal from external pathology providers, and proved increasingly expensive during the study period. Delivery of results in a manner where they could be utilized by the patient was challenging in cases where patients were referred late in their natural history. A custom hybrid capture panel worked reliably. A total of 300 patients were recruited to the study, of which 288 had at least one sample received. Accounting for attrition, 214 patients or 71% went onto the main study. The spectrum of mutations and copy number alterations found in this study was similar to published cohorts. There were few differences between primary and metastatic lesions on average. Paired primary and metastatic samples however displayed discordance for both copy number and mutations. This was the case for actionable alterations in ESR1, ERBB2 and very rarely PIK3CA. Approximately 50% of patients had an actionable alteration. Of these 14% of the overall cohort received a therapy matched to their genomic profile. Five patients received matched therapy off trial and 26 received matched trial therapy. Three were zero responses in the off-trial group, and a response rate of 27% in the matched trial therapy group. To explore genomic heterogeneity in greater resolution, 4 patients with advanced breast cancer underwent rapid autopsies to collect large numbers of metastatic samples. Whole exome sequencing was performed on multiple lesions per patient which allowed inference of the subclonal structure. All patients displayed a monophyletic architecture, with truncal driver alterations giving rise to subclones with differing genomic profiles. One patient with a long disease free interval from primary to metastasis showed the acquisition of a new driver in the metastatic lesion. Driver alterations appeared to shape subsequent evolution.