Pathology - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 2 of 2
  • Item
    Thumbnail Image
    The genomic landscape of phaeochromocytoma
    Flynn, Aidan ( 2015)
    Phaeochromocytomas (PCC) and paragangliomas (PGL) (collectively PPGL) are rare neural crest-derived tumours originating from adrenal chromaffin cells or extra-adrenal sympathetic and parasympathetic tissues. More than a third of PPGL cases are associated with heritable syndromes involving 18 or more known genes. These genes have been broadly partitioned into two groups based on pseudo-hypoxic and receptor tyrosine kinase (RTK) signalling pathways. Many of these genes can also become somatically mutated, although up to one third of sporadic cases have no known genetic driver. Furthermore, little is known of the genes that co-operate with known driver genes to initiate and drive tumourigenesis. To explore the genomic landscape of PPGL, exome sequencing, high-density SNP-array analysis, and RNA sequencing was applied to 36 PCCs and four PGL tumours. All tumours displayed a low mutation frequency in combination with frequent large segmental copy-number alterations and aneuploidy, with evidence for chromothripsis seen in a single case. Thirty-one of forty (77.5%) cases could be explained by germline or somatic mutations or structural alterations affecting known PPGL genes. Deleterious somatic mutations were also identified in known tumour-suppressor genes associated with genome maintenance and epigenetic modulation (e.g. TP53, STAG2, KMT2D). A multitude of other genes were also found mutated that are likely important for normal neuroendocrine cell function (e.g. ASCL1, NCAM1, GOLGA1). In addition, the existing paradigm for gene-expression subtyping of PPGL was further refined by applying consensus clustering to a compendium of previously published microarray data, enabling the identification of six robust gene-expression subtypes and subsequent cross-platform classification of RNA-seq data. The majority of cases in the cohort with no identifiable driver mutation were classified into a gene-expression subtype bearing similarity to MAX mutant PPGL, suggesting there are yet unknown PPGL cancer genes that can phenocopy MAX mutations. The cross-platform classification model was then further refined to develop a 46-gene Nanostring-based diagnostic tool capable of classifying PPGL tumours into gene-expression subtypes. The strong genotype-to-subtype relationship in PPGL makes subtyping a powerful tool that can be used clinically to guide and interpret genetic testing, determine surveillance programs and aid in better elucidation of PPGL biology. In applying the diagnostic assay to a test set of 38 cases, correct classification into one of the six subtypes was achieved for 34 (90%) samples based on the known genotype to gene-expression subtype association. The observation that at least one of the six subtypes is likely defined by the presence of non-neoplastic cells led to further refinement into five, four, and three-class architectures, further improving classification accuracy. Increasingly tumour heterogeneity is being recognised as one of the most significant challenges facing modern oncology. Genomically diverse tumour regions create additional complexity in predicting treatment response and metastatic potential through biopsy. Multi-region sampling of multiple synchronous primaries from patients with a predisposing germline mutation was used to explore tumour evolution and heterogeneity in PPGL and concomitant medullary thyroid carcinoma. Evolutionary reconstruction of a single primary PPGL demonstrated periods of both branched and linear evolution resulting in a high degree of intratumoural heterogeneity. Comparison of multiple synchronous primaries provided strong evidence of convergent evolution through recurrent chromosomal aberrations, indicating these may be obligate events in tumourigenesis, and as such, may indicate potential novel therapeutic targets.
  • Item
    Thumbnail Image
    Molecular profiling of ovarian cancer to guide targeted treatment
    Kondrashova, Olga ( 2015)
    Ovarian cancer is a complex disease composed of multiple distinct molecular and clinical subtypes. The survival rate for ovarian cancer has remained largely unchanged over the past three decades, despite the rapid advancement of the knowledge of the molecular and genetic mechanisms underlying most of the subtypes of ovarian cancer. There is, therefore, an urgent need to rapidly translate this knowledge into improved clinical outcomes for patients with ovarian cancer. There have been significant clinical responses of certain types of cancer to targeted therapies that are designed to inhibit specific molecular defects that some tumours appear to be dependent upon. To assist in allocating patients with ovarian cancer to targeted therapies, two customised assays for mutation and copy number alteration detection were developed for molecular profiling. A panel of 29 genes, which are commonly mutated in ovarian cancer, and are potentially therapeutically targeted, was selected to be screened using an amplicon-based assay, designed for next generation sequencing. Seventy six ovarian cancer cases with matched formalin-fixed paraffin- embedded tumour tissue, snap-frozen tumour tissue and blood samples were used for the assay validation and estimation of the diagnostic yield. A panel of 11 commonly copy number altered genes in ovarian cancer was also selected for screening with a herein developed method for multiplex low-level copy number detection. Furthermore, a thorough assessment and optimisation of the available and developed analysis methods was performed to ensure accurate analysis and reporting of mutations and copy number alterations. Thirty five patients with advanced ovarian cancer were tested using the developed assays as part of the ALLOCATE study, with genetic changes detected in 90.9%, demonstrating a high diagnostic yield. Molecular profiling of these cases was not only useful in identification of possible targeted treatment strategies with the aim of improving clinical outcomes, but also assisted in determining the correct diagnosis. Moreover, a novel algorithm was proposed for the prediction of individual tumour response to PARP inhibitors, a promising targeted treatment in high-grade serous ovarian cancer.