The Princess Margaret Clinical Genomics Program: Hospital-based Cancer Genome Analysis
The Pugh lab is focused on the application of genome sequence analysis as a routine clinical test, particularly as modern cancer treatments are increasingly predicated on genetic information. I am particularly interested in genome analysis of serial biopsies and circulating DNA collected during clinical trials, tumors of unknown etiology, and rare pediatric cancers. I also spend part of my time supporting clinical molecular testing as a clinical molecular geneticist through the Advanced Molecular Diagnostics Laboratory.
Cancers arise due to changes in genetic sequence and structure that alter the biology of normal cells. Advances in DNA sequencing technologies have enabled the detection of these alterations in nearly a million human tumours, resulting in over 680,000 unique mutations and 16,000 rearrangements that are listed in the Catalogue of Somatic Mutations in Cancer (COSMIC) as of September 2013. Large-scale studies have uncovered differing mutation rates across cancer types, with the lowest rates found in pediatric (Pugh et al. 2012, Pugh et al. 2013) and hematologic malignancies (Wang et al., 2011) and the highest rates in environmentally associated cancers, such as lung cancer (smoking, Imielinski et al. 2012) and melanoma (sun exposure, Berger et al. 2012). Recurrent mutations have been found within and across cancer types, leading to new understandings of mechanisms that are disrupted in specific cancers and the identification of new biological subtypes.
With this knowledge in hand, therapies that target disrupted pathways have shown remarkable efficacy. Examples of this paradigm include inhibition of the BCR-ABL fusion protein by imatinib in leukemia (Druker et al. 2001), erlotinib and gefitinib efficacy against a spectrum of activating EGFR mutations in lung adenocarcinoma (Paez et al. 2004), and dramatic response to vemurafenib in melanomas with BRAF p.V600E mutations (Chapman et al. 2011). Conversely, treatment with targeted therapies without foreknowledge of underlying molecular alterations has been largely unsuccessful, including an early trial of gefitinib in unselected lung cancer patients (Thatcher et al. 2005). However, while genome “landscapes” have been described for many cancer types, few recurrently mutated genes have been linked to clinical outcome and the majority of tumours do not have mutations suggestive of a clear clinical course. Many of these studies drew upon ideal tissues from banks of tumours collected without consistent clinical indication or annotation, thereby challenging associations between mutation and clinical outcome. An additional challenge to isolated analysis of mutation or copy number variation has been the observation that different types of genome variation can target a cancer gene to similar effect (e.g. amplification, hypo-methylation, up-regulation, or base-pair mutation) and several members of an oncogenic pathway or protein complex may be individually altered at low frequency within a tumour type (Pugh et al 2012, Pugh et al 2013). Furthermore, there is evidence that common oncogenic mechanisms are appropriated across cancer types and that no mutation is truly confined to a single tissue of origin. This observation emphasizes a need to transition from tissue to molecular-based classification of tumours, and reveals opportunities for “off-label” use of targeted therapies across tumour types if strong genotype/phenotype associations are known.
To begin linking genomic genotypes to clinical phenotypes, our laboratory seeks to enable comprehensive genomic profiling of consistently ascertained and treated cancer specimens. Specifically, we are conducting high-resolution examinations of tumour DNA, RNA and epigenetic marks from primary tumour biopsies; examinations that are now feasible due to continued advancements in DNA sequencing technology. Termed “next-generation sequencing” (NGS), advanced DNA sequencing methods have enabled routine analysis of all genetic content (whole genome sequencing, WGS), all annotated genes (whole exome sequencing, WES), all expressed genes (RNA sequencing, RNA-Seq) and regulators of gene regulation (e.g. epigenetic marks, histone binding sites, and DNA/protein interactions) in tissues and, more recently, single cells. These data types are highly complementary and analysis of one large-scale data set greatly informs another. Therefore, we are developing laboratory and computational approaches to extract multiple sources of genome variation from suboptimal tumor specimens, and integrating these data types into cohesive portraits of individual tumor biology. Part of our work focuses on translation of these findings into clinical practice through nomination of clinically-informative markers for targeted testing and development of bioinformatics tools to support clinical laboratory workflows.