Michelle Sam

Spatial genomic heterogeneity within localized, multifocal prostate cancer

Herein we provide a detailed molecular analysis of the spatial heterogeneity of clinically localized, multifocal prostate cancer to delineate new oncogenes or tumor suppressors. We initially determined the copy number aberration (CNA) profiles of 74 patients with index tumors of Gleason score 7. Of these, 5 patients were subjected to whole-genome sequencing using DNA quantities achievable in diagnostic biopsies, with detailed spatial sampling of 23 distinct tumor regions to assess intraprostatic heterogeneity in focal genomics. Multifocal tumors are highly heterogeneous for single-nucleotide variants (SNVs), CNAs and genomic rearrangements. We identified and validated a new recurrent amplification of MYCL, which is associated with TP53 deletion and unique profiles of DNA damage and transcriptional dysregulation. Moreover, we demonstrate divergent tumor evolution in multifocal cancer and, in some cases, tumors of independent clonal origin. These data represent the first systematic relation of intraprostatic genomic heterogeneity to predicted clinical outcome and inform the development of novel biomarkers that reflect individual prognosis.

Elevated Coding Mutation Rate During the Reprogramming of Human Somatic Cells into Induced Pluripotent Stem Cells

Mutations in human induced pluripotent stem cells (iPSCs) pose a risk for their clinical use due to preferential reprogramming of mutated founder cell and selection of mutations during maintenance of iPSCs in cell culture. It is unknown, however, if mutations in iPSCs are due to stress associated with oncogene expression during reprogramming. We performed whole exome sequencing of human foreskin fibroblasts and their derived iPSCs at two different passages. We found that in vitro passaging contributed 7% to the iPSC coding point mutation load, and ultradeep amplicon sequencing revealed that 19% of the mutations preexist as rare mutations in the parental fibroblasts suggesting that the remaining 74% of the mutations were acquired during cellular reprogramming. Simulation suggests that the mutation intensity during reprogramming is ninefold higher than the background mutation rate in culture. Thus the factor induced reprogramming stress contributes to a significant proportion of the mutation load of iPSCs. STEM CELLS 2012;30:435–440

Genomic hallmarks of localized, non-indolent prostate cancer

Prostate tumours are highly variable in their response to therapies, but clinically available prognostic factors can explain only a fraction of this heterogeneity. Here we analysed 200 whole-genome sequences and 277 additional whole-exome sequences from localized, non-indolent prostate tumours with similar clinical risk profiles, and carried out RNA and methylation analyses in a subset. These tumours had a paucity of clinically actionable single nucleotide variants, unlike those seen in metastatic disease. Rather, a significant proportion of tumours harboured recurrent non-coding aberrations, large-scale genomic rearrangements, and alterations in which an inversion repressed transcription within its boundaries. Local hypermutation events were frequent, and correlated with specific genomic profiles. Numerous molecular aberrations were prognostic for disease recurrence, including several DNA methylation events, and a signature comprised of these aberrations outperformed well-described prognostic biomarkers. We suggest that intensified treatment of genomically aggressive localized prostate cancer may improve cure rates.

Abstract B129: Clinical implications of inter- and intra- prostatic heterogeneity.

Men with localized prostate cancer vary widely in clinical outcome, with a 30-50% failure rate after primary treatment. There is thus significant interest in developing genomically refined prognostic groups. We sought to evaluate the extent of genetic heterogeneity, both between patients (inter-prostate) and within individual prostate glands (intra-prostate) to assess the impact of genetic heterogeneity on risk stratification within a tight clinical cohort. Copy number aberrations (CNAs) from 75 Gleason 7 patients were determined by OncoScan SNP microarrays. We measure the percentage of genome involved in a CNA, termed percent genome aberration (PGA), a proxy for genomic instability. Additionally, whole genome sequencing was applied to 10 intermediate-risk prostate tumours and matched blood, including multiple manually macro-dissected regions from 5 of the prostates (range 2 to 9). Somatic single nucleotide variants (SNVs) and genomic rearrangements (GR) were extracted from each patient. We find a high degree of inter-prostatic heterogeneity between the 75 Gleason 7 patients, with the number of CNAs per patient ranging from 0 to 929, corresponding to PGA 0 to 16.7%. Known prognostic markers can differentiate between patients at higher risk for biochemical recurrence, but only account for a fraction of the cohort. Notably, when these prognostic genes are examined within multiple regions of five independent tumours, they differ in copy number between cancerous regions of the same prostate. For example, TP53 is deleted in 1/2, 1/3, 4/9, 0/4, and 4/5 prostate regions. Indeed, phylogenetic analysis of geographically distinct regions revealed multi-clonal disease in two of the five patients; separate analyses based on SNVs, CNAs, and GRs all concluded that these patient have two genetically distinct cancers within their prostate. We demonstrate dramatic levels of inter- and intra- prostate genetic heterogeneity within pathologically identical or similar cancers. The observed intra-prostatic genomic heterogeneity, both in terms of multi-focal and multi-clonal disease, has critical implications for clinical management. Prognostic information obtained by biopsy may be inconsistent depending on the site of biopsy, and applying personalized medicine to prostate cancer will be challenging. This study highlights the need for further evaluation of how intra-prostatic heterogeneity is related to patient prognosis. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B129. Citation Format: Emilie Lalonde, Paul C. Boutros, Michael Fraser, Richard de Borja, Nicholas J. Harding, Dominique Trudel, Alice Meng, Pablo H. Hennings-Yeomans, Andrew McPherson, Amin Zia, Jianxin Wang, Timothy Beck, Natalie S. Fox, Taryne Chong, Michelle Sam, Jeremy Johns, Lee Timms, Nicholas Buchner, Sohrab Shah, Cenk Sahinalp, Thomas J. Hudson, John D. McPherson, Theodorus van der Kwast, Robert G. Bristow. Clinical implications of inter- and intra- prostatic heterogeneity. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B129.

Abstract 2003: A molecular portrait of potentially curable prostate cancer.

Intermediate risk prostate cancer (CaP) with Gleason score (GS) of 7 show up to 100x variability in genetic instability. As CaP is multifocal and likely multiclonal, there is a need to characterize heterogeneity for patient stratification, which would increase the ability to act on genomic information by adding adjuvant therapies to offset systemic occult metastases that currently limit cure in ∼30% of patients. Individual genetic portraits could be used to improve cure on combined clinical-molecular staging criteria. We undertook a pilot study to assess the genetic heterogeneity of potentially curable GS=7 CaP. We selected 10 men with GS=7 CaP; 5 treated with external beam radiotherapy (frozen pre-treatment biopsies) and 5 treated with radical prostatectomy (RadP, frozen tumour). Additionally, DNA from 18 distinct formalin-fixed, paraffin-embedded (FFPE) foci from the 5 RadP were analysed. Each of these 28 foci were subjected to whole-genome sequencing (WGS) and OncoScan SNP arrays to yield comprehensive genetic profiles. mRNA expression was evaluated on frozen RadP by microarray. Germline DNA from whole-blood was also analysed. Following independent pathology reviews and manual macro-dissection of tumour areas of ≥70% cellularity, WGS (≥50x tumour, ≥30x germline) was performed on as little as 50 ng genomic DNA, and OncoScan arrays were performed using as little as 30ng DNA using either amplified or innate genomic DNA. Regions of CaP in FFPE RadP were recorded using a tissue map to identify independent malignant foci, and ERG immunostaining was performed to assist in the identification. In cases where ERG-positive and -negative foci were adjacent, ERG staining was repeated on an un-stained slide to confirm separate foci based on 3D multi-section analyses. ERG fusion status was also assessed in frozen samples by aCGH or IHC. Validation of SNVs via SNP array and deep-resequencing showed ∼99% accuracy. Tumour cellularity was estimated using Qpure and was >60% for all samples. Phylogenetic techniques were used to demonstrate clear multi-clonality in two tumours. Across all tumours, ∼50% of SNVs were specific to an individual tumour-region. Phylogenies were confirmed with both SNVs and CNAs, but CNAs generally exhibited greater concordance amongst different regions of the same tumour. Some previously observed recurrent mutations were previously identified as recurrent in CaP (e.g. SPOP), and the overall mutation rate for intermediate-risk CaP was only somewhat below that reported for castrate-resistant disease (11,230 somatic SNVs per tumour). Our studies support the concept that a complete characterization of inter- and intra-CaP heterogeneity is possible in fresh and archival tissues; the latter is important for correlations to clinical outcome. These approaches can then be streamlined for high-throughput analyses within personalized medicine laboratories leading to “point of care” molecular tests and individualization of therapy. Citation Format: Michael E. Fraser, Richard de Borja, Dominique Trudel, Nicholas J. Harding, Pablo H. Hennings-Yeomans, Alice Meng, Emilie R. Lalonde, Andrew Brown, Natalie S. Fox, Taryne Chong, Amin Zia, Michelle Sam, Jianxin Wang, Michelle A. Chan-Seng-Yue, Jeremy Johns, Lee Timms, Nicholas Buchner, Ada Wong, Fouad Yousif, Rob Denroche, Gaetano Zafarana, Maud HW Starmans, Hanbert Chen, Shaylan Govind, Francis Nguyen, Melania Pintilie, Neil Fleshner, Stanislav Volik, Lakshmi Muthuswamy, Colin C. Collins, Thomas J. Hudson, Lincoln D. Stein, Timothy Beck, John D. McPherson, Theodorus van der Kwast, Paul C. Boutros, Rob G. Bristow. A molecular portrait of potentially curable prostate cancer. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2003. doi:10.1158/1538-7445.AM2013-2003

Abstract 3184: Whole genome sequencing of low-input fresh frozen prostate cancer biopsies

Proceedings: AACR 103rd Annual Meeting 2012‐‐ Mar 31‐Apr 4, 2012; Chicago, IL Prostate cancer is the most commonly diagnosed malignancy among men in the United States. Due to an aging population, prostate cancer incidence has been increasing, with an estimated 200,000 men being diagnosed in 2010 and more than 32,000 deaths resulting from this disease. Better predictors of patient prognosis and treatment outcome are required to individualize prostate cancer treatment. High-throughput genomic sequence-based approaches offer a unique opportunity to identify biomarkers of disease-progression, thereby enabling more individualized therapy. The Canadian Prostate Cancer Genome Network (CPC-GENE) is an outcomes-based initiative that will sequence 500 specimens from 350 prostate cancer patients over a 5-year time span. Previously, whole genome sequencing efforts from biopsy specimens have been hindered by insufficient quantities of extracted DNA required as input for sequencing library construction. As a proof of concept to demonstrate the ability to sequence low input amounts of DNA from prostate biopsies, whole genome sequencing has been initiated for 50 prostate tumor biopsy samples along with their matched blood-derived reference sample. An on-bead sample preparation protocol was optimized using decreasing quantities of input DNA and used to construct sequencing libraries from as low as 100ng of DNA derived from macrodissected fresh frozen prostate biopsies (>70% cellularity). Sequencing is performed on the Illumina HiSeq 2000 platform to generate coverage depths of 50x for tumor samples and 30x for reference samples. Following alignment using NovoAlign and variant-calling using GATK, we compared our results to genotyping-array results generated using the Affymetrix OncoScan platform. Single-nucleotide variants detected using arrays were validated >99% of the time by sequence data, confirming that the use of a low-input library did not hinder mutation detection. Sequencing does not exhibit significant genome-wide coverage biases, and CNV calls were compared between the genotyping arrays and the next-generation sequencing data. Outcomes from the sequencing and analysis of the initial 50 sample sets will similarly be applied over a 5-year period to characterize an additional 450 prostate specimens. The ability to whole genome sequence specimens where minimal amounts of extracted DNA exist presents new opportunities to sequence many samples previously deemed unusable, while also providing encouraging prospects for whole genome sequencing applications for future studies using biopsy specimens. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3184. doi:1538-7445.AM2012-3184

Abstract B13: Whole-genome mutation landscape in pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDAC) is a rare cancer with a very high mortality rate. Because it is extremely difficult to detect at an early stage; PDAC tumors often spread to regional lymph nodes or distant metastases by the time they are diagnosed. Published reports have already identified a number of chromosomal alterations at many genomic levels; however PDAC still lacks a comprehensive catalogue for the whole genome mutation spectrum. The goal of our study is to annotate all types of identifiable genomic aberrations based on whole genome sequencing of 5 PDAC tumors. It is a common knowledge that surgical primary tissues of PDAC have very low tumor content. Hence, for our study all five primary tumors have been modeled as xenografts using NOD-SCID mice to enrich for tumor cells. Here, we report on the cancer-specific genome alterations in 5 PDAC tumors and show that xenograft models do represent genomic landscape of primary tumors. All samples were whole-genome sequenced using Illumina HiSeq to give a minimum coverage of 30X. We have developed an analysis pipeline to identify somatic single nucleotide variations (SNVs) using The Genome Analysis Tool Kit (GATK), copy number alterations (CNAs) using KSseg (in-house CNV algorithm) and structural variations using Geometric Analysis of Structural Variants (GASV). A number of filters have been implemented to separate germline variants and mouse derived contamination from the cancer specific somatic variation. Our analysis has identified an average of 1527 SNVs, 1555 INDELs, and 53 CNAs per PDAC genome (combined for primary and xenograft). All somatic SNVs were verified using Ion Torrent based sequencing technology with a verification rate of 93%. CNAs were verified using Nimblegen 2.1M Array-based Comparative Genome Hybridization technology, and produced a verification rate of greater than 98% for losses and 60-97% for gains. We also observed a high level of overlap between primary tumor and xenograft samples, with 84% of total primary tumor SNVs and 61% of INDELS (called across all samples) being found in the correlating xenograft genome. After verification of SNVs by deep sequencing, we observe an additional 50% of SNVs that were called only in the xenograft samples validate in the primary sample. Our results show that the somatic single nucleotide mutation rate is in the range of 1 - 4 SNVs/Mb and there is a statistically significant increase in the G>T transversions. It is well known that methylated CpG dinucleotides are the preferred sites for G > T transversions and we are investigating the role played by DNA methylation alterations. All somatic variants were annotated using an in-house software package based on Sequence Ontology classification of variant effects to integrate different types of variations and provide a functional interpretation. Our analysis has identified 290 genes that are functionally impacted in 4 or more genomes by any type of mutation. They include 6 known oncogenes, 10 protein kinases, 9 cell differentiation markers, 17 transcription factors and 6 cytokines and growth factors. Functional enrichment analysis on this gene set using MSigDB v3.0 database shows important cancer-related pathways including the NK cells pathway, the Adherens junctions interactions pathway, and the axon guidance signaling pathway. Citation Format: Carson Holt, Fouad Yousif, Lee Timms, Michelle Sam, Kimberly Begley, Thomas Hudson, John D. McPherson, Lincoln D. Stein, Lakshmi B. Muthuswamy, Christina Yung, Tim Beck, Bojan Losic, Niloofar Arshadi, Christine Ouelltt, Irinia Kalatskaya, Richard de Borja, Robert Denroche. Whole-genome mutation landscape in pancreatic ductal adenocarcinoma. [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Progress and Challenges; Jun 18-21, 2012; Lake Tahoe, NV. Philadelphia (PA): AACR; Cancer Res 2012;72(12 Suppl):Abstract nr B13.

Abstract A9: The pancreatic ductal adenocarcinoma project at the Ontario Institute for Cancer Research.

Pancreatic cancer is the fifth leading cause of cancer-related death with a poor prognosis and 5-year survival rates of less than 5%. As a contributing member of the International Cancer Genome Consortium (ICGC), the Ontario Institute for Cancer Research (OICR) has committed to generating a comprehensive catalogue of genomic abnormalities found in pancreatic ductal adenocarcinomas (PDAC). Using next-generation sequencing technologies, 375 independent pancreatic tumors and their matched controls will be characterized over a 3- to 5-year time span. Many of the samples also have derived matching xenografts and a few have cell lines derived from the xenograft tumors. To date, this project has collected 173 ICGC consented matched samples comprised of 85 PDAC tumor/reference and 21 xenograft/ reference pairs. 32 sample sets have matched tumor/xenograft/reference. In addition, there are 16 cell lines derived from some of these xenografts. Whole-genome and exome target-enrichment sequencing is currently performed using the HiSeq 2000 platform. For efficient variant calling, all whole-genome analyses used a minimal depth of 30x reference and 50x tumor/xeno/cell line and exome analyses used greater than 100x target coverage, with most samples exceeding these targets. Sequence alignment is primarily performed with Novoalign (Novocraft.com) and variants called with the Genome Analysis Toolkit (McKenna et al., Genome Res. 20:1297) to identify germline and somatic variants in all matched sample sets. On average 35 coding non-synonymous variants have been observed per primary tumor exome. Validation of the coding non-synonymous variants is ongoing with data deposited in the ICGC Data Coordinating Center (www.icgc. org). Initial data have been combined with PDAC data sets from the Australian ICGC effort (S. Grimmond and A. Biankin) and from sequencing efforts at the Baylor College of Medicine (R. Gibbs) to increase analytical power. Sequencing of primary tumors and xenografts was complicated by human and mouse stroma, respectively. Complete whole-genome sequencing of the host mouse strain was needed for removal of “interspecies SNP” observed due to alignment of both species to regions of similarity in the human reference sequence. The “SNPs” are falsely identified as somatic variants after subtracting the human germline SNP. Methods are under development to improve this removal process. Current sequencing is focusing on expanding the data set of matched normal and primary tumor pairs. It is anticipated that 150 exome pairs will be sequenced by 4th quarter 2012 with whole-genome sequencing following closely. Continued development of xenograft resources is ongoing in parallel. Whole genome sequences will be generated from selected xenografts. Collectively, the sequence data, xenografts, and cell line models will make a rich resource for studying PDAC. Citation Format: Kimberly N. Begley, Debabrata Mukhopadhyay, Gloria M. Petersen, Alexei Protopopov, Sarah Thayer, Lynda Chin, Emin Ibrahimov, Patricia Shaw, Thomas Hudson, Steve Gallinger, Ming-Sound Tsao, Lincoln Stein, John D. McPherson, Lakshmi Muthuswamy, Timothy Beck, Christina Yung, Michelle Sam, Lee Timms, Carson Holt. The pancreatic ductal adenocarcinoma project at the Ontario Institute for Cancer Research. [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Progress and Challenges; Jun 18-21, 2012; Lake Tahoe, NV. Philadelphia (PA): AACR; Cancer Res 2012;72(12 Suppl):Abstract nr A9.

Abstract B18: Genomic analysis of pancreatic ductal adenocarcinoma.

Pancreatic cancer is the fifth leading cause of cancer deaths. Five-year survival rate is In our initial screen, whole-exome sequencing of 33 primary PDAC tumors and matched controls has been performed on the Illumina HiSeq 2000. Sequence alignment and variant calling have been performed using Novoalign and GATK, respectively. After manual review and validation on the Ion Torrent platform, we have identified 648 somatic mutations, 471 of which are non-silent mutations that impact 444 genes. Our results confirm several known mutations in PDAC such as KRAS, p53 and SMAD4. However, their mutation frequencies are lower than expected due to tumor cellularity. We have also screened for copy number alterations (CNAs) using Illumina Omni1-Quad BeadChip. Analysis was performed using Genome Studio, KSseg and PennCNV. In the 33 primary tumors, a median of 90 regions with copy number gain, copy number loss, or copy-neutral LOH have been detected per sample. Median genomic lengths are 19Mb and 20Mb in regions with copy number gain and loss, respectively. Annotation of the altered regions has identified 9152 protein-coding genes, miRNA and non-coding RNA that are altered in 4 or more tumors. To identify the pathways that contribute to PDAC, we have analyzed the genes with somatic mutations or CNAs by means of a functional interaction (FI) network. The FI network consists of curated pathways from Reactome and other databases and a high confidence set of functional interactions predicted by machine learning techniques. A PDAC-specific subnetwork is constructed by projecting the altered genes onto the FI network, and subsequently analyzed by a community clustering algorithm to identify network modules. These modules have been identified as KRAS, p53, TGFβ, Hedgehog, Integrin, Cadherin, Wnt, Rho GTPase and G-protein signaling pathways. While our effort in identifying driver mutations is ongoing, our initial screen has identified candidate genes that will be targeted for deep sequencing in all primary tumors. We will continue to perform whole-exome sequencing of other primary tumors along with xenografts derived from some of the primaries and cell lines derived from some of the xenografts. In addition, whole-genome sequencing of selected specimens is being performed to complement the exome data. The wealth of data will help to characterize the genomic abnormalities in PDAC. Citation Format: Christina K. Yung, Christine Ouellete, Lee Timms, Michelle Sam, Kimberly Begley, Thomas J. Hudson, John D. McPherson, Lincoln D. Stein, Timothy Beck, Lakshmi Muthuswamy, Richard De Borja, Carson Holt, Rob Denroche, Fouad Yousif, Zheng Zha, Niloofar Arshadi. Genomic analysis of pancreatic ductal adenocarcinoma. [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Progress and Challenges; Jun 18-21, 2012; Lake Tahoe, NV. Philadelphia (PA): AACR; Cancer Res 2012;72(12 Suppl):Abstract nr B18.