Swee Seong Wong

Cistrome: an integrative platform for transcriptional regulation studies

The increasing volume of ChIP-chip and ChIP-seq data being generated creates a challenge for standard, integrative and reproducible bioinformatics data analysis platforms. We developed a web-based application called Cistrome, based on the Galaxy open source framework. In addition to the standard Galaxy functions, Cistrome has 29 ChIP-chip- and ChIP-seq-specific tools in three major categories, from preliminary peak calling and correlation analyses to downstream genome feature association, gene expression analyses, and motif discovery. Cistrome is available at http://cistrome.org/ap/.

Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes

Gastric cancer, a leading cause of cancer-related deaths, is a heterogeneous disease. We aim to establish clinically relevant molecular subtypes that would encompass this heterogeneity and provide useful clinical information. We use gene expression data to describe four molecular subtypes linked to distinct patterns of molecular alterations, disease progression and prognosis. The mesenchymal-like type includes diffuse-subtype tumors with the worst prognosis, the tendency to occur at an earlier age and the highest recurrence frequency (63%) of the four subtypes. Microsatellite-unstable tumors are hyper-mutated intestinal-subtype tumors occurring in the antrum; these have the best overall prognosis and the lowest frequency of recurrence (22%) of the four subtypes. The tumor protein 53 (TP53)-active and TP53-inactive types include patients with intermediate prognosis and recurrence rates (with respect to the other two subtypes), with the TP53-active group showing better prognosis. We describe key molecular alterations in each of the four subtypes using targeted sequencing and genome-wide copy number microarrays. We validate these subtypes in independent cohorts in order to provide a consistent and unified framework for further clinical and preclinical translational research.

Genomic landscape and genetic heterogeneity in gastric adenocarcinoma revealed by whole-genome sequencing

Gastric cancer (GC) is the second most common cause of cancer-related deaths. It is known to be a heterogeneous disease with several molecular and histological subtypes. Here we perform whole-genome sequencing of 49 GCs with diffuse (N=31) and intestinal (N=18) histological subtypes and identify three mutational signatures, impacting TpT, CpG and TpCp[A/T] nucleotides. The diffuse-type GCs show significantly lower clonality and smaller numbers of somatic and structural variants compared with intestinal subtype. We further divide the diffuse subtype into one with infrequent genetic changes/low clonality and another with relatively higher clonality and mutations impacting TpT dinucleotide. Notably, we discover frequent and exclusive mutations in Ephrins and SLIT/ROBO signalling pathway genes. Overall, this study delivers new insights into the mutational heterogeneity underlying distinct histologic subtypes of GC that could have important implications for future research in the diagnosis and treatment of GC.

Progenitor-like Traits Contribute to Patient Survival and Prognosis in Oligodendroglial Tumors

Purpose: Patient-derived glioma-propagating cells (GPC) contain karyotypic and gene expression profiles that are found in the primary tumor. However, their clinical relevance is unclear. We ask whether GPCs contribute to disease progression and survival outcome in patients with glioma by analyzing gene expression profiles. Experimental Design: We tapped into public sources of GPC gene expression data and derived a gene signature distinguishing oligodendroglial from glioblastoma multiforme (GBM) GPCs. By adapting a method in glioma biology, the Connectivity Map, we interrogated its strength of association in public clinical databases. We validated the top-ranking signaling pathways Wnt, Notch, and TGFβ, in GPCs and primary tumor specimens. Results: We observed that patients with better prognosis correlated with oligodendroglial GPC features and lower tumor grade, and this was independent of the current clinical indicator, 1p/19q status. Patients with better prognosis had proneural tumors whereas the poorly surviving cohort had mesenchymal tumors. In addition, oligodendroglial GPCs were more sensitive to Wnt and Notch inhibition whereas GBM GPCs responded to TGFβR1 inhibition. Conclusions: We provide evidence that GPCs are clinically relevant. In addition, the more favorable prognosis of oligodendroglial tumors over GBM could be recapitulated transcriptomically at the GPC level, underscoring the relevance of this cellular model. Our gene signature detects molecular heterogeneity in oligodendroglial tumors that cannot be accounted for by the 1p/19q status alone, indicating that stem-like traits contribute to clinical status. Collectively, these data highlight the limitation of morphology-based histologic analyses in tumor classification, consequently impacting on treatment decisions. Clin Cancer Res; 18(15); 4122–35. ©2012 AACR.

Abstract 4104: Intrinsic and acquired resistance to cetuximab in colorectal cancer patients

Anti-EGFR antibodies, such as cetuximab, are effective therapies for many late-stage colorectal cancer (CRC) patients; unfortunately, many tumors are initially unresponsive while others show initial efficacy but eventually develop acquired resistance. Genomic studies of patient tumors, cell lines, and xenograft models have identified putative anti-EGFR resistance markers, including mutations in KRAS, NRAS, BRAF, PIK3CA, and the EGFR extracellular domain, as well as amplifications in ERBB2 and MET. In order to further confirm and identify new resistance mechanisms to anti-EGFR treatment in CRC, we performed retrospective genomic profiling of 25 CRC patients treated at Samsung Medical Center from 2006-2015. Patients received cetuximab containing chemo regimens with varying duration of responses, including acquired resistance cases. Our analysis identifies mutations in receptor tyrosine kinases, such as EGFR, NTRK1, and PDGFRA, as well as RAS/MAPK pathway genes that affect cetuximab response. We also uncover genomic alterations in ERBB2 and c-KIT as potential novel mechanisms regulating sensitivity to anti-EGFR antibodies. Additional genomic analyses of acquired resistance tumors and in vitro studies of a patient-derived cell line provide added insights into clonal selection and signaling pathways that bypass the EGFR blockade. Overall, our study elucidates important new facets in the landscape of anti-EGFR resistance mechanisms. Note: This abstract was not presented at the meeting. Citation Format: Steven M. Bray, Jeeyun Lee, Seung Tae Kim, Philip J. Ebert, John N. Calley, Isabella H. Wulur, Thejaswini Gopalappa, Swee Seong Wong, Hui-Rong Qian, Jason C. Ting, Jiangang Liu, Melinda D. Willard, Amit Aggarwal, Ruslan D. Novosiadly, Hee-Cheol Kim, Christoph Reinhard. Intrinsic and acquired resistance to cetuximab in colorectal cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4104. doi:10.1158/1538-7445.AM2017-4104

Abstract 29: Concordance of somatic mutations found in primary tumors and plasma circulating-free DNA from early and late-stage colorectal cancer patients

Circulating-free DNA (cfDNA) holds great potential for non-invasive somatic mutation detection in cancer patients as a so-called “liquid biopsy”. A liquid biopsy is especially valuable when tumor tissue is not available and for longitudinal monitoring of tumor burden or emerging resistance. In such cases, tumor-specific mutations are not known a priori and in order to detect tumor-derived somatic mutations it is best to interrogate a broad panel of cancer genes. The primary obstacle to sequencing cfDNA with broad mutation panels is achieving the necessary limit of detection to identify small amounts of tumor-derived cfDNA relative to the predominant wild-type cfDNA in the plasma of cancer patients. Tumor-derived cfDNA levels correlate with tumor stage, also making it an important consideration when investigating the sensitivity of this approach. We demonstrate the detection of somatic variants in several cancer genes in the plasma of early and late-stage colorectal cancer (CRC) patients by deep sequencing (> 8,000X) the cfDNA and matched normal DNA of N=33 CRC patients using the AmpliSeq cancer panel. The AmpliSeq panel requires less than 10ng of input DNA and amplifies hotspot loci of 50 known cancer genes for next-generation sequencing (NGS). We observe at least one somatic mutation with greater than 0.1% variant allele frequency in the cfDNA of all 33 CRC patients. We also show the sensitivity of detecting tumor-derived somatic mutations in cfDNA by exome sequencing of the primary tumor tissue. We find higher sensitivity for tumor mutations in cfDNA of late-stage patients compared to early-stage patients. Overall, we demonstrate the feasibility of using NGS on a small sized cancer panel for identifying somatic mutations, including those with low variant allele frequency, in cfDNA of early and late-stage CRC patients. Although a cfDNA approach holds promise as a tool to aid in pre-clinical and clinical research, more work is needed to understand its performance under different assay and disease conditions. Citation Format: Steven M. Bray, Philip J. Ebert, John N. Calley, Richard E. Higgs, Isabella H. Wulur, Swee Seong Wong, Candice L. Horn, Ricardo Martinez, Christoph Reinhard. Concordance of somatic mutations found in primary tumors and plasma circulating-free DNA from early and late-stage colorectal cancer patients. [abstract]. In: Proceedings of the AACR Precision Medicine Series: Integrating Clinical Genomics and Cancer Therapy; Jun 13-16, 2015; Salt Lake City, UT. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(1_Suppl):Abstract nr 29.

Abstract LB-313: Integrative genomic profiling of Asian gastric cancers identifies four subgroups with distinct pathobiology and prognosis

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA Gastric cancer is the 2nd most common cause of cancer related mortality and 4th most common cancer worldwide. The molecular classification of Gastric Cancers and the relevance of pre-clinical models are not well established, creating challenges in discovering novel molecularly targeted therapies. In order to address that, we conducted integrated molecular data analysis of 300 Asian Gastric tumors through the Asian Cancer Research Group (ACRG). We identified four cancer subtypes, based on RNA/DNA profiling, Laurens histological classification (Intestinal and Diffused) and Epstein Barr Virus (EBV) status, exhibiting differential pathobiology as well as prognosis. The groups are 1) Mesenchymal subgroup characterized by Diffused tumors with hallmarks of Epithelial to Mesenchymal transition such as CDH1 loss and co-occurring with IGF2 over-expression; 2) Microsatellite instable (MSI) subgroup characterized by predominantly hypermutated Intestinal tumors (including majority of mutations in KRAS) with likely MLH1 loss through promoter methylation; 3) TP53 pathway active subgroup with Epstein-Barr virus (EBV) infection or mutated oncogenes (e.g. PIK3CA) and 4) TP53 pathway inactive characterized by p53 loss through deleterious mutations in TP53 or MDM2 amplification and further characterized by both focal amplifications in oncogenes such as HER2, EGFR, cMET, CCNE1 as well as large scale chromosomal gains and losses. The above subtypes exhibited differential prognosis with the Mesenchymal subtype displaying the worst prognosis and the MSI subtype the best prognosis among the subtypes. The subtypes and their association with prognosis were independently validated in an additional large Gastric cancer cohort (N=277). We studied the applicability of this classification in other gastrointestinal (GI) cancers and show the presence of our proposed molecular subtypes of Gastric cancer in Colorectal cancers as well thereby suggesting commonalities in biological processes that give rise to Gastric and Colorectal tumors and providing a common ground to classify GI cancers. We also checked the presence of Gastric cancer subtypes in pre-clinical models of GI tract cancers and found that cell line panels often used for drug discovery shown an under-representation of p53 pathway active subtype, thus possibly creating challenges in translation to clinical studies. Overall, we provide a stratification that will lay a more solid groundwork for rationally targeting Gastric Cancer by helping focus on specific altered mechanisms and/or oncogenes as well as allowing for a more rational choice of pre-clinical models in drug discovery and development. Citation Format: Razvan Cristescu, Jeeyun Lee, Michael Nebozhyn, Amit Aggarwal, Jason Ting, Swee Seong Wong, Yong Yue, Christoph Reinhard, Kyoung Kim, Ingu Do, Hongyue Dai, Andrey Loboda. Integrative genomic profiling of Asian gastric cancers identifies four subgroups with distinct pathobiology and prognosis. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr LB-313. doi:10.1158/1538-7445.AM2014-LB-313

Abstract 817: Genomic evolution landscape of patient tumor derived xenograft of gastric cancer.

Proceedings: AACR 104th Annual Meeting 2013; Apr 6-10, 2013; Washington, DC Patient-derived xenograft (PDX) models have emerged as key model systems to understand the efficacy of anti-cancer agents. PDX systems have demonstrated superiority over cell lines with regards to higher histological resemblance to primary tumor, presence of stroma and also in mimicking response to therapeutic agents. However, the genomic evolution landscape of model establishment is not a well studied topic. In order to do that, we systematically collected materials from the first three passages of xenografts derived from eight gastric cancer patients from Seoul National University, Korea from 2008-2011. We subsequently generated comprehensive genomic profiles of the first three passages, parent primary tumor, and matched normal tissues for the eight patients. The genomic profiling data included whole-exome sequencing data (Agilent exome capture, paired-end sequencing on Illumina HiSeq 2000), mRNA expression (Affymetrix U133 Plus2.0), copy number data (Affymetrix SNP6), miRNA expression (Agilent miRNA array v16.0) and DNA methylation data (Illumina HumanMethylation27). The analysis is focused on two types of events: a) Cancer-related genomic changes derived from normal vs. tumor comparison, and their subsequent assessment in early passages; b) Passage-specific genomic changes derived from tumor vs. passages comparison. A further refinement of passage-specific genomic changes is done to differentiate ‘true’ passage-specific events from the ones that appear due to the difference in tumor purity that is lower in primary tumor due to the normal contamination. Overall, we describe the genomic landscape of evolution along xenograft establishment in gastric cancer and provide a comprehensive picture of genetic and genomic similarities (and differences) of xenografts to the primary tumor. This analysis will help us to better interpret the in vivo results emanating from experiments using xenografts and to translate the findings appropriately to the clinic. Citation Format: Kun Yu, Swee Seong Wong, Jason C. Ting, Thompson N. Doman, Yong Yue, Amit aggarwal, Gregory P. Donoho, Rebekka Krumbach, Heiner H. Feibig, Seong-Ho kong, Woo-ho Kim, Han-kwang Yang, Christoph Reinhard. Genomic evolution landscape of patient tumor derived xenograft of gastric 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 817. doi:10.1158/1538-7445.AM2013-817 Note: This abstract was not presented at the AACR Annual Meeting 2013 because the presenter was unable to attend.

Abstract 3981: Evaluation of mutation detection sensitivity in exome sequencing and other genotyping platforms

An evaluation was performed to compare mutation calls from exome capture sequencing data to publicly available data sets derived from Sequenom MassARRAY and Sanger sequencing methods. We compared hot-spots KRAS and BRAF mutations using Sanger sequencing method in patient-derived tumor grafts. Mutation data of the tumor grafts were also generated with Sequenom MassARRAY using the OncoCarta panel. We also evaluated the sensitivity of mutation detection for a panel of cancer genes in >100 cancer cell lines against the data from the Cancer Cell Line Encyclopedia Project (OncoMap mutation panel) as well as the Sanger9s Cancer Genome Project. Overall, we observed high sensitivity and concordance of the mutation calls from the exome sequencing with the other platforms and technologies. The above comparisons support the use of exome sequencing as a sensitive mutation detection method for patient-derived tumor grafts and cancer cell lines. 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 3981. doi:1538-7445.AM2012-3981

Abstract 5108: A combined processing and assessment workflow for SNV and indel calling on exome sequencing of tumor samples

A workflow for exome sequencing is established using multiple callers for SNV and indel. A spectrum of measures is looked at to understand the diversity and characteristics of the samples. Two common workflows for variant calling are used, namely GATK unified genotyper and Samtools mpileup. Illumina paired end 90 base reads are mapped to the human genome using BWA. Variants called by GATK and Samtools are compared and merged to give a unified set of calling in VCF format. This approach gives additional confidence in calls supported by both callers as there are technical differences in the two workflows. dbSNP rate and TI/TV ratio are used to assess the general quality (sensitivity and specificity) of the calls. It is observed that tumor samples with very low and high TI/TV novel ratio are largely lung and melanoma samples respectively. They are likely to be UV and smoking induced DNA mutation. This affects the overall TI/TV ratio. A handful of samples also exhibit high variant count that is uniquely found (private) in the sample alone. This correlates with lower than expected dbSNP rate. 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 5108. doi:1538-7445.AM2012-5108