Yoke-Eng Chiew

Patterns of somatic mutation in human cancer genomes

Cancers arise owing to mutations in a subset of genes that confer growth advantage. The availability of the human genome sequence led us to propose that systematic resequencing of cancer genomes for mutations would lead to the discovery of many additional cancer genes. Here we report more than 1,000 somatic mutations found in 274 megabases (Mb) of DNA corresponding to the coding exons of 518 protein kinase genes in 210 diverse human cancers. There was substantial variation in the number and pattern of mutations in individual cancers reflecting different exposures, DNA repair defects and cellular origins. Most somatic mutations are likely to be ‘passengers’ that do not contribute to oncogenesis. However, there was evidence for ‘driver’ mutations contributing to the development of the cancers studied in approximately 120 genes. Systematic sequencing of cancer genomes therefore reveals the evolutionary diversity of cancers and implicates a larger repertoire of cancer genes than previously anticipated.

Novel Molecular Subtypes of Serous and Endometrioid Ovarian Cancer Linked to Clinical Outcome

Purpose: The study aim to identify novel molecular subtypes of ovarian cancer by gene expression profiling with linkage to clinical and pathologic features. Experimental Design: Microarray gene expression profiling was done on 285 serous and endometrioid tumors of the ovary, peritoneum, and fallopian tube. K-means clustering was applied to identify robust molecular subtypes. Statistical analysis identified differentially expressed genes, pathways, and gene ontologies. Laser capture microdissection, pathology review, and immunohistochemistry validated the array-based findings. Patient survival within k-means groups was evaluated using Cox proportional hazards models. Class prediction validated k-means groups in an independent dataset. A semisupervised survival analysis of the array data was used to compare against unsupervised clustering results. Results: Optimal clustering of array data identified six molecular subtypes. Two subtypes represented predominantly serous low malignant potential and low-grade endometrioid subtypes, respectively. The remaining four subtypes represented higher grade and advanced stage cancers of serous and endometrioid morphology. A novel subtype of high-grade serous cancers reflected a mesenchymal cell type, characterized by overexpression of N-cadherin and P-cadherin and low expression of differentiation markers, including CA125 and MUC1. A poor prognosis subtype was defined by a reactive stroma gene expression signature, correlating with extensive desmoplasia in such samples. A similar poor prognosis signature could be found using a semisupervised analysis. Each subtype displayed distinct levels and patterns of immune cell infiltration. Class prediction identified similar subtypes in an independent ovarian dataset with similar prognostic trends. Conclusion: Gene expression profiling identified molecular subtypes of ovarian cancer of biological and clinical importance.

Whole–genome characterization of chemoresistant ovarian cancer

Patients with high-grade serous ovarian cancer (HGSC) have experienced little improvement in overall survival, and standard treatment has not advanced beyond platinum-based combination chemotherapy, during the past 30 years. To understand the drivers of clinical phenotypes better, here we use whole-genome sequencing of tumour and germline DNA samples from 92 patients with primary refractory, resistant, sensitive and matched acquired resistant disease. We show that gene breakage commonly inactivates the tumour suppressors RB1, NF1, RAD51B and PTEN in HGSC, and contributes to acquired chemotherapy resistance. CCNE1 amplification was common in primary resistant and refractory disease. We observed several molecular events associated with acquired resistance, including multiple independent reversions of germline BRCA1 or BRCA2 mutations in individual patients, loss of BRCA1 promoter methylation, an alteration in molecular subtype, and recurrent promoter fusion associated with overexpression of the drug efflux pump MDR1.

Integrated Genome-Wide DNA Copy Number and Expression Analysis Identifies Distinct Mechanisms of Primary Chemoresistance in Ovarian Carcinomas

Purpose: A significant number of women with serous ovarian cancer are intrinsically refractory to platinum-based treatment. We analyzed somatic DNA copy number variation and gene expression data to identify key mechanisms associated with primary resistance in advanced-stage serous cancers. Experimental Design: Genome-wide copy number variation was measured in 118 ovarian tumors using high-resolution oligonucleotide microarrays. A well-defined subset of 85 advanced-stage serous tumors was then used to relate copy number variation to primary resistance to treatment. The discovery-based approach was complemented by quantitative-PCR copy number analysis of 12 candidate genes as independent validation of previously reported associations with clinical outcome. Likely copy number variation targets and tumor molecular subtypes were further characterized by gene expression profiling. Results: Amplification of 19q12, containing cyclin E (CCNE1), and 20q11.22-q13.12, mapping immediately adjacent to the steroid receptor coactivator NCOA3, was significantly associated with poor response to primary treatment. Other genes previously associated with copy number variation and clinical outcome in ovarian cancer were not associated with primary treatment resistance. Chemoresistant tumors with high CCNE1 copy number and protein expression were associated with increased cellular proliferation but so too was a subset of treatment-responsive patients, suggesting a cell-cycle independent role for CCNE1 in modulating chemoresponse. Patients with a poor clinical outcome without CCNE1 amplification overexpressed genes involved in extracellular matrix deposition. Conclusions: We have identified two distinct mechanisms of primary treatment failure in serous ovarian cancer, involving CCNE1 amplification and enhanced extracellular matrix deposition. CCNE1 copy number is validated as a dominant marker of patient outcome in ovarian cancer.

Expression of c-erbB receptors, heregulin and oestrogen receptor in human breast cell lines.

Members of the c‐erbB family have been implicated in poor prognosis in breast cancer. Given the propensity for heterodimerisation within the erbB family, the pattern of co‐expression of these receptors is likely to be as functionally important as aberrant expression of any given receptor alone. Therefore, the patterns of expression of the receptors, epidermal growth factor receptor (EGF‐R), c‐erbB‐2, c‐erbB‐3, c‐erbB‐4, and one of the erbB ligands, heregulin (HRG), were examined in normal and malignant breast cell lines and compared with expression of oestrogen receptor (ER), a classical indicator of good prognosis. There was an inverse correlation between ER and EGF‐R mRNA levels, as previously described, but no correlation between either of these receptors and c‐erbB‐2. c‐erbB‐3 expression was positively correlated with ER. In contrast, HRG expression was inversely related to ER. Expression of antisense‐ER resulted in increased EGF‐R mRNA, demonstrating a functional link between the expression of these 2 genes, however, there was no significant change in c‐erbB‐2 or c‐erbB‐3 mRNA, suggesting that ER is not directly involved in control of expression of these genes. A comparison of individual erbB receptors and HRG revealed that the majority of lines expressing increased levels of c‐erbB‐2 also expressed elevated levels of c‐erbB‐3 mRNA, and none of the cell lines that expressed both c‐erbB‐2 and either c‐erbB‐3 or c‐erbB‐4 expressed the ligand HRG. In summary, the levels of expression of c‐erbB‐1, ‐2, ‐3, and ‐4 varied in this series of breast cell lines, and the pattern of expression and the relationship of each growth factor receptor to the expression of ER was quite distinct. The lack of expression of HRG in cell lines that express receptors may be indicative of paracrine interactions between erbB ligands and their cognate receptors and may suggest that the ligand and receptors are expressed in different subtypes of breast epithelial cells from which the cell lines are derived. Int. J. Cancer 87:487–498, 2000.

Tumor protein D52 (TPD52) is overexpressed and a gene amplification target in ovarian cancer.

Recurrent chromosome 8q gain in ovarian carcinoma is likely to reflect the existence of multiple target loci, as the separate gain of chromosome bands 8q21 and 8q24 has been reported in independent studies. Since tumor protein D52 (TPD52) has been identified as a chromosome 8q21 amplification target in breast and prostate carcinoma, we compared TPD52 expression in normal ovarian epithelium (n = 9), benign serous adenomas (n = 11), serous borderline tumors (n = 6) and invasive carcinomas of the major histologic subtypes (n = 57) using immunohistochemistry. These analyses revealed that all normal ovarian epithelium samples and benign serous tumors were predominantly TPD52‐negative, whereas TPD52 was overexpressed in most (44/57; 77%) ovarian carcinomas regardless of histologic subtype. TPD52 subcellular localization was predominantly cytoplasmic, although nuclear localization was also frequently observed in mucinous and clear cell carcinomas. In an independent cohort of stage III serous carcinomas (n = 18), we also directly compared in situ TPD52 expression using immunohistochemistry and TPD52 copy number using interphase FISH analyses. This revealed that TPD52 dosage and TPD52 expression were significantly positively correlated. TPD52 therefore represents a novel molecular marker in ovarian cancer, which is broadly expressed across the different histologic subtypes and whose upregulation frequently reflects increased TPD52 copy number.

Molecular profiling of low grade serous ovarian tumours identifies novel candidate driver genes

Low grade serous ovarian tumours are a rare and under-characterised histological subtype of epithelial ovarian tumours, with little known of the molecular drivers and facilitators of tumorigenesis beyond classic oncogenic RAS/RAF mutations. With a move towards targeted therapies due to the chemoresistant nature of this subtype, it is pertinent to more fully characterise the genetic events driving this tumour type, some of which may influence response to therapy and/or development of drug resistance. We performed genome-wide high-resolution genomic copy number analysis (Affymetrix SNP6.0) and mutation hotspot screening (KRAS, BRAF, NRAS, HRAS, ERBB2 and TP53) to compare a large cohort of ovarian serous borderline tumours (SBTs, n = 57) with low grade serous carcinomas (LGSCs, n = 19). Whole exome sequencing was performed for 13 SBTs, nine LGSCs and one mixed low/high grade carcinoma. Copy number aberrations were detected in 61% (35/57) of SBTs, compared to 100% (19/19) of LGSCs. Oncogenic RAS/RAF/ERBB2 mutations were detected in 82.5% (47/57) of SBTs compared to 63% (12/19) of LGSCs, with NRAS mutations detected only in LGSC. Some copy number aberrations appeared to be enriched in LGSC, most significantly loss of 9p and homozygous deletions of the CDKN2A/2B locus. Exome sequencing identified BRAF, KRAS, NRAS, USP9X and EIF1AX as the most frequently mutated genes. We have identified markers of progression from borderline to LGSC and novel drivers of LGSC. USP9X and EIF1AX have both been linked to regulation of mTOR, suggesting that mTOR inhibitors may be a key companion treatment for targeted therapy trials of MEK and RAF inhibitors.

Genomic Classification of Serous Ovarian Cancer with Adjacent Borderline Differentiates RAS Pathway and TP53-Mutant Tumors and Identifies NRAS as an Oncogenic Driver

Purpose: Low-grade serous ovarian carcinomas (LGSC) are Ras pathway-mutated, TP53 wild-type, and frequently associated with borderline tumors. Patients with LGSCs respond poorly to platinum-based chemotherapy and may benefit from pathway-targeted agents. High-grade serous carcinomas (HGSC) are TP53-mutated and are thought to be rarely associated with borderline tumors. We sought to determine whether borderline histology associated with grade 2 or 3 carcinoma was an indicator of Ras mutation, and we explored the molecular relationship between coexisting invasive and borderline histologies. Experimental Design: We reviewed >1,200 patients and identified 102 serous carcinomas with adjacent borderline regions for analyses, including candidate mutation screening, copy number, and gene expression profiling. Results: We found a similar frequency of low, moderate, and high-grade carcinomas with coexisting borderline histology. BRAF/KRAS alterations were common in LGSC; however, we also found recurrent NRAS mutations. Whereas borderline tumors harbored BRAF/KRAS mutations, NRAS mutations were restricted to carcinomas, representing the first example of a Ras oncogene with an obligatory association with invasive serous cancer. Coexisting borderline and invasive components showed nearly identical genomic profiles. Grade 2 cases with coexisting borderline included tumors with molecular features of LGSC, whereas others were typical of HGSC. However, all grade 3 carcinomas with coexisting borderline histology were molecularly indistinguishable from typical HGSC. Conclusion: Our findings suggest that NRAS is an oncogenic driver in serous ovarian tumors. We demonstrate that borderline histology is an unreliable predictor of Ras pathway aberration and underscore an important role for molecular classification in identifying patients that may benefit from targeted agents. Clin Cancer Res; 20(24); 6618–30. ©2014 AACR.

Ankyrin Repeat Domain 1, ANKRD1, a Novel Determinant of Cisplatin Sensitivity Expressed in Ovarian Cancer

Purpose: The standard of care for ovarian cancer includes platinum-based chemotherapy. It is not possible, however, to predict clinical platinum sensitivity or to design rational strategies to overcome resistance. We used a novel approach to identify altered gene expression associated with high sensitivity to cisplatin, to define novel targets to sensitize tumor cells to platins and ultimately improve the effectiveness of this widely used class of chemotherapeutics. Experimental Design: Using differential display PCR, we identified genes differentially expressed in a mutagenized cell line with unusual sensitivity to cisplatin. The most highly differentially expressed gene was selected, and its role in determining cisplatin sensitivity was validated by gene transfection and small interfering RNA (siRNA) approaches, by association of expression levels with cisplatin sensitivity in cell lines, and by association of tumor expression levels with survival in a retrospective cohort of 71 patients with serous ovarian adenocarcinoma. Results: The most highly differently expressed gene identified was ANKRD1, ankyrin repeat domain 1 (cardiac muscle). ANKRD1 mRNA levels were correlated with platinum sensitivity in cell lines, and most significantly, decreasing ANKRD1 using siRNA increased cisplatin sensitivity >2-fold. ANKRD1 was expressed in the majority of ovarian adenocarcinomas tested (62/71, 87%), and higher tumor levels of ANKRD1 were found in patients with worse outcome (overall survival, P = 0.013). Conclusions: These findings suggest that ANKRD1, a gene not previously associated with ovarian cancer or with response to chemotherapy, is associated with treatment outcome, and decreasing ANKRD1 expression, or function, is a potential strategy to sensitize tumors to platinum-based drugs.

Genome-wide Analysis Identifies Novel Loci Associated with Ovarian Cancer Outcomes: Findings from the Ovarian Cancer Association Consortium

Purpose: Chemotherapy resistance remains a major challenge in the treatment of ovarian cancer. We hypothesize that germline polymorphisms might be associated with clinical outcome. Experimental Design: We analyzed approximately 2.8 million genotyped and imputed SNPs from the iCOGS experiment for progression-free survival (PFS) and overall survival (OS) in 2,901 European epithelial ovarian cancer (EOC) patients who underwent first-line treatment of cytoreductive surgery and chemotherapy regardless of regimen, and in a subset of 1,098 patients treated with ≥4 cycles of paclitaxel and carboplatin at standard doses. We evaluated the top SNPs in 4,434 EOC patients, including patients from The Cancer Genome Atlas. In addition, we conducted pathway analysis of all intragenic SNPs and tested their association with PFS and OS using gene set enrichment analysis. Results: Five SNPs were significantly associated (P ≤ 1.0 × 10−5) with poorer outcomes in at least one of the four analyses, three of which, rs4910232 (11p15.3), rs2549714 (16q23), and rs6674079 (1q22), were located in long noncoding RNAs (lncRNAs) RP11-179A10.1, RP11-314O13.1, and RP11-284F21.8, respectively (P ≤ 7.1 × 10−6). ENCODE ChIP-seq data at 1q22 for normal ovary show evidence of histone modification around RP11-284F21.8, and rs6674079 is perfectly correlated with another SNP within the super-enhancer MEF2D, expression levels of which were reportedly associated with prognosis in another solid tumor. YAP1- and WWTR1 (TAZ)-stimulated gene expression and high-density lipoprotein (HDL)-mediated lipid transport pathways were associated with PFS and OS, respectively, in the cohort who had standard chemotherapy (pGSEA ≤6 × 10−3). Conclusions: We have identified SNPs in three lncRNAs that might be important targets for novel EOC therapies. Clin Cancer Res; 21(23); 5264–76. ©2015 AACR.