Integrated analyses of the ovarian cancer genome

4 Jul 2011

Ovarian cancer is the fifth most common cancer in women in the UK; worldwide, it accounts for about 4% of all cancers diagnosed in women.

 Its prognosis is good on the rare occasions where it is diagnosed early, but most ovarian cancers are only detected at a late stage when outcomes are poor. The most aggressive common type is high-grade serous ovarian adenocarcinoma; about 13% of cases of this cancer occur in women with germline mutations in BRCA1 or BRCA2

Most cases, however, arise from a complex series of somatic mutations that have not yet been fully elucidated. Understanding more of the genetic picture of serous ovarian adenocarcinoma may help in developing more targeted treatments for this often intractable disease.

The Cancer Genome Atlas project was set up by the US National Institutes of Health in 2008 to collate and catalogue the genetic changes associated with cancer. 

Following a pilot programme, its network of researchers is now analysing the genetics of over 20 different, relatively common tumour types. A group of Cancer Genome Atlas Research Network scientists led by Paul Spellman at the Lawrence Berkeley National Laboratory has now conducted a comprehensive analysis of the genetic changes associated with this disease. 

The researchers sequenced the exomic (protein-coding) DNA regions from tumour and matched normal tissue samples from 489 patients with confirmed diagnoses of Stage II – Stage IV high-grade serous adenocarcinoma of the ovary. Micro-RNA expression, mRNA expression, DNA copy number and promoter methylation analyses were all carried out with the same samples using microarrays. 

The samples reflected a typical pattern of age and cancer stage at diagnosis, tumour grade and outcome for patients with this condition.

The most commonly mutated gene by far was found to be the tumour suppressor TP53, which was found to be mutated in 96% of all tumours.

 Eight other genes were found to be mutated in a significant number of cases. These included BRCA1 and BRCA2, as expected, and also RB1, NF1, FAT3, CSMD3, GABRA6 and CDK12. Of these, CDK12 has previously been implicated in lung and gastro-intestinal cancers.

 Copy number abnormalities were observed at 113 locations, and increases in promoter methlyation (which can silence gene expression) were observed involving 168 genes.

An analysis of the mRNA expression profiles of these tumours led the researchers to classify them into four distinct and statistically separable subtypes. These were termed, respectively, ‘immunoreactive’, ‘differentiated’, ‘proliferative’, and ‘mesenchymal’, with each characterised by a different set of over-expressed genes. Further, less statistically robust subsets could be defined based on DNA methylation patterns.

A separate set of 193 genes was discovered to provide a survival signature, with high expression of 108 of these correlated with poor survival and 85 with good survival.

A combined analysis of mutations, DNA methylation and gene expression patterns based on pathways known to be associated with cancer showed the RB1 and PI3K/RAS pathways to be de-regulated in 67% and 45% of these ovarian cancer cases respectively.

Many of the cancers with wild type BRCA1 and BRCA2 showed defects in DNA repair by homologous recombination, either through epigenetic silencing of BRCA1 or through changes in interacting genes.

The NOTCH and FOXM1 signalling pathways were also frequently found to be altered.

In summary, this comprehensive genomic analysis of high grade serous ovarian adenocarcinoma confirmed the genetic heterogeneity of the condition; identified about half the cases as being deficient in homologous recombination and that might benefit from PARP inhibitors; and defined subtypes that could be used to stratify patients for more personalised treatment.




The Cancer Genome Atlas Research Network (2011). Integrated genomic analyses of ovarian cancer. Nature 474, 609-615. DOI 10.1038/nature10166