By Monica Heger
This story was originally published Feb. 14.
Scientists from the Broad Institute, Weill Cornell Medical College, and the Dana Farber Cancer Institute have sequenced matched tumor/normal pairs from seven individuals to identify large rearrangements that may have implications for prognosis and therapeutics.
The study, described last week in Nature, is the first published example of whole-genome sequencing of a prostate tumor. It is a follow-up of a previous study in which researchers at Weill Cornell Medical College sequenced 25 prostate tumor transcriptomes to identify novel gene fusions (IS 11/2/2010). The work is part of a larger collaboration between the institutes that aims to sequence the transcriptomes and whole genomes of up to 100 prostate cancer tumors.
"The real surprise is that we see major rearrangements," Mark Rubin, a pathologist at Weill Cornell Medical College and a senior author of the paper, told In Sequence. "We were surprised not only by the number, but by the fact that they do not appear random." Rubin said that the rearrangements appear organized and are possibly associated with disease outcome, adding that they probably would not have been detected without whole-genome sequencing.
The team sequenced the samples on the Illumina GA using a paired-end sequencing strategy with 101 base pair reads and 30-fold haploid coverage. They identified a median of 3,866 base mutations per tumor, a rate similar to that observed in AML and breast cancer, but several fold lower than rates for small cell lung cancer and melanoma.
About 20 non-synonymous base mutations were called within protein-coding genes, and the researchers also identified six coding indels, including a 2-base pair frameshift in the tumor suppressor gene PTEN.
They identified a median of 90 rearrangements per sample, a distribution similar to what has been seen in breast cancer. Three of the tumors contained a gene fusion between TMPRSS2, an androgen regulating gene, and the transcription factor ERG, a fusion identified in the previous transcriptome sequencing of 25 tumors.
Interestingly, in the tumors harboring the TMPRSS2-ERG gene fusion, rearrangement breakpoints tended to cluster near open chromatin, androgen receptor, and ERG DNA binding sites. In tumors lacking that fusion, however, rearrangement breakpoints were inversely correlated with those regions, and in fact, more rearrangement breakpoints were observed in closed chromatin.
The pattern suggests different mechanisms for tumorigenesis, depending on whether the tumor has a TMPRSS2-ERG gene fusion or not. A similar pattern has also been observed in breast cancer, where breakpoints were strongly associated with estrogen receptor binding sites.
"What we're finding is that some of the key changes due to alterations in the prostate tumor are probably going to be relevant in breast cancer," Rubin said. "Both of those proteins [estrogen and androgen] probably have a lot of overlap with how they engage the machinery of the tumor cells … These are early observations, but we think that there will be a similarity between the two."
Holger Sültmann, whose team at the German Cancer Research Center is working on prostate cancer sequencing as part of the International Cancer Genome Consortium, thought the rearrangements were surprising because they appear to occur in one single event in which they also cause the loss of tumor suppressors or the gain of oncogenes.
"They are finding basically no loss of genetic material, but very drastic and complex rearrangements, which are occurring in one event," he said. Previously, mutations had been thought to accumulate over time, but the new work shows that a single rearrangement "may lead to the loss of tumor suppressor genes or the gain of an oncogene in one event, instead of a gradual accumulation of mutations."
In four of seven tumors, rearrangements inactivated tumor suppressors. Both PTEN and MAGI2 genes were found to be disrupted. PTEN is a known tumor suppressor in prostate cancer, and while MAGI2 is a novel finding in prostate cancer, it has been found to be disrupted in other cancers where PTEN mutations are common and appears to function in the same pathway.
"It suggests that there's a new way that the pathway can be mutated," which could have therapeutic implications, said Rubin. There are currently some drugs that target the PTEN pathway, and more are in development, and this finding suggests that "potentially more individuals could benefit from the treatment than we previously thought," he said.
Additionally, the loss of these tumor suppressor genes coupled with the TMPRSS2-ERG gene fusion could potentially be a new way to subdivide patients into aggressive and non-aggressive groups, said Rob Bristow from Toronto's Princess Margaret Hospital, who is heading up the Canadian-ICGC prostate cancer sequencing project. In previous studies, the combination of the fusion and loss of the tumor suppressor led to poor prognosis.
If the finding is replicated in additional samples, it suggests that the gene fusion and loss of tumor suppressor combo could serve as a biomarker for predicting disease outcome and making treatment decisions, and could also serve as a way to select patients for clinical trials, Bristow said.
"There are drugs coming down the pipeline that would target PTEN deficient pathways," he said. "So patients could be chosen for a clinical trial that uses one of these agents."
Additionally, because patients with this mutational profile appear to do worse — even after surgical removal their tumors tend to metastasize — clinicians would know early on that the patients would need not only localized treatment such as radiology or surgery, but also a more global treatment, such as a hormone-based therapy or molecular therapy, that would kill the metastases before they developed.
However, Bristow added, that is still well into the future. "We're talking about the whole-genome sequencing of only seven prostate cancers. So in order to understand the utility of these findings, we're going to have to repeat it in hundreds of prostate cancers from patients with different stages of the disease."
Bristow said that it's likely that whole-genome sequencing of prostate tumors could eventually be used to answer three main clinical questions: which cancers are indolent, and therefore don't need treatment; in non-indolent cancers, which ones can be treated successfully with a localized treatment; and finally, which patients are likely to have an aggressive form of cancer that will require individualized treatment targeting the patient's specific mutations.
Have topics you'd like to see covered by In Sequence? Contact the editor at mheger [at] genomeweb [.] com.