Researchers led by the British Columbia Cancer Agency in Vancouver have used high-throughput transcriptome sequencing to discover in a rare type of ovarian cancer a mutation that has diagnostic, as well as likely therapeutic, value.
The study, published online last week in the New England Journal of Medicine, underscores how next-generation sequencing can directly benefit the clinic and indicates that whole-genome sequencing may not always be needed to discover important cancer mutations.
"This is a small example of how these technologies won't just map out the complexity of disease, as has been shown previously in some other cancer genomics studies, [but] produce distinct results which can be immediately usable clinically," said David Huntsman, medical director of the Center for Translational and Applied Genomics at the BCCA, the senior author of the study.
For their study, the researchers focused on granulosa-cell tumors of the ovary, which account for less than 5 percent of all ovarian cancers. Clinically and pathologically, GCTs have not been well defined, and there is no treatment other than surgery, according to Huntsman.
Initially, the team analyzed the transcriptomes of four adult-type GCTs by whole-transcriptome paired-end RNA sequencing, using the Illumina Genome Analyzer platform, and found that all four had a missense point mutation in the FOXL2 gene, which encodes a transcription factor that is involved in the development of granulosa cells.
Analyses of further tumors showed that the same mutation was present in 97 percent of 89 additional adult-type GCTs and absent in 329 unrelated ovarian or breast tumors. It was also present in a smaller percentage of related ovarian cancer types. The researchers are now conducting further validation studies to determine the specificity of the mutation and the best approach to assay it in the clinic.
According to Huntsman, the results not only redefine GCTs of the ovary on a molecular basis, but also provide "the first clue to where a therapeutic target may lie."
"We believe that this discovery will lead to improved treatments, in addition to being a diagnostic for granulosa-cell tumor," he said.
The reason his team chose transcriptome sequencing — a targeted approach that leaves out approximately 98 percent of the genome — over whole-genome sequencing is that the latter was "too big a budget item," Huntsman said.
He said it is unlikely that the researchers missed any diagnostic mutations for GCTs, since there are not many tumors that have more than one of these so-called pathognomonic mutations.
Further sequencing may be warranted in the future, he said, once a treatment for the disease is available, in order to identify mutations that distinguish responders from non-responders.
One reason he and his colleagues were able to make their discovery by sequencing only a small number of samples is that GCTs are genomically stable. For cancers with more unstable genomes, researchers "are probably gong to have to look at many, many cases to map out what's going on," Huntsman said, for example as part of the Cancer Genome Atlas or the International Cancer Genome Consortium.
The BCCA researchers favored transcriptome sequencing over exome capture and sequencing for their study because at the time they did the experiment — more than a year ago — the transcriptome approach was well established at their center, and because it allowed them to obtain both gene-expression and exon-mutation data in a single experiment, according to Huntsman, something that exon capture does not afford.
[ pagebreak ]
"Until very recently, there has not been a viable option for full exome capture," said Martin Hirst, functional genomics group leader at the Genome Sciences Centre at the BCCA, in an e-mail message. "But we are not opposed to [the exon capture] approach and are currently evaluating a number of hybridization-based capture platforms."
Exon sequencing could reduce the amount of sequencing required, he said, because all captured exons should be present in equimolar amounts. He cautioned, though, that exon capture is currently "limited by our incomplete understanding of the transcriptome."
While it is possible that transcriptome sequencing missed certain mutations present in lowly expressed transcripts, or truncating mutations that caused decay of the transcript, "we are happy with our choice," Huntsman said.
According to Hirst, who was in charge of sequencing for the project, the center currently runs seven flowcell lanes of an Illumina GAII paired-end run for each tumor whole-transcriptome shotgun sequencing library, which costs on the order of C$8,000 ($7,000) in reagents. When the Illumina platform reaches an output of 96 gigabases per run, he said, only one lane will be required.
He and his colleagues have developed two WTSS protocols, one using at least 5 micrograms of total RNA that requires no amplification, and another one — suitable for tissue biopsies — that uses as little as 8 nanograms of total RNA and includes an amplification step.
By comparison, researchers at Washington University School of Medicine said last year that the total cost of sequencing an acute myeloid leukemia sample and normal control — including labor, data analysis and storage, and instrument amortization — was $700,000, and estimated the second such study to cost on the order of $200,000, although that cost may since have dropped (see In Sequence 10/14/2008).
Illumina just launched a personal genome sequencing service at a cost of $48,000 (see other article in this issue).
Elaine Mardis, co-director of the Genome Center at Washington University and one of the leaders of the WashU AML study, told In Sequence this week that sequencing both the genome and the transcriptome of tumors is going to be important in cancer genome studies.
While whole-genome sequencing provides information about mutations and structural variants throughout the genome, transcriptome sequencing delivers, in addition, information about alternative splicing, novel genes, and alleles that are mutated but not expressed.
Finding mutations by sequencing the transcriptome alone, however, "is a little harder," she said, since no good models for sequencing depth and coverage are available to determine how much sequencing is required to cover genes that are mutated and expressed at low levels. "The cost can be quite high," she said, and could approach that of whole-genome sequencing.
The Canadian researchers' paper is not the first published tumor transcriptome sequencing study — last year, a team of researchers from Brigham and Women's Hospital, 454 Life Sciences, RxGen, and the National Center for Genome Resources published an article in PNAS in which they described sequencing the transcriptomes of four lung tumors using 454's technology (see In Sequence 2/26/2008).
But the BCCA authors believe that their study is one of the first to demonstrate immediate clinical benefits. "In a very practical sense, this may be the first indication of the power of this technology," Huntsman said. "There [are] obviously going to be other, bigger stories coming down the pipeline as these technologies become more widely used, but it's a small taste of how these technologies are going to completely revolutionize our understanding of the cancer problem," he said.
The BCCA researchers have been using their transcriptome-sequencing approach for a variety of other cancer projects. According to Hirst, his center has prepared and sequenced almost 170 whole-transcriptome shotgun sequencing or miRNA libraries from cancer samples since mid-2007, and used the data to measure expression abundance of genes and exons, detect mutations and polymorphisms, measure alternative splicing, and identify fusions and RNA editing events.
Huntsman said that his ovarian cancer group has been focusing on rare subtypes of the disease, "where we feel we can obtain knowledge, as opposed to just data, by looking at smaller numbers of cases." For more common types of ovarian cancer, he and his team are collaborating with, and providing samples to, TCGA.