NEW YORK (GenomeWeb News) – A new prognostic gene signature for ovarian cancer, described in PLoS Medicine yesterday, is revealing ovarian cancer-related pathways and transcription factors that could help researchers close in on new treatments.
A team of Dutch investigators led by University of Groningen gynecological cancer researcher Ate van der Zee uncovered an 86-gene signature that could distinguish between high- and low-risk individuals with different ovarian cancer survival. Given the limited treatment options for ovarian cancer, such signatures may have less of a role for guiding treatments than for coming up with new therapies — at least for now. But that isn't stopping others from pursuing their own predictive tests for ovarian cancer.
Ovarian cancer affects more than 25,000 American women each year and is linked to roughly 14,000 deaths in the US annually. It is currently treated with surgery to remove the tumor followed by chemotherapy. Even so, ovarian cancer is often detected at an advanced stage and treatment cures only a small minority of patients. Most will relapse following treatment leaving the five-year survival rate at a dismal five to 30 percent.
Ovarian cancer cells often have elevated levels of a protein called CA-125 and blood tests for the antigen can help physicians evaluate how treatment is progressing. The goal, though, is to identify diagnostic, predictive, and prognostic biomarkers that can lead to earlier detection and improved treatment.
BRCA1 and BRCA2 mutations — best known for their link to breast cancer — can also up the chances of ovarian cancer. But these mutations explain only about 10 percent of ovarian cancer risk. And given the relative rarity of the disease, a genetic or other test aimed at diagnosing ovarian cancer early in women without symptoms would have to be extremely accurate to prevent a slew of false positives.
A bit more progress has been made in the search for molecular biomarkers that can classify ovarian cancer patients based on their treatment response and survival profiles.
In the latest such study, van der Zee and his team used approximately 35,000 70-mer oligonucleotide microarrays made by the Netherlands Cancer Institute to assess gene expression in 157 advanced stage serious ovarian cancers. In so doing, they found 86 genes whose expression was linked to overall survival time.
The researchers subsequently demonstrated that 57 of these genes could be used to predict survival profiles in a published study of 118 individuals with ovarian cancer whose gene expression was assessed on a different microarray platform.
The researchers didn't find genes that overlapped with signatures from previous studies, author Steven de Jong, a medical oncology researcher at the University of Groningen, told GenomeWeb Daily News. But, he added, they did find some overlapping pathways and regulatory networks. Sixteen out of the 17 pathways and 12 of 13 transcription factors implicated in the team's original ovarian cancer signature were also affected in the validation group.
"This also gets us some clues to which pathways are important to the biological behavior of these tumors," van der Zee told GWDN.
The study is not the first to look for genetic or other signatures that could predict ovarian cancer outcomes. In 2004, researchers from Harvard University and elsewhere published a paper in the Journal of Clinical Oncology identifying a 115-gene signature that they dubbed the Ovarian Cancer Prognostic Profile. The following year, the same team published a follow-up paper describing a gene expression signature involved in chemotherapy response. Numerous other gene expression studies have followed.
Researchers have looked for other types of prognostic markers, too — examining everything from single protein levels and proteomic profiles to microRNA and genetic variants.
For instance, a team of researchers led by University of Texas MD Anderson Cancer Center researcher Anil Sood found that levels of the small RNA processing proteins Dicer and Drosha were linked to ovarian cancer outcomes.
And in December, another team of researchers retrospectively gauged the levels of 21 protein biomarkers in 500 ovarian tumor samples. Nine of the biomarkers could be used as prognostic indicators across the entire group of individuals with ovarian cancers. But just three stood up prognostically — and only in the serous cell type — when the individuals were considered by cancer subtype.
That group emphasized the need to consider the different ovarian cancer subtypes as distinct conditions during studies aimed at new biomarker discovery. Kate Lawrenson and Simon Gayther, gynecological cancer researchers at the University College London, also stressed the importance of ovarian cancer classification in a perspectives article published in PLoS Medicine yesterday.
Ovarian cancer requires other special considerations besides subtypes, experts say.
In 2006, the US Department of Health and Human Services' Agency for Healthcare Research and Quality prepared a report on "Genomic Tests for Ovarian Cancer Detection and Management." The report noted that the "most studied use of genomic tests was for predicting or detecting response to treatment after debulking therapy and adjuvant chemotherapy."
Even so, the authors found a dearth of clinical data on genomic tests related to ovarian cancer in general. They also suggested that the development of such tests is limited by factors such as small sample sizes and unrealistic estimates for the prevalence of ovarian cancer in the general population.
"Although research remains promising, adaptation of genomic tests into clinical practice must await appropriately designed and powered studies in relevant clinical settings," the AHRQ report's authors concluded.
That means that researchers will need to carefully consider a biomarker's potential applications and do research in an appropriate setting, Bill Lawrence, a researcher with AHRQ's Center for Outcomes and Evidence, told GenomeWeb Daily News.
Still, researchers say such approaches are difficult given the relative rarity of ovarian cancer compared to breast cancer and other diseases. That difference in epidemiology may also explain why gene expression signatures for breast cancer — including Agendia's MammaPrint, which has already secured US Food and Drug Administration approval — are closer to clinical application than ovarian cancer signatures.
For his part, though, van der Zee said the ovarian cancer expression profile he and his team uncovered will likely never be used in a clinic. Instead, he said the signature is valuable for its potential to provide new clues about ovarian cancer biology and plausible drug targets.
Van der Zee argued that, at least at the moment, predicting relapse is not as crucial for ovarian cancer as it is for some other cancers, since most ovarian cancers eventually recur. Whereas genetic tests that can accurately predict outcomes could lead to variable treatments in early stage breast cancer patients, van der Zee continued, ovarian cancers are usually diagnosed at an advanced stage and have very limited treatment options.
But prognostic tests could be more useful for guiding treatment if new therapies become available, van der Zee said. He and his co-workers plan to look at whether they can inhibit pathways in cell culture or tumors and then transfer that information to mice and humans, de Jong explained, "We're trying to connect it with targeted therapies now."
A team of Duke University researchers expressed a similar point of view in a paper published in Clinical Cancer Research in 2005. That group used microarrays to find gene expression patterns associated with better long-term ovarian cancer survival.
At the time, senior author Jeffrey Marks, a Duke pathology researcher, and his colleagues concluded that "[t]he value of gene expression-based predictors of prognosis in advanced ovarian cancers will not be fully realized until additional therapies are available for those destined to have poor survival following conventional chemotherapy."
Still, some are already pursuing clinical prognostic tests. Researchers at Key Genomics, which specializes in bioinformatics and predictive genomics, are using a predictive gene expression algorithm to come up with prognostic gene signatures. The company is collaborating with Transgenomics on the ovarian cancer front.
Key Genomics CEO Tim Gallagher told GenomeWeb Daily News that the company plans to deliver a full ovarian cancer signature to its partners for clinical evaluation some time this year.
Last June, Rosetta Genomics announced that they are developing an miRNA-based biomarker test to distinguish between patients who are most likely to respond to platinum-based chemotherapy and those who will probably need a secondary treatment.
Meanwhile, researchers at Berkeley Lab and the University of California at San Francisco are reportedly developing their own biomarkers for predicting which ovarian cancer patients will require alternative therapies and for coming up with new ovarian cancer treatments.