Researchers at the Biodesign Institute at Arizona State University have identified a panel of 28 blood-based protein biomarkers that could aid in the early diagnosis of breast cancer.
Consisting of autoantibodies thought to be produced in response to tumor-associated proteins, the biomarker panel is one of the first to have shown potential utility as an early detection tool for breast cancer, Joshua LaBaer, director of the Biodesign Institute's Center for Personalized Diagnostics and leader on the study, told ProteoMonitor.
"I think it's fair to say that there aren't any good early detection biomarkers," he said. "People have tried, but perhaps because the [early stage] tumors aren't that vascular and perhaps because they don't start a lot of inflammation, it's been very hard to find markers for the disease."
With those difficulties in mind, LaBaer's team chose to look not for proteins produced by tumor cells but instead for autoantibodies generated in response to those proteins. The hope, he said, was that these antibodies would prove more abundant than the tumor proteins themselves.
"You want to catch cancers at the earliest possible stage, which means the smallest stage, which means that the amount of protein produced by the tumor is going to be at its smallest amount," LaBaer said. "The amount of protein that can come out of a small number of cells creates a very low concentration when diluted in the bloodstream. So, our reasoning was that if that small amount of protein gets out, it triggers an immune response, and that immune response acts like an amplification system."
"You can get multiple antibodies produced against a small amount of antigen," he added. "So that's why we went after the autoantibodies."
It's an old idea, LaBaer said, but new technologies like the nucleic acid programmable protein array, or NAPPA – developed in his lab when he was a researcher at Harvard – have made it more feasible for biomarker discovery (PM 05/15/2008).
"What used to be used was [a method] called SEREX [serological expression cloning], which used phage display libraries that you would probe with the serum of patients," he said. "The problem was that typically you would do some panning experiments to look for antigens specific for cancer and then you would probe them with serum, but you couldn't compare cases to controls easily. You'd have to take each hit and then validate it in an ELISA assay."
With NAPPAs, on the other hand – which use proteins synthesized in situ directly from printed cDNA vectors at the time of the assay – the ASU researchers were able to compare dozens of patients to dozens of controls for hundreds of antigens at once.
In the study, published online recently in the Journal of Proteome Research, the researchers began with a panel of 5,000 antigens, which they winnowed down to 750. These they tested in a cohort involving 50 breast cancer cases and 50 controls, which enabled them to cut the panel down to 100 antigens. They tested these in another cohort involving 50 cases and 39 benign breast disease controls, after which they brought the panel down to 28 analytes, which they tested in a blinded study using another 50 cases and 39 benign breast disease controls.
In the blinded study, each of the biomarkers demonstrated specificity ranging between 75 percent and 100 percent with sensitivities in the 10 percent to 40 percent range. Such low sensitivity is an inherent problem with using autoantibodies, LaBaer said, suggesting that any effective biomarker test using autoantibodies will consist of multiple proteins.
"Even in patients with the disease, only 40 percent of them make [a given] autoantibody, which means that any single autoantibody is not going to be sufficient as a screening tool," he said. "The hope is that by combining multiple autoantibodies, you get some added sensitivity. If antibody A doesn't get it maybe antibody B or antibody C will pick it up."
A related issue, LaBaer noted, is that breast cancer "isn't just one disease. It's probably six or seven different diseases. There's estrogen receptor-positive disease, estrogen receptor-negative disease, HER2-positive disease, triple-negative breast cancer. Different people have different molecular causes for their cancer and may therefore produce different sets of autoantibodies."
This also has implications for the design of the group's validation studies. Given that from a molecular standpoint breast cancer is actually a number of different diseases, the patient pools the group has used for its research thus far may not have been optimal for identifying protein markers for the condition.
"It's possible that if you do a study like we did with 50 cases, what you may really be doing is a study with seven or eight cases of basal [breast cancer], 20 cases with ER positive, 10 cases with ER negative, and so on. So the actual denominator in your study is quite low because you're really looking at multiple diseases," LaBaer said.
In addition to planning validation studies for the panel, the researchers are also working to move the assays from the microarray format to a solution-based immunoprecipitation platform in hopes of upping the test's sensitivity.
"The microarray platform is fine for screening, but it's a one-size-fits-all format. There's no optimization of chemistry or binding," LaBaer said.
The team also plans to investigate if any of the panel's proteins are tied to other diseases, which could hamper their effectiveness as markers specific to breast cancer, he noted.
[ pagebreak ]
While researchers have had little success identifying proteins in blood that are useful in early detection of breast cancer, products like Clarient's protein biomarker test Mammostrat are being used to assess patients' risk of recurrence after surgery and treatment (PM 07/16/2010). In addition, scientists like Duke University's Victoria Seewaldt and Emanuel Petricoin and Lance Liotta at George Mason University are working to identify protein phosphorylation patterns that can predict subjects' risk of progression from pre-cancer to breast cancer, as well their likely responsiveness to various therapies (PM 11/12/2010).
The ASU test, LaBaer said, would most likely be developed as an additional screening tool to be used with mammograms, which, he noted, currently miss roughly 25 percent of breast cancers.
"The hope would be that if you coupled a blood test with mammography you could achieve two potential benefits," he said. "Women who might have been missed by the mammography could potentially be picked up in the blood test. The second use is for women who had a positive mammography but for whom it isn't clear whether it's a cancer or benign. Roughly four out of five women who have a finding on mammography have a benign mass. So four out of five women are getting a surgery they don't need."
The test could be particularly useful in the case of triple-negative breast cancers, which, LaBaer said, typically show up in younger women and are often missed by mammography. The researchers are currently putting together a trial with roughly 200 participants investigating the application of the 28-marker panel to this specific form of the disease.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.