Skip to main content
Premium Trial:

Request an Annual Quote

ASU Researchers Identify Antibody Signatures for Early Detection of Ovarian Cancer


NEW YORK(GenomeWeb) – Researchers at Arizona State University's Biodesign Institute have identified an autoantibody signature that could prove useful in early detection of ovarian cancer.

Using high-density programmable protein microarrays, they identified autoantibodies to 12 protein antigens that could distinguish between ovarian cancer samples and controls. These markers are now being tested further through the National Cancer Institute's Early Detection Research Network, Karen Anderson, an ASU researcher and one of the leaders of the effort, told GenomeWeb.

Anderson and her ASU colleague Joshua LaBaer are also on the scientific advisory board of biomarker firm Provista Diagnostics, which is evaluating the markers as part of a licensing agreement it signed with ASU in 2013, Anderson said.

The agreement, which was expanded earlier this year, gives Provista access to patents, proprietary technology, materials, processing techniques, and protocols for detecting certain autoantibodies, and allows the company to use certain biomarkers to develop diagnostic products for detecting breast and ovarian cancer and human papillomavirus.

Presented in a study published this month in the Journal of Proteome Research, the 12 markers demonstrated sensitivities ranging between 13 percent and 22 percent at specificity of greater than 93 percent. While these sensitivities are too low individually, Anderson said the researchers hope that combining the markers in a panel – or adding them to existing markers – could provide the needed performance.

The EDRN group is also looking at combinations of biomarkers from both the ASU group and other teams, she noted.

"We are currently testing sets of blinded samples [through the EDRN], and other labs are using the same samples and testing other markers, as well," she said. "The EDRN is gathering that data together and will evaluate both the individual markers and also whether there is a panel of biomarkers that could be useful for detection of ovarian cancer."

At this stage of evaluation, the EDRN teams are using samples taken from ovarian cancer patients at diagnosis, Anderson said, adding that she was not certain when data from the analysis would be available.

The next step, provided the markers appear useful, would be testing them in larger, prospective sample collections, looking at their performance "maybe six months before diagnosis or a year before diagnosis," she said.

While protein biomarker work has traditionally focused on identifying proteins over- or under-expressed due to a disease state, looking instead at the body's immune response has certain advantages, Anderson and her co-authors noted in the JPR paper.

For instance, autoantibodies to a protein are often present at levels much higher than the protein itself, providing a sort of natural signal amplification. In addition, autoantibody signatures can persist after the differences in protein expression are gone.

Anderson and her co-authors are not alone among ASU Biodesign researchers in pursuing an autoantibody-based approach to biomarker work. Working independently, another team of Biodesign researchers led by Stephen Johnston has developed a technique that uses random-sequence peptide arrays to identify disease-linked immunosignatures.

In 2010, Johnston and his colleagues Neal Woodbury and John Rajasekaran launched a spinout, HealthTell, to commercialize the approach. The company is currently in clinical feasibility studies and aims to launch its first test by the end of next year.

While HealthTell uses random-peptide arrays, Anderson and her team use microarrays containing full-length proteins generated using the nucleic acid programmable protein array (NAPPA) technology developed by LaBaer. The NAPPA technology uses printed cDNA vectors to synthesize proteins in situ.

Using the approach, the researchers were able to express more than 2,000 proteins per slide in a microarray format. For the JPR study, they used 5,177 candidate proteins, measuring the levels of autoantibodies in sample serum to each of these proteins to determine any differences between cases and controls.

The effort, Anderson said, stemmed from a previous study in which she and her colleagues observed autoantibodies specific to the protein p53 in the serum of ovarian cancer patients.

"So what we wanted to do in this study was ask the broader question of [whether there were] autoantibodies to other proteins," she said. "We know from other studies that cancer patients can develop autoantibodies to a variety of different proteins."

However, Anderson noted, "we don't yet understand the rules by which some proteins are immunogenic and some are not – why some induce immune responses in some cancers and some don't. So we did an unbiased screen."

From the screen of 5,177 antigens, the researchers selected a subset of 741 promising antigens to test against new samples, ultimately identifying the final set of 12, which they again tested with a new set of samples consisting of 60 cases and 60 controls.

Compared to approaches like HealthTell's random-sequence peptide arrays, one possible advantage of the NAPPA approach is that full-length proteins with largely intact structures enable the researchers to present the autoantibodies with the same antigens they were directed against in actual human systems, Anderson said.

However, she added that she thought no one yet knew what would prove the best way of identifying autoantibody signatures in human serum, and that most likely there would not ultimately be a single best way.

"The antibody response is extremely diverse, and it can recognize both linear and conformational epitopes," she said. 

Anderson noted that for clinical purposes it will probably be desirable to move from the NAPPA platform to a more conventional format like ELISAs. In the JPR study the researchers tested the final panel of 12 proteins using a Luminex bead assay.