This story originally ran on Sept. 30.
By Tony Fong
TORONTO – Clinical proteomics may have a spotty and troubled history, but at the Human Proteome Organization's annual conference here this week, the push to move proteomics from the research lab to the clinical setting was everywhere to be seen.
From talks where researchers shared data from studies of a wide variety of diseases to numerous posters on new techniques to study biological samples and processes that could lead to clinical breakthroughs, proteomics researchers continued in their quest to bring clinical relevance to the field.
That may be no different from past HUPO meetings as well as other proteomics meetings. But for a field that has been viewed critically for not having delivered much to the clinical setting, this year's meeting was a striking contrast to the HUPO conference held two years ago in Seoul when leaders in the field took to task what they saw as "sloppy science" being done and called the biomarker field "chaotic" [See PM 10/11/07].
This year, the criticisms were more muted and the work was left to speak for itself. During the weekend, a whole-day session was devoted to highlight some of the challenges and research being done in clinical proteomics. It began with Dennis Hochstrasser, director of clinical pathology at the Geneva University Hospital and vice dean of medicine at the university, who declared that in spite of the perception that proteomics has failed to provide many answers for the clinical setting, "I strongly believe that proteomics has delivered. … not only in the diagnostics area but also in the treatment of diseases."
Indeed, two weeks before the start of the conference, a proteomics test, the OVA1 ovarian cancer triage test, won regulatory approval by the US Food and Drug Administration, one of just a very limited number of proteomics tests to ever be approved by the FDA. The test was co-developed by Vermillion and Quest Diagnostics and is expected to become commercially available in the fourth quarter [See PM 09/17/09].
OVA1 is based on a panel of biomarkers identified by Daniel Chan, director for the Center for Biomarker Discovery at Johns Hopkins University School of Medicine, who said during the weekend session that it took nine years to get regulatory approval for the test.
Still, the history of clinical proteomics has been marked mostly by disappointment. Initially hyped as a research field that would transform medicine, it has instead become a science that, according to some at the conference, has been unable to back up the talk with the walk.
During a session on HUPO's Human Proteome Project, Sudhir Srivastava, chief of the cancer biomarkers research group in the division of cancer prevention at the National Cancer Institute, said that the main obstacle to funding for the project, estimated at $1 billion, is that the broader community is not convinced of the benefits that proteomics will have [See related story this issue].
Other speakers at the meeting reiterated the need for the field to do a better job of marketing itself.
But according to Will Dracup, CEO of software firm Nonlinear Dynamics, clinicians already know all about the potential benefits of proteomics. What concerns them is that proteomics experiments largely have not been reproducible. Until the field does a better job of designing studies, following proper protocols, and creating standards, clinicians will remain unconvinced that proteomics has a place in their world, he said.
A Better Predictive Tool?
One of the promises of proteomics has always been that it could provide information on a level that genomics can't, and in a talk about work being supported by the National Institute of Diabetes and Digestive and Kidney Diseases, Salvatore Sechi, the proteomics program director at the institute, said that several genome-wide association studies suggest that genomics does not provide a very good predictive tool for diabetes.
[ pagebreak ]
In fact, the need for a proteomics marker for diabetes is indicative of the need for such a marker in many diseases, he said: While there is already a marker in existence, the oral glucose tolerance test, the gold standard in diabetes screening, has several drawbacks including the fact that it is not highly reproducible and a patient is required to fast before taking the test, making it "not optimal," Sechi said.
NIDDK is currently supporting four research projects to find new markers, he said: One at the Institute for Systems Biology is performing glycoproteomics discovery work and has identified 22 potential markers. Another project at Pacific Northwest National Laboratory is using an accurate mass tagging approach and has identified 87 peptides that are differentially expressed in three populations.
In addition, Intrinsic Bioprobes, a Tempe, Ariz., firm, is developing mass-spec based immunoassays to capture protein antibodies in high-throughput and identified ApoClll and serum amyloid A as potential biomarkers. The fourth effort, at Johns Hopkins, has identified O-GlcNAc modification on certain sites as a possible biomarker.
Other work discussed during the conference included that of Richard Drake, a professor and co-director of the Eastern Virginia Medical School biomedical proteomics center, who presented work he's conducting using MALDI-TOF mass spectrometry to identify and characterize microorganisms in his facility, including methicillin-resistant Staphylococcus aureus. MALDI-TOF mass spec-based identification of microorganisms, he added, is already being done by hospitals all over the world.
Hochstrasser declared afterward that 20 years from now, "people will say this is when proteomics began," referring to the fact that a simple mass spec platform such as a MALDI-TOF can, and is, being used in the clinical setting.
Still, moving from discovery to the clinic remains a challenge. Several factors have prevented more clinical proteomics products from reaching the market, said Chan, such as the complexity of plasma proteins, the changing nature of proteomic profiles, systematic biases in non-disease associated factors, and biological variations.
In order to translate biomarkers to a clinical diagnostic, they need to have high analytical accuracy, Chan noted. In addition, he said that they need not only be specific and sensitive, but the data must be reproducible and have a lot-to-lot variation of less than 10 percent. The clinical performance, he added, must have a defined intended use and target population and studies need to be designed to limit non-biological variations and effects. Researchers should identify pre-analytical and biological variability, as well.
While immunoassays have been the predominant method used in clinical labs, Chan said that in the future mass specs will be used in combination with immunoassays, but that will happen only when mass specs are "extremely accurate."