This story originally ran on Nov. 11.
By Tony Fong
Discovering protein biomarkers and having them adopted by clinical practitioners are two distinct processes with different criteria.
If clinical proteomics will ever achieve success, proteomics researchers will need to master both processes. That was the take-home from some speakers at a meeting held last week by the American Association of Clinical Chemistry.
Speakers from federal funding and regulatory agencies, the private sector, and the clinical chemistry community also suggested that if finding the biomarkers is hard, getting them into the clinic may be even harder. And unless proteomics researchers know what the process entails, they will always get stuck in the discovery-to-clinic pipeline.
The two-day meeting, held in Bethesda, Md., came on the heels of AACC's proteomics division gaining permanent status after five years as a provisional division [See PM 11/06/09]. Along with highlighting research being done in proteomics, the meeting also was an opportunity for funding agencies to provide an update on the proteomics-focused activities within their agencies, and to assess the field.
Pothur Srininvas, a program director at the National Heart Lung and Blood Institute, said that the NHLBI has "invested in proteomics for eight years" in an effort to help develop technologies and apply the science to answer clinical questions.
But so far, he said, the only clinical traction his agency has seen is in cardiovascular diseases with the discovery and eventual implementation of troponin as a diagnostic marker for heart disease. Since then, however, despite the support from NHLBI, success has been hard to come by, and the effort has experienced the same problems seen by proteomics researchers elsewhere: analytical issues and clinical significance.
Henry Rodriguez, director of the NCI's Clinical Proteomic Technologies for Cancer initiative, added that while "proteomics is delivering … the rate at which it's delivering … is not what people want to see."
One of the highlights from last week's meeting was a session held by the US Food and Drug Administration to provide an overview of its approval process and factors for companies to keep in mind as they move from discovery to clinical translation to regulatory submission.
The FDA process is a mystery to many proteomics researchers, and the fact that proteomics is still developing, and that the technology remains in a formative stage, makes the regulatory process especially challenging.
In his talk, Leigh Anderson, founder and CEO of the Plasma Proteome Institute, reiterated that to his best calculations, only 109 proteins have been cleared by the FDA for clinical use, either by direct approval or approval of the protein or tests containing proteins. None, however, has been discovered by proteomics technologies, but rather by earlier protein research methods.
That figure was arrived at, however, before Vermillion's OVA1 test for ovarian cancer received 510(k) clearance from the FDA in September [See PM 09/17/09].
Regardless, last month Anderson told attendees at a meeting held by the National Cancer Institute's clinical proteomics group that during the past 15 years, the number of proteins approved for clinical tests had remained unchanged at 1.5 per year.
Another 96 proteins are available for diagnostic purposes through laboratory-developed tests. By definition, these are not FDA-approved but overseen by Clinical Laboratory Improvement Amendments regulations.
In an attempt to facilitate greater communication between the FDA and the NCI about the NCI's proteomics initiatives and to address bottlenecks in clinical proteomics, about a year and a half ago an NCI-FDA interagency oncology task force subcommittee on molecular diagnostics decided to focus on proteomics.
[ pagebreak ]
In a workshop held a year ago, the subcommittee decided on an exercise in which mock pre-market 510(k) submissions would be submitted to the FDA through the pre-IDE process, used by companies seeking informal advice from the agency on submissions for novel devices.
The purpose of the mock 510(k) applications was to shed light on the regulatory process for proteomics researchers. Reports detailing the process and comments from the FDA on the submissions are expected in the coming months, Zivana Tezak, a scientific reviewer at the FDA, said last week.
There are currently two mock submissions: one for a multiplex mass spec-based assay and one for a multiplex affinity array.
FDA 101
In her talk, which provided an overview of the FDA approval process, Tezak said that one problem that researchers need to be mindful of is that regulatory goals often are very different from research and exploratory goals, and that "it's not enough to understand the biology." Before going to market with a device, a company needs to prove to the FDA that the device is safe and effective, based on valid science.
There are two pathways to market for a medical device, the pre-market 510(k) notification and the pre-market approval, or PMA, route. The former is for devices that are substantially "equivalent to a predicate," or device already on the market that is considered to have minimal risk to patients. The PMA process is for devices that have the highest risk.
In vitro diagnostic devices are regulated under three classes: Class I devices, which are considered the least risky and subjected to pre-market 510(k) notification, Class II devices are subject to pre-market 510(k) approval, while Class III devices are considered the riskiest and subject to PMA.
A device's risk and what class a submission falls into are "highly dependent on the claim of the test," Tezak said. For example, someone may file an application for a cystic fibrosis transmembrane conductance genotyping multiplex assay. If the claim on the submission is to aid in diagnosis, it could be approvable under the pre-market 510(k) route. But if intention of the test is for fetal screening, it would need to receive PMA to go to market.
Similarly, a breast cancer assay would need to receive PMA for use as a screening diagnostic, but would require only pre-market 510(k) notification if the intended use is for prognosis in a patient who has already been diagnosed for the disease.
Intended use "drives everything else that you do," said Estelle Russek-Cohen, from the division of biostatistics at the FDA's Office of surveillance and biometrics.
So what does the FDA require to approve a medical device? In general, it includes a description of the device, covering both the instrument and reagents. It should also include detailed description of "appropriate internal and external controls [and] calibrations," Tezak said.
The submission also has an analytical performance section in which an applicant will need to provide information about the device's precision, accuracy, and performance around the cut-off point, and other performance measurements.
And the FDA will want information about reproducibility of the device, from lab to lab and instrument to instrument, and major sources of variability.
Even if a device can analytically do what its developer claims, both Tezak and Russek-Cohen said that may not be enough for FDA approval. For example, in the case of a DNA test for human papillomavirus, it may be able to detect the virus as a marker, but to get the nod from the agency the test would need to also show clinical relevance by proving the detected biomarker is an indication of active disease.
[ pagebreak ]
According to James Wittliff, a professor of biology at the University of Louisville, and research professor of surgery at the university's School of Medicine, some considerations in developing FDA-approvable proteomics tests include the complexities of tests using multi-analytes versus a single analyte; the challenges of evaluating a multi-analyte assay when there are so few successful precedents; and the lack of reference materials and standards providing reproducible results of proteomics experiments and biomarker-based tests.
Private Industry is Your Friend
But the FDA approval process is not the only challenge to moving proteomics discoveries to the clinical setting, some said last week. In a session on the role clinical chemistry can have on proteomics and translational medicine, Thomas Moyer, who specializes in clinical biochemistry and immunology at the Mayo Clinic, and a past president of AACC, said that in the US, there is a strong history of government support for scientific discovery.
But when it comes to translating those discoveries into practice, those endeavors have largely been funded by private industry. Indeed, both Jack Ladenson, a professor of clinical chemistry at Washington University School of Medicine, and Kevin Halling, vice-chair of R&D in the department of laboratory medicine and pathology at the Mayo Clinic, both bypassed NIH funding and instead received funding from the private industry for work they did: Ladenson developed the antibodies used for the creatine kinase MB test for myocardial infarction, while Halling developed a fluorescence in situ hybridization assay for bladder cancer detection.
Ladenson's work was funded by Monsanto as part of an agreement between the company and Washington University. Halling collaborated with Vysis, now Abbott Molecular, on the FISH assay.
"Get past this mindset that if you are working with industry, you are working with the dark side," Moyer told the audience.
He also recommended that the bioindustry adapt a patent system similar to the one used by the electronics industry. About 20 years ago, the microprocessor industry was "flooded with patents," and it became bogged down in negotiations with the patent holders, "unable to acquire licenses to the series of patents necessary to create new microprocessors," Moyer said in an e-mail to ProteoMonitor.
To solve this, the microprocessor industry created a clearinghouse for patent holders, by which industry players interested in negotiating for the rights to a patent could do so with the clearinghouse rather than each individual patent holder.
"Patent holders learned that by working in the consortium, they could rapidly realize gain from their patents that was not evident when working outside the consortium concept," Moyer said, adding that the cell phone industry now bears witness to the value of such a consortium model.
The bioindustry is now in the same place as the microprocessor industry two decades ago with "every scientist … patenting every SNP imaginable," slowing down the implementation of new discoveries, Moyer said. "If biotech could agree on some form of consortium concept, implementation of novel technologies would likely happen much faster than it is happening today."