Miami healthcare firm Opko Health is building a protein biomarker-based Alzheimer's diagnostic using a new technique for detecting disease-specific antibodies developed by researchers at the Scripps Research Institute in Jupiter, Fla.
The technique, which is described in a paper published this month in Cell, uses synthetic molecules called peptoids to screen for autoantibodies in blood associated with Alzheimer's disease and could be broadly useful as a protein biomarker discovery tool, said Thomas Kodadek, professor of chemistry and cancer biology at Scripps and a scientific director at Opko.
Traditionally, scientists looking for biomarkers focus on identifying proteins that are differentially expressed in diseased subjects compared to healthy controls. Some researchers have suggested, though, that it might be better to concentrate not on these differentially expressed proteins, but rather on the antibodies a subject generates in response to them.
Such an approach has several potential advantages, Kodadek told ProteoMonitor. Because "antibodies are a lot more stable than most serum proteins, how you collect the sample, how you freeze it isn't terribly important," he said. "Whereas for the guys who are doing standard proteomics, that sort of stuff just plagues them."
More significantly, he added, "the adaptive immune response against an initiating antigen results in millions-fold amplification of the antibodies that recognize it." That amplification makes the antibodies much easier to detect and measure than typical protein biomarkers, many of which are of low abundance.
Researchers have largely avoided such antibody-based approaches, however, due to the perceived difficulty of developing capture agents capable of isolating them in patient samples, Kodadek said.
"The immunology community is kind of stuck in this box where they believe that the only ligand that could bind to these putative antibodies with high enough affinity to pull it out of blood would be the native antigen," he said. "The thinking is that antibodies are incredibly specific receptors, so to pull them out you need to know what the native antigen is, and that's really hard to do. You're forced to speculate as to what would be the initiating antigen, and for the vast majority of diseases, we simply don't know."
Technological advances have provided avenues of attacking this problem. Arizona State's Joshua LaBaer, for instance, used nucleic acid programmable protein arrays – which employ proteins synthesized in situ directly from printed cDNA vectors at the time of the assay – to screen patient samples against roughly 5,000 antigens as part of a study his lab recently performed looking for autoantibody biomarkers for breast cancer. The researchers eventually identified 28 antibodies that could aid in early diagnosis of the disease (PM 12/24/2010).
Kodadek's team tried a different route. Instead of using proteins as antigens, the researchers used peptoids – synthetic N-substituted oligoglycines – to screen for antibodies raised in response to the onset of Alzheimer's.
"We wanted to run the [immunoglobulin G] population past a large collection of molecular shapes, just past lots of different synthetic molecules," Kodadek said. "So it came down to a question of how could we do that most conveniently, and peptoids are among the easiest compounds to synthesize large collections."
In the Cell study the researchers screened patient blood samples against a library of roughly 40,000 peptoids, identifying three peptoids that between them bound to a total of two IgG antibodies present in Alzheimer's patients but not normal controls. Since then, Kodadek's team has synthesized a library of roughly 1 million peptoids. Using this larger library they've identified an additional six Alzheimer's-specific autoantibodies.
Kodadek and Opko have also begun investigating the biomarkers' utility in larger subject pools. To date they've tested them in roughly 250 cases and controls, and are currently obtaining longitudinal sample sets from researchers at Columbia Medical School and Albert Einstein College of Medicine with the aim of identifying which markers might be most useful for early-stage detection of the disease. According to Kodadek, those studies should be finished in the next two to three months.
The company will be ready to submit the test for US Food and Drug Administration approval sometime this year, he said, although, he noted, it might opt out of the FDA route and sell the panel as a laboratory-developed test instead. Given the current lack of treatments for Alzheimer's, an early detection test might right now be most useful as a research tool for pharma companies, as opposed to a general patient screening device, Kodadek suggested.
"There's an almost ethical issue that, since there's no treatment for [Alzheimer's], do you really want early detection?" he said. "What I think the test is really going to be brilliant for is to help pharma companies stratify their clinical trials with the appropriate patients. In the next five years that would be the most useful thing we would do with it."
Much work has gone on in recent years into identifying blood-based protein biomarkers for early detection of Alzheimer's. Last month the NIH-funded Alzheimer's Disease Neuroimaging Initiative released data comprising measurements of 189 proteins in blood from 600 subjects (PM 12/10/2010). There's been little research, however, on an autoantibody-focused approach like Opko's.
In fact, Kodadek said, his team's decision to apply the approach to Alzheimer's was a bit of an accident. He chose to work on the disease because a colleague had ready access to samples – a choice that, looking back on it, he now somewhat regrets.
"To be honest, Alzheimer's was a lousy first choice [for testing the autoantibody approach], and in retrospect I wish I hadn't done it because there's no gold standard we can compare to until people die and we get an autopsy," he said. "It would have been much smarter on my part to pick something like colon cancer where we could have compared to colonoscopy results, which are really quite definitive."
Kodadek is now applying the approach to other diseases including pancreatic and lung cancer. Opko, which licensed the technology in the fall of 2009, is looking to hire around 25 new employees to speed the research along, he said. It hopes to have beta tests for both diseases by the end of the year.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.