Researchers at the University of Medicine and Dentistry of New Jersey have developed an array-based diagnostic for Alzheimer's disease, and are planning to further develop and commercialize the test via a new company, Durin Technologies.
The method uses Life Technologies' Invitrogen ProtoArray v5.0 Human Protein Microarrays to measure certain autoantibodies in serum that the researchers believe to be associated with neurodegenerative disease. In a discovery and testing study, published in PLoS this month, the group established a signature of 10 antibodies that distinguished Alzheimer's disease from healthy controls with 96 percent sensitivity.
Durin Technologies was founded last year with an initial investment of $175,000 from the Foundation Venture Capital Group to develop a blood-based Alzheimer's diagnostic. In subsequent research, the team has also applied their method to Parkinson's disease, achieving similar accuracy, Robert Nagele, a UMDNJ researcher and the founder of Durin Technologies told BioArray News.
According to Nagele, attempts by other researchers to identify a blood-based genomic marker for accurate diagnosis of Alzheimer's have found limited success so far.
He and his colleagues became interested in autoantibodies, he said, after finding that brain-targeting autoantibodies are able to cross the blood brain barrier in AD. The researchers wondered if it would be possible to detect these antibodies while still in the blood.
"It made the question more interesting," he said. "Binding selectively means that the antibodies… are targeting those neurons… That means that everyone, all those Alzheimer's patients we were looking at. Every one of them had to be walking around with antibodies that were originally in their blood that would bind to brain."
To identify autoantibodies that might distinguish AD from healthy controls or other diseases, the researchers used Invitrogen protein microarrays containing 9,486 antigens. "We looked around for the very best arrays we could find and it looked like [Life Tech] had the best arrays," Nagele said.
Serum samples from 50 Alzheimer's patients and 40 non-disease controls were randomly split into a training or a testing set, providing 25 unique AD profiles and 20 healthy ones for each group.
ProtoArray Prospector analysis software identified 451 autoantibodies with a "significantly higher prevalence in the AD group," the authors wrote, and the researchers chose the ten markers with the largest differential between AD and healthy samples as their signature.
Nagele said the softwares involved made the process of marker selection, training, and testing fairly easy.
"These are sort of like intelligent softwares in a way, where you can say to them, 'Learn this, this is what Alzheimer's people look like,'" he said.
"So once we decided on our ten biomarkers, we took the other half of the sample, [and] we asked, 'How well can you distinguish those?' That's where it makes or breaks a diagnostic," he said.
"And it worked wonderfully," Nagele added. "We were very happy."
The researchers were able to distinguish AD samples in the testing set with positive predictive value of 100 percent, a 6 percent prediction error, and a negative prediction value of 87 percent.
Used on a combination of the training and testing samples, 90 samples in all, the ten antibody markers divided AD and healthy samples with 96 percent sensitivity and 92.5 percent specificity.
The researchers also tested their signature against samples with breast cancer and Parkinson's disease to see how well they could distinguish AD from these other conditions. "The purpose of comparing Parkinson's was we decided a true test of the diagnostic would be to distinguish it from another disease that is very similar," Nagele said. "To shoot ourselves in the foot, we picked Parkinson's because we know there are a lot of commonalities in the pathology. So we figured if our test was really good, it should be able to distinguish those two diseases and it did really well" — with a 90 percent sensitivity and approximately 80 percent specificity.
"The only reason those numbers were probably lower is that many Alzheimer's people also have Parkinson's," Nagele said.
[ pagebreak ]
Nagele said the Alzheimer's patients in the AD group were "pretty well distributed" in terms of the stage or extent of their disease, with Mini-Mental State Examination scores between two and 24. The authors reported in their paper that the signature distinguished samples with over 90 percent accuracy in all subgroups, from early- to late-stage AD, and wrote that this suggests the approach may also be useful for early detection. But, according to Nagele, the group isn't planning to go in that direction, at least not yet. "Our purpose was to develop a diagnostic, not so much early detection at this point," he said.
Since completing this study, the group has moved on to apply the same approach to Parkinson's disease alone, for which they were able to achieve even higher diagnostic efficiency, according to Nagele.
"When we did the assay… because the two [diseases] are very similar, we would then expect to see some common targets, and sure enough, when we did the Parkinson's diagnostic, there was some overlap," he said. "That was a very nice confirmation that the approach is probably correct."
Also, he said, "Parkinson's primarily affects a small area of the brain called susbtantia nigra, whereas Alzheimer's is all encompassing. We would expect then that the biomarkers should be even more specific for a disease that affects only a smaller region. And sure enough we were able to get really good results… and trim our biomarkers down to five."
According to Nagele, his team has completed that study and submitted it for publication. He added that the group has also recently applied their method to cancer diagnosis, which he said also "seems to be working pretty well."
Path to Market
Nagele said that Durin Technologies is planning to explore a number of avenues for its autoantibody approach as they simultaneously move forward to commercialize the Alzheimer's diagnostic. "The thing to point out on this is that it's a diagnostic that is based on a common mechanism of diseases," he said. "We think that the antibodies that we're detecting… is the result of sort of a debris-clearing mechanism [associated with cell death.]"
"So we're hopeful, of course we don’t know for sure, but for at least so far we're batting 1.000, [which] would seem to predict that this diagnostic approach should be applicable to a huge number of diseases," he said.
Nagele said the company is "ready to approach" the US Food and Drug Administration.
"We need to find out what they would require to approve a diagnostic of this sort, how many samples need to be done, and all those things you have to go through for FDA approval," Nagele said. "But we feel we can do them really, really fast," he said. "Everything we've done so far has been in one year."
He said the group plans to stick with the Invitrogen ProtoArray platform, though he said there are "other platforms that could be explored," without elaborating.
Nagele said he wasn't aware of any blood-based tests for Alzheimer's that had advanced as far as Durin's method, though one was recently released in Europe: Exonhit Therapeutics earlier this year secured a CE-IVD Mark for AclarusDx, a blood-based test for Alzheimer’s disease based on a signature of around 130 genes that is run on the Affymetrix GeneChip platform. It recently tapped Almac Diagnostics to offer the test to European customers through Almac's UK reference laboratory (BAN 7/5/2011).
Nagele said he hadn't heard of the test and couldn't give a comparison to Durin's method, but said in an email "it all boils down to cost, degree of noninvasiveness and, of course, accuracy." According to a March 2011 press release from Exonhit, AclarusDx has a sensitivity of 81 percent, and a specificity of 67 percent, established in a French multicenter blind validation.
Nagele also mentioned another team from the Scripps Research Institute looking at autoantibodies as a potential AD diagnostic, who he said has taken a different array-based approach.
These researchers, led by Thomas Kodadek, used microarrays of chemicals called peptoids — "little sprigs of synthetic chemical that have a three-dimensional shape that antibodies might recognize," Nagele said — to create antibody signatures that would distinguish AD patients. Their work was published this January in Cell.
Nagele said called this approach "really beautiful," but said that there is "a big difference between the two results."
"In [the Scripps] case… there could be a very prominent potentially diagnostic antibody in the blood and if he doesn't just happen to have the shape of the chemical on the surface, it goes undetected," he said.
Additionally, Nagele said, using protein arrays allows for the direct identification of antibodies and their targets, which could help identify potential new treatments.
"In our case… as soon as we see the antibody binding, we know it's antibody to this protein," said Nagele. "So we can actually give you their names. The nice thing about [that] is if you see a couple names crop up in the disease, it points you to a molecular pathway that might be disease relevant," he said. "And if a pharmaceutical company is looking for some promising targets for therapy, we might be able to give them a list."
Right now, Nagele said, the group hasn't tried to pin down any pathway information for the antibodies in their signature.
"We were not really concerned, at least initially, about the function of the protein targets," he said. "We just sort of said, let's not be biased by looking for proteins we think should be there, let's let the assay method tell us which proteins are important," he added. "So we decided to go that route and I think that’s a wise route to go."
Now, the company is making some critical decisions about how to move forward with the Alzheimer's test, and with a potential collection of diagnostics based on their autoantibody, protein array approach.
The pressing question is how to move forward with commercialization, specifically whether to seek approval of various diagnostics separately, or try to group them as a kit. "These are all considerations we have to think about," Nagele said.
Also, he added, "do we even want to get involved with that? Whereas you have all these companies that already have the infrastructure in place to commercially produce diagnostic kits and take it through FDA approval… We may just want to sell it off or hand it off to them. We haven't decided yet."
He said Durin hasn't leaned one way or another on the question yet. If any companies in the pharmaceutical field express interest, he said, that would "probably push the company to make that decision."
Have topics you'd like to see covered in BioArray News? Contact the editor at mashford [at] genomeweb [.] com.