NEW YORK (GenomeWeb) – In a study published this week in Proceedings of the National Academy of Sciences, researchers at Arizona State University's Biodesign Institute demonstrated the ability of their immunosignaturing diagnostics platform to detect a wide range of diseases, including early stage cancers.
According to Phillip Stafford, a Biodesign researcher and first author on the paper, the studies suggest the platform's potential usefulness as a broad screening tool for early detection of disease in healthy populations.
Immunosignaturing uses random-sequence peptide microarrays to capture antibodies in patient blood samples. Based on the levels and patterns of antibody binding, researchers build antibody expression profiles that can then be correlated with various disease states. The technique has potential advantages over conventional protein biomarker-based tests, Stafford told ProteoMonitor, in that it detects broad immune responses, which may provide a more comprehensive picture of disease states than a smaller set of markers; and the immune cell replication involved in immune response provides a natural amplification of the signal, making it possible to detect disease earlier than with conventional protein markers.
The technique has been developed over the last decade through efforts led by Stephen Johnston, co-director of Biodesign Institute's Center for Innovations in Medicine. In 2010, Johnston, along with his Biodesign colleagues Neal Woodbury and John Rajasekaran, launched the company Healthtell to commercialize diagnostics based on the immunosignaturing technique. Last year the company raised $4 million to support the development of tests aimed at early detection of lung, breast, prostate, and colorectal cancer.
In the PNAS paper, the Biodesign researchers examined the ability of the platform to simultaneously distinguish between controls and patients suffering from a variety of diseases. In the first stage of the study, they looked at 120 total samples: 20 non-cancer controls and 20 samples from each of five cancer cohorts – breast, lung, pancreatic, gliablastoma, and multiple myeloma – collected at five different centers.
Using their immunosignaturing platform, they were able to classify the different patient samples by their disease state with 95 percent accuracy.
They then moved on to a larger cohort, looking at historical samples from 1,516 individuals across 14 different disease cohorts, including 249 healthy controls, and multiple stages of brain cancer, lung cancer, ovarian cancer, pancreatic cancer, pancreatitis, Ewing sarcoma, valley fever, and multiple myeloma.
Applying the platform to this larger cohort, they found that they were able to classify these patients with an average accuracy of 98 percent.
Interestingly, from an early detection perspective, Stafford noted, early stage disease was often easier for the platform to identify than later stage cases, as it presented more uniform immunosignatures across patients.
For instance, "pancreatitis and early forms of pancreatic cancer were actually more immunogenic than late-stage pancreatic cancer, and the same was true for breast cancer – the earlier you went the more likely you were to find a more homogenous signature that was easier to pick out across a large number of people," he said.
"Immunologically, it might be that your tumor is following multiple pathways, and by the time you get to a late stage there might be six or seven different pathways you followed, but there may be only two that you followed to initiate that first tumor," he suggested.
While in the first days of proteomics, many researchers envisioned discovering protein panels capable of screening for early stages of diseases like cancer in healthy populations, the field – somewhat chastened by early failures – has in recent years largely shifted to less ambitious goals like identifying markers useful for testing high-risk populations or monitoring patients for recurrence. Stafford said, however, that such screening of health populations was his research group's ultimate goal for the platform.
To that end, the researchers are currently working under a $30.7 million grant from the US Department of Defense to analyze 500 cases and more than 800 controls from the National Cancer Institute's Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial to see if the platform can detect cancer in apparently healthy subjects one, two, and three years before their ultimate cancer diagnoses.
When screening healthy populations for relatively rare conditions like cancer, extremely high test specificity is required to avoid large numbers of false positives. To provide this level of specificity, the Biodesign researchers have shifted from the platform they used in the PNAS study, which consisted of arrays containing 10,000 synthetic peptides, to larger arrays consisting of 330,000 peptides.
"I think we've pretty much tapped out the performance of the [10,000-peptide] arrays," Stafford said, noting that while an array of that size provided the sensitivity the researchers needed, it didn't have the specificity, particularly if they wanted to test simultaneously for a larger number of diseases than the 14 they looked at in their recent work.
The larger array provides "very strong distinctions between cases and controls" even for cases that "looked a little murky when you ran them on the [smaller array]," Stafford said, adding that data quality appears to scale roughly linearly with the number of peptides on the array.
The DoD grant, which the researchers received in 2012, also funds research into using the platform for early detection of infectious disease. The military, Stafford says, "is interested in [identifying infected soldiers] days before they have symptoms so they can keep them from going out with the rest of the platoon."
In the initial stage of this work, Stafford and his colleagues succeeded in using the platform to distinguish between controls and subjects infected with eight different diseases, including dengue, West Nile virus, and malaria.
Another potentially important aspect of the platform in regards to the Biodesign team's ultimate aim to use it to screen healthy populations is its compatibility with dried blood spot samples.
Because these samples are easy to collect and durable enough to be sent through the mail, they are potentially ideal for regular screening tests. As Johnston described the platform's potential future in a 2011 interview with ProteoMonitor, "our big idea is that this is what healthy people will be sending in on a regular basis for diagnostic purposes. The idea is that people could send a drop of blood through the mail and we could do the [immunosignature] on that."
According to Stafford, the group's initial work suggesting the platform is compatible with dried blood spots has held up to further investigation. "You can just crush it up in buffer and it works as well as a [serum sample]," he said.