Scientists at the University of Georgia and China's Jilin University are developing a urine-based protein biomarker test for gastric cancer that they hope to offer as a direct-to-consumer diagnostic in the US and China.
Using gene expression data combined with structural analyses to determine the proteins most likely to be secreted in urine, the researchers identified endothelial lipase as a potential urinary marker for gastric cancer, using it to distinguish between 21 cases and 21 controls with sensitivity and specificity of roughly 90 percent.
The teams are now in the middle of a study testing the marker against 200 gastric cancer cases and 200 healthy controls, said Ying Xu, director of UGA's Institute of Bioinformatics and one of the leaders of the study. If the biomarker is successful in that study, the researchers plan to develop it as a DTC immunoassay for commercialization in the US and China, he told ProteoMonitor.
"If [the study] is successful we'll make a test kit similar to a pregnancy kit," he said. "People could do it like an annual physical. If you find something abnormal then you'll want to get your stomach checked."
Using Affymetrix GeneChip Human Exon 1.0 ST arrays, they analyzed tissue from gastric cancer patients and healthy controls looking for genes that were differentially expressed, identifying from this analysis 715 genes for further investigation.
They then narrowed down this pool by applying a bioinformatic analysis to determine which of the proteins coded for by these 715 genes would be secreted in urine. This left them with 71 candidate proteins that they further winnowed down to a group of six based on their differential expressions and potential functional relevance to gastric cancer. These six they quantified in the 42 case and control samples by Western blotting, identifying endothelial lipase as the most promising candidate.
The researchers have since developed an ELISA for the protein that they plan to use in the 400-sample validation study, Xu said.
A New Sample Source?
While much protein biomarker work has been done in blood and serum, urine has certain advantages that could make it an excellent sample source for biomarker research, Xu noted. In particular, urine's dynamic range is significantly smaller than blood — roughly four logs compared to 10 or 11 — which makes it a much simpler fluid to analyze. This is especially important for mass spec analyses, which have long had difficulty detecting and quantitating low-abundance protein markers in the presence of much higher abundance proteins in serum.
"The traditional thinking is that healthy kidneys will not let proteins pass through them," Xu said, explaining why proteomics researchers haven't concentrated more on urine as a sample source. "But now people have actually found that's not the case. Healthy kidneys will let some proteins pass through and get into urine; the quantity is just really small. So this is basically a new understanding — urine actually is probably just as good in terms of the information we're looking for, but you need to have more accurate techniques to capture [proteins in urine] because of their very low abundance."
Akhilesh Pandey, a professor at Johns Hopkins' Institute of Genetic Medicine, agreed that urine is a promising sample source for biomarker discovery, telling ProteoMonitor that because kidneys filter out the bulk of high-abundance proteins like albumin and immunoglobulins, urine is a much cleaner sample to work with. Using MRM assays developed on Agilent's 6490 triple quadrupole machine, his lab has detected proteins in urine at nanogram-per-milliliter levels without using any enrichment techniques, he said.
"You don't routinely see studies where unfractionated serum is introduced into a mass spectrometer and you get answers for anything other than the most abundant proteins," he said. "There are technologies like antibody enrichment, but those are not routine methods."
The relative simplicity of taking and working with urine also allows researchers to start with more sample, which can enable better detection of low-abundance proteins, Pandey noted.
"In serum, because of the [high] dynamic range, using 100 μL of unfractionated serum generally won't give you better answers than 10 μL because you're just adding too much background," he said. [In urine] we are happy with what we've seen where adding more material [gives] you enough analyte to detect."
[ pagebreak ]
In April, his group published a study in the Journal of Proteome Research characterizing the human urinary proteome. In total they identified 1,823 proteins in human urine, 671 of which had never been identified in urine before.
Given this success in identifying proteins in urine via mass spectrometry, Pandey questioned the usefulness of the bioinformatic approach taken by Xu's team to determine which proteins were most likely to appear in urine.
"My take on it is that you don't need to do the bioinformatics," he said. "If we find it [using mass spec], we can find it again. That means that particular analyte can now be monitored in the urine."
Xu said the researchers tried using mass spec for discovery, but were unable to measure changes in expression as closely as they wanted. They are now working with Norcross, Ga.-based RayBiotech on custom protein arrays that will let them investigate a larger number of proteins than the six they focused on in the initial research.
Preliminary data from 70 case and control pairs from the ongoing 400-subject study has reinforced the results of the first study, Xu said, and they plan to finish analyzing all the samples by the end of the summer. Additionally, they are looking at 50 samples from subjects with liver, lung, breast, and colon cancers to determine if endothelial lipase is specific to gastric cancer or if it might be a more general cancer marker. They also hope to add a second protein to improve the test's accuracy, Xu said.
Even if the test proves accurate enough, though, commercializing it as a DTC product will likely be a tricky process — at least if the history of DTC genetic testing is any guide.
After launching their first tests three years ago, DTC genomics firms maintained for several years an uneasy relationship with the US Food and Drug Administration, marketing their services largely as laboratory-developed tests over which the agency has traditionally exercised "enforcement discretion." In May 2010, however, in response to Pathway Genomics' plans to sell genomic sample-collection kits at retail stores Walgreens and CVS/Caremark, FDA sent letters to four major genomic testing companies instructing them to submit their products for review and clearance by the agency (PGx Reporter 06/16/2010).
Just over a month later, the Government Accountability Office issued the results of its own year-long investigation of DTC genomic testing, concluding that the companies' test results were "misleading and of little or no practical use to consumers" (PGx Reporter 07/28/2010). Since then, several firms, including Pathway and Navigenics, have changed their business models to require physician involvement in the testing process.
This March, the Molecular and Clinical Genetics Panel of the US Food & Drug Administration's Medical Devices Advisory Committee advised the agency that genetic tests that could potentially inform medical care should require the involvement of a doctor (PGx Reporter 03/09/2011).
Citing existing urine-based DTC products like pregnancy tests, which measure levels of the protein chorionic gonadotropin, and diabetes tests measuring ketone levels, Pandey said he saw no reason a DTC test like Xu envisions couldn't make it to market, however.
"Biomarkers got a bad rep, but now they're coming back," he said. "If we find markers that are specific and detectable in urine, I see no hindrance [to bringing tests to market]."
Xu himself acknowledged the challenges of getting a DTC protein biomarker test through FDA and said that given the difficulties involved, the researchers might first aim to commercialize in China any test that emerges from their work.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.