NEW YORK (GenomeWeb) – Protea Biosciences, a mass spectrometry instrument and services company, is looking to move into the clinical diagnostics space. With several collaborations with academic cancer researchers, the firm is on its way to developing mass spec imaging-based tests, potentially for melanoma.
Protea scientists told GenomeWeb that they're developing a machine learning algorithm that can detect and distinguish between the fingerprints of benign or cancerous tissue. While they haven't yet found specific biomarkers that would be the basis of a regulated clinical diagnostic, remarkably, the method doesn't really need one.
By analyzing mass spectra from both cancerous and benign tissue samples, "[the algorithm] learns what the fingerprint of a malignant lesion looks like and what a benign [lesion] looks like," Erin Seely, head of clinical research at Protea, told GenomeWeb. "Some algorithms use a distinct small subset of peaks. This is using a full mass spec fingerprint," she said.
As proof of concept, Protea is working with Yale pathologist Rossitza Lazova to train the algorithm to discern between benign moles and melanoma, using matrix-assisted laser desorption/ionization (MALDI) mass spec imaging.
"We're doing mass spec imaging of proteins or protein-derived peptides and taking thin sections of tissue like you would for histology," Seely explained. The method allows the researchers to map the spatial location of molecules in the tissue and characterize the samples based on the complete fingerprint.
"It's removing any bias you might have in looking optically at the tissue," Haddon Goodman, Protea's VP of Global Imaging, said. He cautioned that the firm isn't planning to build a first-line melanoma diagnostic test, but rather a confidence-builder in cases of ambiguity. "Ideally, we see ourselves as the company for when you go for your dermatology visit, if anything needs a biopsy, it would be sent to us to perform routine biomarker assessment," he said.
"The approach is reasonable," Richard Yost, a clinical mass spec expert at the University of Florida, who is not affiliated with Protea, told GenomeWeb. "It's always a challenge to come up with meaningful biomarkers. Our old image [of cancer] was that there was normal tissue, which was all alike, and tumor tissue would also all be alike, but [tumors] tend to be very diverse." He also noted that MALDI mass spec requires the tissue to be removed from the patient, while his and other labs are working on diagnostics in real time, diagnosing melanoma in situ.
But the most appealing aspect to Protea's method is that it could be applied to other solid tumors, or even other diseases, Goodman said.
Protea began a concerted push into clinical diagnostics about a year and a half ago, Goodman said. Seely had joined the firm from Vanderbilt University to pursue clinical applications of mass spec and brought with her the collaboration with Lazova, the Yale pathologist. Around the same time, Protea announced collaborations with Dana-Farber Cancer Institute and Memorial Sloan Kettering Cancer Center focused on lung cancer.
The Yale collaboration involves looking at skin tissue sections stained for histological analysis. "Lazova is choosing the area she wants to analyze and we're going in and targeting just those specific regions corresponding to the melanocytic lesions in these skin samples, collecting spectra from them," Seely said. So far, the collaboration has yielded a study of 25 well-characterized malignant samples and 25 benign ones, which have been used to generate the algorithm, and a smaller set of samples to validate it.
"It's pattern matching, you don't have to have a specific list of targets," Seely said. "Whereas in some other approaches you need to know exactly what protein to target, this is just a fingerprint type of analysis. Obviously when you start looking into treatment, knowing what those [proteins] are becomes important, but for diagnostics, the algorithm just needs a pattern."
The collaboration has reached the point where the scientists are evaluating how well the algorithm works in more ambiguous cases and comparing the mass spec results to long-term clinical follow up, Seely said.
"Our goal is to then push this into the clinical diagnostics space and develop a laboratory test based on the work we're doing here," Goodman said.
If Protea is going to successfully develop a clinical test, it will eventually need to zero in on specific analytes. "I think absolutely we can discover new biomarkers," Goodman said. With mass spec, he said the firm will be able to determine the specific upregulated proteins that are causing melanoma, or other conditions.
With the Yale collaboration, Protea is set up to capitalize on any biomarkers discovered. Any intellectual property will be jointly owned and Protea has first rights to license biomarkers and develop and offer a test based on them. Goodman declined to disclose whether the company is close to securing any IP. According to the US Patent and Trademark Office, neither Protea nor Lazova has patented any target molecules.
The market for resolving ambiguous skin biopsies is somewhat limited. According to the American Cancer Society, approximately 75,000 melanomas will be diagnosed this year mostly by pathologists' visual abilities. However, disagreement between individuals and inconclusive results are common.
For Protea, it's an accessible first step for a company that doesn't even yet have a CLIA-certified laboratory. "We're at the point where the data is looking quite promising for us," Goodman said. "Melanoma is our focus right now but we do want to investigate other [cancers] as well."
First, the firm will have to invest more into its lab infrastructure and collaborations to find specific markers of interest. Goodman declined to comment on how Protea would raise more money to kick off its clinical diagnostics business. Protea trades on the Nasdaq over the counter markets under ticker symbol "PRGB." Other areas of interest for the company are to expand the Dana Farber and Memorial Sloan Kettering collaborations, which haven't progressed as far as the Yale collaboration.
"I think that kind of goes hand in hand with our funding strategy," Goodman said. "To move forward we do need investment to our infrastructure to be able to provide this as a service."