NEW YORK – After several years marketing direct-to-consumer, microbiome wellness testing, Bellevue, Washington-based Viome is taking its first step into the clinical diagnostics market, announcing this month that it is developing a saliva-based oral and throat cancer detection test that has received Breakthrough Device Designation from the US Food and Drug Administration.
The predictor was developed using the company's metatranscriptomic technology, which analyzes messenger RNA that originates from human tissues and the microbes that colonize our various body systems.
The company's oral cancer discovery was informed by the same methods behind its commercial DTC wellness business, which markets vitamin supplements and dietary guidelines that it claims it can personalize by analyzing mRNA in an individual's stool and blood.
Naveen Jain, Viome's founder and CEO, said that oral and throat cancer detection is the first of several diagnostic applications that the company plans to pursue based on biomarkers and drug targets gleaned from its commercial testing database.
"Because we have a very large population of consumers on our platform, we are able to find many of these disease phenotypes through the rich metadata that we collect from them. We are able to do … the first step of the discovery process very fast and very cheaply, and using that, then we can get into the validation trials and independent clinical scenarios," said Guruduth Banavar, Viome's chief technology officer.
Jain added the firm has analyzed samples from over a quarter million people thus far. "We currently have over 30 different predictive biomarkers [and] we're going to be going after all of them," he said.
Operating in a gray area of supplements and dietary advice, omics-based wellness offerings have largely been able to skirt the FDA regulations that apply to diagnostics or prescription drugs.
But in this regulatory vacuum, leading microbiome researchers have still questioned whether there is sufficient utility evidence to support selling these types of "personalized" recommendations, even when the risk of harm is low.
In pivoting to diagnostics and therapeutics, Viome is entering a space where it will be required to demonstrate validity and utility for its products much more definitively. Its FDA Breakthrough Designation for oral cancer detection is a positive first indication that it could meet these demands.
"We were able to convince the [program] that this is much better than the standard of care [which] is only 72 percent sensitivity," said Banavar. "We were also able to do this in a reliable and consistent way, whereas the current practice is much more subjective. You essentially have a dentist who is checking the oral cavity just by visual and tactile methods, and they're going to decide whether something is wrong and whether to send you to the secondary care center and so forth. That's a very subjective, skill-based type of process which can go wrong."
The firm hasn't yet published peer-reviewed data on its oral cancer detection approach but Banavar said that a study describing the development and initial validation of the assay has been accepted for publication.
In a preprint of the paper already online, company investigators and academic collaborators describe their development of a machine learning classifier trained on data from Viome's metatranscriptomic analysis of saliva samples from oral cancer patients, normal controls, and individuals with premalignant lesions in their oral cavity.
"We were able to find a diagnostic biomarker, which [reflects] an interaction between the microbiome and the human genes. That molecular signature was able to get a specificity of almost 97 percent and about 90 percent sensitivity for oral squamous cell carcinoma," said Banavar.
More specifically, authors wrote that their classifier could identify oral premalignancies and cancers from healthy controls with an area under the receiver operating curve of 0.87.
When restricted only to average-risk oral cancers, detection improved to an AUC of over 0.9.
According to Banavar, the team has also been able to show that it can detect cancers in the throat with high sensitivity.
In the preprint, Viome describes how the assay could be used to screen high-risk populations — adults of either sex 50 years or older or those with a history of tobacco use. A positive result would trigger follow up — "either a detailed physical examination and/or a biopsy by an appropriate medical practitioner," the authors wrote, adding that the primary risk is the possibility of a false prediction. "In a situation where the device produces a false negative result, there is a chance that a case … could go undetected, but this risk is no greater than what exists under current standard of care," they added.
Dennis Lo, a professor of chemical pathology at the Chinese University of Hong Kong and an expert in noninvasive cancer detection, said in an email this week that the Viome preprint doesn't raise any major red flags in terms of the company's fundamental methodology or early results, but did say that variables associated with saliva collection — such as time of day, hydration status, and meals — have challenged other efforts and will have to be watched closely as the firm moves forward.
Jain said Viome is now working on a multi-center validation that it hopes can replicate its initial sensitivity results and extend its evidence for robustness and reproducibility in order to support FDA premarket approval.
"We are working very closely with FDA to essentially get to the right path of making it happen as quickly as possible," he said, as well as considering offering a laboratory-developed test in the interim.
Some of the most prominent companies pursuing noninvasive cancer detection tests have coalesced around DNA methylation as a necessary signal for their assays, either alone or in combination with other targets like DNA mutations, fragmentation patterns, RNA, and proteins.
According to Jain, there is biological rationale to view methylation and mRNA as representative of some of the same biological activities or alterations.
"If you think about the purpose of methylation being to essentially dampen down certain gene expression, by looking at the gene expression itself you get the methylation automatically. You also get the histones, so you can see what genes are overexpressed and which are underexpressed," he said. "By looking at the transcriptome, you are automatically taking into account all of the epigenetics, including methylation. And in addition to that, we are looking at microbial cell-free RNA as well, so we have both the microbial side and the human side."
Other promising technologies expected to enter the market for early detection of cancers in the mouth, nose, and throat include several blood-based approaches, one of them spearheaded by Lo, which analyze circulating cell-free DNA for tumor-associated sequences from Epstein-Barr and human papillomaviruses.
While Viome's approach does involve measurement of microbial gene expression, it isn't specific to viral tumors, Banavar said. "What we see is sort of the more general hallmarks of cancer, for example, apoptosis, or p53 pathways, interferon alpha, interferon gamma. These are all the sort of human cancer hallmarks that we were able to see, which are very well established in the cancer literature across many different cancers. And of course, on the microbial side, also, we were able to see pathways that are generally associated with carcinogenesis," he explained.
The ability to detect a broader slate of oral cancers rather than just viral-associated tumors could be attractive to clinicians. At an American Association for Cancer Research liquid biopsy conference last year, University of Chicago surgical oncologist Nishant Agrawal argued that the field would benefit from a combined viral and non-viral approach that could be applied across the spectrum of head and neck cancers — not only HPV-associated cancers but also other squamous cell tumors of the mouth and oral cavity, in which viral DNA plays no role.
Agnosticism toward particular pathogens also makes Viome hopeful that its platform could identify signatures for detection of other cancers, as well tools for disease detection and monitoring in other non-cancer settings.
"We started the company not as a gut microbiome company, but our fundamental belief was to understand the biochemical activities and the functions that are being performed by the gut microbiome and how those activities and the molecules from the microbiome actually interact with the host … and I think that if you look at the future of health care, mRNA is going to be probably the key," Jain said.
He said Viome is currently in active trials with Advent Health focused on metabolic disorders including nonalcoholic fatty liver disease and with Mayo Clinic on insomnia and obesity.
"We also have research going on in ankylosing spondylitis and in [colorectal cancer] … and we are currently looking at pancreatic cancer and bile cancer and have already announced ovarian cancer, so we're going to be really going after a very large set of targets," Jain added. Other future targets include mental health, as well as other metabolic and immune disorders.
Building on this data, the company also believes that its platform's ability to profile host-microbiome interaction on the transcriptome level can also identify targets for new therapeutics, with two vaccines already in early development — one for colorectal cancer and the other for autoimmune disease, according to Jain.
"We saw that [between] people who have colorectal cancer and people who don't have colorectal cancer, if you match them by age and gender you can [see that] a particular organism has 500-fold higher activity in people who have colorectal cancer ... even though the prevalence [of this organism] is almost identical in both cases," Jain said. "If we were simply looking at metagenomics, we would have completely missed that [because] only through the messenger RNA are you able to see the activity of what is going on."
Emmanuel Hanon, who recently left the vaccine research division of GlaxoSmithKline to join Viome as global head of research and development, further noted that "Microbes are associated with the onset of several chronic disorders, and with a vaccine, you can actually interact or interface between that specific connection of the microbe with the host."
"We have identified one specific bacterium that potentially could be involved with what we call autoimmunity [and] antigen mimicry, where the body reacting against the bacteria is actually at the same time reacting against a protein of the body and that's what [triggers] the onset of the autoimmune disorder," Hanon added. "The hope is that with a vaccine, you can actually prevent that specific event … provided you have identified the individuals that have a specific microbiome signature."
According to Jain, Viome also plans to continue its DTC wellness business as these ventures move forward. After a few years in that market, the company has not proven that its molecularly informed dietary or supplement recommendations improve their health over standard-of-care guidelines or placebo.
However, Banavar said that the firm is about to change that with new data showing improvements in scores that rate diabetes risk, depression symptoms, irritable bowel syndrome, and anxiety.
This data has not been peer-reviewed but has been published by the company as a preprint. Among limitations, authors highlight that the experiments conducted were single-arm interventional studies without control arms, albeit ones in which participants were blinded to the fact they were participating in an interventional study to minimize placebo effect. Subjects' adherence to their nutritional recommendations was also not controlled for in the examinations.