NEW YORK – At-home gut microbiome testing is a burgeoning branch of the overall genetic testing field, but consumers may be buying tests that offer little useful information at best and disinformation at worst.
As genetic testing in general gains broader consumer acceptance, and the link between gut microbe composition and disease becomes clearer, an industry has grown up around offering consumers personalized health advice based on the metagenomic analysis of their microbiomes.
While experts caution that such direct-to-consumer (DTC) microbiome tests often go beyond what the research can support, the large databases that some of these companies generate are being leveraged to explore further diagnostic possibilities of the microbiome, such as cancer diagnosis and prognosis, via peer-reviewed research.
Understanding what actionable information consumers can expect from the gut microbiome will only grow in importance as the field expands, experts also note.
The microbiome testing field can be broadly divided into targeted tests meant to diagnose either specific or a narrow range of disorders, and those meant to assess an individual's overall wellness, although the latter category does sometimes provide information on specific microbiome-associated conditions.
Firms such as Doctor's Data and Genova, for instance, have marketed several physician-ordered tests. The products these companies offer tend to be highly targeted toward infectious microorganisms such as those responsible for meningitis and pneumonia, and for dysbiosis specific to gastrointestinal disorders such as inflammatory bowel disease (IBD) and the related irritable bowel syndrome (IBS).
Also in the targeted testing area of the market, but further from commercialization, are firms such as Enterome, Metabiomics, and Origin Sciences with products in various stages of clinical testing that target progressive liver diseases such as nonalcoholic fatty liver disease and colon cancer.
Meanwhile, companies such as Viome, Biohm, and Sun Genomics have a wider scope, coupling genomic technologies such as shotgun sequencing to computational algorithms trained to analyze genomic and/or transcriptomic data in an effort to measure a broader set of health-related markers.
Viome's DTC Gut Intelligence Test, for instance, assays for all microbes present in a sample and builds an overall profile of an individual's gut health based on that.
This is where the science gets murky, according to experts.
"Microbiomes are somewhat unique, akin to a fingerprint," said Leigh Frame, program director of integrative medicine at George Washington University. "Your version of a healthy microbiome is probably different than my version of a healthy microbiome. And that makes it really difficult to figure out what a healthy microbiome looks like when there are many, many, many healthy microbiomes."
The majority of microbiome tests, whether DTC or physician-ordered, use metagenomic sequencing and informatics techniques to assess microbial taxonomy, or which bugs are present and in which proportion in any given sample. While the majority of microbiome samples come from stool, some developers have also begun investigating saliva and blood.
"That only gives us so much information," Frame said, "and frankly, misinformation."
For example, the microbiome exists in a constant state of flux, responding to a wide variety of nutritional and environmental stimuli. Taking a sample at any given time doesn't guarantee that that sample is or will remain representative of its host's microbiome at another time.
"I will guarantee my microbial community yesterday at lunchtime looked very different to my microbial community today," said Jack Gilbert, a professor of pediatrics at the University of California, San Diego, whose research has long focused on the microbiome. "If I'm looking for a biomarker to predict disease or predict my personalized nutrition, [then] I need to understand when that biomarker might show up. And that's a pretty hard thing to predict if you don't [measure] it regularly."
The degree to which these variations affect how our microbiomes might impact our health or how we metabolize different foods and medicines remains far from completely understood. Even diseases with a well-established microbiome contribution, such as IBD, are very heterogeneous, presenting differently in each person, which creates opportunities for individualized therapy.
"That's where I think the microbiome has a lot of promise," Frame said, commenting that people with conditions such as IBD often respond differently to treatments.
"Right now, we don't understand why, but it's probably that they have a different microbiome," she said. "We're giving them the same therapy, but they're reacting differently to it."
Sunny Jain, founder and CEO of San Diego-based Sun Genomics, which markets the DTC gut health test Floré, made IBS his company's main focus because it is one of the microbiome-related disorders that affects most people.
Sun Genomics sponsored a study into the gut microbiome composition of 490 individuals with IBS (and 122 healthy controls) and identified four subtypes that were defined both by unique microbiome compositions and sets of symptoms, research that was published in 2021 in the journal mSystems.
Company researchers used participants' gut microbe profiles to design individual probiotic formulas and followed them for approximately a year, measuring changes in gut microbe diversity across three time points. Although the researchers noted no significant change in overall microbial diversity among the participants, they did observe significant changes in IBS subtype-associated species and related biological pathways.
"We were the first group to show the longitudinal benefit of being on a precision synbiotic [a mixture of probiotics and prebiotics] and how that would shift your microbiome over time," Jain said.
"When we started the company, nobody thought you could modulate the microbiome," he continued. "We're one of the first companies to show that yes, you can modulate and change the microbiome through supplements, diet, and other interventions, and be healthier."
In fact, one of the obstacles to the wider adoption of microbiome-based diagnostics in general has been the lack of associated therapeutic options. While there is a dizzying array of pro- and prebiotic supplements available, these are considered food products rather than medicines and are therefore unregulated by bodies such as the US Food and Drug Administration.
This is beginning to change, however. The FDA recently approved two fecal microbial transplant pills: Rebyota from Ferring Pharmaceuticals and Vowst from Seres Therapeutics. Both treatments consist of bacterial spores purified from healthy donor feces and are used to treat Clostridium difficile infections.
Bellevue, Washington-based Viome also offers consumers personalized supplements and dietary advice based on an analysis of patient samples collected via at-home kits. In contrast to competitors, however, it focuses on microbial gene expression patterns instead of taxonomical distributions. Rather than asking which species are present in a sample, that is, Viome focuses more on what they are doing.
The idea behind this strategy is that bacteria and fungi can enter dormant states, meaning that not all microbes present in a sample are meaningfully active; not all bacteria of the same species or strain may express pathogenic genes; and biological pathways associated with expression patterns may provide information on an individual's likely response to therapy, as in the case of cancer treatments, as well as their nutritional needs.
"There are a lot of people with P. gingivalis in their oral biome," said Guruduth Banavar, Viome's cofounder and chief technology officer, "but there is a particular gene in the P. gingivalis genome that is considered pretty bad if expressed, but that is not expressed in most people."
Because of this, Banavar explained, analyzing the gene activity of the microbiome should be more informative that simply identifying the species present.
GWU's Frame agreed that this approach to microbiome analysis in principle holds some promise, saying that "just because that single bug isn't there doesn't mean the function of that bug isn't there. There's a lot of redundancy throughout the ecosystem."
Banavar said that last year the company analyzed the microbiomes of two cohorts of people, one from the US and another from Japan, as part of a study that has not yet been published. In terms of species abundance, he said, the two cohorts were "quite different, with only a small overlap."
Looking at the functions of expressed genes, however, Banavar said that there was a "huge overlap between the two populations." Essentially, the various microbiomes were doing the same work, regardless of which species were present in which proportions.
Assays such as Viome's Full Body Intelligence Test use this data to identify microbial metabolic pathways suggestive of how well a person's microbiome can digest the nutrients found in certain foods.
A result that finds a deficiency in a person's ability to eliminate oxalates, for instance, would trigger a recommendation that that individual avoid high-oxalate foods such as spinach.
"If my activity for eliminating oxalate is not good," Banavar said, "then if I eat too much spinach [for example], I'm going to create kidney stones from the oxalates that accumulate."
Gilbert agrees that a better understanding of the functional genes at play in the microbiome holds strong potential for biomarkers of numerous conditions as well as therapeutic targets.
"Mechanistically it's very, very important that we understand what functional genes are there so that we can understand what might be driving the [biological] signatures that we see," he said.
To Gilbert and other experts, however, the current state of knowledge in that arena is insufficient for making claims as detailed as which specific foods to eat or avoid. A large part of why it may be insufficient is that relatively little data exists on how individual microbiomes change over time and in response to the variety of daily dietary and environmental stimuli a person regularly experiences. What data does exist, Gilbert said, comes from small studies that may be representative of larger populations.
The few small studies that do serial measurements generally look at taxonomical changes rather than functional ones. Less is known about how gene function might change relative to taxonomical changes, Gilbert said, adding that there are currently a wide range of opinions on that matter within the microbiome field.
"Some of my colleagues will say that it doesn't change very much, and others will say well, the abundant things don't change very much, but some of the rare things do," Gilbert said. "And other people will say, well, that's not really important because what you really need to know is what's being actively transcribed, not just what's there."
Furthermore, Frame explained that in most cases she is aware of, personalized dietary advice rarely outperforms general dietary advice. Saying that one should avoid large quantities of spinach, for example, essentially amounts to a rephrasing of "eat a balanced diet," which is a cheaper and more easily obtained recommendation than any offered as a result of metagenomic testing.
"I think the [companies] that tell you they can personalize a diet to you are going too far," Frame said. "Because we don't have that level of clarity yet. But if they're saying that we can make recommendations on how to shift your microbiome based on what we see, then that is, I think, an accurate statement."
Ginger Hultin, a registered dietitian nutritionist and author of several books on the effects of diet on health, concurred, saying that except for some specific cases, telling someone which specific food items to eat or to avoid "makes no sense at all."
One such relatively well-known case of where science does back up a specific food recommendation concerns people who harbor certain bacteria that convert trimethylamine (TMA) into trimethylamine-N-oxide (TMAO), a known independent risk factor for heart disease.
"Eating less red meat if you have TMA-producing bacteria or eating more plants is a great suggestion," Hultin said. "[And] to say 'eat more broccoli' doesn't seem immediately harmful. But I think the problem is that then people are falsely stating that this is because of my microbiome. [That's] based on faulty research and that is problematic, I think."
Although Hultin is unaware of specific cases where personalized dietary advice resulted in serious harm, she echoed Gilbert's concern that such an outcome shouldn't be ruled out. Especially problematic, Hultin said, are recommendations about which foods to avoid, as this can trigger or aggravate a condition called orthorexia nervosa, in which one carries the idea of eating healthy too far. To people with that condition, as well as other disordered eating conditions such as anorexia nervosa, being told to avoid certain foods by a source of authority could worsen one's fear of food, leading to unnecessarily reducing overall consumption or cutting certain foods out of one's diet.
"Our society has a problem with that," Hultin said. "I fear that these direct-to-consumer tests that aren't well thought out or aren't translating the research correctly could be really harmful to people's actual mental health and relationship to food."
Fortunately, that problem so far appears to remain theoretical.
The cancer connection?
The other side of Viome's business consists of leveraging the revenue brought in by test kit sales and wellness services subscriptions to fund research in microbiome-based cancer diagnosis and drug response, as metabolites produced by the human microbiome alter drug metabolism and by extension efficacy, which likely explains some of the variance in individual drug response and is an area of active research.
In the cancer diagnosis arena, Viome has taken a particular interest in developing a microbial diagnostic assay for the early detection of oral cancers, noting that this indication currently lacks good early diagnostic tools, while multiple studies have uncovered strong correlations between oral cancers and changes in the microbiome. In 2021, the company and academic collaborators published a study in NPJ Genomic Medicine identifying an oral cancer-associated metatranscriptomic classifier that incorporated both taxonomical and functional information.
The firm is now conducting a clinical study of that classifier.
Viome also plans to explore the role of nutrition and the gut microbiome in ovarian cancer patients post-treatment in order to develop personalized food recommendations to support and restore the immune system, appetite, and sleep patterns that are often weakened by cancer treatment.
That a company based largely on nutrigenomics should foray into cancer diagnostics may not be as strange as it first sounds. Numerous studies have found associations between diet and cancer risk and progression. Salmonella enterica subspecies S. typhi, for instance, has been associated with gallbladder cancer, Streptococcus bovis with colon cancer, and Chlamydia pneumoniae with lung cancer.
Nicholas Chia, associate director of the Mayo Clinic Microbiome Program, recounts that the previous director, Heidi Nelson, once told him that patients came to her all the time after they received a colon cancer diagnosis to ask what they should eat.
It was a question that neither Nelson nor Chia had an answer to, Chia recalled. And despite leading cutting-edge research into how to leverage the microbiome for individualized therapies, point-of-care devices, and cancer risk prediction, he's still looking for one.
"I wouldn't say we have firm answers yet," he said, "but … we already know that some cancers, for example, colorectal cancers, can be exacerbated by some metabolites, and other cancers in animal models will actually be the opposite. So there is definitely [room for] personalization there, but we just don't have enough evidence to make use of that right now."
Greg Sepich-Poore, cofounder and chief analytics officer of microbiome analytics firm Micronoma, said that many microbes may be complicit in cancer development, although he doubted that many microbes are "bona fide" carcinogens.
"The tumor microenvironment itself can enable microbial seeding and growth of certain species, and when present, those microbes can accelerate carcinogenesis," he said.
Pancreatic cancer, he continued, is one of the better studied examples of this.
"Microbes are enriched roughly 1,000-fold in malignant pancreas tissue compared to nonmalignant counterparts," he explained, "and after seeding the tumor, they can negatively modulate anti-tumor immunity or spur pro-inflammatory pathways."
Furthermore, while the proportion of intratumoral microbes that are involved in cancer processes versus simply "passengers" remains unknown, even the passenger microbes may make suitable biomarkers, as long as they are consistently present.
"Categorizing which microbes are causal versus complicit versus passengers is an extremely difficult if not impossible task, since the answer can change depending on the particular pathophysiological circumstances, body site, and/or time point," Sepich-Poore said via email, "in the same way that Staphylococcus species on the skin can be considered normal 'passengers' on healthy skin but causal 'pathogens' in endocarditis, which could also make them complicit agents."
Sepich-Poore acknowledged that microbial functional variation can enable cancer diagnosis and type discrimination, but cautioned that there is much nuance to this and that past studies have shown limited performance in predicting immunotherapy response from fecal sample-derived microbes.
One recent meta-analysis of four prior shotgun metagenomic studies of the response of metastatic melanoma patients to immune checkpoint inhibitors, for example, "did not find anything remarkable" with respect to the functional consequences of differential microbial gene expression, nor associated metabolic pathways.
Sepich-Poore, Banavar, and Gilbert all called for more meta-analyses, as well as more and larger studies generally, of microbiome-based diagnostic uses.
"The gold standard for the best quality research is a randomized, controlled clinical trial," Banavar said.
What sets clinical trials apart in the microbiome space, he explained, largely comes down to the size of the datasets involved.
"The data sets required in order to make the right discoveries are generally much larger," he said, "because the number of features that you have to deal with is much larger."
Sandrine Miller-Montgomery, CEO and cofounder of Micronoma, concurred, saying "if you look at circulating tumor DNA, we have only 23,000 human genes to mine for [a] mutation or a [genetic] profile. Now, if you look at the circulating microbial DNA, we suddenly have about 2 million genes."
Despite increasing experimental complexity, Miller-Montgomery explained that the microbiome's expanded genetic library is also an advantage, as the number of circulating microbial cells far exceeds that of cancer cells.
For every circulating cancer cell, she said, there may be 10 to 500 viruses, bacteria, and fungi.
"When it comes specifically to the early stage [of cancer], you have a ratio that favors microbial DNA," she said.
This logic underlies Micronoma's Oncobiota platform for early detection of multiple cancers via microbial DNA. The company has begun with lung cancers, for which it claims it has found a specific microbial footprint.
"What we discovered is that if you look at the cancer tissue that's in the lung, the microbes you find in it are going to be very different from the ones you find in breast cancer tissue, for example, or pancreatic cancer tissues," Miller-Montgomery said.
Despite the strides that Micronoma has been making into microbiome-based cancer diagnostics, its research also serves as a reminder of just how unsettled the field is with respect to the interplay between cancer and the microbiome.
A recent study, currently available as a preprint, called into question the validity of the results of a 2020 study by Micronoma that found widespread evidence of cancer-specific microbial signatures. This study is the basis of the company's technology, underlying its OncobiotaLung assay, which garnered breakthrough device designation from the FDA in January.
The authors of the preprint claim to have found significant contamination of microbial DNA by human sequences and that the computational method used in the study created artificial signatures for each cancer.
Micronoma refuted the findings of the more recent study in a GitHub post.
The 2020 study has been cited in many other cancer-microbiome research papers since then, including Viome's 2021 oral cancer study, which cited it as evidence supporting the company's reasons for conducting that study.
Viome did not respond to questions about how the ongoing controversy might affect its own plans.
Signal-to-noise for the consumer
With dozens of companies involved in every aspect of microbiome research, from therapeutics and diagnostics to nutraceuticals and diet/lifestyle advice, and some of those companies straddling the lines between wellness products and rigorous, peer-reviewed research, some experts worry that consumers are not effectively equipped to make sense of what our microbiome can and can't tell us, and of which information might be actionable.
"It takes a lot of effort on the consumer's part to be able to weed through what's in the market and ensure that they are not subscribing to something that is not [legitimate]," said Sai Krupa Das, senior scientist on the Energy Metabolism Team at the Human Nutrition Research Center on Aging at Tufts University.
The fact that even experts in the field are grappling with uncertainty, as seen in the controversy surrounding Micronoma's cancer association study, likely complicates the task even further. Scientific uncertainty, however, does not prevent the precision nutrition arm of the human microbiome market from moving forward.
"The market is way ahead of the science and sometimes wrongly so," Das said, adding that consumers need better education and tools to properly discern validated science from marketing spin.
The market is also large and expanding, implying that the issues detailed above will grow with it.
"Precision medicine has come to precision nutrition," Sun Genomics' Jain said. "It's … going to be the huge market that everyone goes after."
An expanding horizon
As the field grows, the human microbiome's potential diagnostic uses continue to expand.
Sun Genomics, for instance, has been studying potential microbial biomarkers and therapeutic targets for autism spectrum disorder (ASD) and manipulating an individual's mood, both of which have held longstanding associations with changes in gut flora diversity.
ASD in particular, has attracted considerable scientific and popular attention, as gastrointestinal (GI) complaints associated with gut microbe dysbiosis are common throughout the disorder. No specific microbial signature of ASD gut dysbiosis has been established, and it remains unknown if dysbiosis is a cause or an effect of the altered eating habits common among those with ASD. Nonetheless, the consistent presence of such altered gut flora suggests both diagnostic therapeutic targets.
Sun Genomics' ASD study is currently online as a preprint on MedRxiv.
GI complaints are also common in Parkinson's disease, and research done to better understand them has led to the theory that the gut microbiome composition can facilitate the spread of the pathogenic alpha-synuclein aggregates that are characteristic of the neurological decay seen in that disease. More recently, investigations into that connection have uncovered evidence that the oral microbiome also changes in people with Parkinson's, in ways that may support diagnosis.
Evidence also links changes in the microbiome to other neurological conditions like post-traumatic stress disorder (PTSD), which currently lacks any established biomarkers.
Illana Gozes, a researcher at Tel Aviv University, collected saliva samples from a cohort of 189 Israeli veterans who had participated in Israel's 1982 war with Lebanon and identified microbial signatures that discriminated veterans with PTSD from those without, as well as correlations between air pollution and PTSD symptoms and related psychopathologies. Her work was published last year in Molecular Psychiatry.
A biomarker test for PTSD could prove instrumental in helping veterans obtain necessary psychiatric support and financial compensation. In both Israel and the US, compensation is tied to the severity of one's PTSD, the measurement of which can be somewhat subjective. If a veteran complains of PTSD-associated symptoms and difficulties but fails to convince an interviewing psychologist, an objective lab test may aid in diagnosis.
"If the psychologist says that [the veteran] doesn't have PTSD, he won't get compensation, but if a psychologist was aided by something that is totally objective, then maybe he will," Gozes said.
Researchers and test developers alike are largely optimistic that microbiome-based assays will eventually become standard of care in many settings, and that despite advances, the field overall is still in a fairly young state. While most experts remain highly skeptical about the capabilities and value of microbiome-informed precision nutrition, companies within that space appear likely to play strong roles in advancing the science.
One key advantage of developing a consumer base, for instance, is that consumer data contributes to the kind of large metagenomic databases needed to conduct studies with high statistical power.
"We've had over 100,000 customers come through," Jain said, "and we have an active database that is growing in the tens of thousands. We'll soon be at our goal of getting to the 100,000 mark on the database side."
Similarly, Viome maintains a database of over 400,000 samples and estimates that over 200,000 people have bought the company's microbiome-based wellness tests.
Nonetheless, most experts agree that more remains unknown than known about the human microbiome and how it interacts with the rest of the body to impact things such as health, disease, and drug metabolism.
"It's challenging because we don't [know] as much as we'd like to say, but there are a lot of advances that are being made," Tuft's Das said. "The field has definitely exploded and is going in the right direction. And there's a lot of pieces of the puzzle coming together."