Karen Titus
August 2019—Even though he practices in Houston, James Versalovic, MD, PhD, swears he can see the coastline. As he and colleagues at Texas Children’s Hospital delve into research related to the human microbiome, several diagnostic and treatment possibilities are starting to appear tantalizingly close.
Comparing their endeavors to ocean explorers of yore, he says, “We’re not quite sure exactly what is in front of us. We can see land—but we’re not quite there.”
But unlike Captain Cook and company, Dr. Versalovic and others in the field have access to next-generation sequencing (though Dr. Versalovic jokes, “It’s been around long enough to be called this-generation sequencing”). In pediatric irritable bowel syndrome, physicians are using both the composition and function of gut microbiome to identify those with the disease, to help stratify patients, and to match them to the appropriate treatment.
His excitement is palpable. “What we are trying to do is bring precision medicine to gastroenterology by using the microbiome as a window to chronic disease,” says Dr. Versalovic, pathologist-in-chief at Texas Children’s and professor of pathology and immunology at Baylor College of Medicine.
His sailing analogy is apt in many ways as physicians and researchers consider the potential uses of the microbiome, itself a vast and changing sea of sorts. Whole new worlds could open up in medicine, including pathology. “Our job now is to bring this ship to shore and make something out of it,” Dr. Versalovic says.
[dropcap]F[/dropcap]or those who prefer the steadiness of a terra firma analogy, Alexander Khoruts, MD, is happy to oblige. Dr. Khoruts, who is professor of medicine, Division of Gastroenterology, Hepatology, and Nutrition; director, luminal GI section; and medical director, Microbiota Therapeutics Program, University of Minnesota, calls his work—developing microbiota-based therapeutics—a “new frontier.”
Targets include several clinical conditions “where the microbiome is totally decimated with antibiotics,” he says, including cases of recurrent Clostridium difficile or in patients receiving high doses of antibiotics in the context of intensive chemotherapy. He and his colleagues are also targeting ulcerative colitis, metabolic syndrome, autism, and advanced liver disease; their hope is to use microbiome-based diagnostics to stratify patients for treatment.
Stepping even further inland, Colleen Kraft, MD, compares microbiome-based research to the Wild West. “Approaches and analysis are not yet standardized—we’re still learning. This is an exciting field,” she says. She anticipates a more focused approach will emerge eventually. “I think it has to be closely paired with therapeutics,” says Dr. Kraft, associate professor, pathology and laboratory medicine, and associate professor of medicine, Division of Infectious Diseases, Emory University School of Medicine. Once a therapeutic gets FDA approval, the diagnostics will evolve as well. “Because those companies will be looking for more indications, and they’re going to be looking at primary prophylaxis, etc. That will pivot the whole field, if and when that happens,” says Dr. Kraft, who is also medical director, microbiology section, Emory Medical Laboratories, and cofounder (along with gastroenterology colleague Tanvi Dhere, MD) of Emory’s Microbiota Enrichment Program.
Regardless of the metaphor used, it’s clear that something big is happening with microbiome testing.
Dr. Versalovic points to a study—he is a coauthor—published in May as launching “a new era, I think it is fair to say, in applying the science of the microbiome to pathology and diagnosis and monitoring of chronic disease conditions.” The study focused on recurrent abdominal pain, which, coupled with constipation, diarrhea, or a combination of the two over a period of time, constitutes IBS (Hollister EB, et al. J Mol Diagn. 2019;21[3]:449–461).
Currently physicians rely on clinical criteria for diagnosis. The JMD study used whole-genome shotgun metagenomics and global unbiased fecal metabolomic profiling to characterize fecal microbial communities, along with correlation-based approaches and machine learning algorithms to identify associations between microbes, metabolites, and abdominal pain.
It turns out the intestinal microbiome in children with IBS is qualitatively different from the gut microbiome in healthy children, Dr. Versalovic says. “What was important in our study is that we have clearly stratified patients with IBS by examining both the composition and function of the gut microbiome, without examining human DNA.” The stratification allows physicians to identify patients who might benefit from dietary interventions. By knowing which microbes—primarily bacteria, in the case of IBS—and genes are present, pathologists should be able to classify patients in ways that are similar to stratifying patients with cancer or with other chronic diseases.
Among other studies, Dr. Kraft is involved in a clinical trial that enrolls renal transplant recipients who have been infected with a drug-resistant organism to receive fecal enemas. These patients typically have urinary tract infections with ESBL-producing Enterobacteriaceae, and they may have carbapenem-resistant Enterobacteriaceae or vancomycin-resistant enterococci infections as well, she says. While the infection has been documented, “We perform selective culture on their stool to see if they’re still colonized,” since infections typically persist in the gut.
Can resistant bacteria on a selective agar become a surrogate for a bad gut? “I would argue that if you can culture CRE from your stool, or rectal swab, then your gut microbiota is really decimated,” Dr. Kraft says. “I’m going to warrant that one day we may be intervening on these people when they have positive stool screening cultures, with microbiome restoration-type therapeutics.” In addition to looking at stool enemas (an approach “that is only mildly acceptable,” she acknowledges with a laugh), she and her colleagues have an IND from the FDA to look at use of stool pills.
They’re also looking at healthy patients with so-called normal microbiome, in a clinical trial being done with Bio-Rad Laboratories. Though results are not yet available, she says the literature suggests that patients who have 40 to 60 percent of an anaerobe (such as Bacteroides) in their gut post-stool transplant “usually means that you’re doing pretty well. So we want to see a predominant bacteria in the form of a strict anaerobe.” This represents a delicate balance, she says, and it doesn’t take much—overuse of antibiotics being a prime example—to destroy it.
[dropcap]T[/dropcap]he technical challenges of doing this work have made for a rough voyage. For starters, microbiome testing, at present, is about as standardized as interpretive dance.
Collection and preservation of primary samples vary, which has a tremendous impact on data analysis, says Dan Knights, PhD, assistant professor in the Department of Computer Science and the BioTechnology Institute, University of Minnesota, and CEO of CoreBiome. Unlike human genome sequencing, which involves looking for mutations within copies of one genome, the microbiome is a complex mixture of hundreds of genomes, all mixed together at different ratios.
With that as a starting point, he continues, “You can imagine all the things that can go wrong.” DNA extraction requires a balanced method that doesn’t bias certain species up or down. “And that’s just one step out of dozens in the process that can introduce bias into the final readout.” Researchers are trying to understand those sources of bias and lock down a protocol that provides consistent, verifiable answers, Dr. Knights says, whose academic research is focused on the gut microbiome.
For now, stool is the most accessible sample, and it makes practical sense for sampling the intestinal microbiome. Says Dr. Khoruts: “That’s where most of the microbes are.” And even stool can be challenging to work with, producing noisy data, he says, though use of shotgun sequencing is helping researchers generate higher-resolution data.
Dr. Kraft poses a number of questions related to her research. Is swabbing heavily on a selective agar sufficiently quantitative? What is the real sensitivity of this approach? “One of the reasons we’ve introduced selective agars on stool as our endpoint is because I wanted it to be something extremely practical—we can’t be doing 16S sequencing on demand for everybody.” But that leaves the not-so-simple matter of figuring out the best way to endure total sensitivity.
Dr. Khoruts finds himself wondering if researchers are focused on the right microbiome. Other microbiomes have their own sample types as well, obviously, including from skin, oral cavity, respiratory system, and reproductive tract. At his institution, sampling has expanded to include colon mucosa obtained during colonoscopies, which may have better resolution for distinguishing disease versus normal. Small bowel and related organs remain unexplored territories but may end up being more important than colon, Dr. Khoruts says.
The microbiome itself is terra incognita for laboratory testing. “There are no normal values for anything,” says Dr. Khoruts. Even with the most abundant microbes, like Bacteroides in stool, there can be three to four logs difference between healthy individuals. “We still haven’t learned enough science to know how to present these data in meaningful ways—and even what, in fact, is dysbiosis.”
While the picture may be clear in some cases (when a patient has taken a lot of antibiotics, for example, “It’s very easy to look at the microbial community and say, ‘This is not normal,’” says Dr. Khoruts), there are no accepted signatures for conditions such as metabolic syndrome. Even abnormalities that have been described in the literature vary across publication—no pattern has emerged, in other words. And even if one had, he continues, it’s not clear whether a signature drives disease or results from it.
Given that an individual’s microbiome fingerprint is more unique than their genome, “We’re sort of stuck in the middle right now,” says Dr. Knights. “We know this is important, but it’s too complex, too variable, to know what’s normal. You really need to have a whole set of normals that represent the different types of microbiomes we see in healthy people.”
[dropcap]H[/dropcap]aving come this far, however, no one wants to return empty-handed. The potential riches are too enticing.
There’s data to suggest microbiome testing might be useful in staging liver disease, based on a fairly consistent pattern that has appeared in a small number of publications, says Dr. Khoruts. “There is a progression in what happens through the microbiome, and one can distinguish whether someone has liver fibrosis based on their microbiome, at least according to the research literature.” As always, validation remains the stubborn missing link between clinical use and what looks good in print. “But there is some hope there,” Dr. Khoruts says.
Also in the works: a microbiome test that would evaluate patients’ burden of antibiotic-resistant organisms. “As we are anticipating a tsunami of infections with these kinds of organisms, that could be clinically useful,” Dr. Khoruts says.
Equally interesting is the possibility that the vaginal microbiome might be predictive of response to antiretroviral drugs—nonresponders appear to metabolize to the anti-HIV drug tenofovir, says Dr. Khoruts. A microbiome-based test might be useful to select patients for treatments or drug trials.
“And we all kind of fantasize still about stratifying patients who have metabolic syndrome,” a complex, likely multifactorial disease. Imagine, he says, using the microbiome to identify a subgroup of patients who are likely to respond to diet or to microbial therapy. “We don’t know why bariatric surgery works in some but there is substantial recidivism in others.”
Dr. Versalovic has oncology, including colorectal cancer, in his sights and says that in the future the microbiome may be important in assessing colorectal cancer risk. “It may be that pathologists will obtain tissue and also look at the microbiome associated with that sample.”
Yet another area of interest is checkpoint immunotherapy and microbiome. Again, the research is intriguing and seems to suggest an association, but each publication recognizes different patterns. “There’s no agreement on what exactly the diagnostic would be. So there’s hope that we’ll sort through this mess. But it’s still years away,” Dr. Khoruts says.
To get there will require new tests. Looking at DNA and microbial composition doesn’t necessarily provide much useful information, just the assemblage of microbes, although measuring its fluctuation might be an interesting diagnostic. Everyone’s microbiome “generally stays in its own cloud,” Dr. Knights says. “Your microbiome is kind of bouncing around in a certain radius within some high-dimensional boundaries.” Small changes occur constantly. “But when is the change large enough to make a difference for a diagnosis? I don’t think the scientific field has the answer to that yet.”
Apart from DNA, researchers are more interested in functionality, Dr. Khoruts says. These sorts of tests tend to be more expensive and difficult to perform. “And then, how deep are you going to sequence?” he asks. Even one species—E. coli, say—can have multiple strains that vary dramatically in their functional potential (e.g. enterohemorrhagic E. coli, toxigenic E. coli).
Functionality depends on which genes are being transcribed and what products are being produced. Says Dr. Khoruts: “It gets you into metabolomics. We’re very invested in that, and it’s much harder to work with.”
All this complexity is pointing physicians in a new direction, according to Dr. Knights. “It’s probably the first time medicine is having to learn ecology,” he says. “In most diseases the microbiome signature of the disease is not just a single bad microbe, or lack of a single good microbe, but instead is more about the healthy ecosystem. And that’s causing people to rethink how to do the clinical test.”
He predicts there may even be a role for an artificial intelligence-based approach. “I think that’s going to show up from time to time in the clinic. Sometimes you really need that, because there are so many different types of healthy microbiome. You can’t just pin it down and say, ‘This is the target.’” But, Dr. Knights notes, the FDA is still learning how to regulate AI-based approaches.
[dropcap]O[/dropcap]ddly enough, this spring another federal agency, the FBI, showed an interest in microbiome testing, raiding the offices of a startup health company, uBiome, that offered direct-to-consumer microbiome testing.
It’s likely the agency is more concerned with billing irregularities than laboratory medicine, but it highlights a real concern: Is consumer hype getting ahead of clinical realities? Some see worrisome parallels to the companies that lump genetic testing in with DNA ancestry testing.
Dr. Khoruts says his gastroenterologist and other clinical colleagues fairly regularly are handed results from patients who have availed themselves of a direct-to-consumer microbiome test. The turmoil is real. “The vast majority of physicians are unprepared to deal with this. It sounds scientific-y, and I think many view these tests with the appropriate skepticism, but they still don’t know how to handle the information.” For now, it’s simply led to a scenario that could have been a subject for Rembrandt: The Angst of the Physician.
“We’re definitely seeing more of that,” Dr. Versalovic agrees. “Our pulmonologists and gastroenterologists are getting asked often now about the microbiome,” including how it relates to antibiotic use, preventive medicine, and nutrition. “The media has been very effective at publicizing this,” he says, as well as the potential for probiotics. Fecal microbiota transplant and C. diff infection has even made its way into popular culture, including an episode of Tig Notaro’s show One Mississippi several years ago. Dr. Versalovic calls the uBiome incident sobering, involving an “overzealous company that got ahead of itself in promoting microbiome testing.”
Nevertheless, even the most cautious labs need to know that microbiome testing is a reality, says Dr. Khoruts. A large number of for-profit companies are developing microbiome-related therapeutics and diagnostics, and not always with the most virtuous intentions. “What we have,” he says, “is unregulated industry, in the wellness arena, that is already peddling promises that are totally unsubstantiated.” He worries that the FDA is taking a hands-off approach to the matter for now, which does not sit well with him. Watching patients pay for dubious, expensive tests “seems predatory,” he says. “We’re failing them.”
Dr. Kraft agrees: “Whether or not we give a million disclaimers, people will still try to use this information clinically.”
[dropcap]W[/dropcap]ith consumers already tapping at the door, the onus is on medical researchers to bring microbiome testing safely into practice.
Up to now, pathologists have played no role in helping gastroenterologists identify and type IBS. “As we refine this—and there’s plenty of work to do—as we continue to advance our understanding of the microbiome in chronic disease,” says Dr. Versalovic, pathologists will be able to pair treatment (including new medications) with more effective diagnoses via the microbiome and molecular pathology.
Dr. Kraft suspects the laboratory may be involved in donor testing related to fecal microbiota transplants. That raises another issue, however. Test validation, though very surmountable, is definitely something that concerns her colleagues, she says. “Is donor testing appropriate if I’m using it on a platform that’s not validated for normal stool?” Moreover, increased screenings might put an unnecessary burden on labs if no intervention is available, no matter how compelling the test. In her work, she thinks selective agar on stool should be reserved for people at high risk whose disease can be treated. “We’re looking for all these screens, but we still don’t have a reliable way to restore the microbiome” in most scenarios, she says.
Dr. Versalovic foresees an era of metagenomic medicine—looking at the microbiome and human host simultaneously. Two lab subspecialties will be affected, he predicts: molecular pathology and medical microbiology. “As we take a more holistic view of human health, we are closely examining our microbial and human genes together. Considering human microbiology much more broadly, beyond infectious diseases, provides a whole new opportunity to think about the role of the microbiome in chronic disease.”
None of this surprises him. “For those of us who were privileged to be part of the initial phase of the [NIH-funded] Human Microbiome Project, which extends back to 2008, it’s fair to say that these implications and future applications became quite obvious early on,” he says. “But we had so much work to do in terms of characterizing the microbiome at different body sites.”
Some considered microbiome the so-called forgotten organ, Dr. Versalovic continues. “It has as much if not more metabolic capacity than the human liver.” It was obvious different microbes working together as microbial communities have an impact on human biology, health, and disease states. “We knew that at some point the dam was going to break.” He compares the pace to the Human Genome Project, which required at least a decade to manifest applications in pathology. “Now we’ve seen so many examples of that in human genetics and oncology. We’re just beginning to enter that phase with the human microbiome and human disease.”
That’s taking shape at Texas Children’s Hospital, where he and his colleagues have created a microbiome center within the Department of Pathology. In addition to allying research scientists with medical microbiologists, pathologists, and molecular pathologists, they’ve created a medical metagenomics fellowship program. All this pulls them closer to that aforementioned shore, he says.
Dr. Khoruts says he, too, is optimistic and predicts a time when labs set up microbiome departments. “Of course, just because something seems clinically useful doesn’t guarantee it will make it as a diagnostic,” he says. “There’s always so much hope and hype.” It’s clear that normalizing the microbiome can be used to treat recurrent or refractory C. difficile infection, for instance, and the type of normalization doesn’t seem to matter (as long as it’s not the antibiotic-decimated version, he adds). But he calls this fairly low-hanging fruit.
Dr. Knights agrees. “On the one hand, it’s a miracle cure, and there are several companies moving toward an FDA-approved solution for that. On the other hand, we’re just scratching the surface of all of the other diseases that might be impacted, and what microbes might be used to treat those diseases.” He says he’s been “pleasantly surprised by just the sheer volume of disease association” with the microbiome. And while he remains a fan of healthy skepticism, he’s encouraged by the ongoing research. “The amount of science that’s going on, the amount of mechanistic studies that are coming out on almost a weekly basis, is impressive.”
He predicts a range of testing scenarios. Some diagnostics might be simple multiplex PCR that look for a small number of known targets. That is scalable and can be handled with existing laboratory capabilities. Other approaches will be far more complex and are likely to be relegated to specialty labs.
Dr. Versalovic says microbiome testing could eventually be within reach of many labs, given the widespread use of DNA extraction methods and DNA sequencing platforms (not to mention mass spectrometry, which is now being applied to human and microbial metabolites that may prove to be functionally important). The fulcrum will be data analysis, Dr. Versalovic says. As pathology departments across the country develop clinical informatics programs, they’ll need to keep a sharp focus on applying data analysis to metagenomics and the microbiome. “We need to take established technologies and apply them in new ways by analyzing the data differently. That’s going to be the big leap now in terms of bringing the boat to shore.”
“For pathology departments, it’s: How can we take established technologies in clinical chemistry, mass spectrometry, nucleic acid isolation, nucleic acid sequencing, then harness the information?” he adds. “That will require a whole new branch of clinical informatics to do full-scale metagenomics.”
Despite all the enthusiasm in the field, Dr. Versalovic says conversations with his pathologist colleagues tend to arc like a flare, sparking brightly and fading quickly. “Most pathologists still are—rightfully—questioning where this is headed, and whether it will be useful in the near or distant future.”
But for his part, Dr. Versalovic says, “I’m firmly convinced that the human microbiome is a big part of a more holistic approach to human health and disease. As pathologists, it is incumbent on us to begin to sort out how we can apply this new knowledge to our specialty. It’s now a matter of pushing and probing and leveraging the science.” The effort will take years, he concedes. “But as long as we can develop a critical mass of people in pathology—and outside pathology—who can advocate for this field and help translate this into the practice of pathology, our patients will ultimately benefit.”
Karen Titus is CAP TODAY contributing editor and co-managing editor. Dr. Versalovic and Dr. Kraft, and Christoph Thaiss, PhD, of the University of Pennsylvania, will speak on Sept. 22 during CAP19 at the scientific plenary on the microbiome.