December 4, 2014 | This November, MIT and Massachusetts General Hospital announced $25 million in funding for a new Center for Microbiome Informatics and Therapeutics, for researchers from both institutions to explore links between human disease and the bacteria and other microbes that live in and on our bodies. With a flagship program studying inflammatory bowel disease, and support for multiple smaller “innovation projects,” the new Center has a mandate to spend its entire startup fund in five years, with the explicit aim of producing new therapies and bringing them into clinical testing.
This joint initiative comes at a time of intense interest in microbiome-based therapies, including from venture funds and biotech entrepreneurs. Just this week, a company focused on microbiome-derived drugs, Seres Health of Cambridge, Mass., closed a $48 million funding round to advance a therapeutic into Phase III trials. As these therapies look increasingly likely to enter medical practice, now is an important moment to examine their promises, limitations, the pipeline from basic science to new therapeutics, and the regulatory structures that could govern them.
To learn more, Bio-IT World correspondent Aaron Krol spoke to Eric Alm, associate professor of biological engineering at MIT, who will serve as co-director of the new Center alongside MGH’s chief of gastroenterology Ramnik Xavier.
Bio-IT World: The Center for Microbiome Informatics and Therapeutics has come into being just as a lot of companies and investors are getting excited about producing new therapies based on the microbiome. Do you think the time is right for this field to be moving aggressively into medical interventions?
Eric Alm: One of the reasons the field is really taking off right now is the Human Microbiome Project, characterizing microbial diversity in healthy populations, and that project was fueled by the fact that we’ve gotten very good at sequencing DNA. We built that technology so we could sequence the human genome, and with the human genome there has always been the promise that we would be able to turn our DNA sequence into drugs. That’s been true to some extent, but it’s a very difficult process, to say the least. There’s no obvious roadmap to go from sequencing a genome, or sequencing a hundred genomes, to a new drug to treat condition X, Y or Z. With the microbiome, the potential is there for a much faster route from DNA sequence to something we can translate into the clinic.
We’re seeing new correlations between the microbiome and human disease pretty much every week. And that’s not surprising, because the microbiome is important. It’s been described as a virtual organ. So the interesting shortcut here is, if any of those relationships are causal — if what makes you sick is having too much of a certain microbe, or not enough of some other microbes — then we actually have something we can test in the clinic. We can go to healthy people, culture all of these strains, and if sick people don’t have enough of them, we take them out of the freezer, grow them up, and put them in a pill. Or maybe you’ve got too much of some species, so you need to raise a phage against it, or a narrow-spectrum antibiotic, or put in a whole new landscape with fecal transplants. At the Center, we’re starting to build that toolkit to engineer the microbiome so we can deliver these new types of therapies.
Unlike someone trying to go from the human genome to a new therapy, you also have the advantage that the differences between people’s microbiomes can be larger and more apparent than the differences between their genomes.
Right, and this is also dynamic data, so something we’re very interested in at the Center is longitudinal sampling. One of the things we’re doing is building a Smart Toilet, which will collect these samples automatically when you use it. Eventually, a later version of the Smart Toilet, in addition to collecting samples, could also start processing them. We’re also looking at micro-blood draws to get a longitudinal, daily picture of what the microbiome is doing, and what the host is doing.
With an inflammatory bowel disease patient, for instance, having that person’s genome is helpful, because it might say in the course of your lifetime you’re more likely to get IBD. But we think this dynamic data could tell us: you’ve got IBD, and you’re headed to a flare-up next Tuesday, and here are the activities you’ve engaged in in the past that lower your biomarker levels to the safe zone, and change your microbial composition. So it’s not just about your predisposition to disease, but where are you on the spectrum of healthy versus diseased, and are you starting to veer away from healthy?
MIT partnered with Massachusetts General Hospital to create this Center. Has MGH been active in using microbiome-based therapies that already exist, like fecal transplants for C. difficile infections?
Absolutely. We’re collaborating on a number of fecal transplant studies now with MGH, and we had a paper out recently
characterizing what happens after a fecal transplant for C. diff
. We’re working on a follow-up paper now that gets into it at a much deeper genomic level.
I’m excited to see that “Informatics” is right in the name of the Center. What types of analytical tools do we still need to really understand the microbiome?
Dealing with these kinds of data presents a lot of new problems that haven’t come up in other fields of bioinformatics. One problem, for example, is that whenever we go into a sample and look at how many microbes are there, we’re not really counting those cells. We’re looking at relative amounts. So you can get changes where one bug will bloom in one person, or at one point in time, and then there’s an ambiguity. Did that one organism increase in number, or did everything else just decrease in number? When we first started working in this field, you would find a lot of these correlations, and most of them would be dead wrong. It turned out that it was all a mathematical artifact of dealing with these relative proportions. That hasn’t really cropped up in bioinformatics previously, because there’s a certain type of data that’s very prone to this.
Another problem is we really want to look at time series, and a lot of the methods for analyzing time series data, for various reasons, are not appropriate for the type of data we collect. It’s more than just collecting lots and lots of big data. We need to fundamentally develop new analytical approaches, just like they did in the early days of gene expression microarrays. And of course, there are large quantities of data, and that makes it even more difficult for clinicians to get involved.
Have clinicians been reluctant to look into therapies based on the microbiome?
To some extent, we have carried out microbiome-based research in silos, where we have the basic scientists working with the computer scientists to analyze the data, and then we have the folks doing clinical trials. There are a lot of people doing clinical trials, even here in the New England area, who are collecting these samples, but they’re not generating the molecular analyses that they could. We see that as a huge lost opportunity. When really exciting clinical samples are collected, as part of some clinical trial, or better yet early in the experimental design, we want to get engaged with clinicians to say, you guys should collect samples at these different time points, and all these different types of samples. And we will help you process those samples and get the big molecular data we think is appropriate, and we’ll also help you do the data analysis.
Is that the advantage the Center gets from pairing an institution like MIT with a research hospital like MGH?
Absolutely. That’s reason one, and reason two is to keep us focused, so that there’s a pressure on us. After you’ve done the basic science, and you’ve analyzed the clinical samples, there’s another step, and that’s to give something back to those clinical sites that they can use to treat patients and run trials. We want to get their data into the MIT part of the Center, but then we want to send something back — a new microbiome-based therapeutic to test or a new strategy for treating disease. That’s the vision.
At your lab, you’ve also taken a very practical, policy-oriented approach to microbiome therapies. How should regulators and government address this area?
One of the biggest challenges we face is navigating a complex regulatory environment. That’s something basic scientists don’t always think about, but it’s important to us because we also have a clinical mission. Members of my lab founded OpenBiome
, the nation’s first stool bank, and they’re leading the charge on the regulatory front. My personal philosophy is, if we’re only talking about stool, we already know that it works, at least for C. difficile
infection. So it should not be regulated as a drug. We’ve argued that it should be regulated as a tissue product, because regulation as a drug might open up opportunities for companies to get some market exclusivity, which I think would be bad for everyone. It’s one thing to get market exclusivity after you spend $100 million in R&D on improving a product, but if you’re just putting poop in a bag, that’s kind of abusing the system the FDA has worked out. It’s a human-derived product, and there should be regulations to make sure there is no infectious material present, but we shouldn’t force stool into the same regulatory framework as a conventional drug if it doesn’t fit. Maybe it fits better as a tissue product, or maybe it needs its own category — and that’s a challenge for the FDA, because they have to work within the existing framework.
So you don’t see the general argument for market exclusivity, that it defrays the cost of drug development, applying to the microbiome space?
We’ve already got a non-profit stool bank. You wouldn’t shut down the Red Cross so Pfizer can own the market for blood. Instead, we should be pushing for better therapeutics, synthetic microbial products where you can control the composition and make it from batch to batch very reliably, and meet all of the very rigorous standards the FDA has for the manufacture of drugs. I think it’s going to be a non-trivial expense to develop a synthetic microbial therapeutic, so I can see those products being regulated in a very similar way to conventional drugs, because they can be manufactured in a similar way to conventional drugs.
In addition to your flagship IBD program, part of the Center’s $25 million fund will be going to small innovation projects. How are these different from typical research grants?
One of the basic philosophies of the Center, at least getting it off the ground, is that a lot of the action is going to happen in the next five years. So the entire $25 million is going to get spent down in the next five years. It’s not going to be just a few endowed graduate fellowships. There will be a lot of people engaged in substantial, multi-investigator, multi-year innovation projects. With an innovation grant, you can think about trying to develop a therapeutic, for example, for some indication that’s not currently being looked at in the Center.
Does microbiome research have a difficult time attracting funding from traditional sources?
In general, this is a good time for funding for microbiome research. But at the same time, we’re very cognizant of our mission to translate science, and not just fund what could probably be funded elsewhere. The things that are really difficult to fund are bold projects that say, for instance, I have a real interest in rheumatoid arthritis, I think I know how to make a drug, and I want to run clinical trials. That would be very difficult to fund at NIH. There are many steps along the way where you could fail, it’s very high risk, and with funding rates as low as they are, you’d have to be a lot more conservative. You’d have to break it into much smaller pieces, spread over a much longer period of time. We’re in a position to go after a few different diseases as aggressively as possible, and really see what works, and that’s why I think this is a very good model. If our fundamental premise is wrong, and there’s not a lot of low-hanging fruit in the microbiome space, then I guess we won’t be successful. But if our fundamental premise is right, and some of the disease-microbiome associations are causal, then I think we have the model to get there very quickly.
Microbiome research tends to attract press that perhaps overstates the significance of certain findings. What do you think we should keep in mind when we consider the medical implications of the microbiome, to keep things in perspective?
There is a lot of excitement in the press when we see a study correlating the microbiome with disease. And some of those associations might be causal, some of them might not. If they are causal, they might be treatable by perturbing the microbiome, and they might not. The thing to keep in mind is that there are many steps to translating a scientific discovery into a therapy. Is it correlation or causation? And if it’s causation, and you’ve taken two samples and put them into different mice, and seen a small change in phenotype, that’s still a long way from taking those bugs and putting them into a person. We know that putting a completely different microbiome in an animal will have a big effect. The microbiome is important. We’ve proved that beyond a shadow of a doubt. So the results you see may actually be specifically carrying over the phenotype you care about, but you might also just be perturbing the system in a way that happens to affect the phenotype you care about.
But I would also say, I know there’s a lot of talk about overhyping the microbiome, but I don’t actually think there’s more overhyping of the microbiome compared to any other field in science. I think there’s such a flurry of activity, and so much excitement, that we’re just hearing about it more often.
And there is a flip side to that hype, which is that clearly people are very interested in the microbiome. Why do you think this field has been so exciting for the public?
Well, I think you do get a boost in readership every time you talk about poop, so that might have something to do with it. But this is also kind of surprising to people. We don’t normally think about our microbes, and just to hear these facts, that there are ten times as many microbial cells in the body as human cells, or a hundred times as many microbial genes, and their metabolic capacity is equal to or greater than the liver, from this collection of things that are basically invisible — it’s a really interesting concept. Most people were not aware of it a few years ago, and it changes the way you think about a lot of things. Maybe ten or twenty years from now, everybody will be so comfortable and familiar with the microbiome that it’s not as big a topic of conversation. It’s like talking about how important the liver is. People aren’t doing that at dinner, and they’re not reading about it in Newsweek.