In 1957, the legendary microbiologist Stanley Falkow started asking patients at his hospital to swallow their own poop.

Back then, people who were about to undergo operations had to take large doses of antibiotics to kill bacteria that might cause infections. But the drugs also destroyed their normal gut microbes, leaving them with diarrhea and indigestion. To forestall these problems, Falkow and his colleagues collected their patients’ stools and piped the excrement into capsules. By swallowing these after the operations, the patients could replenish their depleted gut communities.

This continued for months. “The chief hospital administrator discovered what was up,” Falkow later wrote. “He confronted me and exclaimed: Falkow, is it true you’ve been feeding the patients shit?” Yes, Falkow confessed, and was fired—only to be re-hired two days later. He never collected data from his ersatz trial, but anecdotally, the patients who took the capsules felt better.

Sixty years later, it seems that progress has been both great and minimal. On the one hand, the concept of fixing gut problems by loading people with stools from healthy donors—a procedure aptly known as a fecal transplants—is gaining in acceptance. On the other hand, the disappointing results from a new clinical trial suggest that there’s still a lot we don’t know about why this unorthodox technique works.

Fecal transplants are most commonly used to cure infections of Clostridium difficile—a weedy bacterium that causes severe, recurring diarrhea. The idea is that a donor’s microbes would reset the recipient’s diseased community, much like re-turfing a lawn that’s overrun by dandelions.

It sounds unorthodox but it clearly works. There are hundreds of stories of miraculous recoveries. And in one clinical trial, fecal transplants cured 94 percent of C-diff patients, while a standard antibiotic cured just 27 percent. Backed with such strong evidence, the technique has become increasingly accepted—a seemingly fringe treatment that is working its way into the mainstream.

Still, there are problems. Fecal transplants are undeniably gross. They can spread pathogens, so doctors typically put potential donors through rigorous screening. And poop is a medicine unlike any other: a largely uncharacterized community of microbes, viruses, and other constituents that varies greatly from donor to donor. This variability causes headaches for regulators—how do you approve a product that comes out of human backsides rather than manufacturing plants, and so changes every time it is produced?

There are two main solutions. The first is to freeze the stools from only the healthiest of donors, store them in stool banks, and use them repeatedly. That’s the approach used by the non-profit organization OpenBiome, as well as facilities in various U.S. hospitals. The second approach is to concoct stool substitutes—cocktails of gut microbes that can replicate the benefits of a fecal transplant, but without the feces.

Several companies are going down the latter route, including Seres Therapeutics, a startup based on Cambridge, Massachusetts.  Their lead product—SER-109—consists of a single capsule containing 100 million spores from 50 species of gut microbes, purified from the stools of healthy human donors. It’s designed to treat multiply recurring C-diff infections.

It seemed promising. As Nidhi Subbaraman wrote for Buzzfeed, Seres was a “favorite among a crop of young ventures making drugs containing bacteria.” In September 2014, the company announced that in an early Phase 1b trial, which focuses on safety, SER-109 had cured 29 out of 30 patients with recurring C-diff. In June 2015, the U.S. Food and Drug Administration (FDA) gave the drug breakthrough status—a label intended to accelerate the approval of especially promising therapies.  

But last Friday, those early hopes were dashed when Seres revealed that SER-109 had pooped out of a phase 2 trial (which focuses on efficacy as well as safety).

The company had enrolled 89 patients with C-diff infections that had recurred at least three times, and randomly assigned them to either SER-109 or a placebo. Critically, neither doctor nor patients knew which treatment they were getting—a rarity for fecal transplant trials.  If it had worked, it would have been a big milestone. But after eight weeks, the patients who got SER-109 were just as likely to have suffered another bout of C-diff as those who got the placebo, and just as likely to have experienced diarrhea, abdominal pain, and flatulence.

In other words, the pill didn’t work. Seres stock plummeted by more than 69 percent.

“The results were surprising, especially since the 1b data was eye-popping,” says Roger Pomerantz, the company’s CEO and president, and an infectious disease doctor who has treated many C-diff patients. “We’re now trying to dissect why there was a difference in the two trials.” They’re looking at whether the patient pools differed: Those from the first trial were based on four prominent academic centers, while those from the second came from 37 different sites around the country, including private practices and regional hospitals. They’re also checking if the microbes in the pills actually established themselves in the recipients’ guts.

This failure doesn’t refute the value of fecal transplants, which have already proven their worth in treating C-diff infections. It does, however, show how hard it will be to distill the technique into a more palatable and marketable form. “Obviously there’s something very effective that people are getting when they take the whole stool,” says Diane Hoffmann from the University of Maryland, who is leading a working group that will advise the FDA on regulating microbiome transplants. “Something in there is working. We just don’t know what it is, and it might be hard to deconstruct.”

Or, as genomicist Nathan Pearson neatly put it, “There’s no silver bullshit.”

It’s not even clear if making drugs from stool is the right move. Elaine Petrof from Queen’s University thinks it is. When I was researching my book, she told me that “stool is a stopgap,” and will inevitably be replaced by precise mixtures of well-defined microbes, which can be cooked up again and again to the same recipes. That would be easier to regulate and to check for safety. She and her colleagues are working on one such formulation, which they call RePOOPulate.

But Alexander Khoruts, from the University of Minnesota, felt differently. “The full spectrum of microbes harvested from donors has been designed by nature, and has a proven safety track record in the original host,” he told me. “That’s a very hard benchmark to improve upon with any kind of synthetic.”

The SER-109 story supports his view. It may be that the drug failed because there’s something about the full community of microbes in a donor stool that’s irreducibly beneficial. Pull out one or two beneficial species, and you lose the power of the whole. Perhaps it’s ten species that matter. Or hundreds. Or the chemical milieu of the poop itself.

As Noah Fierer, a microbiome scientist at the University of Colorado, Boulder, noted on Twitter, “Ecology matters.” The human microbiome is not a simple thing that we can tweak by just adding this microbe or removing that one. It’s a vast ecosystem, akin to a rainforest or a grassland. To control it with any kind of success, we need a much better understanding of what its members do, how they compete and cooperate with each other, how they interact with their hosts (that’s us), and how they fare when transplanted in a new host. These are complex questions, which we are only just beginning to answer.

* The article originally intimated that Seres’ Phase 2 trial was the first double-blinded, randomized, placebo-controlled trial for fecal transplants, when there have been others.