Three types of phages, used to prepare diluted solutions in a French hospitalROMAIN LAFABREGUE / AFP / Getty

In 2010, when Lilli Holst scraped a lump of soil from the underside of a rotting eggplant, she had no idea that this act would help to save the life of a British teenager, eight years later and 6,000 miles away.

Holst, an undergraduate at the University of KwaZulu-Natal, in South Africa, was participating in a project in which students search through local soil samples for new phages—viruses that infect and kill bacteria. Holst found several, and gave them all names. In a worm farm, she discovered Liefie. In an aloe garden, Lixy. And from that decaying eggplant, Muddy. All three viruses infect a common bacterium called Mycobacterium smegmatis. And all of them were new to science.

Samples of Muddy and the other phage viruses made their way to the lab of Graham Hatfull, a phage expert at the University of Pittsburgh. He stored them in a freezer, along with at least 10,000 others that had also been discovered and named by students: Mariokart, TGIPhriday, Chupacabra, Benvolio, ChickenNugget, IAmGroot, and more. They were sitting there, in the cold, when in late 2017 Hatfull got a call from doctors at Great Ormond Street Hospital, in London.

The London team, led by the pediatrician Helen Spencer, had been treating a 15-year-old girl with cystic fibrosis—a genetic disorder that leads to persistent lung infections. To prepare for a double lung transplant, the girl had been taking drugs to suppress her immune system, and these allowed an already present microbe called Mycobacterium abscessus to run amok through her body. She had new lungs, but also heavy infections in her liver, limbs, buttocks, and torso, and in the surgical wound on her chest. Antibiotics weren’t working, and the outlook was grim. The team put her on a palliative-care plan.

But Hatfull has spent decades studying phages that attack mycobacteria, the group to which the girl’s life-threatening microbes belonged. Her doctors wanted to know whether he had anything in his arsenal that might kill those particular strains. He looked in his database—and found Muddy.

In laboratory tests, Muddy efficiently destroyed the exact strain of M. abscessus that was itself destroying the London patient’s body. “It was good that we found one,” Hatfull says. “But it was bad that we only found one,” because bacteria can easily evolve to resist any single phage.

His team eventually found two more phages— BPs and ZoeJ—that had the potential to kill M. abscessus, but weren’t doing it very well. Some phages kill the bacteria they infect by reproducing frantically and bursting out in fatal fashion, but others opt for a more tranquil existence of harmlessly hiding in their hosts. BPs and ZoeJ naturally go for the latter path, so Hatfull’s team modified them by deleting the gene that keeps them peaceful. Unrestrained, these modified microbes could kill M. abscessus as well as Muddy.

Last June, the London team started injecting all three phages—one natural and two modified—into the patient. She didn’t experience any major side effects, and after a month of twice-daily doses, the infection in her chest began to disappear. Shortly after, her liver cleared up. After six months, almost all the other lesions had faded. “It’s not like she’s out of the woods, in the sense that she has cystic fibrosis and a new set of lungs,” Hatfull says, “but she’s in very good general health.”

As with any single case of medical success, it is impossible to truly know whether the supposed treatment was what eventually saved the patient: That’s why doctors run clinical trials. But Benjamin Chan of Yale University says that this “fantastic” study “very nicely shows a probable impact of the phages.” After all, the patient’s infections clearly weren’t going away on their own, and they weren’t responding to other treatments.

Phages were commonly used to treat infections in the 1920s, and though they’re still used in Russia and parts of eastern Europe, they largely fell out of favor in the West. But they’ve stepped back into the limelight after a growing line of dramatic success stories. The most famous case involves Steffanie Strathdee, an epidemiologist who led the hunt for phages that ultimately cured her husband, Tom Patterson, of a life-threatening infection. Such successes have prompted a renewed interest in phage therapy, especially in the era of antibiotic-resistant superbugs.

The London patient’s case is a milestone—she is the first person to be treated with phages that have been genetically engineered. “It requires trust to take a leap off the edge into completely unknown medicine,” says Hatfull, who appreciates that many people might be unnerved by his team’s work. “The idea of using a virus in the first place is challenging, let alone messing around with it,” he acknowledges.

He clarifies that his team simply deleted a gene that both BPs and ZoeJ would switch off naturally, when they eventually decide to flip from passive stowaways to active killers. The team also didn’t add any genes from other organisms into the phages—an important distinction, which meant that, under UK and European Union regulations, the viruses didn’t count as genetically modified organisms.

This case also represents a second milestone: It’s the first time phages have successfully treated a mycobacterial infection in a human. That’s huge. These microbes include the one that causes tuberculosis. They also include a group of more than 100 species called the NTMs, which often hit people with cystic fibrosis. M. abscessus and other opportunistic germs belong to this group.“Treating NTMs is a big deal,” Chan says. “It’s a very unmet need in the cystic-fibrosis community.”

But Muddy, ZoeJ, and BPs aren’t cure-alls for these infections. M. abscessus is incredibly diverse, and a phage that kills one strain might do nothing against another. The London team learned that the hard way. It treated a second young girl with cystic fibrosis, who also had a double lung transplant, and who came down with a different strain of M. abscessus. And against that strain, Hatfull struggled to find any effective phages, despite his extensive collection. By the time he identified one, it was too late. The second patient had died.

To reliably treat any given NTM infection, scientists will need much larger phage libraries. “That’s not the case for many other bacteria, like E. coli or Staphylococcus aureus, where strains are more broadly susceptible to commonly isolated phages,” Chan says.

But even in those cases, it’s time-consuming to identify, grow, and perhaps even modify the right virus for every single case. “The challenge is whether you could ever make phage therapy broad enough so you could have an off-the-shelf set at your disposal, which you knew could infect any strain that was out there,” Hatfull says. He doubts that’s possible for M. abscessus, or indeed for most infections. That’s probably why at least one trial, in which doctors tested the same cocktail of 12 phages in patients with infected burn wounds, was a bust.

“For the many infections where you have this great variability, it’s going to be hard to figure out how to get phages to span it all,” he says. And without that consistency, it might also be hard for phages to get significant investments from pharmaceutical companies, and to go from the stuff of individual miracles to the stuff of generalized medicine.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.