Rethinking How to Fight the World’s Deadliest Infection
Scientists still don't fully understand what protects people from tuberculosis. Could an overlooked part of the immune system finally lead to better vaccines?

When Mehak Khan told her mother she planned to focus her PhD research on tuberculosis, her mother was skeptical: “Who even gets that anymore?,” she asked.
It’s a common misperception that tuberculosis is a bygone illness. The reality is that it kills more than 1.2 million people per year, making it the deadliest infectious disease on Earth. But that’s not the only aspect of it that’s poorly understood.
For decades, scientists have struggled to develop a new vaccine for a disease that remains shrouded in mystery. TB is famously sneaky—the bacterium that causes it can hide inside human immune cells and actively subvert or evade the body’s defenses. The only licensed vaccine to combat it, Bacille Calmette-Guérin, or BCG, is more than 100 years old and is not reliably effective in adults. M72, the current frontrunner among newer vaccine candidates, is now in a 20,000-person Phase 3 trial. In an earlier trial, its efficacy was around 50 percent.
Historically, scientists assumed that T cells are the body’s main defense against intracellular bacteria like the one that causes tuberculosis. The role of antibodies—produced by B cells—was largely dismissed. “We had this very black and white view of what T cells do versus what B cells do,” says Khan, a biomedical researcher specializing in infectious disease immunology. “It was always thought that T cells are the only arm of immunity that would be important.”
But recently, researchers have gained a better understanding of the role that antibodies might play in protection. The problem has been that bacteria are genomically enormous—Mycobacterium tuberculosis alone encodes roughly 4,000 proteins, more than 40 percent of which are uncharacterized. Because of this, says Khan, “we never knew which antigens really matter from a vaccine development point of view.”
That might not be the case for long. Earlier this year, Khan and her team used machine learning to develop a comprehensive atlas that maps antibody responses across the full spectrum of TB, potentially offering new clues for how immunologists could approach immunity against a disease that has plagued human beings for 9,000 years.
Using protein microarrays, the researchers mapped the entire Mycobacterium tuberculosis proteome across three different groups—people with active TB, people with latent TB, and “resisters,” those who remain negative on standard TB infection tests despite repeated exposure. They also included rhesus macaques administered BCG either through the skin or intravenously.
Scanning a whole proteome for antibody responses is not in itself new. But previous work focused almost entirely on people who are sick, cataloging which proteins the antibodies of active TB patients recognized.
By including resisters, people with latent infection, and intravenously vaccinated monkeys—three populations that kept the bacterium in check—Khan and her colleagues could look for proteins that surfaced repeatedly in protection rather than illness.
They found that many of the proteins used in experimental TB vaccines, including M72, elicited little antibody recognition in groups whose immune systems appeared to be controlling the bacterium. Instead, researchers identified a different set of proteins that were consistently recognized across these groups.
“[This research] is really opening up the dialogue about the role of antibodies and what part they could play in protection,” says Birgitte Giersing, the unit head of vaccine R&D and policy at the World Health Organization’s department of immunizations, vaccines, and biologicals. “The fact that this analysis is identifying a whole host of new targets is really exciting.”
(What to know about rising tuberculosis cases in the U.S.)
To appreciate the potential impact of Khan and her team’s work, you have to look at the outsized impact TB has had on human history—and the decades of failed efforts to develop a vaccine that works.
Tuberculosis is an ancient disease; evidence of its presence has been found in the spines of Egyptian mummies. You may know it by its former name, consumption, though it’s also had colorful epithets such as the great white plague, the robber of youth, and the king’s evil.
Because it tended to drain people of both life and color, consumption was bizarrely romanticized in 18th- and 19th-century Europe. The combination of pallor and sickly thinness became the standard of Victorian-era beauty; the many artistic lives it claimed, from Keats to the Brontë sisters, cemented its status as a fashionable way to die. Over in the U.S., the disease convinced some New Englanders that they were being preyed upon by vampires.
In reality, tuberculosis was—and still is—spread through the air. While it remains prevalent globally, the majority of its fatalities occur among those living in close quarters, with poor access to sanitation and nutrition. In other words, “it is a disease of poverty,” says Krish Ramanathan, the interim CEO of Gates Medical Research Institute, a subsidiary of the Gates Foundation and sponsor of the M72 vaccine’s current Phase 3 trial. “Most of the wealthy countries have forgotten about TB.”
Technically, there are three ways of getting rid of tuberculosis, says Richard White, a professor of infectious disease modeling at the London School of Hygiene and Tropical Medicine.
The first is to ensure that everyone on Earth has adequate access to nutritious food and healthcare. The second is to screen everyone and treat them. While perhaps not quite as intractable as solving global poverty, option two has its own challenges—many people infected with TB are asymptomatic, and treatment can require a combination of several antibiotics taken over the course of many months.
The last option is to develop an effective vaccine. “Of all the interventions, it's the last one standing when everything else is impossible,” White says. “Getting to somebody once, giving them a vaccine—you can do that in a war zone.”
But the road to finding a vaccine that works has been fraught with challenges. BCG was developed back in the early 1900s, when Albert Calmette and Camille Guérin—a physician and a veterinarian, respectively—cultured tubercle bacilli on slices of glycerin-treated potato mixed with ox bile. It’s still the only vaccine administered today, and it does a remarkable job of protecting children against severe forms of TB. But in adults, its efficacy varies wildly—as low as zero in some populations.
The main difficulty in developing a new vaccine is that researchers have never really understood what protective immunity against TB actually looks like.
TB lacks what is called a “correlate of protection”—a biological marker that indicates whether a person is protected against an infectious disease. In some cases, finding a correlate of protection is as simple as measuring antibody levels—above a certain threshold, it can be reasonably assumed a person is protected.
(This ”bacteria hunter” discovered the causes of tuberculosis, anthrax, and cholera.)
That’s not the case with TB, says Helen McShane, a professor of vaccinology at Oxford University. TB is incredibly complex—it hides in cells and is very adept at evading and subverting host immunity. “It’s unlikely that there is going to be a very simple correlate of protection. And frankly, if there was, we’d have found it by now.”
Picking which of the roughly 4,000 proteins produced by Mycobacterium tuberculosis to target has never been a perfect science—historically, they’ve been the ones that were most abundant, easiest to spot, or had already been researched. “It’s been very difficult to identify which of the antigens may work because we haven’t necessarily had the tools to test them,” the WHO’s Giersing says. “We have a few fundamental building blocks that are missing.”
In addition to lacking correlates of protection, there are no known animal models that can reliably predict a potential vaccine’s efficacy in humans. In short, “the only real way to work out what you need to do to get a vaccine that works, is to get a vaccine that works,” says White. “And unfortunately, just because of the nature of tuberculosis, those experiments which tell us whether it works or not are very expensive.”
Indeed, cost has been a significant bottleneck. The countries where TB is deadliest are also those least equipped to pay for its research. Private companies are reluctant to finance expensive trials, which require thousands of participants followed over the course of years, when the risk of failure has historically been so high. The current M72 trial is estimated to cost $550 million and is backed entirely by philanthropy.
No one expects the new atlas to bear fruit overnight. It will contribute to the long, iterative process of understanding what protection looks like. For example, the research found that antibody levels against one protein in M72, called PPE18, were significantly higher in people with active TB than in those with latent infection. Antibody levels were also higher in monkeys given standard skin-injected BCG than in those given the more protective intravenous version. In groups with greater control over TB, including people with latent infection, the resisters, and the protected monkeys, responses fell below the level the researchers could reliably detect.
Resisters—heavily exposed populations whose bodies nevertheless shrug off the disease—remain an enduring puzzle in TB research. The study found that antibodies from these protected groups converge on a shared set of the bacterium’s proteins—for a handful of them, the resisters’ antibodies were better at helping immune cells recognize and engulf their targets.
That suggests their antibodies may be doing something other than simply marking exposure. But, Khan says, the design can’t establish cause. “Antibodies only mature with help from T cells, so a strong antibody response may be a readout of good T cell immunity rather than a mechanism of protection in its own right.”
As for the antigens recognized by those who mount a protective response, more testing is needed—for example, are the newly identified antigens actually protective? If so, the next step would be to test their protective potential in animal models before eventually moving to human trials. None of this new groundwork means that M72 is doomed to fail.
After all, it was not designed to produce antibodies; its two proteins were chosen precisely because they provoke T cells. It’s simply more information in a field where breakthroughs have not been forthcoming. Based on what the atlas adds, McShane expects the next generation of vaccines will use a mixture of T cell-inducing and antibody-inducing antigens. “I think a balanced approach where we have cellular and humoral immunity is likely to be the most effective vaccination strategy,” she says.
(The “Black Angels” cared for tuberculosis patients when no one else would.)
For now, results from M72’s Phase 3 are expected in late 2028, and though it’s the leading vaccine candidate, several others are in the pipeline, all using different approaches.
“Having that variety gives us three very different bites of the cherry,” White says. While no one is sure which, if any, of the upcoming candidates will be effective, every fragment of information gets researchers closer to understanding TB. That itself is reason for hope, he adds.
“I'm optimistic that by the middle of the next decade, we might have a vaccine which we can generate for a reasonably low price, and really, really make a difference to TB and stop these people dying.”