As a physician-researcher specializing in infections in newborns, Tobi Kollmann is laser focused on how to stop babies from dying. While mortality for kids under five years old has dropped substantially in recent decades, he says, there has been little improvement in the rate of deaths in the first week of life. Indeed, of the 5.2 million children under 5 years old the World Health Organization estimates died last year of preventable or treatable causes, nearly half—2.4 million—died before reaching four weeks of age, and most newborn deaths occurred in the first week after birth. Having spent time in African countries where neonatal mortality is common and witnessed the frustration of colleagues there with their limited options for preventing it, Kollmann, now at the Telethon Kids Institute in Perth, Australia, says he wanted to find solutions that could be implemented immediately.
So when Kollmann came across a 2011 study that found giving the tuberculosis vaccine (called bacille Calmette-Guérin, or BCG) to low-weight infants in Guinea-Bissau at birth cut their mortality rate in their first four weeks of life in half—not because of the protection it offered against tuberculosis itself, but by reducing the risk of sepsis and other infections—it caught his attention. “That to me was just unimaginable,” he says. Around 2012, Kollmann attended a meeting of researchers studying such off-target effects of vaccines and sat in on a talk by one of the authors of the 2011 study, Dutch anthropologist Peter Aaby, founder of the Bandim Health Project. Aaby spoke about the history of research on two vaccines linked to declines in all-cause mortality: BCG and the measles vaccine, which both incorporate live, attenuated pathogens. Kollmann “was just shocked how such a tool that could potentially be so incredibly effective . . . could have been kept under the wraps for so, so long.”
Trials are now underway testing the BCG vaccine to see if it can provide partial, temporary protection against COVID-19—marking the first time a vaccine has been trialed against a specific pathogen other than the one it was designed for.
Kollmann didn’t wait long to begin studying what he terms the pathogen-agnostic effects of BCG, coauthoring several reviews on the topic and, more recently, studying the mechanisms of the effects in animals. Other vaccines, such as the smallpox and live polio vaccines, have also yielded hints in observational studies that they could provide some protection, at least temporarily, from pathogens other than those they target. In recent years, some randomized controlled trials have provided more-solid evidence for such protections.
Despite decades of reports of vaccines’ off-target effects, however, the phenomenon has received relatively little attention, and researchers have only a nascent understanding of the mechanisms. While critics argue that some of the claims made about pathogen-agnostic effects outstrip the available evidence, Kollmann says he thinks many experts have been slow to embrace their potential of pathogen-agnostic effects because of dogmatic thinking about how the immune system works. “People think in boxes,” he says.
The question of how much protection vaccines such as BCG provide against off-target pathogens, and under what circumstances, has acquired more urgency as COVID-19 sweeps the globe. Trials are now underway testing the BCG vaccine in groups at high risk of exposure to SARS-CoV-2 to see if it can provide partial, temporary protection against the novel coronavirus—marking the first time a vaccine has been trialed against a specific pathogen other than the one it was designed for. “It’s obviously very important to emphasize this does not replace a corona-specific vaccine, because, of course, that’s what the world needs,” says Nigel Curtis, a clinician and infectious disease researcher at Murdoch Children’s Research Institute and the University of Melbourne who is leading a trial of BCG for COVID-19. “But this might provide a stopgap until one comes along, because it looks like it might be many months or years before that.”
Early hints at vaccines’ off-target effects
One of the first researchers to note the apparent nonspecific effects of a vaccine was Soviet virologist Marina Voroshilova, who together with her husband Mikhail Chumakov, the director of the Polio Research Institute in Moscow, observed an unexpected benefit while spearheading large-scale testing of a new polio vaccine beginning in 1959. Unlike Jonas Salk’s famous vaccine, which used killed virus, the new, oral version, developed by Albert Sabin, used live, attenuated virus. Voroshilova and Chumakov didn’t hesitate to give the inoculation to their own children as well as to millions of other young people. Voroshilova noticed that in addition to conferring lasting protection against polio, the vaccine also seemed to be associated with lower-than-expected rates of influenza and other respiratory infections in the months after it was given, an effect backed up by later controlled trials. It wasn’t clear what explained the off-target effect, but Chumakov, Voroshilova, and their colleagues found that the vaccine raised subjects’ blood levels of interferons, proteins that instruct immune cells to launch an antiviral response.
Other studies looking into the Sabin vaccine’s effect on diseases besides polio supported this observation. And as it turned out, the oral polio immunization wasn’t unique in providing nonspecific protection. In controlled trials of the BCG vaccine conducted between 1948 and 1961 in the US and the UK, researchers found an average reduction in mortality from causes other than tuberculosis of about 25 percent among those who received the shot compared with those who did not. BCG even proved itself an early form of immunotherapy after researchers at Queen’s University in Ontario published a 1976 report of nine patients with bladder cancer who received an injection of the BCG vaccine directly into the affected organ and had a lower-than-expected recurrence rate. The vaccine is still used to treat bladder cancer today.
Another vaccine with apparent bonus protective effects was one designed to prevent measles. In the 1970s, vaccinating children in very poor countries against measles wasn’t a priority for international aid groups because, some believed, the children who died from the disease were malnourished and would have died soon of something else. But when Aaby and his colleagues visited an area of Guinea-Bissau in 1979 to investigate this hypothesis, they found that the children who died of measles weren’t more likely to be malnourished than those who survived the disease. After public health officials implemented a vaccination program with a live, attenuated measles inoculation in the area, mortality within one year among 6- to 35-month-old children declined sharply, Aaby and his team found—an effect not fully explained by the drop in measles infections. The study was not a controlled trial, however, so it remains impossible to say if the vaccine was responsible.
To examine this apparent link between the measles vaccine and non-measles mortality in other settings, Aaby and his colleagues later analyzed the results of 10 cohort and case-control studies on measles vaccination performed in seven developing countries. The drop in mortality rates after vaccination ranged from 30 percent to 86 percent, outstripping the reductions that could be explained by measles mortality rates alone. “These observations suggest that standard titre measles vaccine may confer a beneficial effect which is unrelated to the specific protection against measles disease,” the authors wrote in their 1995 BMJ paper. They detected no such effects with two other, non-live vaccines, diphtheria-tetanus-pertussis (DTP) and a polio immunization.
It was around that time that Christine Stabell Benn, then a medical student at Aarhus University, joined Aaby’s team at the Bandim Health Project. She never left, and the pair has now studied the nonspecific effects of vaccines together for more than two decades, becoming perhaps the most visible advocates of those effects’ significance. Through further studies, largely in Guinea-Bissau, Benn and Aaby concluded that vaccines made of live, attenuated pathogens—including BCG, the live polio vaccine, and measles and smallpox immunizations—provoke beneficial nonspecific responses, while vaccines that use bits of killed virus do not. In fact, Benn suggests that non-live viruses actually make some kids temporarily more susceptible to illnesses other than the one they’ve been vaccinated against. (See “A Nonspecific Vaccine Risk?” below.)
Critics have questioned the existence and strength of nonspecific effects, both beneficial and deleterious. “A lot of the evidence is very weak,” says Paul Fine, an epidemiologist at the London School of Hygiene & Tropical Medicine who has written letters in the past to journals challenging the interpretations of study results that indicate such effects. One problem with observational studies, he says, is that “vaccines in routine practice are not allocated randomly, and kids who receive vaccines, in any population in the world outside a vaccine trial, are not like the kids who don’t receive vaccines. That’s been a main issue” that can confound results, he says.
To sort through the evidence on nonspecific effects and what implications they might have for vaccine schedules, the World Health Organization (WHO) convened a working group in 2013 that included Fine, Benn, and other experts. (The WHO formulates recommendations for low- and middle-income regions, but each country develops a vaccination schedule based on its own disease threats and other circumstances.) As part of its research, the group commissioned an independent systematic review of studies on nonspecific effects of BCG, DTP, and the measles vaccine. The review identified five clinical trials of BCG, including the study of low birthweight infants in Guinea-Bissau that had caught Tobi Kollmann’s attention a few years earlier, that found a 30 percent average reduction in mortality with the BCG vaccine. The review’s authors noted that “tuberculosis is now an infrequent cause of death in infants and young children, so if BCG has an effect on all cause mortality it is unlikely to be entirely due to fewer deaths from that disease.” Similarly, the review identified four controlled trials of the measles vaccine that found an average decline in mortality of 26 percent, with deaths from measles contributing negligibly to that finding, suggesting that the vaccine also protected from other causes of death. The reviewers found no clinical trials of nonspecific effects for DTP, and all of the observational studies addressing the question were classified as having a high or very high risk of bias and generated widely varying results.
The working group concluded in 2014 that “the systematic review neither excludes nor confirms the possibility of beneficial or deleterious non-specific immunological effects of vaccines on all cause-mortality,” and also noted a lack of data on what the mechanisms for such effects might be in humans. The authors suggested that the WHO develop standard protocols for studying nonspecific effects in an ethical and rigorous way, but it’s unclear what the current status of that effort is. A WHO official told The Scientist in July that she would look into the matter but never provided the information despite several follow-up inquiries.
How Vaccines Train Innate Immunity
While researchers have observed for decades that certain vaccines seem to help recipients ward off more than just the target pathogen, only in recent years have they identified possible mechanisms for these bonus benefits. For example, in a study published this year (depicted here), researchers examined immune cells from the blood and bone marrow of healthy adults before and after they received a live tuberculosis vaccine known as bacille Calmette-Guérin, or BCG.
© TERESE WINSLOW
In the bone marrow post-vaccination, genes are expressed that trigger hematopoietic stem and progenitor cells to differentiate into monocytes, neutrophils, and other so-called myeloid cells. In a separate analysis of the effects of BCG in newborns, the researchers found that the vaccine ramped up the number of neutrophils in babies’ blood compared with unvaccinated infants.
© TERESE WINSLOW
Monocytes from the blood displayed epigenetic changes after vaccination that opened chromatin harboring multiple genes involved in driving an inflammatory response, making them more accessible for transcription. Meanwhile, chromatin closed around genes associated with immune tolerance.
When exposed to the fungal pathogen Candida albicans in vitro, immune cells sampled from patients’ blood 90 days after vaccination released more of the cytokine interleukin 1β, which mediates inflammation, than did cells from blood drawn from the same individuals before vaccination.
Putting nonspecific effects of vaccines to the test
New studies on nonspecific effects have been published since the working group concluded, but a full picture of the phenomenon has not yet come into focus. In a 2017 study of low–birth weight infants in Guinea-Bissau, Benn and her colleagues found a 43 percent reduction in the mortality rate from infectious disease in the first 28 days of life among babies who received BCG soon after birth compared with those who got it at around six weeks of age, on the typical local schedule. Meanwhile, a study comparing the effects of live and non-live polio vaccine on toddlers in Bangladesh found that children who received the live vaccine experienced fewer episodes of diarrhea over the following year than did those who received the killed version. And this September, infectious diseases specialist Mihai Netea of Radboud University in the Netherlands and colleagues published interim results of a randomized trial of BCG in elderly people in Greece that showed that people who received the vaccine were less likely than controls to develop infections of any kind in the following year; BCG appeared to be most protective against respiratory viruses.
Other results complicate the case for nonspecific effects. In a Danish trial of BCG also led by Benn, there was no difference between babies who received the vaccine at birth and those who didn’t get it at all in terms of hospitalization rates for infection over the first 15 months of life—except among BCG-vaccinated children whose mothers had also been BCG-vaccinated at some point prior to their pregnancies; these babies had a 35 percent lower hospitalization rate than their non-vaccinated peers. And two trials of a different strain of the BCG vaccine in low–birth weight infants in India, one of which administered BCG along with live polio vaccine, found no effect on mortality in the first 28 days after birth.
See “An Old TB Vaccine Finds New Life in Coronavirus Trials”
Nonspecific effects have garnered some high-profile attention as a potential near-term tool for fighting COVID-19. Robert Gallo, best known for his role in identifying HIV as the cause of AIDS in the 1980s, joined with Aaby, Benn, and other authors to pen an opinion article in a June issue of Science calling for studies to determine whether the live polio vaccine could protect against COVID-19. (In rare cases, one of the three viral strains can mutate and cause disease, but Gallo suggests this could be avoided by only using the other two strains.) Gallo, a cofounder of both the Institute of Human Virology at the University of Maryland School of Medicine and the Global Virus Network, wasn’t successful in persuading the National Institutes of Health to back such trials, he says, but he’s working with partners on clinical studies outside the US, chiefly in India. Netea says he knows of 15 current trials testing BCG against COVID-19, and two each of the oral polio and the measles, mumps, and rubella (MMR) vaccines.
If ongoing trials demonstrate that BCG reduces the incidence or severity of COVID-19, that finding could have implications beyond the current pandemic, says Curtis. It would indicate that the inoculation could also serve as a ready-made stopgap against future outbreaks of new viruses, he notes. He adds that his current study is examining whether BCG affects rates not just of COVID-19, but also of more-familiar respiratory infections such as the flu. If so, BCG “might be something you would just use in population groups like the elderly that are more susceptible” to infection.
Potential mechanisms of vaccines’ broad protection
While observations of nonspecific effects are decades old, hints about the mechanisms that might explain the phenomenon have only begun to emerge. Vaccines are designed to train the adaptive immune system—that is, primarily the T cells that recognize and kill specific pathogens and the B cells that make antibodies—to fight off a particular infection. To do this, vaccines typically include components of the pathogen they aim to protect against, or live viruses or bacteria that are closely related to the target pathogen but don’t cause illness. So it seems counterintuitive that their effects would extend beyond that target pathogen.
Netea, who collaborates with Benn, Aaby, and Curtis, says he thinks the lack of a mechanistic explanation for the non-specific effects seen in early studies led people to “kind of forget about it.” He himself only learned about the existing literature on such effects about a decade ago, when a student working in his lab was testing the immune response to the tuberculosis-causing bacterium after BCG vaccination. The student had used exposure to a pathogenic fungus as a control, and Netea was surprised to see that BCG helped the cells drawn from human volunteers who received the vaccine mount an immune response not only against Mycobacterium tuberculosis, but the fungus as well.
While observations of nonspecific effects are decades old, hints about the mechanisms that might explain the phenomenon have only begun to emerge.
To find out why this was, Netea led a larger study in which he and his colleagues again tested the blood of healthy volunteers before and after they were inoculated with BCG. Isolating pathogen-chomping innate immune cells known as monocytes and exposing them to a variety of pathogens in vitro, the team found that monocytes drawn after inoculation released 50 percent to 100 percent more of the immune cell–stimulating cytokines TNF and IL-1β. The researchers also identified specific post-inoculation changes to the methyl and acetyl tags on histones packaging the monocytes’ DNA. The authors concluded that BCG trains innate immunity via epigenetic alterations. “It’s like putting marks on the genes that are necessary for host defense,” Netea says. “Then, when you get an infection . . . those genes are marked by certain chemical modifications of the histones, and then the transcription factors can bind easier,” ramping up gene transcription to support a more robust immune response.
His and other groups have since elucidated similar processes in other components of the innate immune system, such as natural killer cells, Netea adds. And in a study published earlier this year, he and colleagues found that BCG vaccination in adults appears to shift gene transcription patterns in immune precursor cells in the bone marrow to a program needed to form a class of leukocytes known as myeloid cells. Similarly, BCG-vaccinated newborns quickly boost production of neutrophils, a type of myeloid cell, relative to their unvaccinated peers. (See illustration.)
Why would an infant’s immune cells need to be trained, rather than having the epigenetic marks for a robust immune response from the start? Netea says he suspects that the answer to that question has to do with the requirements of human pregnancy—namely, that the parent’s body not reject the fetus. Tamping down the parent’s immune system during pregnancy could also affect the fetus’s, he suggests, leading to babies born vulnerable to infection.
That early vulnerability, counteracted by live-vaccine immune training, may help explain the dramatic lowering of all-cause mortality found with early BCG vaccination in Guinea-Bissau. But epigenetic effects don’t appear to be the full story on how BCG—and potentially other live vaccines—provide nonspecific protections, at least in neonates.
In a study published earlier this year, a team led by Kollmann and Telethon Kids Institute researcher Nelly Amenyogbe (and also including Aaby and Benn) treated some newborn mice with BCG and then injected fecal bacteria into their abdomens, and into the abdomens of control animals, to simulate sepsis. They found that BCG initiated a process known as emergency granulopoiesis, which ramped up the production of neutrophils in the bone marrow. These cells “sit in the bone marrow, waiting . . . to gobble up bacteria and many other things,” Kollmann says. “And then, the moment we induce sepsis in these mice that have received BCG, [the neutrophils are] released, go to the site of infection . . . and knock off these bacteria within hours of the infectious process starting.”
What most struck Kollmann was the timeline for the emergency granulopoiesis-based protection: the process began within hours of BCG inoculation, and the peak of the neutrophil-based protection came three days later. It was fast, and it lined up with the findings in the 2011 study that had first drawn his attention to pathogen-agnostic effects. The speed of the nonspecific protection also meshed with results from two other trials that compared the effects of giving BCG at birth versus later—namely, that the vaccine provides dramatic protection that begins almost immediately after inoculation.
Kollmann says he hopes that the finding of a mechanism to back up the clinical trial data will lead to a more uniform practice of giving BCG to babies in high-risk settings immediately after birth. Longer term, he adds, more research is needed to better understand BCG’s beneficial effects. “We need to figure out: What is it about BCG that makes this happen? Can we formulate this, can we optimize this, can we make it better, faster?”
A Nonspecific Vaccine Risk?
As a growing number of studies—epidemiological, clinical, and mechanistic—suggests that some vaccines arm the immune system for battle against pathogens other than the ones they were designed to target, a smaller and less rigorous body of evidence indicates that in some circumstances, other vaccines may do the opposite. Studies by Christine Stabell Benn and Peter Aaby of the Bandim Health Project and their colleagues point to non-live vaccines such as the diphtheria-tetanus-pertussis (DTP) shot possibly making girls—though not boys—temporarily more susceptible to illnesses other than the ones they’d been vaccinated against, Benn says. In a 2012 cohort study in Guinea-Bissau, for example, the researchers found that the group of girls who received DTP when they were two months old had more than twice the mortality rate between two and six months of age of those who hadn’t been vaccinated . But this study and others that have found a risk of increased all-cause mortality following non-live vaccines have drawn criticism for being observational, and—due to ethics concerns about delaying vaccination for experimental reasons, Benn says—no controlled trials have been published on whether such an effect exists.
In a recent study, Benn and her colleagues compared the activity of immune cells from women in the Netherlands whose most recent vaccine was DTP with those whose most recent vaccine was the tuberculosis vaccine (called bacille Calmette-Guérin, or BCG), which consists of live, attenuated bacteria and has been linked to protection against a range of infections. The cells from the former group were less responsive to non-target pathogens than the cells from women who’s had the BCG vaccine, they found, leading Benn to suspect that while live vaccines train the immune system to go after any threat more effectively, non-live vaccines train it to instead tolerate invaders other than the vaccine’s target. She doesn’t advocate for doing away with DTP and other non-live vaccines, but argues that girls, at least, should receive a live vaccine shortly after non-live jabs to retune their immune systems. “What all these studies suggest is that it would be very wise to design different vaccination programs for boys and girls,” she says.
Such potentially deleterious nonspecific effects received some attention last year with the rollout of a new malaria vaccine in a pilot program in three African countries, Malawi, Ghana, and Kenya. Benn and Aaby have voiced concerns that the vaccine, which is non-live, could increase all-cause mortality rates in girls, and they point to a post-hoc analysis of data from a Phase 3 trial that showed a higher mortality rate in infant girls who received the vaccine compared with those who didn’t. Kate O’Brien, the director of the World Health Organization’s (WHO) Department of Immunization, Vaccines and Biologicals, says that study wasn’t designed to test all-cause mortality, and notes in an email to The Scientist that “the female mortality in the [malaria] vaccine arm was similar to male mortality in the control and . . . vaccine arms.” The finding of higher all-cause mortality among girls was likely due to chance, she says. Nevertheless, the WHO is monitoring all-cause mortality in the ongoing trial of the vaccine, and so far the data indicate that there is no mortality increase associated with it, O’Brien notes.
Correction (November 13): Incorrect labels of the open and closed chromatin in the graphic have been corrected. The Scientist regrets the error.