Read our COVID-19 research and news.

An antibody (red/pink) latches on to the surface protein of SARS-CoV-2 (green). Many scientists believe the level of so-called neutralizing antibodies triggered by a vaccine predicts how well it protects.


Can immune responses alone reveal which COVID-19 vaccines work best?

Science’s COVID-19 reporting is supported by the Heising-Simons Foundation.

Other than running a placebo-controlled, clinical trial lasting many months and involving tens of thousands of people, is there any way to be sure a COVID-19 vaccine will work? Many researchers contend that the success of several vaccines now widely in use offers a shortcut: Simply gauge a vaccine’s ability to elicit so-called neutralizing antibodies, which bind to the virus and prevent it from entering cells. But several recent studies, the latest published as a preprint on 24 June, point to other “correlates of protection”: “binding” antibodies—which latch on to the virus but don’t block entry—and another set of immune warriors called T cells.

Vaccine decisions may soon depend on a better understanding of these supporting actors. Several companies are developing updates of their COVID-19 vaccines tailored to protect against new viral variants, and they hope regulatory agencies won’t require that they show efficacy in big clinical trials, which are not only time-consuming and expensive, but also increasingly ethically fraught because some of the participants receive a placebo even though proven vaccines are now available.

With an established correlate of protection, trials can give an updated vaccine to a much smaller group of participants and then check whether they produce the telltale immune responses. (That’s how the annual updates of flu vaccines are approved.) Health officials may also turn to correlates when they consider prioritizing existing COVID-19 vaccines, authorizing new “mix and match” combinations, or even when making decisions about entirely new vaccines.

But finding robust correlates has been challenging. During the megatrials that led to the authorization of COVID-19 vaccines, investigators monitored antibody responses and tried to correlate them with their odds of participants getting sick. Different trials, however, used different antibody assays and different definitions of mild COVID-19, the main endpoint in the trials. “It’s anarchy because it’s always been anarchy,” says John Moore, an immunologist at Weill Cornell Medicine. “You’re dealing with different academic labs and different companies, and companies tend not to talk to each other.” Many trials also lacked the statistical power to measure protection from hospitalization and death, arguably a COVID-19 vaccine’s most important task. And few trials even looked carefully at T cells, which are far more cumbersome to measure.

Still, two studies—first published as preprints in March here and here—confirmed the prediction by Moore and many other scientists that neutralizing antibodies (“neuts”) play a key role. To “normalize” the different assays used in the trials, they compared levels of antibody elicited by each vaccine with antibodies found in people who naturally became infected in the trial’s placebo group. In both analyses, the vaccines that triggered higher levels of neuts than the ones typically seen in recovered people offered the best protection—strong evidence of a correlation, Moore says.

“That’s a great relief to me,” says Penny Moore, a virologist at the National Health Laboratory Service in South Africa, who helped measure protective immune responses in different vaccine trials and was “really puzzled” by the results. But she and others suspected neuts are far from the whole story. “I just cannot work out for the life of me how much [other immune responses] are contributing and where they’re contributing,” she says.

During the efficacy trials of the messenger RNA (mRNA) vaccines made by the Pfizer-BioNTech collaboration and Moderna, for example, the first shot of both vaccines triggered barely measurable levels of neutralizing antibodies, but still offered substantial protection. “It suggests there’s more than neutralizing antibodies going on here,” says David Montefiori, an immunologist at Duke University who runs a lab that measures neuts for a handful of COVID-19 vaccine trials sponsored by the U.S. government. Neuts skyrocketed only after the second mRNA shot, when protection rose to more than 90%.

T cells, which coordinate the B cells that produce antibodies but also clear infected cells when neuts falter, appear to bolster the defense. In a study published in February that included 12 patients whose COVID-19 ranged from mild to fatal, a team led by immunologist Antonio Bertoletti of the Duke–National University of Singapore Medical School reported that patients who early on had the highest levels of immune system messengers that kick T cells into action—an indirect, but relatively simple, way to measure their presence—had milder disease because they cleared the infection faster.  

Penny Moore and colleagues also found support for a role for T cells. In an 11 June preprint, they reported that 96% of participants in an efficacy trial of the COVID-19 vaccine produced by Johnson & Johnson (J&J) made antibodies that neutralized a viral strain from early in the pandemic but only 19% had antibodies that neutralized the Beta variant, which is widespread in South Africa and infamous for dodging neuts. Yet despite the variant, the vaccine remained protective against moderate and severe COVID-19. “I think it’s entirely plausible … that T cells are doing something really useful here,” Penny Moore says. A monkey study with this vaccine, published in Nature last year, also showed that T cells contributed substantially to control of the virus if neut levels weren’t high enough to do the job.

The immune system figures out how to use all the weapons at its disposal.

Shane Crotty, La Jolla Institute for Immunology

Binding antibodies may also be more important than researchers assumed. The 24 June preprint, by researchers from the University of Oxford, found that high levels of neuts correlated with the 80% protection achieved 28 days after participants in the United Kingdom received two shots of the vaccine the team developed with AstraZeneca. But digging more deeply into the data revealed that binding antibodies were as good as a correlate—if not better.

It’s not clear exactly why, because binding antibodies don’t directly block the infection process. One possibility is that they make the virus more susceptible to being gobbled up by macrophages or other cells that ingest intruders. This mechanism, called phagocytosis, protected children from severe COVID-19, immunologist Galit Alter of the Ragon Institute of MGH, MIT and Harvard reported in Nature Medicine in March. Then again, it may be that binding antibodies are produced in lockstep with neuts, but at higher levels, and are simply a surrogate marker.

Work by virologist Shane Crotty and Alessandro Sette of the La Jolla Institute for Immunology has shown that people handle SARS-CoV-2 most effectively if they have T cells and antibodies working in sync, as they showed in a study of the immune reactions of 24 COVID-19 patients whose disease ranged from mild to fatal. “The immune system figures out how to use all the weapons at its disposal,” Crotty says.

South Africa, which has fewer than 1% of its population fully vaccinated amid an exploding epidemic, has shown the potential pitfalls of overemphasizing neuts. In February, the country abandoned the AstraZeneca-Oxford vaccine after it had a disappointing 22% efficacy against mild disease in a large trial. Test tube analyses seemed to support the decision: Antibodies triggered by the vaccine had far less neutralizing power against the Beta variant, which then accounted for nearly all infections. But Penny Moore’s study of the J&J vaccine has subsequently shown that disappointing levels of neutralizing antibodies don’t keep a vaccine from providing good protection against severe disease. “Our obsession with neuts may mean that we missed an opportunity here for AstraZeneca,” she says.

Other scientists counter that it makes sense to use neuts as a gauge to rank the relative powers of different vaccines, but acknowledge that this will require standardized assays. Chinese researchers in the 23 June issue of Vaccine published national standards for SARS-CoV-2 neutralization assays. “This has not been the most important priority, but it’s going to become one if we move away from phase 3 trials,” John Moore says.

With the picture still muddy, regulators need to decide whether correlates of protection should offer vaccinemakers a shortcut to bringing improved products to market. Pfizer and Moderna are developing next generation candidates designed to create high levels of neutralizing antibodies against the Beta variant, and the U.S. Food and Drug Administration (FDA) has signaled that it will accept this correlate of protection for approval decisions. “Even though we might not get the perfect surrogate—it might mediate partial protection—that could be good enough,” says Peter Gilbert, a biostatistician who designs clinical trials at the Fred Hutchinson Cancer Research Center. “We don’t need perfection here.”

But Alter worries regulators relying on neuts might approve unnecessary booster shots simply because they outdo existing shots on that measure. “If [regulators] don’t adapt, we’re going to end up overboosting, and we’re going to be making the drug companies really happy,” she says.

It’s also unclear whether a convincing correlate from a vaccine that uses, say, mRNA, applies to one that uses a different platform like a viral vector. “We’re hoping to have more immune correlate of protection information before updates on that,” says Peter Marks, who heads FDA’s vaccine division.

With more than a dozen vaccines now in use, that information may arrive soon, Sette says. Although companies typically control the data from clinical trials, academic labs can now compare recipients of different vaccines, he says. “In the next few months, all the different labs will be generating analyses of what different vaccines do and a large amount of data will be generated in academic labs,” Sette says. “There’s going to be a fundamental wealth of information.”