The first wave of Covid-19 tests asked, Do you have the SARS-CoV-2 virus right now? But as the world starts to eye an exit from lockdown, it needs tests that can answer a different question: Did you have the virus in the past? If so, a person should have antibodies: proteins in the blood that will recognize and grab onto the virus if it comes back, and — hopefully — block reinfection.

Testing for antibodies could answer key questions about the virus, helping epidemiologists track exposures and potentially leading to the issuing of “immunity passports” for people who’ve beaten back the infection and can return to work. “I think you’re going to see [antibody testing] explode,” says allergist Mitchell Grayson, chief of allergy and immunology at Nationwide Children’s Hospital and Ohio State University in Columbus.

But even though it’s likely that someone carrying antibodies to SARS-CoV-2 would be protected from repeat infection, scientists don’t yet know that. The World Health Organization recently warned against issuing immunity certificates based on antibody tests for that reason. And the tests themselves are plagued with high rates of inaccuracy. The United Kingdom, for example, paid $20 million for antibody tests that Oxford University later deemed too insensitive to detect antibodies in all but the sickest patients.

In the United States, eight organizations had received emergency use authorization from the Food and Drug Administration for antibody tests, as of April 30. But the FDA announced in March that test makers could sell Covid-19 diagnostics as soon as they’re satisfied with their own test and have notified the FDA, before the agency has reviewed their data. Dozens of tests have come out under these guidelines.

Here’s some of what we know about antibody tests for Covid-19, and what scientists still need to find out.

How does an antibody test work?

It’s a different kind of test than the ones used to diagnose an active SARS-CoV-2 infection, which detect viral genetic material in a person’s nose, throat or saliva.

Antibody tests, in contrast, look for the presence of antibodies against the virus in people’s blood — a sign that they have fought off the infection in the past. And these kinds of tests are much more difficult to get right, says Linfa Wang, a virologist and director of the program in emerging infectious diseases at the Duke-National University of Singapore Medical School.

Most of the existing antibody tests use an engineered piece of the virus — frequently based on the aptly named “spike protein,” which sticks out from the surface of the virus. During infections, the virus uses the spike protein to gain entry to cells, by attaching to cell-surface molecules called ACE2 receptors. Presumably, people who have been infected with the virus will generate antibodies that bind to the spike and other parts of the virus. Thus, some antibodies should also stick to the engineered protein piece in the antibody test.

Often, these antibodies are detected via a test called an ELISA (enzyme-linked immunosorbent assay), which is typically run in a lab, taking a few hours. An alternative test, using the same principles, takes place on a paper strip, much like a pregnancy test. This test, called a lateral flow assay, can be used in a doctor’s office, with results in minutes. But the World Health Organization has warned that the faster tests are less accurate.

Close-up photograph of a hand wearing a blue glove. Underneath that person’s hand is the finger of an individual who is getting the test. The finger has been pricked and blood is being drawn into a small tube. To the left is the small testing strip to which the blood will be added.

A medical technician draws blood for a fast antibody test that will screen for two types of antibody — IgG and IgM — that react to portions of the SARS-CoV-2 virus. The blood is added to the white strip seen on the left. Such tests usually contain a piece of the spike protein that sticks out from the surface of the virus. A positive result means the blood contains antibodies that will bind to the spike protein, presumably because the person has been infected by the virus in the past. However, antibody tests — especially the rapid ones done at doctors’ offices — are far from perfectly accurate.

CREDIT: REUTERS / ADREES LATIF

How accurate are the tests?

A good test has two qualities. It should be sensitive, meaning it accurately identifies true positive signals. In this case, that means it picks up almost everyone who has SARS-CoV-2 antibodies. The first commercial antibody tests to gain the FDA’s nod — from Cellex, Ortho Clinical Diagnostics and Chembio Diagnostic Systems — have sensitivity rates from 94 to 100 percent. The higher the sensitivity, the less likely a test is to give a false negative result — saying someone doesn’t have antibodies when they really do. A person may also get a false negative result if they’re tested too early, before antibodies begin to accumulate around one to two weeks into an infection.

A good test should also be specific, meaning it correctly identifies people who don’t have the antibodies. For those first three commercial tests, specificity ranged from 83 to 96 percent. If a test gives a false positive — indicating someone has antibodies when they in fact don’t — a person might think they’re safe when they’re not.

Part of the challenge with specificity in this context is that SARS-CoV-2 is not the only member of the coronavirus family to infect human beings. Two other coronaviruses, SARS and MERS, caused earlier multi-country outbreaks. Moreover, there are at least four coronaviruses that cause a fraction of cases of the common cold.

Many people have been exposed to these common-cold coronaviruses, and could be carrying antibodies against them. So any Covid-19 test must be able to distinguish between the novel coronavirus and the more common, less dangerous ones.

A lot of the tests being used right now fail to do so, says Sylvia Daunert, a bioanalytical chemist and director of the JT Macdonald Foundation Biomedical Nanotechnology Institute at the University of Miami in Florida. Daunert coauthored a paper on the use of antibody-like molecules in biotechnology for the Annual Review of Analytical Chemistry in 2017.

Several organizations, including the FDA, the Centers for Disease Control and Prevention, and the National Institutes of Health, are now working to validate some tests.

If an accurate antibody test is positive, does that mean the person is protected?

With SARS-CoV-2 so new, it’s hard to be sure. Not all infections create immunity; tetanus, for example, often doesn’t.

Animal research does suggest that SARS-CoV-2 antibodies may appear and confer some immunity. Studies with the coronavirus that caused the 2003 SARS outbreak found that mice developed protective antibodies. And in a recent study, not yet reviewed by other scientists, researchers in Beijing infected two monkeys with SARS-CoV-2. After the monkeys recovered, the scientists tried to re-infect them, but the animals did not sicken again — suggesting that immunity is possible.

A large red blob (molecular structure of the spike protein) is attached at one side to a large blue blob (with the characteristic molecular shape of an antibody). The red blob is protruding through a linear orange structure that is the lipid membrane of the virus.

Three-dimensional rendering of an antibody (blue) binding to the coronavirus spike protein (red), which is embedded in the viral membrane (orange).

CREDIT: JUAN GAERTNER / SHUTTERSTOCK

On the other hand, reports out of Asia suggest that some people can get Covid-19 twice: Some people have tested positive for viral genes, then negative, then positive again. However, many virologists believe this might be due to faults in the viral genetic tests; these, like antibody tests, are far from perfect. One such test was reported to have a 15 percent false negative rate.

Complicating matters further, says Daunert, one person’s immune system is not like another’s. Some people will produce tons of antibodies. Others, such as those who are on immunosuppressive treatment due to another condition, might make fewer antibodies and for only a short time. A mild case of Covid-19 might not yield abundant, long-term antibodies. In China, one-third of survivors had very low antibody levels that might not protect them in the future.

The nature of the antibody tests also complicates the answer. ELISA and lateral flow assays identify any antibody that can stick to the viral protein fragment. But what they don’t reveal, Wang says, is whether those antibodies can block infection.

Antibodies that can actually prevent the virus from entering cells, called neutralizing antibodies, are the most important ones. Detecting those is an involved process: Technicians culture cells along with the virus in a lab, then add a sample of a person’s blood to see if any antibodies protect the cells. This takes a few days, costs much more and must be done by trained personnel wearing protective gear in certified labs.

How do scientists and health officials want to use antibody tests?

The expanding availability of tests will help researchers start to answer many questions. “The thing about Covid-19 is, we still do not know exactly when immunity starts developing, and then, how long after it has developed does immunity persist,” says Chockalingam Palaniappan, chief innovation officer at Ortho Clinical in Rochester, New York. “I think these tests will allow you to measure it.”

One issue is quite immediate: Physicians hope that they can transfer the antibodies from a survivor to a person with current Covid-19, so that those antibodies could attack the virus. Antibody testing would indicate which people would make the best donors. For these antibodies, or immunity certificates, Wang says he’d only trust a cell-culture, neutralizing antibody test.

Antibody tests are also of great use in epidemiology, says Daunert. The fact that many people carrying the coronavirus exhibit no symptoms makes it difficult for epidemiologists to track its spread. Knowing how many people have been infected would help leaders make decisions about lifting or relaxing lockdown orders.

Moreover, knowing the true infection rate would improve estimates of Covid-19’s fatality rate. According to global data from the Johns Hopkins University of Medicine, about 7 percent of confirmed cases have led to fatalities as of April 30. The mortality rate in the 2003 SARS outbreak was 10 percent. But if more people have been infected with SARS-CoV-2 than are counted in current tallies, that death rate might be lower.

Antibody testing could also help solve the mystery of which kind of animal SARS-CoV-2 came from before it jumped to people, perhaps in a wet market in Wuhan, China. Wang is developing a test to check for antibodies in diverse creatures.

What have scientists learned so far from antibody tests?

Thus far, antibody test-based studies in all but the hardest-hit places have suggested that fewer than 10 percent of people have been infected — putting populations nowhere near the 70-plus percent “herd immunity” rate that would help protect still-vulnerable individuals.

In a recent study, Stanford University researchers used antibody tests to estimate that as many as 48,000 to 81,000 people in Santa Clara County, California, had been exposed to SARS-CoV-2 by early April, even though the county had just 1,100 confirmed cases at the time. Similarly, researchers in Los Angeles County, using the same test donated by a company in Minnesota, announced that 221,000 to 442,000 adults had been infected by early April, when fewer than 8,000 cases had been reported.

Those data seem to indicate that 4 to 5 percent of people in those places had already been exposed, giving Covid-19 a fatality rate closer to 0.1 to 0.2 percent. But those studies, which have not yet been reviewed by other scientists, have come under fire, partly due to the high rate of mistaken results in the test, which has not undergone FDA review and has a known tendency toward false positives.

If a test gives a false positive — indicating someone has antibodies when they in fact don’t — a person might think they’re safe when they’re not.

New York made its own test and screened 3,000 grocery shoppers to calculate that as many as 2.7 million people — 14 percent of the population — in New York City had been exposed. That would make the mortality rate about 0.5 percent. But that test had a specificity somewhere between 93 and 100 percent, and the data did not include people who don’t go out right now, such as many elderly or immunocompromised individuals and children.

What are the biggest questions about SARS-CoV-2 antibodies?

Assuming that antibodies are generated, what will the immune system do upon re-exposure to SARS-CoV-2? For now, it’s a mystery. The hope is that those antibodies will launch an immune attack on the virus before it gets a foothold, so a person would get a lower degree of illness, or none at all.

But there’s another, unpleasant possibility, Wang says. If the antibodies stick to the virus but don’t neutralize it, then they can create a phenomenon known as antibody-dependent enhancement. Because cells have surface receptors for antibodies, the binding can actually create a second entryway for the virus to infect cells, making the infection worse.

The next question is, how long do these antibodies last? Some infections, like measles or chicken pox, produce long-lasting immunity. Others, not so much. Based on experience with influenza and other RNA-based viruses, Grayson predicts that SARS-CoV-2 immunity might last “about a year, possibly longer.”

In the case of the original SARS virus, one study indicated that most people who were infected kept their antibodies for two years, but by the third year, half of them were antibody-negative. However, a recent, as-yet-unreviewed study suggested that SARS antibodies could last for 12 years.

Reliable testing should — with time — answer these key questions. “With all these tools, in combination,” says Palaniappan, “I think we can get a pretty good handle on the disease.”