Future Tense

What We Really Know About the Risk of Coronavirus Reinfection

A woman coughs into her elbow with a visualization of the SARS-CoV-2 virus behind her.
Photo illustration by Slate. Photo by Bertrand Blay/iStock/Getty Images Plus and dragana991/iStock/Getty Images Plus.

Since the beginning of the coronavirus pandemic, there have been anecdotal reports of COVID-19 patients becoming reinfected with SARS-CoV-2. Until this week, there was no concrete evidence demonstrating that SARS-CoV-2 reinfection could occur. A man who previously recovered from COVID-19 in March tested positive at an airport screening site this month while returning to his home in Hong Kong. According to a paper from researchers at Hong Kong University, viral genome sequencing showed that he was infected with two different genetic variants of SARS-CoV-2 and was therefore authentically reinfected, although the authors have not presented data to rule out contamination, which can readily occur in labs that process many clinical samples that may all contain different variants of the virus. (The paper was recently accepted for publication in Clinical Infectious Diseases.) Since then, two more cases of reinfection have been reported in the Netherlands and Belgium, although no data has been made available to confirm these findings.

But this is not cause for alarm. You cannot tell whether something is going to have a major impact on the pandemic by looking at a solitary case, or even a handful of cases. Maria Van Kerkhove, head of the emerging viruses unit at the World Health Organization, emphasized that this needs to be studied at the population level before making any broader conclusions. Currently there’s very little evidence to suggest that reinfection is common.

So what can we learn about reinfection from these case reports and the relatively scant data they provide? Not much. But now we can begin to see how it aligns with other evidence about reinfection and immunity to SARS-CoV-2 and develop hypotheses that can then be tested in more rigorous clinical studies or in experimental models.

Before we get into the details of the study, we need to go over some basics. (Bear with me.) Antibodies are proteins that bind specific antigens, or viral proteins that are recognized by the immune system. When an antibody binds its antigen and renders the virus noninfectious, it is referred to as a neutralizing antibody. There are several different types of antibodies, but those that are most likely to be neutralizing are called IgG, so these are most commonly measured to see if someone has developed neutralizing antibody responses. Serum is the liquid component of blood that contains circulating antibodies. The concentration of antibodies in the blood is referred to as the titer.

Most commonly, a PCR test is used to diagnose SARS-CoV-2 infection, which specifically detects and amplifies a small piece of the viral genome directly. But antibody tests, which are what’s most relevant here, quantify antibody titers. Some can also determine how many neutralizing antibodies in a serum sample can inactivate either authentic SARS-CoV-2 or a surrogate virus.

Now that the immunology lesson is out of the way, we can look at several interesting observations in the Hong Kong case. One was that the patient did not have detectable serum IgG antibodies 10 days after symptom onset. This suggests that the first infection, in which the patient reported mild COVID-19 symptoms, did not trigger a particularly robust neutralizing antibody response. This is consistent with reports that antibody titers correlate with disease severity—the bigger the viral load, the sicker the patient and the bigger the antibody response—and may suggest that either the patient didn’t mount an immune response at all, or at least mounted weak antibody responses below the detection threshold for the test that was used. The absence of detectable serum antibodies does not necessarily demonstrate absence of functional immune memory, as antibodies are not the sole indicator of immunity.

This leads to a second interesting observation: The patient was completely asymptomatic throughout the course of the second infection, and afterward, he had detectable IgG antibodies to the SARS-CoV-2 nucleocapsid protein. This could be explained by the presence of immunological memory, or immune cells that “remember” the virus from the first infection, even if antibody titers were too low to detect using a lab test. In this scenario, the immune memory wasn’t robust enough to provide sterilizing protection and completely block infection, but it was sufficient to reduce disease pathology. This is consistent with what we know about how the immune system works: After recovery from an initial infection, antibody titers normally decrease to a baseline level, which may be too low to detect, although the memory B cells that make antibodies are likely still present. Akiko Iwasaki, a virologist and immunologist at Yale who studies SARS-CoV-2, called it “a textbook example” of a functional immune system responding to reexposure of a prior pathogen. Even if preexisting immunity doesn’t completely protect against infection, it is still thought to reduce disease severity.

This type of nonsterilizing immune protection against disease has been observed in all of the studies testing vaccines in rhesus macaques: Not all the vaccinated animals were protected against infection, but they were much less sick, at least in terms of their clinical presentation, laboratory values, respiratory rates, and tissue pathology, compared with control animals. None of this is surprising, nor does it suggest that reinfection is going to be a huge problem for public health, at least by being frequently associated with severe COVID-19. In fact, it doesn’t tell us much at all because you can’t definitively investigate what happened in a single case study in a rigorous or controlled way.

The most important question for reinfection, with the most serious implications for controlling the pandemic, is whether reinfected people can transmit the virus to others. The Hong Kong study did not measure infectious virus shed after reinfection, so it’s not possible to determine whether the patient might be capable of spreading the virus to others based on these data. The patient was isolated after testing positive, so there’s no epidemiological evidence to suggest that this patient transmitted SARS-CoV-2 to anyone else during his secondary asymptomatic infection. If reinfected patients can spread the virus, it could increase community transmission if those cases are not detected through testing. The ability to detect those cases is likely going to further decline due to new Centers for Disease Control and Prevention guidance that people should not seek testing unless they are symptomatic, even if they have a known exposure to a confirmed case. That’s why it’s crucial to make a more informed estimate of the prevalence of reinfection. If reinfection is a rare event, then it is likely to have less of an impact on transmission within communities.

Not observing reinfection prior to this point may mean that either it is rare or it is rarely detected. There have been multiple reports of reinfection throughout the pandemic, but these have been either anecdotal or inconclusive. By February, there were multiple reports of “repositive” patients who tested negative after recovering from COVID-19 and then subsequently tested positive again. Most of these repositive cases were thought to be the result of residual viral RNA, the SARS-CoV-2 genetic material that is detected by the test but does not always correspond to the presence of infectious virus and can persist for weeks after recovery. This was confirmed by a study conducted by the South Korean CDC in May on more than 200 repositive cases, which demonstrated that repositive patients were not associated with new transmission clusters, nor were they shedding infectious virus, and were therefore likely not infected at all despite testing positive. In the Hong Kong case, reinfection was confirmed by sequencing the full viral genome rather than detecting a small piece of genetic material using the diagnostic test, and observing that the genome sequences in their entirety were distinct during each episode of infection.

There are far fewer reports of people testing positive more than four months past recovery, although that may be because people who have recovered from COVID-19 are less likely to seek testing months after their symptoms resolve or they test negative, or if they have had an antibody test and are seropositive. However, if reinfection commonly causes severe COVID-19, no evidence of that has been found to date. With more than 24 million SARS-CoV-2 cases worldwide, it’s likely that reinfection is probably rare if it hasn’t been observed until now. We also can’t exclude the possibility that this sample may be a false positive resulting from contamination of either sample, in which case we are still waiting for a confirmed report of bona fide reinfection.

Going forward, if we are to understand anything about the implications of reinfection for transmission and immunity, we must design studies that can address these questions. We need to understand how common it is and whether reinfected patients develop severe COVID-19, we need to determine the relationship of reinfection to immunity, and we need to assess whether these patients are capable of shedding infectious virus and transmitting SARS-CoV-2 to others. That information can’t be obtained from a single case. Until we have evidence that reinfection is something to worry about, we should instead focus on reducing the transmission risk for everyone in the community—regardless of whether they have already been infected with SARS-CoV-2.

Future Tense is a partnership of Slate, New America, and Arizona State University that examines emerging technologies, public policy, and society.