Infection, immunity, and surveillance of COVID-19

Dr Amitabh Suthar and Dr Christopher Dye give their perspective on infection, immunity and surveillance of COVID-19.

provide some evidence that stringent public health and social measures limited SARS-CoV-2 transmission, as reflected by lower seroprevalence rates. Their data also reinforce concerns about inequitable access to vaccines: Seroprevalence changes due to vaccination were more common in high-income countries while seroprevalence changes due to infection were more common in low-and middle income countries. The data also point to uneven access to health services and diagnostics because the ratio of infections to reported cases was high in resourceconstrained regions of the world, particularly Africa. But the analysis by Bergeri and colleagues also poses questions about the current and future value of serosurveillance for SARS-CoV-2 and other emerging pathogens. We comment on 3.
The first concerns the precision of the serological assays used. At their core, accurate measures of seroprevalence depend on having antibody tests with high sensitivity and specificity. On sensitivity, Bergeri and colleagues found that seroprevalence was relatively low in children less than 10 years old. Perhaps children were less frequently exposed to infection; but low prevalence might also be explained by the milder infections experienced by children, which perhaps stimulated weaker antibody responses and more false negatives. Antibody titers also tend to be lower in asymptomatic cases, a proportion of which may never become positive during the course of infection [7]. Another challenge to serosurveillance is that infection can be confounded by vaccination. Bergeri and colleagues countered this by using antinucleocapsid (N) antibodies to measure infection in countries where vaccines using only spike (S) protein antigens, i.e., mRNA vaccines, where delivered. However, in many low-and middle-income countries inactivated vaccines, such as Sinovac's CoronaVac, Sinopharm's BBIBP-CorV, or Bharat Biotech's BBV152 COVAXIN, are also delivered [9]. Inactivated vaccines elicit both anti-S and anti-N responses and therefore antinucleocapsid (N) antibodies would not differentiate between infection and vaccination. In these countries, seroprevalence measurements had to be adjusted using accessory data on the fraction of people vaccinated. Given the challenges of tracking vaccinations administered, this may have biased estimates.
Second, serosurveillance has limited utility in tracking rapidly spreading infections. Point seroprevalence is an aggregate between seroconversion and seroreversion [10]. For SARS--CoV-2, the median time from exposure to seroconversion is about 3 weeks; the time to reversion is about 25 weeks [11]. So serosurveillance captures neither recent infection nor past reversion (Bergeri and colleagues did not allow for reversion in their estimation of seroprevalence). In a rapidly growing epidemic with a doubling time of less than 1 week [12], seroprevalence lags far behind the spread of infection. In general, failing to allow for antibody dynamics will typically underestimate the cumulative prevalence of infection. In the extreme, if serological surveys are spaced too far apart, they could entirely miss explosive, short-lived outbreaks of disease (or waves of transmission).
Third, Bergeri and colleagues argue that anti-SARS-CoV-2 antibodies are highly predictive of immune protection, as stated in WHO guidelines [8]. However, the detection of antibody does not guarantee immunity, whether it be protection from SARS-CoV-2 infection or from COVID-19 illness and death, nor does the absence of antibody reliably indicate susceptibility to infection or disease. The relationship between antibody and protection against SARS-CoV-2 or COVID-19 requires quantitative calibration [13,14], recognizing that protection depends both on humoral (antibodies and memory B cells) and cellular immunity (T cells) [15]. The calibration is necessarily different for infection and disease, and no general rules yet exist. It is telling that just 6 (0.6%) of the serological studies described by Bergeri and colleagues were based on tests that detect neutralizing antibodies-the antibodies that are most closely linked to functional immunity. ABS contributed to this article in his personal capacity. The views expressed are his own and do not represent the views of the Centers for Disease Control and Prevention or the United States government.

Future of serosurveillance
Bergeri and colleagues have shown how serosurveillance can help to characterize nearly 3 years of the COVID-19 pandemic. They do not discuss, either on technical grounds or with respect to the limited financial resources of many national health services, how to prioritize serological surveys alongside other key elements of disease surveillance systems and health system strengthening. While core surveillance systems serve priority objectives (Table 1), WHO gives serological surveys a limited role during COVID-19 outbreak investigations, tracking infection, and retrospectively measuring the attack rate or the size of an outbreak [3]. Furthermore, serosurveillance is not considered to be a source of information to guide public health and social measures [4]. As we learn how to safely live with SARS-CoV-2, the experience that lies behind nearly a thousand serological surveys will be valuable in updating WHO guidance on the role, requirements, and use of serosurveillance data for SARS-CoV-2 and future health emergencies. Those updated recommendations should inform the decision of whether and how to invest, as Bergeri and colleagues propose, in "a global system or network for targeted, multi-pathogen, high-quality, and standardized collaborative serosurveillance" to monitor COVID-19 and other emerging pathogens.