As the COVID-19 pandemic advances, it’s become increasingly clear that national tallies of case numbers and deaths are only telling part of the underlying dynamics of the epidemic.
COVID-19 caught health systems around the world unawares, and the development and distribution of testing lagged behind the ferocious spread of this virus.
Research is now catching us back up and trying to understand retrospectively what has happened in our communities. However, in resource-poor health systems, this is a particular challenge, as virus testing may not have been achievable.
Fortunately, our body’s response to all infections – whether it’s a virus, bacterium or parasite – has some consistent characteristics.
And our team have been adapting lessons learnt in the decades-long fight to control malaria to now enhance surveillance for COVID-19.
Undetected COVID-19 cases
Early in the epidemic, when many countries’ diagnostic and clinical services were overwhelmed by the rapidly rising number of COVID-19 cases, only people with significant clinical symptoms or clear risk factors were tested for the SARS-CoV-2 virus.
Mild and asymptomatic cases were either not tested at all or may have returned negative test results because of late testing when the viral ribonucleic acid (known as RNA) load had already decreased below the test’s detection limits.
As diagnostic and contact tracing capacities ramped up and the epidemic curve started to bend, increased testing revealed large numbers of mild and even asymptomatic SARS-CoV-2 infections. This tells us that a significantly larger proportion of populations may have been exposed than previously thought.
How much this burden is being underestimated will vary greatly between countries.
In countries like Australia, South Korea or Germany, where testing rates were and remain high, we have a much better understanding of the extent of population exposure than in countries like Italy, Spain or the USA where health systems were overwhelmed by the virus with long delays in increasing testing capacity.
The epidemic curve is least understood in low and middle-income countries (LMICs) with weak health systems and limited or no pre-existing molecular testing capacity. There is growing evidence that the epidemic progression in many LMICs is different from that observed in wealthier countries.
However, the lack of good surveillance data makes it very hard to understand the combined effects on viral transmissions of differences in population age structure, settlement and mobility patterns and climate.
It also makes it very hard to accurately predict the future course of the epidemic in much of the world.
Detecting past infections
Our immune systems are finely tuned to fight a variety of invading pathogens, but they do this with a fairly limited number of weapons.
Antibodies – proteins that are released by immune B cells and circulate in the blood for years – are an important arm of our immune defence.
Specific antibodies are produced to particular ‘antigens’ contained in a pathogen, and the antibodies present in our blood give a detailed history of our disease exposure and the vaccinations we may have received. They also give information about how protected we are against future infections.
Determining the prevalence of anti-SARS-CoV-2 antibodies in a population, known as sero-surveillance, may provide a way to gain a more complete understanding of the progression of the COVID-19 epidemic.
As anti-SARS-CoV-2 antibodies only appear seven to 10 days after symptoms they are not suitable for clinical diagnosis.
However, as even mild and asymptomatic infections result in sero-conversion (this is the generation of detectable antibodies), serology is better able to provide an in-depth understanding of COVID-19 transmission, as well as a more accurate estimation of case fatality rates, and an assessment of population-level immunity.
Good quality sero-surveillance requires not only representative, unbiased population sampling, but also low-cost and high-throughput tests with both high sensitivity (greater than 95 per cent) and excellent specificity (approximately 99 per cent).
This isn’t easy to achieve.
While antibody responses to an infection are long-lasting, antibody levels vary greatly over time. After a rapid rise, responses peak at two to four weeks after infection and then decay in two phases.
Firstly, they decay rapidly over the first three to four months. This is followed by a much slower decay over the following months and years. These changes over time mean that serological tests achieve their best sensitivity in the relatively short period of peak antibody levels, but have substantially lower performance several months after an infection.
A similar pattern of antibody level change over time is observed for P. vivax malaria, and we have harnessed this to develop a sero-surveillance tool able to identify individuals infected with P. vivax in the previous nine months.
This is an important advance because it enables people with parasites hidden in their liver to also be detected, for the first time.
Lessons from malaria sero-surveillance
Malaria is a parasitic disease that is prevalent in many countries around the world – many of which are LMICs.
Our extensive studies of serological markers of exposure to Plasmodium vivax malaria infections, published recently in Nature Medicine, offer important lessons for the development of assays for COVID-19 sero-surveillance.
Firstly, the choice of antigen matters. This is what the antibodies bind to. For our malaria assay we screened over 340 antigens to arrive at a panel of eight best performing antigens.
Secondly, combining antibody measurements to more than one antigen results in both increased sensitivity and specificity, with performance increasing notably for combinations of up to five antigens.
Thirdly, taking the kinetics of antibody responses into account, it is possible to develop more complex antibody signatures that can pinpoint when exposure had taken place.
And lastly, it is possible to build a reliable, cost-efficient high-throughput testing platform for sero-surveillance.
We are now applying the same design principals and technology to develop a high-performance assay for COVID-19 sero-surveillance. Our team has started to screen a large number of SARS-CoV-2 protein fragments to select the best performing antigens.
In a first pilot assay (or test) with four SARS-CoV-2 proteins, we were able to identify people with previously confirmed SARS-CoV-2 infections with a sensitivity of 96.4 per cent and a specificity of 99.1 per cent.
When fully developed our assay will measure antibodies to five different SARS-CoV-2 antigens together with a panel of antigens from seasonal coronaviruses (milder coronaviruses that circulate regularly in the community) to control for potential cross-reactivity.
We think this test will have the most benefit for LMICs, allowing a snapshot of a community’s exposure – and immunity – to COVID-19.
In malaria-endemic LMICs, we can combine the COVID-19 assay with our malaria antigens. This would mean countries not only getting accurate data on COVID-19, but also being able to monitor potentially detrimental effects of the epidemic on other important infectious diseases.
The advances we made over the last six years in developing sophisticated sero-surveillance assays for P. vivax malaria have put us in an excellent position to rapidly respond to the extraordinary challenges posed by COVID-19.