Antibodies Fade Fast: Malawi Study Maps Hidden Waves of COVID-19 Infection

A large serological study from Malawi has revealed just how quickly immunity to SARS-CoV-2 fades after infection, and how much of the pandemic’s true course went unseen in a country with limited routine surveillance. By analysing more than 15,000 neutralising antibody measurements from nearly 1,700 unvaccinated, HIV-uninfected participants, researchers reconstructed individual infection histories across urban Lilongwe and rural Karonga between February 2021 and April 2022. Their findings, published in PLOS Global Public Health, paint a picture of rapid antibody decline, frequent reinfection, and striking regional differences in exposure to the virus.

The research team, led by Mhairi J. McCormack of the University of Liverpool and colleagues, measured neutralising antibody titres against four distinct versions of the virus: the ancestral B.1 lineage, the Beta variant, the Delta variant, and the Omicron subvariants BA.1 and BA.2. Rather than relying on standard assays that simply flag whether someone has antibodies, the team used a pseudotyped virus neutralisation assay, in which harmless HIV-based particles carrying the SARS-CoV-2 spike protein are tested for their ability to infect cells in the presence of participant serum. The lower the serum concentration needed to block infection, the higher the neutralising titre, providing a quantitative, variant-specific readout of immune protection.

Quantitative titres alone, however, are difficult to interpret. Antibody levels rise sharply after an infection and then decay, so a modest measurement could reflect an old, strong immune response or a recent, weak one. To untangle this, the researchers built a multi-level Bayesian model of antibody kinetics. The model simultaneously estimated how much each type of infection boosts antibody levels, how fast those boosted levels wane, and how responses vary between individuals. Crucially, it used the full longitudinal trajectory of each participant’s titres across multiple variants to infer when infections most likely occurred, rather than relying on a single threshold to declare a sample positive.

This modelling approach paid off immediately. Across the 1,675 participants, the model identified an estimated 429 infections, with a 95 percent credible interval of 417 to 441. Traditional seroconversion-based thresholds, which typically classify someone as infected only when antibody levels cross a fixed cut-off, missed 39 of these infections, roughly 9 percent of the total. Many of these hidden infections would have been reinfections in people whose antibodies, though boosted, never rose above the conventional detection threshold. In settings where surveillance is sparse, the authors argue, such missed reinfections can seriously distort estimates of how much transmission has actually occurred.

The antibody kinetics themselves were sobering. After an infection, neutralising antibody levels fell steeply: on average, only 48 percent of the acute boost remained after three months, and just 5 percent survived a full year. In practical terms, the molecular signature of an infection becomes nearly invisible within months, which explains why threshold-based methods struggle and why serological surveys conducted long after a wave can dramatically undercount it. The rapid waning also has direct implications for protection, since circulating neutralising antibodies are a key correlate of defence against infection, even though cellular immunity may persist longer.

Not all infections were equal. Pre-Omicron variants, including Beta and Delta, generated stronger antibody boosts than Omicron infections did. Responses also varied widely between individuals: the model found that participants clustered into low and high responder groups, meaning that some people mounted substantially stronger neutralising responses to infection than others, independent of the variant involved. This heterogeneity adds another layer of complexity to interpreting a single blood sample, and it is precisely the kind of variation that the Bayesian framework was designed to capture rather than average away.

Cross-reactivity, the ability of antibodies raised against one variant to neutralise others, extended across considerable antigenic distance in the dataset. Infections with Omicron BA.1 or BA.2 induced notably broader antibody responses than earlier variants, consistent with Omicron’s heavily mutated spike protein sitting further from the ancestral virus on the antigenic map. The practical message is one of partial immunity: antibodies from a prior infection with an earlier variant offer some cross-protection against later ones, but that protection is incomplete, and the breadth of immunity depends heavily on which variant caused the infection.

The geographic and demographic patterns were equally revealing. Estimated seroincidence, the rate of new infections inferred from the antibody data, was higher in urban Lilongwe than in rural Karonga, at 0.41 versus 0.27 infections per person per three-month period. Much of this gap was driven by the early 2022 Omicron wave, which swept through the urban population with particular force. Reinfections were common throughout the study period, and were especially frequent among adults and among residents of the urban site, patterns that likely reflect greater exposure opportunities in denser, more connected communities.

For sub-Saharan Africa, where documented COVID-19 case counts were widely believed to understate true transmission, the study offers a methodological template. Serological data are among the few reliable windows into past infection in regions with limited testing capacity, but their value depends on correctly accounting for antibody waning and boosting. By modelling these dynamics explicitly and using variant-specific titres, the researchers showed that infection histories can be reconstructed with enough resolution to distinguish first infections from reinfections and to time waves that no surveillance system recorded.

The authors conclude that neutralising antibody responses following infection are both heterogeneous and short-lived, a combination that underscores the importance of vaccination for sustaining protection as variants evolve. Their results also highlight a practical priority: identifying reinfections, not just first infections, is essential for understanding transmission patterns and for pinpointing populations at elevated risk of repeated infection. In places like Malawi, where routine genomic and case surveillance remains limited, longitudinal serology paired with kinetic modelling may be one of the few tools capable of revealing who has been infected, how often, and with what.

Subject of Research: Neutralising antibody kinetics and reconstructed SARS-CoV-2 infection histories in unvaccinated populations in urban and rural Malawi

Article Title: SARS-CoV-2 neutralising antibody profiles reveal variant specific antibody dynamics and regional differences in infection histories in Malawi

Article References: McCormack, M. J., Banda, L., Kasenda, S., Hughes, E. C., Crampin, A., Amoah, A. S., Read, J. M., Ho, A., Willett, B. J., & Hay, J. A. (2026). SARS-CoV-2 neutralising antibody profiles reveal variant specific antibody dynamics and regional differences in infection histories in Malawi. PLOS Global Public Health, 6(10), e0007133. https://doi.org/10.1371/journal.pgph.0007133

Image Credits: AI Generated

DOI: 10.1371/journal.pgph.0007133

Keywords: SARS-CoV-2, neutralising antibodies, serology, Malawi, Omicron, antibody waning, reinfection, Bayesian modelling, pseudotyped virus assay, seroincidence, sub-Saharan Africa, variant cross-reactivity