The World Health Organization estimates that every year there are around a billion cases of seasonal influenza, causing up to 650,000 deaths globally.
Providing new insights into the epidemiology of influenza, a new study published in PLOS Biology has estimated which strains of influenza A/H3N2 – and when – infected and reinfected individuals over time, and how this affected influenza rates in and around Guangzhou, China, between 1968 and 2014.
Influenza subtype A/H3N2 is a key contributor to seasonal flu outbreaks and causes repeated reinfections. But how immunity builds up over a lifetime and determines long-term epidemiology is not fully understood.
To address this, the research project, which involved UK and international institutions including the Pandemic Sciences Institute and Imperial College, London, analysed data from the Fluscape cohort study in and near Guangzhou. Researchers reconstructed individual infection histories and historical influenza patterns by applying a mathematical model to antibody data from blood samples collected from 1,130 individuals.
Researchers found that influenza infection rates in the region averaged 19% annually with substantial variation over time – a higher rate than suggested by routine surveillance data. Reinfections occurred roughly once every five years but tended to be more frequent in younger children and decreased in frequency into adulthood.
In alignment with previous research, the study found that infection risk decreased with increasing antibody levels – but also that this relationship became weaker with age, suggesting that other immune factors, not just antibodies, become more important correlates of risk as we get older.
Lead author Dr James Hay, Wellcome Trust Early Career Fellow at PSI, said: “Our study demonstrates how detailed insights into influenza epidemiology and immunity can be obtained by analysing these complex, multi-strain antibody profiles – a level of detail which is extremely difficult to obtain through routine surveillance.”
Steven Riley, Professor of Infectious Disease Dynamics at Imperial College London, said: “This modelling approach could be applied to understand the long-term immune dynamics of other viruses which cause repeated lifetime infections, such as coronaviruses, norovirus, and many others.”
Decoding antibody profiles
In counting cases of influenza-related illness, surveillance data may underestimate real infection incidence. Serological studies provide more information; however, observed antibody levels are the result of complex interactions of multiple immunological responses from all past infections. Disentangling the contribution of each infection to an individual’s antibody profile is a challenging decoding problem.
This complex reactivity is reflected in the study, as antibody levels were measured against 20 strains of A/H3N2 representing viruses which circulated from 1968 up until 2014. Individuals’ antibody levels were mostly low or non-existent for strains detected long before they were born, but many younger participants had antibodies against strains that circulated in the years before birth, suggesting the presence of cross-reactive antibodies. Antibody levels tended to be highest against strains which were detected around or just after an individual was born.
The resulting multi-strain antibody profiles were analysed using serosolver, an open-source statistical software previously developed by Dr Hay and colleagues to reconstruct individual-level infection histories. The combined analysis of all infection histories allowed A/H3N2 incidence to be estimated in the region over almost five decades.
Not only does the approach developed by Dr Hay and colleagues add to the current understanding of who gets infected with influenza and when, it is also relevant to viruses such as SARS-CoV-2, which reinfect people over the course of their lives. Such epidemiological insights contribute to the body of research informing public health planning and infectious disease outbreak response.