Reconstructing infection histories and antibody dynamics in parallel to quantify dengue risk using mathematical models
Link to the paper: https://go.nature.com/2kiKGPA
Dengue continues to infect millions of people each year throughout the central tropical and subtropical band of the world. However, most people don’t even realize they’ve been infected while others will get severely sick or even die. We still do not know exactly why this happens, but individuals with pre-existing antibodies to dengue from previous infections appear to be at increased risk of severe illness compared to those who have never previously been infected.
Efforts to combat the burden from dengue would be greatly aided if we could identify the subset of individuals at risk of getting really sick for careful monitoring. We’d also like to understand how stable that risk is. In order to get close to these goals, we first need to understand how antibody concentrations behave over time and link them to disease outcomes. This is complicated to do. The regular presence of silent infections and frequent clinical misdiagnoses mean we cannot simply ask people whether they’ve had dengue before or not. Also, by the time we test patients who show up with dengue at a clinic, their antibody levels will often have changed in response to the infection. We need to be able to turn back the clock to see what their antibody levels were before infection to assess whether there are precise antibody concentrations associated with disease.
To do this, we forged collaborations with research groups in Thailand and the US that had run a cohort study where blood was taken every 3 months for up to 5 years from ~3,500 school-aged children in a rural district in Northern Thailand; an area where dengue has circulated endemically for decades. Antibody titers were measured in each blood sample to each of the four serotypes – an enormous undertaking consisting of over 140,000 tests. We used advanced statistical methods, called data augmentation, to reconstruct an individual’s infection history (including from these silent asymptomatic infections) in parallel to reconstructing their daily antibody levels. Using this framework we could then assess whether antibody levels are linked to the risk of being infected or sick in the future from dengue. We found there was a clear window of risk, where those with pre-existing antibody titers of under 1:40 (as measured through tests called hemagglutination inhibition assays) had over 7 times the risk of developing dengue hemorrhagic fever (the severe form of dengue) compared to those who were naïve. Those with higher titers appeared also to be protected. We also found that individuals developed a stable setpoint antibody load a year following infection, which strongly suggests risk remains largely unchanged from this time point.
These findings have implications for the development of dengue vaccines. The only dengue vaccine licensed to date, Dengvaxia, seems to shift your antibody level but not sufficiently high in those that are completely naïve at the time of vaccination to prevent infection or disease. In fact, it seems vaccination in previously naïve individuals places them squarely in the window of risk. By contrast, individuals who are already within this window from historic infections appear to be lifted out of it by the vaccine. These findings further support the testing of individuals for presence of antibodies before vaccination.