How is a dengue vaccine feasible?
A dearth of information on individual susceptibility and correlatives of protection is a major problem in understanding dengue infections and pathogenesis. While recent monitoring data on a large Nicaraguan child cohort has provided much need information on the importance of antibody titres in protection vs. pathogenesis, more studies are needed to reach a full understanding of the disease. An additional difficulty is to determine infection dynamics, as most infections are missed because they are asymptomatic.
Data acquisition and modelling from a Thailand cohort
In this study, Salje and colleagues collected data for 5 years from a school-based cohort in Thailand, where blood from 3,451 children was collected every 90 days, or with increased frequency during illness or seasonal dengue peak times. Samples were tested by haemagglutination inhibition assay (HAI), plaque reduction neutralisation assay (PRNT), and inhibition ELISAs, showing a good correlation between the three tests, although measured titre values differed, particularly between labs. This is an important aspect to take into consideration that strongly suggests the need for a universal reference standard to normalise data, as already happens for other viruses.
Over time, the study recorded 274 symptomatic infections, of which 62 (23%) required hospitalisation and 36 (13%) lead to dengue haemorrhagic fever (DHF). By applying data augmentation methods to impute undetected infection, as well as the infecting serotype (rather than relying on set cut-offs that are sensitive to assay variability), the authors show a mean increase in log2 titres of 7.6 for the infecting serotype and of 6.6 for non-infecting serotypes in primary infections (a small difference that makes HAI assays unable to distinguish between the two), and a mean increase of 5.8 across serotypes in post-primary infections, decreasing by 76% after 1 year. An interesting observation was that the titres for DENV 2 were always lower than for other serotypes, although it is unclear whether this is a limitation of the assays used or is something specific to the biology of this serotype. Using the same methods, the authors probabilistically identified 1,149 asymptomatic infections, of which 507 occurring during the period of more active surveillance. This suggests that only 35% of DENV infections are symptomatic. Furthermore, 34% of the asymptomatic infections were re-conducted to DENV 4, suggesting a reduced risk of disease from this serotype, independently of high transmissibility. So if transmissibility does not necessarily predict disease, is there anything that can?
The ambivalent role of antibodies: further evidence
The authors’ conclusions are similar to what we know from Katzelnick’s study on the Nicaraguan cohort: the probability of being infected and developing disease is strongly linked to the mean antibody titres. An individual with mean log2 titres < 2 has a 17% annual risk to develop infection; an individual with mean log2 titres between 2 and 3 has an annual risk of 16%, and this is reduced to 11% for an individual with mean log2 titres > 3. In other words, for log2 titres > 2, each unit increase in log2 titres is associated with a 0.71-times relative risk of infection. These are the annual probability figures suggested by the authors for individual with A) a primary infection, B) a secondary infection with log2 titres <3, and C) a secondary infection with log2 titres >3, respectively:
Symptomatic infection: 6.4%; 8.4%; and 4%
Hospitalisation: 1.2%; 2.4%; and 0.3%
DHF: 0.2%; 1.5%; and 0.0%
These data are in line with the results from Katzelnick and colleagues, and demonstrate that the highest risk of infection is associated with lower antibody titres, followed by the total lack of pre-existing antibodies, and finally by high antibody titres (> 3). While the authors do not measure or specifically mention antibody-dependant enhancement (ADE) in their analysis, these data strongly support a role for ADE in dengue pathogenesis and severe disease, and in particular the relatively novel concept that titres (rather than serotype specificity, multiple infections or age) are determinant of disease severity; although all these phenomena are necessarily correlated, as age correlates with the probability of contracting multiple infections, and encountering the same serotype accelerates expansion of high-titres neutralising antibodies. Interestingly, 54% of the population has a mean log2 titre of less than 3, suggesting higher vulnerability to dengue and severe dengue.
Titres and time
Another important aspect to take into consideration is time, as antibody titres steadily decrease already 1 year after infection, and again the log2 antibody titre correlates well with the risk of infection and DHF. This confirms that what is likely to have gone wrong with Dengvaxia in seronegative vaccines; the antibody raised by the vaccine in naïve individuals seems to have placed them in the low antibody titre window that increases the risk of diseases. With this new set of measurements, the authors suggest that PRNT titres > 1:100 or HAI titres > 1:40 are a good starting point to determine the correlative of protection for a dengue vaccine; anything below might increase the risk. However, what happens over time when antibody titres fade? Is a boosting strategy sufficient to restore long-term high antibody levels? Moreover, what about the high variations that this study suggests exist between individuals?
Outlook
While mostly confirming conclusions that other studies have reached in the past few years, this work provides a methodological and mathematical framework that can help future vaccine testing and field studies. It also strongly highlights the problems inherent to the development of new vaccines, as it remains unclear whether the current candidates in the trial, although hopefully raising higher antibody levels, are going to address these newly emerging issues specifically. Equally, assay standardisation and optimisation to reflect neutralisation of the circulating rather than lab-adapted strains of dengue should be pursued. From what we have learnt so far, the key for a successful dengue vaccine is to induce high enough levels of antibodies to be maintained over time, and after previous verification of the serostatus of each. To date, these remain significant challenges which might require some disruptive thinking.
Flaviviruses and dengue virus, in particular, constitute one of our main research interests. At Virology Research Services we support studies into dengue vaccine, diagnostics, and antiviral development. Our research also aims to improve the assays currently in use to maximise their relevance and power. If your research involves dengue or other flaviviruses, including molecular mechanisms of host-virus interaction, contact us to find out how we can help.
