Serological inference of past primary and secondary dengue infection: implications for vaccination

Abstract

Owing to the finding that Dengvaxia^® (the only licensed dengue vaccine to date) increases the risk of severe illness among seronegative recipients, the World Health Organization has recommended screening individuals for their serostatus prior to vaccination. To decide whether and how to carry out screening, it is necessary to estimate the transmission intensity of dengue and to understand the performance of the screening method. In this study, we inferred the annual force of infection (FOI; a measurement of transmission intensity) of dengue virus in three locations in Vietnam: An Giang (FOI = 0.04 for the below 10 years age group and FOI = 0.20 for the above 10 years age group), Ho Chi Minh City (FOI = 0.12) and Quang Ngai (FOI = 0.05). In addition, we show that using a quantitative approach to immunoglobulin G (IgG) levels (measured by indirect enzyme-linked immunosorbent assays) can help to distinguish individuals with primary exposures (primary seropositive) from those with secondary exposures (secondary seropositive). We found that primary-seropositive individuals-the main targets of the vaccine-tend to have a lower IgG level, and, thus, they have a higher chance of being misclassified as seronegative than secondary-seropositive cases. However, screening performance can be improved by incorporating patient age and transmission intensity into the interpretation of IgG levels.

Keywords: IgG antibody; Vietnam; dengue; force of infection; serostatus; vaccination.

Figure 1.

Geographical locations and population demographics of An Giang (AG), Ho Chi Minh City (HC) and Quang Ngai (QN). The demographic statistics were taken from the population census of 2014, published by the General Statistics Office (GSO) of Vietnam. The annual reported dengue incidence rates were calculated based on the number of cases reported between 2000 and 2010 [39] and the population sizes in 2009 (published by the GSO). (Online version in colour.)

Figure 2.

The seroprevalence of dengue in the study populations. The bars represent the proportion (with 95% confidence intervals) of samples in each age group of 4 years that have IgG levels above 10 Panbio units. The lines show the estimates of seroprevalence from 1000 simulations, of which the parameters were drawn from the posteriors of the following models: AgiCon33 (for An Giang), HcmCon31 (for Ho Chi Minh City) and QngCon31 (for Quang Ngai). (Online version in colour.)

Figure 3.

The histogram of the measured IgG levels (bars) and the inferred IgG-level distributions (lines) of the three exposure classes (seronegative, primary seropositive and secondary seropositive) across the study populations. The IgG-level distributions were simulated 1000 times based on the posteriors of the following models: AgiCon33 (for An Giang), HcmCon31 (for Ho Chi Minh City) and QngCon31 (for Quang Ngai). (Online version in colour.)

Figure 4.

Demonstration of using a continuous model to estimate the probability of a given sample being primary seropositive. This estimation method makes use of not only the IgG levels (measured by ELISAs) but also the ages of patients and the force of infection of dengue in the population. This probability table was built based on the IgG-level distributions and the median FOI estimate of An Giang, inferred from the AgiCon33 model. The left-most area of the table (in blue) represents individuals who are likely to be seronegative. Meanwhile, individuals falling into the rightmost area (also in blue) are likely to be secondary seropositive. (Online version in colour.)