Serotype-specific epidemiological patterns of inapparent versus symptomatic primary dengue virus infections: a 17-year cohort study in Nicaragua

Abstract

Background: Dengue is the most prevalent mosquito-borne viral disease and a major public health problem worldwide. Most primary infections with the four dengue virus serotypes (DENV1-4) are inapparent; nonetheless, whether the distribution of symptomatic versus inapparent infections by serotype varies remains unknown. Here, we present (1) the evaluation of a DENV1-4 envelope domain III multiplex microsphere-based assay (EDIII-MMBA) to serotype inapparent primary infections and (2) its application leveraging 17 years of prospective sample collection from the Nicaraguan Pediatric Dengue Cohort Study (PDCS).

Methods: We analysed primary DENV infections in the PDCS from 2004 to 2022 detected by inhibition ELISA (iELISA) or RT-PCR. First, we evaluated the performance of the EDIII-MMBA for serotyping with samples characterised by RT-PCR or focus reduction neutralisation test. Next, we analysed a subset of inapparent primary DENV infections in the PDCS with the EDIII-MMBA to evaluate the epidemiology of inapparent infections. Remaining infections were inferred using stochastic imputation, taking year and neighbourhood into account. Infection incidence and percentage of inapparent, symptomatic, and severe infections were analysed by serotype.

Findings: Between Aug 30, 2004, and March 10, 2022, a total of 5931 DENV-naive participants were followed in the PDCS. There were 1626 primary infections (382 symptomatic, 1244 inapparent) detected by iELISA or RT-PCR over the study period. The EDIII-MMBA demonstrated excellent overall accuracy (100%, 95% CI 95·8-100) for serotyping inapparent primary DENV infections when evaluated against gold-standard serotyping methods. Of the 1244 inapparent infections, we analysed 574 (46%) using the EDIII-MMBA. We found that the majority of primary infections were inapparent, with DENV3 exhibiting the highest likelihood of symptomatic (pooled odds ratio compared with DENV1: 2·13, 95% CI 1·28-3·56) and severe (6·75, 2·01-22·62) primary infections, whereas DENV2 was similar to DENV1 in both analyses. Considerable within-year and between-year variation in serotype distribution between symptomatic and inapparent infections and circulation of serotypes undetected in symptomatic cases were observed in multiple years.

Interpretation: Our study indicates that case surveillance skews the perceived epidemiological footprint of DENV. We reveal a more complex and intricate pattern of serotype distribution in inapparent infections. The substantial differences in infection outcomes by serotype emphasises the need for vaccines with balanced immunogenicity and efficacy across serotypes.

Funding: National Institute of Allergy and Infectious Diseases (National Institutes of Health) and Bill & Melinda Gates Foundation.

Translation: For the Spanish translation of the abstract see Supplementary Materials section.

Figure 1

Overall symptomatic and inapparent DENV infections in the PDCS from 2004 to 2022 (A) Yearly counts of DENV primary infections by symptomatic or inapparent status, as indicated. (B) Yearly cumulative incidence proportion of primary inapparent and symptomatic DENV infections. (C) Yearly proportion of inapparent infections among total primary DENV infections. The pink shaded area in 2016 is the period when ZIKV circulated in the PDCS. The grey area in panel B covers the pooled incidence proportions of overall primary DENV infections by serotype. The dashed line in panel C is the yearly average of inapparent primary DENV infections across years, and the grey area represents the 95% CI. The x-axes show the cohort year; cohort year 2021 ended in March, 2022. DENV=dengue virus. PDCS=Pediatric Dengue Cohort Study. ZIKV=Zika virus.

Figure 2

Concordance of relative EDIII-MMBA values from participants with primary DENV infection with values from RT-PCR and FRNT (A) Symptomatic primary DENV infections confirmed by RT-PCR. (B) Inapparent primary DENV infections confirmed by FRNT. Within the heatmaps, each cell indicates the MFI. MFI values per participant were transformed by dividing the MFI of each EDIII antigen by the highest MFI observed in the set (MFI/maxMFI). Adjacent to each heatmap, dendrograms display the hierarchical clustering of individuals based on Euclidean distances calculated for the normalised MFI values. On the right of the dendrograms, a colour-coded column classifies individuals according to the gold standard assay results, providing a reference for true infection status. DENV=dengue virus. EDIII-MMBA=envelope domain III multiplex microsphere-based assay. FRNT=focus reduction neutralisation test. MFI=mean fluorescence intensity. ZIKV=Zika virus.

Figure 3

Serotype distribution in primary symptomatic and inapparent infections by serotype from 2004 to 2022 in the PDCS (A) Pooled relative distribution of DENV serotypes circulating yearly in inapparent and symptomatic primary DENV infections across multiple imputations. Transparency was added in years with infection count <30. The grey panel shows the total incidence of primary DENV infections. (B) Temporal dynamics of the percent of primary DENV infections by infection outcome (shown in upper and lower panels) and serotype across imputations. The grey area covers the pooled incidence proportions of overall primary DENV infections by infection outcome. The x-axes show the cohort year; cohort year 2021 ended in March, 2022. DENV=dengue virus. PDCS=Pediatric Dengue Cohort Study. *p<0·05. †p<0·01. ‡p<0·001.

Figure 4

Temporal dynamics of symptomatic, inapparent, and overall primary DENV infections by serotype from 2004 to 2022 in the PDCS (A) Pooled percentage of primary DENV infections by infection outcome and serotype, as indicated in left, middle, and right panels across imputations. The grey area covers the pooled incidence proportions of overall primary DENV infections by serotype. (B) Proportion of symptomatic infections among total primary DENV infections by serotype (indicated in left, middle, and right panels). This analysis excludes years when no symptomatic infections or <10 primary infections were reported. This is the result of the pooled predicted marginal probabilities of disease given infection from logistic regressions adjusting for year, compared with reference years (dashed line) of the most intense outbreak (DENV1: 2012, DENV2: 2019, DENV3: 2009). The x-axes show the cohort year; cohort year 2021 ended in March, 2022. DENV=dengue virus. PDCS=Pediatric Dengue Cohort Study.

Figure 5

Spectrum of disease of primary DENV infections (A) Pooled counts of symptomatic and overall primary DENV infections with crude percentages of symptomatic primary DENV infections by DENV serotype and pooled ORs adjusting by year showing the association between dengue case and serotype among primary DENV infections. (B) Pooled counts of severe and overall primary DENV infections with crude percentages of severe primary DENV infections by DENV serotype and pooled ORs adjusting by year showing the association between dengue case and serotype among primary DENV infections. (C) Spectrum of infection outcome and disease severity according to the 2009 classification as the crude pooled percentages of primary DENV infections. DENV=dengue virus. OR=odds ratio.