This prospective study has captured the entire spectrum of clinical disease severity among infants with primary DENV infections—ranging from inapparent infections, mild outpatient febrile illnesses, hospitalized illnesses without evidence of DHF, and unambiguous hospitalized DHF/DSS. DENV3 was the predominant infecting serotype in the 2007 dengue season covered by this report. We therefore focused on the potential role of maternally derived anti-DENV IgG in shaping DENV3 disease severity. Our data support an initial in vivo protective role for high levels of maternally derived anti-DENV3 IgG at birth. The estimated in vitro anti-DENV3 neutralizing capacity at birth (maternal DENV3 PRNT50
geometric mean titer [GMT]
413, range 41–5,690) positively correlated with the infant age of symptomatic primary DENV3 illness (all disease severity groups). A similar correlation was previously reported for estimated anti-DENV2 neutralizing capacity at birth in 13 DHF infants with primary DENV2 infections 
. A more recent study did not report a correlation between estimated serotype-specific neutralizing capacity at birth and the infant age at dengue illness 
. However, this study combined serotype-specific PRNT50
values for infants with DENV1-3 infections in their analysis and performed a different in vitro neutralization assay.
All the infants in our study with symptomatic DENV3 infections had a DENV3 PRNT50
≤50 at the time of infection, and there were no correlations between the neutralizing capacity at time of infection and DENV3 disease severity. These data support that primary DENV3 febrile illnesses occurred in infants when maternally derived anti-DENV3 IgG were below effective in vivo neutralizing concentrations. Identifying levels of neutralizing antibody associated with clinical protection is an important issue for dengue vaccine development and testing, as such correlates of protective immunity have not been previously well defined. The estimated neutralizing antibody titers at the time of primary DENV infection in infants represent a natural study of passive immunization. For DENV3 infections, our results suggest that a DENV3 PRNT50
>50 is likely to correlate with clinical protection but a measurable DENV3 PRNT50
≥10 and ≤50 is not. These findings are remarkably consistent with an earlier study we conducted in older children with secondary DENV3 infections 
. Using the same DENV3 strain in the neutralization assays, we reported that only a pre-illness DENV3 PRNT50
≥100 was associated with lower viremia and milder DENV3 disease severity. We anticipate that similar neutralizing antibody data for other DENV serotypes will emerge as our infant clinical study continues.
The current ADE model predicts that infant DHF develops when subneutralizing maternally derived anti-DENV IgG enhances DENV infection in Fc receptor-bearing cells to higher levels than in infants with subneutralizing anti-DENV IgG levels who do not develop DHF 
. We found that essentially all infants with symptomatic DENV3 infections had subneutralizing plasma IgG levels and measurable Fc receptor-dependent DENV3 ADE activity at the time of infection. The infants who developed DHF did not have significantly higher frequencies or levels of DENV3 ADE activity compared to symptomatic infants without DHF. We recognize that some hospitalized infants classified as “not DHF” may have met the WHO criteria for DHF with more intensive investigations. However, there were no significant differences in DENV3 ADE activity between the most severe hospitalized DENV3 illnesses and mild outpatient DENV3 illnesses, or between all hospitalized and nonhospitalized infants. Our data cannot exclude that mean ADE levels in symptomatic infants without DHF were actually ≤1.0 log10
DENV3 genome eqs/ml lower than in the DHF infants. Notably, ADE and early viremia levels varied widely across all the symptomatic DENV3-infected infants, even among those with mild symptomatic primary DENV3 illnesses. DENV3 ADE activity at the time of infection and early viremia levels were also not correlated, although a significant positive association with a regression slope <0.2 may have been missed. Our prospective study is the first to directly examine ADE activity at the time of infection among infants exhibiting a wide spectrum of dengue disease severity. The results suggest that measurable ADE activity is common and varied across all symptomatic DENV3-infected infants, and no significant associations with DHF have yet emerged. There has been only one previous study that examined and reported a direct association between ADE activity and infants with primary DHF 
. The authors reported that peak DENV2 ADE activity in 5/13 diluted sera from mothers of infants with primary DENV2 DHF were above a cutoff value compared to 2/22 sera from mothers of infants with nondengue febrile illnesses and toddlers with secondary DHF. The dilution of maternal sera that produced peak ADE activity was not directly linked to the time of DENV2 infection in the DHF infants. The reported data were from a single experiment, and there was no comparison to primary DENV2-infected infants without DHF.
We measured Fc receptor-dependent ADE of DENV3 infection using the K562 human cell line and a highly reproducible DENV3 qRT-PCR assay. Experimental conditions for ADE assays have not been standardized and have been previously debated 
. The K562 cell line has been used by others to reliably measure ADE of DENV infection in vitro 
and support the temporal association between ADE activity and infant DHF 
. DENV qRT-PCR is a sensitive, accurate, and reproducible method to measure viral production, and has always correlated with plaque-forming virus titers or percentage of infected cells in ADE assays 
. The ADE results and conclusions in this report are restricted to the DENV3 serotype, though the ADE model has been proposed as a general mechanism for all infant DHF. Future data from the ongoing clinical study should help to determine whether the current lack of a positive correlation between ADE activity and symptomatic dengue disease severity will hold up within narrower confidence intervals, and also extend our findings beyond primary infant DENV3 infections.
There is the possibility that ADE of DENV infection is important in the pathogenesis of all symptomatic primary infant dengue (including mild, outpatient febrile illnesses) compared to inapparent/asymptomatic dengue. In , the observed infant ages may be clustered in a window period after maternally derived anti-DENV3 IgG drop below effective in vivo neutralizing concentrations. If inapparent DENV3 infections are not similarly clustered (i.e., shifted to the right of the symptomatic DENV3 infections), then it would suggest that an antibody-mediated effect is associated with symptomatic DENV3 illness. We could not pinpoint the time of inapparent DENV3 infections and therefore could not include this group in or assess their relevant ADE activity. Additional studies that can accurately capture the ages of inapparent infant DENV infections will be important in this regard.
Interestingly, we found that a higher weight-for-age z-score in the first few months of life was a risk factor for developing DHF from a subsequent primary DENV infection during infancy. This association persisted at the time of illness presentation among infants with symptomatic primary DENV infections. Some earlier studies have suggested that infants with DHF are less likely to be malnourished at presentation compared to healthy or nondengue illness controls 
. A younger infant age at the time of a primary DENV infection has also been previously reported as a risk factor for DHF 
. Our study demonstrated this trend but notably it was independent of ADE activity. The anthropometric and age-related observations in this study should stimulate investigations into novel potential host mechanisms involved in the pathogenesis of infant DHF. These observations should also alert clinicians caring for infants in dengue-endemic countries. A high index of suspicion and close monitoring for DHF should take place particularly for febrile infants ≤8 mo old with above average WHO weight-for-age z-scores for their population.
In summary, our prospective nested case-control study found that DENV-infected infants exhibited a wide range of disease severity. We identified levels of maternally derived neutralizing antibody associated with clinical protection against symptomatic DENV3 illness. Measurable ADE activity at illness onset and early viremia levels varied widely across all symptomatic DENV3-infected infants, including those with mild outpatient illnesses. The results should encourage rethinking or refinement of the currently promulgated ADE model for infant DHF, promote prospective studies of infant dengue, and stimulate new directions of research into novel potential mechanisms for infant DHF.