Postdoctoral Research Fellow Merck & Co, United States
Objectives: Serum antibody (Ab) titers are thought to be a potential predictor for protection against symptomatic dengue infection. Despite extensive research on the relationship between vaccine-induced titers and protection, it remains unclear how to best estimate the risk of contracting dengue after natural infection. This study aims to quantify the effectiveness of natural immunity by assessing the relationship between post-natural infection (PNI) Ab titers and the probability of disease (PoD), i.e., the probability of contracting symptomatic dengue within a specified period. We propose a novel framework that allows us to compute PoD curves for various prediction periods (henceforth referred to as N-month PoD curves) given a longitudinal subject-level dataset of titers and disease status.
Methods: We used generalized linear models (GLM) to infer the relationship between PNI Ab titers and symptomatic disease using data from a longitudinal cohort study in school children in Thailand [1]. We expanded on the previously established approach of using one time point as a predictor for disease outcome [2], and instead used all titer measurements taken prior to disease, or all measurements if disease was not observed. To qualify our approach, we conducted simulations based on a “true” PoD curve and assessed our ability to reproduce it. To overcome the limitation of the PoD prediction period being equal to the data sampling frequency, we propose three novel methods: subsampling data, a mathematical approximation based on the short-term PoD curve, and a weighted GLM approach. These methods allow us to extend the prediction period beyond the time in between sample collection.
Results: We estimated the 3- and 12-month PoD curves for school children in Thailand between 1998 and 2003. The results indicate that PNI Ab titers can be used as a correlate of risk for symptomatic dengue. Using simulated data sampled every 3 months, we showed that our approach accurately estimates the parameters of the true 3-month PoD curve and consistently estimates a unique 12-month PoD curve, regardless of which method is used. These results provide compelling numerical evidence that multiple PoD curves can be estimated for different prediction periods on the same data set.
Conclusions: This work enhances the understanding of dengue-related diseases by providing a framework for predicting risk based on time-dependent PNI Ab titers. These methodological advancements improve our ability to estimate PoD curves by utilizing the majority of a given dataset and enabling predictions further into the future. Better estimates of PNI Ab titer-based protection can also aid vaccine development by serving as a reference point, with which vaccine-induced protection can be compared. Whether used as a standalone or to compare PNI Ab titers to vaccine induced Ab titers, this work can be used for optimizing of dengue outbreak prevention strategies.
Citations: [1] Salje, H. et al. Reconstruction of antibody dynamics and infection histories to evaluate dengue risk. Nature 557, 719–723 (2018). [2] Katzelnick, L. C. et al. Neutralizing antibody titers against dengue virus correlate with protection from symptomatic infection in a longitudinal cohort. PNAS 113, 3, 728-33 (2016).
