Cholera is an acute dehydrating diarrheal disease caused by infection with Vibrio cholerae. It is endemic in over 50 countries, affecting up to 3 million people and causing more than 100,000 deaths annually. A renewed global effort to fight cholera is underway, catalyzed by the large on-going epidemic in Haiti and now aided by new generation oral cholera vaccines. Identifying key populations at high risk of cholera is essential to guide these activities. Current methods to estimate cholera burden are largely based on clinical reporting with infrequent microbiological confirmation. These methods are limited by the sporadic nature of outbreaks, poor surveillance infrastructure, and fundamental uncertainties in the number of asymptomatic or mildly symptomatic cases. Improved methods of detecting cholera exposure and risk are urgently needed. Detection of immune responses in serum (serosurveillance) can provide a new avenue for rapid and accurate estimates of cholera exposure and risk. We currently do not understand what immunological and clinical parameters are most predictive of recent exposure, nor whether immune responses in areas with different levels of endemicity are similar. In preliminary studies, we have used machine learning methods on antibody response data from cholera patients in Bangladesh to classify whether individuals had been exposed in the previous 30-, 90-, or 360-days with high sensitivity and specificity. In this application, we propose to use longitudinal antibody response kinetics, from populations with diverse genetic and epidemiologic profiles, paired with novel statistical and machine learning approaches to provide generalizable tools to estimate the incidence of exposure to Vibrio cholerae from cross sectional serosurveys.
In Aim 1, we will develop models to estimate the time since exposure to Vibrio cholerae and exposure incidence from cross-sectional antibody profiles and demographic data using previously collected data from a cohort in Bangladesh. These results will allow us to identify the antibodies and demographic factors that are most useful for prediction of time-since-exposure.
In Aim 2, we will collect longitudinal antibody data from a cohort of cholera cases and household contacts in Haiti to develop models for estimating exposure incidence from cross-sectional serosurveillance. This cohort will also enable us to compare the models developed for moderate/severe cases and mild/asymptomatic cases.
In Aim 3, we will optimize and validate field-adapted methods to measure cholera-specific antibodies, including the use of dried blood spot and lateral flow assays. We will conduct a proof-of-concept cross-sectional serosurvey using these methods in rural Haiti. Upon the completion of these aims, we will have provided a number of new tools for measure of susceptibility to cholera in a population. These tools will have the potential to transform cholera control efforts from the current reactive strategies to proactive ones, with the potential to contribute to disease elimination.
Cholera is a diarrheal disease affecting millions of people worldwide each year, with the majority of deaths due to cholera occurring in places with poor disease surveillance infrastructure. To optimize cholera control efforts, accurate estimates of cholera exposure and risk are needed. We propose studies using computational methods to determine how antibody decay data may help us estimate cholera exposure incidence.
