Nipah virus dynamics in bats and implications for spillover to humans

Abstract

Nipah virus (NiV) is an emerging bat-borne zoonotic virus that causes near-annual outbreaks of fatal encephalitis in South Asia-one of the most populous regions on Earth. In Bangladesh, infection occurs when people drink date-palm sap contaminated with bat excreta. Outbreaks are sporadic, and the influence of viral dynamics in bats on their temporal and spatial distribution is poorly understood. We analyzed data on host ecology, molecular epidemiology, serological dynamics, and viral genetics to characterize spatiotemporal patterns of NiV dynamics in its wildlife reservoir, Pteropus medius bats, in Bangladesh. We found that NiV transmission occurred throughout the country and throughout the year. Model results indicated that local transmission dynamics were modulated by density-dependent transmission, acquired immunity that is lost over time, and recrudescence. Increased transmission followed multiyear periods of declining seroprevalence due to bat-population turnover and individual loss of humoral immunity. Individual bats had smaller host ranges than other Pteropus species (spp.), although movement data and the discovery of a Malaysia-clade NiV strain in eastern Bangladesh suggest connectivity with bats east of Bangladesh. These data suggest that discrete multiannual local epizootics in bat populations contribute to the sporadic nature of NiV outbreaks in South Asia. At the same time, the broad spatial and temporal extent of NiV transmission, including the recent outbreak in Kerala, India, highlights the continued risk of spillover to humans wherever they may interact with pteropid bats and the importance of limiting opportunities for spillover throughout Pteropus's range.

Keywords: Nipah virus; Pteropus; bats; disease modeling; henipavirus.

Fig. 1.

Map showing age-stratified seroprevalence in P. medius colonies, Bangladesh. Bats from eight colonies were sampled and tested for anti-NiV IgG antibodies: four within the “Nipah belt” (orange shaded) and four outside. Seroprevalence of adults (A), juveniles (J), and total seroprevalence (T) are shown with 95% CI error bars. The shaded region represents the “Nipah belt,” where previous human NiV outbreaks have been reported.

Fig. 2.

Results of Bayesian generalized linear model of factors affecting Nipah serostatus in bats in cross-sectional study. Bars indicate ORs and 50% (inner) and 95% (outer) credible intervals for model parameters. Factors with asterisks (*) have 95% CIs that do not overlap one. Model intercept (predicted probability of seropositivity for a juvenile, female bat outside the Nipah belt of mean size and good body condition) was 0.26 (95% CI 0.12 to 0.56).

Fig. 3.

Serodynamics of the Faridpur bat population, measured and fit to a GAM. (A) Adult seroprevalence over time, with measured values and 95% CI in blue and mean GAM prediction and 95% shown with line and surrounding shaded areas. Point from February 2006 (purple) is shown separately due to ELISA vs. Luminex measure. Periods of significant change (where GAM derivative 95% does not overlap zero) are shown in red (increasing) and green (decreasing). Periods of increase indicate viral-circulation events in the adult population; these do not occur with consistent periodicity or seasonality. Counts of primary human cases from local district (dark gray) and bat viral detections (orange; Table 1) are shown on bottom for comparison. (B) Juvenile seroprevalence during the first year of life (“yearlings”). All years’ measurements are collapsed onto the scale of a single year overlain to show yearling dynamics. Measured values and 95% CIs are shown in blue, and mean and 95% CIs for the GAM model pooled across cohorts are shown with line and surrounded shaded areas. GAM realizations for individual years are in gray and overall effect in black. The period of significant decline in the GAM is shown in red. Juvenile seroprevalence decreases over the course of the year and is not distinguishable from a simple linear decrease (ΔAIC < 1, dotted line).

Fig. 4.

Longitudinal data and fitted model for NiV serological dynamics in adult and juvenile bats. Red and black points show observed data (±1 SE), and solid lines show the fitted model (thick lines show the trajectory for the model with maximum-likelihood parameter estimates; thin lines show realizations for parameter estimates drawn from the estimated distributions) for the fraction of adults and juveniles seropositive for NiV (left axis), and the model-estimated number of infected adult and juvenile bats (bottom and right axes). See Methods for details of model structure.

Fig. 5.

Satellite telemetry and home-range analysis. Location data from satellite collars (n = 14) placed on 11 bats from Faridpur and 3 bats from Cox’s Bazaar, Chattogram, collected between 2009 and 2011 were used to calculate local and long-range movement patterns and home range for these two groups.

Fig. 6.

NiV partial N-gene phylogeny (224 nt). Phylogenetic neighbor-joining tree created in Geneious Prime 2019 using a Tamura-Nei model with 1,000 bootstrap replicates and Hendra virus as an outgroup is shown. Branch lengths are shown as the number of substitutions per site. Sample collection date, location, and GenBank accession numbers are included in the label for each sequence, except P. medius sequences we collected (GenBank accession nos. MK995284–MK995302). Blue labels indicate bat sequences from Faridpur and Bhanga (an outbreak response in Faridpur). Purple sequences are from P. medius from other roosts sampled during the longitudinal study. Red sequences are from P. medius in Cumilla. Green sequences are human NiV sequences from Bangladesh and India.