Serological analysis in humans in Malaysian Borneo suggests prior exposure to H5 avian influenza near migratory shorebird habitats

Abstract

Cases of H5 highly pathogenic avian influenzas (HPAI) are on the rise. Although mammalian spillover events are rare, H5N1 viruses have an estimated mortality rate in humans of 60%. No human cases of H5 infection have been reported in Malaysian Borneo, but HPAI has circulated in poultry and migratory avian species transiting through the region. Recent deforestation in coastal habitats in Malaysian Borneo may increase the proximity between humans and migratory birds. We hypothesise that higher rates of human-animal contact, caused by this habitat destruction, will increase the likelihood of potential zoonotic spillover events. In 2015, an environmentally stratified cross-sectional survey was conducted collecting geolocated questionnaire data in 10,100 individuals. A serological survey of these individuals reveals evidence of H5 neutralisation that persisted following depletion of seasonal H1/H3 HA binding antibodies from the plasma. The presence of these antibodies suggests that some individuals living near migratory sites may have been exposed to H5 HA. There is a spatial and environmental overlap between individuals displaying high H5 HA binding and the distribution of migratory birds. We have developed a novel surveillance approach including both spatial and serological data to detect potential spillover events, highlighting the urgent need to study cross-species pathogen transmission in migratory zones.

Fig. 1. Distribution of sampled villages (blue) and migratory shorebird sightings (red) in Sabah.

Sampled villages are in the Northern Sabah region, while migratory shorebird sightings are distributed throughout Sabah. This figure was generated in QGIS 3.30.2. Natural Earth raster map data was used to generate the canvas map. Source data for migratory shorebird sightings are provided as a Source Data file.

Fig. 2. Individuals living in proximity to migratory shorebirds have increased H5 HA binding and neutralisation.

As the data is non-parametric, the medians of each group were compared using a two-tailed Mann–Whitney test in GraphPad Prism 10.0.3. a Represents ELISA-binding data of n = 2000 samples, where the population is 51% (n = 1024) poultry owners and 23% (n = 454) were living in proximity to wild shorebirds (<10 km). Statistically significant p-values are 0.0174 (H1N1 binding, poultry ownership), 0.0156 (H5N2 binding, proximity), and 0.0012 (H5N1 binding, proximity). The nonsignificant p-values were 0.4707 (H3N2 binding, poultry ownership), 0.0612 (H3N2 binding, proximity), 0.7503 (H1N1 binding, proximity), 0.7633 (H5N2 binding, poultry ownership), and 0.1997 (H5N1 binding, poultry ownership). All samples were tested in duplicate at a 1:50 dilution. b Displays NT50 values for H5N1 pseudovirus microneutralisation assays (n = 204 for A/Indonesia/05/2005 and A/Bar headed goose Qinghai/1A/2005; n = 174 for A/chicken/Malaysia (Sabah)/6123/2018). NT50 is here defined as the dilution factor of the serum necessary to reach 50% inhibition of the pseudovirus. The samples tested against A/Indonesia/05/2005 and A/Bar headed goose Qinghai/1A/2005 pseudoviruses were 52% (n = 108) poultry owners, and 34% (n = 69) lived within 10 km of a shorebird sighting. The samples tested against A/chicken/Malaysia (Sabah)/6123/2018 pseudovirus (n = 178) were 50% poultry owners (n = 89), and 33% (n = 58) lived within 10 km of a migratory shorebird. The p-value for proximity comparisons for H5N1 Qinghai NT50s was <0.0001 (0.000007), and the p-value for poultry ownership for H5N1 Sabah NT50s was p = 0.0035. The nonsignificant p-values were 0.0950 (H5N1 Indonesia, poultry ownership), 0.1461 (H5N1 Indonesia, proximity), 0.4328 (H5N1 Qinghai, poultry ownership), and 0.0687 (H5N1 Sabah, proximity). The figure was generated in GraphPad Prism 10.0.3. Source data are provided as a Source Data file.

Fig. 3. H5 neutralisation is not depleted upon treatment with seasonal flu beads.

NT50 values were calculated using nonlinear regression in GraphPad Prism 10.0.3. Each plot displays the nonlinear curve fit used to calculate NT50 and 95% confidence interval bands of the nonlinear curve fit. Each plot is the average of two replicates minimum (Negative 1, n = 3; Negative 2, n = 5). LogNT50 values before and after treatment were compared via a sum-of-squares F Test in GraphPad (two-sided). All p-values for statistically significant differences in logNT50 are listed, while non-significant (p > 0.05) changes are listed as ns. a Displays the control serum for the depletion assay. Negative controls 1 and 2 are H1N1 convalescent sera (low and medium titre, respectively), while positive control pooled serum was exposed to a sub-virion H5N1 vaccine. Both sets of controls were obtained through BEI Resources. Positive pool 1 is a medium titre pool and positive pool 2 is a low titre pool, according to the supplier. For negative controls 1 and 2, the change in NT50 was statistically significant (p = 0.0010 and 0.00004, respectively), while for the positive pools, these changes were not significant (p = 0.8467 and 0.5636, respectively). Eight samples with unknown exposure history from the Malaysian cohort were treated by the same method in b. All samples were treated with five rounds of fresh protein-conjugated beads in a cocktail mixture at a concentration of 100 beads/μL per bead and aspirated upon incubation with a magnet. The changes in NT50 were not significant for any of these samples (p = 0.0958, 0.3534, 0.0813, 0.7387, 0.8763, 0.6636, 0.4321, 0.8414 for high neutraliser 1, 2, and low neutraliser 1 through 6, respectively). Source data are provided as a Source Data file.

Fig. 4. Spatial overlap between wild shorebird contact and H5 binding exists that is not seen when comparing domesticated poultry contact and H5 binding.

a Estimated mean probability of wild shorebird contact and estimated mean probability of high H5 binding (to A/Duck/Laos/3295/2006) from Bayesian geostatistical models predicted across the study site in Northern Sabah (Fig. 1). Posterior probabilities were estimated using 1000 posterior samples. These posterior probabilities were then used to predict the probability of the outcome variable (H5 binding or species distribution) across the whole study region at 30 m spatial resolution. A total of n = 2000 samples were tested in duplicate by ELISA at a 1:50 dilution, and IgG binding to H5 HA was measured in absorbance units (AU). High H5 binding is considered greater than 2.0 AU based on comparison to the control cohorts. b Adjusted odds ratios for environmental predictors in mixed effects modelling of contact and H5 binding are shown along with 95% confidence intervals for each odds ratio and statistical significance, where p ≤ 0.10 (.), p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***). Odds ratios greater than 1 (shown in blue) indicate a factor that increases the odds of high H5 binding or of migratory shorebird sightings. Odds ratios <1 (shown in red) decrease the odds of those outcomes. p-values for the risk factors for the odds of wild bird sightings are 3.07 × 10⁻⁶ (model intercept), 6.88 × 10⁻¹⁰ (Euclidean distance from roads), 0.001 (elevation in meters above sea level), 0.001 (distance from the sea in meters), 6.90 × 10⁻⁵ (maximum temperature of warmest month in °C), 4.68 × 10⁻⁹ (precipitation of the wettest month in mm), 0.094 (distance from bush forest in meters) and 0.01 (distance from old forest in meters). p-values for the risk factors for the odds of H5N1 binding are as follows: 0.018 (normalised differential vegetation index), 3.93 × 10⁻⁶ (elevation in meters above sea level), 1.67 × 10⁻⁶ (distance from the sea in meters), 4.79 × 10⁻⁵ (mean diurnal range in °C), 4.85 × 10⁻⁷ (minimum temperature of coldest month in °C), 0.04 (precipitation of the wettest month in mm), 0.084 (precipitation seasonality), and 0.027 (distance from irrigated farmland in metres). Source data are provided as a Source Data file including mean posterior estimates, odds ratios, and confidence intervals.