Epidemiology and evolutionary dynamics of H9N2 avian influenza virus in Bangladesh

Abstract

Low pathogenicity avian influenza (LPAI) H9N2 has been enzootic in Bangladeshi poultry since 2006. H9N2 outbreaks can decrease egg production and growth and pose a risk to human health. The role of avian hosts in the persistence, evolution, and dispersion of H9N2 is poorly understood in Bangladesh. Hence, this study unveils the intricate role of major host species in virus maintenance and evolution and the temporal and seasonal patterns of H9N2 in Bangladesh from 2006 to 2023. Multinomial logistic regression analysis indicated that the circulation of H9N2 in different species and interfaces is significantly influenced by the seasons. Bayesian phylogenetic analysis of H9N2 sequences in Bangladesh revealed two distinct lineages: G1 and Eurasian. The G1 lineage split into two clusters, coexisting until 2019, at which point only one cluster persisted. Bayesian phylodynamic analysis of G1 lineage unveiled frequent bidirectional viral transitions among ducks, chickens, and quails. Chickens might be a pivotal source of H9N2 in Bangladesh, with a higher number of viral transitions from chickens to ducks and quails. Quails appear to acquire most of their viral transitions from chickens rather than ducks, suggesting that quail epizootics are primarily triggered by spillover events from chickens. Our results suggest viral circulation in commercial chickens despite vaccination. The vaccination approach should be revised, assess vaccine efficacy, and extension of vaccination to backyard chickens and quails.

Keywords: AIV; LPAI; Phylodynamics; poultry; surveillance; vaccination; zoonotic.

Figure 1.

Temporal Trend of H9N2 Cases in Bangladesh (2006–2022) by Species. Each column represents a month, with dotted lines denoting year transitions. The color-coded bars within each month reflect the number of cases reported from different species.

Figure 2.

Seasonal patterns of H9N2 cases for species and interfaces. The P-value is obtained from Chi-squared test.

Figure 3.

Odds ratio with 95% confidence interval of the variables of the multinomial logistic regression model relating the variation of H9N2 in species and interface with season in Bangladesh (*P < 0.05, **P < 0.01, ***P < 0.001).

Figure 4.

Bayesian Phylogenetic tree of the HA gene of H9N2 in Bangladesh from 2006–2023, depicting the various clades. While the two large grouping at the top of the tree (in blue and yellow) indicates the G1 lineage and the small group (in purple) indicates Eurasian lineage. The scale bar indicates the number of substitutions per site. Colored boxes indicate both HA clade and HA-NA subtype combination. Branches are coloured according to host type.

Figure 5.

Time-scaled phylogenetic tree of HA sequences of H9N2 in Bangladesh. Branches are colored according to host type and the thickness of branches indicates posterior probabilities of the ancestral host type. Genetic subgroups are identified by differently colored boxes and marked R1-R11.

Figure 6.

Circos plot of Markov jump counts (reflected by arrow width) illustrating the viral transitions chicken, duck and quail. The direction of arrows indicates the direction of the Markov jumps. Following Jeffreys [46], viral transition rates were considered statistically supported when Bayes Factor >3.0 (arrows with orange borders), strong support when BF > 10 (arrows with red borders) and decisive support when BF >100 (arrows with black borders).