Beatriz Martínez-López, Alda F. A. Pires, Dale Polson. ISERPD 2015.
Since its first description in the US in the late 80s, porcine reproductive and respiratory syndrome virus (PRRSV) is still one of the main causes responsible for large economic losses in the US swine industry -estimated in more than $664 million per year (1)- and worldwide. PRRSV is a highly variable RNA virus of the family arteriviridae that is classified into genotypes by restriction fragment and length polymorphism (RFLP) (2). While the direct and indirect transmission of PRRSV has been demonstrated under experimental conditions, little observational studies have been conducted to quantify the role of each transmission route under diverse epidemiological conditions (i.e., different site types, farm sizes, animal movement patterns or nearby farm densities). Certainly a better knowledge of the dynamics of PRRSV direct or indirect transmission under field conditions will inform decisions regarding the implementation of specific intervention strategies (e.g., increase of biosecurity, pre-movement testing, area vaccination, air filtration, etc.) to more cost-effectively prevent and control PRRSV.
Here we proposed the use of multilevel models in combination with phylogenetic, space-time and social network analyses (SNA) to get further insights into the PRRSV spread dynamics and transmission routes under diverse epidemiological conditions.
Materials and methods
More than 200 PRRSV sequences (ORF 5 region) collected over 3 years from PCR-positive field samples were used to estimate the genetic diversity of PRRSV. More than 150,000 animal movement records during the same period of time were used to evaluate the animal movement patterns and characteristics. The spatial location of sites (i.e., geographical coordinate), the time when the PRRSV sequence is collected as well as other on site characteristics (i.e., type of farm, etc.) were used to compute the spatial and temporal distance and to evaluate differences among types of sites.
A multilevel regression model was used to quantify the association between the pairwise genetic distance of two isolates (ij) belonging to the same genotype, the spatial and temporal proximity and the frequency and characteristics of swine movements (i.e., bidirectional number of incoming and outgoing shipments, pairwise comparison of the farm centrality measures and of the number of animals moved). Phylogenetic analysis and pairwise comparisons were done with Disease BioPortal® and SNA metrics and multilevel modeling in R-language v.3.1.1 (R Core Team, 2014).
Model results show that PRRSV isolates are spatially and temporally clustered [Spatial and Temporal distance_ij] and differ with the type of site [Diff_site_ type_ij] (Table 1). Similarly, there was a significant but negative association (OR<1) between the genetic diversity of two PRRSV isolates and sharing pigs [Movs_between_ij] or having high centrality measures [indegree_ij and betweenness_ij] in sites i and j.
Table 1. Multilevel logistic regression model results
We used phylogenetic, space-time and social network analyses in combination with multilevel regression modeling to evaluate the association between PRRSV genetic diversity (which is a proxy for multiple PRRSV wild-type infections or high PRRSV evolution rate) and space, time and pig movement characteristics in swine production systems of US. Results strongly suggest evidence of PRRSV transmission occurring by pig movements between sites. We found also a significant, but smaller, association between genetic distance and space-time distance, suggesting area spread among herds. This may be associated to aerosol transmission or other factors clustered in space. Results are in agreement with hypothesis and previous studies conducted in Canada.