Association Between PRRSV ORF5 Genetic Distance and Differences in Space, Time, Ownership and Animal Sources Among Commercial Pig Herds

Abstract

The objective of this study was to investigate associations between genetic distance of porcine reproductive and respiratory syndrome virus (PRRSV) detected in Ontario swine herds, and the distance between the herds with respect to space, time, ownership and animal sources. PRRSV sequence data between September 2004 and August 2007 were obtained from the Animal Health Laboratory of the University of Guelph. Geographical coordinates were obtained from the Ontario Pork marketing board, and network information about ownership and animal suppliers was obtained using a telephone interview. The matrices of sequence, spatial, temporal and network distances were generated and were analysed using the Mantel test, and using linear-mixed models with P-values based on random permutations. A total of 438 PRRSV isolates from 329 premises and 232 ownerships were originally included; 57 of the isolates were considered vaccine type. The Mantel correlation test indicated that there was positive correlation between sequence distance and geographic distance (r = 0.11, P = 0.001), as well as sequence distance and temporal distance (r = 0.03, P = 0.03), with similar results reported after adjusting for the ownership distance. Mantel correlogram suggested existence of spatial correlation up to ~30 km distance. Multivariable linear-mixed model for association between genetic distance and space-time distance was characterized by the three-way interaction among space, time and ownership (P < 0.001). It suggested that positive association between sequence similarity and spatial proximity exists in herds under different ownerships, but its magnitude is very small. In contrast, for pairs of herds under identical ownership, the spatial association was more complex. This could be a consequence of interactions within ownerships, or alternatively decisions made about sampling of herds for diagnostic purposes. Of the networks evaluated, ownership (P < 0.001) and gilt supplier (P < 0.001) showed the highest magnitude of association with genetic distance and should be investigated further for their impact on disease spread.

Lessons learned and knowledge gaps about the epidemiology and control of porcine reproductive and respiratory syndrome virus in North America

“…Some studies relied on serologic evidence of PRRSV circulation in various herd types. Some investigators have relied on descriptive statistics, 41,140 contingency tables, 40 Cox proportional hazard models, 73,142 generalized additive models and methods to detect spatial and space-time clustering and clusters commonly used in spatial epidemiology, 141 statistical models of individual-level infectious disease transmission, 35 results of the Mantel correlation test, 39,[143][144][145] and linear regression based on a permutation approach. 40,[143][144][145] The analytic approach has also differed among studies.…”

“…140 Others have involved broad classifications based on PRRSV types (European vs North American) on the basis of serologic results, 77 genotypes on the basis of RFLP patterns, 35,141 classification based on phylogenetic trees and different cutoff points to categorize variants into groups, 40,41,142 or matrices of similarities to assess correlations between the similarity of nucleotide sequences and spatial and other proximities among herds. 145 As a part of such investigations, one of the largest uncertainties continues to be importance of area spread (also commonly referred to as local spread), which is often discussed in the context of airborne transmission. Some investigators have relied on descriptive statistics, 41,140 contingency tables, 40 Cox proportional hazard models, 73,142 generalized additive models and methods to detect spatial and space-time clustering and clusters commonly used in spatial epidemiology, 141 statistical models of individual-level infectious disease transmission, 35 results of the Mantel correlation test, 39,[143][144][145] and linear regression based on a permutation approach.…”

“…Investigators of epidemiological studies 40,77 identified area spread and argued that aerosol transmission played a critical role in transmission. 141 Few studies have been conducted to investigate membership in important networks such as ownership, 35,41,141,144,145 sources of animals and semen, 35,77,140,141 or transportation or service providers. Furthermore, several authors reported the possibility of area spread for some discrete genotypes, although such spread was not identified as the dominant mechanism of spread 41,141 or could be explained in some cases because of common sources of animals.…”

Novel analytic tools for the study of porcine reproductive and respiratory syndrome virus (PRRSv) in endemic settings: lessons learned in the U.S.

“…Many studies on the evolutionary epidemiology of PRRSv have been attempted with the ultimate goals of assessing the evolutionary patterns of the virus and guiding the decision making process related to the control and prevention resources. Some studies focused on establishing association between the evolutionary features of PRRSv and epidemiological characteristics of outbreaks in different geographical levels [52][53][54][55][56][57]; others have focused on discriminating between novel and preexisting strains to model their spread and maintenance within affected populations [19,[58][59][60]. However, most of those observational studies used classical phylogenetic methods to either genotype newly emerging PRRSv strains on the basis of restriction fragment length polymorphism (RFLP) patterns [19,59,61], or assess correlations between the similarities of nucleotide sequences and other epidemic features [19,52,57,62].…”

“…Accordingly, the Bayesian approach provides a natural way to estimate (and accommodate) uncertainty in the phylodynamic model parameters, including the virus phylogeny, divergence times, and history of geographic spread [66]. Bayesian models have been demonstrated, for example, that the difference in the number of nucleotides between PRRSv sequences is an inaccurate measure of true phylogenetic distance [58].…”