Modeling Marek's disease virus transmission: A framework for evaluating the impact of farming practices and evolution

Abstract

Marek’s disease virus (MDV) is a pathogen of chickens whose control has twice been undermined by pathogen evolution. Disease ecology is believed to be the main driver of this evolution, yet mathematical models of MDV disease ecology have never been confronted with data to test their reliability. Here, we develop a suite of MDV models that differ in the ecological mechanisms they include. We fit these models with maximum likelihood using iterated filtering in ‘pomp’ to data on MDV concentration in dust collected from two commercial broiler farms. We find that virus dynamics are influenced by between-flock variation in host susceptibility to virus, shedding rate from infectious birds, and cleanout efficiency. We also find evidence that virus is reintroduced to farms approximately once per month, but we do not find evidence that virus sanitization rates vary between flocks. Of the models that survive model selection, we find agreement between parameter estimates and previous experimental data, as well as agreement between field data and the predictions of these models. Using the set of surviving models, we explore how changes to farming practices are predicted to influence MDV-associated condemnation risk (production losses at slaughter). By quantitatively capturing the mechanisms of disease ecology, we have laid the groundwork to explore the future trajectory of virus evolution.

Managing Marek’s disease in the egg industry

“…In all scenarios, if the outbreak persisted, disease spread was rapid, with the majority of hens infected by the end of a cohort. The high prevalence is consistent with other modelling studies of MDV transmission within broiler farms that featured significantly shorter cohorts durations (Atkins et al, 2013b; Kennedy et al, 2018). The lower laying-hen numbers in Aviary systems carry an additional disadvantage, as lower stocking densities and more open designs render flocks more prone to hen cannibalism (Ahammed et al, 2014).…”

“…All model parameters were extracted from available data (Atkins et al, 2013a, 2011a; Cui et al, 2016; Kennedy et al, 2018; Ralapanawe et al, 2016b, 2016a; Zhang et al, 2015). Parameters associated to barn characteristics where obtained from communication with Burnbrae Farms in Ontario Canada, or the literature.…”

“…Data on the virulence level of the MDV pathotype and mortality, in addition to the standard assumption of Exponentially distributed parameters that is typical of compartmental models(Greenhalgh and Day, 2017; Hethcote and Tudor, 1980), was used to estimate MDV mortality rates ( ν ) for each MDV strain (Ralapanawe et al, 2016a). We base the viral particle removal rate ( δ ) on the barn ventilation system and estimates of the decay rate of the viral particles (Kennedy et al, 2018). Therefore, δ is taken as the sum of the decay rate and the average air exchange rate of a typical barn (Table 1, Webappendix).…”

“…As infected laying-hens are likely symptom free due to vaccination, and cohort sizes range from 30,000-100,000 hens (Holt et al, 2011), the identification and removal of MDV infected hens is difficult. Additionally it has been shown that MDV is often reintroduced to barns as often as once per month (Kennedy et al, 2018). Thus, other measures are needed to prevent or limit MDV infection.…”

Pathogen transmission from vaccinated hosts can cause dose-dependent reduction in virulence

“…This would suggest that contacts exposed to vaccinated shedders experienced overall lower cumulative exposure dose, including from other infected contacts, over the course of 8 weeks, resulting in milder symptoms and lower mortality. This negative feedback on the environmental pathogen burden strongly advocates the application of MD intervention strategies that reduce either within-host or environmental virus load, even if only moderately 41 . In general, depending on the relationship between exposure dose and subsequent within-host replication in any particular system, targeting reduction in pathogen load in intervention strategies may have greater positive knock-on effects than currently assumed.…”