2.1. Geographical zones
The survey was conducted in Gironde County, southwestern France, characterized by its contrasting landscape. The epidemiological parameters of the three defined zones were determined. The first zone, named Confluence, corresponds to the area where WNV-infected horses were reported in 2022 (Fig. 1). The confluence lies at the junction of two rivers, the Garonne and the Dordogne rivers, where large populations of mosquitoes can be expected. The second zone corresponds to the Arcachon basin. The Arcachon basin lies at the heart of the East Atlantic Flyway, one of the eight most important bird migration corridors on the planet. No cases of WNV infection were reported here. These two zones offer a protected habitat for resident and migratory bird species (represented in green in Fig. 1) that use wetlands as a stopover during their long journeys. These avian tranquility zones encourage interactions between local mosquitoes and competent avian hosts which promotes the establishment of the bird-mosquito-bird transmission cycle of WNV and USUV. The third zone covers the area between the Confluence and the Arcachon basin.
2.2. Study design and sample collection
The study was a cross-sectional prospective study. It took place in April and May 2023, prior to the occurence of seasonal orthoflavivirus activity, to reflect the exposure of horses to WNV, USUV and TBEV during the summer 2022. Animals were recruited by the Clinique Equine de Conques (Saint Aubin de Branne, Gironde, France) through social media and client mailing, according to our three different zones. All owners signed a consent form and treating veterinarians were informed of the procedure. This study was approved by the Ethics Committee for Clinical Research (ComERC) of the Veterinary School of Alfort (EnVA) (Agreement number: 2023-06-23).
Serum samples were collected from 494 adult (> 2-year-old) client-owned horses, ponies, and donkeys residing in 39 different stables of the three studied zones: 306 equids from 25 stables in the Confluence zone, 77 equids from 5 stables in the Arcachon zone, and 111 equids from 9 stables in the intermediate zone (Fig. 1). The animals were unvaccinated against WNV and had been living in the sampling area for the past 6 years throughout 2022 without traveling to the Mediterranean region.
The number of equids sampled in a stable depended on the stable size: all horses were sampled in small stables (fewer than 10 animals), 10 horses were sampled in medium stables (between 10 and 15 animals), and 15 horses were sampled in large stables (more than 15 animals). All animals were asymptomatic at the time of sampling. The 3 cases of 2022 were not included in our cohort, although two of the three corresponding stables were included (Fig. 1).
Serum samples were collected from each included animal. Blood samples were collected in vacutainer dry tubes and centrifuged at 5000 rpm for 5 minutes within 24 hours of collection. The serum was separated, stored at 4°C, and sent to the French National Reference Laboratory for West Nile virus, ANSES (Maisons-Alfort, France) for further analysis.
In addition to sample collection, each stable manager/owner completed a survey consisting in a personal face-to-face interviews to gather information on individual characteristics and stable management. This information included age, sex, hair colour, and type of housing. We also collected information about the environment, such as the distance to the nearest water surface, and the distance to the nearest special protection areas. SPAs have been created for the protection of wild bird species listed in the Annex 1 of the EU bird directive (Directive 79/409/EEC), or that serve as breeding, moulting, wintering, or staging areas for migrating birds. Five SPAs have been defined in the study area around wetlands (Fig. 1).
2.3. Serological testing
ELISA tests
Serum samples were, first, tested for IgG flaviviruses using the commercial Pan-Flavivirus ELISA “ID Screen Flavivirus competition” (Innovative Diagnostic, Montpellier, France). The protocol was performed according to the manufacturer's instructions. The results are expressed as %S/N. If a sample had a %S/N less than or equal to 40%, the sample was considered positive. If it is strictly greater than 50%, the sample is negative. If the result is between 40% and 50%, the sample is considered doubtful. Positive and doubtful results were confirmed for the presence of WNV, USUV and TBEV specific neutralizing antibodies using specific virus neutralization tests.
Virus-neutralization test (VNT)
To identify specifically against which flaviviruses the IgG antibodies detected by ELISA are directed, a virus neutralization test (VNT) was carried out for the three main flaviviruses circulating in France: WNV, USUV and TBEV. This test was performed in 96-well plates as described in Beck et al.[18] .
Briefly, the sera were diluted in a cascade by a factor of 2, from 1:5 to 1:160, in a final volume of 100µL. Fifty microliters of virus was added at an infectious dose of 100 TCID50. After 1h30 of incubation at 37°C and 5% CO2, 100µL of Vero cells was added at a concentration of 2.105 cells/mL. The neutralizing antibody titer was obtained by observing the cytopathic effect (CPE) under a light microscope after 3-, 4- and 5-days of incubation at 37°C and 5% CO2 for WNV, USUV and TBEV respectively. The last dilution of serum showing no CPE or a proportion of CPE less than a quarter of the well observed, was noted as the dilution of serum considered where there is still protection against the virus. The neutralizing antibody titer was therefore the inverse of the serum dilution.
Sera neutralizing more than one virus were considered positive for the virus neutralized at a fourfold higher dilution than all other viruses. If a fourfold difference in neutralizing antibody titers was not reached, the serum was considered undifferentiated between the viruses. Sera that tested positive by competitive ELISA and negative by VNT were considered negative.
Statistical analysis
All collected variables were described in terms of frequency distribution (qualitative data) or median and range (quantitative data) classified by the serological status.
Data were analysed using mixed effects logistic regression models, using the stable as a random effect. The dependent variable was the serological status of the animals, according to the seroneutralization test. Fixed effects were animal age (years), sex (male or female), hair colour (light or dark), the type of housing (always on pasture, or both on pasture and indoors), the distance in kilometres (km) to the nearest water surface (intermittent or permanent, of any type: river, canal, estuary, lake, marsh, reservoir etc.), and the distance in km to the nearest special protection area (SPAs).
The exponentiated regression coefficients produced odds ratio (OR) as a measure of the effect of the variables. Starting from the model with all the fixed effects, we selected the most parsimonious model using backward model selection based on the likelihood ratio test, p-value < 0.05.
We used the nonparametric approach proposed by Vaumourin et al. (2014) to test whether the association between positive VNT results against WNV, USUV and TBEV (i.e. the number of equids positive for 1, 2 or 3 of these viruses) significantly differed from the expected distribution under the null hypothesis of independent positive results.
All the statistical analyses were performed using R 4.3.3 (R Core Team 2024). Mixed effects logistic regression models were fitted using the lme4 package [20]. Model selection was performed using the buildmer package [21].