Molecular prevalence of equine alphaherpesvirus-1 shedding in healthy broodmares in Ontario

Abstract

Equine herpesvirus-1 (EHV-1) is ubiquitous in the horse population, but prevalence estimates have ranged from 3 to 88% depending on the population and method of sampling. No prevalence studies have been carried out in Ontario, Canada. The objective of this study was to measure the prevalence of EHV-1 shedding in healthy broodmares in Ontario. A total of 381 mares from 42 farms in Ontario were sampled, including pregnant and barren broodmares. Samples were collected from the nose, vagina, and blood of each mare up to 6 times from December 2016 through October 2017 using a cross-sectional study design. The EHV-1 glycoprotein B (gB) copy number was measured using droplet digital polymerase chain reaction (ddPCR). A survey was completed at time of sampling regarding signalment, pregnancy status, and vaccination. Overall, 85% of the mares sampled were positive for EHV-1 from at least one site, on at least one occasion. Samples were positive 8.1%, 15.8%, and 17.2% of the time from the nose, vagina, and white blood cells, respectively. Pregnant mares had increased odds of shedding virus from the nose [odds ratio (OR) = 1.50, 95% confidence interval (CI): 1.03 to 2.18, P = 0.037]. Vaccination only reduced the odds of virus presence in blood (OR = 0.70, 95% CI: 0.49 to 0.99, P = 0.043). Advanced gestation appeared to also have a sparing effect on virus presence in blood (OR = 0.89, 95% CI: 0.85 to 0.93, P < 0.001). Most mares in Ontario were positive for EHV-1 despite being healthy and vaccinated, although the amount of viral DNA detected was extremely low. In addition, the vagina was identified as a source of viral shedding.

Introduction

Equine herpesvirus-1 (EHV-1), now renamed equine alphaherpesvirus-1, causes disease in horses of all ages. Foals are suspected of being exposed to, and are frequently infected with, EHV-1 from their dams (1,2), after which the virus may live latently within the bronchial lymph nodes, trigeminal ganglia, and peripheral blood mononuclear cells (PBMCs) (3,4).

The clinical signs produced by EHV-1 most commonly include cough and nasal discharge, distal limb edema, late-term abortion, and myeloencephalopathy (5–7). Throughout their lives, horses are at risk of repeated exposure to EHV-1 from direct contact with other horses, indirect contact with contaminated objects, called fomites, or recrudescence of a latent virus.

As recrudescence of latent EHV-1 is possibly the most important factor in precipitating natural outbreaks of EHV-1 neurological and abortive disease rather than novel infection (8), a better understanding of the population prevalence of EHV-1 would be beneficial for understanding individual risk. The reported prevalence of EHV-1 shedding ranges from 3 to 88%, depending on the sampled population and method of detection (7,9,10).

In an attempt to increase the sensitivity of conventional polymerase chain reaction (PCR), Allen et al (8), for example, reported a prevalence of 54% using a magnetic bead-based, sequence-capture, nested PCR on postmortem samples from the submandibular lymph nodes of Thoroughbreds in Kentucky, USA. In another study, it took repeated sampling over 6 y to get a final prevalence estimate of 88% in abattoir samples in Brazil (10).

Due to the threat to horses of all ages, vaccines for EHV-1 have been distributed since the 1960s. Although these vaccines were initially lauded for the reduction in abortion frequency (11), outbreaks of disease and EHV-1-associated abortions continue to occur around the world. No studies have directly assessed the prevalence of the EHV-1 within a population prior to and after vaccination. As commercial vaccines have been unsuccessful at preventing infection, they have been labeled as an aid in reducing clinical signs only. However, a study from Australia confirmed that despite aggressive vaccination, abortion rates remained largely unchanged after vaccination (12). Similarly, outbreaks of both myeloencephalitis and abortion associated with EHV-1 have occurred in vaccinated herds in Ontario, suggesting an incomplete protection in the Ontario equine population. Shedding and reactivation is thought to occur intermittently in response to external stimuli, such as stress and other unknown factors (13,14).

During the reactivation period, humoral immunity is rendered essentially ineffective as viral spread can occur from cell-to-cell, partially explaining the lack of protection from vaccination (15). Two in-vitro models discovered evidence of direct cell-to-cell transfer of virus allowing infection of uterine and nervous endothelial cells by infected PBMCs, as the presence of virus neutralizing (VN) antibodies did not alter the infection rate significantly (16–18). Furthermore, virus shedding has been identified without clinical signs and recrudescence (8,14). The objective of this study was to measure the prevalence of EHV-1 shedding in healthy broodmares in Ontario with repeated sampling up to 6 times from December 2016 through October 2017. Shedding from the mucosal surfaces of the nose and vagina was explored using the more sensitive droplet digital PCR (ddPCR). The association of shedding with vaccination, signalment, and breeding status was also investigated.

Materials and methods

Farm participation

Based on a suspected prevalence of EHV-1 in 80% of the population from previous data (10) and a total active breeding population of 17 000 mares in Ontario according to data from 2011 provided by Equine Canada (19), it was determined that a sample size of at least 243 mares was required to achieve 5% precision [95% confidence interval (CI)] in estimating the current prevalence.

A total of 38 participating farms were identified using a survey distributed to Ontario horse breeders. Survey results and data were published previously (20). Four additional farms were added after the survey, via communication with farms. A broodmare survey was completed at the time of each sample collection. Data collected included: age; breed; pregnancy status; breeding and foaling dates; EHV-1 vaccination status and most recent vaccination date; EHV-1 vaccine product used; and any history of abortion.

The Institutional Animal Care Committee at the University of Guelph approved the study (AUP#3257), which conformed to the standards of the Canadian Council on Animal Care.

Sample collection

To assess the impact of the sampling site on the prevalence of EHV-1 and association with demographic and management variables, research team members sampled from the nose, vagina, and blood. Samples were collected bimonthly from December 2016 through October 2017 by farm personnel or the authors (CJC, MMB, and LGA). The authors wore disposable gloves between farms and followed each farm’s biosecurity protocols, including boot covers and protective frocks.

A minimum of 10 mL of whole blood was collected in an evacuated ethylenediaminetetraacetic acid (EDTA) vacutainer tube and an additive-free serum vacutainer tube (Monoject; Tyco HealthCare Group, Mansfield, Massachusetts, USA) by direct venepuncture with a 20-gauge needle. Nasal and vaginal secretions were collected using 2 15-cm rayon-tipped swabs (CultureSwab Liquid Stuart; Becton Dickinson, Sparks, Nevada, USA). The swabs were advanced into the ventral meatus of either the right or left nostril or the vagina and rotated to improve mucus yield. Samples were kept refrigerated and transported overnight to the laboratory at the University of Guelph, Ontario. Serum was obtained by centrifugation (5000 × g for 10 min). Swab and serum samples were stored at −20°C until processing and labelled with an anonymous number the researchers could decode for subsequent analyses.

White blood cell DNA isolation

For white blood cell (WBC) nucleic acid extraction, the buffy coat was aspirated using a single-use transfer pipette after centrifugation (5000 × g for 10 min) and placed in a 3-mL conical Eppendorf tube. Any remaining red blood cells (RBCs) were lysed with 2 mL of MilliQ water. The samples were centrifuged (10 000 × g for 2 min) to form a pellet. The supernatant of lysed RBC debris was discarded and the pellet was frozen at −20°C.

DNA isolation

Viral genomic deoxyribonucleic acid (DNA) was extracted using an EZNA Viral DNA Isolation Kit (Omega Bio-tek, Norcross, Georgia, USA). For WBCs, 40 μL of TL buffer was mixed with approximately 30 μL of WBC material and incubated for at least 4 h at 40°C. A sterile steel bead was added and the sample was placed in a TissueLyzer (QIAGEN, Valencia, California, USA) for 3 min to homogenize the sample and disrupt the WBC membranes. This step was crucial to the maximal yield of DNA from WBCs since without this step the cellular debris would occasionally congeal and clog the columns provided.

A protease enzyme and BL buffer were added and the sample was incubated at 70°C for 20 min. Once ethanol was added, the entire volume was loaded into the EZNA DNA isolation columns. The columns were spun at 14 000 × g for 1.5 min and then washed twice with DNA wash buffers, as per the kit protocol. DNA was eluted into 100 μL of elution buffer and frozen at −40°C.

Nasal and vaginal swab tips were placed in conical Eppendorf tubes and covered with TL buffer and protease enzyme prior to incubation as previously described. As the volume of reagents had to cover the entire rayon tip, the volumes of TL buffer and protease enzyme were doubled to maintain the concentration of the final solution. Volumes of BL buffer and ethanol were also increased in compensation to maintain the effective ratios described by the manufacturer. Once the samples were centrifuged through the DNA isolation columns, the protocol was the same as previously described. The final DNA concentration was determined by a spectrophotometric method with a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, Delaware, USA).

Droplet digital PCR

Droplet digital PCR (ddPCR) was carried out by the Agriculture and Food Lab, Lab Services Division, University of Guelph using ddPCR with a Taq polymerase and previously published glycoprotein B (gB)-specific primer (21) to quantify viral pathogen load in each sample.

The ddPCR reaction mixture consisted of 1× ddPCR Supermix for Probe (Bio-Rad, Mississauga, Ontario), with previously published primers (21), at a concentration of 96 nM [EHV-1F (5′CATGTC AACGCACTCCCA3′), EHV1R (5′GGGTCGGGCGTTTCTGT3′)], 64 nM concentration of the EHV-1 probe (5′FAM-CCCTACGCTG CTCC-MGB-NFQ3′), and 4 μL of sample DNA in a final volume of 25 μL.

From each PCR reaction mixture, 20 μL were mixed with 70 μL of Droplet Generation Oil for Probes (Bio-Rad) in a DG8 Cartridge (Bio-Rad). The PCR droplets were then generated using a QX200TM Droplet Generator (Bio-Rad). From each droplet mix, 20 μL were transferred to a 96-well PCR plate (Bio-Rad). The plate was sealed with a foil heat seal using a PX1TM PCR Plate Sealer (Bio-Rad). Validation involved serial dilution of an artificial fragment (synthetic control) to determine a suitable range for ddPCR quantitation.

For data analysis runs, each plate included 92 samples, 2 positive laboratory controls with known EHV-1 concentrations, 1 positive control (synthetic gene fragment), and 1 negative control (water). PCR amplification was carried out on a GeneAmpTM PCR System 9700 (Applied Biosystems, Foster City, California, USA) under the following settings: 95°C for 10 min, followed by 48 cycles at 95°C for 20 s and 62°C for 40 s, followed by 1 final cycle at 98°C for 10 min.

After amplification, droplets from each well were read automatically on a QX200TM Droplet Reader (Bio-Rad). ddPCR data were acquired and analyzed with QuantaSoft Software (Bio-Rad) and recorded as copies/μL. The final viral load copies/mL of each sample was determined by multiplying the average copy number/μL of PCR mixture in each well by the sample dilution factor.

Statistical analysis

A cross-sectional study design was used to estimate the prevalence of EHV-1 within the Ontario broodmare population and the associations between EHV-1 status and the following independent variables: month of sampling, age, breed, pregnancy status at time of sampling, whether or not the mare had been vaccinated for EHV-1 within the previous 2 y or previous 2 mo, which EHV-1 containing vaccine was used most recently, whether the mare had a history of abortion, month of gestation (if applicable), and the number of months since most recent foaling (if applicable).

Descriptive statistics, including medians, ranges, and interquartile ranges, were reported for continuous variables and proportions with their 95% CIs were generated for categorical variables. EHV-1 status was dichotomized and considered “positive” when DNA was detected using ddPCR, whereas samples with no detectable DNA were reported as “negative.” Multi-level logistic regression models were used to examine the association between demographic and management characteristics of broodmares and their EHV-1 ddPCR results at each sampling site and overall status based on a parallel interpretation of multiple sampling sites (i.e., considered positive if any site tests positive).

The assumption of linearity between continuous independent variables and the log odds of the outcome was examined using lowest curves (i.e., locally weighted regression). If the assumption was not met, the independent variable was categorized or, if appropriate, a quadrative relationship was modeled. Random intercepts were included to account for autocorrelation within individual animals and within a farm. If a variance component was less than 10⁻¹⁰ and if the model fit was not improved based on changes in Akaike’s Information Criterion (AIC), that random effect was excluded from the final model.

In addition, for these multi-level models, model fit was assessed by examining the assumptions of homogeneity of variance and normality of the best linear unbiased predictors (BLUPS) graphically; Pearson residuals were also examined to identify outlying observations/covariate patterns. If the assumptions concerning the BLUPS were not met, we compared the fit of the multi-level model to an ordinary logistic regression model to determine if model fit was improved by including the random effect(s) using AIC. All statistical tests were conducted using Stata Intercooled 16.0 (StataCorp, College Station, Texas, USA).

Results

Descriptive statistics

In total, 381 broodmares on 42 farms were sampled. On 41 of the farms, at least 1 mare tested positive. Using ddPCR of the EHV-1 gB gene, the presence of EHV-1 DNA was quantified in nasal and vaginal secretions and PBMCs. The numbers of EHV-1 gB copies/μL of original DNA extract identified using ddPCR ranged from 0.00 to 55 625.00 in nasal samples, 0.00 to 23.75 in vaginal samples, and 0.00 to 618.13 in white blood cells (WBCs). The cumulative period prevalence of EHV-1 in broodmares sampled was 85.0% (95% CI: 81.1 to 88.5%) of mares sampled (Table I, Figure 1).

Table I.

Prevalence estimates of the EHV-1 status^a of healthy Ontario broodmares and farms (overall^b and stratified by sampling site) per sampling period from December 2016 to October 2017.

	December 2016		February 2017		April 2017		June 2017		August 2017		October 2017		Overall
	N	n (%) positive	N	n (%) positive	N	n (%) positive	N	n (%) positive	N	n (%) positive	N	n (%) positive	N	n (%) positive
Mares sampled	352	147 (41.7%)	356	90 (25.3%)	311	116 (37.3%)	307	127 (41.4%)	262	94 (35.9%)	266	92 (34.6%)	381	324 (85.0%)
Farms sampled	39	36 (92.3%)	42	32 (76.2%)	39	34 (87.2%)	35	31 (88.6%)	36	24 (66.7%)	35	26 (74.3%)	42	41 (97.6%)
Nasal positive	351	28 (8.0%)	353	29 (8.2%)	308	41 (13.3%)	293	12 (4.1%)	261	11 (4.2%)	266	28 (10.5%)	1832	149 (8.1%)
Vaginal positive	351	45 (33.5%)	354	47 (13.3%)	306	69 (22.6%)	307	50 (16.3%)	261	30 (15.3%)	265	40 (15.1%)	1844	291 (15.8%)
WBC positive	351	100 (45.2%)	354	19 (5.4%)	305	19 (6.2%)	292	83 (28.4%)	261	57 (21.8%)	265	38 (14.3%)	1828	316 (17.2%)

^aEHV-1 status based on droplet digital polymerase chain reaction (ddPCR).

^bMare classified as positive if any sampling site during the testing period was positive.

Figure 1.

Molecular prevalence of equine herpesvirus-1 (EHV-1) in nasal, vaginal, and white blood cell count samples, and overall in healthy broodmares according to month of sampling.

The largest number of mares were sampled in February 2017, but the largest percentage of positive samples occurred in December 2016 (Table I, Figure 1). Prevalence of EHV-1 was lowest in nasal swabs (149/1832, 8.1%, 95% CI: 6.9 to 9.4%) and highest in WBCs (316/1828, 17.3%, 95% CI: 15.6 to 19.1%). There was no association between sampling sites, with very different outcome odds at each site when investigating each risk factor.

Risk factors analyzed

Month

A notable variation in prevalence of positive samples was observed in the different months sampling occurred, with a lower percentage of positive samples overall in February (Table I, Figure 1). The odds of broodmares testing positive based on overall status were significantly lower in February compared to all other months (Table II). The odds of positive nasal and vaginal swab samples were significantly higher in April than in February, but WBC positivity was not significantly different (Table II). Viral presence in nasal swabs was significantly lower in June than in February [odds ratio (OR) = 0.48, 95% CI: 0.24 to 0.96%, P = 0.039]. Except for in April, all months sampled had higher odds of WBC samples being positive than in February (Table II).

Table II.

Results of univariable multi-level^a logistic regression models examining the associations between EHV-1^b status for each sample type and sampling month, demographic characteristics, vaccination status, and reproduction history.

Variables	Nasalc			Vaginald			WBCe			Overall positivity
	OR	95% CI	P-value	OR	95% CI	P-value	OR	95% CI	P-value	OR	95% CI	P-value
Month
February			< 0.001			0.012			< 0.001			< 0.0001
December	0.97	0.56 to 1.66	0.900	0.96	0.62 to 1.49	0.851	7.28	4.33 to 12.26	< 0.001	2.13	1.55 to 2.94	< 0.001
April	1.72	1.04 to 2.85	0.036	1.93	1.28 to 2.92	0.002	1.20	0.62 to 2.31	0.596	1.77	1.27 to 2.48	0.001
June	0.48	0.24 to 0.96	0.039	1.27	0.82 to 1.96	0.288	7.29	4.28 to 12.41	< 0.001	2.10	1.51 to 2.93	< 0.001
August	0.49	0.24 to 1.00	0.051	1.21	0.76 to 1.93	0.415	5.34	3.07 to 9.31	< 0.001	1.70	1.19 to 2.41	0.003
October	1.31	0.75 to 2.27	0.338	1.19	0.75 to 1.89	0.454	3.09	1.73 to 5.53	< 0.001	1.59	1.12 to 2.26	0.010
Age	0.99	0.95 to 1.03	0.546	1.00	0.97 to 1.03	0.759	0.99	0.96 to 1.02	0.410	0.99	0.97 to 1.01	0.434
Breed
Other			0.277			0.943			0.087			0.205
Draft	1.60	0.60 to 4.22	0.346	0.89	0.38 to 2.08	0.780	1.52	0.73 to 3.19	0.264	1.60	0.88 to 2.90	0.121
Standardbred	1.09	0.44 to 2.71	0.853	0.94	0.42 to 2.08	0.877	0.78	0.40 to 1.53	0.469	1.12	0.65 to 1.92	0.681
Thoroughbred	0.88	0.36 to 2.13	0.776	0.95	0.45 to 2.03	0.895	0.76	0.41 to 1.42	0.388	0.98	0.59 to 1.64	0.948
Warmblood	0.79	0.31 to 1.98	0.610	0.79	0.36 to 1.73	0.553	0.91	0.47 to 1.76	0.787	1.05	0.62 to 1.78	0.853
Pregnant	1.50	1.03 to 2.18	0.037	0.89	0.68 to 1.16	0.377	1.05	0.81 to 1.36	0.710	1.06	0.87 to 1.30	0.562
Ever vaccinated	1.00	0.60 to 1.67	0.991	1.04	0.69 to 1.58	0.850	0.70	0.49 to 0.99	0.043	0.84	0.64 to 1.12	0.230
Simple product
Innovator			0.403			0.445			0.019			0.775
Other	1.11	0.39 to 3.21	0.841	5.71	0.65 to 50.30	0.117	0.29	0.11 to 0.78	0.015	1.45	0.61 to 3.47	0.401
Pneumabort	1.50	0.85 to 2.64	0.166	7.72	1.00 to 59.74	0.050	0.30	0.14 to 0.62	0.001	1.18	0.67 to 2.10	0.566
Prestige	0.92	0.40 to 2.12	0.839	6.44	0.81 to 51.41	0.079	0.32	0.14 to 0.73	0.006	1.06	0.55 to 2.05	0.854
Prodigy	1.55	0.85 to 2.83	0.157	6.70	0.86 to 52.31	0.070	0.30	0.14 to 0.63	0.002	1.36	0.75 to 2.47	0.313
Vetera	Omitted due to collinearity			6.37	0.82 to 49.79	0.077	0.41	0.19 to 0.86	0.019	1.14	0.63 to 2.07	0.667
Abortions	0.68	0.36 to 1.27	0.222	0.97	0.64 to 1.48	0.887	0.90	0.59 to 1.36	0.603	0.93	0.68 to 1.26	0.625
Gestation (mo)	1.04	0.98 to 1.11	0.228	0.96	0.92 to 1.01	0.164	0.89	0.85 to 0.93	< 0.001	0.95	0.91 to 0.98	0.006
Months since foaling	0.98	0.85 to 1.12	0.728	0.93	0.85 to 1.03	0.158	1.04	0.96 to 1.14	0.345	0.99	0.92 to 1.06	0.724
Months since vaccination	0.95	0.89 to 1.01	0.078	1.01	0.97 to 1.04	0.751	1.03	1.00 to 1.06	0.098	1.00	0.97 to 1.02	0.840
Vaccinated < 2 mo	1.37	0.93 to 2.03	0.112	0.95	0.71 to 1.27	0.754	0.64	0.49 to 0.85	0.002	0.92	0.74 to 1.14	0.441

^aRandom intercept for farm.

^bEHV-1 status based on droplet digital polymerase chain reaction (ddPCR).

^cWithin nasal sampling models, the farm variance ranged from 0.11 to 0.25.

^dWithin vaginal sampling models, the farm variance ranged from 0.08 to 0.44.

^eWithin WBC sampling models, the farm variance ranged from 0.01 to 0.20.

^FWithin “Overall” sampling models, the farm variance ranged from 0.01 to 0.0.

OR — Odds ratio; CI — Confidence interval.

Age

The ages of broodmares sampled varied from 2 to 27 y (median: 10, IQR: 7 to 14). The presence of EHV-1 particles in the nose, vagina, and blood was not associated with age (Table II).

Breed

The sampled population included a large variety of breeds, representative of those present in Ontario: Thoroughbreds, Standardbreds, Dutch and Hanoverian Warmbloods, Belgians, Clydesdales, Friesians, Shires, Percherons, Canadian draft horses, Rocky Mountain and Quarterhorses, Fjord, and pony breeds.

Breeds were reduced to 5 categories for analysis: Thoroughbreds (589/1853, 31.8%, 95% CI: 29.7 to 34.0%); Standardbreds (565/1853, 30.5%, 95% CI: 28.4 to 32.6%); Warmbloods (Dutch and Hanoverian; 446/1853, 24.1%, 95% CI: 22.1 to 26.1%); Draft horses (Belgian, Clydesdale, Friesian, Shire, Percheron, and Canadian); 154/1853, 8.3%, 95% CI: 7.1 to 9.7%); and Other (Rocky Mountain and Quarter Horse, Fjord, and pony breeds; 80/1853, 4.3%, 95% CI: 3.4 to 5.3%). No breed was associated with viral presence (Table II).

Pregnancy and foaling

Over the sampling period, 64.2% (1189/1853, 95% CI: 61.9 to 66.4%) of the samples were collected from pregnant mares. Pregnancy had no significant effect on the odds of testing positive for EHV-1 overall, either in vaginal or WBC samples (Table II). In nasal swabs, however, the odds of testing positive were significantly greater in pregnant mares (Table II).

During gestation, the odds of testing positive significantly decreased based on WBC sample testing and overall status (Table II). Finally, neither the month since foaling nor a history of abortions was associated with the odds of testing positive for EHV-1 at any sampling site or based on overall status (Table II).

Vaccination

Most mares sampled had been vaccinated at the time of sampling (85.6%, 1586/1853, 95% CI: 83.9 to 87.2%). The vaccines used included multivalent vaccines from Prestige (Merck), Innovator (Zoetis), and Vetera (Boehringer Ingelheim) product lines and Calvenza (Boehringer Ingelheim); as well as monovalent vaccines labeled to aid in preventing EHV-1-associated abortion (Prodigy from Merck, Pneumabort K + 1b from Zoetis).

The vaccines used were reduced to 6 categories: Pneumabort K + 1b (572/1589, 36.0%, 95% CI: 33.6 to 38.4%); Prodigy (381/1589, 23.9%, 95% CI: 21.9 to 26.2%); Vetera (351/1589, 22.1%, 95% CI: 20.1 to 24.2%); Prestige (182/1589, 11.5%, 95% CI: 9.9 to 13.1%); and Innovator (38/1589, 2.39%, 95% CI: 1.7% to 3.3%, as well as an “Other” category (65/1589, 4.1%, 95% CI: 3.2 to 5.2%) that included an “unknown combination vaccine” and Calvenza. The odds of EHV-1 DNA being present in WBCs were lower in vaccinated mares (OR = 0.70, 95% CI: 0.49 to 0.99%, P = 0.043). Similarly, recent vaccination (within 2 mo of sampling) was associated with lower odds of WBC samples being positive (Table II).

Individual vaccine products were investigated for their role in the presence of EHV-1 particles and all vaccine products had lower odds of being positive on WBC samples than the referent, “Innovator” (Table II).

Random effects and diagnostics

Random intercepts were included in all models to control clustering by farm. The variance component at the mare level was very small (i.e., < 10⁻¹⁰) and was therefore removed. The best linear unbiased predictions (BLUPs) generally met the assumptions of homoscedasticity and normality. For most models, the fit was improved with the inclusion of the random intercept for farm based on the AIC. All potential outlying observations were investigated, but their removal did not impact the final models.

Discussion

The presence of EHV-1 viral DNA particles was essentially ubiquitous in the study population of Ontario horses. Using ddPCR, EHV-1 was found in 97.6% of Ontario herds sampled and in 85.0% of broodmares. The ddPCR period prevalence is comparable to the 86% prevalence of seroconversion described by Matumoto et al (22). Interestingly, the difference in sampling techniques and demographics appears to play a role in the prevalence estimates reported in the literature. In addition to the well-described nose and blood sources, to the authors’ knowledge, this is the first report describing the use of vaginal swabs for detecting EHV-1 DNA.

Droplet digital PCR (ddPCR) has been commercially available since 2011 (23). As with quantitative PCR (qPCR), ddPCR technology uses Taq polymerase in a standard PCR reaction to amplify a target DNA fragment, in this case gB. However, ddPCR partitions the PCR reaction into thousands of individual reaction vessels (droplets) prior to amplification and acquires raw data at the reaction end point in the form of a positive-negative result for every droplet.

The advantage of quantification of DNA without standard curves or conversions is improved precision and reproducibility of data compared to qPCR, especially in the presence of sample contaminants. Therefore, ddPCR technology can be used for extremely low-target quantitation or in variably contaminated samples, in which the sample dilution requirements would likely generate undetectable target levels using qPCR (24).

In this population, the majority of mares were positive for EHV-1 and were so in multiple sampling sites at each timepoint, despite being healthy and vaccinated. To the best of the authors’ knowledge, only one other study has investigated the presence of EHV-1 in a clinically healthy, non-experimental cross-sectional study, but this was done in a postmortem setting (8).

As shedding of EHV-2 and EHV-5 has been identified in otherwise asymptomatic horses (25,26), it would follow that EHV-1 would be found in this otherwise healthy population. A study from Australia also described EHV-1 silently circulating throughout an equine population (12). Virus transmission has been described through nose-to-nose contact and fomites, as large numbers of viral particles have been noted in nasal secretions, with and without the presence of viremia (27). This is also consistent with the findings in this study in which the presence of virus at each sampling site appeared to occur independently of each other. Although stress can induce herpesvirus recrudescence (28,29) in these cases, shedding was increased from nasal samples in pregnant mares but was not increased by foaling.

The presence of viral particles in the vagina in the present study is a novel finding but is unsurprising as the virus is known to affect the uterus and reproductive organs of male and female equids. Neither pregnancy, nor foaling increased the odds of finding EHV-1 in vaginal samples, even though virus can be collected from vaginal secretions after an EHV-1-induced abortion.

As with the nose, vaginal shedding in this study was not affected by vaccination, which indicates that additional biosecurity measures, such as antiviral sanitation during breeding examinations, may be required to limit spread between mares. As the amount of viral DNA retrieved was very low, additional studies would be required to determine the biological and clinical relevance of this finding. However, Lee and Lee (26) also identified EHV-2 and EHV-5 in genital swabs from healthy mares in South Korea. More recently, EHVs have also been reported in the feces of wild zebras (30). Exposure of foals to EHV-1 may therefore not only come from the nose but also from the mare’s vagina and contamination of the tail region, which indicates the need for further research into this possible mode of transmission.

Consistent with other studies, vaccination appeared to lower the odds of virus presence in blood when horses were vaccinated, especially when vaccinated within 2 mo of sample collection (31). Vaccine studies typically test for viral presence and seroconversion within days or weeks after vaccination as evidence of protection (6,14,32). In the spring prior to the study, the Innovator vaccine (Zoetis) was most recently used by only 3 farms. This limits the significance of the increased odds of a positive WBC test compared to the other categories of vaccine products, as none of those horses was recently vaccinated. Currently available vaccine strains of EHV-1 appear to stimulate an IgG1 and IgG4/IgG7 humoral response (31,33,34), which may reduce viral shedding at the nose, in addition to lowering levels of viremia.

Vaccines that are labeled to aid in the prevention of abortion associated with EHV-1 tend to have higher antigen loads than vaccines containing multiple antigens (35). To take advantage of this protective effect, the American Association of Equine Practitioners (AAEP) strongly recommends that mares be vaccinated at 5, 7, and 9 mo of gestation, as well as using a combination vaccine that includes EHV-1 and EHV-4 antigens 30 d prior to foaling (36). As would be expected with breeders following vaccination recommendations, associations were identified between month of gestation, vaccination within 2 mo of sample collection, and the month of sampling for WBC samples in the present study.

A sparing effect in WBC samples was identified in this study as the month of gestation increased, which suggests a protective host response to viremia in these healthy broodmares. In vitro, susceptibility to EHV-1 infection is higher in cells with the MHCI-B2 genotype (37). Kydd et al (37) reported that mares carrying the B2 allele may be associated with increased rates of pregnancy loss in Thoroughbreds but could not confirm increased susceptibility to EHV-1. Similarly, Dunuwille et al (38) could not identify any genetic association to cases with equine herpesvirus myeloencephalopathy, which indicates that there is evidently more work to be done in the host-factor realm.

Interestingly, different immune responses have also been associated with the different equine leukocyte antigen (ELA) haplotypes, in which the ELA-A10 haplotype produced lower titers (13). Further investigation into the response of vaccination in the host, IgG response, and virus shedding is recommended, especially in healthy versus clinically affected horses. Because the limit of detection with ddPCR is very low, the clinical significance of finding EHV-1 in this study is unclear, which is a major limitation of the study. There was little correlation within a horse with repeated sampling, suggesting that the presence of very low levels of EHV-1, as identified by ddPCR, may occur randomly within a horse or be inconsistently detected. Future work should compare the results of ddPCR detection with other PCR detection methods to determine the clinical applicability of this new assay.

As with any PCR investigation, we are unable to differentiate virus DNA from infective virus. Based on these results, ddPCR would be of value in an experimental setting to compare shedding sites and perhaps better identify latency. Further investigation would be required in order to determine if there is infectious virus shedding from the vaginal mucosa, especially during an EHV-1 abortion.

There is also a need for a comparative investigation of DNA variation of herpesviruses implicated in the clinically affected population and those present endemically. Some viral strains have been more effective than others at producing higher amounts of shedding in experimental studies (39). As only the EHV-1 gB gene was investigated in this study, we could not differentiate between the wildtype N752 or neuropathogenic D752 strains or the possibility of a modified-live vaccine (MLV) strain (40). Furthermore, although the primer used to identify the gB gene could, in theory, amplify EHV-8, which was previously called asinine herpesvirus-3 (41), no samples containing EHV-8 have been reported in Canada to the best of the authors’ knowledge.

Although this population was geographically and characteristically diverse, which makes the study highly representative of horses in Ontario, this could have limited the study’s power due to age and breed categorization. Some mares were lost to follow-up due to sale, movement, or death; however, over 80% of the mares that enrolled in this study completed at least 4 of 6 sampling periods. Although all mares in this study remained healthy during the sampling period, abortions and respiratory disease have occurred sporadically since.

In summary, this study identified a period prevalence for EHV-1 of 85.0% in a subset of the healthy Ontario broodmare population. Virus DNA was identified not only in the nose and peripheral blood, but also in vaginal secretions, which has not been previously reported. Because detectable levels of virus were very low, ddPCR may be a useful tool for prevalence investigations, although the clinical relevance of these low shedding levels is unknown. Known stressors, such as pregnancy, did increase virus presence in nasal samples, but foaling did not.

This study was also in agreement with the findings of previous studies that virus levels in peripheral blood were lower after recent vaccination (< 2 mo) (31). As viral shedding from the vagina was not reduced after vaccination, it is recommended that the vagina be further investigated as a source of infection or environmental contamination. Lastly, the odds of viral DNA being present appeared to decrease as gestation progressed in this group of healthy broodmares, which suggests that there may be a host factor that is overcome during abortive-EHV-1 infection despite vaccination.

Funding Statement

The authors thank Bailey Fuller for her assistance during the analysis of these data. This research was funded by Equine Guelph and Zoetis Inc. Graduate student stipend support was provided to Dr. Carina Cooper by the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA) and the Ontario Veterinary College (OVC), which also provided stipend support for summer students involved in the project