Materials and Methods

Sample population—Sera obtained from horses in Kentucky were used in 2 experiments. Experiment 1 was a retrospective serologic evaluation of samples obtained in 2001 and 2002, stored frozen at the UKLDDC, and shipped to the Laboratory for Calicivirus Studies at Oregon State University for analysis.

Included in the serum samples collected in 2001 from 112 Thoroughbreds were 58 sera from mares that aborted or mares on farms with a history of abortion problems (group 1); 25 sera from randomly selected, 3- to 13-year-old healthy mares sold at auction (group 2; breeding-age control mares); and 29 serum samples randomly obtained from mixed-age males and yearling females subsequently sold at auction (group 3). Group 3 was a negative control group for abortion. Because of the age or sex of the horses in group 3, they could not have aborted but could have been exposed to vesivirus.

Experiment 2 comprised serum samples obtained before and after ETC exposure from each of 29 pregnant non-Thoroughbred broodmares. The experiment was conducted in 2002 at the UKLDDC to determine whether MRLS could be experimentally linked to ETCs.²⁷ The 29 pregnant mares were assigned to 3 groups to mimic on-farm conditions for exposure of pregnant mares to ETCs. Group 1 (10 mares) was allowed to graze on pastures where they were exposed for two 10-day periods to dense populations of ETCs and frass as the ETCs migrated through the pasture-grass forage. Group 2 (9 mares) was managed in a similar manner, but these mares were exposed to frass only. Group 3 (10 mares) was intended to represent nonexposed control mares; however, some ETCs were able to circumvent the caterpillarguard fencing, and these mares were inadvertently exposed to migrating ETCs and frass (although at lower amounts of exposure than for groups 1 and 2).

Procedure—To ensure that investigators were not aware of the identity of any mares in both experiments 1 and 2, the UKLDDC accession numbers and all information except species were removed from the serum samples and a unique identification code was assigned to each sample. Samples were then shipped to the Laboratory for Calicivirus Studies for testing. After testing, the code was translated and an analytic database constructed by use of a computer spreadsheet.^b Coded information included date of sample collection, sex, age, and sample type (ie, acute, convalescent, or randomly obtained serum).

For experiment 2, at least 3 serum samples were obtained from each of the 29 pregnant mares. The first sample was obtained before ETC exposure, and at least 2 samples were obtained after exposure at approximately 2-week intervals.

Separate shipments of ETCs and tissues collected from 20 aborted fetuses were stored frozen and sent to the Laboratory for Calcivirus Studies. Fetal tissues were badly autolyzed at the time of arrival because of high ambient temperatures during shipment. Although the samples were not suitable for critical analyses, virus isolation, direct electron microscopy, fluorescent antibody examination, and reverse transcriptase–PCR assays were attempted on autolyzed fetal tissues and frozen ETCs, as described elsewhere.^15,17,a

Antibody testing—A validated indirect ELISA²⁸ with a recombinant vesivirus antigen (D3A) cloned from the capsid protein gene of SMSV-5 was used for detection of antibodies. When experimentally inoculated into horses, the parent virus (SMSV-5) can cause disease.^13,26

The D3A antigen has been further characterized. The SMSV-5 also infects > 20 species as phylogenetically diverse as shellfish, teleosts, pinnipeds, swine, and humans.^16,24,25 The D3A antigen^28,a has been tested against vesiviruses isolated from 10 host species, yielding 14 serotypes of vesiviral typing sera (all validated by the Plum Island Animal Disease Laboratory), including that for SMSV-5. The D3A peptide is patented (US patent No. 6,593,080 B1) and is a vesivirusspecific group antigen that cross-reacts against multiple vesivirus neutralizing–antibody serotypes.²⁸ It has been purified into a 293-amino acid peptide (termed caliciscreen)¹⁷ and adjusted as a stock antigen to achieve a concentration of 1.2 mg/mL, which has then been used in an indirect ELISA for detection of vesiviral antibodies.^28,a

The OD values were determined at 620 nm by use of a plate reader.^c Net OD values were calculated as test sample OD against D3A antigen minus the OD of control wells treated the same in all aspects except for the addition of test serum. A cutoff value of 0.14 for net OD for the ELISA was established after calculating the mean OD (ie, 0.065) plus 2 times the SD (ie, 2 × 0.034) of the net OD readings for the 29 sera obtained from the healthy mixed-age males and yearling females sold at auction (ie, group 3 of experiment 1). This approach established a calculated specificity of approximately 97.5%.²⁹

Samples were considered to have positive results when the net OD value exceeded the cutoff value of 0.14. A single, randomly selected, OD-negative serum sample was used as a negative control sample for each ELISA plate (n = 14), and the mean of the net OD for that sample among plates was 0.055 (range, 0.041 to 0.065; SD, 0.007; mean SE, 13%). A positive control sample was used for each plate. It consisted of a single, OD-positive serum sample randomly selected from the pool of mares that yielded positive results. The sample selected yielded a mean net OD for the 14 plates of 0.35 (range, 0.274 to 0.425; SD, 0.038; mean SE, 11%).

Statistical analysis—Data were analyzed by use of a statistical program.^d The distribution of sera with positive results and the association between seroprevalence and abortion in horses were evaluated for all sample sets by use of the Fisher exact test, Yates corrected χ² test, or Mantel χ² test, as appropriate. A Fisher exact test was used when expected values were < 5. For each of these comparisons, values of P < 0.05 were considered significant. An OR within the lower limit of the 95% CI (CI > 1) was interpreted as a significant association.

Results

Experiment 1—Results of the retrospective serologic evaluation were summarized (Table 1). Adult mares that had aborted or came from farms with mares that had recently aborted were identified as the group at highest risk for becoming antibody-positive. A significant (P = 0.039) difference for antibodies against vesivirus was observed between groups 1 and 2 (OR, 2.64; 95% CI, 1.01 to 6.92). Thus, mares in group 1 (ie, mares associated with abortions) had a higher probability (2.64 times higher) of being seropositive for vesivirus, compared with mares in group 2 (ie, breeding-age control mares). Similarly, significant (P < 0.001) differences were observed between the seronegative control group (group 3; males and yearling females sold at auction) and the breeding-age control mares (OR, 18.67; 95% CI, 2.2 to 160.0) and between the male and yearling females sold at auction and the mares associated with abortion (OR, 49.3; 95% CI, 6.3 to 389.0).

Experiment 2—Seven of 10 pregnant mares exposed to ETCs and frass and 7 of 9 pregnant mares exposed to frass only aborted. Three of 10 mares in the control group aborted; however, as mentioned previously, mares in the control group were accidentally exposed to a low number of migrating ETC larvae that had circumvented the caterpillar-guard fencing that was intended to keep them away from the mares of the control group.

Results for vesivirus-specific antibodies in serum samples obtained from the 29 pregnant mares of experiment 2 were summarized (Table 2). Of the 29 samples obtained before exposure, 18 (62.1%) had positive test results for antibodies against vesivirus. The number of samples with positive results increased to 25 of 29 (86.2%) for the second sample obtained after exposure. Eight of 10 (80%) mares in the control group were initially seropositive, and these mares remained seropositive through the second sample after exposure. The number of seropositive mares in the group exposed to ETCs and frass increased from 5 (50%) before exposure to 9 (90%) for the second sample after exposure. Similarly, the number of seropositive mares among the 9 mares exposed only to frass increased from 5 (55.6%) before exposure to 8 (88.9%) for the second sample after exposure.

Of 17 mares in groups 1 and 2 that aborted, 8 (47.1%) were seropositive for vesivirus antibodies before exposure but 7 additional mares seroconverted (15/17 [88.2%]) by the time the second sample after exposure was obtained. For group 1, 4 of 7 (57.1%) mares that aborted were seropositive before exposure, which increased to 7 of 7 (100%) seropositive mares for the second sample after exposure. Of the 7 mares in group 2 that aborted, 3 (42.9%) were seropositive before exposure to frass; however, this increased to 6 of 7 (85.7%) that were seropositive for the second sample after exposure. Of 3 mares that aborted in the control group, 1 was seropositive before the inadvertent exposure to ETC and the remaining 2 subsequently seroconverted.

The proportion of mares that were seropositive increased significantly (P = 0.03) from before exposure (18/29 [62.1%]) to the time the second sample was obtained after exposure (25/29 [86.2%]), with an OR of 3.8. For mares that aborted, there was a significant (P = 0.013; OR, 8.4) increase in the proportion that was seropositive before exposure (8/17 [47.1%]) and the proportion that was seropositive at the time the second sample after exposure was obtained (15/17 [88.2%]).

Five mares in the study had a 3-fold or greater increase in the serum net OD value, which suggested a primary or secondary immune response to vesivirus (Figure 1). Three of these 5 mares aborted during the experiment. Although those 3 mares were initially seronegative, they seroconverted to vesivirus during the experiment. The other 2 mares with a > 3-fold increase in net OD value did not abort during the experiment; however, they were seropositive before exposure.

View Full Size

Optical density values of serum samples obtained from 5 mares before exposure to ETCs and frass or frass alone (week 0) or at approximately 2-week intervals after exposure. Each symbol represents results for 1 mare. Results of an indirect ELISA revealed a 3-fold or greater increase in net OD values during the experimental period. An OD value of 0.14 was designated as the cutoff value for positive results (dotted horizontal line). Notice that 3 of the mares were initially seronegative, but they seroconverted during the study; all 3 of these mares aborted. In contrast, 2 mares were initially seropositive, and neither of these mares aborted during the experiment.

Citation: American Journal of Veterinary Research 67, 6; 10.2460/ajvr.67.6.1033

• Download Figure
• Download figure as PowerPoint slide

Examination of the aborted fetal tissues and ETC samples yielded negative results for testing conducted by use of in vitro virus isolation, direct electron microscopy, and reverse transcriptase–PCR assay. However, there was a weak positive signal in tissues from 12 of 20 fetuses during fluorescent antibody testing.^a

Discussion

The data reported here linking abortion in horses and seroconversion to vesivirus were statistically robust for both the retrospective serologic evaluation and the ETC experiment, although testing of fetal tissues by use of in vitro virus isolation, direct electron microscopy, and PCR assay to provide additional evidence of vesivirus infection yielded negative results. Tissues from 20 aborted fetuses that had been stored frozen and shipped for examination by use of the aforementioned procedures were badly autolyzed because they thawed during shipment. As would be expected, virus isolation and electron microscopy efforts were unrewarding, and PCR assay also failed, perhaps in part because universal PCR primer sets for vesivirus were unavailable. However, 12 of 20 samples did yield weakly positive test results (a faded, diffuse fluorescent signal^a) when tested by use of a fluorescent-tagged monoclonal antibody²³ that is reactive against > 30 serotypes of vesivirus. This reactivity could have been nonspecific background fluorescence associated with poor sample quality. However, 8 of the 20 similarly handled samples yielded negative results, whereas those 12 yielded weakly positive results. This argues against the possibility that fluorescence observed in the 12 samples was the result of false-positive reactions.^a

Samples of ETCs and frass were also examined for vesivirus by use of the same aforementioned tests used for fetal tissues; all results for ETCs and frass were negative. However, examination of fetal tissues and ETCs should be repeated on new sample sets obtained and tested under optimized conditions.

Although type-specific neutralizing antibodies against vesivirus have been reported³⁰ in free-ranging feral donkeys, and horses have been experimentally infected with vesivirus,^21,26 evidence of natural exposure and spread of vesivirus among horses has not been described. Seroprevalence for antibodies against vesivirus was 40% for 25 mares of reproductive age sold at auction in experiment 1, but it was 62.1% for 29 healthy pregnant broodmares used for experiment 2. Seroprevalence in these populations seems extremely high, compared with seroprevalence for the 29 mixedage males and yearling females, all of which were seronegative. These differences could be related to age, which is obviously true for many disease agents. Increases in age (eg, time) result in an increase in the probability of exposure.

However, 3 observations of data independent of age provided evidence suggesting an association between seroconversion to vesivirus and abortion. First, there was a 3-fold or greater increase in vesiviral antibody titer detected in some mares, as determined by evaluation of paired serum samples obtained before and after abortion. Second, significant differences in prevalence of antibodies against vesivirus were detected between breeding-age mares associated with farms that had a history of abortion problems and breedingage mares sold at auction that were not associated with a reported history of abortion. Finally, in experiment 2, 15 of 17 (88.2%) mares that aborted seroconverted, whereas only 2 of 12 (16.7%) mares that did not abort seroconverted. These serologic data are indicative of vesiviral infection and when compared with data of probable ETC-induced abortion, 17 of 29 (58.6%) mares exposed to ETCs and ETC frass in experiment 2 aborted, whereas 12 of 29 (41.4%) mares did not abort.

For experiment 1, the 54 negative control sera obtained from a nonselective population of Thoroughbreds in Kentucky had a common criterion that the horses were sold through auction at sales barns. Thus, those horses received physical examinations and did not have evidence of active infectious disease. These horses were then allocated into 2 subgroups (29 mixed-age males and yearling females, none of which could have aborted, and 25 breedingage mares [3 to 13 years old] whose reproductive histories were unknown).

Reexamination of the ELISA results by use of a standard^15,25 cutoff value of 0.10 for the OD results, rather than the cutoff value of 0.14 derived from the 29 horses incapable of aborting and whose ELISA OD values were surprisingly all < 0.14, yielded greater sensitivity and less specificity. By use of a cutoff value of 0.10, 64 of 112 (57.1%) horses in the population were seropositive, 17 of 54 (31.5%) horses sold at auctions were seropositive, 4 of 29 (13.8%) mixed-age males and yearling females were seropositive, 13 of 25 (52.0%) 3to 13-year-old mares were seropositive, and 47 of 58 (81.0%) abortion-associated mares were seropositive. These data were statistically more robust than those reported here and clearly illustrate that raising and lowering the cutoff value for the OD results would not alter the overall outcome of our reported findings.

Not all mares in experiment 1 from farms with a history of abortion problems actually aborted; therefore, a seroprevalence of 63.8% for this group may well be an underestimate. Conversely, it is unknown whether any pregnant broodmares used for the ETC exposure in experiment 2 or the 3- to 13-year-old mares sold at auction in experiment 1 had aborted. Therefore, the percentages of seropositive mares from these 2 groups (62.1% and 40.0%, respectively) could be an overestimate.

As stated previously, the 29 pregnant broodmares (non-Thoroughbred mares used for nursing of Thoroughbred foals) used in experiment 2 for the ETC exposure did not have a history of abortions but did have a high seroprevalence (62.1%) for antibodies against vesivirus. Increases in the number of sera with positive results for this study were detected in mares that aborted, primarily mares fed ETCs and frass or frass alone. Ten mares were purposely fed ETCs and frass. Of these, 7 aborted during the experiment and all 7 were seropositive, whereas the remaining 3 mares that did not abort did not become seropositive during the experiment. An increase of > 40% for seroconversion to vesivirus among mares that aborted is evidence of an association between being seropositive for vesivirus and abortion.

Little is known about the immune response in animals after a primary or secondary exposure to vesivirus,²¹ although limited data have been published on experimental infections in calves²³ and primates.³¹ Varied primary responses have been experimentally induced for various vesivirus serotypes in nonequine species, ranging from no detectable neutralizing antibodies to a rapid response resulting in high antibody titers.^{16,23,31–33} Therefore, the > 3-fold increase in the serum antibody response in 5 mares during experiment 2 (3 of which aborted during the experiment), should be examined in more detail. The 3 mares that aborted had a similar pattern of antibody increase, despite being seronegative for antibodies against vesivirus before the start of the experiment. This suggests that they became infected during the experiment with vesivirus strains that had antigenic profiles reactive to the D3A group antigen. The other 2 mares that did not abort during the experiment were seropositive before ETC exposure and had a sharp increase in antibody titer. This increase in antibody titer suggests a secondary immune response following reexposure to vesiviral antigen.

Despite the lack of a 3-fold or greater increase in immune response, the remaining 4 mares that aborted had an increase in antibodies against vesivirus and could have had a primary infection with vesivirus. Although only 3 serum samples (up to 33 days after exposure) were available from these 4 mares, they all had the same pattern for antibody titers (1.5- to 2.5-fold increase). Had another serum sample been obtained subsequently from these 4 mares, it is postulated that an even greater increase in antibody titer may have been seen. Although antibody titers for some mares increased between day 33 and 50, neither of 2 mares that were seronegative before the start of the experiment and did not abort during the experiment had a similar pattern of antibody change.

The original intent of the ETC exposure experiment was to establish an association between ETCs and MRLS^9,27 and explore a possible linkage between calicivirus-infected ETCs and abortion in mares, which would fit the ecologic profile for calicivirus-infected naval orange worms.³⁴ However, no association could be detected between seroconversion to vesivirus and ETC exposure, which suggests that vesivirus and ETCs may be independently associated with abortion in mares and that ETCs may not be a primary vector of transmission for abortogenic variants of vesivirus belonging to the family Caliciviridae.

The high seroprevalence for antibodies against vesivirus in mares from farms with a history of abortions, which differed significantly (P = 0.039) when compared with the seroprevalence for a control group of 3- to 13-year-old mares sold at auction in experiment 1, and the correlation between seroconversion to vesivirus and experimentally induced abortions in mares in experiment 2 provides corroborating evidence of vesivirus as an etiologic agent associated with abortion in horses. Lack of seropositive samples for the negative control group suggests that males and young females may be less immunologically responsive or had less exposure to vesivirus. In contrast, detectable immunoreactive exposure to vesivirus was found for approximately half of all breeding-age mares tested.

Because seroprevalence was ≥ 62% in a population of broodmares with unknown reproductive histories, it seems likely that horses become infected through routes of exposure accounted for by differing husbandry or herd management practices or as a result of differences in the response to vesiviral exposure when comparing pregnant mares or broodmares to other populations of horses. Ingestion is an established route of infection for vesivirus (eg, contaminated sources of dietary protein, such as fish, fish meal, fish lysates, carcass wastes, or other animal products). Similarly, direct and indirect (fomites) contact may result in ingestion of fecal matter, urine, nasal or oral secretions, or fetal fluids or residues.^13,16,35 A likely point source of infection would be animals with acute or chronic infections with persistent shedding of virus. This concept of vesiviral persistence is also supported by descriptions of in vitro infections with vesivirus in which high viral titers were sometimes attained without cytopathic effect²⁴ and for in vivo studies¹⁶ of infections in fish, shellfish, primates, and cattle in which signs of clinical disease were sometimes ambiguous. Transmission via direct contact has been reported repeatedly in swine, cattle, and reptiles during experimental infectivity studies with vesivirus^{16,23,31–33} and could represent the most common route of transmission in domestic animals.

Analysis of the serologic data in horses reported here suggests that broodmares are commonly exposed to vesivirus from unknown sources and that there is a significant seroconversion rate associated with mares that abort in Kentucky. Whether exposure to horses (most notably broodmares that abort) is an increased public health risk is not known, but because some members of vesiviruses are zoonotic^15,16,25 and there is evidence of their association with abortions in humans,¹⁶ this is a reasonable concern, especially for pregnant women. Analysis of our data suggests that screening for vesiviruses should be added to the panel of diagnostic tests for horses that abort and that additional in-depth investigations of the role of vesivirus in horses be conducted.