• Open Access

Accurate Antemortem Diagnosis of Equine Protozoal Myeloencephalitis (EPM) Based on Detecting Intrathecal Antibodies against Sarcocystis neurona Using the SnSAG2 and SnSAG4/3 ELISAs


  • Results were presented in part at the 2010 American College of Veterinary Internal Medicine Forum, Anaheim, CA

Corresponding author: S.M. Reed, Rood and Riddle Equine Hospital, Lexington, KY 40511-9072; e-mail: sreed@roodandriddle.com.



Recent work demonstrated the value of antigen-specific antibody indices (AI and C-value) to detect intrathecal antibody production against Sarcocystis neurona for antemortem diagnosis of equine protozoal myeloencephalitis (EPM).


The study was conducted to assess whether the antigen-specific antibody indices can be reduced to a simple serum : cerebrospinal fluid (CSF) titer ratio to achieve accurate EPM diagnosis.


Paired serum and CSF samples from 128 horses diagnosed by postmortem examination. The sample set included 44 EPM cases, 35 cervical-vertebral malformation (CVM) cases, 39 neurologic cases other than EPM or CVM, and 10 non-neurologic cases.


Antibodies against S. neurona were measured in serum and CSF pairs using the SnSAG2 and SnSAG4/3 (SnSAG2, 4/3) ELISAs, and the ratio of each respective serum titer to CSF titer was determined. Likelihood ratios and diagnostic sensitivity and specificity were calculated based on serum titers, CSF titers, and serum : CSF titer ratios.


Excellent diagnostic sensitivity and specificity was obtained from the SnSAG2, 4/3 serum : CSF titer ratio. Sensitivity and specificity of 93.2 and 81.1%, respectively, were achieved using a ratio cutoff of ≤100, whereas sensitivity and specificity were 86.4 and 95.9%, respectively, if a more rigorous cutoff of ≤50 was used. Antibody titers in CSF also provided good diagnostic accuracy. Serum antibody titers alone yielded much lower sensitivity and specificity.

Conclusions and Clinical Importance

The study confirms the value of detecting intrathecal antibody production for antemortem diagnosis of EPM, and they further show that the antigen-specific antibody indices can be reduced in practice to a simple serum : CSF titer ratio.


antibody index


blood–brain barrier


central nervous system


cerebrospinal fluid


cervical veterbral malformation


enzyme-linked immunosorbent assay


equine protozoal myeloencephalitis


indirect fluorescent antibody test


percent positivity

Equine protozoal myeloencephalitis (EPM) is one of the most important infectious neurological diseases in horses in the Americas (reviewed extensively in Refs. [1, 2]). The clinical signs of EPM vary and can include asymmetrical weakness, muscle atrophy, and ataxia, but none can be considered pathognomonic. The disease is most commonly caused by the protozoan parasite Sarcocystis neurona, which is transmitted in feces from infected opossums,[3, 4] although sporadic cases of EPM caused by the related parasite Neospora hughesi have been observed.[5, 6] EPM occurs when parasites invade and propagate in the central nervous system (CNS), where they may infect cells randomly within the white and gray matter of the brain or spinal cord of the infected horse. However, it is important to note that parasite proliferation in the CNS and associated immunopathology does not occur in every horse infected with S. neurona. Indeed, studies in various regions of the United States have shown seroprevalence rates against S. neurona ranging from approximately 33–89%,[7-14] but the annual incidence of clinical EPM has been estimated at less than 1%.[15] Therefore, it is apparent that S. neurona infection does not always progress to neurologic disease. Factors influencing the occurrence of EPM are not well understood, but stress, concurrent disease, and general anesthesia have been implicated.[1, 2]

Antemortem diagnosis of neurological diseases in horses can be challenging. For EPM, detection of antibodies against S. neurona commonly has been used as an adjunct to aid diagnosis, and several ancillary serologic tests have been developed for EPM diagnosis. These include Western blot,[16, 17] a direct agglutination test,[18] an indirect fluorescent antibody test (IFAT),[19] and enzyme-linked immunosorbent assays (ELISAs) based on single[20] or multiple[21, 22] S. neurona surface antigens (SnSAGs). Varying levels of sensitivity and specificity for detection of anti-S. neurona antibodies have been achieved with these assays, but their utility for EPM diagnosis has been limited primarily by the incongruity between S. neurona infection and the occurrence of clinical disease. Specifically, the presence of serum antibodies against S. neurona is an indicator of infection, but not necessarily disease. Detection of S. neurona-specific antibodies in cerebrospinal fluid (CSF) was logically proposed as an approach to confirm CNS infection and EPM. However, passive diffusion of proteins occurs normally across the blood–brain barrier (BBB), and serum antibodies are found at proportional concentrations in the CSF.[23, 24] Therefore, a horse that has even modest serum antibody titers against S. neurona may have detectable concentrations of antibodies in the CSF, thereby resulting in a false positive result.

Application of either the Goldmann–Witmer coefficient (C-value) or the antigen-specific antibody index (AI) to assess antibody titers obtained with an ELISA based on recombinant SnSAG2[21] was shown recently to allow accurate discrimination of horses with EPM.[25] These 2 CSF indices are tests of proportionality that determine if the amount of antigen-specific antibody in the CSF is greater than would be expected from normal transfer across the BBB, which is indicative of intrathecal antibody production because of infection in the CNS. To expand on this previous work, we have used the SnSAG2 ELISA and a new ELISA based on the SnSAG4/3 chimeric fusion protein[22] to examine paired serum/CSF from an extensive dataset of 128 neurologically diseased and normal horses for which a diagnosis was confirmed by necropsy. Based on the supposition that the concentration of specific proteins, including antibodies, in the serum relative to the CSF typically will fall within a standard range approximating 130 : 1–250 : 1 in clinically normal horses,[25, 26] we investigated whether serum : CSF antibody titer ratios were appreciably lower than normal in horses diagnosed with EPM. In the present study, we show that use of the SnSAG2 and SnSAG4/3 serum : CSF titer ratios alone was sufficient to detect intrathecal antibody production and represents an accurate and reliable approach to diagnose EPM by immunologic testing.

Materials and Methods

Sample Set

A total of 128 horses were included in the study. All had a postmortem diagnosis and paired serum and CSF available for testing. The cases originated from Rood and Riddle Equine Hospital, Lexington, Kentucky (n = 43), New Bolton Center, University of Pennsylvania School of Veterinary Medicine (n = 37), M.H. Gluck Equine Research Center, University of Kentucky (n = 26), Marion DuPont Scott Equine Medical Center, Virginia-Maryland Regional College of Veterinary Medicine (n = 12), University of Florida College of Veterinary Medicine (n = 7), and referral cases from private practice (n = 3). The 37 cases from New Bolton Center were included in a companion study comparing the SnSAG2 and SnSAG4/3 ratios with the IFAT for EPM diagnosis.[31] Diagnostic categories for this study were “EPM” (n = 44), cervical vertebral malformation (“CVM”; n = 35), “Other Neurologic” (non-EPM and non-CVM; n = 39), and “Non-neurologic” (n = 10). All necropsies were performed, reviewed, or both by board-certified (ACVP) veterinary pathologists. Confirmed positive cases in the “EPM” category had neurologic deficits as determined by attending clinicians and postmortem lesions consistent with EPM as determined by veterinary pathologists. Criteria for a postmortem diagnosis of EPM included the presence of multifocal, asymmetric myelitis with or without encephalitis consisting of infiltrates of lymphocytes, macrophages, and occasional eosinophils forming wide cuffs of cells around blood vessels. There was extension of the inflammation into the adjacent neuropil with associated tissue degeneration. These cases had no additional histopathologic changes, allowing exclusion of other neurologic conditions. Immunohistochemistry or PCR testing was used to demonstrate the presence of S. neurona at the discretion of the pathologist, but was not performed in every case. Horses were grouped into the “CVM” category based on neurologic examination, myelogram showing compression, and postmortem examination demonstrating pathologic changes indicative of CVM. The “Other Neurologic” category included non-EPM, non-CVM cases that included a variety of diagnoses, including equine herpesvirus myeloencephalopathy, tumors, vertebral or skull fractures, or epilepsy. Lesions consistent with EPM were not noted in the pathology reports for all CVM and “Other Neurologic” cases. The “Non-neurologic” cases were neurologically normal horses that had been euthanized and necropsied for reasons such as laminitis or colic.

SnSAG2 and SnSAG4/3 ELISAs

Recombinant SnSAG2 and SnSAG4/3 were prepared and ELISAs were conducted as described previously.[21, 22] All test samples and standards (positive and negative sera) were run in duplicate wells, and the average OD450 was determined for each sample. To account for interplate variation, the average OD450 of each test sample was expressed as a percent positivity (PP) value.[27] Samples were diluted serially starting at 1 : 250 for serum and 1 : 2.5 for CSF. Endpoint titers were determined for each serum and CSF sample pair using a PP cutoff of 10%, and serum : CSF titer ratios were calculated from the reciprocal titers obtained from each ELISA. If serum antibodies were not detected at the lowest dilution, the sample was considered negative and the ratio was calculated using <250 as the numerator. For example, a ratio of <50 would be recorded for a case with a nondetectable serum titer, but a CSF titer of 1 : 5 (<250/5). If there were no detectable CSF antibodies, a negative result was recorded for both the sample and for the ratio because this indicated no intrathecal antibody production. To circumvent the known antigenic diversity of S. neurona and the likely variation in antibody responses elicited in horses, the SnSAG2 and SnSAG4/3 ELISAs were run and analyzed independently, as recommended previously.[22] Although each serum and CSF sample pair was tested with both ELISAs, the results from the 2 assays (herein designated SnSAG2, 4/3) were considered collectively. That is, the higher endpoint titer for each sample and the lower serum : CSF ratio for each sample pair were considered the relevant results and used for the final interpretation. For each case, the relevant titers and ratio were not always obtained with the same ELISA. Likelihood ratios were calculated for 3 separate outcome variables: SnSAG2, 4/3 serum titer, SnSAG2, 4/3 CSF titer, and SnSAG2, 4/3 titer ratio. Sensitivity and specificity at various cutoff values were calculated from 2 × 2 contingency tables, and likelihood ratios for a positive test were determined by the standard equation LR+ = sensitivity/(1 − specificity).


To establish the optimal accuracy of EPM diagnosis by serologic testing, paired serum and CSF from 128 horses diagnosed postmortem were analyzed with the SnSAG2, 4/3 ELISAs. When serum alone was considered, antibody titers up to 1 : 4,000 were found in cases from all diagnostic categories (Fig 1). Although EPM was the most common diagnosis for cases exhibiting endpoint antibody titers from 1 : 1,000 to 1 : 4,000, the total number of cases in the 3 non-EPM categories (30) outnumbered EPM cases (26) at these titers. Eight cases were found to have endpoint titers >1 : 4,000; five were diagnosed as EPM cases and three were diagnosed as “Other Neurologic” cases. The highest antibody titer in serum, 1 : 16,000, was detected in 2 EPM cases and 1 “Other Neurologic” case. Serum antibodies were not detected in 3 EPM cases, although antibodies were detected in CSF from three of these cases (see Table S1). Likelihood ratios calculated using serum titers as the outcome indicated that EPM was only about 2.5 times more likely in a horse exhibiting a very high serum titer of ≥1 : 4,000 and about 1.5 times more likely in a horse that had a serum titer of 1 : 1,000 or 1 : 2,000 (Table 1).

Table 1. Likelihood ratios for postmortem cases using SnSAG2, 4/3 ELISA serum titers
Reciprocal TiterEPM+EPM−True+False+Likelihood Ratio
Figure 1.

Distribution of cases based on the SnSAG2, 4/3 antibody titers in serum. The SnSAG2 and SnSAG4/3 ELISAs were run independently, with the higher end-point titer recorded for each sample. Cases from all diagnostic categories were distributed across serum dilutions up to 1 : 4,000, and both EPM and non-EPM cases were observed at the highest antibody titer detected (1 : 16,000). Samples were considered negative if antibodies were not detected at the 1 : 250 dilution.

The SnSAG2, 4/3 endpoint antibody titers in CSF samples from the 128 postmortem cases ranged from negative to 1 : 2,560. Antibody titers up to 1 : 20 in CSF were found for cases from all diagnostic categories (Fig 2). However, titers >1 : 20 were observed primarily in EPM cases (n = 34), with only 3 “Other Neurologic” cases exhibiting CSF titers >1 : 20 (2 = 1 : 40; 1 = 1 : 80). Antibody titers ≥1 : 160 were observed in CSF from 21 horses, all of which had a postmortem diagnosis of EPM. Two EPM cases did not have detectable antibody in CSF. The highest CSF endpoint titer detected by the ELISAs (1 : 2,560) was observed in 2 EPM cases. As expected, the majority of antibody-negative CSF samples were non-EPM cases (n = 40). Likelihood ratios calculated using CSF titers as the outcome indicated that the likelihood of EPM was over 21 times higher in a horse exhibiting a CSF titer of ≥1 : 40 (Table 2).

Table 2. Likelihood ratios for postmortem cases using SnSAG2, 4/3 ELISA cerebrospinal fluid titers
Reciprocal TiterEPM+EPM−True+False+Likelihood Ratio
Figure 2.

Distribution of cases based on the SnSAG2, 4/3 antibody titers in cerebrospinal fluid (CSF). The SnSAG2 and SnSAG4/3 ELISAs were run independently, with the higher end-point titer recorded for each sample. Cases from all diagnostic categories were distributed across CSF dilutions up to 1 : 20, but EPM cases were predominate at CSF antibody titers of 1 : 40 and above. Samples were considered negative if antibodies were not detected at the 1 : 2.5 dilution.

Based on our previous work showing that accurate serodiagnosis of EPM could be achieved by using CSF indices to detect intrathecal antibody production against S. neurona,[25] we examined whether a simple ratio of serum titer to CSF titer would provide a similarly sensitive and specific diagnosis of an EPM case. As shown in Figure 3, the SnSAG2, 4/3 serum : CSF titer ratios exhibited pronounced demarcation between EPM cases and all other non-EPM diagnostic categories. Of the 44 EPM cases, 38 had a titer ratio that was <100. Two EPM cases had CSF titers that exceeded serum titers, thus resulting in a titer ratio <1. As mentioned previously, there were 4 EPM cases that were negative for serum antibodies by both ELISAs, but CSF antibodies were detected (see Table S1). In contrast to the SnSAG2, 4/3 titer ratios observed for the EPM cases, 81 of 84 non-EPM cases had no detectable CSF antibody (designated negative) or had a titer ratio that was ≥100 (Fig 3). Of the 3 non-EPM cases that yielded a titer ratio <100, 2 had a ratio of 50 and 1 had a ratio of 25 (see Table S1). Likelihood ratios calculated using the SnSAG2, 4/3 titer ratio as the outcome indicated that a horse was approximately 24 times more likely to have EPM if the ratio was <100 (Table 3).

Table 3. Likelihood ratios for postmortem cases using combined SnSAG2, 4/3 ELISA serum : cerebrospinal fluid antibody ratios
Titer RatioEPM+EPM−True+False+Likelihood Ratio
>100 & negative3700.0680.8330.08
Figure 3.

Distribution of cases based on the SnSAG2, 4/3 serum : cerebrospinal fluid (CSF) titer ratio. The SnSAG2 and SnSAG4/3 ELISAs were run and analyzed independently, with the lower serum : CSF titer ratio recorded for each sample. A majority of EPM cases exhibited a SnSAG2, 4/3 serum : CSF titer ratio that was less than 100, while a serum : CSF titer ratio that was greater than 100 was observed for most of the non-EPM cases.

As shown in Table 4, comparison of the SnSAG2, 4/3 ELISA sensitivity and specificity using the serum titer, the CSF titer, or the serum : CSF titer ratio as the diagnostic criterion indicated that the most favorable accuracy was achieved with the titer ratio at a cutoff of either ≤50 or ≤100. A titer ratio cutoff of ≤50 provided 86.4% sensitivity and 96.4% specificity. Diagnostic sensitivity was improved to 93.2% if a ratio cutoff of ≤100 was used, although the specificity of the assay was decreased to 83.3%. A CSF titer alone also was a reliable diagnostic criterion, with very good sensitivity and specificity observed using a cutoff of either 1 : 10 or 1 : 20 (Table 4). In contrast to the serum : CSF titer ratios and the CSF titers, serum titers alone did not provide good diagnostic accuracy. Although sensitivity of 86.4% could be achieved at a low cutoff of 1 : 500, specificity was unacceptable at 36.9% (Table 4). Similarly, specificity of 88.1% was obtained if a high endpoint titer of 1 : 4,000 was used as the cutoff, but this resulted in very poor sensitivity (29.5%).

Table 4. Comparison of assay sensitivity and specificity using serum titers, cerebrospinal fluid (CSF) titers, or serum : CSF ratios at different cutoff values
Serum Reciprocal TiterCSF Reciprocal TiterSerum : CSF Ratio
Cutoff% Sensitivity% SpecificityCutoff% Sensitivity% SpecificityCutoff%Sensitivity% Specificity
  1. a

    95% confidence intervals.

≥4,00029.5 (18.2–44.2)a88.1 (79.5–93.4)≥4077.3 (63.0–87.2)96.4 (90.0–98.8)≤2570.5 (55.8–81.8)98.8 (93.6–99.8)
≥2,00052.3 (37.9–66.2)75.0 (64.8–83.0)≥2084.1 (70.6–92.1)90.5 (82.3–95.1)≤5086.4 (73.3–93.6)96.4 (90.0–98.8)
≥1,00070.5 (55.8–81.8)60.7 (50.0–70.5)≥1093.2 (81.8–97.7)75.0 (64.8–83.0)≤10093.2 (81.8–97.7)83.3 (73.9–89.8)
≥50086.4 (73.3–93.6)36.9 (27.4–47.6)≥595.5 (84.9–98.7)58.3 (47.7–68.3)≥20095.5 (84.9–98.7)57.1 (46.5–67.2)

Because testing should be conducted only on horses showing signs of neurologic disease, diagnostic sensitivity and specificity also were calculated after removing the 10 non-neurologic horses from the sample set. All 10 of the normal horses had no detectable CSF titer or a serum : CSF titer ratio that was >100 (Fig 3), and removal of these cases resulted in a small decrease in specificity. At a titer ratio cutoff of ≤50, sensitivity remained unchanged at 86.4%, whereas specificity decreased from 96.4 to 95.9%. At a cutoff of ≤100, sensitivity remained at 93.2%, and specificity decreased from 83.3 to 81.1%.


The inability to distinguish horses that have been exposed to S. neurona but are clinically unaffected from horses that actually have the disease has long been a major impediment to accurate antemortem diagnosis of EPM. However, recent work has shown the diagnostic value of using antigen-specific CSF indices to demonstrate intrathecal antibody production against S. neurona.[25] These tests of proportionality rely on measurement of antigen-specific antibodies in the serum and CSF to determine if the amo unt of antibody in the CNS is greater than would be expected from normal passive transfer across the BBB. Consistent with previous findings,[28] we have shown in the present study that the antigen-specific antibody indices can be reduced to a simple ratio of serum antibodies to CSF antibodies to identify intrathecal antibody production. By using the SnSAG2, 4/3 titer ratio as the diagnostic criterion, very good sensitivity and specificity were achieved when examining the samples from 128 horses that had undergone postmortem examination. The optimal titer ratio cutoff for distinguishing horses with EPM from non-EPM horses approximated 100, consistent with the reference range of concentrations of specific proteins (eg, albumin and IgG) in the serum and CSF of horses.[25, 26] In many of the EPM horses, the titer ratio was extremely low, with some cases exhibiting CSF titers that approached or exceeded the serum titers. It must be noted, however, that titer ratios were not similarly low for some EPM and non-EPM cases (eg, ratio = 100), and these results will be encountered when the assay is used for antemortem diagnosis. For these cases, calculation of the AI or C-value can be performed to obtain a more accurate assessment of intrathecal antibody production. An anomalous titer ratio also may result from conditions that can disrupt the BBB, such as neuroborreliosis or viral encephalomyelitis. Therefore, it is advisable to conduct a more thorough analysis such as a C-value calculation, particularly if either the albumin quotient or CSF albumin concentration is high suggesting loss of BBB integrity.[26]

Very good diagnostic accuracy was achieved using a titer ratio cutoff of either ≤50 or ≤100 (Table 4). The decision to choose 1 of these cutoffs over the other should be predicated on the purpose for conducting the test. If a horse is being tested in a clinical setting to diagnose EPM, then it seems advisable to use ≤100 as the cutoff because there is a much greater cost in a false negative result that removes EPM from the differential diagnosis and leads to the decision not to treat the horse with anticoccidial drugs. Although the likelihood ratio for a horse that exhibits a serum : CSF titer ratio of 100 suggests that the horse is less likely to have EPM than another neurologic condition (Table 3), inclusion of a titer ratio = 100 increases the sensitivity to 93.2% and decreases the possibility of obtaining a false negative result. In contrast, a titer ratio cutoff of ≤50 would be advisable if using the assay to screen candidate horses for inclusion in an EPM study because this cutoff provides test specificity of 96.4% (Table 4) and horses grouped in this category were 24.2 times more likely to have EPM (Table 3).

Although 10 non-neurologic cases were included in the sample set for this study, it is important to emphasize that testing of horses without clinical signs of neurologic disease is not recommended, regardless of the diagnostic assay being used. However, inclusion of these samples in this study helps to demonstrate the antibody titers and serum : CSF ratios that will be seen in horses that do not have active S. neurona infection in the CNS. Furthermore, it is very likely that some samples submitted for testing will come from horses with conditions that are not neurologic in origin (eg, lameness). Therefore, the normal horses in the current sample set serve as valuable negative controls that further illustrate the utility of the serum : CSF ratio for eliminating EPM from the differential diagnosis.

Each serum and CSF sample pair was tested with both the SnSAG2 and the SnSAG4/3 ELISA, but the results from the 2 assays were considered collectively. That is, the higher end-point titer for each sample or the lower serum : CSF ratio for each sample pair was considered the relevant result and used for the final analysis. This was performed to account for the varied antibody responses that individual horses may exhibit against each of the SnSAGs (see Table S1 and other studies[21, 22]). By considering the results of the 2 assays collectively, greater diagnostic sensitivity can be achieved. Indeed, only 79.5% sensitivity (SnSAG2) or 88.6% sensitivity (SnSAG4/3) was observed in the present study when serum : CSF ratios from the 2 ELISAs were considered independently. Although a single ELISA incorporating a mixture of the SnSAG antigens would be more efficient, previous experiments suggested that this approach decreases the sensitivity of antibody detection.[22] Consequently, a collective result from the 2 independent ELISAs currently is preferred.

The results of this study highlight the value of collecting CSF from horses with suspected EPM to allow assessment of intrathecal antibodies against S. neurona. The serum : CSF titer ratio proved to be best for obtaining an accurate clinical diagnosis of EPM, but CSF titers alone were also found to be useful for identifying EPM horses. The likelihood ratios calculated for CSF titers indicated that there is a high probability of EPM in horses exhibiting CSF titers ≥1 : 40 (LR = 21.644; Table 2), and a CSF titer cutoff of either 1 : 20 or 1 : 10 provided good diagnostic sensitivity and specificity (Table 4). Minor blood contamination of CSF during sample collection is not a concern because antibody measurements are not impacted until a cell count of approximately 10,000 RBCs per μL of CSF is surpassed.[25, 29] This degree of blood contamination will impart an obvious red tinge to CSF, and these samples should be evaluated more thoroughly by calculating a C-value or AI. As mentioned above, CSF samples also should be considered suspect if noticeably cloudy or xanthochromic. For these cases, the albumin quotient or CSF albumin concentration should be carefully considered to assess whether the BBB has been compromised, in which case a C-value determination would be advisable.

Serum titers alone were a poor indicator of disease. Multiple EPM cases were observed with low or nondetectable antibodies in serum, despite appreciable antibody concentrations in the CSF, and numerous non-EPM cases had high serum titers (Fig 1). Although horses with high serum titers (≥1 : 4,000) were approximately 2.5 times more likely to have EPM (Table 1), use of a cutoff at this level provided very poor diagnostic sensitivity of 29.5% (Table 4). Sensitivity can be improved to suitable levels by decreasing the cutoff to a low serum titer of 1 : 500, but this will result in a concomitant decrease in diagnostic specificity to an unacceptable 36.9% (Table 4). These findings contrast with previous work that suggested IFAT serum titers provided very good accuracy for EPM diagnosis.[28] It is unlikely that this discrepancy is because of differences in the assays because there should be good concordance between the SnSAG2, 4/3 ELISA titers and the IFAT titers, albeit on a different dilution scale. Instead, it is probable that the previous findings obtained with the IFAT were biased by geography because all of the cases were collected from a single state (California). Indeed, serum antibodies were not detected in approximately 88% of the non-EPM horses in the sample set from this previous study,[28] which is very different from the seroprevalence that has been documented in populations of horses from most regions of the United States where EPM is considered enzootic.[7-14] Collectively, the data from the current study along with findings from a related study comparing the SnSAG2, 4/3 ELISAs and IFAT[31] indicate that serum titers do not provide an accurate diagnosis of EPM.

Demonstration of intrathecal antibody production requires a serologic test that can quantify antigen-specific antibodies in the serum and CSF. ELISAs are useful for quantifying antibodies because they are relatively simple, inexpensive, and generate objective data. The SnSAG2 and SnSAG4/3 assays used in this study are indirect ELISAs that have been shown to be very reliable and accurate for detecting antibodies against S. neurona.[21, 22, 30] These ELISAs also exhibit excellent analytical sensitivity,[22] which allows for analysis of serum and CSF pairs at dilutions that will encompass the expected normal antibody ratios (ie, those >100). Although the SnSAG2 and SnSAG4/3 ELISAs were shown in this study and previously[25] to be ideal assays for demonstrating intrathecal antibody production, it should be feasible to obtain similar results using other tests that can quantify antigen-specific antibodies, such as the IFAT.[19] Indeed, previous serologic investigation using the IFAT suggested that diagnosis of EPM was enhanced by examining the ratio of serum to CSF titers.[28] However, serum : CSF titer ratios obtained with different serologic tests may not be directly comparable, and optimal ratio cutoffs will need to be determined empirically if using assays other than the SnSAG2 or SnSAG4/3 ELISAs.

In summary, the findings presented here reaffirm that intrathecal antibody production against S. neurona is a valuable indicator for diagnosing EPM. Our results further show that the basic premise of the C-value and the antigen-specific AI can be reduced in practice to a simple ratio between serum and CSF antibody titers to assess whether the S. neurona-specific antibody concentrations present in the CNS are greater than expected from normal passive transfer of serum antibodies across the BBB. The C-value and the AI provide greater accuracy and confidence, and it is prudent to employ 1 of these indices when the diagnosis remains ambiguous (eg, ratio = 100) or there is suspicion that the BBB has been compromised. However, the serum : CSF titer ratio likely will be sufficient in many cases to achieve a dependable and accurate diagnosis of EPM.


We thank Whitney Mathes (Rood and Riddle Equine Hospital) for assisting with sample and data collection. Gratitude goes to all of the veterinary pathologists who conducted postmortem examinations of the cases compiled for this study. This research was supported by funds from the Amerman Family Equine Research Endowment.

Conflicts of Interest: S. Reed is a consultant for Equine Diagnostic Solutions, LLC. J. Morrow and A. Graves are co-owners of Equine Diagnostic Solutions, LLC, which has licensed the assays used in the study. D. Howe is the inventor listed on patents covering the assays used in the study.