PresentationPDF Available

PERFORMANCE ASSESSMENT OF DIFFERENT DIAGNOSTIC ASSAYS TO IDENTIFY EPM-AFFECTED HORSES IN A CLINICAL SETTING

Authors:
  • Amwell Data Services LLC

Abstract

Presented by R. Belgrave at the 2nd EPM Society Workshop, Lake Tahoe, NV, October 2017
PERFORMANCE ASSESSMENT OF DIFFERENT DIAGNOSTIC ASSAYS TO
IDENTIFY EPM-AFFECTED HORSES IN A CLINICAL SETTING
Rachel Lemcke1, Rodney Belgrave1, Jennifer Morrow2, Nicola Pusterla3
1Mid-Atlantic Equine Medical Center, Ringoes, NJ 08551
2Equine Diagnostic Solutions, Lexington, KY 40511
3School of Veterinary Medicine, University of California, Davis, CA 95616.
Current antemortem EPM diagnostic strategies determine serum, CSF, and serum: CSF antibody
titers to differentiate acute or heightened infection from general exposure to S. neurona. To more
effectively identify EPM-affected patients and better evaluate assay performance, we compared
paired results from two different EPM assays (the IFAT and the SAG 2, 4/3 ELISA) within a
subpopulation of patients at an equine clinical hospital. Sampled across four years, IFAT CSF
samples, in addition to paired SAG serum and CSF, were submitted soon after collection (n=88
horses). For a subpopulation (n=18), antibody ratios for serum: CSF for both IFAT and SAG
were calculated on paired samples and tested using the same aliquots. EPM-positive samples
were defined as: ≥1:160 serum and ≥1:5 CSF antibodies on IFAT; ≥1:250 serum and ≥1:2.5 CSF
antibodies on SAG; ≤64 and <100 ratios on serum: CSF ratio for IFAT and SAG assays,
respectively. Descriptive statistical analysis, including overall percent agreement (OPA), positive
percent agreement (PPA), negative percent agreement (NPA), Cohen's kappa coefficient (k), and
McNemar p test, was performed to compare assay components, as a perfect reference EPM test
standard is nonexistent. When paired results for the SAG serum: CSF ratio were compared to the
IFAT CSF, the OPA, PPA, and NPA were 87.50, 47.37, and 98.55, respectively, indicating that
both tests can equally diagnostically exclude EPM (p<0.02; n=88). Additionally, the SAG ratio
was interpreted as EPM less frequently than the IFAT CSF; moderate agreement between the
assays (k=0.55, 95% CI 0.33-0.78) also points to this disparity in EPM identification. Similarly,
comparison between the SAG to IFAT serum: CSF ratio results revealed an OPA, PPA, and
NPA of 88.89, 85.71, and 90.91, respectively, with stronger agreement between the tests
(k=0.77, 95% CI 0.46-1.00). These assay results did not demonstrate an identification bias
toward either assay (n=18; p=0.48), suggesting the assays could be potentially substituted for
each other, though a larger sample size should be considered in future analysis. Comparisons
between IFAT or SAG serum versus the SAG or IFAT ratio, respectively, showed a poor
agreement of results: OPA and k were 44.44 and -0.23 (95% CI -0.63-0.17), and 44.44 and 0.07
(95% CI, -0.07-0.22), respectively. Additionally, while ratio calculations for both assays require
positive CSF samples, positive SAG CSF samples occasionally yielded negative ratios (NPA
60.76; k=0.24 (95% CI 0.10-0.38); n=88). Interestingly, the IFAT CSF perfectly correlated with
the ratio (OPA 100; k=1.00; n=18). A lack of corresponding necropsy results on sampled horses,
and a small subpopulation for direct ratio comparisons, unfortunately limits assessment of EPM
assay accuracy (i.e. sensitivity and specificity), though past research has evaluated assay
accuracy (Johnson et al., 2013). Clinicians relying solely on serum testing for a diagnosis may
likely mistake infection for exposure given the high seroprevalence (Witonsky, 2016). CSF
testing alone seemed appropriate for the IFAT assay, but did not seem sufficient on the SAG
assay. In conclusion, clinicians selecting the IFAT assay could potentially only analyze CSF,
while the SAG assay seems to perform best when using the serum: CSF ratio.
Johnson, A.L., Morrow, J.K., Sweeney, R.W., 2013. Brief Communication 596599.
Witonsky, S., 2016. ACVIM Consensus Statement 491502. doi:10.1111/jvim.13834
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.