This work was done at Ontario Veterinary College, University of Guelph, Guelph, ON, Canada. Results were presented as a Paper Abstract at the ACVIM Forum 2008, San Antonio, TX.
Validation of a Commercial Enzyme Immunoassay for Detection of Clostridium difficile Toxins in Feces of Horses with Acute Diarrhea
Article first published online: 6 APR 2010
Copyright © 2010 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 24, Issue 3, pages 628–632, May/June 2010
How to Cite
Medina-Torres, C.E., Weese, J.S. and Staempfli, H.R. (2010), Validation of a Commercial Enzyme Immunoassay for Detection of Clostridium difficile Toxins in Feces of Horses with Acute Diarrhea. Journal of Veterinary Internal Medicine, 24: 628–632. doi: 10.1111/j.1939-1676.2010.00506.x
- Issue published online: 7 MAY 2010
- Article first published online: 6 APR 2010
- Submitted September 3, 2009; Revised January 15, 2010; Accepted February 22, 2010.
Background: Clostridium difficile infection (CDI) is a recognized cause of colitis in the horse. Identification of its toxins is important for management of individual cases and for prevention of transmission and zoonosis. In humans, CDI diagnosis is performed with enzyme immunoassays, none of which have been validated for horses.
Hypothesis/Objectives: (1) Establish which test for CDI diagnosis was more frequently used by diagnostic laboratories, (2) determine the identified test's performance, sensitivity, and specificity, and (3) validate its use in diarrheic horses.
Animals: Samples were obtained from 72 horses presented with acute diarrhea and hospitalized at the Ontario Veterinary College, University of Guelph.
Methods: A survey was conducted to establish which of the tests for CDI diagnosis in horses is most commonly used throughout North America. A questionnaire was sent to all laboratories registered in the Veterinary Infection Control Society and the American Association of Veterinary Laboratory Diagnosticians. The performance of the test was evaluated by comparison to a cell cytotoxicity assay (CTA), the accepted Gold Standard for C. difficile toxin detection.
Results: The Techlab C. difficile Tox A/B II ELISA was the most frequently used test. Compared with the CTA, no significant difference was observed, and a good level of agreement (93%) was obtained. The diagnostic performance of the ELISA test was adequate (84% sensitivity and 96% specificity).
Conclusions and Clinical Importance: Results demonstrate that the Techlab C. difficile Tox A/B II ELISA is a reliable, adequate, and practical tool for identification of C. difficile toxins in horse feces.
Clostridium difficile infection
cytotoxicity inhibition assay
minimum essential medium
mean unbiased estimates
polymerase chain reaction
Veterinary Teaching Hospital
Identification of an etiologic agent in horses with acute diarrhea remains a difficult and often unrewarding task.1 An underlying cause is not identified in >50% of horses with acute diarrhea.2Clostridium difficile is a Gram positive, rod-shaped, catalase-negative, oxidase-negative, spore-forming, strictly anaerobic bacterium.3 It is a recognized cause of gastrointestinal disease in many species and has been implicated as a causative agent of acute diarrhea in adult horses4 and foals.5,6 Clinical presentation can range from a mild self-limiting form to per-acute and fatal disease. Furthermore, C. difficile infection (CDI) cannot be clinically differentiated from other causes of acute diarrhea in the horse.1 Thus, CDI should be considered an initial differential diagnosis in all horses with acute colitis.1,7
Accurate diagnosis of CDI is necessary for adequate management and specific therapeutic intervention.3 The gold standard for the diagnosis of CDI is the cytotoxicity inhibition assay (CTA).8 However, the CTA is labor-intensive and of limited availability, and thus, not conducive to routine testing.
In humans, CDI diagnosis has been performed by a variety of methods including direct visualization via sigmoidoscopy or colonoscopy for pseudomembranous colitis,9 culture of C. difficile,10,11 and detection of the C. difficile toxins by various enzyme immunoassays (EIAs).12 The sensitivity and specificity of EIAs are reportedly high for the diagnosis of CDI in humans13 and pigs,14 and low in dogs.15 Issues regarding diagnosis of CDI remain controversial, with some authors suggesting the combined use of culture and toxin detection, and others recommending toxin testing alone.3 The objectives of this study were (1) to determine which of the available tests for the diagnosis of CDI was most frequently used by veterinary diagnostic laboratories, (2) to determine the performance, sensitivity, and specificity of the identified test, and (3) to validate its use as a diagnostic tool for the detection of CDI in horses with acute diarrhea.
Materials and Methods
A survey of veterinary diagnostic laboratories throughout North America was conducted to identify the most frequently used test for the identification of CDI in horses. A questionnaire inquiring whether testing for C. difficile was performed, what tests were used, what types of specimens were evaluated, how discrepancies in the results were interpreted in laboratories where >1 test was performed, and if culture was routinely performed was used (Appendix 1). The questionnaire was sent by email to diagnostic laboratories registered in the Veterinary Infection Control Society (n = 27) and the American Association of Veterinary Laboratory Diagnosticians (n = 41) listserves. The survey was not pretested, and follow-up of nonresponders was attempted by resending the questionnaire on a 2nd occasion.
The most frequently used diagnostic test, the Techlab C. difficile Tox A/B IIa (Techlab ELISA), was evaluated in parallel with a CTA on fecal samples from diarrheic horses. A convenience sampling was performed and all samples were prospectively processed. A requirement of a minimum of 66 samples was estimated after performing a power calculation (β= 0.20) based on the performance of the ELISA test in humans. From March 1 to August 31, 2007, fecal samples were collected consecutively at admission from horses presented with acute diarrhea (<24 hours onset of fluid defecation) to the Veterinary Teaching Hospital (VTH), Ontario Veterinary College (OVC), and on the day of onset from patients that developed diarrhea while hospitalized. Samples were collected into a sterile container during spontaneous defecation, stored at 4°C within 2 hours of collection, and processed within the next 2 weeks. Results of the Techlab ELISA were compared with the CTA, using the latter as the gold standard for detection of the presence of C. difficile toxins. All samples also were cultured anaerobically after enrichment.
The Techlab ELISA was performed according to the manufacturer's instructions. The CTA was performed on VERO (African green monkey kidney) cellsb with the C. difficile Toxin/Antitoxin Kita according to the manufacturer's instructions. VERO cells were cultured in 96-well flat-bottom tissue culture plates with lids,c at 37°C in a 5% CO2, 95% filtered air, humidified incubator. Minimum essential medium (MEM) was used, with d-glucose at 6000 mg/L, Earle's salts, sodium bicarbonate at 3700 mg/L, without l-glutamine,b supplemented with 5% (v/v) fetal calf serum,b and 2% (v/v) antibiotic/antimycotic solution with 10,000 U of penicillin (base), 10,000 μg of streptomycin (base), and 25 μg of amphotericin B/mL.b Cells were incubated for 24 hours.
Test specimens were prepared by filtering 2 mL of the diarrheic fecal sample through a 0.45 μm filter.d Fifty microliters of the ultrafiltrate were added in duplicate to a microtiter well in a 96-well plate previously seeded with VERO cells as described above, after replacing the culture medium with 100 μL of fresh MEM. One well was inoculated with 50 μL of the kit antitoxin antibody, previously prepared and stored at −20°C until needed (cytotoxicity inhibition well). The other well (cytopathic well) received 50 μL of sterile phosphate-buffered saline (PBS). Plates were placed back in the incubator and observed under an inverted phase microscope for identification of cell rounding indicating cytotoxicity and toxin presence at 12, 18, and 24 hours postinoculation. When cell rounding was observed in the cytopathic well, but was not present in the cytotoxicity inhibition well, toxin inhibition was confirmed, and the sample was considered positive for the presence of C. difficile toxin B. The sample was considered negative when no cytopathic effect was observed.
For C. difficile isolation, a 1 cm3 aliquot of each sample was incubated for 7 days at 37°C in 9 mL of C. difficile moxalactam norfloxacin enrichment broth.e The broth then was vortexed and 4 mL shocked with anhydrous ethanol in a 1 : 1 dilution for 1 hour. After alcohol shock, samples were centrifuged at 3500 ×g for 10 minutes. The supernatant was discarded, and the pellet was cultured anaerobically on blood agare for 48 hours. Plates were read at 24 and 48 hours. C. difficile colonies were identified by morphology, Gram stain, odor, and detection of l-proline aminopeptidase activity.f
Statistical analysis of the results was performed by the McNemar's Test, and the odds ratio (OR) was determined to establish if OR was significantly different from 1 (null hypothesis [H0] of OR = 1). Kappa was used to establish the agreement between the Techlab ELISA and the CTA, following the Landis and Koch guidelines.16 Kappa was considered significant if P < .05. The diagnostic performance was established using exact binomial distribution procedures including computation of shortest (Sterne) exact 95% confidence intervals (CI). The sensitivity and specificity of the Techlab ELISA were established. The mean unbiased estimates (MUE) were calculated. When the expected mean of several random samples equals the population parameter, the sample mean is an unbiased estimate of the population.
Of 68 laboratories contacted in the survey, 27 (40%) responses were obtained. Twenty-five of these (93%) reported testing for C. difficile. The most commonly used diagnostic test was the Techlab ELISA (52%) followed by fecal culture (44%) (Table 1). Other diagnostics tests reported (used in <10% of laboratories) included the Triage C. difficile Panel (tests for C. difficile common antigen and toxin A), the Meridian BioSc Premier toxins A & B test (tests for toxins A and B), the CTA, polymerase chain reaction (PCR), and gross pathology and histopathology. Fecal culture use was reported in combination with one of the other diagnostic tests on 9 (36%) occasions. Combined diagnostic testing was reported by 10 (40%) of the 25 respondents, and where discrepant results were obtained, none reported retesting of the sample, 3 of the 10 (30%) relied on a toxin test result, and 2 of 10 (20%) had never seen discrepant results. The remaining 5 laboratories (50%) of the 10 in which >1 test was conducted on each sample did not indicate how discrepancies were interpreted. Criteria for testing a sample included testing every fecal sample (n = 5), testing upon clinician or pathologist request (n = 9), diarrheic stools only (n = 7), horses with colitis (n = 1), age—neonatal diarrhea (n = 2), and “when warranted by the case” (n = 3).
|Test||# of Laboratories||%a|
|Techlab C. difficile Tox A/B II ELISA||13||52|
|Triage C. difficile Panel||2||8|
|Meridian BioSc Premier toxins A & B||2||8|
|Toxin test (not specified)||3||12|
|Cell cytotoxicity assay||2||8|
|Gross pathology and histopathology||1||4|
For validation of the Techlab ELISA, fecal samples from 72 horses (39 male [54%], 33 female [46%]; age range, 4 days to 13 years; mean, 3.26 years) were collected. Of these, 56 (78%) samples were obtained upon admission from patients with diarrhea, and 16 (22%) were collected on the day of onset from previously hospitalized patients that developed diarrhea while at the VTH OVC. Results are presented in Table 2. No significant difference was observed between the Techlab ELISA and CTA (OR = 0.66; 95% CI: 0.08, 4.28; 2-tailed P= 1.00). Using the Landis and Koch guidelines for interpretation,16 an almost perfect agreement (93.1%) between the Techlab ELISA and the CTA was observed (κ= 0.82; CI: 0.67, 0.97; P-value under H0 of no agreement between tests is <.0001).
|Techlab ELISA+||16 (84.2% TP)||2 (3.8% FP)||18 (25.0%)|
|Techlab ELISA−||3 (15.8% FN)||51 (96.2% TN)||54 (75.0%)|
|Culture+||18 (94.7% TP)||1 (1.9% FP)a||19 (26.4%)|
|Culture−||1 (5.3% FN)b||52 (98.1% TN)||53 (73.6%)|
|Total||19 (26.4%)||53 (73.6%)||72|
Exact binomial distribution procedures for the Techlab ELISA performance yielded a sensitivity of 84% (95% CI: 60, 96%; MUE, 84%) and a specificity of 96% (95% CI: 86, 99%; MUE, 96%).
Fecal anaerobic culture results are presented in Table 2. The sensitivity and specificity of fecal culture, using the CTA as the gold standard, were 95% (CI: 72, 99%) and 98% (CI: 89, 99%), respectively. A 97.2% agreement between these 2 tests was established (κ= 0.93; CI: 0.83, 1.00).
The survey results showed the Techlab ELISA to be the most frequently used and readily available EIA for detecting CDI in horses in diagnostic laboratories throughout North America. This finding is consistent with human medicine practice, where EIAs are considered the clinical standard.12,13,17 There is considerable debate as to the optimal method of testing and it is suspected that false negative results are common in human medicine.17 Although some tests have good correlation with the CTA,17 its use as the gold standard also has been questioned because of a presumed relatively low sensitivity.12 The lack of equine-specific information was a limiting factor, and species-specific testing was required.
In addition, when asked about the combined use of tests and the interpretation of discrepant results (Appendix 1, question 3; a question that was prepared foreseeing separate, but simultaneous antigen and toxin testing), only 3 of 10 laboratories reported relying on and reporting the toxin test result in the face of discrepancy. This would be the most adequate way to interpret discrepancies, considering that demonstration of C. difficile toxin presence is required for the diagnosis of CDI.8 Two of the 7 remaining laboratories that performed simultaneous testing indicated that antigen-negative/toxin-positive samples had never been observed. However, although it might occur with relatively low frequency, toxin-positive samples in which antigen is not detected do occur when performing culture combined with either the CTA or the Techlab ELISA.
The questionnaire demonstrated the wide use of nonvalidated tests for the diagnosis of CDI in horses. In addition, criteria for conducting tests, sample testing, and result interpretation are not standardized. The sole use of fecal culture demonstrates the presence of C. difficile in the tested subject, but does not confirm the production of its toxins, which are necessary for the occurrence of CDI. It appears that a substantial number of nondiarrheic horses carry C. difficile in their colonic flora,18 with a prevalence of 7.6% recently identified in healthy racehorses in active training in Southern Ontario.19 PCR-based methods do not seem advantageous for CDI diagnosis in horses when compared with routine, nonmolecular-based procedures,12 although they might be useful in hospitalized humans, because of a lower apparent colonization rate. Furthermore, PCR methods, as is the case for the EIAs, have not been validated for the diagnosis of CDI in horses. The utilization of gross pathology and histopathology can be suggestive of the occurrence of CDI based on characteristic lesions, but this is not specific, and still requires demonstration of C. difficile toxins for confirmation.8
Another interesting observation is the apparent inclusion of normal stool in the tested samples. Some laboratories (18.5%) reported testing every fecal sample, and just 7 (28%) indicated that only diarrheic stools were processed. Testing of nondiarrheic feces is considered inappropriate in humans and it is hard to envision a justification for testing in nondiarrheic horses.3 With nondiarrheic samples, the pretest likelihood of a true positive result is very low, and with a low population prevalence, the positive predictive value decreases if the test is not 100% specific. However, requests by clinicians and pathologists were the most frequently encountered criteria for sample testing, which could partially explain this apparently inadequate sample inclusion.
The results obtained in the survey indicated a marked preference for the use of the Techlab ELISA in veterinary diagnostic laboratories throughout North America. Although the sensitivity and specificity of the Techlab ELISA is high for the diagnosis of CDI in humans (84.6 and 98.2%),13 and the test was shown to have a good correlation with the CTA in samples from pigs (91 and 86%),14 its performance was very poor on dog fecal samples (33 and 65%).15
The results of the present study demonstrate good correlation of the Techlab ELISA with the CTA on equine diarrheic fecal samples. With 84% sensitivity, the Techlab ELISA will potentially yield some false negative results. However, the high specificity (96%) indicates that the test will identify most negative horses as truly negative.
Because of convenience in laboratory work, samples were stored for 2 weeks before processing. Because this is not how the test would be used for diagnostic purposes, sample storage is a possible limitation of the present study. However, Weese et al20 demonstrated that C. difficile toxins were stable and could be detected by an EIA test after storage at 4°C for 30 days. Furthermore, the authors of the present study previously tested sample storage of toxin-positive C. difficile inoculated horse feces for 150 days, with no change in the detection of toxins by both the Techlab ELISA and the CTA for up to 3 months (unpublished data).
Anaerobic culture results, although not without discrepancy, appeared to correlate closely with the results of the Techlab ELISA and the CTA. However, the use of fecal culture alone should be discouraged, because positive fecal culture results for C. difficile are obtained in the general, nondiarrheic horse population.19
Overall, the performance of the Techlab ELISA was adequate. Sensitivity and specificity in diarrheic horses are comparable to results in humans,13 slightly lower and higher, respectively compared with results in pigs,14 and higher than results obtained in dogs.15
Although other EIAs could be valuable tools for the diagnosis of CDI in horses, results of these tests must be interpreted with caution until validation studies are conducted, and performance is documented. Meanwhile, the reliability of the results reported by diagnostic laboratories using methods other than the Techlab ELISA and gold standard (CTA) might be unreliable, and must be interpreted cautiously when treating horses with clinical manifestations resembling those of CDI.
In conclusion, the high level of agreement between the Techlab ELISA and the CTA, the acceptable but not ideal sensitivity, and the high specificity, indicate that the Techlab ELISA is an adequate diagnostic tool and a reliable and practical method for clinical diagnosis of CDI in horses.
aTechlab Inc; Blacksburg, VA
bGIBCO, Invitrogen Corp; Invitrogen Canada Inc, Burlington, ON, Canada
cSARSTEDT Inc, Newton, NC
dFisherbrand Syringe Filter 25 mm, 0.45 μm, NY, Sterile; Fisher Scientific Company; Ottawa, ON, Canada
eOxoid Ltd; Oxoid Company, Nepean, ON, Canada
fRemel PRO disc, Remel Inc; Remel Products 12076, Santa Fe, Lenexa, KS
Authors are grateful to Joyce Rousseau for excellent laboratory technique instruction and practice, and to William Sears for invaluable advice on statistical analysis. Funding provided by the Ontario Ministry of Agriculture and Food Research (OMAFRA) Equine Program—Equine Guelph.
- 2The etiology and treatment of idiopathic colitis: Recent studies. Proc Am Assoc Equine Pract 1992;38:433–439., ,
- 13Performance of the TechLab C. DIFF CHECK-60 enzyme immunoassay (EIA) in combination with the C. difficile Tox A/B II EIA kit, the Triage C. difficile panel immunoassay, and a cytotoxin assay for the diagnosis of Clostridium difficile-associated diarrhea. J Clin Microbiol 2004;42:4863–4865., , , et al.
- 19Fecal prevalence of Salmonella, Clostridium difficile and intestinal parasites in race horses in Ontario. J Vet Intern Med 2007;21:664., , , et al.
|1. Do you test for Clostridium difficile? Yes / No|
|2. Which test(s) do you use? ___________________________________________________________________________________________|
|3. If using a combination of antigen and toxin detection tests, how do you interpret discrepant results? (Antigen positive / Toxin negative or Antigen negative / Toxin positive) ____________________________________________________________________________________|
|4. Do you consider any criteria as a requirement for testing (e.g. consistency of the stools)? Or do you test every submitted fecal sample?|
|Every fecal sample___________________________|
|Diarrheic stools only_________________________|
|5. Do you routinely culture for Clostridium difficile? Yes / No|