Evaluation of Two Rapid Assays for Detection of Clostridium difficile Toxin A in Stool Specimens

Abstract

Rapid laboratory diagnosis of Clostridium difficile-associated diarrhea (CDAD) is highly desirable in the setting of hospital cost containment. We tested 654 stool specimens to compare the performance of two assays for rapid detection of toxin A, the Immunocard Toxin A test (Meridian Diagnostics, Inc.) and the Culturette Brand Toxin CD enzyme immunoassay (EIA) (Becton Dickinson Microbiology Systems), with a cytotoxin assay (Cytotoxi Test; Advanced Clinical Diagnostics) and culture on cycloserine-cefoxitin-fructose agar followed by determination of the production of toxins A and B. A chart review was performed for patients whose stool specimens provided positive results on one to three of the assays. With the “gold standard” of all four assays positive or chart review evidence of CDAD, 97 (14.8%) stool specimens were positive by one or more assays and 557 (85.2%) were negative by all methods. Total agreement for all assays was 90.5% (592 of 654). The sensitivity, specificity, positive predictive value, and negative predictive value for toxigenic culture were 94.7, 98.6, 87.1, and 99.5%, respectively, for toxigenic culture; 87.7, 98.6, 86.2, and 98.8%, respectively, for the cytotoxin assay; 71.9, 99.3, 91.1, and 97.3%, respectively, for the Immunocard; and 68.4, 99.1, 88.6, and 96.9%, respectively, for the Culturette EIA. While easy to perform and highly specific, these rapid assays do not appear to be sufficient for accurate diagnosis of CDAD.

Clostridium difficile can cause pseudomembranous colitis and up to 20% of cases of antibiotic-associated diarrhea without colitis (16). The organism produces two exotoxins: toxin A, which is probably responsible for most of the gastrointestinal symptoms seen because it functions as an enterotoxin, and toxin B, which is a potent cytotoxin. Rapid laboratory diagnosis of C. difficile-associated diarrhea (CDAD) is highly desirable in the setting of hospital cost containment. A variety of methods are presently available for the laboratory diagnosis of CDAD, including cell culture assay for the presence of cytotoxin, anaerobic culture of stool specimens for the organism followed by testing for the production of toxin (toxigenic culture), latex agglutination for the detection of C. difficile-associated antigen in stools, and enzyme immunoassays (EIAs) for the detection of toxin A, toxin B, or both (17). Although the cytotoxin assay has been considered the “gold standard” method for the diagnosis of CDAD, none of the methods mentioned above have yet been able to offer a combination of high sensitivity and specificity with ease and rapidity of test performance (19). In this study we have evaluated two rapid assays, both EIAs, for the detection of C. difficile toxin A in stool specimens and have compared their performance with those of a cytotoxin assay and toxigenic culture.

Stool specimens.

A total of 654 fresh stool specimens submitted to the Microbiology Service of the National Institutes of Health (NIH) Warren G. Magnuson Clinical Center and to the Microbiology Laboratory of Suburban Hospital with a request for C. difficile toxin detection were used in this evaluation and were tested by all assays in parallel. Specimens were stored at 4°C and tested within 72 h of collection. Testing was performed on Mondays, Wednesdays, and Fridays, and all C. difficile tests (culture and the three assays for detection of toxin in stool specimens) were set up on the same day in the NIH laboratory.

Immunocard Toxin A test.

The Immunocard Toxin A test (Meridian Diagnostics, Inc., Cincinnati, Ohio) is a rapid membrane EIA using a monoclonal antibody specific for C. difficile toxin A. Testing was performed according to the manufacturer’s instructions. Briefly, 25 μl of the patient’s stool specimen was diluted 1/15 in sample diluent and enzyme conjugate (horseradish peroxidase-conjugated anti-C. difficile toxin A monoclonal antibody), and 150 μl of the resulting suspension was added to each of two lower sample ports. The sample was allowed to enter the test card device and migrate along the membrane and through the upper reaction ports for 5 min, after which 3 drops of wash and 3 drops of substrate were added to the reaction ports. Samples that appeared to clog the sample ports were recorded as slow-flow samples, and a plastic or wooden applicator stick was used to gently stir the stool suspension in the sample ports in an effort to encourage migration along the membrane, although this procedure was not recommended by the manufacturer. The upper left reaction port contained immobilized toxin A and served as a control port. The upper right reaction port contained an immobilized anti-C. difficile toxin A capture antibody and served as the patient test port. Reaction ports were then observed for the development of any blue color after 5 min at room temperature. The development of a blue color in both the left (control) and right (test) reaction ports indicated a valid positive test result. Visually detectable blue color in the control reaction port with no blue color in the test reaction port indicated a negative result. Results with no blue color in the control reaction port were interpreted as invalid, and testing was repeated.

Culturette Brand Toxin CD assay.

The Culturette Brand Toxin CD assay (Becton Dickinson Microbiology Systems, Cockeysville, Md.) is a rapid EIA using microwells coated with an anti-toxin A polyclonal capture antibody. Testing was performed according to the manufacturer’s instructions. Briefly, 100 μl of the patient’s stool specimen was diluted 1/3 in sample buffer. Conjugate reagent (horseradish peroxidase-conjugated anti-C. difficile toxin A polyclonal antibody) and 100 μl of diluted stool sample were added to the microwells and incubated at 35°C for 1 h. After the microwells were washed to remove unbound conjugate, a chromogenic substrate was added and incubated at room temperature for 10 min. Stop solution was added, and the reactions were read spectrophotometrically at 450 nm in an MR5000 microplate reader (Dynatech Laboratories, Chantilly, Va.). A low-positive control well and a negative control well were included with each test run. An absorbance reading of ≥0.15 was considered a positive test result. Readings of ≥0.10 but <0.15 were considered indeterminate results, and testing on those specimens was repeated the next day samples were tested.

Cytotoxin assay.

The Cytotoxi test (Advanced Clinical Diagnostics, Toledo, Ohio) is a commercial cytotoxin assay which uses a mammalian epithelial cell line in microtiter plates. Testing was performed according to the manufacturer’s instructions. Briefly, 1 part of the stool specimen was mixed with 3 parts of dilution buffer, followed by centrifugation and filtration of the supernatant through a 0.45-μm-pore-size filter. Control wells for each test included a well with toxin only (positive control), a well with antitoxin only (antitoxin control), a well with toxin plus antitoxin (neutralization control), and a well with buffer only (blank control). All patient specimens were tested in two wells: one contained specimen filtrate plus diluent, and the other contained specimen filtrate plus antitoxin. The final specimen dilution for the test wells was 1:32. The presence of C. difficile toxin was detected in a stool sample by the appearance of a cytopathic effect (CPE) that could be neutralized by C. difficile antitoxin. Positive wells had CPE represented by ≥50% of the cells affected. Inoculated wells were examined for CPE after 24 and 48 h of incubation. Specimens for which CPE was observed in both test wells (i.e., antitoxin did not neutralize the CPE) were retested by using diluted specimen filtrate with a final dilution of 1:96.

Toxigenic culture.

All stool specimens were anaerobically cultured for C. difficile, with determination of toxin A and B production on all C. difficile isolates. Spores of C. difficile were selected for by alcohol treatment. Equal volumes of the stool specimen and absolute ethanol were mixed and incubated at room temperature for 1 h. The alcohol-treated stool specimen was inoculated onto CDC anaerobe blood agar (Remel, Lexena, Kans.) and anaerobically reduced cycloserine-cefoxitin-fructose agar (CCFA) (Anaerobe Systems, San Jose, Calif.) and was incubated anaerobically at 37°C in an anaerobic chamber for 72 h. Colonies on CCFA that were suspected of being C. difficile were identified by using the PRO Disc (Carr Scarborough Microbiologicals, Decatur, Ga.) as previously described (9) and were tested for toxin production. For toxin B production, an agar plug adjacent to C. difficile colonies was removed and tested by the cytotoxin assay by substituting the agar plug for the patient specimen filtrate in the cytotoxin assay. For toxin A production, colonies of C. difficile were inoculated into chopped meat glucose broth (Carr Scarborough Microbiologicals) and incubated anaerobically at 37°C for 5 days, at which time the broth supernatant was tested with the Culturette EIA.

Analysis of results.

The cytotoxin assay, toxigenic culture, Immunocard, and Culturette EIA were performed and interpreted independently by different individuals who did not know the results of the other assays. By using the cytotoxin assay and/or toxigenic culture as a gold standard method, the sensitivities, specificities, positive predictive values, and negative predictive values of the Immunocard and Culturette assays were calculated. In addition, performance parameters were calculated for all assays following a review of patient charts. Patient charts were reviewed (when available) when a specimen yielded positive results from three or fewer of the four assays tested. A patient was considered to have CDAD if his or her stool specimen gave a positive result in all four of the assays or if the following chart review criteria were met: (i) at least six loose or watery bowel movements within a 48-h period, (ii) antimicrobial therapy within 8 weeks preceding the onset of diarrhea, (iii) appropriate positive response to therapy with either metronidazole or vancomycin or improvement of diarrhea after antibiotic withdrawal, (iv) exclusion of other etiologies of diarrhea, or (v) demonstration of pseudomembranous colitis by lower gastrointestinal endoscopy (1, 6, 10, 20, 22).

Of 654 specimens tested, 97 (14.8%) were positive by one or more assays and 557 (85.2%) were negative by all assays. All four assays were positive for 35 specimens, giving a 90.5% (592 of 654) total agreement for all assays. Culture provided 111 isolates of C. difficile, of which 33 (29.7%) were nontoxigenic and 78 (70.3%) were toxigenic. All 78 toxigenic organisms produced both toxin A and toxin B. There were 56 specimens with both cytotoxin- and toxigenic culture-positive results. The performance characteristics of the Immunocard and Culturette EIA compared to those of the cytotoxin assay, toxigenic culture, and the cytotoxin assay plus toxigenic culture are shown in Table 1.

TABLE 1.

Performance characteristics of the cytotoxin assay, toxigenic culture, Immunocard Toxin A assay, and Culturette toxin A EIA

Gold standard	Assay	Result	No. of gold standard results:		Performance characteristica (%)
			Positive	Negative	Sens	Spec	PPV	NPV	Accuracy
Cytotoxin assay	Toxigenic Culture	Positive	56	22	83.6	96.2	71.8	98.1	94.8
		Negative	11	564
	Immunocard	Positive	39	8	58.2	98.6	82.9	95.3	94.5
		Negative	28	579
	Culturette EIA	Positive	36	8	53.8	98.6	81.8	94.9	94.0
		Negative	31	579
Toxigenic culture	Cytotoxin	Positive	56	11	71.8	98.1	83.6	96.3	94.9
		Negative	22	565
	Immunocard	Positive	44	3	56.4	99.5	93.6	94.4	93.9
		Negative	34	570
	Culturette EIA	Positive	41	3	52.6	99.5	93.2	93.9	93.9
		Negative	37	573
Cytotoxin assay + toxigenic culture	Immunocard	Positive	39	8	69.6	98.6	82.9	97.2	96.2
		Negative	17	590
	Culturette EIA	Positive	36	9	64.3	98.5	76.6	96.7	95.6
		Negative	20	589

^aSens, sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value. 

Charts were reviewed for 46 patients to determine their CDAD status. Charts were unavailable or incomplete for the remaining 16 of the 62 patients whose stool specimens were positive in three or fewer of the four assays performed. CDAD was indicated for 57 patients, resulting in a prevalence of 8.9% (57 of 638). The performance characteristics of all four assays for the detection of CDAD are shown in Table 2.

TABLE 2.

Performance of assays for the detection of CDAD

Assay	Result	No. of specimens from patients with CDAD:		Performance characteristica (%)
		Positive	Negative	Sens	Spec	PPV	NPV	Accuracy
Cytotoxin	Positive	50	8	87.7	98.6	86.2	98.8	97.6
	Negative	7	573
Toxigenic culture	Positive	54	8	94.7	98.6	87.1	99.5	98.3
	Negative	3	573
Immunocard	Positive	41	4	71.9	99.3	91.1	97.3	96.8
	Negative	16	577
Culturette EIA	Positive	39	5	68.4	99.1	88.6	96.9	96.8
	Negative	18	576

^aSens, sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value. 

We observed that 16.2% (106 of 654) of the specimens appeared to clog the sample ports (slow-flow specimens) in the Immunocard test card. The distribution of stool consistency in our study was 53.8% soft (unformed), 41.9% liquid, and 4.3% formed. The distribution of consistency of slow-flow specimens was 73.4% soft, 18.3% liquid, and 8.3% formed. With these slow-flow specimens, we found that migration of the stool specimen could be aided by using a plastic or wooden applicator stick to gently stir the stool suspension in the sample ports. Slow-flow specimens included 8 of the 41 (19.5%) true-positive specimens detected by the Immunocard. We believe that some of these positive specimens might have been missed if we had not aided the migration of the stool suspension.

Indeterminate results with the Culturette EIA were found in 5.2% (34 of 654) of specimens. Upon repeat testing only half of the indeterminate results were resolved. No specimens with indeterminate results after repeat testing were considered truly positive. One true positive was detected as a result of repeat testing of specimens with indeterminate results.

The sensitivities of the rapid assays in this study were much lower than those of the cytotoxin assay or toxigenic culture (Table 2). To date, only abstracts of evaluations of the Immunocard and the Culturette EIA have been published. Reported sensitivities of the Immunocard have ranged from 70 to 90.6% (2, 3, 7, 8, 13, 21). When chart reviews were performed to establish clearer diagnoses of CDAD, the sensitivities of the Immunocard have been reported to be 73, 77.3, and 85% (2, 3, 7). These results are similar to our results. Reported sensitivities for the Culturette EIA are similar to the 68.4% found in our study, with 71% reported when Culturette EIA results were compared to cytotoxin results alone and 65% reported when a chart review was performed (4, 23). In a study to determine the analytical sensitivity of five EIA kits for the detection of C. difficile toxin A, the Culturette EIA required the largest amount of toxin (12,770 pg/ml of stool) for a positive result (14). This was also apparent in our study because both rapid tests had difficulty detecting toxin in specimens with weak cytotoxin activity (not cytotoxin positive until the 48-h reading). Eleven of the 56 cytotoxin- and toxigenic culture-positive specimens were cytotoxin positive after 48 h, and only one of these was positive in a rapid test (Culturette EIA).

The poor performance of the two rapid assays in this study and in others suggests that these are inadequate choices for use as the sole assay for the diagnosis of CDAD in a clinical laboratory. Another option is to use one of these rapid assays as a screening test with confirmation of positive results by another method, such as the cytotoxin assay. The Immunocard evaluated in this study detected toxin A. A previous study evaluated the performance of another Immunocard assay for the detection of C. difficile glutamate dehydrogenase, which is a marker antigen for C. difficile (22). The sensitivity of this Immunocard C. difficile test was reported to be 83%, with a larger number of false-positive results (37 of 139) than false-negative results (20 of 720). The performance of the two rapid assays for toxin A in the present study was hampered by a higher number of false-negative results than false-positive results (Table 2). This is unfortunate, since a screening test with a large number of false-negative results is not useful. Most specimens for C. difficile testing are truly negative (91.1% in this study). All negative specimens would have to be tested with the confirmatory assay if a screening test has a high false-negative rate. The reason for performing a screening test for CDAD is to reduce the workload by selecting specimens that are likely to be positive for further testing.

After a review of patient charts, toxigenic culture was found to be the best laboratory method for the diagnosis of CDAD tested in this study (Table 2). Culture will on occasion result in the detection of toxin-producing C. difficile isolates that are not causing disease. One of our patients who was tested had CDAD demonstrated by cytotoxin assay, toxigenic culture, and clinical criteria; a month later, this patient had toxigenic culture-positive but toxin-negative stools, with a few loose stools and no other clinical features of CDAD. Goodpasture and Bridge reported that culture coupled with a confirmatory toxin assay on all C. difficile isolates is the most accurate and least expensive of the methods presently used to diagnose CDAD (12). Unfortunately, toxigenic culture is not a rapid method. Culture can be more rapid when alcohol shock, CCFA, and the PRO Disc are used to isolate and identify C. difficile (9, 11). Our data support the contention of Peterson et al., who believe that the methodology presently available necessitates both testing for the presence of toxin (either A or B or both) and culturing for toxigenic C. difficile for maximum sensitivity in the diagnosis of CDAD (19). Kader et al. report that detection of toxin A or toxin B alone is inadequate for the diagnosis of CDAD in pediatric patients (15). In light of recent reports of C. difficile isolates that do not produce both toxins, testing for both toxin A and toxin B may be warranted (5, 18). We also understand that rapid diagnosis is necessary in many institutions in order to initiate specific antibiotic treatment for C. difficile. Our results do not allow us to recommend the Immunocard Toxin A test or the Culturette EIA as the sole assay for the diagnosis of CDAD. It remains clear that laboratory results for the diagnosis of CDAD must be correlated and interpreted with the clinical presentation of the patient (20).