Comparison of culture I cytotoxin assay and two EIA tests with clinical diagnosis of Clostridium diffici / e-associated diarrhea

OBJECTIVE
The most common etiology of infectious diarrhea in hospitalized patients is Clostridium difficile. No single laboratory test yields a definitive diagnosis. Four methods were evaluated for their sensitivity and specificity in patients who had clinically defined C difficile-associated diarrhea.


METHODS
Clinical criteria for C difficile-associated diarrhea were defined. All adult in-hospital patients whose stools were tested for C difficile were prospectively followed. Stools were examined with culture on a selective medium, a commercial cytotoxicity assay (cta), and two commercially available enzyme immunoassays (eias) for toxin A (Meridian) and toxin AB (cbc).


RESULTS
During the study period 235 stool specimens from 185 patients were tested. Fifty-one patients were positive for C difficile or its markers, cta was most sensitive (80%), whereas cbc-eia was most specific (98%). Differences in the sensitivities of cta and Meridian-eia were minor (80% versus 73.3%) and they were equally specific (95.5%).


CONCLUSIONS
The sensitivity and specificity of eia for toxin A is similar to other tests. However, due to rapidity and ease of performance, it may be a more practical test for the diagnosis of C difficile-associated diarrhea, especially if the cytotoxin assay is not available.

C LOSTRIDIUM DIFFICILE IS AN OPPORTUNISTIC PATHOGEN that can cause intestinal infection in any setting and following any procedure that destabilizes the normal protective intestinal flora.Antibiotic therapy is, by far, the most common predisposing factor, implicated in more than 98% of cases .The organism produces two toxins, an enterotoxin (toxin A) a nd a cytotoxin (toxin B) ( 1 ,2).Toxin A is believed to play the primary role in the pathogenesis of the disease because of its enterotoxicity.Toxin B is cytotoxic and possibly produces tissue damage after the initial action of toxin A.
C dijficile is a major cause of hospital acqu ired dianhea.Toxigenic C dijficile, its toxins or both have been detected in 12 to 19% of all fecal specimens submitted to microbiology laboratories (3,4).The spectrum of disease caused by C dijficile eA.'i.ends from asymptomatic carriage through mild self-limited diarrhea to severe pseudomembranous coli tis .In the 1970s, detection of toxin B by tissue cu lture was shown to have a good conelation with the endoscopic finding of pseudomembranous disease.However.pseudomembranous disease is now rarely seen because most C dijficile infections are diagnosed and treated earlier and endoscopy is not warranted (5).Consequ ently, in the recent past there h ave been very few reports in which endoscopy findings were used for diagnosis.Many have used the clinical criteria as a gold standard to assess the performance of C dijficile diagnostic methods.Presently.five different types of tests are available to detect C difficile or its toxins in feces.We tested four assays and compared the results with the clinical diagnosis of C difficile-associated dianhea.

MATERIALS AND METHODS
Stool specimens submitted to the Department of Microbiology, Victoria Hospital.London, Ontario from July l.1992 to December 31.1992 were studied.Only adu lt patients with a hospital stay of more than 72 h were included.Stools were investigated for SaLmonelLa species.ShigelLa species, Yersinia species and Campy-L obacterspecies , as well as for Escl1ericl1iacoLi0157:H7.In addition, stools were tested for the presence of C dljficile and its toxins A and B. At the time of culture, a portion of each specimen was aliquoted and stored at -70°C for la ter cytotoxicity testing by tissue culture method.C difficile culture: Stools were subj ected to alcohol shock treatment (6) to kill vegetative cells .After removal of alcohol by centrifugation, the pellet was used to inoculate a selective medium fo r C dijficile containing D-cycloserine.cefoxitin and fructose (CCF medium; Oxoid, Unipath).The plates were incubated for up to 72 h at 35°C under anaerobic conditions.Spreading yellow colonies with ground glass appearan ce resembling C dijficile were subcultured onto 5% horse blood plates containing Columbia base agar.One plate was incubated aerobically and the oU1er anaerobically for 48 h.The anaerobic isolates were considered to be C dijficile if lliey had typical colony morphology on the selective plates and produced butyric, isobutyric, valerie and isocaproic acid (7).Toxin A by EIA (Meridian-EIA): Toxin A was detected by enzyme immunoassay (ElA) using a Premier kit (Meridian Diagnostics, Inc, Ohio) .Stool specimens were prepared, processed and results were interpreted according to the instru ctions of the manufacturer.The kit employs polyclonal antitoxin A capture antibody adsorbed to breakaway microwells and requires 2.5 h to perform.Results were read visually.

RESULTS
Patients included in this study were all 18 years of age or older.During the study period of July l to December 31.1992, 235 stools samples from 185 adult in-hospital patients were received.Fifty samples were repeat specimens from 1 7 patients who were negative for C dijficile markers.It is our po licy not to test repeat specimens from patients who are positive for any of the C dijficile markers.One hundred and two patients had significant diarrhea as defined by this study.One hundred and sixteen pat.ients had received antibiot.icspreviously.None of the stool samples were positive for Salmonetla species, Shigetla species.Ye rsinia species.E coli Ol57:H7 or Campylobacier species.Of 185 patients 51 were posit.ivefor C dij]i.cile or its markers.C dij]i.cile was isolated from the stools of 41 patients.Toxin B was detected in stool samples of 31 patients by CTA.The Meridian-EtA detected toxin A in 29 patients.whereas CBC-ElA was positive in only 22 patients.Eighteen pat.ients were posit.ivefor all C dij]i.cile tests.The study included only two patients who underwent sigmoidoscopy.Neither had visual or h istological evidence of pseu domembranous colitis .Using clin ical and diagnostic criteria, 30 patients were considered to have diarrhea due to C dij]i.cile.Of 155 pat.ients considered not to have C dijfi.cile-associated diarrhea.four met all four clinical criteria.Of 30 patients with C dijficile disease, 23 were culture-positive , 24 were positive by CTA, 22 by Meridian-EtA and 19 were positive by CBC-ElA.SiA1.een of the 30 patients were positive by all four tests.Since repeat specimens from patients wiU1 positive markers were not tested .mulliple specimens from negative patients were disregarded and were not included in U1e calculations.Sensitivities.specificities, and positive and negative predictive values of each test are shown in Table l.
In patients who had diarrhea but were negative for C dijficile markers.diarrhea was attributed to hyperalimentat.ion in 39 cases, ulcerative colitis or Crohn's disease in six, gastrointestinal malignancies in 10 and necrotizing colitis in one.

DISCUSSION
The investigation of fecal specimens from patients suspected of C dijfi.cile-associated diarrhea has become a major task for most clin ical laboratories.A s imple, rapid and reliable test for C dij]i.cile is h ighly desirable.Currently, five kinds of commercial tests are available as diagnostic aids for C dijficile infection: culture, latex agglutination and toxin assay by tissue culture.ElA and dot immunobinding assay.
C dijficile culture requires up to three days for results.Compared with clinical criteria, the culture tech-niqu e has been shown to be the most sensitive (4,8 -11) though this was not our experience.False positive rates with this test are high as C difficile can be recovered from a high proportion of hospitalized patie nts rece iving antibiotics without evidence of disease (12,13).Twothirds of all cases of antibiotic-associated diarrhea are not du e to C difficile (14) and it appears that in the a bsen ce of de tectable toxin.the organism does not cause disease (2 , 15).In a ddition.a significant proportion of C difficile isolates.up to 40%.from hospitalized patients are nontoxigenic (4.8, 16) .Culture, like th e latex agglutination test, does not differentiate between toxigenic and non toxigenic strains of C difficile.
Earlier studies using endoscopy demonstrated a good correlation between CTA and pseudomembranous colitis (13,17,18).However, the results of a cytotoxin assay depend on the severity of the disease.CTA is positive in over 90% of patients with pseudomembranous colitis (19 ,20) and the toxin detection rate in fecal filtrates increases with the severity of the disease (21).It is.however, the less severe disease that poses a diagnostic dilemma.The sensitivity of CTA has been reported to range between 67 and 78% , using clinical criteria similar to ours (4,(8)(9)(10)(11).Contrary to the findings of most investigators , in the pres ent study CTA was more sensitive than culture.In this regard our findings are not unique.DiPersio and investigators (22) found U1at among pa tients with clinical C diffi-cile-associated dia rrhea, CTA detec ted more cases than any other lest, including culture.Furthermore, our sensitivity of 80% was comparable to previously reported rates .Many factors including final dilution of inoculum, age and type of the cell line employed influence CTA results (23)(24)(25).As we batched our specimens and the cell cultures were fresh , it is doubtful that the res ults of CTA could be duplicated in the routine laboratmy setting.Repeat inc ubation of microtrays causes progressive deterioration of cell lines.This ma y cause false n egative results towards to the end of the shelf life , as the cells may not be viable.
Since 1988, enzyme immunoassays for C difficile toxin A have been commercially available.At present.
only Cambridge Biotech Corporation markets an ElA lest that detects both toxins , A and B. Practically all strains of C difficile tested to da te produce either both toxins or no detectable toxin (2,26).Therefore, for diagnostic purposes it matters little which toxin one detects.CTA detects toxin B, which is 1000-fold more cytotoxic than toxin A. CTA is also more sensitive than ElA, as il can detect picogram a mounts of toxin B, whereas ElA tests require 1 to 10 ng of toxins/mL for detection.ElA results are also depe ndent on the quality and specificity of antibodies used for liga nd capture (2).Indeterminate tests, although not encountered in this study, can be a problem.Doern and coworkers (27) found the CBC-ElA to be more sensitive than the Meridia n -EIA: 84.5% versus 69% .Their criteria of C difficileassociated diarrhea were different from ours and were b ased on the results of four assays, in a ddition to clinical assessment of patients.It should also be noted U1at in their study, nine specimens yielding false n egative results on initial testing with Meridian-EtA tested positive with this assay on repeat testing.Sensitivity and specificity of the Meridian-EtA in the present study are similar to those found by DiPersio eta! (22).
In conclusion, although not shown by our study, CTA is considered the single most useful laboratory test for the diagnosis of C diffi-cile-associated dia rrhea.In our h ands the Meridian-EtA was only slightly less sensitive.Of 30 patients with C diffi-cile-associated diarrhea, Meridian -EtA detected only two fewer patients than CTA a nd was equal to CTA in specificity.The CBC-ElA was the most specific test, but least sensitive.Though the data are based on a relatively small number of patients with C diffi-cileassociated diarrhea, it appears tl1at the Meridian-EIA can be used as a primary test for detecting C diffi-cileassociated diarrhea.if CTA is not available.In a ddition , ElA has the advantage of results being available in a short lime.Regar-dless of the method of diagnosis, stools from patients witl1out diarrhea and no history of prev ious antibiotic therapy should not be tested for C difficile.Due to frequent colonization of hospitalized patients with this organism, results from patients without symptoms are like ly to be less specific .
CAN J INFECT D1s VoL 5 No 4 JuLY 1 AuGusT 1994