- Email: [email protected]
Detection of toxin production in Clostridium difficile strains by three different methods
ORIGINAL ARTICLE Detection of toxin production in Clostridium difficile strains by three different methods M. Yu¨cesoy1, J. McCoubrey2, R. Brown2 and I. R. Poxton2 1
Dokuz Eylu¨l University, Medical School, Department of Microbiology and Clinical Microbiology, Izmir, Turkey and 2Department of Medical Microbiology, University of Edinburgh Medical School, Edinburgh, UK
Objective To compare two immunoassays for detection of toxins produced in vitro by isolates of Clostridium difficile with the standard tissue culture assay, to help in the diagnosis of C. difficile-associated diarrhoea.
Toxin production was investigated in 42 strains of C. difficile of various serotypes, ribotypes and S-protein types. These included strains from our laboratory collection, strains freshly isolated from stool specimens of patients suspected of suffering from C. difficile-associated disease or of carrying it asymptomatically, and one reference strain (NCTC 11223). Toxin was assayed by (i) a rapid slide immunoassay (C. difficile toxin A test, Clearview, Oxoid), (ii) an enzyme-linked microplate immunoassay (C. difficile toxin A/B test, Techlab), and (iii) a tissue culture assay. The rapid slide assay and the enzyme immunoassay were performed according to the manufacturers’ recommendations. The tissue culture assay was performed using Vero cells. Methods
Thirty of the 42 strains (71%) were shown to be positive for toxin A by the slide immunoassay and 34 of the strains (81%) were found to be toxin A/B producers by the enzyme immunoassay. The same 34 strains that were positive in the enzyme immunoassay also produced toxin B (cytotoxin) in the tissue culture assay. The sensitivity, specificity, and positive and negative predictive values for the rapid slide immunoassay method were calculated to be 88.2%, 100.0%, 100.0% and 66.7%, respectively, when compared to tissue culture assay results as the reference method. These values for the enzyme immunoassay method were all 100.0%. In this study eight strains were found to be non-toxin-producing by all methods. It is possible that there were four strains that only produced toxin B (A– Bþ), and were missed by the rapid A-only assay.
Results
We can recommend the use of the Techlab A þ B enzyme immunoassay for the detection of toxin production by C. difficile strains because of its high sensitivity and specificity, its ease of use, and its capability of detecting both A- and B-type toxins.
Conclusions
Keywords Clostridium difficile, toxin A, toxin B, immunoassay, antibiotic-associated diarrhoea, tissue culture, cytopathic effect Accepted 8 January 2002
Clin Microbiol Infect 2002; 8: 413–418
INTRODUCTION
Corresponding author and reprint requests: M Yu¨cesoy, Dokuz Eylu¨l University, Medical School, Department of Microbiology and Clinical Microbiology, Izmir, Turkey Tel: þ90 232 2777777 4505 Fax: þ90 232 2590541 E-mail: [email protected]
Until the mid-1970s, the clinical importance of Clostridium difficile was not understood. However, it is now clearly recognized that this organism is responsible for antibiotic-associated gastrointestinal diseases ranging from benign diarrhoea to life-threatening pseudomembranous colitis [1,2]. There are a number of virulence factors associated
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases
414 Clinical Microbiology and Infection, Volume 8 Number 7, July 2002
with this organism. Toxins A and B are considered the most important but there are also various extracellular enzymes and an ADP-ribosylating toxin, and even spore formation is an important attribute [1]. S-layer proteins, which vary considerably between strains, both in molecular mass and immunogenic reactivity, might have a role in virulence by promoting adherence and colonization or by evasion of the immune system [3,4]. The pathogenicity of C. difficile is clearly associated with the production of the two exotoxins, toxin A and toxin B, two of the largest bacterial toxins known, with molecular weights of 308 000 and 270 000, respectively. Toxin A is primarily an enterotoxin and causes hemorrhage and fluid secretion, and is also chemotactic for neutrophils. It elicits the release of cytokines that play important roles in its pathogenesis. Both toxins enter the cell primarily by receptor-mediated endocytosis. Toxin B is lethal and very cytotoxic for most cell lines, while toxin A has some cytotoxic activity but much less than toxin B. The toxins are believed to exert an additive effect in vivo causing destruction of the cellular cytoskeleton by inducing depolymerization of actin. For a recent review on the actions of the toxins, and other virulence factors of C. difficile, see Poxton et al. [5]. C. difficile-associated diarrhea is an emerging nosocomial problem in many countries [6,7]. Evidence suggests that the nosocomial spread of this organism can occur by direct patient to patient contact, by transmission to patients from the hands of the hospital personnel, or by acquisition from the environment [8]. There is still debate as to whether culture or direct detection of toxin in stool specimens is the preferred method for diagnosis. Culture is obviously important for epidemiological studies, while toxin detection in stools can give a rapid indication that the patient is being exposed to biologically active toxin. Many studies have compared various commercially available immunoassays for direct detection of toxin. However, if culture is considered important then it is necessary to be able to demonstrate toxin production by the isolates. The aim of our study was to compare three simple methods for detection of toxin (toxin A-only immunoassay, a toxin A þ B immunoassay and the tissue culture) from a wide range of isolates of C. difficile. These 42 strains varied in cell surface phenotype (different serotypes and S-protein types), genotype (ribotype), and came from a wide
range of patients and geographical origins. It is important to be able to detect the production of toxins in isolates of C. difficile to help in the diagnosis of C. difficile-associated disease. MATERIALS AND METHODS Bacterial strains Forty-two C. difficile strains of various ‘Delme´ e’ serotypes (A, B, C, D, F, G, H, I, K and X), ‘Cardiff’ ribotypes (10, 23, 26, 56 and 106), ‘Edinburgh’ Sprotein types (5336, 5242, 5144 and 5941; see [4] for description of ‘S-type’), and a selection of uncharacterized isolates from patients with symptoms and from symptomless babies were used in the study. These included strains from our laboratory collection, strains freshly isolated from the stool specimens of patients suspected of suffering from C. difficile-associated disease (CDAD), symptomless patients and one NCTC strain (NCTC 11223). Culturing for toxin testing The strains were inoculated into brain heart infusion, proteose peptone medium [9] and incubated under anaerobic conditions at 37 8C for 5 days. Toxin was measured in supernates that were obtained after cultures had been centrifuged at 3000 g for 20 min. Toxin tests The toxin production by the strains was investigated by (i) a rapid slide immunoassay (C. difficile toxin A test, ClearviewTM Oxoid, Basingstoke, UK), (ii) a microplate enzyme immunoassay (EIA; C. difficile toxin A þ B, Techlab, Blacksburg, VA, USA), and (iii) a tissue culture assay (see below). The rapid slide immunoassay and the microplate assay were both performed according to the manufacturers’ recommendations. For the rapid slide immunoassay 100 mL culture supernatant was added to 1 mL sample diluent and vortexed for 10–15 s. This sample (125 mL) was inoculated to the test unit. Blue latex particles, which had been sensitized with antibody specific to C. difficile toxin A, bound to toxin A in the sample and this complex was trapped by antibody to toxin A, forming a visible blue line. The results of the rapid slide assay were read 30 min after the application of the sample to the test unit. A blue line in the control and result windows indicated a positive result.
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
Yu¨cesoy et al
For the EIA, 50 mL of the culture supernatant were added to 200 mL of the specimen diluent and vortexed for 10 s. After adding one drop of conjugate to the wells, two drops of diluted sample were inoculated. One negative and one positive control well were included for each plate. The plate was incubated in a shaker at 37 8C for 50 min. After washing, one drop of substrate A and one of substrate B were added to the wells. After 10 min one drop of stopping solution was added. The results of the EIA were read visually and spectrophotometrically at a dual wavelength 450/620 nm and the wells with an optical density (OD) <0.080 were interpreted as negative while those with an OD of 0.080 were considered positive. The assay took approximately 1.5 h. Tissue culture assay For the tissue culture assay Vero cells were cultured in Eagle’s minimal essential medium (alpha modification: Sigma, St Louis, MO, USA) supplemented with fetal calf serum (7%), L-glutamine (2 mM), penicillin (200 units/mL) and streptomycin (200 mg/mL) at 37 8C in 5% CO2. The cells were grown to confluent monolayers in 75 cm2 flasks, counted with Trypan blue in an improved Neubauer counting chamber and resuspended in fresh medium at a concentration of 105 cells/mL. This was used to seed 96-well microtitre plates, adding 200 mL volumes to each well of a 24-well plate.
Table 1 The distribution of the tissue culture results according to the positive results obtained for >50% (3 þ) endpoint
Measurement of C. difficile toxin 415
Plates were incubated for 20–24 h before the inoculation of the culture supernatants. Serial dilutions of the C. difficile culture supernatants were made in sterile saline and 20 mL of these diluted samples was inoculated into the wells. The plates were incubated at 37 8C in 5% CO2 and assay results were determined for typical changes after 24 and 48 h according to a scale from 0 to 4þ [4þ being the titre showing a 95–100% cytopathic effect (CPE) compared to the control wells]. The cytotoxic titres were expressed as the highest dilution exhibiting >50% (3þ) CPE. RESULTS Thirty of the 42 strains (71%) were shown to be positive for toxin A by the slide immunoassay and 34 of the strains (81%) were found to be toxin A/Bproducing by the microplate assay. The visual and spectrophotometric dual wavelength results gave exactly the same result. The same 34 strains that were positive in the EIA also produced toxin B (cytotoxin) in the tissue culture assay. The distribution of the tissue culture results according to the titres is shown in Table 1. The sensitivity, specificity and positive and negative predictive values (PPV, NPV) for both methods, when compared to tissue culture assay results as the reference method, are shown in Table 2. The agreements between rapid immu-
Dilutions Incubation period
4 n (%)
16 n (%)
24 h 48 h
3 (8.8) 2 (5.9)
4 (11.8) 8 (23.5) 14 (41.2) 3 (8.8) 7 (20.6) 14 (41.2)
Table 2 The sensitivity, specificity, positive and negative predictive values calculated for rapid slide and enzyme-linked immunoassays Sensitivity (95% CI) Specificity (95% CI) Positive predictive value (95% CI) Negative predictive value (95% CI)
64 n (%)
256 n (%)
1024 n (%)
4096 n (%)
>4096 n (%)
4 (11.8) 1 (2.9) 5 (14.7) 2 (5.9)
– 1 (2.9)
Rapid slide immunoassay (toxin A test)
Enzyme-linked immunoassay (toxin A/B test)
88.2% (71.6, 96.2) 100.0% (59.8, 100.0) 100.0% (85.9, 100.0) 66.7% (35.4, 88.7)
100.0% (87.4, 100.0) 100.0% (59.8, 100.0) 100.0% (87.4, 100.0) 100.0% (59.8, 100.0)
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
416 Clinical Microbiology and Infection, Volume 8 Number 7, July 2002
noassay and tissue culture results and between EIA and tissue culture results, were 90.5% and 100%. In this study eight strains (serotypes D, I, X, A7 and A10; ribotype 10, S type 5144/ribotype 26 and an uncharacterized isolate from a symptomatic patient) were found to be non-toxin-producing by all methods. It is possible that there were four strains (serotypes K, A8, A9 and an uncharacterized strain isolated from a symptomless baby) that only produced toxin B (A– Bþ), and were missed by the rapid A-only assay. DISCUSSION The widespread use of broad-spectrum antibiotics is causing an increasing incidence of pseudomembranous colitis and antibiotic-associated diarrhoea [10]. Clostridium difficile causes disease in hospitalized adults, primarily the elderly and those receiving antibiotics such as clindamycin, ampicillin, or oral cephalosporins [11]. The presence of diarrhoea and a history of recent antimicrobial therapy are only suggestive of CDAD; laboratory tests are necessary for the confirmation of the diagnosis [10]. Detection of toxin in the stool of a symptomatic patient is one of the most accurate methods for the diagnosis of these infections. Although important for epidemiology, the detection of just the organism may have limited value because the asymptomatic carriage rate may be high in hospitalized patients receiving antibiotics. Organisms that do not produce toxin are thought to be avirulent [12], and thus interpretation of a positive culture result alone is not possible. Toxin production by isolates is therefore a useful piece of extra evidence when attempting to make a definite diagnosis based on culture alone. The detection of toxin B by the tissue culture assay is considered to be the standard [13]. However, tissue culture techniques are not routinely available in most diagnostic laboratories. The technique may require an incubation period of up to 24 h (although strong positives can be seen after a few hours) and there are problems with the standardization of the method. The laboratory protocols vary in the dilutions and cell lines used, and in the interpretation of the assay endpoint [14]. Some authors accept a 50% CPE as a positive result, however, some consider a >90% CPE as the endpoint.
Yolken et al. [15] first developed and evaluated an EIA for the detection of C. difficile toxin in 1981. After this attempt, further developments have continued and a number of commercial diagnostic kits are now available for the detection of either toxin A or both toxins. In this study of strains in pure culture the tissue culture cytotoxicity assay was compared with two of these tests (a slide immunoassay for toxin A only and a microplate immunoassay for toxins A and B). The tests we used to detect the toxins from the culture supernatants of the isolated strains were found to react in the same way when detecting toxin from specimens (unpublished data). The principle of the microplate immunoassay is based on a monoclonal antibody to toxin A and the antigen–antibody complex is visualized by a precipitate, which is formed by blue latex coated with antibody. The sensitivity, specificity, PPV and NPV obtained for the toxin A test were 88.2%, 100.0%, 100.0%, 66.7%, respectively. Bentley et al. [16] used this test to detect toxin A in fecal specimens and reported the sensitivity and specificity values as 83.1% and 96.9%, respectively. Vanpoucke et al. [17] used the same test and found the sensitivity, specificity, PPV and NPV to be 89%, 83%, 71%, 94%, respectively, for the detection of toxin in stool samples. Our rates on pure cultures are concordant with the results of these studies, and it seems reasonable to use this test for the detection of toxin both in stool specimens and C. difficile culture supernatants. It has the advantage of being rapid (the result is ready after 30 min), extremely easy and one is able to work with one specimen at a time instead of collecting samples. The main disadvantage is that it will not detect toxin B produced by A– Bþ strains. However, the occurrence of this phenotype is uncommon and localized. We also used the toxin A þ B EIA test for the detection of both toxins. In our study the sensitivity, specificity, PPV and NPV were found to be 100.0%. In one multicenter study, this kit was used to detect the toxins in 1152 fecal specimens and the values obtained for the sensitivity ranged between 83.3 and 96.0%, for the specificity between 99.3 and 100% and for NPV between 90.0 and 99.5%. They found PPV to be 100% and the correlation as 94.9–99.5% when compared with the tissue culture assay [18]. Aldeen et al. [19] compared the same toxin A þ B test to cell culture cytotoxin assay in 1109 diarrhoeal stool samples. They got the sensitivity and specificity as 94.3%,
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
Yu¨ cesoy et al
99.3%, initially. However, after resolution of six discrepancies, the detected sensitivity, specificity, PPV and NPV were 94.5%, 100%, 100% and 98.8%, respectively. The agreement between toxin A þ B and tissue culture assay was 98.5%. Our results for pure cultures are in parallel with the above fecal toxin results. This test combines the advantages of an EIA method, such as rapidity, ease of use and sensitivity, with the capability of detecting both toxins. Also, EIAs have objective endpoints determined with an EIA reader. De Girolami et al. [8] compared visual and spectrophotometric readings and have reported good correlations. Our results are parallel to this finding. The reported sensitivity and specificity values for other available commercial EIA kits which can detect toxins A and B were between 79.6 and 96.2%, 93.5 and 100%, respectively, and PPV and NPV were between 86.7 and 100% and 95.7 and 99%, respectively [14,18, 20,21]. These results are similar to, or a little lower than, those obtained for the Techlab A/B test. Four of our strains appear to be toxin A– Bþ strains which were negative in toxin A EIA but positive in the toxin A/B test. It is possible that the sensitivity levels of the toxin A test we used may be low and the production of toxin A is missed. On the other hand, they might be true toxin A– Bþ strains that could have caused CDAD in the patients. Lyerly et al. [18] mentioned the presence of a toxin A– Bþ isolate from a patient with a clinical history consistent with C. difficile disease. In recent years toxin A– Bþ strains have been described in symptomatic adults in Japan and the USA [22,23]. Alfa et al. [24] reported that the toxin A– Bþ strain was responsible for a nosocomial outbreak of CDAD. Brazier et al. [25] also reported the same phenotype C. difficile strains from symptomatic patients. It has been suggested that these strains might have caused disease due to a more active toxin B [18]. Another group reported that it was possible either that toxin B alone was capable of causing diarrhoea or that the strain contained other factors capable of causing disease. Lyerly et al. [18] also found that two serogroup type F strains showed toxin A– Bþ phenotype in their study. This was in agreement with the findings of Depitre et al. [26]. The ribotypes of A– Bþ strains, which were reported by Brazier et al. [25], belonged to ribotype 17. Our apparent A– Bþ strains belonged to serogroups A and K (K, A8, A9). We also had a strain which belonged to
Measurement of C. difficile toxin 417
serogroup F which was found to produce both A and B toxins. Eight of the isolates were shown to be non-toxinproducers. This reflects the fact that there are nontoxigenic C. difficile strains which exist in the gastrointestinal tracts of hospitalized patients as normal colonizers. It is possible, but so far not proven, that non-toxigenic strains may cause disease depending on the predisposing factors of the host and the virulence of these strains. In other studies it is well recognized that a proportion of strains isolated from patients whose clinical course was consistent with C. difficile-associated disease were found to be non-producers of toxin, e.g. 3% (two of 63) in the study by Clabots et al. [27]. From the result of our study we can recommend the use of the Techlab toxin A/B enzyme immunoassay for the detection of toxin production by C. difficile isolates because of its high sensitivity, specificity, ease of use and its capability of detecting both A- and B-type toxins. However, the Oxoid, ClearviewTM system is so easy to use, and as it is designed for single samples, its usefulness should not be overlooked, especially if there is no history of Toxin A– Bþ strains in the locality. It is recommended, however, that isolates of C. difficile are regularly monitored to see which toxins they are producing. ACKNOWLEDGMENTS Presented as a poster at 11th European Congress of Clinical Microbiology and Infectious Diseases (1–4 April 2001, Turkey). REFERENCES 1. Murray PR, Rosenthal KS, Kobayashi GS, Pfaller MA. Clostridium. Medical Microbiology, 3rd edn. St Louis Missouri: Mosby Inc., 1998; 296–307. 2. Brazier JS, Borriello SP. Microbiology, epidemiology and diagnosis of Clostridium difficile infection. In: Aktories K, Wilkins TD, eds. Current Topics in Microbiology and Immunology Clostridium Difficile. Berlin: Springer, 2000; 1–33. 3. Sharp J, Poxton IR. The cell wall proteins of Clostridium difficile. FEMS Microbiol Letts 1988; 55: 99–104. 4. McCoubrey J, Poxton IR. Variation in the surface layer proteins of Clostridium difficile. FEMS Immunol Med Microbiol 2001; 31: 131–5. 5. Poxton IR, McCoubrey J, Blair G. The pathogenicity of Clostridium difficile. Clin Microbiol Infect 2001; 7: 421–7.
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
418 Clinical Microbiology and Infection, Volume 8 Number 7, July 2002 6. Macgowan AP, Brown I, Feeney R et al. Clostridium difficile-associated diarrhoea and length of hospital stay (letter). J Hosp Infect 1995; 31: 241–4. 7. Riley TV, O’Neil GL, Bowman RA, Golledge CL. Clostridium difficile-associated diarrhoea: epidemiological data from Western Australia. Epidemiol Infect 1994; 113: 13–20. 8. De Girolami PC, Hanff PA, Eichelberger K et al. Multicenter evaluation of a new enzyme immunoassay for detection of Clostridium difficile enterotoxin A. J Clin Microbiol 1992; 30: 1085–8. 9. Poxton IR, Brown R, Fraser AG et al. Enteropathogenic clostridia and Clostridium botulinum. In: Collee JG, Fraser AG, Marmion BP, Simmons A, eds. Mackie & McCartney’s Practical Medical Microbiology, 14th edn. Edinburgh: Churchill Livingstone, 1996; 546. 10. Schue´ V, Green GA, Monteil H. Comparison of the ToxA test with cytotoxicity assay and culture for the detection of Clostridium difficile-associated diarrhoeal disease. J Med Microbiol 1994; 41: 316–18. 11. Finegold SM. Clinical considerations in the diagnosis of antimicrobial agent-associated gastroenteritis. Diagn Microbiol Infect Dis 1986; 4: 87S–91S. 12. Barlett JG. Clostridium difficile: clinical considerations. Rev Infect Dis 1990; 12: S243–S251. 13. Jacobs J, Rudensky B, Dresner J et al. Comparison of four laboratory tests for diagnosis of Clostridium difficile-associated diarrhoea. Eur J Clin Microbiol Infect Dis 1996; 15: 561–6. 14. Merz CS, Kramer C, Forman M et al. Comparison of four commercially available rapid enzyme immunoassays with cytotoxin assay for detection of Clostridium difficile toxin (s) from stool specimens. J Clin Microbiol 1994; 32: 1142–7. 15. Yolken RH, Whitcomb LS, Marien G. Enzyme immunoassay for the detection of Clostridium difficile antigen. J Infect Dis 1981; 144: 378. 16. Bentley AH, Patel NB, Sidorczuk M et al. Multicentre evaluation of a commercial test for the rapid diagnosis of Clostridium difficile-mediated antibioticassociated diarrhoea. Eur J Clin Microbiol Infect Dis 1998; 17: 788–90. 17. Vanpoucke H, De Baere T, Claeys G, Vaneechoutte M, Verschraegen G. Evaluation of six commercial
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
assays for the rapid detection of Clostridium difficile toxin and/or antigen in stool specimens. Clin Microbiol Infect 2001; 7: 55–64. Lyerly DM, Neville LM, Evans DT et al. Multicenter evaluation of the Clostridium difficile TOX A/B Test. J Clin Microbiol 1998; 36: 184–90. Aldeen WE, Binham M, Aiderzada A, Kucera J, Jense S, Carroll KC. Comparison of the TOX A/B test to a cell culture cytotoxicity assay for the detection of Clostridium difficile in stools. Diagn Microbiol Infect Dis 2000; 36: 211–13. Barbut F, Kajzer C, Planas N, Petit JC. Comparison of three enzyme immunoassays, a cytotoxicity assay, and toxigenic culture for diagnosis of Clostridium difficile-associated diarrhea. J Clin Microbiol 1993; 31: 963–7. Arrow SA, Croese L, Bowman RA, Riley TV. Evaluation of three commercial enzyme immunoassay kits for detecting faecal Clostridium difficile toxins. J Clin Pathol 1994; 47: 954–6. Kato H, Kato N, Fukui K, Ohara A, Watanabe K. High prevalence of toxin A negative/toxin B positive Clostridium difficile strains among adult inpatients. Clin Microbiol Infect 1997; 3: 220–2. Sambol S, Gerding D, Merrigan M, Lyerly D, Johnson S. Severe truncation of the toxin A gene in a pathogenic Clostridium difficile (CD) strain not detectable by toxin A immunoassay. Clin Infect Dis 1998; 27: 946–9. Alfa MJ, Kabani A, Lyerly D et al. Characterization of a toxin A-negative, toxin B-positive strain of Clostridium difficile responsible for a nosocomial outbreak of Clostridium difficile-associated diarrhea. J Clin Microbiol 2000; 38: 2706–14. Brazier JS, Stubbs SL, Duerden BI. Prevalence of toxin A negative /B positive Clostridium difficile strains. J Hosp Infect 1999; 42: 248–9. Depitre C, Delmee M, Avesani V et al. Serogroup F strains of Clostridium difficile produce toxin B but not toxin A. J Med Microbiol 1993; 38: 434–41. Clabots CR, Johnson S, Olson MM, Peterson LR, Gerding DN. Acquisition of Clostridium difficile by hospitalized patients: evidence for colonized new admissions as a source of infection. J Infect Dis 1992; 166: 561–7.
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
Dokuz Eylu¨l University, Medical School, Department of Microbiology and Clinical Microbiology, Izmir, Turkey and 2Department of Medical Microbiology, University of Edinburgh Medical School, Edinburgh, UK
Objective To compare two immunoassays for detection of toxins produced in vitro by isolates of Clostridium difficile with the standard tissue culture assay, to help in the diagnosis of C. difficile-associated diarrhoea.
Toxin production was investigated in 42 strains of C. difficile of various serotypes, ribotypes and S-protein types. These included strains from our laboratory collection, strains freshly isolated from stool specimens of patients suspected of suffering from C. difficile-associated disease or of carrying it asymptomatically, and one reference strain (NCTC 11223). Toxin was assayed by (i) a rapid slide immunoassay (C. difficile toxin A test, Clearview, Oxoid), (ii) an enzyme-linked microplate immunoassay (C. difficile toxin A/B test, Techlab), and (iii) a tissue culture assay. The rapid slide assay and the enzyme immunoassay were performed according to the manufacturers’ recommendations. The tissue culture assay was performed using Vero cells. Methods
Thirty of the 42 strains (71%) were shown to be positive for toxin A by the slide immunoassay and 34 of the strains (81%) were found to be toxin A/B producers by the enzyme immunoassay. The same 34 strains that were positive in the enzyme immunoassay also produced toxin B (cytotoxin) in the tissue culture assay. The sensitivity, specificity, and positive and negative predictive values for the rapid slide immunoassay method were calculated to be 88.2%, 100.0%, 100.0% and 66.7%, respectively, when compared to tissue culture assay results as the reference method. These values for the enzyme immunoassay method were all 100.0%. In this study eight strains were found to be non-toxin-producing by all methods. It is possible that there were four strains that only produced toxin B (A– Bþ), and were missed by the rapid A-only assay.
Results
We can recommend the use of the Techlab A þ B enzyme immunoassay for the detection of toxin production by C. difficile strains because of its high sensitivity and specificity, its ease of use, and its capability of detecting both A- and B-type toxins.
Conclusions
Keywords Clostridium difficile, toxin A, toxin B, immunoassay, antibiotic-associated diarrhoea, tissue culture, cytopathic effect Accepted 8 January 2002
Clin Microbiol Infect 2002; 8: 413–418
INTRODUCTION
Corresponding author and reprint requests: M Yu¨cesoy, Dokuz Eylu¨l University, Medical School, Department of Microbiology and Clinical Microbiology, Izmir, Turkey Tel: þ90 232 2777777 4505 Fax: þ90 232 2590541 E-mail: [email protected]
Until the mid-1970s, the clinical importance of Clostridium difficile was not understood. However, it is now clearly recognized that this organism is responsible for antibiotic-associated gastrointestinal diseases ranging from benign diarrhoea to life-threatening pseudomembranous colitis [1,2]. There are a number of virulence factors associated
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases
414 Clinical Microbiology and Infection, Volume 8 Number 7, July 2002
with this organism. Toxins A and B are considered the most important but there are also various extracellular enzymes and an ADP-ribosylating toxin, and even spore formation is an important attribute [1]. S-layer proteins, which vary considerably between strains, both in molecular mass and immunogenic reactivity, might have a role in virulence by promoting adherence and colonization or by evasion of the immune system [3,4]. The pathogenicity of C. difficile is clearly associated with the production of the two exotoxins, toxin A and toxin B, two of the largest bacterial toxins known, with molecular weights of 308 000 and 270 000, respectively. Toxin A is primarily an enterotoxin and causes hemorrhage and fluid secretion, and is also chemotactic for neutrophils. It elicits the release of cytokines that play important roles in its pathogenesis. Both toxins enter the cell primarily by receptor-mediated endocytosis. Toxin B is lethal and very cytotoxic for most cell lines, while toxin A has some cytotoxic activity but much less than toxin B. The toxins are believed to exert an additive effect in vivo causing destruction of the cellular cytoskeleton by inducing depolymerization of actin. For a recent review on the actions of the toxins, and other virulence factors of C. difficile, see Poxton et al. [5]. C. difficile-associated diarrhea is an emerging nosocomial problem in many countries [6,7]. Evidence suggests that the nosocomial spread of this organism can occur by direct patient to patient contact, by transmission to patients from the hands of the hospital personnel, or by acquisition from the environment [8]. There is still debate as to whether culture or direct detection of toxin in stool specimens is the preferred method for diagnosis. Culture is obviously important for epidemiological studies, while toxin detection in stools can give a rapid indication that the patient is being exposed to biologically active toxin. Many studies have compared various commercially available immunoassays for direct detection of toxin. However, if culture is considered important then it is necessary to be able to demonstrate toxin production by the isolates. The aim of our study was to compare three simple methods for detection of toxin (toxin A-only immunoassay, a toxin A þ B immunoassay and the tissue culture) from a wide range of isolates of C. difficile. These 42 strains varied in cell surface phenotype (different serotypes and S-protein types), genotype (ribotype), and came from a wide
range of patients and geographical origins. It is important to be able to detect the production of toxins in isolates of C. difficile to help in the diagnosis of C. difficile-associated disease. MATERIALS AND METHODS Bacterial strains Forty-two C. difficile strains of various ‘Delme´ e’ serotypes (A, B, C, D, F, G, H, I, K and X), ‘Cardiff’ ribotypes (10, 23, 26, 56 and 106), ‘Edinburgh’ Sprotein types (5336, 5242, 5144 and 5941; see [4] for description of ‘S-type’), and a selection of uncharacterized isolates from patients with symptoms and from symptomless babies were used in the study. These included strains from our laboratory collection, strains freshly isolated from the stool specimens of patients suspected of suffering from C. difficile-associated disease (CDAD), symptomless patients and one NCTC strain (NCTC 11223). Culturing for toxin testing The strains were inoculated into brain heart infusion, proteose peptone medium [9] and incubated under anaerobic conditions at 37 8C for 5 days. Toxin was measured in supernates that were obtained after cultures had been centrifuged at 3000 g for 20 min. Toxin tests The toxin production by the strains was investigated by (i) a rapid slide immunoassay (C. difficile toxin A test, ClearviewTM Oxoid, Basingstoke, UK), (ii) a microplate enzyme immunoassay (EIA; C. difficile toxin A þ B, Techlab, Blacksburg, VA, USA), and (iii) a tissue culture assay (see below). The rapid slide immunoassay and the microplate assay were both performed according to the manufacturers’ recommendations. For the rapid slide immunoassay 100 mL culture supernatant was added to 1 mL sample diluent and vortexed for 10–15 s. This sample (125 mL) was inoculated to the test unit. Blue latex particles, which had been sensitized with antibody specific to C. difficile toxin A, bound to toxin A in the sample and this complex was trapped by antibody to toxin A, forming a visible blue line. The results of the rapid slide assay were read 30 min after the application of the sample to the test unit. A blue line in the control and result windows indicated a positive result.
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
Yu¨cesoy et al
For the EIA, 50 mL of the culture supernatant were added to 200 mL of the specimen diluent and vortexed for 10 s. After adding one drop of conjugate to the wells, two drops of diluted sample were inoculated. One negative and one positive control well were included for each plate. The plate was incubated in a shaker at 37 8C for 50 min. After washing, one drop of substrate A and one of substrate B were added to the wells. After 10 min one drop of stopping solution was added. The results of the EIA were read visually and spectrophotometrically at a dual wavelength 450/620 nm and the wells with an optical density (OD) <0.080 were interpreted as negative while those with an OD of 0.080 were considered positive. The assay took approximately 1.5 h. Tissue culture assay For the tissue culture assay Vero cells were cultured in Eagle’s minimal essential medium (alpha modification: Sigma, St Louis, MO, USA) supplemented with fetal calf serum (7%), L-glutamine (2 mM), penicillin (200 units/mL) and streptomycin (200 mg/mL) at 37 8C in 5% CO2. The cells were grown to confluent monolayers in 75 cm2 flasks, counted with Trypan blue in an improved Neubauer counting chamber and resuspended in fresh medium at a concentration of 105 cells/mL. This was used to seed 96-well microtitre plates, adding 200 mL volumes to each well of a 24-well plate.
Table 1 The distribution of the tissue culture results according to the positive results obtained for >50% (3 þ) endpoint
Measurement of C. difficile toxin 415
Plates were incubated for 20–24 h before the inoculation of the culture supernatants. Serial dilutions of the C. difficile culture supernatants were made in sterile saline and 20 mL of these diluted samples was inoculated into the wells. The plates were incubated at 37 8C in 5% CO2 and assay results were determined for typical changes after 24 and 48 h according to a scale from 0 to 4þ [4þ being the titre showing a 95–100% cytopathic effect (CPE) compared to the control wells]. The cytotoxic titres were expressed as the highest dilution exhibiting >50% (3þ) CPE. RESULTS Thirty of the 42 strains (71%) were shown to be positive for toxin A by the slide immunoassay and 34 of the strains (81%) were found to be toxin A/Bproducing by the microplate assay. The visual and spectrophotometric dual wavelength results gave exactly the same result. The same 34 strains that were positive in the EIA also produced toxin B (cytotoxin) in the tissue culture assay. The distribution of the tissue culture results according to the titres is shown in Table 1. The sensitivity, specificity and positive and negative predictive values (PPV, NPV) for both methods, when compared to tissue culture assay results as the reference method, are shown in Table 2. The agreements between rapid immu-
Dilutions Incubation period
4 n (%)
16 n (%)
24 h 48 h
3 (8.8) 2 (5.9)
4 (11.8) 8 (23.5) 14 (41.2) 3 (8.8) 7 (20.6) 14 (41.2)
Table 2 The sensitivity, specificity, positive and negative predictive values calculated for rapid slide and enzyme-linked immunoassays Sensitivity (95% CI) Specificity (95% CI) Positive predictive value (95% CI) Negative predictive value (95% CI)
64 n (%)
256 n (%)
1024 n (%)
4096 n (%)
>4096 n (%)
4 (11.8) 1 (2.9) 5 (14.7) 2 (5.9)
– 1 (2.9)
Rapid slide immunoassay (toxin A test)
Enzyme-linked immunoassay (toxin A/B test)
88.2% (71.6, 96.2) 100.0% (59.8, 100.0) 100.0% (85.9, 100.0) 66.7% (35.4, 88.7)
100.0% (87.4, 100.0) 100.0% (59.8, 100.0) 100.0% (87.4, 100.0) 100.0% (59.8, 100.0)
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
416 Clinical Microbiology and Infection, Volume 8 Number 7, July 2002
noassay and tissue culture results and between EIA and tissue culture results, were 90.5% and 100%. In this study eight strains (serotypes D, I, X, A7 and A10; ribotype 10, S type 5144/ribotype 26 and an uncharacterized isolate from a symptomatic patient) were found to be non-toxin-producing by all methods. It is possible that there were four strains (serotypes K, A8, A9 and an uncharacterized strain isolated from a symptomless baby) that only produced toxin B (A– Bþ), and were missed by the rapid A-only assay. DISCUSSION The widespread use of broad-spectrum antibiotics is causing an increasing incidence of pseudomembranous colitis and antibiotic-associated diarrhoea [10]. Clostridium difficile causes disease in hospitalized adults, primarily the elderly and those receiving antibiotics such as clindamycin, ampicillin, or oral cephalosporins [11]. The presence of diarrhoea and a history of recent antimicrobial therapy are only suggestive of CDAD; laboratory tests are necessary for the confirmation of the diagnosis [10]. Detection of toxin in the stool of a symptomatic patient is one of the most accurate methods for the diagnosis of these infections. Although important for epidemiology, the detection of just the organism may have limited value because the asymptomatic carriage rate may be high in hospitalized patients receiving antibiotics. Organisms that do not produce toxin are thought to be avirulent [12], and thus interpretation of a positive culture result alone is not possible. Toxin production by isolates is therefore a useful piece of extra evidence when attempting to make a definite diagnosis based on culture alone. The detection of toxin B by the tissue culture assay is considered to be the standard [13]. However, tissue culture techniques are not routinely available in most diagnostic laboratories. The technique may require an incubation period of up to 24 h (although strong positives can be seen after a few hours) and there are problems with the standardization of the method. The laboratory protocols vary in the dilutions and cell lines used, and in the interpretation of the assay endpoint [14]. Some authors accept a 50% CPE as a positive result, however, some consider a >90% CPE as the endpoint.
Yolken et al. [15] first developed and evaluated an EIA for the detection of C. difficile toxin in 1981. After this attempt, further developments have continued and a number of commercial diagnostic kits are now available for the detection of either toxin A or both toxins. In this study of strains in pure culture the tissue culture cytotoxicity assay was compared with two of these tests (a slide immunoassay for toxin A only and a microplate immunoassay for toxins A and B). The tests we used to detect the toxins from the culture supernatants of the isolated strains were found to react in the same way when detecting toxin from specimens (unpublished data). The principle of the microplate immunoassay is based on a monoclonal antibody to toxin A and the antigen–antibody complex is visualized by a precipitate, which is formed by blue latex coated with antibody. The sensitivity, specificity, PPV and NPV obtained for the toxin A test were 88.2%, 100.0%, 100.0%, 66.7%, respectively. Bentley et al. [16] used this test to detect toxin A in fecal specimens and reported the sensitivity and specificity values as 83.1% and 96.9%, respectively. Vanpoucke et al. [17] used the same test and found the sensitivity, specificity, PPV and NPV to be 89%, 83%, 71%, 94%, respectively, for the detection of toxin in stool samples. Our rates on pure cultures are concordant with the results of these studies, and it seems reasonable to use this test for the detection of toxin both in stool specimens and C. difficile culture supernatants. It has the advantage of being rapid (the result is ready after 30 min), extremely easy and one is able to work with one specimen at a time instead of collecting samples. The main disadvantage is that it will not detect toxin B produced by A– Bþ strains. However, the occurrence of this phenotype is uncommon and localized. We also used the toxin A þ B EIA test for the detection of both toxins. In our study the sensitivity, specificity, PPV and NPV were found to be 100.0%. In one multicenter study, this kit was used to detect the toxins in 1152 fecal specimens and the values obtained for the sensitivity ranged between 83.3 and 96.0%, for the specificity between 99.3 and 100% and for NPV between 90.0 and 99.5%. They found PPV to be 100% and the correlation as 94.9–99.5% when compared with the tissue culture assay [18]. Aldeen et al. [19] compared the same toxin A þ B test to cell culture cytotoxin assay in 1109 diarrhoeal stool samples. They got the sensitivity and specificity as 94.3%,
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
Yu¨ cesoy et al
99.3%, initially. However, after resolution of six discrepancies, the detected sensitivity, specificity, PPV and NPV were 94.5%, 100%, 100% and 98.8%, respectively. The agreement between toxin A þ B and tissue culture assay was 98.5%. Our results for pure cultures are in parallel with the above fecal toxin results. This test combines the advantages of an EIA method, such as rapidity, ease of use and sensitivity, with the capability of detecting both toxins. Also, EIAs have objective endpoints determined with an EIA reader. De Girolami et al. [8] compared visual and spectrophotometric readings and have reported good correlations. Our results are parallel to this finding. The reported sensitivity and specificity values for other available commercial EIA kits which can detect toxins A and B were between 79.6 and 96.2%, 93.5 and 100%, respectively, and PPV and NPV were between 86.7 and 100% and 95.7 and 99%, respectively [14,18, 20,21]. These results are similar to, or a little lower than, those obtained for the Techlab A/B test. Four of our strains appear to be toxin A– Bþ strains which were negative in toxin A EIA but positive in the toxin A/B test. It is possible that the sensitivity levels of the toxin A test we used may be low and the production of toxin A is missed. On the other hand, they might be true toxin A– Bþ strains that could have caused CDAD in the patients. Lyerly et al. [18] mentioned the presence of a toxin A– Bþ isolate from a patient with a clinical history consistent with C. difficile disease. In recent years toxin A– Bþ strains have been described in symptomatic adults in Japan and the USA [22,23]. Alfa et al. [24] reported that the toxin A– Bþ strain was responsible for a nosocomial outbreak of CDAD. Brazier et al. [25] also reported the same phenotype C. difficile strains from symptomatic patients. It has been suggested that these strains might have caused disease due to a more active toxin B [18]. Another group reported that it was possible either that toxin B alone was capable of causing diarrhoea or that the strain contained other factors capable of causing disease. Lyerly et al. [18] also found that two serogroup type F strains showed toxin A– Bþ phenotype in their study. This was in agreement with the findings of Depitre et al. [26]. The ribotypes of A– Bþ strains, which were reported by Brazier et al. [25], belonged to ribotype 17. Our apparent A– Bþ strains belonged to serogroups A and K (K, A8, A9). We also had a strain which belonged to
Measurement of C. difficile toxin 417
serogroup F which was found to produce both A and B toxins. Eight of the isolates were shown to be non-toxinproducers. This reflects the fact that there are nontoxigenic C. difficile strains which exist in the gastrointestinal tracts of hospitalized patients as normal colonizers. It is possible, but so far not proven, that non-toxigenic strains may cause disease depending on the predisposing factors of the host and the virulence of these strains. In other studies it is well recognized that a proportion of strains isolated from patients whose clinical course was consistent with C. difficile-associated disease were found to be non-producers of toxin, e.g. 3% (two of 63) in the study by Clabots et al. [27]. From the result of our study we can recommend the use of the Techlab toxin A/B enzyme immunoassay for the detection of toxin production by C. difficile isolates because of its high sensitivity, specificity, ease of use and its capability of detecting both A- and B-type toxins. However, the Oxoid, ClearviewTM system is so easy to use, and as it is designed for single samples, its usefulness should not be overlooked, especially if there is no history of Toxin A– Bþ strains in the locality. It is recommended, however, that isolates of C. difficile are regularly monitored to see which toxins they are producing. ACKNOWLEDGMENTS Presented as a poster at 11th European Congress of Clinical Microbiology and Infectious Diseases (1–4 April 2001, Turkey). REFERENCES 1. Murray PR, Rosenthal KS, Kobayashi GS, Pfaller MA. Clostridium. Medical Microbiology, 3rd edn. St Louis Missouri: Mosby Inc., 1998; 296–307. 2. Brazier JS, Borriello SP. Microbiology, epidemiology and diagnosis of Clostridium difficile infection. In: Aktories K, Wilkins TD, eds. Current Topics in Microbiology and Immunology Clostridium Difficile. Berlin: Springer, 2000; 1–33. 3. Sharp J, Poxton IR. The cell wall proteins of Clostridium difficile. FEMS Microbiol Letts 1988; 55: 99–104. 4. McCoubrey J, Poxton IR. Variation in the surface layer proteins of Clostridium difficile. FEMS Immunol Med Microbiol 2001; 31: 131–5. 5. Poxton IR, McCoubrey J, Blair G. The pathogenicity of Clostridium difficile. Clin Microbiol Infect 2001; 7: 421–7.
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418
418 Clinical Microbiology and Infection, Volume 8 Number 7, July 2002 6. Macgowan AP, Brown I, Feeney R et al. Clostridium difficile-associated diarrhoea and length of hospital stay (letter). J Hosp Infect 1995; 31: 241–4. 7. Riley TV, O’Neil GL, Bowman RA, Golledge CL. Clostridium difficile-associated diarrhoea: epidemiological data from Western Australia. Epidemiol Infect 1994; 113: 13–20. 8. De Girolami PC, Hanff PA, Eichelberger K et al. Multicenter evaluation of a new enzyme immunoassay for detection of Clostridium difficile enterotoxin A. J Clin Microbiol 1992; 30: 1085–8. 9. Poxton IR, Brown R, Fraser AG et al. Enteropathogenic clostridia and Clostridium botulinum. In: Collee JG, Fraser AG, Marmion BP, Simmons A, eds. Mackie & McCartney’s Practical Medical Microbiology, 14th edn. Edinburgh: Churchill Livingstone, 1996; 546. 10. Schue´ V, Green GA, Monteil H. Comparison of the ToxA test with cytotoxicity assay and culture for the detection of Clostridium difficile-associated diarrhoeal disease. J Med Microbiol 1994; 41: 316–18. 11. Finegold SM. Clinical considerations in the diagnosis of antimicrobial agent-associated gastroenteritis. Diagn Microbiol Infect Dis 1986; 4: 87S–91S. 12. Barlett JG. Clostridium difficile: clinical considerations. Rev Infect Dis 1990; 12: S243–S251. 13. Jacobs J, Rudensky B, Dresner J et al. Comparison of four laboratory tests for diagnosis of Clostridium difficile-associated diarrhoea. Eur J Clin Microbiol Infect Dis 1996; 15: 561–6. 14. Merz CS, Kramer C, Forman M et al. Comparison of four commercially available rapid enzyme immunoassays with cytotoxin assay for detection of Clostridium difficile toxin (s) from stool specimens. J Clin Microbiol 1994; 32: 1142–7. 15. Yolken RH, Whitcomb LS, Marien G. Enzyme immunoassay for the detection of Clostridium difficile antigen. J Infect Dis 1981; 144: 378. 16. Bentley AH, Patel NB, Sidorczuk M et al. Multicentre evaluation of a commercial test for the rapid diagnosis of Clostridium difficile-mediated antibioticassociated diarrhoea. Eur J Clin Microbiol Infect Dis 1998; 17: 788–90. 17. Vanpoucke H, De Baere T, Claeys G, Vaneechoutte M, Verschraegen G. Evaluation of six commercial
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
assays for the rapid detection of Clostridium difficile toxin and/or antigen in stool specimens. Clin Microbiol Infect 2001; 7: 55–64. Lyerly DM, Neville LM, Evans DT et al. Multicenter evaluation of the Clostridium difficile TOX A/B Test. J Clin Microbiol 1998; 36: 184–90. Aldeen WE, Binham M, Aiderzada A, Kucera J, Jense S, Carroll KC. Comparison of the TOX A/B test to a cell culture cytotoxicity assay for the detection of Clostridium difficile in stools. Diagn Microbiol Infect Dis 2000; 36: 211–13. Barbut F, Kajzer C, Planas N, Petit JC. Comparison of three enzyme immunoassays, a cytotoxicity assay, and toxigenic culture for diagnosis of Clostridium difficile-associated diarrhea. J Clin Microbiol 1993; 31: 963–7. Arrow SA, Croese L, Bowman RA, Riley TV. Evaluation of three commercial enzyme immunoassay kits for detecting faecal Clostridium difficile toxins. J Clin Pathol 1994; 47: 954–6. Kato H, Kato N, Fukui K, Ohara A, Watanabe K. High prevalence of toxin A negative/toxin B positive Clostridium difficile strains among adult inpatients. Clin Microbiol Infect 1997; 3: 220–2. Sambol S, Gerding D, Merrigan M, Lyerly D, Johnson S. Severe truncation of the toxin A gene in a pathogenic Clostridium difficile (CD) strain not detectable by toxin A immunoassay. Clin Infect Dis 1998; 27: 946–9. Alfa MJ, Kabani A, Lyerly D et al. Characterization of a toxin A-negative, toxin B-positive strain of Clostridium difficile responsible for a nosocomial outbreak of Clostridium difficile-associated diarrhea. J Clin Microbiol 2000; 38: 2706–14. Brazier JS, Stubbs SL, Duerden BI. Prevalence of toxin A negative /B positive Clostridium difficile strains. J Hosp Infect 1999; 42: 248–9. Depitre C, Delmee M, Avesani V et al. Serogroup F strains of Clostridium difficile produce toxin B but not toxin A. J Med Microbiol 1993; 38: 434–41. Clabots CR, Johnson S, Olson MM, Peterson LR, Gerding DN. Acquisition of Clostridium difficile by hospitalized patients: evidence for colonized new admissions as a source of infection. J Infect Dis 1992; 166: 561–7.
ß 2002 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 8, 413–418