Logo of jcmPermissionsJournals.ASM.orgJournalJCM ArticleJournal InfoAuthorsReviewers
J Clin Microbiol. 1999 Dec; 37(12): 4012–4019.

Predominant Staphylococcus aureus Isolated from Antibiotic-Associated Diarrhea Is Clinically Relevant and Produces Enterotoxin A and the Bicomponent Toxin LukE-LukD


Staphylococcus aureus was isolated as the predominant or only isolate from cultures of stools of 60 patients over 2 years in a university hospital, leading to the collection of 114 isolates. Diarrhea was observed in 90% of the patients. Ninety-eight percent of the patients had received antibiotics in the month before the diarrhea. Ninety-two percent of the S. aureus isolates were methicillin resistant. S. aureus was encountered with antibiotic-associated diarrhea among 47 quite elderly patients affected or not affected by a gastrointestinal disease. Among the antimicrobial treatments, cessation of the previous therapy when possible or rapid application of oral vancomycin therapy was the most appropriate. Analysis of total DNA by pulsed-field gel electrophoresis revealed 27 different SmaI pulsotypes distributed in 15 clusters. The pulsotypes never differed for related isolates from a single patient, even if they originated from patients with bacteremia. S. aureus was not isolated as the predominant isolate in cultures of stools of 57 patients who received an antimicrobial treatment for more than 5 days without diarrhea. Occurence of production of both enterotoxin A and the bicomponent leucotoxin LukE-LukD by the S. aureus isolates was significantly different from that by random isolates. The results strongly suggest that when predominant in stool samples, S. aureus should be considered a possible etiologic agent for some cases of antibiotic-associated diarrhea.

From 1955 to 1970, Staphylococcus aureus was suspected as a cause of antibiotic-associated diarrhea (AAD) (30, 36). Investigations established a frequent relation between pseudomembranous colitis and administration of tetracycline with the isolation at high yields of S. aureus from stools (21, 32, 36). Altemeier et al. (2) studied patients with S. aureus-associated enteritis; the majority of the patients had undergone gastrointestinal surgery. One-third of the patients had undergone no operation before development of diarrhea. S. aureus was the only species isolated from stools on culture and was resistant to tetracycline. Hinton et al. (17) emphasized the emergence of staphylococci in stools as important pathogens responsible for AAD for three reasons: first, the close time relation between the onset of fulminating diarrhea and the occurrence of a heavy intestinal staphylococcal flora; second, the time relation between the defervescence from the disease and the decrease in abundance of staphylococci in stools after an appropriate antibiotic therapy; and third, the relation between the intestinal lesion and the location of the organism in patients who died (17, 30). Therefore, isolation of S. aureus as the predominant organism in the fecal flora should lead to treatment. S. aureus strains previously incriminated in enterocolitis (36) were reported (6) to be secondary enterotoxin A (SEA) producers.

In the 1970s, pseudomembranous enterocolitis was observed in patients treated with clindamycin and other antibiotics (4, 14). Isolation and identification of Clostridium difficile, which produces an enterotoxin (toxin A) and a cytotoxin (toxin B), favored the hypothesis that this bacterium was the cause of pseudomembranous colitis associated with AAD (3, 4, 14). The role of S. aureus was neglected and became controversial. However, methicillin-resistant and gentamicin-resistant S. aureus strains were recognized as the causes of outbreaks of enteritis in hospitalized patients who were treated with extended-spectrum antibiotics (25, 33) and in whom mild to fatal illnesses were observed.

The aim of the work described here was to study the clinical significance of S. aureus strains isolated for 2 years as the predominant bacteria from stools in a routine laboratory and to study the toxins production in comparison to those produced by isolates from other sites of infection. As LukE-LukD seemed to be frequently present in strains isolated from stools (15), the members of this bicomponent leucotoxin family and the staphylococcal enterotoxins were typed.


Clinical data. (i) Group A patients.

S. aureus was isolated as the only isolate or as the predominant isolate in cultures of stools of 60 patients whose hospital charts were reviewed. A standardized questionnaire was completed and included age, sex, hospital unit, reasons for admission, medical history, presence and duration of clinical diarrhea, additional symptoms, and a list of other samples that had been collected. Receipt of antimicrobial treatment or a gastrointestinal tract decontamination procedure within 1 month before the occurrence of diarrhea was recorded. The therapeutic outcomes in terms of the evolution of diarrhea and death as a direct consequence of this complaint were recorded.

Clinical diarrhea was defined as three or more liquid or soft stools a day for at least 2 days. Resolution of diarrhea was defined as normal stool frequency and normal consistency for at least 3 consecutive days.

(ii) Group B patients.

Stools from 57 different patients who originated from two medical care units and who had received antibiotics for at least 5 days without diarrhea were systematically processed from March to June 1998. Samples were inoculated onto a 5% (vol/vol) sheep blood agar plate and onto Chapman-Stone medium for selective isolation of staphylococci (Difco, Detroit, Mich.). The presence and relative abundance (on blood agar) of S. aureus isolates were recorded.

(iii) Random isolates.

The toxin production study included 146 consecutive strains that were collected during two periods (July 1996 and March 1998) and that originated from different sites of infection, including 11 from blood, 38 from skin, 51 from the throat, mouth, and nose, 18 from the urinary tract, 23 from abcesses and infected materials, and 5 from stools.

Bacteriologic analyses. (i) S. aureus isolates.

All stool samples referred by clinicians from patients hospitalized in Strasbourg University Hospital, Strasbourg, France, between July 1996 and June 1998 were evaluated. Stools and other samples (urinary tract, slough, vagina, and nose) were inoculated onto 5% sheep blood agar plates, and the plates were incubated for 24 h at 37°C. S. aureus was identified in the routine laboratory by conventional methods (19). Only cultures with abundant S. aureus (approximately more than 90% of the total CFU) were used in the study. Blood specimens were processed under aerobic culture conditions (BACTEC Plus+ Aerobic/F* culture vial; Becton Dickinson).

Antibiograms for the isolates were obtained by the disk diffusion method. Isolates were inoculated according to the recommendations of the National Committee for Clinical Laboratory Standards (28). The following antibiotics were tested: amikacin, amoxicillin, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, neomycin, nitrofurantoin, rifampin, tobramycin, trimethoprim-sulfamethoxazole, and vancomycin. Evaluation of methicillin resistance was performed by plating the strains on buffered Mueller-Hinton agar with 2% (wt/vol) NaCl at 37°C for 24 h of incubation with a disk of oxacillin (5 μg; Sanofi-Diagnostic Pasteur, Marnes la Coquette, France), according to the standards of the French Committee for Antibiogram (34).

Tests for detection of enteric pathogens, such as Salmonella, Shigella, Campylobacter, and Yersinia, were done by conventional methods (1, 7, 27).

(ii) C. difficile isolates.

C. difficile was identified after plating stools on George-selective agar medium (13) containing cefoxitin (10 μg/ml) and cycloserine (250 μg/ml) and incubating the plates at 37°C for 48 h under strict anaerobic conditions. Colonies suspected as being C. difficile were confirmed by biochemical tests. Both toxins A and B were detected in fresh stool specimens. The toxin A test was determined by using the C. difficile Toxin A Test (Oxoid, Hampshire, England) as recommended by the manufacturer. The toxin B test was based on detection of a cytotoxic activity in a McCoy cell culture (23). Detection of C. difficile only was not considered a substitute for the one-toxin assay for the bacterial diagnosis of C. difficile-associated postantibiotic intestinal disorders (37).

Pulsed-field gel electrophoresis (PFGE).

S. aureus strains at midexponential growth in 25 ml of TY (BioTrypcase, 1.6% [wt/vol]; yeast extract, 1% [wt/vol]; NaCl, 0.5% [wt/vol]) were embedded in agarose plugs as described previously (31). After lysis of the plugs for 18 h at 56°C, the plugs were washed and stored at 4°C in 10 mM Tris-HCl–1 mM EDTA (pH 8.0). Macrorestriction of the DNA was accomplished with SmaI (New England Biolabs, Beverly, Mass.). Plugs (agarose plugs, 2.5 by 5.0 by 1 mm) were equilibrated in 300 μl of restriction enzyme buffer for 30 min at 0°C. They were incubated for 4 h at room temperature in 60 μl with 10 U of enzyme. Electrophoresis was performed at 12°C with a Beckman Geneline Transverse Alternating Field Electrophoresis (TAFE) system at 150 mA in 0.5× TAFE system buffer (20× TAFE system buffer is 0.2 M Tris, 10 mM free acid EDTA, and 87 mM CH3COOH [pH 8.2]) in a 1% (wt/vol) agarose gel. Electrophoresis of SmaI DNA fragments began with 2-s pulses for 1 h, followed by 14-s pulses for 1 h, 12-s pulses for 1.5 h, 10-s pulses for 2.5 h, 8-s pulses for 6 h, and 6-s pulses for 6 h. The bacteriophage lambda PFG Marker (New England Biolabs) was used as a molecular marker. The gels were stained with ethidium bromide and were photographed under UV transillumination. Pulsotypes were compared and were classified in a dendrogram by using the Dice coefficient and the unweighted pair group method with arithmetic mean clustering provided by Molecular Analyst (version 1.5) and Fingerprinting (version 1.12) software (Bio-Rad, Ivry sur Seine, France).

Toxin determinations. (i) Enterotoxins A, B, C, and D and TSST-1.

S. aureus strains were detected as recommended by the manufacturer (Oxoid) by detection of enterotoxins A, B, C, and D, and toxin shock syndrome toxin 1 (TSST-1) by semiquantitative tests with reversed passive latex agglutination (RPLA) toxin detection kits (SET-RPLA and TST-RPLA [Oxoid], respectively). These tests were reported to be sensitive and specific, although large amounts of enterotoxin E seemed to be detected as low levels of enterotoxin A (8).

(ii) Enterotoxin E.

The total DNAs of the isolates were digested by EcoRI restriction of total DNA. After agarose gel electrophoresis of the restricted DNAs, entE was probed by Southern blotting (35) with 5′-32P-labelled specific oligonucleotide, position 636–5′-TTCTGAAGGGTCCACGGTAAGTTAT-3′–position 661 (11).

(iii) Bicomponent leucotoxins and exfoliative toxins.

Components of different leucotoxins (Panton-Valentine leucocidin, LukE-LukD, LukM-LukF′-PV, gamma-hemolysin, and exfoliative toxins A and B) were detected by radial gel immunodiffusion with component-specific rabbit polyclonal and affinity-purified antibodies prepared as described previously (15). Culture supernatants were tested after 18 h of growth in YCP (yeast extract, Casamino Acids, pyruvate) medium (15).


Clinical results. (i) Group A patients.

During the study period, a total of 60 patients were enrolled in group A. Forty-one patients (68.3%) were admitted to medical units, 10 (16.7%) were in surgical units, and 9 (15%) were in pediatric wards. The mean age was 60.0 ± 29.0 years (age range, 1 month to 95 years). The male/female ratio was 1.2:1. One hundred fourteen S. aureus isolates from these patients were evaluated: 81 were from stools, 28 were from blood, and 5 were from other samples (urinary tract, slough, vagina, and nose). During the same period, stools of 3,437 different patients were examined. Culture and/or toxin assays for C. difficile were positive for 460 patients. For 247 of these patients, a diagnosis of C. difficile-associated AAD was made. The distribution of the C. difficile isolates according to age (mean age, 63.7 ± 24.3 years; age range, 3 months to 99 years) and medical units was similar to that for S. aureus isolates.

Among the 60 patients in group A, 6 had no clinical diarrhea. The mean age of these patients was 1.8 ± 3.3 years (age range, 1 month to 9 years). All were hospitalized in pediatric units and presented no more than two watery stools. These patients had normal body temperatures, but two presented with moderate elevations in C-reactive protein levels.

Fifty-four (90%) patients had clinical diarrhea, from most of whom it was green and watery. The patients had a mean of 4.1 liquid or soft stools per day (range, 3 to 10) liquid or soft stools). The mean age of the patients was 66.5 ± 22.6 years (age range, 1 month to 95 years). Only 3 of 54 patients did not receive antibiotics in the month before the onset of diarrhea. For the first of the latter patients without a specific medical background, Campylobacter jejuni was found in the stools in association with S. aureus, which was predominant. For the second patient, a 3-month-old, an enteropathogenic Escherichia coli strain was found. The third patient was a premature infant who presented with an elevation in the C-reactive protein level and from whom a methicillin-susceptible S. aureus strain was isolated. A diagnosis of colitis was made, and the patient was treated with intravenous vancomycin. C. difficile was isolated from the stools of four patients. The tests for enterotoxin A and cytotoxin B were also positive. Patients were treated with metronidazole, and the diarrhea disappeared rapidly, but the tests for C. difficile remained positive. After a 3-week remission, the diarrhea reappeared, and both C. difficile and S. aureus were still isolated from the stools. The test for toxin B was positive, but at a lower level. For three of these patients, oral treatment with vancomycin, 500 mg every 8 h, was prescribed, and the diarrhea disappeared within 3 days but the stools remained positive for C. difficile and cytotoxin B only. For the fourth patient, the remission of diarrhea was longer (2 months); he was then treated with metronidazole but soon died.

The 47 remaining patients presented with diarrhea, and no enteropathogens other than S. aureus were isolated as the predominant isolate in culture or were associated with AAD. The mean age of these patients was 68.6 ± 18.2 years. Table Table11 summarizes the different antibiotics prescribed for these 47 patients. Diarrhea began an average of 11.7 ± 5.4 days (range, 4 to 30 days) after administration of the first antibiotic dose. For 30 of 47 (63.8%) patients, combination therapy with two or three antibiotics was administered. Thirty-four of the 47 (72.3%) patients received fluoroquinolones: 11 as monotherapy and 23 in combination with other agents.

Antimicrobial therapy administered to 47 patients within the month before the onset of diarrheaa

The patients with AAD were then distinguished according to their underlying diseases. Twenty-nine patients had not undergone abdominal surgery and had no digestive disease. These patients were admitted for cardiac pulmonary decompensation, cerebrovascular accident, or general deterioration. All of these patients presented or had presented with an infection: pneumonia, urinary tract infection, or an infection of undetermined origin. These infections were assumed to be the reason for hospitalization or were considered nosocomial infections. These patients were hospitalized for 22.2 ± 33.9 days (range, 0 to 170 days) before the onset of diarrhea. Eighteen other patients had a particular underlying disease with digestive manifestations: one had pancreatitis, one had cirrhosis, three had digestive carcinoma with recent surgery, four had digestive carcinoma without recent surgery, one had multiple endocrine neoplasia, three had surgery for malformation, one had diverticulosis, two had liver graft, one had bone marrow graft, and one had graft-versus-host disease (digestive form). Among the latter group of 18 patients, 7 patients had received an immunosuppressive therapy before the onset of diarrhea: 4 had received chemotherapy for neoplastic disease and three had received chemotherapy for grafts. Two patients, one with multiple neoplasia and one with graft-versus-host disease, had frequent watery stools, but the stools became diarrheic with isolation of S. aureus. These 18 patients were hospitalized for 15.9 ± 10.1 days (range, 0 to 37 days) before the onset of diarrhea.

The patients had normal body temperature just before the beginning of the diarrhea. The clinical manifestations of diarrhea in patients were a rise in body temperature for 28 of 47 (59.6%) patients and a rise in biological parameters of infection for 26 of 47 (55.2%) patients. Finally, 11 (23.4%) patients did not present any general or biological signs of infection except the diarrhea.

Treatment and evolution of disease in group A patients.

The outcomes of the diarrhea for the 47 patients with AAD were considered according to the antimicrobial treatment. For eight patients, oral vancomycin was applied. For six of these patients, the first dose of glycopeptide was administered 5.1 days (range, 3 to 8 days) after the beginning of diarrhea, which disappeared within 4.1 ± 1.8 days (range, 2 to 8 days). For one patient, the same treatment began after 31 days, and then the diarrhea disappeared rapidly. The remaining patient, an 81-year-old, had three recurrent episodes of diarrhea. A first diarrhea remission of 15 days was obtained with discontinuation of the previous antibiotic. Treatment with oral vancomycin was induced and achieved a further remission of 10 days, but the diarrhea developed again and stool, vaginal, nasal, and slough specimens became positive for S. aureus. The patient was again treated with oral vancomycin, but the patient presented with a general deterioration dominated by S. aureus-positive diarrhea and died.

Of the 47 patients, cultures of blood for 10 patients were positive for S. aureus. For one patient only one blood sample was positive by culture, and the culture was assumed to be contaminated with S. aureus. For the other nine patients, two or more blood samples were positive for S. aureus by culture. These patients had intravascular catheters or central or peripheral lines, and none of them was infected. The nine patients were treated with parenteral vancomycin when blood cultures became positive. Three of the patients died after development of an advanced shock when the diarrhea worsened, despite antimicrobial therapy. For these patients, the bacteremia arose after a duration of diarrhea of 32.3 ± 23.6 days (range, 10 to 65 days). For one of them, an ischemic colitis was suspected and a partial colectomy was performed. However, the diarrhea continued and S. aureus was isolated in pure culture, but the patient was negative for C. difficile and toxins. Histology revealed a pseudomembranous enterocolitis. Within 16 days following surgery this patient developed a septic shock positive for S. aureus and died, despite the antimicrobial therapy that had been prescribed. The six remaining patients survived, and the diarrhea disappeared with parenteral vancomycin given 4.2 ± 3.9 days (range, 1 to 11 days) after the beginning of the diarrhea. Eleven other patients without positive blood cultures received intravenous vancomycin because of signs of infections. For seven patients the diarrhea stopped within 4.8 ± 1.2 days, but four patients died.

For nine patients, the measure taken was discontinuation of antibiotics 3.7 days (range, 0 to 11 days) after the beginning of diarrhea, which stopped within 4.5 days (range 2 to 7 days). The patients had lower signs of infection; 5 of the 10 patients had no rise in body temperature. Diarrhea disappeared from all patients in this group, with a rapid improvement in their general condition.

For the remaining 10 patients, including the patient with blood contamination, the previous antimicrobial treatment was replaced by another drug but not vancomycin. The diarrhea stopped for two patients. Eight patients died, and the diarrhea was an important factor in the changes in their health.

Group B patients.

The mean age of the 57 group B patients was 75.8 ± 11.2 years. They received antibiotics for at least 5 days before stool sampling. Two patients had digestive surgery, and one patient had amputation surgery. Seven patients had digestive diseases but had not undergone surgery. Twenty patients had complicated diabetes. All 57 patients presented with infectious diseases, which affected, in order of frequency, the lungs, urinary and digestive tracts, and skin. Therefore, the medical background for the group B patients was similar to that for the group A patients. Table Table11 summarizes the antibiotics prescribed. Eighteen (31.6%) patients received amoxicillin and 22 (38.6%) received fluoroquinolones. Colonies of S. aureus were found on blood agar plates and on Chapman medium plates for two patients. For one of these patients, an Enterococcus sp. was predominant (>90% of the total CFU), while for the second patient, S. aureus seemed to be present in amounts equal to the amounts of Enterococcus. For three patients, 1 to 10 CFU of S. aureus was isolated only on Chapman medium. For the latter three patients and the 54 remaining patients, the predominant isolated floras were an Enterococcus sp., gram-negative rods, and yeast and coagulase-negative Staphylococcus.

Biological results. (i) Antimicrobial susceptibilities of S. aureus strains from group A patients.

Isolates from 58 patients (97%) were resistant to amoxicillin and 55 (92%) were methicillin-resistant S. aureus (MRSA), but all were susceptible to vancomycin (100%). They were susceptible to trimethoprim-sulfamethoxazole (58 of 60; 97%), nitrofurantoin (58 of 60; 97%), rifampin (49 of 60; 82%), and chloramphenicol (53 of 60; 88%). Susceptibility to the other antibiotics was variable: 59 (82%) were resistant to ciprofloxacin, 48 (80%) were resistant to kanamycin, 48 (80%) were resistant to tobramycin, 15 (25%) were resistant to gentamicin, 45 (75%) were resistant to neomycin, and 48 (80%) were resistant to amikacin.

(ii) Differentiation of the S. aureus strains by DNA pulsotyping.

The total DNAs of all 114 S. aureus isolates from the patients were analyzed by SmaI macrorestriction and PFGE. Figure Figure11 provides a schematic representation of the 27 different pulsotypes, 26 of which were for isolates from stools. The number of isolates corresponding to each pulsotype was noted. Taking into consideration the profiles obtained for three different isolates retested on a simple or different gel, standard deviations of ±5% were recorded for various DNA fragments. For example, PFGE patterns 3, 4, and 5 differed only in the locations of single DNA fragments of 180, 197, and 165 kb, respectively. PFGE profiles 10 and 13, which differed only in the locations of single DNA fragments of 450 and 420 kb, respectively, were also considered to be similar, even though the other fragments were similarly separated. PFGE fingerprints 20 and 21 also differed because of an additional fragment of 420 kb in pattern 21. The pulsotypes were distributed in 15 clusters, with the cutoff for a cluster being 80% similarity.

FIG. 1
Schematic representation and computer-aided dendrogram analysis of the 27 SmaI pulsotypes of S. aureus DNA originating from the 60 patients in group A. Column a, pulsotype numbers; column b, number of patients with isolates corresponding to each pulsotype; ...

Considering the distribution of isolates from the 60 patients into 27 PFGE fingerprints, it appeared that isolates of pulsotypes 13, 10, 16, and 17 were the most frequently encountered (10, 7, 7, and 7 isolates from different patients, respectively) (Fig. (Fig.1).1). In all cases, when S. aureus was isolated several times from a given patient, all related stool isolates had the same pattern (Fig. (Fig.2).2). For one patient, with a clinical history of diarrhea for 3 months, the three isolates from stools and isolates originating from the nose and the urinary tract were pulsotype 16, whereas isolates from a slough and the vagina were pulsotype 17. For each patient with septicemia, isolates from stools and blood cultures were of the same pulsotype (Fig. (Fig.2).2). PFGE profiles 14, 21, and 22 were observed for isolates from single patients whose stools and blood were sampled. The isolate from one positive blood culture considered to be contaminated with S. aureus and those from stools of a single patient were of pulsotypes 24 and 10, respectively.

FIG. 2
DNA fingerprint analysis of the SmaI-restricted total DNAs from 22 S. aureus isolates by TAFE system PFGE. Molecular masses (in kilobases) were obtained from a bacteriophage lambda PFG Marker (New England Biolabs) (lanes T). The samples (a) and the pulsotype ...

(iii) Toxin production.

The results of toxin production are summarized in Table Table2.2. The newly discovered bicomponent leucotoxin LukE-LukD was produced by 93.6% of the 47 representative isolates from patients with S. aureus-associated AAD, 76.9% of the 13 S. aureus isolates not associated with AAD, and 33.6% of the 146 random isolates. SEA enterotoxin A was produced by 85.1, 46.1, and 54.1% of the three groups of isolates cited above, respectively. Both groups of toxins were produced by 80.9, 22.6, and 30.8% of the isolates in the same groups, respectively.

Production of different toxins by S. aureus isolatesa

Among the six isolates from young patients without diarrhea in whom S. aureus was the predominant organism, three produced LukE-LukD, two produced SEA, and two produced staphylococcal enterotoxin D but no isolate produced both SEA and LukE-LukD. The isolate from the young patient presenting with enterocolitis with diarrhea produced LukE-LukD but none of the staphylococcal enterotoxins.

All 47 isolates were negative for production of enterotoxins B, C, D, E, TSST-1, and Panton-Valentine leucocidin. Only one isolate produced the exfoliative toxin A. Among the 146 random S. aureus isolates, 6 isolates produced enterotoxin B, 8 produced enterotoxin C, 10 produced enterotoxin D, 15 produced TSST-1, 2 produced Panton-Valentine leucocidin, and 2 produced one of the exfoliative toxins. The gamma-hemolysin was produced by all the isolates studied; LukM-LukF′-PV (15) was never produced. Of the five isolates from patients who did not have diarrhea and who received antibiotics for at least 5 days, all produced SEA; only two isolates produced LukE-LukD.


With the identification of C. difficile and its toxins, the etiologic role of S. aureus in AAD became controversial (4, 14). The arguments developed were (i) the increase in the carriage rate of S. aureus following admission to the hospital (29), (ii) the lack of clear evidence of the association of S. aureus with pseudomembranous colitis, (iii) culture conditions more favorable to the growth of S. aureus (5), and (iv) the efficacy of vancomycin, which became apparent in 1959, against both S. aureus and C. difficile, although C. difficile was not identified at that time (5). However, there is some evidence against the etiologic role of S. aureus and C. difficile as well. As opposed to S. aureus, a satisfactory animal model of colitis evidenced the role of C. difficile in these infections (9). The frequency with which each bacterium was associated with AAD was not assessed. The rate of lack of recovery of C. difficile despite detection of toxin in stools is generally estimated to be no more than 5%. Recent studies indicate that when S. aureus is the predominant isolate in a culture it is a possible cause of AAD (20, 25, 33).

Over 2 years, 60 patients for whom S. aureus was isolated from stool samples as the predominant or only isolate were considered. Fifty-four patients (mean age, 66.5 years) had diarrhea, whereas the others were young infants without diarrhea. For 6 of the 54 patients, the role of S. aureus could not be clearly established. One young patient with diarrhea from whom S. aureus was recovered did not receive antibiotics. Finally, 47 patients were assumed to have S. aureus associated AAD. The frequency of S. aureus-associated AAD was a fifth of that of C. difficile-associated AAD and was therefore significant. Most patients with S. aureus-associated AAD presented with signs of infection, whereas this was more seldom the case for those with C. difficile-caused AAD (4, 14).

To fulfill the second Koch's postulate, stools from 57 patients (group B) who did not have diarrhea after at least 5 days of antimicrobial treatment in two medical units also affected by AAD and who were examined over 4 months during the 2 years of the study. It appeared that an Enterococcus sp. was the predominant organism, and S. aureus was isolated from only five patients and was obtained on a selective culture medium (Chapman medium) for three of the patients. The frequency of S. aureus isolation from stools differed for group A and B patients, although their medical backgrounds were similar. However, previous antimicrobial therapy also differed for the two groups, since fluoroquinolones alone or in combination with another antibiotic were mostly administered to group A patients (P < 0.001), whereas amoxicillin was more frequently administered to group B patients (P < 0.02). The reason why fluoroquinolones may select for MRSA over bacteria like Enterococcus spp. remains unexplained. The rate of carriage of S. aureus increases with the length of hospitalization (38). However, the predominance of S. aureus in stools is probably an essential condition for induction of the disease, as is the case for C. difficile. The pathologic yields of abundance of S. aureus and C. difficile in stools remain to be defined.

Discontinuation of antibiotics seemed to be a good choice for resolution of the S. aureus-associated AAD when patients presented with few clinical signs of infection. Otherwise, oral vancomycin applied rapidly after the beginning of AAD was satisfactory. Intravenous vancomycin was more or less efficient, in accordance with previous data (20, 25, 33). For every patient in the present study, S. aureus was no longer isolated after relapse of diarrhea consecutive to a dedicated treatment.

For the six young patients, the rate of occurrence of S. aureus in stools was generally frequent, especially for patients fed nasogastrically (4, 16), and the significance of these isolates rarely seemed to be pathogenic. The absence of diarrhea may be due to the absence of the association of SEA and LukE-LukD or, as for the C. difficile toxins (4), to the nonsusceptibility of young children. In such patients, cases of diarrhea due to S. aureus were reported in a previous study (16).

The frequency of 91% MRSA strains among the isolates from 60 group A patients (97% for the isolates from the 47 patients with AAD) with S. aureus in stools differed from the 40% frequency of MRSA found in the same hospital (22). The S. aureus strains originating from stools in this study appeared to be polymorphic when analyzed by PFGE, since 26 different pulsotypes were distinguished, although 15 clusters were evidenced, which favors a nonepidemic origin. However, pulsotype 13 corresponded to pulsotype 1 found a few years ago for nosocomial MRSA strains in the same hospital (31). Eleven patients were infected with such an isolate. Therefore, strains associated with AAD cannot be easily defined as being responsible for hospital-acquired infections or being of an endogenous origin. In fact, little is known about (i) the presence at very low yields and frequency of S. aureus in the gastrointestinal flora, (ii) the selective pressure of feeding on this ecology for the selection of resistant strains, (iii) the possibility of several types of strains in the fecal flora, and (iv) the more or less rapid growth of contaminating MRSA in antibiotic-treated gastrointestinal fluids. The spread of S. aureus strains from patients with AAD must be considered a possible and important source of nosocomial infections that require specific hygienic measures in hospitals. Moreover, PFGE analysis revealed pulsotypes which never differed for isolates from a single patient, including those from patients with bacteremia in association with AAD. These data strongly suggest that S. aureus-associated AAD may constitute an entry to septicemia.

The production of LukE-LukD and SEA was significantly different between the S. aureus strains associated with AAD and the group of random isolates (P < 0.001; χ2 = 51.36 and 14.45, respectively). The 44 representative isolates positive for LukE-LukD among the S. aureus isolates associated with AAD were polymorphic, with 18 different pulsotypes distributed in 10 clusters. Most of these isolates were MRSA (97%), but only 24 of the 49 random isolates that produced LukE-LukD were MRSA. Production of this toxin was not linked to methicillin resistance.

The frequency of production of SEA in the random group of isolates was comparable to the results in the literature, despite variability in this frequency (10, 24, 26). The general occurrence of TSST-1 production by S. aureus isolates from clinical samples appeared to be quite frequent (10 to 12%), but data were not significant for isolates originating from feces (18). A comparable frequency was observed for random isolates in this study, but because of the absence of such isolates from stools, the frequency cannot be assessed. The production of both SEA and LukE-LukD was significantly different among the 47 isolates associated with AAD and the random group of isolates (P < 0.001; χ2 = 51.87). The frequencies of the two toxins were not significantly different from the observed frequencies of association (Z test, P < 0.05; Z = 0.122 and 1.063). There is no genetic link between the two toxins.

In four other patients, the presence of S. aureus and diarrhea disappeared after oral vancomycin treatment, while C. difficile and cytotoxin B were still detected. The patients concerned were treated with metronidazole, with relapses in 3 to 6 weeks. When the diarrhea appeared again, S. aureus and C. difficile and its toxins were still isolated, although at lower yields than in the first event. Three of these patients were then treated with oral vancomycin. The diarrhea disappeared and S. aureus was no longer isolated, whereas they remained positive for C. difficile and its toxins. The last patient was treated with metronidazole, and the outcome was fatal, with diarrhea as an important factor. These four S. aureus isolates produced LukE-LukD, and three produced SEA. A recurrent C. difficile-associated diarrhea (12) and a complication of diarrhea caused by S. aureus were also envisaged. The particular profile of toxin production may be an argument favoring a staphylococcal etiology.

In conclusion, S. aureus infection in association with AAD exists and must be taken into consideration, especially for relatively older patients, even though it is less frequent than AAD caused by C. difficile. Rapid and specific care is recommended for such patients and to prevent the spread of MRSA. Production of both SEA and LukE-LukD is associated with these isolates. An attempt is being made to develop experimental models.


We are grateful to the staff of Strasbourg University Hospital for cooperation in the collection of clinical data. Thanks are given to N. Boord for reading of the English. We greatly appreciate the technical assistance of B. Muller and D. Keller and the help of O. Meunier and J.-M. Rousée for the computer analysis of the PFGE fingerprints.

This work was supported by grant EA1318 from “Direction de la Recherche et des Etudes Doctorales.”


1. Aleksic S, Bockemühl J. Yersinia and other Enterobacteriacea. In: Murray P R, Baron E J, Pfaller M A, Tenover F C, Yolken R H, editors. Manual of clinical microbiology. 7th ed. Washington, D.C: ASM Press; 1999. pp. 483–496.
2. Altemeier W A, Hummel R P, Hill E O. Staphylococcal enterocolitis following antibiotic therapy. Ann Surg. 1963;157:847–858. [PMC free article] [PubMed]
3. Barbut F, Corthier G, Charpak Y, Cerf M, Monteil H, Fosse T, Trévoux A, De Barbeyrac B, Boussougant Y, Tigaud S, Tytgat F, Sédallian A, Duborgel S, Collignon A, Le Guern M, Bernasconi P, Petit J. Prevalence and pathogenicity of Clostridium difficile in hospitalized patients—a French multicenter study. Arch Intern Med. 1996;156:1449–1454. [PubMed]
4. Bartlett J G. Clostridium difficile: history of its role as an enteric pathogen and the current state of knowledge about the organism. Clin Infect Dis. 1994;18:S265–S272. [PubMed]
5. Bartlett J G, Taylor N S, Chang T W, Dzink J A. Clinical and laboratory observations in Clostridium difficile colitis. Am J Clin Nutr. 1980;33:2521–2526. [PubMed]
6. Bergdoll M S. Enterotoxins. In: Easmon C S F, Adlam C, editors. Staphylococci and staphylococcal infections. Vol. 2. London, United Kingdom: Academic Press, Inc.; 1983. pp. 559–598.
7. Bopp C A, Brenner F W, Wells J G, Strockbine N A. Escherichia, Shigella, and Salmonella. In: Murray P R, Baron E J, Pfaller M A, Tenover F C, Yolken R H, editors. Manual of clinical microbiology. 7th ed. Washington, D.C: ASM Press; 1999. pp. 459–474.
8. Brett M M. Kits for the detection of some bacterial food poisoning toxins: problems, pitfalls and benefits. J Appl Microbiol. 1998;84(Suppl.):S110–S118. [PubMed]
9. Chang T W, Bartlett J G, Gorbach S L, Onderdonk A B. Clindamycin-induced enterocolitis in hamsters as a model of pseudomembranous colitis in patients. Infect Immun. 1978;20:526–529. [PMC free article] [PubMed]
10. Coia J E, Browning L, Haines L, Birkbeck T H, Platt D J. Comparison of enterotoxins and haemolysins produced by methicillin-resistant (MRSA) and sensitive (MSSA) Staphylococcus aureus. J Med Microbiol. 1992;36:164–171. [PubMed]
11. Couch J L, Soltis M T, Betley M J. Cloning and nucleotide sequence of the type E staphylococcal enterotoxin gene. J Bacteriol. 1988;170:2954–2960. [PMC free article] [PubMed]
12. Fekety R, McFarland L V, Surawicz C M, Greenberg R N, Elmer G W, Mulligan M E. Recurrent Clostridium difficile diarrhea: characteristics of and risk factors for patients enrolled in a prospective, randomized, double-blinded trial. Clin Infect Dis. 1997;24:324–333. [PubMed]
13. George W, Sutter V, Citron D, Finegold S. Selective and differential medium for isolation of Clostridium difficile. J Clin Microbiol. 1979;9:214–219. [PMC free article] [PubMed]
14. George W L. Antimicrobial agent-associated colitis and diarrhea: historical background and clinical aspects. Rev Infect Dis. 1984;6:S208–S213. [PubMed]
15. Gravet A, Colin D A, Keller D, Girardot R, Monteil H, Prévost G. Characterization of a novel structural member, LukE-LukD, of the bi-component staphylococcal leucotoxins family. FEBS Lett. 1998;436:202–208. [PubMed]
16. Gutmann L T, Idriss Z H, Gehlbuch S, Blackmon L. Neonatal staphylococcal enterocolitis: association with indwelling feeding catheters and Staphylococcus aureus colonisation. J Pediatr. 1976;88:836–839. [PubMed]
17. Hinton N A, Taggart J G, Orr J H. The significance of the isolation of coagulase-positive staphylococci from stool. With special reference to the diagnosis of staphylococcal enteritis. Am J Clin Pathol. 1960;33:505–510. [PubMed]
18. Jaulhac B, De Buyser M L, Dilasser F, Prévost G, Piémont Y. Screening of staphylococci for the toxic shock syndrome toxin-1 (TSST-1) gene. Lett Appl Microbiol. 1991;13:90–92. [PubMed]
19. Kloos W E, Bannerman T L. Staphylococcus and Micrococcus. In: Murray P R, Baron E J, Pfaller M A, Tenover F C, Yolken R H, editors. Manual of clinical microbiology. 7th ed. Washington, D.C: ASM Press; 1999. pp. 264–282.
20. Kodama T, Santo T, Yokoyama T, Takesue Y, Hiyama E, Imamura Y, Murakami Y, Tsumura H, Shinbara K, Tatsumoto N, Matsuura Y. Postoperative enteritis caused by methicillin-resistant Staphylococcus aureus. Surg Today. 1997;27:816–825. [PubMed]
21. Lundsgaard-Hansen P, Senn A, Roos B, Waller U. Staphylococcic enterocolitis. Report of six cases with two fatalities after intravenous administration of N-(pyrrolidinomethyl tetracycline) tetracycline. JAMA. 1960;173:1008–1013. [PubMed]
22. Mahoudeau I, Elkhaïli H, Delabranche X, Freitas I, Meunier O, Prévost G, Monteil H. Epidémiologie de Staphylococcus aureus dans les Hôpitaux Universtaires de Strasbourg. J Med Strasb. 1997;28:153–159.
23. Maniar A C, Chubb H, Louie T J, Williams T W, Forsyth W, Wilt J C. Detection of Clostridium difficile toxin with McCoy cell monolayers and cell suspensions and comparison with HeLa cell assay. J Clin Microbiol. 1984;19:294–295. [PMC free article] [PubMed]
24. Mauff G, Röhrig I, Ernzer U, Lenz W, Bergdoll M, Pulverer G. Enterotoxigenicity of Staphylococcus aureus strains from clinical isolates. Eur J Clin Microbiol. 1983;2:321–326. [PubMed]
25. McDonald M, Ward P, Harvey K. Antibiotic-associated diarrhoea and methicillin-resistant Staphylococcus aureus. Med J Aust. 1982;1:462–464. [PubMed]
26. Melconian A K, Brun Y, Fleurette J. Enterotoxin production, phage typing and serotyping of Staphylococcus aureus strains isolated from clinical materials and food. J Hyg Camb. 1983;91:235–242. [PMC free article] [PubMed]
27. Nachamkin I. Campylobacter and Arcobacter. In: Murray P R, Baron E J, Pfaller M A, Tenover F C, Yolken R H, editors. Manual of clinical microbiology. 7th ed. Washington, D.C: ASM Press; 1999. pp. 716–726.
28. National Committee for Clinical Laboratory Standards. Performance standards for antimicrobial disk susceptibility tests, 6th ed. Approved standard M2-A5. Wayne, Pa: National Committee for Clinical Laboratory standards; 1997.
29. O'Connor T W. Pseudomembranous enterocolitis: a historical and clinical review. Dis Colon Rectum. 1981;24:445–448. [PubMed]
30. Oeding P, Austarheim K. The occurrence of staphylococci in the intestinal content after treatment with antibiotics. Acta Pathol Microbiol Scand. 1954;35:473–483. [PubMed]
31. Prévost G, Jaulhac B, Piémont Y. DNA fingerprinting by pulsed-field gel electrophoresis is more effective than ribotyping in distinguishing among methicillin-resistant Staphylococcus aureus isolates. J Clin Microbiol. 1992;30:967–973. [PMC free article] [PubMed]
32. Prohaska J V. Pseudomembranous enterocolitis. The experimental induction of the disease with Staphylococcus aureus and its enterotoxin. Arch Surg. 1959;79:197–206. [PubMed]
33. Scopetti F, Orefici G, Biondi F, Benini F. Staphylococcus aureus resistant to methicillin and gentamicin as a cause of outbreak of epidemic enteritis in a hospital. Boll Ist Sieroter Milan. 1983;62:406–411. [PubMed]
34. Société Française de Microbiologie. Communiqué 1998 du Comité de l'Antibiogramme de la Société Française de Microbiologie. Pathol Biol. 1998;46:I–XVI.
35. Southern E. Detection of specific sequence among DNA fragments separated by gel electrophoresis. J Mol Biol. 1975;98:503–517. [PubMed]
36. Surgalla M J, Dack G M. Enterotoxin produced by micrococci from cases of enteritis after antibiotic therapy. JAMA. 1955;158:649–650. [PubMed]
37. Varki N M, Aquino T I. Isolation of Clostridium difficile from hospitalized patients without antibiotic-associated diarrhea or colitis. J Clin Microbiol. 1982;16:659–662. [PMC free article] [PubMed]
38. Zubrzycki L, Spaulding E H. Studies on the stability of the normal human fecal flora. J Bacteriol. 1962;83:968–974. [PMC free article] [PubMed]

Articles from Journal of Clinical Microbiology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Cited in Books
    Cited in Books
    NCBI Bookshelf books that cite the current articles.
  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem chemical substance records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records.
  • Taxonomy
    Taxonomy records associated with the current articles through taxonomic information on related molecular database records (Nucleotide, Protein, Gene, SNP, Structure).
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...