A study of the prevalence of cytotoxic and non-cytotoxic Klebsiella oxytoca fecal colonization in two patient populations

1Dalhousie University, Halifax, Nova Scotia; 2Department of Medicine, Saint John Regional Hospital, Saint John, New Brunswick; 3Division of Gastroenterology; 4Department of Pathology and Laboratory Medicine, Queen Elizabeth II Health Sciences Centre, Halifax, Nova Scotia Correspondence: Dr Kevin R Forward, Divison of Microbiology, Queen Elizabeth II Health Sciences Centre, 5788 University Avenue, Halifax, Nova Scotia B3H 1V8. Telephone 902-473-4109, fax 902-473-4432, e-mail kevin.forward@cdha.nshealth.ca Antibiotic-associated hemorrhagic colitis (AAHC) was first described in 1978 (1). The condition has been associated with beta-lactam, quinolone and pristinamycin therapy (2-4). AAHC usually resolves rapidly following discontinuation of the inciting antibiotic (4,5). The accumulated evidence implicates Klebsiella oxytoca as a probable cause of AAHC (2,5-12). This hypothesis was recently confirmed when Koch’s postulates were fulfilled (4). The pathogenic mechanism by which K oxytoca causes hemorrhagic colitis is thought to involve a chromosomally encoded low-molecular-weight toxin that inhibits nucleic acid synthesis (13,14). The cytotoxin has been shown to induce mucosal hemorrhage in rabbit ileum and is toxic to HEp-2, Vero, CHO and HeLa cells (13,15). K oxytoca also produces a chromosomally encoded beta-lactamase that renders it resistant to aminopenicillins. Therapy with these antibiotics and others originAl Article

to which K oxytoca is resistant presumably contributes to its overgrowth and the development of AAHC.
The ability of penicillins to select for K oxytoca has been exploited by researchers to more easily culture the organism from stool (5).However, this technique still allows for the growth of a significant number of aerobic stool organisms in addition to K oxytoca.We formulated a modified version of MacConkey agar that facilitates the selection and differentiation of Klebsiella species from other fecal flora.
We used this modified MacConkey agar and a simple screening process to determine the background prevalence of both cytotoxic and noncytotoxic K oxytoca in patients at our institution undergoing screening for other pathogens both related and unrelated to antibiotic-associated diarrhea.Few studies have investigated the prevalence of K oxytoca in stool, yet such knowledge is necessary to interpret the significance of cultures performed in patients with suspected K oxytocaassociated hemorrhagic colitis and to determine if the spectrum of illness caused by K oxytoca extends beyond frankly hemorrhagic diarrhea.

Patient selection
Two hundred rectal swabs submitted for vancomycin-resistant enterococci (VRE) screening and 429 Clostridium difficile cytotoxin-negative stool specimens were screened for K oxytoca.Subsequent specimens from previously screened patients were excluded.Specimens were submitted to the microbiology laboratory at the Queen Elizabeth II Health Sciences Centre in Halifax, Nova Scotia, between June and August, 2007.Approval to use clinical specimens for the present study and to review the clinical records was granted by the Capital Health Research Ethics Board.

Culture and screening for K oxytoca
MacConkey agar was modified from the common formulation (16) in two ways: the agar was supplemented with 30 mg/L of ampicillin, and 10 g/L of adonitol was substituted for lactose.Colonies that appeared pink or red, indicating adonitol fermentation, were subcultured onto blood agar and an indole spot test was performed.Positive colonies were further screened with a motility-indole-ornithine test, a triple sugar iron test and a phenylalanine deaminase test.These tests were performed as described in the Clinical Microbiology Procedures Handbook (17).Bacterial isolates that exhibited typical reactions were identified using the Vitek 1 automated instrument (bioMérieux Canada, Inc).Bacterial isolates were considered to be K oxytoca if identified as such by Vitek 1 with at least 90% confidence.

Cytotoxicity assays
C difficile cytotoxicity assays were performed using standard methods.Briefly, cell-free extract from each stool sample was introduced into two microtitre wells lined by human foreskin cell monolayers: one well with C difficile-neutralizing antitoxin (TechLab, USA) and the other without.Samples were considered to be C difficile cytotoxin-positive if cytopathic effects were noted by 48 h in wells without antitoxin and absent in wells containing antitoxin.
K oxytoca cytotoxicity assays were performed using previously described methods (4,5,18).HEp-2 monolayers were assessed for cell rounding and cell death at 24 h and 48 h.K oxytoca 06/19 O was used as a positive control and K oxytoca ATCC 13182 (Cedarlane Laboratories, Canada) was used as a negative control.

Chart review
Medical records for a subset of the patients colonized with K oxytoca were reviewed retrospectively and the incidence of diarrhea and bloody stool was noted.Data were available for 13 K oxytoca-positive patients who were screened for VRE and 27 K oxytoca-positive patients who underwent C difficile cytotoxicity testing.In addition, records of a convenience sample of 93 K oxytoca-negative patients who were also negative for C difficile cytotoxin were reviewed.

Statistical methods
Significance of association between variables was measured using c 2 analysis, except when an expected value was less than 10, in which case Fisher's exact test was employed.

RESULTS
Prevalence of K oxytoca K oxytoca was isolated from 14 of 200 patients (7.0%) tested for VRE and 42 of 429 patients (9.8%) screened for C difficile cytotoxicity (P>0.25), as shown in Table 1.One of the 14 isolates (7.1%) from patients screened for VRE and 10 of the 42 isolates (23.8%) from patients tested for C difficile cytotoxicity were cytotoxic to HEp-2 cells (P>0.25), as shown in Table 2.

Selective and differential capability of MaConkey adonitol media and screening process
The substitution of adonitol for lactose within MacConkey agar allowed for the differentiation of K oxytoca from various other Gram-negative ampicillin-resistant organisms.A total of 222 of the 629 specimens (35.3%) examined grew one or more adonitol-negative colony morphotype, allowing ready differentiation from K oxytoca.One or more adonitol fermenting bacterial strain other than K oxytoca was isolated from 243 (38.6%) of the specimens screened; 187 (77.0%) of these specimens were excluded based on negative spot indole tests and a further 49 (20.2%) because of atypical biochemical reactions (Figure 1).Sixty-seven bacterial isolates from 66 clinical specimens were subjected to definitive examination by Vitek 1; only eight of these isolates were not K oxytoca.Other organisms isolated included Raoultella ornithinolytica, Escherichia hermannii, Escherichia coli and Enterobacter aerogenes.

Chart review
Only two of 13 K oxytoca-positive patients (15.4%) from the group screened for VRE and three of 27 (11.1%)from the group screened for C difficile cytotoxicity, included in the chart review, had documentation of visibly bloody stool (Table 3).None of the five patients included with cytotoxic K oxytoca strains from the group tested for C difficile cytotoxicity had documented bloody stool.Blood was noted in the stool of 11 of 93 K oxytoca-negative control patients (11.8%) from the group tested for C difficile cytotoxicity.All 93 control patients and 27 K oxytoca-positive patients tested for C difficile cytotoxicity were presumed to have diarrhea as an indication for C difficile cytotoxicity testing.Only one patient in the K oxytoca-positive group screened for VRE had documented diarrhea.
The three groups examined did not differ significantly with respect to age or sex, as shown in Table 3. Onset of diarrhea occurred in hospital for 17 K oxytoca-positive patients (63%) in the group screened for C difficile cytotoxicity and 42 K oxytoca-negative patients (45.2%) from the convenience sample (P>0.1).

DISCUSSION
Koch's postulates were recently fulfilled to confirm that K oxytoca causes at least some cases of AAHC (4).We have used a modified form of MacConkey agar to elucidate the background prevalence of the organism by screening patients undergoing testing unrelated to AAHC (VRE screening).This is the first report of the prevalence of K oxytoca fecal colonization at a centre in North America.Previous studies in Austria and France have reported the prevalence of K oxytoca colonization in asymptomatic individuals as 1.6% and 9.0%, respectively (4,5).The latter study found 42.9% of K oxytoca isolates from asymptomatic individuals to be cytotoxic to HEp-2 cells (5).We isolated K oxytoca from 14 of 200 patients (7.0%) being screened for VRE; one isolate (7.1%) was cytotoxic to HEp-2 cells.These results are similar to those of a recent series that identified one cytotoxic and four noncytotoxic K oxytoca isolates in a cohort of 111 hospitalized patients with no associated antibiotic exposure or active diarrhea (19).
We also identified K oxytoca in the stool of 42 of 429 patients (9.8%) who had specimens submitted for C difficile cytotoxicity testing.We assume that the primary indication was clinical suspicion of antibiotic-associated diarrhea.Ten (23.8%) of these isolates were cytotoxic to HEp-2 cells.There was a trend toward more frequent isolation of cytotoxinproducing strains of K oxytoca in the group tested for C difficilecytotoxicity relative to those screened for VRE, but it did not approach statistical significance (P>0.25).It is likely that most patients tested for C difficile cytotoxicity had recent antibiotic treatment, but this did not appear to increase the likelihood of K oxytoca-positive cultures in this population.Hospital records of prior antimicrobial use were incomplete and were not incorporated into this analysis.
The proportion of K oxytoca-positive patients for whom bloody stool was noted in the clinical record did not differ significantly between those screened for C difficile cytotoxicity relative to VRE.Nor did these rates differ significantly from the incidence of bloody stool within the K oxytoca-negative control population.None of the five patients with cytotoxic K oxytoca from the group screened for C difficile cytotoxicity whose charts were available had a notation of bloody stool in their record.Others have also observed that cytotoxic variants of K oxytoca are often isolated in the absence of disease (5).An alternative hypothesis is that the spectrum of diarrheal disease caused by cytotoxin-producing K oxytoca extends beyond AAHC; although the best evidence accumulated to date suggests otherwise, the possibility of such an association cannot be conclusively rejected (19).It is also possible that hematochezia was present in some of our patients but was not documented.This is the first report of the use of a selective and differential growth medium incorporating adonitol for the isolation of Klebsiella species from stool.There is currently no gold standard growth medium used for this purpose.The most recent large-scale   Because K oxytoca constitutively expresses a beta-lactamase and overgrowth of stool flora can decrease bacterial isolation rates, it is likely that the use of ampicillin here and by others (5) improves detection of K oxytoca within stool.The differential property of our modified MacConkey agar also facilitated the isolation of K oxytoca from stool.Because K oxytoca (and other Klebsiella species) ferments adonitol and produces pink/red colonies on this modified MacConkey agar, the medium allows one to differentiate K oxytoca from organisms that do not ferment adonitol.This property allowed us to exclude bacterial colonies as possible K oxytoca isolates in 35.3% of the fecal specimens examined and decrease resource utilization.We have also demonstrated a simple screening process that allowed for the elimination of 97.2% of fecal specimens containing adonitol-positive isolates other than K oxytoca without the need for automated testing.The majority of such specimens (77%) were eliminated using simple spot indole tests.
We have determined the background prevalence of cytotoxic and noncytotoxic K oxytoca in stool and demonstrated a growth medium and screening process combination that allows for easy isolation of the organism.Such studies are necessary to facilitate the interpretation of positive K oxytoca cultures and cytotoxicity assays.Our data clearly show that both cytotoxic and noncytotoxic forms of K oxytoca are common within our institution.These results suggest that current understanding of the occurrence and causality of AAHC is insufficient for clinical laboratories to begin culturing K oxytoca and testing for cytotoxicity; positive and negative predictive values remain unknown.