Surveillance of Antibiotic Resistance among Hospital- and Community-Acquired Toxigenic Clostridium difficile Isolates over 5-Year Period in Kuwait

Clostridium difficile infection (CDI) is a leading and an important cause of diarrhea in a healthcare setting especially in industrialized countries. Community-associated CDI appears to add to the burden on healthcare setting problems. The aim of the study was to investigate the antimicrobial resistance of healthcare-associated and community-acquired C. difficile infection over 5 years (2008–2012) in Kuwait. A total of 111 hospital-acquired (HA-CD) and 35 community-acquired Clostridium difficile (CA-CD) clinical isolates from stool of patients with diarrhoea were studied. Antimicrobial susceptibility testing of 15 antimicrobial agents against these pathogens was performed using E test method. There was no evidence of resistance to amoxicillin-clavulanic acid, daptomycin, linezolid, piperacillin-tazobactam, teicoplanin and vancomycin by both HA-CD and CA-CD isolates. Metronidazole had excellent activity against CA-CD but there was a 2.9% resistance rate against HA-CD isolates. Ampicillin, clindamycin, levofloxacin and imipenem resistance rates among the HC-CD vs. CA-CD isolates were 100 vs. 47.4%; 43 vs. 47.4%; 100 vs. 100% and 100 vs. 89%, respectively. An unexpected high rifampicin resistance rate of 15.7% emerged amongst the HA-CD isolates. In conclusion, vancomycin resistance amongst the HA-CD and CA-CD isolates was not encountered in this series but few metronidazole resistant hospital isolates were isolated. High resistance rates of ampicillin, clindamycin, levofloxacin, and imipenem resistance were evident among both CA-CD and HA-CD isolates. Rifampicin resistance is emerging among the HA-CD isolates.


Introduction
Clostridium difficile is a Gram-positive, spore-forming anaerobic bacteria. It has an important role in hospital-acquired diarrhoea ranging from mild cases to severe pseudomemraneous colitis, collectively named C. difficile infection (CDI). The number of CDI has increased dramatically in term of frequency, severity, occurrence in outbreaks setting and recurrence rate since 2002 in North America [1], Europe [2] and Australia [3]. This is partially related to the emergence of the hypervirulent strains i.e. C. difficile B1/ NAP1/027 [4]. Another hypervirulent strain i.e. C. difficile ribotype 078 has emerged, whose incidence increased from 3 to 13% during 2005-2008 in the Netherlands [4]. In addition, new risk groups have been added to the list which include community-onset CDI [5,6], CDI in children as well as peripartum ladies [7].
Antimicrobial therapy plays an important role in the development of CDI. This risk increases if C. difficile is resistant to the offending or the used antimicrobial agent [8]. One of the main theories for the rise in the reported cases as well as outbreaks of CDI is the circulation of flouroquinolone-resistant C. difficile ribotype 027 at the same time as the wild use of flouroquinolones in hospitals [9]. Antibiotic resistance are important in the emergence of epidemic clones and persistence of specific types over time in the hospitals.
CDI plays a significant burden on the healthcare setting and financial resources. Treatment of C. difficile is difficult because of its direct causal relationship with antibiotic use. The two most commonly prescribed antimicrobial agents for treatment of CDI are metronidazole and vancomycin with high recurrence rates. Antimicrobial susceptibility testing is not done routinely for anaerobes including C. difficile. Although C. difficile has been reported as susceptible to metronidazole and vancomycin, there are some reports of reduced susceptibility of C. difficile to metronidazole [10,11]. High percentage of multi-drug resistant (resistant to 3 or more drugs) C. difficile was found in certain European hospitals [12]. There is no clear relationship or association between treatment failure and reduced susceptibility or resistance to metronidazole or vancomycin. The aim of this study was to investigate the trend of antibiotic resistance of healthcare-associated and community-acquired C. difficile infection (HA-CDI; CA-CDI) over 5 years (2008)(2009)(2010)(2011)(2012) in Kuwait.

Definitions
Diarrhea was defined as loose stools, i.e. taking the shape of the container or corresponding to Bristol stool chart types 5-7, plus a stool frequency of three stools in 24 h or fewer consecutive hours or more frequently than is normal for the individual (definition of World Health Organization, http://www.who.int/topics/diarrhoea) [13,14,15]. According to the European Centre for Disease Prevention and Control [16], an episode of CDI was defined as a patient with diarrhea whose stool takes the shape of the container, and it is positive for C. difficile toxin A and/ or B without other etiology or endoscopic evidence of pseudomembranous colitis.
CA-CDI was defined, in this study, as the onset of symptoms occurring while the patient was outside a healthcare facility and the patient had not been discharged from a healthcare facility within 12 weeks before symptom onset (community onset/community-acquired); or the onset of symptoms occurring within 48 h upon admission to a healthcare facility and the patient had no prior stay in a healthcare facility within the 12 weeks prior to symptom onset (healthcare facility onset; community-acquired) [16].

Bacterial strains
Stool samples were collected from diarrheagenic patients suspected of C. difficile infection acquired in the hospital or community from 2008-2012. They were processed in the Anaerobe Reference Laboratory, (ARL, Ministry of Health), Faculty of Medicine, Kuwait. Stool specimens were cultured on selective media (cycloserine-cefoxitin fructose agar; CCFA, Oxoid limited, Basingstoke, Hampshire, UK) after heat-shock procedure. Suspected C. difficile isolates were identified by Gram staining, typical morphology on agar plates, as well as characteristic odour. Then it was identified further by biochemical tests using API 20A (bioMerieux, Marcy I'Etoile, France) and confirmed by VITEK MS (bioMerieux). Toxin production was investigated by GeneXpert TM C difficile assay (Cepheid, Sunnyvale, CA, USA) and EIA method using C. diff Quik Chek Complete kit (QCC) (TechLab, Blacksburg, VA, USA) according to manufacturer's instructions.

PCR ribotyping
PCR ribotyping of all isolates was performed as previously described [20].

Statistical analysis
Fisher's exact test (two-sided) was used to test for difference between CA-CD and HA-CD with those differences showing p <0.05 deemed to be statistically significant.

Ethics statement
Collection of the strains was conducted according to the Declaration of Helsinki and with particular institutional ethical and professional standards. A written informed consent was not obtained from patients or parents of children because the bacterial isolates studied were collected from the routine work of clinical microbiology laboratory for patient care and no additional clinical specimens were collected for the purpose of the study. It is a standard practice not to get written informed consent for use of bacterial isolates unlinked to patient identity from the routine clinical laboratory. Therefore, the waiver for informed consent was granted and the study was approved by the Medical Ethics Committee of Ministry of Health, Kuwait (permit number 2093/MTT).

Results
A total of 111 hospital-acquired and 35 community-acquired toxigenic C. difficile isolates were analyzed. The distribution of HA-CD and CA-CD over 5 years is shown in Table 1.  Table 2.

Discussion
Knowing the antimicrobial susceptibility is critically important when treating patients with CDI in hospital as well as in community settings. Only one study on hospital-acquired C. difficile antimicrobial susceptibility in Kuwait has been published and that was 13 years ago [21]. In the previous study, antimicrobial susceptibility of 15 antimicrobial agents was determined by E test methodology for hospital acquired C. difficile isolates collected from patients in intensive care units of 3 hospitals over a period of 2 years in Kuwait [21]. There have been no study in Kuwait or elsewhere comparing the susceptibility of HA-CD and CA-CD.
In this study, we have evaluated 15 antimicrobial agents, including the two antibiotics currently used as standard therapy for CDI, vancomycin and metronidazole, against 111 hospital-     acquired and 35 community-acquired C. difficile isolates. Our data showed that 2.9% of HA-CD were resistant to metronidazole while all CA-CD isolates were susceptible. However, worthy of note was the higher MIC 90 values demonstrated by current CA-CD isolates compared to earlier study for HA-CD, 0.75 vs. 0.19μg/ml, respectively. This is in contrast to the surveillance studies in Spain and Texas, which reported much higher resistance rates of clinical C. difficile isolates to metronidazole (6.3% and 13.3%), respectively [22,23], unlike lower resistance level of 0.11% reported in the European surveillance study by Freeman et al [24]. In the Pelaez et al study, metronidazole resistance was heterogeneous and not related to the presence of nim genes [22]. All our CA-CD and HA-CD isolates were susceptible to vancomycin howbeit with higher MIC 90 values compared to earlier study for HA-CD (3.0 vs. 0.75μg/ml). This finding is consistent with previous reports [25,26] which demonstrated MIC 90 of 0.75 and 2μg/ml, respectively. Contrastingly, a pan-European surveillance study in 22 European countries reported a 0.9% vancomycin resistance rate [24]. However the role of vancomycin resistant C. difficile is not clear because of the high concentration of vancomycin in the colon of patients after oral vancomycin therapy [27]. In our study, non-susceptibility to imipenem was high in both CA-CD and HA-CD (100 vs. 89%) whereas meropenem resistance was high for only CA-CD (43 vs. 0%). Imipenem resistance in this study is similar to our previous study [21] for HA-CD isolates as well as to the study in Poland [25] which reported that 86%, 87.9%, respectively, of their strains were resistant to imipenem. These reports differ from other studies in Korea and Europe where resistance of HA-CD isolates to imipenem was relatively low 7.4% and 8%, respectively [26,28]. While resistance to meropenem was zero in the previous study in Kuwait as well as in Australia [21,26] for HA-CD isolates, it was relatively high with the CA-CD isolates. Imipenem and meropenem have been used liberally in Kuwait hospitals for the last two decades, which may have led to the selection pressure for resistance to imipenem and meropenem secondary to drug exposure.
Clindamycin resistance was observed in 43% and 47% of CA-CD and HA-CD isolates, respectively, a finding similar to results of our previous study (48%) [21] and a European study (49.62%) [24] but much higher than that reported previously in Poland (27.7%) [25]. It is, however, lower than reports from Australia [26] and Korea [28] where 84.3% and 81%, respectively were resistant. Interestingly, there was a remarkable difference in the resistance levels to erythromycin which was observed in 14.3% of CA-CD isolates compared to 58% of HA-CD isolates. The Korean [28] and Polish [25] studies reported high level resistance to erythromycin, of 80% and 85.5%, respectively.
Rifampicin resistance was 16% among our HA-CD isolates. This is similar to reports from Poland and Europe of 13.40% and 18%, respectively [24,25], but unlike reports from Australia were the resistance rate to rifaximin was almost 0% [26]. The relatively high rifampicin resistance rate in our study may be explained, in part, by the concurrent high rifampicin resistance rates of 10% amongst the Staphylococcus aureus isolates in Kuwait [29] and of 9.2% rate amongst the Mycobacterium tuberculosis isolates [30]. In Poland, rifampicin resistance has emerged in an outbreak of C. difficile ribotype 046 in patients on long term treatment of rifampicin for tuberculosis [31].
We found a striking difference between the susceptibilities of our isolates (HA-CD and CA-CD) to the quinolones, in particular levofloxacin, and the Australian isolates. While all our isolates were resistant as in our previous study of 97% to trovafloxacin, only 3.4% of the Australian isolates were reported as resistant to moxifloxacin [26].
One of the limitations of this study is the relatively small number of CA-CDI isolates, which is probably a true reflection of the paucity of CDI in our country. The second limitation was the use of E test for testing the susceptibility of the isolates to metronidazole. Although CLSI recommends agar dilution method to test for metronidazole MIC, comparison of MICs determined by agar dilution and E test has shown good correlation (+/-2 dilutions) of 86.6, 95.9, and 99% for metronidazole, vancomycin and teicoplanin, respectively, as reported by Barbut et al [32]. In addition, antimicrobial susceptibility testing of C. difficile by E test method has been well documented in previous reports in the literature [21,24,31].

Conclusions
We did not encounter any vancomycin-resistant isolate amongst the HA-CD and CA-CD isolates in this series but few metronidazole resistant hospital isolates were isolated. Ampicillin, clindamycin, levofloxacin and imipenem resistance were evidently at unacceptable levels for both CA-CD and HA-CD isolates. We noted that rifampicin resistance is emerging among the HA-CD isolates and this calls for caution in the use of rifampicin for infections other that Mycobacterium tuberculosis. Therefore, we recommend periodic surveillance and regular antimicrobial susceptibility testing for all toxigenic C. difficile isolates as an informed guide to empiric antibiotic use.