CTX-M β-Lactamases in Escherichia coli from Community-acquired Urinary Tract Infections, Cambodia

The prevalence of CTX-M β-lactamases has reached a critical level, which highlights the need for study of their spread in developing countries.

and cotrimoxazole is often associated with CTX-M production (16,17), thereby limiting the choice of effective antimicrobial drugs to carbapenems or colistin.
Some studies have shown high prevalence of ESBLs among various Enterobacteriaceae in hospitals in developing countries (20)(21)(22). However, little information is available regarding prevalence within the community. Few bacteriologic data are available for Cambodia, and no studies of resistance of Enterobacteriaceae to antimicrobial agents have been reported. We investigated E. coli CA-UTIs in a 2-year prospective study. Our aims were 1) to establish the prevalence of community-acquired urinary E. coli resistance to a wide range of antimicrobial drugs and 2) to characterize the mechanisms underlying E. coli resistance to β-lactams.

Clinical Isolates from Patient Samples
Our laboratory at the Institut Pasteur in Phnom Penh, Cambodia, is a community laboratory, involved in various biological analyses. We receive biological samples from outpatients and hospitals that do not have bacteriology laboratories. Included in the study were patients who visited our laboratory with a suspected UTI from January 2004 through December 2005. Currently hospitalized patients were excluded, yet previous hospitalization was not an exclusionary criterion. We collected basic clinical data for each patient, including age and gender, whether antimicrobial drugs had been taken within the month preceding the sample submission, UTI history, and recent hospital visit. A UTI was defi ned by >10 4 leukocytes/mL urine and >10 5 CFU/mL urine. We used CLED (cystine-lactose-electrolyte defi cient) agar (Dynamic Pharma, Phnom-Penh, Cambodia) for culture. Bacteria were identifi ed with the API20E identifi cation gallery (bioMérieux, Marcy l'Etoile, France). Strains were named (CEC [Cambodian Escherichia coli]) and numbered independently from isolation date.

PCR Amplifi cations
Template DNA was prepared by boiling. Briefl y, 5 colonies were suspended thoroughly in 1 mL DNase-and RNase-free water and boiled for 10 min. After centrifugation, supernatant was used as template DNA. We amplifi ed the ampC upstream region, bla TEM , bla SHV , bla CTX-M , bla VEB, bla OXA-1 , and bla CMY by PCR, using specifi c oligodeoxynucleotides (Table 1). PCR was performed in a 25-μL mixture of 1× buffer (supplied with Taq polymerase), 2.5 mmol/L MgCl 2 , 2.5 U of FIREPol DNA polymerase (Solis BioDyne, Tartu, Estonia), 200 μmol/L of each deoxynucleoside triphosphate, and 25 pmol of each primer. The PCR mixture was subjected to a 5-min denaturation step at 94°C, followed by 30 cycles of 45 s at 94°C, 45 s at 55°C, and 60 s at 72°C, and a fi nal elongation step of 5 min at 72°C. PCR products were separated by 100-V electrophoresis in a 2% agarose gel for 30 min, after which they were stained with ethidium bromide.

PCR Product Sequencing
Amplifi cation products were purifi ed with Montage PCR Filter Units (Millipore, Billerica, MA, USA). Sequencing reactions were performed in a PTC-225 Peltier Thermal Cycler (MJ Research, Waltham, MA, USA) by using an ABI PRISM BigDye Terminator Cycle Sequencing Kit with AmpliTaq DNA polymerase (Applied Biosystems, Branchburg, NJ, USA), according to the manufacturer's instructions. Each template was sequenced with the appropriate primer. Fluorescence-labeled fragments were purifi ed from the unincorporated terminators with an ethanol precipitation protocol. The samples were resuspended in distilled water and subjected to electrophoresis in an ABI 3730xl sequencer (Applied Biosystems).

Phylogenetic Group Determination
To determine phylogenetic group (i.e., A, B1, B2, and D), we performed triplex PCR for all strains (n = 93) as described previously (29). We used chuA and yjaA genes and an E. coli DNA fragment, TSPE4.C2.

Repetitive Extragenic Palindromic PCR
The clonality of ESBL-positive strains was assessed by repetitive extragenic palindromic (rep)-PCR. DNA was extracted by using the QIAGEN Mini kit (QIAGEN, Courtaboeuf, France). REP-PCRs were performed for all strains in the same batch with primers rep-1R and rep-2T, as described previously (28). The resulting products were separated by 70-V electrophoresis in a 1% agarose gel for 3 h, after which they were stained with ethidium bromide. Strains with suspected similar migration profi les were then migrated together with the ESBL-positive strains to assess their similarity. Photographs of the gels (Figure) have been harmonized to be more informative.

Statistical Analysis
Data were analyzed by using Epi Info version 3.2 (www.cdc.gov/epiinfo). Risk factors for ESBL-producing E. coli were assessed by using univariate analysis with the χ 2 or Fisher exact tests; however, we used analysis of variance to determine whether age had an association with ESBL-producing E. coli. Associations between ESBL type, co-resistance, mutations in ampC promoter regions, and phylogenetic groups were separately tested by using the χ 2 or Fisher exact tests. The signifi cance threshold was 0.05.

Population Characteristics
Of the 861 urine samples, 194 were positive for UTI and 163 contained E. coli. Samples from hospitalized patients and samples isolated from the same patient in a short period were excluded from the study. Overall, 93 patients with E. coli-related CA-UTIs were recruited for the study, all of whom were living in Phnom Penh. Their mean age was 43 years (range 1-88 years, median 40 years), and the M:F ratio was 0.21. No pregnant women were included. Twenty-eight (30.1%) patients had a history of UTI (with no details available), 24 (25.8%) had taken antimicrobial drugs, and 9 (9.7%) had sought consultation at a hospital in the month preceding the urine sampling. Patients with a history of UTI had taken signifi cantly more antimicrobial drugs than had persons without UTIs (p<0.0001).

Prevalence of Antimicrobial Drug Resistance in E. coli Strains
Disk-diffusion susceptibility testing indicated high prevalence of resistance to various antimicrobial agents: 88 (94.6%) strains were resistant to amoxicillin and ticarcillin, and 20 (21.5%) were immediate susceptible to cefoxitin. We observed synergy between clavulanic acid, oxyimino- cephalosporins, and aztreonam for 34 strains (36.6%). Decreased susceptibility to cefoxitin was signifi cantly associated with ESBL expression (p<0.001). All strains (n = 93) were susceptible to imipenem. One strain (CEC93) showed a high level of resistance to cefoxitin, cefotaxime, and ceftazidime but not to cefepime; double-disk synergy test results were negative for clavulanate. We observed a substantial level of resistance to quinolones; 63 strains (67.7%) had an intermediate level of resistance or were fully resistant to nalidixic acid, norfl oxacin, and ciprofl oxacin. Quinolone resistance was strongly associated with ESBL expression (p<0.0001). Of the 34 ESBL-carrying strains, only 2 were susceptible to quinolones. Cotrimoxazole resistance was observed in 70 strains (75.3%) but was not signifi cantly associated with ESBL expression (p = 0.06). Aminoglycoside resistance was also signifi cantly associated with ESBL expression (p<0.01). Thirty-nine strains (41.9%) were resistant to at least 1 of the 4 tested aminoglycosides. The most frequently observed phenotypic profi le included resistance to gentamicin, tobramycin, and netilmicin (37.6% of the strains). None of the strains were resistant to fosfomycin or nitrofurantoin ( Table 2).

Risk Factors for ESBL Carriage
Univariate analysis showed no risk factors for ESBL carriage. Carriage was not signifi cantly associated with gen-der, age, previous hospitalization (within the past month), antimicrobial drug treatment (within the last month), or history of UTI.

β-Lactamase Characterization
Thirty-four strains were tested for ESBL identifi cation. They were all positive for bla CTX-M and negative for bla VEB and bla SHV . We detected bla TEM (Table 3). Only 1 strain did not display any mutations. Most mutations occurred within the transcriptional attenuator region; the most frequently observed profi le were +22 C>T, +26 T>G, +27 A>T, +32 G>A, and +70 C>T (11 strains). The other strains each exhibited a different profi le. Strain CEC57 had the profi le -32 T>A with an additional nucleotide change, which signifi cantly increases ampC expression (31). Strain CEC92 had the combination -42 C>T and -28 G>A, which also increases ampC expression (25).

Phylogenetic Groups
Most of the strains studied belonged to groups B2 (53%; n = 49) and D (28%; n = 26), as do most pathogenic E. coli (32). Twelve strains belonged to group A (13%) and 6 to group B1 (6%). Among ESBL-carrying strains, 53% belonged to group B2 (n = 18), 38% to D (n = 13), 6% to A (n = 2), and 3% to B1 (n = 1). Resistance to cotrimoxazole was less prevalent in B2 strains than among strains from the other groups (p<0.05). Resistance to quinolones was more prevalent among group D strains than in strains from other groups (p<0.005). Furthermore, a strong association (p<0.0001) was found between decreased susceptibility to cefoxitin and group D strains. No association was shown between phylogenetic group and other characteristics such as ESBL carriage and aminoglycoside resistance. Among ESBL-carrying strains, none of the 4 groups were associated with 1 particular type of CTX-M.

Rep-PCR Findings
Twenty-one ESBL positive strains and the CMY-2producing strain were divided into 4 clusters; the remaining strains were genetically unrelated (online Appendix Table, available from www.cdc.gov/EID/content/15/5/741-appT. htm; Figure). Cluster C1 consisted of 8 strains with an identical ampC mutation profi le, all belonging to phylogenetic group D. However, β-lactamase content was variable:

Discussion
We surveyed antimicrobial drug resistance in Cambodia. Our prospective study in this developing country focused on CTX-Ms CA-UTIs caused by E. coli. Although the patients included came from the community, antimicrobial drug resistance was prevalent among UTI-causing strains, particularly to β-lactams (including extended-spectrum cephalosporins). Our fi ndings suggest that CTX-Mtype β-lactamases are widespread in Cambodia. CTX-M production was signifi cantly associated with resistance to quinolones and aminoglycosides and with decreased susceptibility to cefoxitin, leading to a high prevalence of multiresistant strains. The spread of CTX-M in the community has already been described through prospective studies in industrialized countries such as Canada (33), France (34),  16  Yes  D  T  G  T  A  T  CEC3  32  No  D  T  G  T  A  T  CEC4  32  No  D  A  T  T  CEC6  32  Yes  D  T  G  T  A  T  CEC8  32  Yes  D  T  G  T  A  T  CEC9  32  Yes  D  T  G  T  A  T  CEC10  32  Yes  D  T  G  T  A  T  CEC13  32  Yes  D  T  G  T  A  T  CEC14  16  Yes  D  T  CEC15  16  Yes  D  T  CEC19  32  No  D  T  G  T  A  T  CEC21  16  Yes  D  A  T  A  CEC22  16  Yes  D  T  CEC31  16  Yes  D  A  T  A  CEC57  32  No  D  A  T  G  T  A  T  CEC88  32  No  B2  A  A  CEC89  32  Yes  D  T  G  T  A  T  CEC92  >128  Yes  D  T  A  T  T  CEC93  >128  No  D  T  G  T  A  T *CEF, cefoxitin; CEC, Cambodian E. coli. †According to numbering of Jaurin and Grundström (30). and the United Kingdom (12); however, we found higher prevalence of CTX-Ms in Cambodia than that reported in these previous studies. We propose 3 possible explanations for the situation in Cambodia. First, although no reliable data were available, it is well known that many persons in the community self-medicate or obtain prescriptions for non-adapted drugs (drugs taken without any antibiotic susceptibility testing of the strain that causes the infections); these practices suggest that uncontrolled consumption of antimicrobial agents is likely to play a major role. This factor, together with the likely substandard quality of some drugs, undoubtedly contributes to the high prevalence of resistance. Second, hygiene in Cambodia is poor, and the population density in Phnom Penh is high. The saturated sewer system, particularly during the rainy season, likely facilitates effi cient propagation and spread of bacteria within the community. Grenet et al. proposed that that poor hygiene in a French-Guyanese Indian community led to the spread of resistant bacteria despite low antibiotic pressure (35). Similarly, poor hygiene conditions may have led to the spread of resistant strains in the Cambodian community studied here. The TEM-type and SHV-type ESBL-until recently the predominant ESBL family subtypes-have never been implicated in the spread of ESBL in the community. Why CTX-M strains are the only ESBL types to spread in the community is not clear. CTX-M strains may have a particularly high capacity to disseminate or an ecologic advantage over other ESBLs and thus persist in the community, whereas other ESBL may be progressively eliminated with decreasing antibiotic pressure.
One limitation to our study is that patients at our institute are not necessarily representative of the Cambodian population. For most persons in Cambodia, antimicrobial agents and biologic analyses are expensive. Therefore, our patients may have been wealthier and may have taken more courses of antimicrobial drug treatment than other Cambodians.
REP-PCR yielded 4 clusters of strains, consistent with their identical ampC mutational profi les, yet with various contents of β-lactamases. Because all included patients were living in Phnom Penh, mediocre hygiene might have favored the diffusion of clones within an urban area, a phenomenon that had previously been observed in large-scale studies undertaken in the United Kingdom (12), Canada (36), Italy (37), and Brazil (15) but not in Hong Kong (13). In contrast to our results, clones identifi ed in those studies harbored the same β-lactamases. Further investigation using multilocus sequence typing would be necessary to identify the molecular determinants of the CTX-M-carrying E. coli pandemic in Cambodia.
Branger et al. have observed that group D E. coli were more frequently resistant to quinolones (38) than were non-D E. coli; these fi ndings are consistent with our results. Moreover, we found a strong association (p<0.001) between group D strains and decreased susceptibility to cefoxitin (secondary to the effect of mutations in the ampC promoter region). This association was also present when all strains were taken into account (p<0.001). However, although CTX-M carriage was more frequently observed for group D strains than for non-D strains, the association was not signifi cant. Given that quinolone resistance and ampC hyperexpression involve several mutations, a possible explanation for this association with group D strains may be a stronger mutation capacity for these strains than for strains belonging to other groups. Further investigations will be required to explain this phenomenon.
Community-isolated ESBL-carrying strains are an emerging challenge for community practitioners and hospitals. Information is not readily available in either developing countries or in industrialized countries, and UTI treatment guidelines remain unchanged. In Cambodia, and probably in many other developing countries, resistant E. coli strains are endemic to the community. Investigating the current situation in Cambodia may improve our understanding of the situation in industrialized countries, where ESBLs are no longer uncommon in the community. According to our experience in Cambodia, measures should focus on improving hygiene and appropriate prescribing of antimicrobial agents. In conclusion, we suggest that this high prevalence of β-lactam resistance in Cambodia is due to the intrinsic capacity of CTX-M-encoding genes to disseminate through communities where hygiene and living conditions are poor and antimicrobial drug consumption is uncontrolled. Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 15, No. 5, May 2009 The opinions expressed by authors contributing to this journal do not necessarily refl ect the opinions of the Centers for Disease Control and Prevention or the institutions with which the authors are affi liated.