Characterization of Fecal Extended-Spectrum-β-Lactamase-Producing Escherichia coli in a Remote Community during a Long Time Period

Subjects.

The study was performed in October 2010. The Trois-Sauts village then had a population of 597 inhabitants and around 10 civil servants, including school teachers and nurses. The lifestyle of the population was still largely traditional, without major changes compared to 2001 (5) and 2006 (6), although access by motor-driven boats and schooling of the oldest children outside the village at around 150 km to the north had increased. The only source of antibiotics was still the health post, where drugs continued to be delivered free of charge upon prescription by the resident nurses, who also kept a precise record of all distributions. As before, only visitors who benefit from permits which are rarely delivered by the regional authority were admitted in the village. We identified the 163 adult volunteers from the 2006 cohort and asked them to participate again. A small financial reward was given for their participation, as during our previous sampling campaign. Volunteers were included if they were healthy at physical examination and signed an informed consent form. Then, they were also expected to reply to a standard questionnaire and provide a freshly passed fecal sample. The study was approved by the Regional Ethics Committee (Comité de Protection des Personnes Sud-Ouest et Outre Mer III, 2010-A00682-37).

Bacteriological methods.

The bacteriological techniques used were essentially the same as those used during the previous campaigns in order to ensure optimal comparative analyses. Fresh fecal samples were inoculated immediately onto Drigalski agar slants in screw-cup tubes and sent to mainland France within 2 weeks, at room temperature. There, the whole culture from each tube was suspended in 1.5 ml of brain heart infusion broth with 10% glycerol and stored at −80°C. Upon harvesting, 100-μl aliquots of each broth were cultured on agar plates selective for extended-spectrum cephalosporin-resistant strains containing 1.5 mg/liter cefotaxime and 2 mg/liter ceftazidime (ESBL; AES Chemunex, Bruz, France). All Enterobacteriaceae with different morphotypes that grew on these plates were identified at the species level using a Maldi Biotyper system (Bruker Daltonics, Bremen, Germany). Their antimicrobial susceptibility was determined using the disk diffusion method, as previously described (http://www.sfm-microbiologie.org/). Class A ESBLs were detected using a double-disk synergy test (7). The phylogenetic relationships of E. coli isolates were determined using the triplex PCR method (8) and multilocus sequence typing (MLST) (9), using 40 strains representing species diversity as references (10). Isolates of the present sampling were compared with each other and also with isolates from 2006 from the same volunteers by repetitive extragenic palindromic PCR (rep-PCR), using a DiversiLab strain typing system (bioMérieux, Marcy l'Etoile, France). The presence of extraintestinal (hly, aer, papC, iroN, traT, ompT, fyuA, hra, and kpsE) and intraintestinal (stx1, stx2, ipaH, heat-labile toxin [LT] and heat-stable toxin [ST] genes, aatA, aaiC, afaD, eae, and bfpA) virulence factor genes was assessed in isolates representative of each rep-PCR pattern, as described previously (10).

Resistance genes, including bla_CTX-M (groups 1, 2, 8, 9, and 25), bla_SHV, and bla_TEM, were amplified with specific primers, as described previously (11). All amplification products were sequenced and submitted to the National Center for Biotechnology Information Library (http://blast.ncbi.nlm.nih.gov) for precise identification.

The transferability of ESBL genes was assessed by mating with E. coli J53rif (12). When mating was negative, transformation into E. coli TOP10 (Invitrogen, Saint-Aubin, France) was attempted by electroporation of whole-plasmid DNA, as described previously (6). Transformants were selected on Drigalski agar with 1 mg/liter cefotaxime, and their antimicrobial susceptibility patterns were assessed. Plasmid replicons from parental strains, transconjugants, and transformants were typed by PCR, as described previously (13). Finally, when plasmids from strains of different genetic backgrounds bore the same CTX-M allele and were either undistinguishable by replicon typing or shared one (or more) common replicon(s), plasmidic DNA was purified from E. coli transconjugants or transformants using the alkaline lysis procedure (Large-Construct kit; Qiagen, Courtaboeuf, France), as recommended by the manufacturer. Restriction fragment length polymorphisms of plasmid DNA were analyzed after cleavage with EcoRI and 0.8% agarose gel electrophoresis. The same procedure was also applied to the plasmids isolated from the same volunteers during the 2006 campaign, when relatedness was suspected according to the same criteria.

Epidemiological data.

Demographic data (age, sex, marital status, and number of children), data on lifestyle (including travel frequency, duration, and destinations) and the environment (size and location of households, contact with animals, daily activities, and baby-sitting of children under 5 years old), and data on medical history (current pregnancy, chronic disease, previous hospitalizations and surgery, and oral and parenteral antibiotic intake during the year before sampling) were collected from each volunteer in a standardized manner using data recorded at the health post and interviews of local informers. Antibiotic treatments prescribed to all villagers were available for 92% of them (552/597) by extraction from the health records, which are stored in the health post. Familial status and the location of the household of each villager were also recorded.

Statistical methods.

All data were analyzed blindly with respect to the identity of the villagers. However, each inhabitant was assigned a code number which allowed us to perform comparisons over time. We compared the epidemiological characteristics of ESBL-E carriers versus noncarriers by means of R software (version 2.13.0 [http://www.cran.r-project.org]). A univariate analysis of discrete variables was performed using a two-sided Pearson chi-square test and Fisher's exact tests. The Student t test and Wilcoxon's test were used to testing continuous variables. All tests performed were two sided, and the significance level was set at 5%. Due to the large number of explanatory variables tested, the results of the univariate analysis were adjusted using Holm's test for multiple testing. We evaluated for a 9-year period (from 2001 to 2010) the evolution of the ESBL-E carriage rate according to overall antibiotic exposure in the year preceding each campaign using a linear regression model and Pearson's correlation. We used McNemar's test to compare the ESBL-E carriage rates between 2006 and 2010.

Characteristics of the study population.

The village population increased from 525 subjects in 2006 to 597 in 2010 without significant structural changes (Table 1). Of the 163 volunteers from the 2006 campaign, 152 were present in 2010, and 151 (99%) of them volunteered again (Table 1).

ESBL-E fecal carriage.

The ESBL-E carriage rate dropped from 8.0% (13/163) in 2006 to 5.3% (8/151) in 2010. Three additional volunteers in 2010 (1.8%) carrying isolates were further identified as producing CMY-2 plasmid-borne cephalosporinase (see below) versus only one study volunteer in 2006.

Antibiotic exposure.

Overall antibiotic exposure decreased in the village from 1.09 treatments/subject/year in 2006 to 0.67 treatments/subject/year in 2010 (P < 0.001), a figure close to that observed in 2001 (5) (Table 2). This decrease was observed for all antibiotic classes and was highest for penicillins and metronidazole, which were also the most frequently used antibiotics. This was also true for cohort volunteers (P = 0.06) (Table 2). Of note, carboxi-iminopenicillins, cephalosporin, or carbapenems were not available, and exposure to fluoroquinolones was minimal (four courses of antibiotics during the year preceding the sampling campaign).

Risk factors for ESBL-E carriage.

We found no significant differences in demographic data, lifestyle environment, or medical history between ESBL-E carriers and noncarriers after adjustment for multiple testing (Table 3). The overall level of antibiotic consumption in the village and the ESBL-E carriage rate are presented in Fig. 1. A trend toward a correlation was found using linear regression (y = 8.675x − 1.396; R² = 0.87). However, Pearson's correlation did not reach significance.

Genetic backgrounds.

Among the eight ESBL-E carriers identified in 2010, four, three, and one were carrying, respectively, one, two, and three different isolates, all of which were E. coli. Thus, a total of 13 ESBL E. coli isolates were isolated (Table 4). As in 2006 in the same volunteers, most of the isolates (12/13) were from commensal A and B1 phylogroups (14). A single strain was from phylogroup D. Rep-PCR showed two patterns (M and N) that were shared each by three different isolates. Each pattern was present in three separate volunteers. Interestingly, isolates exhibiting patterns M and N coexisted in two of them (Table 4). These four volunteers (two carrying both isolates with the M and N patterns and two carrying only one isolate) lived in the same hamlet, and three of them were direct neighbors, although not family related. The other seven isolates were singletons, as confirmed by MLST (Fig. 2). Strikingly, rep-PCR patterns from 2010 isolates were all different from those of 2006 (see Fig. S1 in the supplemental material). Just as in 2006, the virulence gene content of ESBL-E isolates, which was analyzed in strains representative of the rep-PCR patterns, was low (Table 4). Of note, one strain was positive for aatA and aaiC, which allowed us to identify an enteroaggregative pathotype.

Susceptibility pattern and identification of ESBL and cephalosporinases.

ESBL-E strains were coresistant to sulfamethoxazole (77%, or 10/13), tetracycline (69%, or 9/13), trimethoprim (46%, or 6/13), nalidixic acid (23%, or 3/13), and kanamycin (8%, or 1/13). Three strains had no coresistance at all (Table 4).

All 2010 ESBL isolates were CTX-M types, including six (46%) CTX-M-1, five (38%) CTX-M-8, and two (15%) CTX-M-2 (Table 4). CMY-2 was identified in the three cases suspected of producing a transferable cephalosporinase. Transfer experiments showed that cotransfer of other resistant traits together with ESBL genes had occurred in only 4/9 cases and was limited to sulfamethoxazole, trimethoprim, and tetracycline (Table 4).

Plasmid characterization.

Plasmids present in CTX-M-8 strains were positive for incompatibility group I1/Iγ in strains K, Q, R, and S isolated in 2010 but not typeable in strains A and P, isolated in 2006 and 2010, respectively. Restriction patterns of I1/Iγ plasmids from the four CTX-M-8 strains (K, Q, R, and S) were similar (β pattern) (Table 4). Of note, three of these strains (Q, R, and S) were cocarried by one volunteer, whereas the last strain (K) was found in another volunteer who was unrelated and lived in a remote hamlet. In the two M and N CTX-M-1 strains, plasmids also belonged to incompatibility group I1/Iγ and shared the same γ restriction pattern (Table 4). Some minor polymorphisms were observed between plasmids exhibiting β and γ patterns, suggesting microevolution in different bacterial hosts (data not shown).