Antimicrobial-resistant Escherichia coli in public beach waters in Quebec

P Turgeon, P Michel, P Levallois, al. Antimicrobial-resistant Escherichia coli in public beach waters in Quebec. Can J Infect Dis Med Microbiol 2012;23(2):e20-e25. INTRODUCTION: Human exposure to antimicrobial-resistant bacteria may result in the transfer of resistance to commensal or pathogenic microbes present in the gastrointestinal tract, which may lead to severe health consequences and difficulties in treatment of future bacterial infections. It was hypothesized that the recreational waters from beaches represent a source of antimicrobial-resistant Escherichia coli for people engaging in water activities. ObJeCTIve: To describe the occurrence of antimicrobial-resistant E coli in the recreational waters of beaches in southern Quebec. MeTHODS: Sampling occurred over two summers; in 2004, 674 water samples were taken from 201 beaches, and in 2005, 628 water samples were taken from 177 beaches. The minimum inhibitory concentrations of the antimicrobial-resistant E coli isolates against a panel of 16 antimicrobials were determined using microbroth dilution. ReSULTS: For 2004 and 2005, respectively, 28% and 38% of beaches sampled had at least one water sample contaminated by E coli resistant to one or more antimicrobials, and more than 10% of the resistant isolates were resistant to at least one antimicrobial of clinical importance for human medicine. The three antimicrobials with the highest frequency of resistance were tetracycline, ampicillin and sulfamethoxazole. DISCUSSION: The recreational waters of these beaches represent a potential source of antimicrobial-resistant bacteria for people engaging in water activities. Investigations relating the significance of these findings to public health should be pursued. H uman exposure to antimicrobial-resistant bacteria may result in the transfer of antimicrobial resistance to commensal or pathogenic microbes present in the gastrointestinal tract and may lead to severe health consequences. An increase in antimicrobial resistance can lead to greater morbidity and mortality due to bacterial infections, limited therapy alternatives and delayed adminis-tration of effective therapies. Bacterial resistance capacity can be intrinsic or can be the result of mutation or transfer of resistant genes from other bacteria (1). Microorganisms can be resistant to one or more antimicrobial agents and can come from different sources, including the food chain and the environment (2,3).

INTRODUCTION : L'exposition humaine à des bactéries résistant aux antimicrobiens peut provoquer le transfert de la résistance à des microbes commensaux ou pathogènes présents dans le tube digestif, ce qui peut avoir de graves conséquences sur la santé et compliquer le traitement de futures infections bactériennes. On a soulevé l'hypothèse que les eaux de baignade des plages représentent une source d'infection à l'Escherichia coli résistant aux antimicrobiens pour les personnes qui s'adonnent à des activités aquatiques. La présente étude visait principalement à décrire l'occurrence d'E coli résistant aux antimicrobiens dans les eaux de baignade du sud du Québec. H uman exposure to antimicrobial-resistant bacteria may result in the transfer of antimicrobial resistance to commensal or pathogenic microbes present in the gastrointestinal tract and may lead to severe health consequences. An increase in antimicrobial resistance can lead to greater morbidity and mortality due to bacterial infections, limited therapy alternatives and delayed administration of effective therapies. Bacterial resistance capacity can be intrinsic or can be the result of mutation or transfer of resistant genes from other bacteria (1). Microorganisms can be resistant to one or more antimicrobial agents and can come from different sources, including the food chain and the environment (2,3). There are multiple environmental sources, which can include water exposure through drinking or recreational activities. Therefore, aquatic ecosystems may represent a reservoir for antimicrobial-resistant bacteria and a potential medium for the spread and evolution of antimicrobial resistance (4,5).
To our knowledge, few studies have described the exposure to antimicrobial-resistant bacteria through recreational waters for people engaged in water/bathing activities. The present study was part of the Canada-wide research initiative entitled "Prospective Multi-Province Surveillance for Antimicrobial-Resistant Escherichia coli in Drinking and Recreational Source Waters: Impact on Humans and the Environment" (Antimicrobial Resistant Organism -ARO Water Study). The main objective of the present study was to determine the occurrence of antimicrobial-resistant microorganisms in recreational waters of beaches in southern Quebec. We hypothesized that recreational waters from beaches represent a source of antimicrobial-resistant Escherichia coli for people engaging in water activities. E coli is used as a biological indicator for fecal contamination in water sources. These bacteria are known to be a reservoir of resistance genes, and horizontal gene transfer between different strains of E coli and other intestinal organisms has been demonstrated (6,7).

MeTHODS
The present prevalence study aimed to estimate the occurrence of antimicrobial-resistant E coli in public beaches located on lakes in southern Quebec. The beaches included in the present study participated in the provincial public beach surveillance program during the summers of 2004 and 2005 ( Figure 1).

Sampling
Water samples were obtained from the provincial public beach surveillance program during the summers of 2004 and 2005. In this program, beaches were sampled continuously throughout the bathing period, usually from mid-June to the first week of September, and the sampling frequency for each beach was determined by the results of water quality testing from the preceding year. For this surveillance program, beaches were classified in four sanitary groups according to the fecal coliform concentration found in the sampled water. A group A beach has zero colony-forming units (CFU)/100 mL to 20 CFU/100 mL and is tagged as 'excellent quality'. Similarly, a group B beach ('good') has 21 CFU/100 mL to 100 CFU/100 mL, group C ('poor') has 101 CFU/100 mL to 200 CFU/100 mL and group D ('polluted') has more than 200 CFU/100 mL. Beaches in group A are sampled at least once every two years, with the possibility of being sampled annually or more frequently. Beaches in class B are sampled at least three times a year and beaches in classes C and D and new beaches are sampled at least five times a year. For each sampling session, or harvesting, the number of samples collected is a function of the beach length. The number of samples taken ranges from six, for a beach ≤60 m, to 30 for a beach ≥721 m. Each sample is taken inside the bathing area (delimited by floating cables) and the distance between each sampling location is set at equal lengths across a given beach. Water samples are taken at a 15 cm depth, and the sampling pattern is adapted to the beach layout (linear or circular) to make sure that the spatial distribution of the contamination is well assessed. For a linear beach with a bathing area depth of >1.2 m, sampling is performed using a 'W' pattern, while for a linear beach having a bathing area with a water depth <1.2 m, a linear sampling pattern is used. For a circular beach, a linear sampling pattern is also performed for the entire circumference of the bathing area with no consideration of water depth. All water samples are collected using a 250 mL sterile polypropylene bottle containing sodium thiosulfate, although beaches using chlorination (n=9) were not included in the present study. The bottle was submerged in water and the cap was removed only at the sampling time and replaced immediately after collection. Samples were preserved at 4°C and transported to the laboratory within 24 h. A detailed description of this beach surveillance program is available (8).

Microbial analysis
As part of the same surveillance program, beach waters were quantified by membrane filtration and microbiologically tested for fecal coliforms on membrane fecal coliform (mFC) agar (Dalynn, Canada), following the procedure described by the Centre d'expertise en analyse environnementale du Québec (9). For each harvesting of a beach in the present study, up to five different morphotypical colonies suggestive of fecal coliforms were selected from five randomly chosen mFC agar plates. Selected colonies were cultured on a tryptic soy base (TSB) agar slant (Difco, USA) for 24 h at 44°C. Slants were then transferred on ice by priority airmail to the study laboratory in Calgary, Alberta, to ensure delivery in <24 h. Growth from the slant was frozen in skim milk upon receipt and processed within one year from the date of water sampling. Recovery of bacteria when samples were processed immediately versus after freezing for up to one year was not affected by prolonged storage (data not shown). To differentiate other fecal coliforms from E coli, bacteria archived in skim milk were cultured onto X-Gluc agar (Dalynn, Canada) for 18 h to 24 h at 35°C. Up to five blue/green colonies suggestive of E coli of different morphotypes were selected and inoculated into TSB broth and incubated at 35°C for 4 h to 6 h to promote growth.
An aliquot of 0.01 mL of TSB broth bacterial suspension was screened by an agar plate method for antimicrobial resistance using a MacConkey agar control plate without antimicrobials and six MacConkey plates each containing a different antimicrobial: ampicillin 8 µg/mL, gentamicin 8 µg/mL, nalidixic acid 4 µg/mL, streptomycin 32 µg/mL, sulfamethoxazole 156 µg/mL and tetracycline 4 µg/mL. Based on results from a pilot experiment, the concentration of antimicrobial used in the agar screen method was selected to be lower, if possible, than the human resistance breakpoint to be sufficiently sensitive for screening. The choice of these concentrations was made to improve the sensitivity of the test and to capture a maximum amount of antimicrobial-resistant isolates to be processed by the National Antimicrobial Resistance Monitoring System (NARMS) (Centers for Disease Control and Prevention, USA). These concentrations also contributed to preventing inadvertent background overgrowth by fecal coliforms other than E coli. For an isolate to be classified as 'resistant', the minimum inhibitory concentration (MIC) determined by NARMS had to be greater than or equal to the human breakpoint.
Antimicrobial screen plates were manually read to identify growth of morphologically presumptive E coli colonies. The pilot study determined that in a given water sample, multiple E coli colonies are most often clonal when typed by pulsed field gel electrophoresis regardless of the antimicrobial plate from which the isolates were selected, and results from the agar screen method were highly correlated with those generated by the microbroth dilution method (Table 1). Because the pilot study also determined that the highest frequency of resistance was detected on tetracycline media, if growth was seen on multiple antimicrobial screen plates, a single isolate was picked from the tetracycline plate and the selected isolate was assumed to be representative of the predominant E coli clone in the water sample. To ensure that the agar screening method did not miss any resistance, one presumptive E coli isolate was also selected from one in 10 (in 2004) and one in 20 (in 2005) water samples found to have no resistance by the agar screen method, and submitted for testing by microbroth dilution.

Statistical analysis
Fisher's exact test was performed to determine whether the proportions of resistant E coli isolates and the proportion of beaches with at least one resistant E coli isolate were different between 2004 and 2005. Differences among the proportion of E coli isolates resistant to the three antimicrobials most represented were assessed with a McNemar test. To determine whether the level of resistance differed across the various beach sanitary groups, a χ 2 test was performed. All statistical procedures were performed using SAS version 9.1 (SAS Institute Inc, USA).

ReSULTS
According to the provincial surveillance program, sampling for both summers happened from mid-June to mid-August and was continuous throughout these periods. In 2004, 201 beaches were included in the study and were sampled, on average, three times in the summer (range: one to eight), with a mean of eight samples per harvesting (range: six to 12). In 2005, 177 beaches were included in the study and were sampled, on average, three times in the summer (range: one to six), with a mean of eight samples per harvesting (range: six to 18). A total of 674 E coli isolates from the 201 beaches in 2004, and 628 E coli isolates from the 177 beaches in 2005 were analyzed for antimicrobial resistance by microbroth dilution, with a mean of three isolates per beach for each summer. In total, 237 unique beaches were sampled during the study period over the two summers and 141 beaches were sampled in both 2004 and 2005. The smallest beach was 10 m long, the longest was 400 m, and the majority (60%) were <60 m. Figure 2 shows the distribution of antimicrobial resistance to the 16 antimicrobials tested for all resistant E coli isolates for both summers.
The three antimicrobials most represented were tetracycline, ampicillin and sulfamethoxazole.    Table 3 shows the distribution of general resistance for each beach group.
A χ 2 test was performed to determine whether the level of resistance differed across the various beach sanitary groups, which found no statistically significant difference in the percentage of beaches with at least one E coli isolate resistant to one or more antimicrobial(s) between the group 'A' (Excellent) and 'B' (good) beaches for the year 2004 (P=0.251) and 2005 (P=0.192). Statistical testing was not performed for groups 'C' and 'D' due to very low numbers of beaches falling into these categories during the two years of the study. In terms of consistency of antimicrobial resistance results for the same beach, 22 beaches had at least two E coli resistant isolates in 2004 and of these, 10 (45.4%) had all their isolates resistant to the same antimicrobials, and nine (40.9%) had all their isolates resistant to two or more antimicrobials. In 2005, 24 beaches had at least two E coli resistant isolates and of these, 12 (50%) had all their isolates resistant to the same antimicrobials and 11 (45.8%) had all of their isolates resistant to two or more antimicrobials. Of the 141 beaches sampled in both summers, 17 (12.1%) had at least one isolate in each summer with general resistance and 65 (46.1%) beaches from both years had all their isolates susceptible to the 16

DISCUSSION
Overall, these results are in agreement with another study showing that 14% of E coli isolates from Great Lakes recreational waters (Ontario) carried antimicrobial resistance genes and the most frequently found genes were those coding for resistance to tetracycline, ampicillin and streptomycin (12). Resistance to these three antimicrobials is also present in other sources of human exposure. Interestingly, tetracycline and ampicillin were the two types of antimicrobials against which resistance was detected in a high percentage of E coli isolated from pork and beef retailed meat products, according to figures presented in the 2004 and 2005 reports of the Canadian Integrated Program of Antimicrobial   (7) Trimethroprim-sulfamethoxazole 8 (5) 2 (7) 0 1 11 (5) (2,13). Our results are also in agreement with a previous study investigating the prevalence of antimicrobial-resistant bacteria in drinking water of private wells in Ontario, which found that the highest frequencies of antimicrobial resistance were for the similar antimicrobials (tetracycline, sulfamethoxazole and streptomycin) (Nguon RS, unpublished master thesis, Université de Montréal). We also noted that over the two years investigated, an average of one beach out of three demonstrated the presence of E coli with general resistance (resistance to a least one antimicrobial) and that more than 10% of the resistant isolates were resistant to at least one antimicrobial in the very high importance category for human medicine. These findings are noteworthy in terms of public health, given that these antimicrobials are important for the treatment of serious bacterial infections in humans. There is a concern that emergence of resistance to these agents may result in limited, or lack of, alternative effective treatment options. In light of these preliminary findings, it also appears that public beaches in southern Quebec may represent a source of exposure to antimicrobial-resistant bacteria, which could constitute a health risk for people engaging in water activities at these locations. Although there are currently no published data confirming direct transfer of antimicrobial resistance genes from the environment to humans, some studies have shown that the in vitro and in vivo intestinal transfer of genetic material with resistance genes is possible among transient or commensal resident bacteria (7,(9)(10)(11)(12)(13)(14)(15)(16). Evidence for transfer of CMY-2 AmpC beta-lactamase plasmids between E coli and Salmonella species isolates from food animals and humans have been reported (7); these plasmids can readily move between different organisms and can harbour multiple antimicrobial resistance genes. In a previous report, an acquired, plasmid-mediated, AmpC beta-lactamase gene, CMY-2 has also been identified in E coli from water samples (18). Similarities of the CMY-2-containing plasmids identified in E coli isolates of water origin, as well as in human and animal clinical isolates, highlight the potential for transmission of multidrug resistance phenotypes from water sources. Antimicrobial-resistant bacteria ingested with water during bathing activities could thus transfer their antimicrobial resistance genes to bacteria of the intestinal flora of bathers. Transferred genes can act as a reservoir of genetic material available to transient or colonized bacterial populations, including pathogens. Horizontal transfers like these can involve genetic material such as plasmids and transposons that carry more than one marker of resistance and have the potential to transfer many mechanisms of resistance against different antimicrobials in a single transfer (17). The presence of antimicrobial-resistant E coli may also indicate the presence of other antimicrobial-resistant bacteria, due to a transfer of antimicrobial resistance genes in the water environment (19)(20)(21). Overall, antimicrobial resistance in bacterial populations could have serious public health consequences, including prolonged disease duration and increased frequencies of septicemia, hospitalization and death (22).
One of the limitations of the present study was the choice of E coli as the indicator of the presence of antimicrobial-resistant bacteria. E coli are used as a biological indicator for fecal contamination in water sources and are also known to be a reservoir of resistance genes, but they are also naturally resistant to some antimicrobials, such as glycopeptides. For this reason, we did not test our water samples for this class of antibiotic. A combination of two indicators, such as E coli (Gram-negative) and Enterococcus species (Gram-positive) could have resulted in a more accurate estimate of the occurrence of antimicrobial resistance level in this setting, allowing us to test for a broader antimicrobial spectrum. The extent of resistance in beach water samples may have been underestimated given that we selected only one representative E coli isolate from each water sample to be screened for resistance. This isolate was only a representative of the composite of all the E coli from a water sample, and the antimicrobial resistance phenotype represented, at the least, the predominant clone within the sample. The intent of the present study was to establish whether water is a potential reservoir for antimicrobial resistance, and if so, to determine the spectrum of antimicrobial resistance phenotypes. Further studies would be needed to determine the ecology of antimicrobial-resistant E coli in this environment, and the dynamics of transmission and persistence over time. Another limiting element of the present study relates to the sampling frequency of beaches under investigation. Because the sampling frequency of a given beach in the provincial surveillance program is determined by the result of the previous year's water quality, beaches with a better quality (the most common in the present study) are only sampled a few times. This low power of detection may have affected the representativeness of beaches with excellent water quality and introduced a bias in the estimation of resistance prevalence, giving a lower prevalence of beaches with at least one water sample resistant to antimicrobial.
Beach contamination with fecal bacteria carrying antimicrobial resistance characteristics involves an interaction of complex phenomena, including factors related to human and animal population densities, medical and veterinary use of antimicrobials, meteorological events and landscape features. Public health risk related to human exposure to water containing antimicrobial-resistant bacteria is difficult to quantify. Ongoing initiatives and surveillance programs aimed at monitoring microbial water quality should include activities specific to the detection of resistance bacteria from recreational water to further assess the significance and the geographical extent of this concern in various populations.