*3.2. ESBL-Producing E. coli, ESBL, and Quinolone Resistance Genes*

In the rural hospital, 76 *E. coli* isolates were ESBL-producing (48%). Among them, *blaTEM* was detected in 97% of isolates and *blaCTX-M* in 76%. Both *blaCTX-M* and *blaTEM* were detected in 75% of isolates. Quinolone-resistance gene (*qepA*) was detected in 72% of ciprofloxacin-resistant isolates. All three genes were detected in 51% of ciprofloxacin-resistant isolates (Table 3).

**Table 3.** Genetic analysis of extended-spectrum beta-lactamase (ESBL)-producing and ciprofloxacin-resistant *Escherichia coli* strains found in hospital wastewater.


WAT: wastewater after treatment; WBT: wastewater before treatment.

In the urban hospital, 39 *E. coli* isolates were ESBL-producing strains (36%). Among them, *blaTEM* was detected in 95% of isolates and *blaCTX-M* in 41%. Both *blaCTX-M* and *blaTEM* were detected in 41% of isolates. Quinolone-resistance gene (*qepA*) was detected in 86% of ciprofloxacin-resistant isolates. All three genes, *blaTEM*, *blaCTX-M*, and *qepA,* were detected in 36% of ciprofloxacin-resistant isolates (Table 3).

#### **4. Discussion**

Our novel findings show that, in Vietnam, bacteria resistant to commonly used antibiotics along with genes coding for resistance are present in hospital wastewater, even after treatment. Prior to our study, Duong et al. examined *E. coli* resistance to ciprofloxacin and norfloxacin in hospital wastewater of another Hanoian hospital and reported that *E. coli* strains isolated from WAT samples were susceptible [22]. In their study, water samples were collected over two days using grab sampling and only 15 *E. coli* isolates, including three isolates from treated water samples, were tested. Conversely, in our study, water samples were collected every month over one year using continuous sampling and a total of 265 *E. coli* isolates were tested.

In both hospitals, *E. coli* isolates were most resistant to co-trimoxazole (around 70% of isolates). Previous reports show comparatively lower prevalence rates. In 2004–2005, an Indian study indicated that 55% of enteric bacteria found in hospital wastewater were resistant to co-trimoxazole [23]. A study from Poland showed that 20% of *E. coli* isolates from hospital wastewater collected before 2013 were resistant to co-trimoxazole [24]. Co-trimoxazole is a combination of sulfamethoxazole and trimethoprim. Sulfamethoxazole, one of the first antibiotics to be developed, was put into clinical use in 1935 and trimethoprim was first used in 1962. The two antibiotics first started to be used in combination in 1968. Over the past decades, its extensive use in clinical settings to treat a variety of bacterial infections, such as urinary tract infections and respiratory tract infections, which was also the case in the studied hospitals, might explain the high occurrence of co-trimoxazole resistance [25]. Moreover, both sulfamethoxazole and trimethoprim are not readily degradable and their residues found every month over the studied period in the same hospital wastewater could favour the development of co-trimoxazole resistance in the bacteria [4].

Cephalosporin resistance was also found in higher proportions than other antibiotics investigated. Similar prevalence rates of cephalosporin resistant bacteria in hospital effluent were shown by Chagas et al. [26]. Resistance mechanisms to second and third generation cephalosporins differ, with ESBLs being the most important [27]. ESBL enzymes are capable of hydrolysing and inactivating beta-lactam antibiotics and are often plasmid-mediated [28]. The plasmid genes encoding ESBLs can be transferred between different bacterial strains (horizontal gene transfer), facilitating easy spread of antibiotic resistance within as well as between species. Moreover, plasmid-encoded ESBL-producing bacteria can show co-resistance to quinolones, aminoglycosides, and sulfonamides [29]. Consequently, infections caused by ESBL-producing bacterial strains can be difficult to treat due to the restricted amount of antibiotics left for successful treatment.

ESBLs were first isolated in the 1980s [30]. In a study conducted in an urban and rural hospital in central India, Chandran et al. reported a very high prevalence of ESBL-producing *E. coli* in the hospital wastewater (96%) [31]. In our study, the prevalence was around 40%, lower than the aforementioned study but relatively higher than the figures reported by Diwan et al. (25%), Abdulhaq et al. (25%), and Korzeniewska et al. (37%) [22,32,33]. Among genes coding for ESBL production, TEM and CTX-M are the most common [34]. Our findings indicate the presence of *blaCTX-M* and *blaTEM* in ESBL-producing strains, with the *blaTEM* gene being predominant. This is in accordance with the findings of Varela et al., where *blaTEM* was found to be the most prevalent ESBL-encoding gene, followed by *blaCTX-M* [35]. In contrast, Chandran et al. reported higher prevalence of *blaCTX-M* than *blaTEM* in hospital wastewater in India [31]. However, not all the *blaTEM* genes are responsible for ESBL, and in our study, we were not able to do further sequencing to show the frequency of the ESBL *blaTEM*. In addition, according to the PCR protocol used, the detected *blaCTX-M* were restricted to CTX-M group 1 including CTX-M-1, CTX-M-3, and CTX-M-15. Ciprofloxacin resistance was detected in our samples, with gene *qepA* coding for high proportions of the resistant strains. Similar to ESBL-coding genes *blaCTX-M* and *blaTEM*, quinolone resistance gene *qepA* is plasmid-mediated and capable of horizontal gene transfer [36]. Of note is that co-existence of *blaCTX-M*, *blaTEM*, and *qepA* was detected, genetically proving co-resistance in the bacterial strains. In our study, the prevalence of MDR found phenotypically was around 35% with the detection of co-resistance to six out of eight of the studied antibiotics.

Fosfomycin-resistance and imipenem-resistance were also detected among the *E. coli* isolates in our study. Fosfomycin has broad activity against Gram-negative and some Gram-positive bacteria. In some countries, it is recommended as one of the first-line drugs to treat uncomplicated urinary tract infections because of increasing *E. coli* resistance towards other commonly used antibiotics, such as ciprofloxacin and co-trimoxazole [37]. Imipenem, the first carbapenem developed, is used to treat infections caused by β-lactamase-producing bacteria and should be saved to treat infections not readily treated by other antibiotics [38]. High prevalence of carbapenem-resistance in clinical isolates in Vietnamese hospitals has been reported [39]. The detection of resistance to these last-line antibiotics in bacterial isolates from hospital wastewater is of concern, since this can contribute to the spread of resistance among bacterial populations in the environment.

Although there is increasing evidence of the occurrence of antibiotic resistant bacteria in the environment, there are no standardized methods for antibiotic susceptibility testing which are directly applicable to environmental samples so far [2]. Epidemiological cut-off values developed by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) can be used for the interpretation of antibiotic resistance in environmental bacteria. The ECOFF values separate bacteria with acquired resistance mechanisms (non-wild type) from the wild type population (having no resistance) [15]. We found in general decreased susceptibility in the *E. coli* isolates when using ECOFF values.

The role of hospitals in the environmental release of antibiotic-resistant bacteria and ARGs has been demonstrated and has become a growing concern for public health [40–47]. Hospital WWTPs can harbour antibiotic-resistant bacteria and ARGs [48–50]. Antibiotic residues in WWTPs can favour the development of antibiotic resistance due to the selective pressure placed on bacteria. In our previous study, we found that antibiotic concentrations in wastewater collected from hospital WWTPs were often higher than the reported predicted no-effect concentrations for resistance selection as well as the minimum selective concentrations, meaning that the selection of antibiotic-resistant bacteria can occur [4]. Published studies have reported the enrichment of antibiotic-resistant *E. coli* in WWTPs [51,52]. Our findings show significant reductions as well as the enrichment of antibiotic-resistant *E. coli* in WAT. Resistance to at least one studied antibiotic in the *E. coli* isolates from WAT was still detected in high proportions. Consequently, certain amounts of antibiotic-resistant bacteria, along with ARGs, are released into the ambient aquatic environment. They can then enter water bodies used for agriculture, irrigation, or household purposes, which poses a threat to public health. The problem can be aggravated if hospitals do not have WWTPs and the wastewater is discharged directly into the environment, which is common practice in Vietnam as well as many other low- and middle-income countries [53]. It has been shown that the prevalence of antibiotic-resistant bacteria is significantly reduced by advanced wastewater treatment processes such as ozone, UV, and ultrafiltration [54,55]. However, even in such advanced plants, resistant bacteria are not completely removed, therefore, hospitals must invest in effective WWTPs with treatment processes that completely eliminate antibiotic-resistant bacteria.

It is plausible that the dissemination of antibiotic-resistant bacteria and ARGs in the environment can result in their transmission to humans [56], however, direct evidence for this is very scarce [57]. So far, the strongest evidence available has shown the genetic similarities between human-related bacterial strains and environmental isolates collected at exposure-relevant sites [58]. Further studies are needed to identify links between the discharge of antibiotic-resistant bacteria by hospital WWTPs, their occurrence in the ambient environment, and their acquisition by humans via environmental exposure.

Our study has some limitations. Importantly, the prevalence of antibiotic resistance in *E. coli* isolates presented here might not be representative for the whole *E. coli* population in the hospital wastewater because of the limited number of *E. coli* isolates from each water sample. Moreover, due to financial constraints, we were not able to study more antibiotics and ARGs than what we have done. Screening for *blaSHV* gene, which is also common in ESBL-producing *E. coli,* and genes coding for imipenem resistance would make the study more comprehensive. We were also not able to do sequencing for *blaTEM* genes to show the frequency of the genes encoding for ESBL. Another limitation is that data from the urban hospital were unavailable for three months, as sampling could not be carried out due to construction work at the hospital. Furthermore, a detailed description of the functioning of the WWTPs was not available for us.
