Identification and assessment of antimicrobial resistance bacteria in a hemodialysis water treatment system

Recently, the number of people with chronic kidney disease who undergo hemodialysis (HD) has increased significantly. The quality of water used for dialysis is an essential factor to consider since the blood of dialyzed patients encounters 300–600 liters of water per week through a non-selective membrane at each stage of the purification process (e.g., dialysis water, generator, and vascular access) (Ward 2011), which causes the risk of microorganism transmission to be high. Bacterial infections are a major cause of morbidity and mortality in patients with chronic HD (i.e., the 2nd leading cause after cardiovascular disease). On the one hand, kidney failure directly or indirectly alters neutrophil and lymphocyte functions, and, on the other hand, extrarenal purification induces cytokine release. In addition, dialysis patients often have comorbidities that make them more vulnerable to side effects (e.g., hypertension, diabetes, and cardiovascular disease) (Coresh et al. 2007).

Generally, the input water used for dialysis is usually obtained from municipal tap water. Subsequently, tap water is treated extensively by sand filtration and softening methods. Carbon filtration and reverse osmosis (RO) are applied to meet the water standards for dialysis (Tong et al. 2001; Damasiewicz et al. 2012). Sand filtration is mainly used to remove residual particles, while water softening is primarily used to reduce hardness. A carbon filtration device is installed to adsorb residual chlorine. This step is essential because residual chlorine can corrode the RO membrane and reduce its service life. In addition, excessive transfer of chlorine into the bodies of dialysis patients can lead to hemolytic anemia (Junglee et al. 2010). The membrane used in RO can retain 95–98% of the salt and nearly all microorganisms present in the water (Pontoriero et al. 2004). However, after RO, treated water may still present a health risk concerning antibiotic resistance.

Prior studies, in different countries, have shown the presence of several microorganisms in water and have found that dialysis water is an important means of spreading antibiotic-resistant bacteria (Montanari et al. 2009; Damasiewicz et al. 2012; Heidarieh et al. 2016; Shahryari et al. 2016). Hence, in order to reduce the health risks incurred by patients in HD centers, management and microbiological monitoring of the quality of the dialysis water becomes imperative. This study has been conducted to determine, for the first time, the bacteriological quality of the water used in a public HD center at the Provincial Hospital Center (PHC) of Mohammedia, Morocco, based on phenotypic and molecular identification of the predominant bacteria, as well as the study of their profile and their resistance to antibiotics.

This study was conducted in a public HD center with 15 dialysis machines, which included a device explicitly used for the HD treatment of patients infected with the hepatitis B virus. Each year, the center conducts more than 5,000. A water treatment system is an integrated unit that includes a water tap that is pretreated with a filtration system, water softener, and activated carbon filter, followed by the final RO purification step. The treated water is stored in a tank, distributed to the dialysis machines. After each session, the devices are rinsed and disinfected according to the protocols in force.

This study was conducted over 5 months, from February 2019 to June 2019. A total of 54 water samples were collected, and the water samples (250 mL) were taken under aseptic conditions according to the established water sampling guidelines (DES 2018). The sampling ports are shown in Figure 1. After sampling, the bottles were clearly labeled and transported immediately to the epidemiology and hygiene laboratory in the facility to be analyzed within a maximum period of 1–2 h after collection.

A filtration technique was used to count the suspended bacteria in the water samples. A 100 mL fraction of each diluted sample was filtered through a membrane with 0.45 μm diameter pores, and the membrane was then placed on plate count agar (PCA, Oxoid^®) and incubated at 21±2 °C for 7 days. The number of colonies found was expressed as the average number of colony-forming units (CFU/mL), according to international standards for microbiological cultures of dialysis water and dialysis liquids (ANSI/AAMI/ISO13959 2014). Sterile water for injection was used as a negative control.

Distinct colonies were isolated from each positive sample and replicated on mannitol salt agar for the genus Staphylococcus, cetrimide agar for Pseudomonas, and Mac Conkey for other non-fermenting Gram-negative bacilli (NFGNB) and were incubated at 37 °C for 24 h. In total, 47 colonies were selected and checked for purity by streaking on agar plates and were identified using commercially available biochemical galleries: an API 20NE gallery (Bio Mérieux) for identifying bacteria classified as no fermentative and coagulase for staphylococcal species were used.

A molecular approach from BHI cultures performed isolate identifications. Genomic DNA was extracted using a MAGPurix extractor by following a modified protocol. The quality and quantity of the extracted materials were assessed by a Nanodrop 8000 system (Thermo Fisher Scientific, Waltham, MA, USA). The 16S rRNA gene was amplified using a Bioline HS PCR kit and universal primers, which resulted in ∼1,420 bp fragments. Finally, sequencing was performed using an ABI-3130 XL Analyzer (Thermo Fisher Scientific). DNA extraction, 16S rRNA gene amplification, and sequencing were subcontracted to the Platform of Functional Genomics (Center National pour la Recherche Scientifique et Technique (CNRST) of Rabat city, Morocco). The raw data were processed, and the closest strains were identified using the BLASTN program (Altschul et al. 1990).

The antibacterial susceptibility testing was performed using the Kirby–Bauer disk diffusion method on Mueller–Hinton agar. Thirteen antibiotic discs (Oxoid Ltd, Basingstoke, England) were selected, which represented six classes: the penicillin consisted of ticarcillin (T/75 μg), ticarcillin–clavulanic acid (TCA/75/10 μg), piperacillin (P/100 μg), piperacillin + tazobactam (TZP, 100/10 μg), cephalosporins like ceftazidime (CAZ/30 μg), and cefepime (FEP/30 μg). The carbapenems included imipenem (IMI/10 μg), the antibiotics amikacin (AKN/30 μg), gentamycin (GEN/10 μg) among the aminoglycosides, ciprofloxacin (CIP/5 μg), and belong to the fluoroquinolone class and chloramphenicol (C/30 μg) to phenols. Trimethoprim–sulphamethoxazole (TSU, 75/25 μg) is an antagonist to folic acid biosynthesis. The results were interpreted according to the Clinical and Laboratory Standards Institute guidelines (CLSI 2014). Pseudomonas aeruginosa ATCC 27853 was employed as a control strain. The isolates were classified as extensively drug-resistant (XDR), multidrug-resistant (MDR), and non-MDR according to the method of Magiorakos et al. (2012). According to the definitions, XDR is defined as non-sensitive to at least one agent for all but two or fewer antibiotics; MDR is defined as non-sensitive to at least one of three or more resistant agents; and non-MDR isolates show no sensitivity to one or more agents in one or two antibacterial classes.

None of the HD patients experienced febrile/pyrogenic reactions during sample collection. According to the European standards for HD fluids and those of the Moroccan Health Ministry (ANSI/AAMI/ISO13959 2014; Ministère de santé 2018), the treated final water and the final dialysate were unacceptable if one or more of the following criteria were met: HPC of >100 CFU/ml, presence of coliforms in 100 mL, endotoxin level of >0.25 EU/mL.

Among the 54 samples, 27 (50%) presented positive cultures, and 5.5% (3/54) of the water samples were above standard, with 122, 165, and 178 CFU/mL; these three samples are from point 2 after CF. The culturable bacteria levels averaged only 10 CFU/mL in the sand filter effluents. After carbon filtration, this number increased considerably, with an average of 87 CFU/mL and a peak of 178 CFU/mL. The cultured bacteria decreased after RO treatment but were still close to 60 CFU/mL. After the entire treatment process, the treated water arrived at the generator with approximately 43 CFU/mL on average. Gram staining results showed that 79% of the isolated strains were Gram-negative bacteria, while 21% were Gram-positive bacteria isolated from mixed cultures. During the study period, monitoring the contamination level showed about the same bacterial count level before and after disinfection of the dialysate system.

Out of 54 samples from all locations at the HD center, 47 isolates were studied. According to the phenotypic characteristics and by sequencing the 16S rRNA gene, eight species of NFGNB were identified, and the frequencies of detection are listed in Table 1 according to the type of sample analyzed. The most frequently isolated species was Stenotrophomonas maltophilia (34%), followed by Pseudomonas spp. (25.5%) and 8.5% of Burkholderia cepacia. The identified bacterial species and their accession numbers for some sequenced 16S rRNA genes are shown in Table 2.

As shown in Table 3, Pseudomonas spp. was resistant to most of the tested antibiotics: piperacillin (83.33% resistance), piperacillin–tazobactam (58.33% resistance), ticarcillin (50% resistance), and ticarcillin–clavulanic acid (16.67% resistance) were the most active compounds tested and were followed by chloramphenicol (16.67% resistance), imipenem (8.33% resistance), and levofloxacin (100% susceptible). Interestingly, the S. maltophilia isolate exhibited high resistance to most antibiotics tested; levofloxacin was the active compound (100% susceptible). B. cepacia showed increased sensitivity to nearly all tested antibiotics. Almost half (43.75%) of the isolated strains tested exhibited antibiotic resistance, 31.25% (10/32) of them were classified as MDR, and 12.5% (4/32) remained susceptible to only one or two categories of antibiotic (XDR). Interestingly, S. maltophilia showed a very high degree of resistance to most of the evaluated antibiotics; among 16 isolated strains, 6 were non-MDR, 7 were MDR, and 3 were XDR. Among the 12 strains of Pseudomonas spp., 8 were non-MDR, 3 were MDR, and 1 was XDR (Figure 2).

The microbiological control of HD fluid plays a crucial role in preventing HD-associated illness. However, several studies have suggested that the potential presence of bacteria, even in small amounts, may reduce dialysate quality (Gomila et al. 2005). Thus, this study investigated for the first time the quality of treated water that was used for dilution of HD concentrates at a public HD unit in Mohammedia.

The results of the bacteriological analysis showed that the contamination rate was 5.5% (94.5% compliance). Oumokhtare et al. found a high level of compliance (100% in an HD unit in Fez, Morocco) (Oumokhtar et al. 2013). In Japan, HD centers in nine hospitals reported high contamination levels of dialysates, and 42.5% were above the AAMI standard (Oie et al. 2003). In addition, in Tuscan hospitals in Italy, seven of the nine water treatment plants for dialysis (78%) recorded negative results for all the studied parameters (Totaro et al. 2017).

In general, most cultivable bacterial counts were consistent with the standards. The worrying result in this study was the increased number of bacterial cultures after carbon filtration, which is mainly attributed to chlorine residue removal by carbon filtration. Microorganisms are capable of reproducing in the absence of such growth inhibitors as well as on the rough surfaces of carbon particles, which are the leading causes of biofilm formation (Mezule & Juhna 2016). In addition, the RO effluent exhibited high bacteria count levels (i.e., a peak of 178 CFU/mL), which was incompatible with its role and its capacity to retain particles; the input dialysis generator, as the last point in the loop, also recorded positive cultures of the different strains found at previous points. During this procedure, suspended bacterial reproduction and biofilm formation on tube walls may occur. The cell densities could even reach 106 CFU/cm² in the biofilm attached to the tubes of the HD machine (Man et al. 1998; Smeets et al. 2003). However, the presence of low flow rates and stagnation points causes the distribution network and water treatment plants located downstream of the filters to be vulnerable to bacterial growth and biofilm formation.

The identification of the isolated strains during our study revealed the predominance of S. maltophilia with a frequency of 34%. Our results agree with those of Rocha et al. (2020). The isolates of this species in this study exhibited high resistance to the antibiotics tested: 43.75% (7/16) of strains were MDR, and 18.75% (3/16) were XDR, while 12.5% were resistant to TSU, which is the antibiotic of choice for treating S. maltophilia infections (Anđelković et al. 2019). Recently, several cases of bacterial infections due to S. maltophilia have been reported (Thet et al. 2019; Rocha et al. 2020). This bacterium is related to a wide range of infections, with bloodstream infections and pneumonia being the two most common manifestations (Looney et al. 2009). In addition, infections associated with S. maltophilia pose serious treatment challenges; they are inherently resistant to many antibiotics and can acquire antibacterial resistance by multiple mechanisms (Chang et al. 2015). Therefore, the World Health Organization (WHO) has classified this bacterium as one of the MDR bacteria in hospital settings (Brooke 2014).

Pseudomonas was also isolated with a frequency of 25.53%. These results are similar to those reported by Oumokhtar et al. (2013). Among our 12 strains of Pseudomonas spp., 50% showed high resistance, mainly to piperacillin, piperacillin–tazobactam, ceftazidime, and amikacin, while they remained sensitive at a 91.6% level to imipenem. These results present a cause for concern, given the known resistance of the bacteria of this genus to antibiotics, its virulence factors, its easy acquisition of new resistance factors, and its ability to survive in water for long periods, even without nutrients. Further measures must be implemented to remove it from dialysis water for these reasons. Regarding B. cepacia, the four strains identified were sensitive to most antibiotics; however, this result should not minimize its pathogenicity; it can cause bloodstream infections in HP (Rocha et al. 2020).

The use of 16S rDNA sequence analysis (for the first time) in the dialysis water in Morocco not only allows us to confirm the strains that we identified but also other strains (not identified by conventional tests), such as Bacillus cereus, Staphylococcus warneri, S. pasteuri, and Carnobacterium inhibens. Therefore, the presence of all these bacterial varieties with resistant strains in treated water is a matter of concern in HD centers and should be taken seriously when assessing HD water quality by performing routine monthly disinfections with high-level disinfectants. Other solutions to improve the quality of HD water were recommended: disinfecting machines frequently by updating the disinfection protocol (Asserraji et al. 2014; Verma 2015), biofilm analysis on tubing dialysis machines (Marion-Ferey et al. 2003), improving nursing education on dialysis water (Chamney & James 2008), and conducting monthly quality assurance for risk analysis to the progress of water quality. The choice of piping materials is also essential. International standards (BS EN 12201-3 2003; BS EN 806-5 2012) suggest the use of AISI Type 304/316 stainless steel, copper, polyvinyl chloride, or polypropylene (used in the water treatment system in the present study) in water distribution systems, all of which can withstand physical and chemical shock disinfection treatments.

Our study was set in one dialysis center and there may not be enough data points to generalize our findings to other centers. The other limitation was the shortage of some antibiotics which prevented us from doing the antibiotic test for all the strains in a homogeneous way.

In this study, the hemodialysis water treatment system has been colonized by MDR Gram-negative bacteria. The CFU values for dialysis water exceeded the limit of 100 CFU/mL in three samples after CF. Therefore, strict monitoring of the water delivery system eliminates all bacteria, even those protected in biofilm. To elucidate the bacterial community in dialysis water with precision, we recommend that future studies apply culture-independent protocols to detect viable but non-culturable bacteria (VBNC).

The authors would like to acknowledge the responsibility of molecular biology and functional genomics platform (National Center for Scientific and Technical Research, Morocco), Dr El Fahim M.

L.C., T.C., I.Z., R.A.M., and F.M. conceived and designed the study. N.R., L.C., and R.A.M. undertook experimental procedures. L.C., T.C., and N.R. undertook the analysis. L.C. and R.A.M. drafted the manuscript. All authors approved the final manuscript.

Ethical permission was not required and was not necessary for this study. However, written approval was granted by the Provincial Hospital Center (PHC) Mohammedia administration and by the head of the laboratory where this work was done.

The authors declare that they have no conflict of interest.

All relevant data are included in the paper or its Supplementary Information.