This study allows for a description of the population structure of P. aeruginosa clinical isolates collected from hospitals of north-western Poland. It revealed the presence of 30 clusters of related strains and 86 unique strains. The obtained results indicate a predominantly non-clonal population structure and continuous exchange of P. aeruginosa strains between patients of the same and different hospital wards, and even between different hospitals in north-western Poland. Similar results have been presented in the previous epidemiological study conducted in this region [23]. Authors indicated a frequent transmission of P. aeruginosa strains between patients in medical centres. Our results are also in concordance with other epidemiological studies demonstrating a non-clonal population structure of P. aeruginosa. [12, 15-16]. It has been reported that the relatively high recombination frequency of P. aeruginosa strains is considered to be a key driver of the multiclonal population structure [12, 13].
Our results indicate continuous isolation of clonally related P. aeruginosa pathogens from different patients on the same ward. Hospital staff were not aware about the spread of pathogens during that time and did not report any outbreak. The two largest PFGE types, D and E, were described as endemic. Strains of those groups were isolated 17 and 13 times in hospital wards during a period of one year; however, no increase in infection rates was reported by the hospital staff. Isolates of cluster E were isolated 12 times from patients at ICUs of a hospital in Szczecin (H3), which stands for 25.4% of all isolates from that hospital. Strains of cluster D represented 27.1% of all isolates from the burn unit and ICUs of a hospital in Gryfice (H2) (Figure 2). In other studies, the most prevalent PFGE types were shared by 22%-52% of all P. aeruginosa isolates in a single hospital or medical unit [24-26]. Strains in these studies were continuously isolated from patients of hospital units and were described as endemic. Due to the hospital selection pressure, the resistance of endemic strains to multiple antimicrobial agents appears to be the determining factor of their endemicity [27].
Patients of intensive care and burn units are particularly at risk of nosocomial infections caused by P. aeruginosa [6, 18]. The results of this work demonstrate that the transmission of P. aeruginosa strains was most frequent in intensive care and burn units. Endemic strains of clusters D and E were also mostly isolated from patients of medical wards of these types.
In other medical units, the transmission of strains between patients was also possible, however, less frequent. The risk of colonisation by P. aeruginosa hospital strains at ICUs is higher presumably due to the prolonged stay of patients, the severity of their illness and exposure to invasive medical procedures [28]. Detection of strains disseminated in hospital wards seems critically important for immunocompromised patients and other patients susceptible to P. aeruginosa infections.
In this study clonally related strains were also isolated from patients of two or more hospitals, which indicates spread of these pathogens between medical centres in north-western Poland. It is thought that related strains can be transferred between different hospitals via hospital staff or their residents [29].
Epidemic and endemic strains are often multiple drug resistant which is responsible for the increased mortality of those patients. Inappropriate empirical therapy is perceived as the main factor contributing to increased mortality [30]. This study, in a similar way, demonstrates that MDR strains were more frequently found in clusters of clonally related strains. Among MDR P. aeruginosa strains, 71.13% exhibited clonal relatedness with at least one other strain. Strains of cluster D and E were resistant to imipenem and aminoglycosides. Resistance to other antimicrobials was also common in these clusters.
Detection and elimination of dissemination of high-risk clones in many cases is not possible with the use of simple epidemiological data alone [12, 31]. This is also the case in this study, where we demonstrated the presence of endemic strains at ICU and burn units that were not detected by hospital staff. Epidemiological studies with the use of molecular methods were not conducted at these wards at the time when clonally related strains were isolated. Thus, the use of molecular typing methods is necessary, especially at various ICU and burn units, to establish a possible transmission of clonally related strains between patients that may appear in the future. The establishment of new infection prevention and control strategies should also be considered.
The obtained data regarding antimicrobial resistance indicate that levels of resistance of P. aeruginosa strains isolated from north-western Poland are comparable with results for the whole country [32]. According to a 2017 ECDC survey, the average resistance rate to carbapenems (imipenem+meropenem) was 24.2% across Poland, whereas in this research, 28.71% of strains were resistant to carbapenems. Discrepancies between results can be due to different infection control management in hospitals, misuse of antimicrobial agents, sanitation and distributions of strains in the region [33].
The lowest resistance rate among used aminoglycosides has been observed for amikacin. Relatively lower amikacin resistance rates of P. aeruginosa strains were also noted in other studies. Sader et al. [34] compared resistance rates of P. aeruginosa strains originating from multiple medical centres in the USA, and recorded susceptibility rates to gentamicin, tobramycin, and amikacin of 88%, 90%, and 98%, respectively. Similar data was obtained from research conducted in China, where amikacin was the second (after colistin) most effective antibiotic [35]. Lower amikacin resistance rates comparing to resistance to other aminoglycosides was also observed for multi-drug resistant P. aeruginosa strains [36]. Results presented in this and other epidemiological studies suggest that the development of amikacin resistance is less common than for other aminoglycosides in P. aeruginosa strains, and therefore that the use of amikacin could provide a better chance of success in empiric therapy. The lowest resistance rate indicated was that for colistin (~3%). Previously reported colistin resistance rates among various P. aeruginosa strains worldwide varied between 0% to 36% [37, 38]. Colistin is used to treat P. aeruginosa infection due to the MDR profiles of many strains, which mean alternative antibiotics cannot be prescribed. However, colistin is not used routinely due to its diverse side effects, including neuro- and nephrotoxicity [39].
This work has certain limitations. We acknowledge that the collected strains represented only a part of P. aeruginosa strains isolated from patients at hospitals in north-western Poland. We cannot assure that some unique strains were not in fact clonally related with other strains from other medical centres. However, this is the largest P. aeruginosa genotyping study that was ever performed in the region of north-western Poland. The investigated amount of strains allowed us to demonstrate a non-clonal population structure and reveal the presence of endemic P. aeruginosa strains in hospitals of the selected region. Secondly, this analysis relied on clinical P. aeruginosa strains collected only from patients. Environmental samples were not collected for this study. Additional strains from nosocomial environments could help us determine the possible hospital sources of P. aeruginosa strains in hospitals. Furthermore, it would be interesting to look into the mechanisms of resistance of multidrug-resistant strains, especially clonally related strains that were spreading over the ICU and burn units. Metallo-β-lactamase (MBL) producers are commonly found in hospitals of Western-Europe, and are frequently responsible for outbreaks in ICU units [40-42]. Therefore it would be interesting to know the proportion of MBL producers in our hospitals.
The use of PFGE method for genotyping could also be considered as a drawback of this study. It is time-consuming, relatively expensive, and the results are difficult to compare interlaboratory [43]. Additional implementation of other typing methods would have additional value to the study. Performing MLST would allow us to compare results internationally.
However, there are a number of genotyping studies in which PFGE is still used [44, 45]. Although this method was first introduced in 1986, it is still frequently used for typing clinically relevant microorganisms.