Antibiotic susceptibility situation of environmental Legionella pneumophila isolates in Southern Germany

Legionella belongs to a genus of aerobic, Gram-negative bacteria found in freshwater systems, from which they can colonize free-living amoebae and animal hosts (Price & Abu Kwaik 2021). Legionella can cause severe pneumonia, called Legionnaires' disease (LD), in humans as well as a milder influenza-like disease, called Pontiac fever. Humans are infected by inhaling aerosols carrying Legionella (Graells et al. 2018). More than 50 different species and more than 70 antigenic types of Legionella have been described, but  Legionella pneumophila serogroup 1 (Lp 1) causes most LD cases in the USA and in Europe (Haroon et al. 2012; van Heijnsbergen et al. 2014).

As an intracellular pathogen, L. pneumophila replicates in protozoa or macrophages (Jules & Buchrieser 2007). This makes the antibiotic treatment of patients with legionellosis challenging. The chosen antibiotics need to reach therapeutic intracellular levels. Usually, the antibiotics of choice are macrolides and fluoroquinolones such as levofloxacin (LEV) or ciprofloxacin (CIP) (Sreenath et al. 2019). Macrolides are naturally occurring substances found in wastewater and drinking water (Kümmerer 2009). Therefore, L. pneumophila is exposed to those antibiotics that can contribute to the development of antibiotic resistance. In vitro studies confirmed the selection of L. pneumophila mutants with a high level of resistance by exposing them to subinhibitory concentrations of macrolides (Descours et al. 2017).

Due to the lack of a standardized antibiotic susceptibility testing method for Legionella strains, minimum inhibitory concentrations (MICs) and epidemiological cut-off values (ECOFFs) have not been established for L. pneumophila (Wilson et al. 2018; Portal et al. 2021; EUCAST – European Committee on Antimicrobial Susceptibility Testing). Nevertheless, reduced susceptibility to macrolides, especially azithromycin (AZM), has been observed (Vandewalle-Capo et al. 2017). This finding is strongly related to the presence of a macrolide efflux pump encoded by the lpeAB genes (Vandewalle-Capo et al. 2017). As antimicrobial resistance (AMR) has been officially classified as an increasing problem worldwide by the World Health Organization (WHO), the development of various resistances should be closely monitored (Lorentzen et al. 2023).

Most LD cases are considered to be community acquired; however, travel-associated Legionnaires' disease (TALD) increased by 38% from 2020 compared to 2021 (ECDC, 2023; Jong & Hallström 2021). Common sources of infection for TALD are hotel showers or whirlpools (Guyard & Low 2011). During an LD outbreak in Germany in 2012, a cooling tower was found to be the source of infection (Burckhardt et al. 2016). Nevertheless, exposure to contaminated aerosols in hospitals and long-term healthcare facilities presents a higher risk for LD outbreaks, as the most prevalent group to contract LD are males aged 65 years and above (Pedro-Botet et al. 1995). Therefore, the ECDC recommends regular control of the engineered water systems in such facilities (ECDC, 2023).

This study aimed to investigate the antibiotic susceptibility situation of environmental L. pneumophila isolates from building water systems in Southern Germany. All isolates were obtained from hospitals or long-term healthcare facilities. In hospitals, or healthcare facilities in general, increased antibiotic susceptibility of L. pneumophila would pose a great threat. In total, 10,723 LD cases were reported among 29 European countries in 2021, of which 93% were confirmed (ECDC, 2023). The Centers for Disease Control and Prevention (CDC) recommends macrolides and fluoroquinolones for treatment (CDC 2022). Generating data and monitoring the antibiotic susceptibility situation of environmental L. pneumophila isolates could help in the early detection of possible health risks due to increasing antibiotic resistance and, therefore, help to optimize the LD treatment.

For this study, in total 39 environmental L. pneumophila strains were analyzed. All strains were isolated from public drinking water supplies in hospitals and long-term healthcare facilities in Southern Germany from August 2023 to April 2024. Water samples were collected by trained personnel and brought to the laboratory where the presence of L. pneumophila was determined by culture techniques and confirmed by MALDI-TOF. Each isolate was labeled with a Y number followed by 4–5 digits.

Additionally, L. pneumophila ATCC 33152 (WDCM 00107) (serogroup 1) and L. pneumophila ATCC 33156 (WDCM 001809) (serogroup 4) were tested as reference strains. The isolates were characterized as L. pneumophila serogroup 1 (Lp 1) and L. pneumophila serogroups 2–14 (Lp 2–14) according to their serological agglutination with the latex agglutination test (DrySpot L. pneumophila 1, DrySpot L. pneumophila 2–14, Oxoid, Hants, UK). Isolates were stored at −20 °C in Microbank™ tubes (Pro-Lab Diagnostics™ Microbank™, Pro-Lab, ON, Canada) until MIC testing. Before MIC testing, the isolates were subcultured on BCYE-α agar (Xebios, Duesseldorf, Germany) and incubated for 48 h at 36 °C in a humidified atmosphere.

The following nine antibiotics were used in this study: AZM, clarithromycin (CLR), erythromycin (E), CIP, tigecycline (TGC), LEV, moxifloxacin (MXF), rifampicin (RD), and doxycycline (DXT) (Merck, Darmstadt, Germany). Antibiotic powders were dissolved in dimethyl sulfoxide or water according to their solubility. Stock solutions of 32 mg/L were prepared for each antibiotic and were subjected to serial two-fold dilutions to achieve a final concentration of 0.008 mg/L.

The microbroth dilution method was performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (EUCAST – European Committee on Antimicrobial Susceptibility Testing). Briefly, the isolated strains were suspended in Mueller Hinton II broth supplemented with Legionella Growth Supplement (BCYE) (Oxoid, Hants, UK) and adjusted to a McFarland of 0.8. The bacterial suspension was further diluted 1:10 to achieve a final bacterial concentration of 1 × 10⁶ CFU/mL. Each well was inoculated with 50 μL of the antibiotic solution (concentrations ranging from 0.008 to 32 mg/L) and 50 μL of the bacterial suspension. The plates were incubated for 48 h at 36 °C in a humidified atmosphere. The MIC was determined manually as the first well that showed no visible growth.

The statistical analyses were performed using RStudio Statistical Software (R: A Language & Environment for Statistical Computing 2023). First, the Shapiro–Wilk test was used to evaluate the null hypothesis, where the dataset is derived from a normal distribution. The Mann–Whitney U test was used to evaluate the significance of differences between the MIC₉₀ values of Lp 1 and Lp 2–14 strains. For each test, the cut-off value for significance was defined as p < 0.05.

Overall, the collected environmental L. pneumophila isolates showed elevated MIC₉₀ values compared to the reference strains, with a few exceptions. DXT, for example, is an exception since there were no differences between the collected isolates and the reference strains. The 39 collected L. pneumophila isolates included 11 Lp 1 (28%) and 28 Lp 2–14 strains (72%).

Table 1 shows the cumulative percentages of strains inhibited by each agent (AZM, CLR, E, CIP, TGC, LEV, MXF, RD, and DXT) at the stated antimicrobial concentration. For RD and LEV, complete inhibition of growth was observed at an antibiotic concentration of 0.125 mg/L. The other two fluoroquinolones used in this study (CIP and MXF) completely inhibited the growth of all tested isolates at a concentration of 0.5 mg/L. CLR, a macrolide, inhibited 100% of the used isolates at the same concentration, while the macrolides AZM and E were less effective than CLR and a concentration of 2 mg/L was required to inhibit the growth. The least effective antibiotics were DXT, for which full inhibition was only observed at a concentration of 8 mg/L, and TGC, for which a concentration of ≥16 mg/L was necessary for inhibition of all isolates.

MIC₉₀ values for each antibiotic agent are shown in Table 2. The most active agent was RD with a MIC₉₀ of 0.016 mg/L for L. pneumophila serogroup 1 (Lp 1) and L. pneumophila serogroups 2–14 (Lp 2–14) as well as L. pneumophila ATCC 33152. The MIC₉₀ for the reference strain L. pneumophila ATCC 33156 was even lower than 0.008 mg/L.

The three fluoroquinolone antibiotics (CIP, LEV, and MXF) had similar MIC₉₀ values for Lp 1 strains. MIC₉₀ values for LEV and MXF were 0.125 mg/L and for CIP it was 0.25 mg/L. MXF showed no difference between the serogroups and, therefore, Lp 2–14 strains also had MIC₉₀ values of 0.125 mg/L. LEV and CIP MIC₉₀ values for Lp 2–14 strains were lower (0.064 and 0.125 mg/L, respectively). LEV MIC₉₀ values were lowest for the reference strains (0.032 mg/L for L. pneumophila ATCC 33152 and 0.016 mg/L for L. pneumophila ATCC 33156). The MIC₉₀ values for both CIP and MXF were 0.064 mg/L for L. pneumophila ATCC 33152 and 0.032 mg/L for L. pneumophila ATCC 33156.

Within the class of macrolides, CLR had the lowest MIC₉₀ value (0.25 mg/L for both Lp 1 and Lp 2–14 strains). CLR had a MIC₉₀ value of 0.032 mg/L for L. pneumophila ATCC 33152 and 0.016 mg/L for L. pneumophila ATCC 33156. AZM had a MIC₉₀ value of 0.5 mg/L for Lp 1 and Lp 2–14 strains. Also, both reference strains had the same MIC₉₀ value of 0.125 mg/L. The MIC₉₀ value obtained for E was the highest within the class of macrolides with 1 mg/L for Lp 1, Lp 2–14 strains, and L. pneumophila ATCC 33152. L. pneumophila ATCC 33156, on the other hand, had a much lower MIC₉₀ value of 0.125 mg/L for E.

The overall highest MIC₉₀ values were observed for TGC, with ≥16 mg/L for Lp 1, Lp 2–14 strains, and L. pneumophila ATCC 33152. The MIC₉₀ value for L. pneumophila ATCC 33156 was 8 mg/L.

The efficiency of the antibiotic agents according to their overall MIC₉₀ values can be summarized as follows: RD < LEV < MXF < CIP < CLR < AZM < E < DXT < TGC.

There was no significant difference (p < 0.05) between Lp 1 and Lp 2–14 strain MIC₉₀ values for the tested antibiotic agents, except for DXT (p = 0.01).

The German Ordinance on the Quality of Water Intended for Human Consumption (Verordnung über die Qualität von Wasser für den menschlichen Gebrauch (Trinkwasserverordnung – TrinkwV) 2023) mandates testing for Legionella spp. in hospitals and various other facilities. The incidence of LD in Germany is lower than the European mean value. In Germany, 1.6 LD cases occur per 100,000 inhabitants, whereas the European average is 1.8 infections per 100,000 inhabitants (ECDC, 2019). LD should be considered a concerning infectious disease and monitoring of antibiotic resistance levels is necessary to ensure proper treatment.

The distribution of the tested L. pneumophila serogroups shows that the majority of isolates belongs to the less pathogenic serogroups 2–14 (n = 28). Only 28.21% of the isolates are considered to be serogroup 1 (n = 11). Previous studies also showed Lp 2–14 strains as the predominant serogroup within L. pneumophila (Assaidi et al. 2020; Cocuzza et al. 2021; Cruz et al. 2023). To obtain these results, latex agglutination was performed due to its rapid test results. Also, there are other methods to distinguish L. pneumophila serogroups, such as PCR-based assays, MALDI-TOF–MS identification, or Fourier-transform infrared spectroscopy (Tata et al. 2023). These methods are not considered to be the Gold standard. The Gold standard is ELISA-based typing (e.g. Dresden panel) or agglutination (Helbig et al. 2002). Cross-reaction between antibodies can occur using the latex agglutination test, but still it is considered to be the reliable method (Tata et al. 2023).

Overall, RD achieved the lowest MIC₉₀ values and is, therefore, considered to be the most effective antimicrobial agent according to its MIC₉₀ values. There are no known cases of RD resistance in clinical L. pneumophila strains (Bruin et al. 2012). As the usage of RD in LD therapy is limited, resistance development may not be seen in L. pneumophila strains due to the lack of exposure (Sreenath et al. 2019). There are various drug interactions between RD and opiates, cyclosporine, or antiretroviral protease inhibitors. Therefore, RD is not common for monotherapy but can be used in combination with macrolides or quinolones (Chahin & Opal 2017). Despite the lack of standardization for L. pneumophila susceptibility testing, the results of the present study are in accordance with previously published data (Portal et al. 2020; Cocuzza et al. 2021).

Among the antibiotic class of fluoroquinolones, the lowest MIC₉₀ values were measured for LEV. This is, therefore, considered to be the second most active agent after RD. This finding is similar to results from previous studies (De Giglio et al. 2015; Sikora et al. 2017; Sreenath et al. 2019). Both reference strains showed a similar MIC₉₀ value for MXF and CIP, but the isolated Lp 1 strains had a slightly elevated MIC₉₀ value for CIP compared to MXF. Fluoroquinolone-resistant L. pneumophila strains still pose an exception and are rarely found or described in the current literature (Shadoud et al. 2015). Overall, this study showed an increased activity of fluoroquinolones compared to the activity of macrolides against all tested L. pneumophila isolates, considering the MIC₉₀ values alone.

Of the macrolides tested, CLR was the most effective according to the MIC₉₀ value, followed by AZM. For AZM and CLR, the obtained MIC₉₀ values for L. pneumophila ATCC 33152 agree with previously published values (Cruz et al. 2023). Macrolide susceptibility is linked to an efflux pump system, encoded by the lpeAB genes. As, for example, E is produced by the soil bacteria Streptomyces spp., the efflux pump is likely to be an adaptation to protect L. pneumophila from antibiotics in the environment (Massip et al. 2017). In this study, the MIC₉₀ values for E, with the exception of L. pneumophila ATCC 33156, were higher compared to previously conducted studies (Bruin et al. 2012; Sreenath et al. 2019). The lpeAB genes might be overexpressed in the collected isolates due to previous exposure to macrolide antibiotics. Macrolide antibiotics are commonly used in hospitals and most isolates were obtained from public drinking water systems in hospitals. It would be of interest to determine the protein expression levels of the collected isolates and also screen for mutations around the genomic region of lpeAB to elucidate the mechanisms behind this apparent decrease in the sensitivity of E.

Interestingly, even though TGC and DXT belong to different classes of antibiotics, their mode of action is rather similar. Both antibiotics target the ribosome and bind to the 30S ribosome subunit. MIC₉₀ values for TGC are generally higher than those for macrolides or fluoroquinolones (Bruin et al. 2012). The elevated MIC₉₀ values for TGC in this study could be due to the presence of charcoal in the growth medium. The activity of TGC is influenced by charcoal (March et al. 2019). Nevertheless, this study chose to use the medium containing charcoal in accordance with the EUCAST guidelines for susceptibility testing of L. pneumophila (EUCAST – European Committee on Antimicrobial Susceptibility Testing).

DXT was the only antibiotic agent with a significantly lower MIC₉₀ value for Lp 2–14 strains compared to Lp 1 strains. Previous studies reported a significantly lower MIC₉₀ value for DXT from non-L. pneumophila isolates compared to Lp 1 isolates (Xiong et al. 2016). Since Lp 1 is considered the most pathogenic serotype, the mechanism leading to the elevated DXT MIC₉₀ values in Lp 1 could be of interest in decoding L. pneumophila virulence factors. The MIC₉₀ result for L. pneumophila ATCC 33152 against DXT is in accordance with previously published values (March et al. 2019; Cruz et al. 2023). DXT is not commonly used in LD treatment, and there is only limited clinical experience (Chahin & Opal 2017).

L. pneumophila is an environmentally acquired pathogen; therefore, environmental selection pressure in drinking water systems can be relevant (Park et al. 2020). Antibiotic resistance selection pressure in drinking water can be driven by several factors, such as heavy metals, biocides, or disinfectants Sanganyado & Gwenzi (2019).

Overall, this study contributes a piece of information toward a better understanding of AMR testing of environmental L. pneumophila strains. As the obtained MIC₉₀ values are comparable to previously published values using the same AMR testing method, the microbroth dilution method should be implemented as the Gold standard for L. pneumophila susceptibility testing (Cocuzza et al. 2021; Portal et al. 2021; Cruz et al. 2023).

Further studies should include not only environmental L. pneumophila strains but also clinical strains to promote the determination of MICs and ECOFFs. More detailed studies of strains with decreased susceptibility against specific antibiotics are needed to elucidate the mechanism behind resistance development and to evaluate the risks posed by changes in resistance patterns among environmental Legionella strains. The monitoring of environmental strains is a useful tool for the early detection of antibiotic resistance levels in L. pneumophila.

The majority of the isolated L. pneumophila strains belonged to the less pathogenic serogroups 2–14. The different serogroups showed no significant differences in their MIC₉₀ values for most of the tested antibiotics, with the exception of DXT. DXT was the only antibiotic agent that had a significantly higher MIC₉₀ value for Lp 1 compared to Lp 2–14 strains. Overall, the MIC₉₀ values for Lp1 and Lp 2–14 strains were highest for TGC and lowest for RD. As TGC is not the antibiotic of choice for LD treatment, treatment failure of LD is not an immediate threat.

These findings suggest that L. pneumophila strains in Southern Germany do not display elevated MIC₉₀ values compared to similar studies and, therefore, have not yet developed increased antibiotic resistance levels. Monitoring the antibiotic susceptibility situation of L. pneumophila should be performed regularly for early detection of resistance development. Generating more data can be useful in defining MIC and ECOFF values for L. pneumophila.

The authors would like to thank Linda Pfülb and Sara Gundalach for their assistance in collecting the tested Legionella pneumophila isolates.

H.L. conceptualized the study, did data curation, performed analyses and methodology, and wrote the original draft. R.E.M. wrote and did a critical revision of the article. D.B. did analyses. A.H. did project administration. E.M. supervised the study, wrote and did a critical revision of the article. M.F. conceptualized and supervised the study, wrote and did a critical revision of the article.

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

The authors are employed by a commercial veterinary diagnostic laboratory. This employment did not influence data collection, interpretation, or publication preparation. The authors declare there is no conflict.