Antimicrobial Susceptibility of Bacterial Pathogens from Patients with Ocular Surface Infections in Germany, 2020–2021: A Comparison with the Data from Three Previous National Studies

Bacteria are a major cause of superficial eye infections, especially in children. The present study aimed to (i) determine the antimicrobial susceptibility patterns of ocular bacterial pathogens recovered in 20 laboratories during the period 2020–2021 and (ii) compare these results to those from three studies of the same design conducted in 2004, 2009, and 2015 in Germany. Cut-off values defined by EUCAST were used as breakpoints. A total of 1366 bacterial isolates were collected. The most frequent ocular specimens were conjunctival smears (54.3%). Susceptibility rates of Staphylococcus aureus (n = 594), Haemophilus influenzae (n = 178), and Streptococcus pneumoniae (n = 149) to chloramphenicol, gentamicin, kanamycin, neomycin, levofloxacin, ofloxacin, and oxytetracycline were >90% each. Overall, only minor changes in resistance levels were observed in the period since 2004. Therefore, all tested antimicrobials can still be recommended for local therapy of ocular surface infections.


Introduction
Acute infectious conjunctivitis is a common eye disorder in primary care.Although less common than viruses, bacteria are an important cause of conjunctivitis, particularly in children [1].Leading bacterial pathogens of conjunctivitis are Staphylococcus aureus, Streptococcus pneumoniae, and Haemophilus influenzae.Compared to placebo, the use of antibiotic eye drops (e.g., fluoroquinolones or macrolides, alone or in combination with steroids) is associated with improved clinical and microbiological remission rates [2].However, any use of antibiotics may lead to the emergence of resistant organisms [3].It is, therefore, necessary to regularly determine the antibiotic susceptibility of bacterial pathogens in patients with ocular surface infections.
In Germany, the first multi-center study to investigate the occurrence of resistance in bacterial isolates from ocular surface infections (Ophthalmic Study I [OS1]) was carried out in 2004 [4].A second study was performed in 2009 (Ophthalmic Study II [OS2]) and a third one (Ophthalmic Study III [OS3]) was conducted in 2015 [5,6].There were only minor changes detected in the susceptibility rates of Gram-positive and Gram-negative bacteria against gentamicin, kanamycin, levofloxacin, ofloxacin, and chloramphenicol between 2004 and 2015.The present study (Ophthalmic Study IV [OS4]) was carried out in 2020-2021.
The objectives of the present study were to (i) determine the susceptibility of ocular isolates of S. aureus, Staphylococcus epidermidis, S. pneumoniae, H. influenzae, Moraxella catarrhalis, and Pseudomonas aeruginosa, as well as members of the order Enterobacterales, against seven antibiotics used as ophthalmic drugs, and (ii) compare the susceptibility rates to those from the previous studies (OS1-OS3).The antibiotics studied were chloramphenicol, gentamicin, kanamycin, neomycin, levofloxacin, ofloxacin, and oxytetracycline.Furthermore, the susceptibility rates of oxacillin and cefoxitin were determined to phenotype methicillin-resistant staphylococci.
catarrhalis, and Pseudomonas aeruginosa, as well as members of the order Enterobacte against seven antibiotics used as ophthalmic drugs, and (ii) compare the suscept rates to those from the previous studies (OS1-OS3).The antibiotics studied chloramphenicol, gentamicin, kanamycin, neomycin, levofloxacin, ofloxacin, oxytetracycline.Furthermore, the susceptibility rates of oxacillin and cefoxitin determined to phenotype methicillin-resistant staphylococci.

Results and Discussion
Collecting isolates for this study (OS4) was heavily influenced by the COV pandemic.According to the laboratories involved in OS4, the number of eyes swab for testing fell considerably after the start of the pandemic.For this reason, the coll period had to be extended several times, resulting in a total period of 21 months (1 2020 to 31 December 2021).The extension of the collection period was also nece because only 20 out of 40 laboratories that were asked to participate had the capac take part in this study.The limitations of the study, which could potentially ha impact on the results, were missing data on comorbidities of the patients or anti therapy that the patients received.

Bacterial Isolates and Patient Demographic Data
A total of 1366 isolates were collected.This number was slightly higher than three previous studies (Table S1).More than two thirds of the isolates were obtained outpatients and approximately 20% from inpatients, while, for the remaining isolate status of the respective patients was unknown or not documented.The most com sampling material for isolation was conjunctival swabs, followed by eye swabs w further specification (Figure 1; Table S1).The gender ratio of the patients was approximately equal.The majority of pa belonged to the ≤1 year age group, followed by 60-to 79-year-olds.Overall, just ove of the patients were younger than 10 years, with a median age of 3 years.This distrib was in accordance with previous studies.The type of infection was documented in of cases.Of the 774 patients, approximately 84% had conjunctivitis or keratoconjunc (Table S1).The most frequently isolated pathogenic species was S. aureus, followed influenzae and S. pneumoniae (Figure 2; Table S1).The gender ratio of the patients was approximately equal.The majority of patients belonged to the ≤1 year age group, followed by 60-to 79-year-olds.Overall, just over half of the patients were younger than 10 years, with a median age of 3 years.This distribution was in accordance with previous studies.The type of infection was documented in 43.3% of cases.Of the 774 patients, approximately 84% had conjunctivitis or keratoconjunctivitis (Table S1).The most frequently isolated pathogenic species was S. aureus, followed by H. influenzae and S. pneumoniae (Figure 2; Table S1).

MIC Frequency Distributions and Ratios of Susceptibility to Resistant Isolates
Cut-off values (epidemiological cut-off values (ECOFF), tentative epidemiological cut-off values ((T)ECOFF), or screening cut-off values) defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) were used to distinguish between wild-types (hereafter designated as susceptible isolates) and non-wild-types (hereafter designated as resistant isolates) (see Section 3 and Table S2 for details).
The minimum inhibitory concentrations (MICs) that inhibit 50% and 90% of the isolates (MIC50, MIC90), the percentage of susceptible isolates, and the percentage of resistant isolates for the antibacterial agents tested in this study, in comparison with the results of the three previous studies, are presented in Tables 1-3.MIC distributions of the antibacterial agents tested in the present study are presented in Tables S3-S11.

MIC Frequency Distributions and Ratios of Susceptibility to Resistant Isolates
Cut-off values (epidemiological cut-off values (ECOFF), tentative epidemiological cut-off values ((T)ECOFF), or screening cut-off values) defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) were used to distinguish between wild-types (hereafter designated as susceptible isolates) and non-wild-types (hereafter designated as resistant isolates) (see Section 3 and Table S2 for details).
The minimum inhibitory concentrations (MICs) that inhibit 50% and 90% of the isolates (MIC 50 , MIC 90 ), the percentage of susceptible isolates, and the percentage of resistant isolates for the antibacterial agents tested in this study, in comparison with the results of the three previous studies, are presented in Tables 1-3.MIC distributions of the antibacterial agents tested in the present study are presented in Tables S3-S11.
As expected, some of the test compounds showed greater differences in the resistance rates between isolates of the MSSA (cefoxitin-susceptible) and MRSA phenotypes (Tables 1, S4 and S5).
Two MSSA isolates but no MRSA isolates showed resistance to chloramphenicol.Resistance to oxytetracycline was detected in 13 MSSA and 3 MRSA isolates.Due to cross-resistance, resistance rates of neomycin were identical to those of kanamycin (Tables 1 and S1-S3).
EUCAST has also not established cut-off values for kanamycin and neomycin.When comparing the MIC values, cross-resistance between kanamycin and neomycin appears to be less common in S. epidermidis than in S. aureus.The distribution of kanamycin MICs suggests a (T)ECOFF of 8 mg/L.Using this value, the proportion of resistant isolates would be 24.3%(n = 17).All kanamycin-resistant isolates would also be gentamicin-resistant (MIC values 4-64 mg/L), while the neomycin MIC values were ≤0.25 mg/L (n = 10), 0.5-2 mg/L (n = 5), and 16 mg/L (n = 2).
Resistance to fluoroquinolones was estimated based on the number of isolates with levofloxacin MICs > 0.5 mg/L (32.9%).The level of resistance to levofloxacin was approximately the same in all four studies (29-33%) (Table 2).Resistance against oxytetracycline occurred in 41.4% of the isolates (OS2: 14.3%).However, approximately half of the resistant isolates were inhibited at 4 mg/L oxytetracycline, which is right at the presumed (T)ECOFF (MIC > 2 mg/L) (Table S4).It is also worth noting that oxytetracycline MIC values in this study were, on average, one MIC step higher than the MIC values in OS2, suggesting that the isolates inhibited at 4 mg/L oxytetracycline in the present study belong to the wild-type population.
Low resistance rates in S. pneumoniae have also been reported in other studies, for example, in an Australian study that investigated the epidemiology of microbial keratitis, and in a nationwide study in the USA (Antibiotic Resistance Monitoring in Ocular Microorganisms-ARMOR) that monitored the development of resistance between 2009 and 2018 [12,13].The isolates from the susceptibility study of the Paul-Ehrlich-Society 2019-2020 were also 100% susceptible to levofloxacin [8].
2.2.6.Pseudomonas aeruginosa P. aeruginosa is listed as an "expected resistant phenotype" for chloramphenicol, kanamycin, neomycin, and tetracyclines [14].Two isolates (3.1%) were gentamicin-resistant (MICs ≥ 128 mg/L) (Table S10).Both isolates were obtained from inpatients from the same geographical region.One gentamicin-resistant isolate (2.2%) was detected in OS1 but none were detected in OS2 and OS3 (Table 3).Cut-off values of levofloxacin and ofloxacin were 2 mg/L and 4 mg/L, respectively.Two isolates were classified as levofloxacinresistant (3.1% [OS1: n = 6, 13.3%; OS2: n = 4, 11.1%; OS3: 0%), with MICs of 4 mg/L and 16 mg/L.The resistant isolate inhibited at 16 mg/L levofloxacin was also ofloxacinresistant (MIC ≥ 32 mg/L), while the resistant isolate inhibited at 4 mg/L levofloxacin was classified as ofloxacin-susceptible because the ofloxacin MIC was also 4 mg/L (Table S10).Ofloxacin resistance rates found in the three previous studies were consistent with those of levofloxacin (Table 3).In the United States (ARMOR study), fluoroquinolone resistance rates in P. aeruginosa were at a similar level, with no changes noted between 2009 and 2018 [13].In contrast, in the susceptibility study of the Paul-Ehrlich-Society 2019-2020, more than 12% of isolates obtained from outpatients and inpatients were levofloxacinresistant (MIC > 2 mg/L), while the proportion of isolates with a gentamicin MIC above the ECOFF of 8 mg/L was less than 2% [8].

Enterobacterales
Susceptibility data of the Enterobacterales isolates are summarized in Tables 3 and S11-S17.The level of resistance largely depends on the relative proportion of each Enterobacterales species.A resistance rate above 10% was only determined for ofloxacin (13.2%).Resistance to levofloxacin was detected in 8.5% of the isolates.The difference between the two resistance rates is due to the fact that the screening cut-off values for both drugs are 0.25 mg/L, but levofloxacin was, on average, twice as active as ofloxacin.MIC 50/90 values were 0.063/0.25 mg/L (levofloxacin) and 0.125/0.5 mg/L (ofloxacin).In OS1 (n = 161), three isolates (1.8%) were levofloxacin-resistant and 10 (6.1%) were ofloxacin-resistant, and, in OS2 (n = 129), seven were levofloxacin-resistant (5.4%) and eleven were ofloxacin-resistant (8.5%).Enterobacterales were not tested in OS3.
EUCAST has not defined cut-off values for tetracyclines.The oxytetracycline MICs, however, showed a clear bimodal distribution (Table S11).With a concentration of 16 mg/L as the cut-off value, 25.1% of the tested isolates would be considered resistant.However, significant differences in MIC distributions between species were observed (Tables S12-S17).Enterobacter cloacae complex isolates would be 100% susceptible to oxytetracycline at 16 mg/L, while all but one of the P. mirabilis isolates would be oxytetracycline-resistant.

Study Design
The study was designed as a prospective, multi-center in vitro study, like the three previous studies.Of the 20 study centers, 10 were already involved in OS1, 17 in OS2, and 16 in OS3.The methodology of all studies remained the same, with some minor variations regarding the tested agents and species (for details see Tables 1-3).

Clinical Isolates
The study sites were requested to collect 10 to 40 consecutive clinical isolates of S. aureus, S. pneumoniae, and H. influenzae each, from between 1 April 2020 and 31 December 2021.As for the other bacterial species (i.e., S. epidermidis, M. catarrhalis, P. aeruginosa) and the taxon Enterobacterales, all isolates obtained during the collection period were eligible for inclusion in the study.Only the first isolates of patients with ocular surface infections (e.g., conjunctivitis, keratitis, kerato-conjunctivitis, blepharitis, hordeolum) were permitted.The isolates were conserved until the end of the collecting period at the participating laboratories and subsequently shipped to a central laboratory (Antiinfectives Intelligence, Cologne, Germany).

Breakpoints
Due to a lack of clinical data that directly correlate therapeutic results with the MIC of pathogenic bacteria, EUCAST has not defined clinical breakpoints for topically applied antimicrobial substances.EUCAST advises one to use "regular" (clinical) breakpoints or cut-off values for the classification of bacterial phenotypes as susceptible or resistant.
In this study, the screening cut-offs for phenotypic resistance for detection of phenotypic resistance ("screening cut-off values") listed on page 111 of the EUCAST document "Breakpoint tables for interpretation of MICs and zone diameters Version 13.1" [17] or (tentative) epidemiological cut-off values ("[T]ECOFFs") as shown on the EUCAST website "MIC EUCAST-EUCAST MIC and zone diameter distributions and ECOFFs" [18] were used.
ECOFFs are set on the upper end of the MIC distribution and enable distinguishing between wild-types and non-wild-types (phenotypically resistant microorganisms).The screening cut-off values for cefoxitin and oxacillin for S. aureus and S. epidermidis can be found on page 36 of the above-mentioned EUCAST-document [18].The data generated in the studies OS1 to OS3 were evaluated with the same breakpoints in order to be able to compare them with the results of this study.
However, we did not apply the recently published oxytetracycline ECOFF of 0.5 mg/L for S. aureus [18] because this value is within the recommended MIC range (0.25-1 mg/L) of the oxytetracycline-susceptible reference strain S. aureus ATCC 29213 (Table S18).

Quality Control
The accuracy of susceptibility testing was evaluated using reference strains: S. aureus ATCC 29213 (MSSA), S. aureus ATCC 43300 (MRSA), S. pneumoniae ATCC 49619, E. coli ATCC 25922, P. aeruginosa ATCC 27853, and H. influenzae ATCC 49766 [19].The inoculum was quantified for every test run of the reference strains and 10% of the clinical isolates.
The viable count should be approximately 2 × 10 5 to 8 × 10 5 CFU/mL.In all but three cases, the MIC values were within the respective quality control limits (Table S18).

Statistical Evaluation
In order to evaluate statistical significance between two resistance rates R1 and R2, their 95% confidence intervals (calculated according to the Wilson method) were used [20].Statistical significance was determined if both resistance rates lied outside of the confidence intervals of each other.

Conclusions
The time period for collecting isolates for this study (OS4) was heavily influenced by the COVID-19 pandemic.Overall, there have been only minor changes in the resistance level of the antibiotics tested since the study in 2015 (OS3).Since the first study in 2005 (OS1), the resistance situation has even improved.The level of resistance, however, largely depends on the breakpoints applied.We used ECOFF values, tentative ECOFF values, and screening cut-off values defined by EUCAST to estimate the proportion of isolates with phenotypically detectable acquired resistance mechanisms to the agents tested (non-wildtype isolates).EUCAST has not defined clinical breakpoints for the evaluation of topically applied antibiotics due to a lack of clinical data that establish a correlation between the MIC of the infecting bacteria and therapeutic results.Local application of antimicrobial agents may result in higher concentrations at the infection site as compared to systemic therapy.This also applies to the use of antibiotics to treat ocular surface infections, especially conjunctivitis.Against this background and taking into account the epidemiologically largely unchanged resistance situation, the seven tested antibiotics-chloramphenicol, gentamicin, kanamycin, neomycin, levofloxacin, ofloxacin, and oxytetracycline-can continue to be recommended for use in the local therapy of superficial eye infections.In order to analyze the pathogenicity of resistant isolates, it is planned to include therapeutic data and molecular analysis of the genetic background (e.g., virulence and resistance genes) in future studies.

Table 1 .
In vitro activity of the tested antimicrobial agents against S. aureus: comparison of the results of the Ophthalmic Study (OS4) with the results of the previous studies performed in 2004 (OS1), 2009 (OS2), and 2015 (OS3).

Table 1 .
In vitro activity of the tested antimicrobial agents against S. aureus: comparison of the results of the Ophthalmic Study (OS4) with the results of the previous studies performed in 2004 (OS1), 2009 (OS2), and 2015 (OS3).thenumberand percentage of strains that display MICs lower or equivalent to the lowest concentration tested.Values presented in bold indicate the number and percentage of strains that display MICs higher or equivalent to the highest concentration tested.n.t., not tested;1Rates of MSSA and MRSA were determined using oxacillin as the test compound in OS1 and cefoxitin as the test compound in OS2, OS3, and OS4.

Table 2 .
In vitro activity of the tested substances against S. epidermidis and S. pneumoniae: comparison of the results of the Ophthalmic Study (OS4) with the results of the previous studies performed in 2004 (OS1), 2009 (OS2), and 2015 (OS3).
The epidemiological cut-off values (ECOFF), tentative ECOFF ((T)ECOFF), or screening cut-off values were used for interpretation (see Section 3 and TableS10for details).Abbreviations: CXI, cefoxitin; OXA, oxacillin; CHL, chloramphenicol; GEN, gentamicin; KAN, kanamycin; NEO, neomycin; LEV, levofloxacin; OFL, ofloxacin; OXY, oxytetracycline.Values presented in italics indicate the number and percentage of strains that display MICs lower or equivalent to the lowest concentration tested.Values presented in bold indicate the number and percentage of strains that display MICs higher or equivalent to the highest concentration tested.n.t., not tested; -, no cut-off value defined.

Table 3 .
In vitro activity of the tested substances against Gram-negative bacteria: comparison of the results of the Ophthalmic Study (OS4) with the results of the previous studies performed in 2004 (OS1), 2009 (OS2), and 2015 (OS3).
The epidemiological cut-off values (ECOFF), tentative epidemiological cut-off values ((T)ECOFF), or screening cut-off values were used for interpretation (see Section 3 and TableS10for details).Abbreviations: CHL, Chloramphenicol; GEN, Gentamicin; KAN, Kanamycin; NEO, Neomycin; LEV, Levofloxacin; OFL, Ofloxacin; OXY, Oxytetracycline.Values presented in italics indicate the number and percentage of strains that display MICs lower or equivalent to the lowest concentration tested.Values presented in bold indicate the number and percentage of strains that display MICs higher or equivalent to the highest concentration tested.n.t., not tested; -, no cut-off value defined.