Surveillance of Antimicrobial Susceptibility of Anaerobe Clinical Isolates in Southeast Austria: Bacteroides fragilis Group Is on the Fast Track to Resistance

Anaerobic bacteria play an important role in human infections. Bacteroides spp. are some of the 15 most common pathogens causing nosocomial infections. We present antimicrobial susceptibility testing (AST) results of 114 Gram-positive anaerobic isolates and 110 Bacteroides-fragilis-group-isolates (BFGI). Resistance profiles were determined by MIC gradient testing. Furthermore, we performed disk diffusion testing of BFGI and compared the results of the two methods. Within Gram-positive anaerobes, the highest resistance rates were found for clindamycin and moxifloxacin (21.9% and 16.7%, respectively), and resistance for beta-lactams and metronidazole was low (<1%). For BFGI, the highest resistance rates were also detected for clindamycin and moxifloxacin (50.9% and 36.4%, respectively). Resistance rates for piperacillin/tazobactam and amoxicillin/clavulanic acid were 10% and 7.3%, respectively. Two B. fragilis isolates were classified as multi-drug-resistant (MDR), with resistance against all tested beta-lactam antibiotics. The comparative study of 109 BFGI resulted in 130 discrepancies in 763 readings (17%) with a high number of Very Major Errors (VME) and Major Errors (ME). In summary, resistance rates, with the exception of clindamycin and moxifloxacin, are still low, but we are facing increasing resistance rates for BFGI. Surveillance studies on a regular basis are still recommended.


Introduction
Anaerobic bacteria play an important role in a variety of human infections. Among anaerobic bacteria, the Bacteroides fragilis group represents one of the most important anaerobic clinical pathogens and ranges under the 15 most common pathogens causing nosocomial infections [1,2]. Different Gram-positive anaerobic cocci like Finegoldia magna, Peptostreptococcus anaerobius, and Parvimonas micra account for approximately 25-30% of all isolated anaerobic bacteria from clinical specimens [3]. Within the Gram-positive anaerobe rods, Cutibacterium acnes plays an important role, especially in prosthetic joint infections [4]. Monomicrobial and polymicrobial anaerobic infections are usually treated empirically, based on surveillance reports of the susceptibility patterns of these pathogens. Routine susceptibility testing of anaerobic bacteria is not generally recommended for polymicrobial infections, but should be considered for specific clinical situations and in monomicrobial infections [5,6]. For empiric therapy strategies, resistance-surveillance is recommended due to an increase in resistance, especially in Bacteroides fragilis group isolates (BFGI) [1,7,8]. Nevertheless, cultivation and isolation of anaerobes is time-consuming, especially when susceptibility testing is required. The gold standard of susceptibility testing by agardilution method is not suitable for a routine diagnostic laboratory. Other techniques like the MIC gradient testing method (e.g., ETEST ® ) are costly and not always affordable for smaller laboratories or laboratories of middle-and low-income countries. As studies on the antimicrobial susceptibility profile of anaerobic bacteria are underrepresented in the literature [9], we present antimicrobial susceptibility testing (AST) results of 114 Gram positive anaerobic isolates and 110 BFGI determined by MIC gradient testing derived from clinical specimens in Southeast Austria. Furthermore, we performed disk diffusion testing of BFGI, using the zone diameter breakpoints published by Nagy et al. in 2015 [10], and compared the results of the two methods to each other. Table 1 summarizes the distribution by genus and/or species of the 224 anaerobic bacteria and the in-vitro resistance rates to the tested antibiotics determined by MIC gradient testing. All isolates included in the study were clinically relevant. From 224 isolates, 114 (50.9%) were Gram-positives, including Finegoldia magna (n = 31), Peptoniphilus assacharolyticus (n = 23), Peptoniphilus spp. (n = 13), Peptostreptococcus anaerobius (n = 8), Parvimonas micra (n = 8), Actinomyces spp. (n = 11), Cutibacterium acnes (n = 15), and Cutibacterium avidum (n = 5). Furthermore, 110 (49.1%) strains of the BFGI were tested. The most frequently isolated species among the BFGI was Bacteroides fragilis (n = 51), followed by B. thetaiotaomicron (n = 19), B. ovatus (n = 14), B. vulgatus (n = 11), Parabacteroides distasonis (n = 6), B. uniformis (n = 5), B. caccae (n = 2), and B. stercoris (n = 2).

Disc Diffusion Testing of BFGI
In 109 of 110 BFGI, susceptibility was additionally determined by disc diffusion. Results are depicted in Tables 2 and 3. There were 130 discrepancies in 763 readings (17%) between results determined by disc diffusion compared to those determined by MIC testing. The number of very major errors (VME) was as follows: 50% for imipenem, 37.5% for amoxicillin/clavulanic acid, 7.1% for clindamycin, 10% for moxifloxacin, and 33.3% for metronidazole; there were no VME for piperacillin/tazobactam. The number of major errors (ME) was also high: 40.2% for piperacillin/tazobactam, 8.3% for imipenem, 16.7% for clindamycin, 3.6% for moxifloxacin, and 29.9% for metronidazole. The number of minor errors (mE) was 2.7% for amoxicillin/clavulanic acid, 10.9% for piperacillin/tazobactam, 0.9% for imipenem, and 13.6% for moxifloxacin. Inhibition zone diameters (mean ± standard deviation) for the B. fragilis ATCC 25285 control strain in parallel measurements on 8 different dates were as follows: amoxicillin/clavulanic acid 29.    AMC, amoxicillin/clavulanic acid; TZP, piperacillin/tazobactam; IMP, imipenem; CC, clindamycin; MOX, moxifloxacin; MTZ, metronidazole; mE, minor error (disc diffusion = S or R and reference method = I), ME, major error (disc diffusion = R and reference method = S), VME, very major error (disc diffusion = S and reference method = R); n.a., not applicable, because no I interpretation; * 1 strain not readable in disc diffusion test.

Discussion
Infections due to anaerobic bacteria play an important role in clinical routine. A large study including 365,490 episodes of healthcare-associated infections found that Bacteroides spp. were under the 15 most common pathogens causing nosocomial infections [2]. Most infections, where anaerobic bacteria are involved, were treated empirically, mainly with betalactam antibiotics [5]. Due to worldwide increase in resistance rates, especially for Bacteroides spp., and extensive differences in the antimicrobial susceptibility patterns of anaerobic bacteria isolated within different countries, local surveillance studies on a regular basis are highly recommended [9,12].
In this study, we evaluated the antibiotic susceptibility profile of different Grampositive anaerobic bacteria and BFGI isolated from human clinical specimens such as intra-abdominal infections, wounds, and abscesses. Furthermore, we performed disc diffusion testing of BFGI and compared the results to those determined by MIC gradient testing. The findings of this surveillance study present constantly low resistance levels for Gram-positive anaerobic bacteria in Austria. Remarkable resistance was only observed for clindamycin (21.9%) and moxifloxacin (16.7%). Based on our findings, betalactam antibiotics, vancomycin, and metronidazole remain useful drugs for empiric treatment of anaerobic Gram-positive infections. One exception is Cutibacteria spp. due to their natural resistance to metronidazole [11].
Within the BFGI, we are facing increasing resistance rates. In the year 2007, we found very low resistance rates (<1%) to amoxicillin/clavulanic acid, piperacillin/tazobactam, the carbapenems, and metronidazole [13]. Compared to this study, resistance levels increased for all tested antibiotics. Based on our findings, clindamycin and moxifloxacin showed remarkable resistance (50.9% and 36.4%, respectively). Consequently, they are not recommended for empiric treatment. The resistance rates for carbapenems are still low (0.9% for imipenem and 1.8% for meropenem), but there must be an increasing awareness of multidrug resistance among BFGI. Monitoring the resistance patterns of BFGI in routine laboratories by disc diffusion testing, which is much less expensive and easier to perform, would facilitate surveillance. In 2015, tentative zone diameter susceptibility breakpoints for six different antibiotics for disk diffusion testing of BFGI were published [10]. In this study, we compared results of disc diffusion testing to MIC values determined by gradient testing. The number of discrepancies between the two methods were very high, especially for piperacillin/tazobactam (47 discrepant results), moxifloxacin (21 discrepant results), and metronidazole (33 discrepancies). The number of VME and ME was also high for all tested antibiotics. Most of the isolates that showed discrepancies were close to the breakpoint.
This study has limitations. The results of disc diffusion testing were compared to MICs obtained by MIC gradient test method (ETEST ® ), not the agar dilution method, which would be the preferred technique, but, due to its labor-intensive and time-consuming character, is not suitable for routine diagnostic procedures. In contrast to the study from Nagy et al., bacterial suspension was prepared in Brucella broth, not in 0.85% saline or thioglycolate broth, and all discs were from a different company than those in the study from Nagy et al. (BioRad, France and Oxoid, UK versus Becton Dickinson, Heidelberg, Germany). Therefore, we cannot exclude the possibility that these differences would have led to these results. Furthermore, we had some problems with the metronidazole disc.
The zone diameter of metronidazole of the B. fragilis ATCC 25285 control strain ranged from 20 to 28 mm within eight measurements. Therefore, we cannot ensure that results of metronidazole disc diffusion testing are reliable. As described before, this method may be used as a screening method [14]; we would suggest that all results close to the breakpoint are confirmed by a MIC method.

Susceptibility Testing
In total, 224 non-duplicate anaerobic bacterial isolates were analysed. For susceptibility testing, isolates were thawed on Schaedler agar and sub-cultured once prior to inoculation. For MIC determination ETEST ® (bioMérieux, France) or MIC test strips (Liofilchem, Roseto degli Abruzzi, Italy) with the protocol for anaerobic bacteria were used. For both ETEST ® and disc diffusion, a bacterial suspension was prepared in Brucella broth (Becton Dickinson, Germany) to a density of McFarland 1. AST was performed on Brucella agar with Vitamin K1, 5% blood, and hemin (Becton Dickinson, Germany). The 1515-15-min rule of EUCAST was applied. The plates were incubated at 35 • C in an anaerobic atmosphere for 48 h. For Gram-positive anaerobic bacteria, MICs of the following agents were determined: penicillin, amoxicillin/clavulanic acid, piperacillin/tazobactam, imipenem, meropenem, moxifloxacin, vancomycin, clindamycin, and metronidazole. For the BFGI, MICs of the following agents were determined: amoxicillin/clavulanic acid, piperacillin/tazobactam, imipenem, meropenem, moxifloxacin, cefoxitin, clindamycin, and metronidazole. Interpretation was done according to EUCAST guidelines [15] for anaerobes, except for cefoxitin and moxifloxacin, where the CLSI guideline M11-A8 [16] was applied. Disk diffusion testing was performed for the same antimicrobial agents except for cefoxitin and meropenem. All discs (amoxicillin/clavulanic acid 20/10 µg, piperacillin/tazobactam 30/6 µg, imipenem 10 µg, moxifloxacin 5 µg, clindamycin 10 µg, and metronidazole 5 µg) were obtained from Becton Dickinson (Germany). For interpretation, the 2015 published zone diameter from Nagy et al. was used [10]: amoxicillin/clavulanic acid (≥15 mm), piperacillin/tazobactam (≥25 mm), imipenem (≥29 mm), clindamycin (≥25 mm), moxifloxacin (≥19 mm), and metronidazole (≥24 mm). All the measurements were carried out with the naked eye using a ruler, and all measurements were done by the same person. Zone diameters were read at 100% inhibition after 24 h of incubation. The proportion of VME (disc diffusion = S and reference method = R), ME (disc diffusion = R and reference method = S), and mE (disc diffusion = S or R and reference method = intermediate, I) between disc diffusion and MIC determination (reference method) was calculated [17]. B. fragilis ATCC 25285 was included as the quality control strain.

Conclusions
In conclusion, our data still show low resistance rates for betalactam antibiotics and metronidazole for anaerobic bacteria. Remarkable resistance was only observed for clindamycin and moxifloxacin. There should be awareness of increasing resistance rates, especially for Bacteroides spp. Therefore, disc diffusion testing, which can be performed easily in all microbiological laboratories, would be needed, but the method studied has potential for improvement. Surveillance studies on a regular basis are still recommended.