Prevalence of susceptibility patterns of opportunistic bacteria in line with CLSI or EUCAST among Haemophilus parainfluenzae isolated from respiratory microbiota

The application of CLSI and EUCAST guidelines led to many discrepancies. Various doubts have already appeared in preliminary stages of microbiological diagnostics of Haemophilus spp. A total of 87 H. parainfluenzae isolates were obtained from throat or nasopharyngeal swabs from adults 18 to 70 years old, both healthy volunteers and patients with chronic diseases between 2013 to 2015 in eastern Poland. Haemophilus spp. were identified by colony morphology, Gram-staining, API NH and MALDI-TOF MS technique. Both susceptibility to various antimicrobials and phenotypes of Haemophilus spp. resistance to beta-lactams were determined. Statistically significant association between applied guidelines and drug resistance patterns were observed to as follows: ampicillin, cefuroxime, cefotaxime, amoxicillin-clavulanate, azithromycin, tetracycline and trimethoprim-sulfamethoxazole. Resistance phenotypes according to CLSI vs. EUCAST were as follows: 3.4% vs. 8.0% for BLNAR and 6.9% vs. 19.5% for BLPACR isolates. In conclusion, this is the first study that reports comparative analysis of drug susceptibility interpretation using CLSI and EUCAST of haemophili rods from human respiratory microbiota in Poland. In case of susceptible, increased exposure (formerly intermediate) category of susceptibility within H. parainfluenzae isolates we have observed EUCAST as more restrictive than CLSI. Moreover, BLNAI and BLPAI phenotype isolates have been observed, as well as BLPBR using only CLSI or EUCAST guidelines, respectively.

www.nature.com/scientificreports/ influenzae and H. parainfluenzae. The occurrence of such bacteria was found among ten dominant bacterial species in the nasopharynx of young children, where they constituted a specific core of the local microbiota 1 .
Haemophili rods may be specific component 1 or biomarkers of respiratory microbiota 4,6-8 in people of various age groups and different health states, especially in some chronic diseases. They can play a part in asymptomatic colonization, as well as in infections undergoing in conditions promoting opportunism. Percentage of children colonized by H. influenzae can reach up even above 30%, and its excessive colonization of the newborn's respiratory tract may double the risk of pneumonia, bronchitis and otitis media in the first three years of the child's life 9 .
Due to the well documented virulence of H. influenzae non-typeable (NTHi) strains, presumably other species of Haemophilus genus present in oral cavity may also be pathogenic 10 . Growing evidence of infection caused by these opportunistic microorganisms (a series of chronic or recurrent infections), as well as numerous discrepancies during data interpretation using actual recommendations, and growing resistance against commonly used antimicrobials have made this clear. For that reason, introduction of H. influenzae and H. parainfluenzae in the scientific elaborations and recommendations published by the Clinical and Laboratory Standards Institute (CLSI) 11 in 2006 and by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) 12 over the period 2009-2013, became a huge milestone in this issue. It made possible to interpret obtained results and take correct therapeutic decisions, especially in case of H. parainfluenzae for which the choice of antimicrobials and interpretation manner were analogous to H. influenzae criteria 11,12 . It is unfortunate that the application of both guidelines led to many discrepancies in classification of microorganisms for particular groups of sensitivity to various antimicrobials, therapeutic effect change or predicted clinical outcome. This is particularly important given the emergence of resistance to beta-lactam antibiotics among Haemophilus spp. Especially on the basis of knowledge about Haemophilus spp. bacteria resistance phenotypes and differences in resistance genes presence among opportunistic H. parainfluenzae isolates from respiratory microbiota 13 .
The novelty and huge value of this work is the comparison of susceptibility results on the basis of CLSI and EUCAST recommendations. Study was conducted on 87 H. parainfluenzae isolates being an opportunistic respiratory microbiota compound recovered from healthy volunteers and patients with chronic diseases in eastern Poland over the period 2013-2015, and similar data for haemophili rods has not been found in the literature so far.

Results
Susceptibility categories with cLSi and eUcASt. Due to a switching the interpretational criteria from CLSI into EUCAST brought statistically significant results in categorization of H. parainfluenzae isolates tested (Table 1). In the group of sensitive isolates major differences (p < 0.0001) have occurred in case of: cefuroxime, amoxicillin-clavulanate, ampicillin-sulbactam and azithromycin, as well as in cefotaxime (p = 0.0006) and trimethoprim-sulfamethoxazole (p = 0.0216). Isolates categorized as antimicrobials' resistant differed the most (p < 0.0001) in case of as follows: ampicillin, cefuroxime and chloramphenicol, as well as cefotaxime (p = 0.0223),   www.nature.com/scientificreports/ tetracycline (p = 0.0065) and trimethoprim-sulfamethoxazole (p = 0.0144). In Fig. 1 major differences between sensitive, increased exposure and resistant isolates with previous and actual recommendations were presented.
cLSi vs. eUcASt reliance on susceptibility patterns.       www.nature.com/scientificreports/ isolates resistant to one or more antimicrobials (Table 3) with p = 0.0010, respectively. Isolates sensitive to every used antimicrobials at 16.1% (14/87) were not considered in any further analysis. Relevant differences of AST results within two groups of isolates (from control group and chronic disease patients) were revealed (Fig. 2). Among all isolates, 8.0% (7/87) vs. 9.2% (8/87) were MDR strains according to CLSI-vs. EUCAST-derived drug patterns (Fig. 2). With regard to the results among both tested groups of isolates, in control group of healthy volunteers the number of MDR isolates accounted for 8.0% (7/87) vs. 5 (Fig. 3). CLSI 2020 criteria did not change anything in our results, whereas EUCAST 2020 increased the number of BLPACR isolates to 27.6% (24/87) and decreased of BLPAS isolates to 34.5% (30/87). Prevalence of RPs among isolates selected from healthy volunteers and patients with chronic disease was performed (Fig. 3). We have revealed a seven various patterns in control group of isolates. Among isolates of the second group, five vs. four RPs of CLSI vs. EUCAST have been found, respectively. In CLSI-derived results among patients with chronic diseases a strong dominance of betalactamase-positive phenotypes was assessed.

Discussion
Reliable interpretation of microorganisms' sensitivity data to various antimicrobial drugs may create some difficulties and requires expertise. During our studies, association between applied interpretation criteria and susceptibility patterns, as well as different phenotypes of resistance to selected antimicrobials among opportunistic haemophili rods were detected. Many authors [14][15][16][17][18][19][20] point to the differences occurring at the following fields: categorization of drug susceptibility, assignment to individual beta-lactam phenotypes of resistance in Haemophilus spp. bacteria, as well as CLSI 21 and EUCAST 22 cut-off values. The CLSI 2020 criteria has not changed since 2016, except cefuroxime sensitivity category ranges for this type of bacteria 23 On the contrary, EUCAST 2020 update brought significant changes in cut-off values, especially for ampicillin and amoxicillin with clavulanic acid for haemophili rods 24 . This resulted in an increase of the number of e.g.: isolates resistant to these antibiotics, and BLPACR phenotype isolation, as well as a decrease in the number of BLPAS phenotype and MDR isolates selection consequently. Among isolates selected from healthy volunteers the number of MDR isolates have been higher regardless of criteria used in comparison with diseased patients group in which, by contast, EUCAST-derived dominance of MDR isolates has been observed. www.nature.com/scientificreports/ In this study, we observed both criteria-CLSI and EUCAST coincide with obrained RPs amongst H. parainfluenzae (p < 0.001). Either CLSI-derived BLNAI and BLPAI or exclusively EUCAST-derived BLPBR phenotype defined as beta-lactamase-positive, cefinase-positive, resistant to one or more beta-lactams (benzylpenicillin, ampicillin, cephalosporins or carbapenems) found in 2.3% of H. parainfluenzae isolates, isolates revealed practical disparity. Prevalence of RPs among isolates selected from healthy volunteers revealed greater diversity regardless of guidelines used, whereas among isolates selected from patients with chronic diseases a strong dominance of beta-lactamase-positive phenotypes in CLSI-derived results was assessed.
Key practical differences between each guidelines for disk diffusion method in comparison with the latest ones (EUCAST 2020, CLSI 2020) were performed (Table 2), including both methodologies (e.g., bacteriological media, incubation conditions, antibiotic disc contents, zone diameter breakpoints), as well as reference strains for routine quality control. First doubts have already appeared in the preliminary, microorganisms identification stage. Two types of solid media are in use (HTM, MH-F) for AST. Both contain acidic casein, starch, agar, as well as beef and yeast (HTM) or meat extracts (MH-F), and specific growth factors for Haemophilus spp. bacteria-NAD-phosphate (V) and/or hem/haemin/hematin (X) 25 , in the form of HTMS supplement 26 or 5% mechanically de-fibered horse blood. Many microbiologists preferred to use other media such as MHF or chocolate agar which may support haemophili rods in a better way due to improved nutritive quality, resulting in increased MIC values and decreased growth inhibitory zones 14 . The adoption of richer medium such as chocolate agar or agar enriched with blood may also result in a sudden unwarranted increase of BLNAR strains (from 3.5% on HTM agar, through 14.7% on MHF, to 15.4% on chocolate agar) during agar dilution methods 14,18 . Currently the use of HTM agar as a reference medium in CLSI recommendations is more often questioned 14 .
Introduction of EUCAST can result in significant differences compared to CLSI in classification of microorganisms for particular groups of sensitivity to antimicrobials, being considered as more stringent. Classification of medium-sensitive-intermediate strains became great variability, because EUCAST discarded such category and changed the definition into 'susceptible, increased exposure' in 2019. A microorganism is categorized as above when a high likelihood of therapeutic success is observed because exposure to the agent is increased by adjusting the dosing regimen or by its concentration at the site of infection 22 . In practice, CLSI strains in question are usually assigned on the basis of EUCAST to resistant ones. From a clinical point of view, it deprived therapists the possibility of using a damp antimicrobial drug in therapy. Application of stricter criteria and breakpoints in accordance with EUCAST was aimed at reducing the misuse of antibiotics and controlling increasing level of resistance to them 27 . This is an important issue, because different criteria for interpreting the results of antibiograms could have influenced phenotypic determination of resistance among others to beta-lactam antibiotics. It also could be a critical spectra in context of decisions regarding treatment of chronic diseases, in which formation could have been involved e.g. opportunistic microbes resistant to this group of drugs. In addition to therapeutic aspects, another consequence arising from interpretation of received data were either potential epidemiological consequences or the spread of resistance among Haemophilus spp. being a respiratory tract microbiota component and other bacteria e.g. to beta-lactams. In our study, among H. parainfluenzae isolates switching the criteria from CLSI 2016 into EUCAST 2017 resulted in statistically significant (p < 0.05) modifications in (1) sensitivity mostly to beta-lactams with or without inhibitors (ampicillin, cefuroxime, cefotaxime, amoxicillin-clavulanate, ampicillin-sulbactam), as well as macrolides (azithromycin), tetracyclines (tetracycline) and trimethoprim-sulfomethoxazole; (2) susceptibility, increased exposure (formerly intermediate) mainly to cefuroxime (3.4% vs. 62.1%), azithromycin (0% vs. 100%) and tetracycline (42.5% vs. 10.3%); (3) resistance to ampicillin (8.0% vs. 36.8%), cefuroxime (6.9% vs. 37.9%), tetracycline (23.0% vs. 9.2%) and chloramphenicol (6.9% vs. 26.4%). The comparison of our results with the data obtained during the interpretation with the most current recommendations from 2020 revealed several more important problems. Both CLSI reccomendations were consistent with overall isolates categorization. The update caused only slight changes in the number of sensitive and susceptible, increased exposure of isolates to cefuroxime. In turn, the introduction of actual EUCAST 2020 recommendations caused major statistically significant changes, mainly in relation to beta-lactams with or without beta-lactamase inhibitors, as well as microlides and chloramphenicol or trimethoprim-sulfamethoxazole. This, considering all, may suggest significant discrepancies related to a definition of 'intermediate' category of isolates which may result in confusion and inaccuracy when prescribing treatment for example to urinary tract infections 20 . Generally, older versions of recommendations may have resulted in a lack of clinical response, while newer versions may reduce or even avoid therapeutic errors or failures.
The most frequently used antibacterial drugs in diseases caused by H. parainfluenzae, are beta-lactams used in mono-or in a combination therapy with antimicrobials from other therapeutic classes 12,26 . Analysis of our AST results within two groups of Haemophilus spp. isolates either from control group or chronic disease patients revealed relevant differences, especially significant reliance of applied guidelines on drug resistance patterns to e.g.: ampicillin, cefuroxime and cefotaxime. Moreover, susceptibility to amoxicillin-clavulanate of H. parainfluenzae depended on applied guideline, in a clavulanic acid dose-unrelated manner. On the contrary, high compatibility for amoxicillin-clavulanate between CLSI and EUCAST can be observed 20 . On the other hand, if EUCAST was used, a higher percentage of resistant E. coli and Klebsiella spp. isolates as well as lower one of Proteus spp. can be obtained 15 . Further, many authors noted significant differences between amoxicillin-clavulanate MIC determination methods. Researchers have come to the final conclusion that EUCAST-derived amoxicillinclavulanate MIC values were more predictive of therapeutic failure than CLSI ones 15  www.nature.com/scientificreports/ In our study, according to CLSI 2016 vs. EUCAST 2017, 48.3% vs. 73.6% of H. parainfluenzae isolates were resistant to one or more antimicrobials (p = 0.0010). Relevant discrepancies occurred during analysis of both H. parainfluenzae isolates groups as follows: distinctly more isolates resistant to one or more antimicrobials classes regardless of considered criteria; in the chronic disease group-nearly two times more isolates resistant to one class and two classes of antimicrobials according to EUCAST; in the control group-no isolates resistant to three or more antimicrobials classes observed using CLSI criteria; and three-folded higher number of such isolates when EUCAST applied. EUCAST beta-lactam interpretative criteria of Haemophilus spp. can be considered as close to median MICs for susceptible population, with subtle changes in MICs which slightly influenced on the proportion of strains categorized as resistant at the same time 16,17 . This may also contribute the application of appropriate treatment in patients with infections caused by a microbes producing specific beta-lactamases, either ESBL or AmpC 13,17 .
To conclude, this is the first study that reports comparative analysis of drug susceptibility interpretation using CLSI and EUCAST recommendations of haemophili rods isolated from human respiratory microbiota in Poland. In our study, it was observed EUCAST criteria as more restrictive in case of susceptible, increased exposure category within tested isolates. We have found 65.5% statistically significant covering of resistance phenotypes obtained by both CLSI and EUCAST criteria. Either CLSI-derived BLNAI and BLPAI or EUCAST-derived BLPBR phenotypes were determined. We have also found the EUCAST-derived dominance of MDR isolates amongst isolates from chronic diseased patients.
Prevalence of RPs among isolates selected from healthy volunteers revealed a strong dominance of betalactamase-positive phenotypes in CLSI-derived results in Haemophilus spp. bacteria. In this study, we observed EUCAST recommendations more restrictive in case of susceptible, increased exposure (formerly intermediate) category within tested haemophilic rods isolates. We agree with many other authors who agreeably suggested a still need of EUCAST breakpoints major revision, as well as a careful analyses and changes in EUCAST criteria, comprising all specialists involved in the therapeutic process of infectious diseases. Both incorrect determination of drug resistance of pathogens and the presence of antimicrobial resistance genes among bacteria derived from human microbiota, can play much more important role in the therapy development and therapeutic failure.
In a subsequent step, the minimal inhibitory concentrations (MICs) values for ampicillin (MIC Am ) were determined by the E-test method using the E-test strips (BioMerieux, Craponne, France) with antibiotic at a concentration gradient of 0.016-256 mg/L. Strips were placed on Mueller-Hinton agar with 5% defibrinated horse blood and 20 mg/L β-NAD (MH-F; BioMerieux, Craponne, France) after medium inoculation, incubation was performed in abovementioned microaerophilic conditions for 24 h at 35 °C. MIC Am  Resistance phenotypes (Rps). The multidrug-resistance (MDR) was defined as resistance to one or more antimicrobials from three or more therapeutic classes 22 . Phenotypes of resistance to beta-lactam antibiotics for Haemophilus spp. were identified on the basis of as follows: cefinase test, ampicillin MIC values, susceptibility to beta-lactams, benzylpenicillin 1U disk test, as well as determination of the presence of selected beta-lactamase bla genes described previously 13 . Accordingly, RPs were than classified into nine various categories: beta-lactamase-negative ampicillin sensitive BLNAS, ampicillin intermediate BLNAI, ampicillin resistant low-BLNAR and BLNAR, as well as beta-lactamase-positive ampicillin sensitive BLPAS, ampicillin intermediate BLPAI, ampicillin resistant BLPAR, amoxicillin-clavulanate resistant BLPACR and beta-lactams resistant BLPBR, as described earlier 13 and below the Fig. 3 in this paper.
Statistical analysis. Data were summarized and analysed by using GraphPad InStat 3.00 (GraphPad Software, USA). The p value (p < 0.05 considered as significant), the 95% confidence interval ranges (95% CI) and the relative risk (RR), calculated by the approximation of Katz, Pearson's Chi-square test or the Fisher's exact test were used for analysis of the extent of agreement between CLSI and EUCAST.