Development of a Disk Diffusion Method for Testing Moraxella catarrhalis Susceptibility Using Clinical and Laboratory Standards Institute Methods: a SENTRY Antimicrobial Surveillance Program Report

Abstract

Currently, there is no Clinical and Laboratory Standards Institute (CLSI) disk diffusion method for testing Moraxella catarrhalis susceptibility. We examined 318 clinical strains of M. catarrhalis obtained as part of the SENTRY (Asia-Pacific) Antimicrobial Surveillance Program, plus two ATCC strains. MICs were determined by the CLSI standard broth microdilution method, and zone diameters were determined on Mueller-Hinton agar incubated in 5% CO₂. All strains were examined for the presence of BRO-1 and BRO-2 β-lactamases by using molecular techniques. Tentative zone diameter interpretive criteria were successfully developed for 19 antimicrobial agents, including nine β-lactams, using current MIC interpretive criteria where available or wild-type cutoff values where no prior criteria were available. The proposed interpretive criteria were highly accurate, with ≤0.7% very major (falsely susceptible) and ≤1.0% major (falsely resistant) errors, respectively.

Moraxella catarrhalis is a common commensal and occasionally pathogenic bacterium associated with a range of infections of the respiratory tract, including acute otitis media, acute sinusitis, and acute exacerbations of chronic bronchitis (5, 22, 34). Less commonly, this organism can cause more serious and invasive infection, including pneumonia, septicemia, and meningitis (4, 28). A number of standards-setting organizations, including the Société Française de Microbiologie (29) and the British Society for Antimicrobial Chemotherapy (http://www.bsac.org.uk/susceptibility_testing.cfm), have developed disk diffusion tests and interpretive criteria for M. catarrhalis. Recently, the Clinical and Laboratory Standards Institute (CLSI) Subcommittee on Antimicrobial Susceptibility Testing published an interpretive guideline (6), but only for broth microdilution tests, using criteria essentially the same as those applied to Haemophilus spp. (10). Routine testing was not recommended by CLSI at the time of development of those guidelines.

However, this species is frequently isolated from respiratory tract infection specimens and, when present in large numbers, can be assumed by the laboratory to be playing a pathogenic or copathogenic role and, therefore, can be reported to the clinician. For the many international laboratories using CLSI methods and standards, the lack of a disk diffusion method and interpretive criteria limits their ability to communicate useful information to the clinician for a common potential pathogen. We followed CLSI M23-A3 guidelines (7) for the development of a disk diffusion test for M. catarrhalis, supplemented with molecular tests for some resistance mechanisms to assist in the future enhancement of interpretive criteria.

(This work was presented in part at the European Congress of Clinical Microbiology and Infectious Diseases, Munich, Germany, 2007.)

MATERIALS AND METHODS

Isolates.

Clinical strains of M. catarrhalis isolated from lower respiratory tract specimens and blood cultures were collected in 17 diagnostic laboratories in nine countries participating in the SENTRY Antimicrobial Surveillance Program from 1998 to 2004 (Asia-Pacific region) and referred to a monitoring laboratory in Adelaide, South Australia, for susceptibility testing (1). Strains were confirmed phenotypically to be M. catarrhalis by conventional carbohydrate fermentation reactions (23) and tributyrin hydrolysis (Remel, Lenexa, KS). Identifications were also confirmed using PCR according the method of Hendolin et al. (20, 21). M. catarrhalis ATCC 25238 (β-lactamase negative) and ATCC 43617 (BRO-2 β-lactamase-positive 1908 strain) were used as control strains.

Susceptibility testing.

MICs to antimicrobials were determined using broth microdilution as described by CLSI (8) and custom-made panels (Trek Diagnostic Systems, East Grinstead, United Kingdom). Disk diffusion testing was performed as described by CLSI (9) on Mueller-Hinton agar and incubated for 20 to 24 h in 5% CO₂ by using CLSI disk strengths conventionally used for testing other species. Preliminary studies had suggested that a significant percentage of strains would not grow adequately on Mueller-Hinton agar (Bio-Rad, Mames-la-Coquette, France) in ambient air without the addition of CO₂. It was also shown that while added blood or lysed blood enhanced the growth of strains which grow poorly in ambient air, growth was most reliable when unsupplemented and incubated in CO₂. Zone diameters were read by the Osiris zone reader (Bio-Rad, Mames-la-Coquette, France), a system with proven reliability (26). All results were validated by visual inspection of the captured video image in the system and the system's zone diameter selection.

Antimicrobials.

The antimicrobials examined (MIC range and disk content tested) were benzylpenicillin (0.008 to 256 μg/ml; 10 μg), ampicillin (0.008 to 256 μg/ml; 10 μg), amoxicillin-clavulanate (0.008 to 1 μg/ml; 20/10 μg), ampicillin-sulbactam (0.008 to 1 μg/ml; 10/10 μg), cephalexin (0.06 to 8 μg/ml; 30 μg), cefaclor (0.03 to 64 μg/ml; 30 μg), cefuroxime (0.25 to 16 μg/ml; 30 μg), ceftriaxone (0.002 to 16 μg/ml; 30 μg), meropenem (0.004 to 0.5 μg/ml; 10 μg), azithromycin (0.008 to 1 μg/ml; 15 μg), clarithromycin (0.008 to 1 μg/ml; 15 μg), erythromycin (0.03 to 4 μg/ml; 15 μg), ciprofloxacin (0.004 to 2 μg/ml; 5 μg), moxifloxacin (0.008 to 1 μg/ml; 5 μg), chloramphenicol (0.03 to 4 μg/ml; 30 μg), rifampin (0.008 to 1 μg/ml; 5 μg), gentamicin (0.016 to 1 μg/ml; 10 μg), tetracycline (0.03 to 64 μg/ml; 30 μg), and trimethoprim-sulfamethoxazole (0.06 to 4 μg/ml; 1.25/23.76 μg).

Testing for β-lactamase and tetracycline-resistant genes.

All isolates were examined for β-lactamase by using nitrocefin (Becton Dickinson, Sparks, MD). PCR testing for BRO β-lactamase genes and tetracycline-resistant genes was as previously described (13, 16).

Wild-type cutoff values.

MIC distributions were analyzed and wild-type cutoff (CO_WT) MICs were calculated as previously described (32). Where a significant proportion of results were off scale, or where the number of wild-type strains was low, CO_WT values were estimated by visual inspection. Results were compared to those available on the European Committee for Antimicrobial Susceptibility Testing (EUCAST) website (http://www.srga.org/eucastwt/WT_EUCAST.htm) and also to visual estimates of CO_WT from previous studies which provided data on MIC distributions (2, 15, 24).

Zone diameter interpretive criteria.

Disk diffusion zone diameter interpretive criteria were developed as described by CLSI (7), using current CLSI MIC interpretative criteria (6, 8) where they were available and using CO_WT values when they were not available or inappropriate.

RESULTS

A total of 320 isolates of M. catarrhalis were tested, including 318 clinical strains and the two ATCC control strains. Of these, only 20 were β-lactamase negative. A gene encoding a BRO enzyme was detected in all of the β-lactamase-positive isolates. A total of 281 isolates were BRO-1, and 19 were BRO-2 (including one ATCC strain). Two strains had MICs to benzylpenicillin and ampicillin elevated above those of the wild type, suggesting the presence of either β-lactamases other than those of the BRO type (14) or polymorphisms in BRO genes preventing PCR amplification.

CO_WT values.

Because of the low numbers of β-lactamase-negative (wild-type) isolates found (as expected from clinical practice), CO_WT values could be estimated only visually for β-lactams on the isolates from this study (Table 1). For the non-β-lactams, it was possible to use statistical analysis to determine CO_WT values (Table 1).

TABLE 1.

Estimated CO_WT values

Agent	COWT values (μg/ml) for strains used byc:
	This study (n = 320)	Felmingham et al. (n = 135)	Berk et al. (n = 818)	Karlowsky et al. (n = 36)	EUCAST (n = variable)
Penicillin	0.25a	0.125b
Ampicillin	0.06a			0.06b	0.125
Amoxicillin-clavulanate	0.125a	0.125b		0.03b
Ampicillin-sulbactam	0.06a
Cephalexin	4a
Cefaclor	2a	2b
Cefuroxime	1a	2b		1	4
Ceftriaxone	0.016a	0.25b		0.06b	0.064
Meropenem	0.008a,b
Erythromycin	1		0.5		0.25
Clarithromycin	0.5		0.25		0.064
Azithromycin	0.125		0.125
Telithromycin	0.5
Tetracycline	0.5				1.0
Rifampin	0.125
Chloramphenicol	0.5	1	1		2
Gentamicin	0.25				0.5
Ciprofloxacin	0.06	0.125b	0.125		0.125
Moxifloxacin	0.06				0.25

^aVisual estimate based on 20 β-lactamase-negative strains.

^bVisual estimate due to predominantly off-scale results.

^cReferences for previous studies are as follows: Felmingham et al. (15), Berk et al. (2), Karlowsky et al. (24), and EUCAST (http://www.srga.org/eucastwt/WT_EUCAST.htm).

MIC distributions.

The MIC distributions for β-lactams are shown in Table 2, and those for the non-β-lactams are shown in Table 3. As expected, the presence of BRO-1 increased the MICs of benzylpenicillin and ampicillin substantially (Table 2). It also caused obvious but small increases in MICs for amoxicillin-clavulanate and ampicillin-sulbactam. BRO-1 also increased the MICs of all cephalosporins. For reasons that are unclear, the BRO-1 MIC distributions of benzylpenicillin, amoxicillin-clavulanate, cefaclor, and ceftriaxone appeared to be bimodal. The presence of BRO-2 had similar but less-pronounced effects on MIC distributions; for cephalexin and meropenem, it had no detectable effect. The BRO-2 MIC distribution for ceftriaxone was bimodal.

TABLE 2.

MIC distributions for β-lactams by β-lactamase production

^a Solid, black vertical lines indicate CO_WT values, and hatched vertical bars indicate current CLSI breakpoints.

TABLE 3.

MIC distributions for non-β-lactams

^a Solid, black vertical lines indicate CO_WT values, and hatched vertical bars indicate current CLSI breakpoints.

^b CO_WT and breakpoint are the same.

Disk diffusion testing.

Eighteen strains (5.6%) did not grow well or at all on Mueller-Hinton agar in 5% CO₂ and were excluded from the analyses of zone diameter interpretive criteria, as were four strains for which disk diffusion testing was not performed. All other strains grew well and showed confluent growth. Therefore, interpretive criteria were developed using the remaining 298 isolates. A notable feature of the strains that grew well was large zone diameters for strains with low MICs. These were attributed to the disk contents used, which are the conventional strengths used in CLSI disk diffusion testing.

Selection of MIC interpretive criteria.

Of the agents tested, the current CLSI M45-A guideline for infrequently isolated and fastidious bacteria (6) provides interpretive MIC breakpoints for M. catarrhalis for amoxicillin-clavulanate, cefaclor, cefuroxime, ceftriaxone, azithromycin, clarithromycin, erythromycin, ciprofloxacin, tetracycline, trimethoprim-sulfamethoxazole, chloramphenicol, and rifampin. These interpretive criteria have not been subjected to detailed pharmacodynamic or clinical analysis. They have been adopted from those currently used for Haemophilus species largely because M. catarrhalis is associated with a similar spectrum of diseases and has not undergone formal validation by the CLSI. Additional Haemophilus species MIC interpretative criteria were available in the CLSI M100-S18 (10) document for ampicillin-sulbactam, meropenem, and moxifloxacin. These were selected for zone diameter analysis. For two agents, penicillin and ampicillin, the respective CO_WT values were each selected as a single MIC interpretive criterion, with the knowledge that these would be predictive of the presence of β-lactamases. Similarly, CO_WT values were selected as a single MIC interpretive criterion for the macrolides, because the intrinsic potency of this class is substantially greater against M. catarrhalis than it is against Haemophilus species, and therefore, the use of Haemophilus interpretive criteria would be misleading.

Predicted resistance rates.

Based on the selected MIC breakpoints (Table 4), nonsusceptibility rates of the 318 clinical isolates for the different antimicrobials were as follows: penicillin, 94.6%; ampicillin, 94.3%; amoxicillin-clavulanate, 0%; ampicillin-sulbactam, 0%; cefaclor, 24.8%; cefuroxime, 11.9%; ceftriaxone, 0.6%; meropenem, 0%; azithromycin, 0.3%; clarithromycin, 0.9%; erythromycin, 0.6%; ciprofloxacin, 0.3%; moxifloxacin, 0.3%; chloramphenicol, 0%; rifampin, 0%; gentamicin, 0%; tetracycline, 12.9%; and trimethoprim-sulfamethoxazole, 8.4%. All 41 tetracycline-nonsusceptible strains were categorized as resistant and contained the tet(B) gene. Almost all strains nonsusceptible to any of the agents tested, including non-β-lactams, contained the BRO-1 β-lactamase. The exceptions were two strains with MIC results for penicillin and ampicillin elevated above the wild-type level, which contained neither BRO-1 nor BRO-2.

TABLE 4.

Proposed MIC and tentative zone diameter interpretive criteria

Agent and disk content	Source or reference	MIC interpretive criteria (μg/ml)			% Intermediate; % resistante	Proposed zone diameter interpretive criteria (mm)			Discrepancy rates (%)
		Susceptible	Intermediate	Resistant		Susceptible	Intermediate	Resistant	Very major	Major	Minor
Penicillin, 10 μg	This studya	≤0.25		≥0.5	-; 94.6	≥29		≤28	0	0
Ampicillin, 10 μg	Proposed in this study	≤0.06		≥0.125	-; 94.3	≥33		≤32	0	0
Amoxicillin-clavulanate, 20/10 μg	M45-A (6)	≤4/2		≥8/4	-; 0	≥24		≤23	0	0
Ampicillin-sulbactam, 10/10 μg	M100-S18 (10)b	≤2/1		≥4/2	-; 0	≥29		≤28	0	0
Cefaclor, 30 μg	M45-A (6)	≤8	16	≥32	16.0; 8.8	≥18	16-17	≤15	0	0.3	9.4
Cefuroxime, 30 μg	M45-A (6)	≤4	8	≥16	7.9; 4.1	≥20	17-19	≤16	0	0	9.4
Ceftriaxone, 30 μg	M45-A (6)	≤2			-; 0.6	≥15			0.7	0
Meropenem, 10 μg	M100-S18 (10)b	≤0.5			-; 0	≥33			0	0
Azithromycin, 15 μg	This studyc	≤0.125			-; 0.3	≥26			0.3	0
Clarithromycin, 15 μg	This studyc	≤0.5			-; 0.9	≥24			0.7	0
Erythromycin, 15 μg	This studyc	≤1			-; 0.6	≥21			0.7	0
Ciprofloxacin, 5 μg	M45-A (6)	≤1			-; 0.3	≥21			0	0
Moxifloxacin, 5 μg	M100-S18 (10)b	≤1			-; 0.3	≥23			0	0
Chloramphenicol, 30 μg	M45-A (6)d	≤2			-; 0	≥31			0	0
Rifampin, 5 μg	M45-A (6)d	≤1			-; 0	≥28			0	0
Gentamicin, 10 μg	This studyc	≤0.5			-; 0	≥20			0	0
Tetracycline, 30 μg	M45-A (6)	≤2	4	≥8	0; 12.9	≥29	25-28	≤24	0	0	0
Trimethoprim-sulfamethoxazole, 1.25/23.75 μg	M45-A (6)	0.5/9.5	1/19-2/38	4/76	6.0; 2.5	≥13	11-12	≤10	0	1.0	7.4

^aPrincipally to correlate with β-lactamase results.

^bFor Haemophilus influenzae.

^cFor CO_WT values used.

^dModified as no resistance was detected, “susceptible” MIC was the only breakpoint used.

^e-, not applicable.

Zone diameter interpretive criteria.

Because the MIC interpretive criteria used (described above) have not been subjected to the full analysis required according to the CLSI M23-A3 guidelines (7), it was considered possible to set only tentative zone diameter interpretive criteria (Table 4). The requirements for establishing zone diameter interpretive criteria recommended in M23-A3 for unselected clinical strains, which is what our isolates represent, are simply that very major (falsely susceptible) discrepancy rates should be less than 1.5% and that major (falsely resistant) discrepancy rates should be less than 3%. No restriction is placed on minor discrepancy rates, although total error rates should be ≤10%. Using these criteria and the iterative methods described by Brunden et al. (3), we calculated the zone diameter interpretive criteria shown in Table 4. In addition, in the cases where there was a single MIC interpretive criterion, and where several zone diameter options yielded the same discrepancy rates (penicillin, ampicillin, amoxicillin-clavulanate, ampicillin-sulbactam, ceftriaxone, meropenem, ciprofloxacin, and moxifloxacin), we elected to set a conservative single zone diameter criterion at 1 millimeter less than the smallest zone diameter observed in the susceptible population. We also used a single zone diameter interpretive criterion using the same principle when no or only rare strains above wild-type level were detected (macrolides, chloramphenicol, rifampin, and gentamicin). The relatively poor activity of cephalexin against M catarrhalis, and its unusual distribution of MICs in the BRO-1-positive strains, precluded the setting of zone diameter criteria for this agent. Examples of the zone diameter MIC comparisons for cefuroxime, erythromycin, and tetracycline are shown in Fig. 1.

FIG. 1.

MIC zone diameter scattergrams for cefuroxime (A), erythromycin (B), and tetracycline (C). Solid horizontal lines represent MIC interpretive criteria. Dashed vertical lines represent calculated zone diameter interpretive criteria.

DISCUSSION

There has always been an occasional clinical need to test the susceptibility of M. catarrhalis. If the isolate comes from blood or another sterile body site culture (e.g., pleural fluid), treatment is clearly warranted. The organism is believed to be the third most common bacterium associated with acute exacerbations of chronic bronchitis and acute otitis media (4, 5, 14), both of which require antimicrobial treatment under certain circumstances. If resistance to commonly recommended agents is emerging, then regular testing would also be warranted. The availability of disk diffusion methods and interpretive criteria for M. catarrhalis should go a large part of the way to providing guidance on reporting of pathogenic strains of the species for those laboratories unwilling or unable to conduct dilution testing.

This is not the first attempt at developing a disk diffusion method like that of CLSI for M. catarrhalis. Doern and Tubert (12) used Mueller-Hinton agar that was both unsupplemented and supplemented with IsoVitalex and hemoglobin, incubated in air and 5 to 7% CO₂. Kibsey et al. (25) elected to use Mueller-Hinton agar incubated in 5% CO₂ for their disk diffusion study. They did not find strains with poor growth when agar was incubated in ambient air. Luman et al. (27) studied 231 isolates grown on Mueller-Hinton agar, using CLSI disk contents. Like our study, they too noted that a proportion of strains, 11% in their case, grew poorly in ambient air, and required retesting in 5% CO₂. Fung et al. (17) have also shown the feasibility of disk diffusion on media other than Mueller-Hinton agar.

Treatment recommendations for infections caused by this species are largely based on clinical experience and not rigorous trials. The Sanford Guide (19) recommends amoxicillin-clavulanate or an expanded-spectrum or broad-spectrum oral cephalosporin as primary choices, with alternatives being azithromycin, clarithromycin, dirithromycin, and telithromycin. The Sanford Guide states that erythromycin, doxycycline, and fluoroquinolones are known to be effective. Meza et al. (30) recommend β-lactamase-stable cephalosporins or amoxicillin-clavulanate as the first line but provide information on a broad range of other agents, including trimethoprim-sulfamethoxazole, macrolides, fluoroquinolones, doxycycline, and chloramphenicol, for a number of infections, ranging from mild to serious. These recommendations drove our choice of agents to examine.

We appreciate that the validity of many of the MIC interpretive criteria for M. catarrhalis is preliminary and unconfirmed. For the determination of breakpoints for an antimicrobial against a pathogen or group of pathogens, ideally reliable data are required from studies of the in vitro activity, the pharmacodynamics of the antimicrobial, and the in vivo efficacy rates from prospective clinical outcome studies (33). Unfortunately, reliable clinical outcome data on effective antimicrobial agents are scarce and are insufficient to draw conclusions about eradication rates. Furthermore, target pharmacokinetic/pharmacodynamic values for macrolides, tetracyclines, rifampin, aminoglycosides, or folate antagonists have not been uniformly accepted, As a consequence, we have selected MIC interpretive criteria mostly from Haemophilus influenzae, given the similarity of its clinical conditions to those of Haemophilus spp. in which M. catarrhalis is found. However, we believe the use of Haemophilus-derived breakpoints for the macrolides is inappropriate because the intrinsic activity of macrolides is substantially greater against Moraxella. Instead, we elected to use the CO_WT values found in this study for this class of agents. Acquired resistance mechanisms to macrolides have yet to be convincingly demonstrated in M. catarrhalis, even though occasional strains with elevated MICs are detected in large surveillance program studies (2). Indeed, previous work by one of us has suggested that macrolide resistance is more likely to represent misidentification of commensal Neisseria species as M. catarrhalis (31).

M. catarrhalis isolates frequently harbor β-lactamases, almost always of the BRO type. However, β-lactamase production is not a characteristic of the wild type, as early studies showed that M. catarrhalis was always β-lactamase negative until as recently as 1975 (35). Subsequent studies have shown a steady evolution to more than 80 to 90% of all clinical isolates worldwide being β-lactamase producers (34). The rate of β-lactamase production in our clinical strains, which came from eight countries in the Asia-Pacific region, was even higher, at 302/318 or 94.3%. We confirmed that BRO-1 had higher activity than did BRO-2, which was previously noted in other studies including strains from other regions in the SENTRY program (11, 35). Ours is one of the few studies to examine the effect of this difference on the MIC distributions of β-lactams in a broad population of clinical isolates. Fung et al. (18) showed significant differences between the geometric mean MICs of BRO-1- and BRO-2-producing strains for ampicillin, and they showed smaller differences for amoxicillin-clavulanate, loracarbef, cefixime, and cefetamet. Deshpande et al. (11), looking at SENTRY program isolates worldwide, showed a difference in the MIC distributions of penicillin and ampicillin between 385 BRO-1-positive isolates and 14 BRO-2-positive isolates. We also demonstrated a substantial upward shift in the MICs of BRO-1- and BRO-2-containing isolates for penicillin and ampicillin. However, we also noted a smaller but still significant upward shift in amoxicillin-clavulanate and, to a lesser extent, ampicillin-sulbactam, due to these enzymes. The same was true for cefaclor, cefuroxime, and ceftriaxone, but not for cephalexin, which appears relatively stable to BRO β-lactamases based on standard MIC measurement. For this reason, a small to moderate proportion of the β-lactamase-producing population were categorized as nonsusceptible to cefaclor, cefuroxime, and ceftriaxone. This nonsusceptibility was driven by the choice of MIC breakpoints found in CLSI guideline M45A (6), which in turn originated from those for Haemophilus spp. (10).

Apart from resistance mediated by β-lactamases, resistance to other classes is relatively uncommon. The only resistances that featured in our region were those for tetracycline (12.8%) and trimethoprim-sulfamethoxazole (2.5%). The few strains that were categorized as nonsusceptible to macrolides had MIC results that were one twofold dilution higher than the calculated cutoff values and are unlikely to harbor true resistance mechanisms. Of most interest were the considerable rates of nonsusceptibility and resistance to the oral broad-spectrum cephalosporins (cefaclor and cefuroxime) using the current breakpoints (derived from H. influenzae). These rates suggest that the choice of these agents as empirical therapy may be inadvisable.

There are some potential limitations of our findings. We used a single batch of Mueller-Hinton agar. Although there is not a CLSI requirement to use more than one lot or manufacturer of Mueller-Hinton agar for disk diffusion studies, it is possible that differences may exist between lots and manufacturers, especially for antimicrobials affected by divalent cations. We did not pursue the molecular mechanisms of resistance that might explain resistance to trimethoprim-sulfamethoxazole, although this would benefit from further study.

In summary, we have developed tentative zone diameter interpretive criteria for a range of antimicrobials that might be considered for the treatment of M. catarrhalis infections when required. We expect that future work will result in modifications to these criteria, especially as pharmacokinetic/pharmacodynamic targets become fully developed for all drug classes and as clinical outcome studies provide more information on the efficacy of eradication.