Effects of Following National Committee for Clinical Laboratory Standards and Deutsche Industrie Norm-Medizinische Mikrobiologie Guidelines, Country of Isolate Origin, and Site of Infection on Susceptibility of Escherichia coli to Amoxicillin-Clavulanate (Augmentin)

Abstract

Amoxicillin-clavulanate (Augmentin), as a combination of two active agents, poses extra challenges over single agents in establishing clinically relevant breakpoints for in vitro susceptibility tests. Hence, reported differences in amoxicillin-clavulanate percent susceptibilities among Escherichia coli isolates may reflect localized resistance problems and/or methodological differences in susceptibility testing and breakpoint criteria. The objectives of the present study were to determine the effects of (i) methodology, e.g., those of the National Committee for Clinical Laboratory Standards (NCCLS) and the Deutsche Industrie Norm-Medizinische Mikrobiologie (DIN), (ii) country of origin (Spain, France, and Germany), and (iii) site of infection (urinary tract, intra-abdominal sepsis, or other site[s]) upon the incidence of susceptibility to amoxicillin-clavulanate in 185 clinical isolates of E. coli. Cefuroxime and cefotaxime were included for comparison. The use of NCCLS methodology resulted in different distribution of amoxicillin-clavulanate MICs than that obtained with the DIN methodology, a difference highlighted by the 10% more strains found to be within the 8- to 32-μg/ml MIC range. This difference reflects the differing amounts of clavulanic acid present. NCCLS and DIN methodologies also produce different MIC distributions for cefotaxime but not for cefuroxime. Implementation of NCCLS and DIN breakpoints produced markedly different incidences of strains that were found to be susceptible, intermediate or resistant to amoxicillin-clavulanate. A total of 86.5% strains were found to be susceptible to amoxicillin-clavulanate by the NCCLS methodology, whereas only 43.8% were found to be susceptible by the DIN methodology. Similarly, 4.3% of the strains were found to be resistant by NCCLS guidelines compared to 21.1% by the DIN guidelines. The use of DIN breakpoints resulted in a fivefold-higher incidence of strains categorized as resistant to cefuroxime. There were no marked differences due to country of origin upon the MIC distributions for amoxicillin-clavulanate, cefuroxime, or cefotaxime, as determined with the NCCLS guidelines. Isolates from urinary tract and intra-abdominal infections were generally more resistant to amoxicillin-clavulanate than were isolates from other sites of infection.

Amoxicillin-clavulanate (Augmentin), as a combination of two active agents with similar but not identical pharmacokinetic properties following parenteral administration, poses extra challenges over single agents in establishing clinically relevant breakpoints for in vitro susceptibility tests. In Europe there are at least four authorities that provide guidelines for susceptibility testing and breakpoint criteria. The National Committee for Clinical Laboratory Standards (NCCLS) and the British Society for Antimicrobial Chemotherapy (BSAC) guidelines for the determination of MICs are quite similar, generally differing only in the choice of test media (5, 13). Many European laboratories follow the NCCLS guidelines, while laboratories in France and Germany generally follow the Société Français de Microbiologie (SFM) and Deutsche Industrie Norm-Medizinische Mikrobiologie (DIN) guidelines, respectively (2, 8). There are two key factors responsible for the differences between the BSAC, NCCLS, SFM, and DIN agar dilution MIC susceptibility testing guidelines: the susceptibility breakpoints and, for amoxicillin-clavulanate, the concentrations of clavulanate employed. The NCCLS breakpoint criteria for categorizing strains as susceptible, intermediate, or resistant to amoxicillin-clavulanate permit higher MICs than do the BSAC criteria, which in turn allow higher MICs than those of the SFM and DIN guidelines (Table 1). The BSAC and NCCLS guidelines recommend that amoxicillin-clavulanate MIC tests be conducted with a 2:1 amoxicillin/clavulanate ratio, whereas DIN and SFM guidelines recommend testing clavulanate at a fixed concentration of 2 μg/ml, irrespective of the amoxicillin concentration.

TABLE 1.

Principal differences between the NCCLS, BSAC, SFM, and DIN guidelines for the determination of MICs by agar dilution

Antibiotic	Breakpoints (μg/ml) as determined by methodology of a:
	NCCLS (≤1, 2:1 ratio)			BSAC (0, 2:1 ratio)			SFM (≤3, 2 μg/ml)			DIN (≤4, 2 μg/ml)
	S	I	R	S	I	R	S	I	R	S	I	R
Amoxicillin-clavulanate	≤8	16	≥32	≤8		>8	≤4	8–16	>16	≤2	4–8	>8
Cefotaxime	≤8	16–32	≥64	≤1		>1	≤4	8–32	>32	≤1	2–8	>8
Cefuroxime	≤8	16	≥32	≤4		>16	≤8	16–32	>32	≤4	8	>8

^aValues in parentheses indicate the number of endpoint colonies and the clavulanate concentration criteria used with each set of guidelines. S, susceptible; I, intermediate; R, resistant. 

Escherichia coli strains are normally susceptible (>90% strains) to amoxicillin-clavulanate, and no firm evidence of increased resistance has been found (7). Nevertheless, there is concern in some European markets (1, 11) at a perceived increased resistance to amoxicillin-clavulanate, and resistance values as high as 54% have been reported (3). This finding may be valid, reflecting localized resistance problems, or it may be due to an artifact that is attributable to methodological differences in susceptibility testing and breakpoint criteria.

Resistance mechanisms include hyperproduction of the plasmid-mediated TEM-1 β-lactamase (14), the production of a TEM-1 variant (TRC-1) with decreased susceptibility to clavulanic acid inhibition (9), reduction in outer membrane permeability, or a decreased affinity for the penicillin-binding proteins (6). Although there are reports to the contrary (4), it is possible that isolates from infections at different sites, e.g., urinary tract infections (UTIs) or intra-abdominal sepsis (IAS), may exhibit different levels of susceptibility due to different selection pressures during antibiotic therapy.

The objectives of the present study were to determine the effects of the following three factors upon the incidence of susceptibility to amoxicillin-clavulanate in clinical isolates of E. coli: methodology (we used the extremes of the national guidelines, i.e., NCCLS versus DIN), country of origin (Spain, France, or Germany), and site of infection (UTI, IAS, or infections at other sites).

MATERIALS AND METHODS

Collection of isolates.

Participating facilities in Spain, France, and Germany collected consecutive isolates of E. coli from UTI, IAS, or other infections at various sites. Isolates were coded with country, patient initials, and source of infection. Strains were collected during the period from May to November 1994, cultured on 5-ml nutrient agar slopes, and dispatched to the central laboratory at SmithKline Beecham (Brockham Park, Surrey, United Kingdom) for susceptibility testing. The bacterial species of isolates found to be resistant to amoxicillin-clavulanate by NCCLS criteria were confirmed by the ATB system (Biomerieux, Basingstoke, United Kingdom).

Susceptibility tests.

The susceptibilities of each isolate to amoxicillin-clavulanate, cefotaxime, and cefuroxime were determined in parallel with strict adherence to NCCLS (5) and DIN (2) guidelines for MIC determinations by agar dilution. Principal details of endpoints and breakpoints for intravenous preparations are given in Table 1. Both methods recommend testing a bacterial inoculum of 10⁴ CFU/spot. Isolates were tested in six batches of 35 cultures. On each test occasion, five reference cultures (E. coli ATCC 25922, E. coli ATCC 35218, Pseudomonas aeruginosa ATCC 27853, Staphylococcus aureus ATCC 25923, and S. aureus ATCC 29213) were included to verify the accurate preparation of MIC dilution series.

Antibiotics.

Sodium amoxicillin and lithium clavulanate were obtained from SmithKline Beecham. Sodium salts of cefotaxime and cefuroxime were purchased from Sigma (Poole, United Kingdom).

Statistical analyses.

Chi-square (χ_n²) tests were used to analyze the susceptibility categories following implementation of the NCCLS and/or DIN breakpoints.

RESULTS

Reference strains.

All the MICs of amoxicillin-clavulanate, cefotaxime, and cefuroxime for the five reference strains were within NCCLS and DIN acceptable limits. Accordingly, results for the study strains over the six test occasions were pooled.

Composition of study isolates.

A total of 185 E. coli isolates were submitted for susceptibility testing: 96, 51, and 38 strains from Spain, France, and Germany, respectively. The isolates from the three countries exhibited markedly different incidences of strains obtained from the different sites of infection, e.g., the center(s) in Germany submitted 78.9% IAS isolates, whereas the center(s) in France submitted 76.5% UTI isolates. All of the strains, which were found to be resistant to amoxicillin-clavulanate by NCCLS or DIN methodologies, were confirmed to be E. coli.

Effect of methodology on MIC distributions.

The distribution of amoxicillin-clavulanate MICs derived from NCCLS and DIN methodologies were markedly different (Fig. 1). At an amoxicillin-clavulanate concentration containing amoxicillin at 1 μg/ml, 12% of the isolates were inhibited as determined by the NCCLS methodology compared to 23% by the DIN methodology. In contrast, at amoxicillin-clavulanate concentrations containing amoxicillin at 8 to 32 μg/ml, more isolates were found to be inhibited based on the NCCLS methodology. There were considerably more isolates inhibited by cefotaxime at all concentrations between 0.004 and 0.12 μg/ml with the NCCLS criteria than with the DIN criteria (Fig. 2). However, all but two strains were susceptible to 1 μg/ml, which is the lower DIN breakpoint. The MIC distributions for cefuroxime obtained by the NCCLS and DIN methodologies were very similar, the exception being a higher incidence of strains susceptible to 1 μg/ml by the DIN methodology (Fig. 3).

FIG. 1.

Effect of NCCLS and DIN methodologies on the determination of E. coli susceptibilities (MICs) to amoxicillin-clavulanate.

FIG. 2.

Effect of NCCLS and DIN methodologies on the determination of E. coli susceptibilities (MICs) to cefotaxime.

FIG. 3.

Effect of NCCLS and DIN methodologies on the determination of E. coli susceptibilities (MICs) to cefuroxime.

Effect of methodology on breakpoint categories.

Implementation of breakpoints had a marked effect (χ₂² = 74.4; P < 0.001) upon differences between the NCCLS and DIN susceptibility categories for the amoxicillin-clavulanate results (Table 2). A total of 86.5% of strains were fully susceptible to amoxicillin-clavulanate by NCCLS guidelines in contrast to only 43.8% by DIN guidelines. Similarly, only 4.3% strains were resistant by NCCLS criteria compared to 21.1% by DIN criteria. DIN criteria also indicated a 5.9% resistance to cefuroxime, which was significantly higher (χ₂² = 13.2; P = 0.01) than that noted with the NCCLS methodology (1.1%). All strains were fully susceptible to cefotaxime by NCCLS criteria, but two of the strains (1.1%) were found to be intermediate by DIN criteria.

TABLE 2.

Incidence of susceptible, intermediate, and resistant E. coli strains according to NCCLS or DIN methodology

Response	% Incidence of E. coli strains with:
	Amoxicillin-clavulanate		Cefuroxime		Cefotaxime
	NCCLS	DIN	NCCLS	DIN	NCCLS	DIN
Susceptible	86.5	43.8	95.1	83.8	100	98.9
Intermediate	9.2	35.1	3.8	10.3	0	1.1
Resistant	4.3	21.1	1.1	5.9	0	0

Effect of country.

There were no marked differences in the distributions of amoxicillin-clavulanate MICs, as determined with NCCLS guidelines, between strains from the three countries (Fig. 4). However, analysis of the incidence of susceptible, intermediate, and resistant strains (Table 3) indicated significant differences among the isolates by countries of origin which were principally due (χ₂² = 6.5; P < 0.05) to the high incidence of intermediate (18.4%) relative to resistant (0%) strains from Germany compared to those from Spain and France. There were no differences between countries in the distribution of cefuroxime and cefotaxime MICs and susceptible, intermediate, and resistant categories.

FIG. 4.

Comparison of susceptibilities to amoxicillin-clavulanate (based on NCCLS guidelines) of 185 E. coli isolates from Spain, France, and Germany.

TABLE 3.

Incidence of isolates susceptible, intermediate, or resistant, as based on NCCLS guidelines, to amoxicillin-clavulanate, cefuroxime, and cefotaxime by country of origin

Antibiotic	Category	% Incidence for isolates from:
		Spain (n = 96)	France (n = 51)	Germany (n = 38)
Amoxicillin-clavulanate	Susceptible	89.6	84.3	81.6
	Intermediate	6.3	7.8	18.4
	Resistant	4.2	7.8	0.0
Cefuroxime	Susceptible	93.8	94.1	100
	Intermediate	6.2	2.0	0.0
	Resistant	0.0	3.9	0.0
Cefotaxime	Susceptible	100	100	100

Effect of infection site.

IAS and UTI isolates had similar distributions of amoxicillin-clavulanate MICs and similar breakpoint categories, as determined with the NCCLS guidelines. However, the isolates from the other infection sites were generally inhibited by lower concentrations of amoxicillin-clavulanate, and this was reflected (χ₂² = 7.8; P < 0.05) in the higher incidence of strains categorized as susceptible. There were no significant differences in either the cefuroxime or cefotaxime susceptibilities of strains from different infection sites.

DISCUSSION

It is clear from the results in this report that the perceived problem of differences in E. coli resistance to amoxicillin-clavulanate across Europe is largely attributable to differences in the methodologies and guidelines employed for amoxicillin-clavulanate testing and in the interpretation of clinical susceptibility (i.e., breakpoints) and not to any major difference in the actual susceptibilities of E. coli in the different countries or sites of infection. These susceptibility differences reflect the amount of clavulanate available to inhibit β-lactamase and so protect amoxicillin. SFM and DIN guidelines indicate that clavulanate should be maintained at a fixed concentration of 2 μg/ml irrespective of the amoxicillin concentration, whereas NCCLS guidelines specify a 2:1 ratio of amoxicillin to clavulanate. Consequently, at amoxicillin concentrations of ≤2 μg/ml, SFM and DIN tests incorporate more clavulanate than NCCLS, whereas at amoxicillin concentrations of ≥8 μg/ml the reverse is true. Only at an amoxicillin concentration of 4 μg/ml do the guidelines generate an identical amoxicillin-clavulanate product, and this is reflected in the convergence of susceptibility results at this value (Fig. 1). The level of resistance to amoxicillin-clavulanate observed with NCCLS guidelines, i.e., 4.3%, was considerably less than that obtained with the DIN guidelines, i.e., 21.1%. Clavulanate alone can exhibit a significant antibacterial effect against some species. However, clavulanate concentrations in excess of 16 μg/ml are required to inhibit E. coli (7) and thus are relevant only in NCCLS susceptibility studies in which the amoxicillin component is ≥32 μg/ml.

The effects of fixed concentrations versus fixed ratios of clavulanate upon E. coli susceptibility were investigated previously with 100 isolates from Scotland (10). In that study it was argued that using a fixed clavulanate concentration of 2 μg/ml was the preferred option because higher concentrations prevented detection of strains producing a variant of the TEM-1 β-lactamase (TRC-1) that is resistant to inhibition by clavulanic acid and also because this was a more cautious approach to susceptibility testing. However, the incidence of TRC-1 is difficult to ascertain (9). If such strains have become more prevalent, this is not apparent from an extensive literature review (7). As yet there are no clinical studies to assess which susceptibility testing guidelines offer the most accurate prediction of clinical outcome. The NCCLS breakpoints were found to be fully predictive of in vivo efficacy in a rat abscess model with E. coli strains that had different susceptibilities to amoxicillin-clavulanate by SFM and NCCLS criteria (12). Strains defined as intermediate or resistant in vitro according to SFM breakpoints were susceptible in vivo. Given the greater conservatism of DIN breakpoints than SFM breakpoints, one would expect all DIN intermediate strains, at least, to be found to be susceptible to amoxicillin-clavulanate in vivo.

The MIC distributions found for cefotaxime were also affected by methodology used, although the reason for this variation is unknown. The marked differences in the incidences of cefuroxime susceptibility categories reflected differences in NCCLS and DIN criteria.

This study has highlighted the need for worldwide standardization of the methodologies for determining MICs and for the breakpoint criteria used for their interpretation.