Interlaboratory Comparison of Agar Dilution and Etest Methods for Determining the MICs of Antibiotics Used in Management of Neisseria meningitidis Infections

Abstract

Previous studies have shown that there is considerable variation in the methods and media used to determine the susceptibility of Neisseria meningitidis to antimicrobial agents in different countries. In this study, national and regional reference laboratories used a standardized methodology to determine the MICs of antibiotics used in the management of meningococcal infection. Fourteen laboratories participated in the study, determining the susceptibility to penicillin G, rifampin, cefotaxime, ceftriaxone, ciprofloxacin, and ofloxacin of a collection of 17 meningococci, of which 11 strains were previously defined as having intermediate resistance to penicillin (Pen^I) by sequencing and restriction fragment length polymorphism analysis of the penA gene. The MIC was determined by agar dilution and Etest with Mueller-Hinton agar (MH), MH supplemented with sheep blood (MH+B), and MH supplemented with heated (chocolated) blood. Several laboratories encountered problems obtaining confluent growth with unsupplemented MH. MH+B was considered to give the most congruent and reproducible results among the study laboratories. The modal MIC for MH+B for each antibiotic and method was calculated to define the MIC consensus, allowing assessment of each individual laboratory's data in relation to the others. The agreement in each antibiotic/method/medium combination was defined as the percentage of laboratories with a result within one dilution of the modal result. For the whole study, an agreement of 90.6% was observed between agar dilution and Etest methods. The agreement in each laboratory/antibiotic/method combination ranged from 98.2% to 69.7%, with six laboratories demonstrating agreement higher than 90% and 11 more than 80%. The ability of the laboratories to detect the Pen^I isolates ranged from 18.2% to 100%. The apparent difficulty in interpreting susceptibility to rifampin, particularly with the Etest method, is very interesting.

Neisseria meningitidis, the etiological agent of meningococcal disease, remains susceptible to many antimicrobial agents. Apart from resistance to sulfonamides, meningococci remain susceptible to the antibiotics classically used for treatment and chemoprophylaxis (15). But, since 1985, decreased susceptibility to penicillin caused by alterations in the penicillin-binding proteins (12) has been noted in some countries (11, 14, 15). In addition, several meningococcal strains with a high level of penicillin resistance due to beta-lactamase production have also been described (4). Resistance to rifampin is only occasionally observed, and this may be identified following chemoprophylaxis (8).

Because meningococcal disease is such a serious and rapidly progressing illness, it is very important to monitor trends in the resistance to antimicrobial agents in meningococci in each country. In order to compare data between laboratories and monitor resistance trends, it is critical to use standardized protocols for the determination of susceptibility to antibiotics. The National Committee for Clinical Laboratory Standards (NCCLS) recommends a specific methodology to determine the MIC of antibiotics against meningococci (10), but breakpoints to define different categories of susceptibility have not been established. The NCCLS recommends the use of broth microdilution in cation-adjusted Mueller-Hinton (MH) broth supplemented with lysed horse blood or agar dilution with MH supplemented with sheep blood. But, the Etest method, commonly used in Europe, is a practical alternative for clinical laboratories when a quick result is clinically or epidemiologically important (9). Alternatively, the use of a disk diffusion test with 2 U of penicillin or 1 μg of oxacillin has been proposed for that purpose (7), but there is some concern about the reliability of this methodology (6), and further multicenter studies are required to clarify the use of this method.

In order to identify the methodologies used for MIC determination in different countries, a questionnaire compiled by the European Monitoring Group of Meningococci (EMGM) was sent to 26 national and regional reference laboratories, including not only European centers but also laboratories in Australia, Israel, and the United States (5). The results of the questionnaire highlighted the heterogeneity of methods, particularly with regard to the media used. A similar degree of variability was also observed with respect to the breakpoints used. While susceptibility data can monitor trends within a given country or region, the lack of standardization makes it very difficult to compare susceptibility data between countries.

The EMGM, during its fifth meeting in Greece in 1999, concluded that it would be useful and instructive to apply a standard methodology for determining the MICs of a panel of meningococcal strains, particularly focusing on the agar dilution and Etest methods, which, according to the results of the questionnaire, were the most frequently used across Europe. The results of the project, coordinated by the Spanish Reference Laboratory for Neisseria (SRLN), are reported in this study.

MATERIALS AND METHODS

The SRLN prepared and distributed a collection of meningococcal strains together with 500 g of MH dehydrated medium, the antibiotics, and a protocol for performing dilutions and preparing media to determine the MICs of the strains by the agar dilution and Etest methods. The strains were tested on MH, MH supplemented with sheep blood (MH+B), and MH supplemented with heated (chocolated) sheep blood (MH+CH).

Strains.

A collection of 17 freeze-dried meningococcal strains isolated in Spain between 1992 and 1999 was prepared. The isolates, according to the data of the SRLN, represented a range of levels of susceptibility to antimicrobial agents, particularly to penicillin G and rifampin. The characteristics of these strains are shown in Table 1.

TABLE 1.

Characteristics of the strains used in the EMGM standardization study

EMGM strain no.	Antigen expressiona	Yr of isolation	Originb
1	C:2a:P1.5,2	1999	Blood
2	NG:NT:NST	1992	Nasopharynx
4	B:4:P1.15	1999	Nasopharynx
5	B:1:NST	1999	Blood
6	B:NT:P1.3,6	1999	Nasopharynx
7	B:4:P1.15	1999	Nasopharynx
8	B:4:P1.15	1999	Blood
9	B:4:NST	1998	CSF
10	C:2b:P1.5,2	1999	Blood
11	B:NT:P1.9	1999	CSF
12	B:1:NST	1999	Nasopharynx
13	C:2b:NST	1992	Blood
14	C:2a:P1.5	1999	CSF
15	C:2a:P1.5	1994	Blood
16	C:2b:NST	1992	Blood
17	B:1:NST	1999	Blood
18	C:2b:NST	1992	Blood

^aSerogroup/serotype/serosubtype combination.

^bCSF, cerebrospinal fluid.

^cNG, nongroupable; NT, nonserotypeable; NST, nonserosubtypeable.

Media and antibiotics.

The same designated batch of MH (Oxoid, Basingstoke, Hampshire, England) was distributed to all the participating laboratories. The antibiotics to be used were chosen by consensus and included penicillin G, rifampin, ciprofloxacin, ofloxacin, cefotaxime, and ceftriaxone. Batches of antibiotics were sourced from Cepa Schwarz Pharma S.L. for ciprofloxacin and Sigma-Aldrich, Inc., for the other antibiotics. The Etest strips were obtained from the sole manufacturer (AB Biodisk, Sweden) via local suppliers by each laboratory.

Protocols.

Protocols for producing the dilutions and media to determine the MICs of the strains by agar dilution and Etest methods were distributed to all participating laboratories. Antimicrobial susceptibility was determined by two methods (agar dilution and Etest) with three culture media (MH, MH+B, and MH+CB) as previously described (5, 9).

For agar dilution, the antibiotics were used in serial twofold dilutions, with concentrations ranging from 0.007 to 2 μg/ml for penicillin G, 0.0003 to 0.03 μg/ml for cefotaxime, 0.00007 to 0.06 μg/ml for ceftriaxone, 0.007 to 64 μg/ml for rifampin, and 0.0007 to 0.12 μg/ml for ciprofloxacin and ofloxacin. Although the accuracy of concentrations below 0.004 μg/ml is not very good, a wide range of concentrations was included to be able to compare results at these low concentrations.

The organisms were applied to the plates at a final concentration of 10⁵ CFU per spot. After allowing the surface to absorb the inoculum, all the plates were incubated for 24 h at 35°C in 5% CO₂. The MIC was defined as the lowest concentration of antibiotic that prevented visible growth.

For the Etest method, 0.5 McFarland standard suspensions of the test organisms were prepared and inoculated onto the plate with a nontoxic swab to produce confluent growth. Etest strips containing the antimicrobial agent were placed on the inoculated plates with sterile forceps. Some laboratories used 150-mm plates, placing all six strips on it; others used 90-mm plates, placing two strips in each; but most of the laboratories used 90-mm plates, placing only one Etest strip in each. The plates were incubated for 24 h at 35°C in 5% CO₂ and then the MICs were determined according to the manufacturer's instructions.

Each laboratory used its own method to determine the inoculum concentration for the 10⁵ CFU per spot and 0.5 McFarland.

Sequence analysis and RFLP of the penA gene.

Because intermediate resistance to penicillin is produced by alterations in PBP2, encoded by the penA gene, this gene was sequenced, by the SRLN, in all 17 strains as previously described (3). The Institute Pasteur also analyzed the penA gene by restriction fragment length polymorphism (RFLP) according to a published method (1).

Analysis of the data.

All the data were included in a Microsoft Excel file, and the analysis was done with this software.

The congruence of the data was analyzed to identify the optimal medium. For this purpose, the agreement in each antibiotic/method/medium combination was defined as the percentage of laboratories with a result within one dilution of the modal result.

The modal result was calculated for each strain against each antibiotic with the most optimal medium. These modal results defined the MIC for each strain/antibiotic combination. Two results were considered coincident (the same) if they were within ±1 dilution.

RESULTS

The agreement for each antibiotic/method/medium combination is shown in Table 2. The agreement was very similar in MH+B and MH+CH media, but overall, MH+B offered the best agreement, being the medium with the most homogeneous results among laboratories, and was therefore adopted for most of the analyses. Three laboratories experienced difficulty obtaining confluent growth with MH. No laboratories reported difficulties in the interpretation of the Etest method. We were unable to analyze the agreement for ceftriaxone with the Etest because all the strains had MICs lower than the minimum (<0.002 μg/ml) present on the Etest strips.

TABLE 2.

Agreement obtained in each antibiotic/method/medium combination^a

Drug	Method	Agreement (%)
		MH	MH+B	MH+CH
Penicillin	AD	81.2	91.6	90.8
	Etest	75	87.6	82.9
Rifampin	AD	77	78.8	82
	Etest	70.7	75.6	74.2
Ciprofloxacin	AD	87.6	85.3	88.6
	Etest	77.4	91.6	87.1
Ofloxacin	AD	98.5	94.1	91.5
	Etest	85.9	98.8	100
Cefotaxime	AD	77.3	85.6	81
	Etest	74.7	90.9	86.1
Ceftriaxone	AD	76.6	86	87.7
	Etest	ND	ND	ND
Total	AD	83.0	86.9	86.9
	Etest	76.7	88.9	86.1

^aThe highest percentage of coincident MICs among all participating laboratories. The best agreement in each case is in italics. Two or three different media in bold for the same antibiotic and method indicates that differences were not statistically significant. ND, not determined; AD, agar dilution.

The MICs obtained by calculating the modal result in MH+B are shown in Table 3. The MICs were called the MIC consensus. The MIC was not calculated for ceftriaxone with Etest for the reason stated previously. An agreement of 90.6% was observed between the agar dilution and Etest methods (Table 4). The best agreement was observed with penicillin G, ciprofloxacin, and cefotaxime.

TABLE 3.

Modal MICs obtained in MH+B for the both, agar dilution, and Etest methods^a

Modal MIC (μg/ml)
EMGM strain no.	Method	Penicillin G	Rifampin	Cefotaxime	Ceftriaxone	Ciprofloxacin	Ofloxacin
1	AD	0.03	0.015	0.003	0.0015	0.007	0.03
	Etest	0.032	0.016	0.003		0.003	0.016
2	AD	0.03	>64	0.003	0.0007	0.007	0.03
	Etest	0.023	>32	<0.002		0.003	0.016
4	AD	0.03	0.03	0.003	0.0007	0.003	0.015
	Etest	0.016	0.023	<0.002		0.003	0.016
5*	AD	0.06	0.007	0.0015	0.0007	0.003	0.015
	Etest	0.064	0.008	0.002		0.003	0.012
6*	AD	0.12	0.007	0.007	0.0015	0.003	0.03
	Etest	0.125	0.004	0.004		0.003	0.016
7*	AD	0.5	0.015	0.015	0.0015	0.003	0.015
	Etest	0.38	0.016	0.008		0.003	0.012
8*	AD	0.25	0.03	0.007	0.0015	0.007	0.03
	Etest	0.125	0.023	0.004		0.002	0.012
9*	AD	0.25	0.03	0.007	0.0015	0.003	0.03
	Etest	0.125	0.016	0.003		0.004	0.012
10*	AD	0.25	0.007	0.015	0.003	0.007	0.015
	Etest	0.25	0.004	0.008		0.003	0.016
11	AD	0.015	0.007	0.0007	0.007	0.007	0.015
	Etest	0.012	0.0023	<0.002		0.003	0.016
12*	AD	0.12	0.007	0.003	0.0015	0.003	0.015
	Etest	0.125	0.0023	0.003		0.003	0.012
13*	AD	0.5	8	0.007	0.0015	0.003	0.03
	Etest	0.38	6	0.003		0.007	0.016
14	AD	0.03	0.015	0.003	0.0015	0.003	0.015
	Etest	0.032	0.023	0.002		0.003	0.016
15*	AD	0.5	0.03	0.015	0.003	0.007	0.015
	Etest	0.5	0.032	0.004		0.004	0.016
16*	AD	0.5	8	0.007	0.0015	0.007	0.03
	Etest	0.38	>32	0.004		0.003	0.012
17	AD	0.03	0.015	0.003	0.0007	0.007	0.03
	Etest	0.032	0.008	<0.002		0.003	0.012
18*	AD	0.5	8	0.007	0.0015	0.007	0.015
	Etest	0.38	6	0.004		0.003	0.012

^aResults differing by more than ±1 dilution comparing agar dilution with Etest are in italics. *, strains identified as being intermediate to penicillin by sequencing of the penA gene and RFLP.

TABLE 4.

Agreement obtained between the agar dilution and Etest methods for the determination of the consensus MIC of each antibiotic^a

Antibiotic	Agreement (%)
Penicillin G	100
Rifampin	82.4
Ciprofloxacin	94.1
Ofloxacin	76.5
Cefotaxime	94.1
Total	90.6

^aCeftriaxone is not included because the MIC was not determined by Etest.

The agreement of each laboratory/antibiotic/method combination is shown in Table 5. The results for each antibiotic were heterogeneous, with some laboratories showing large variations in agreement; laboratory number 12 ranged from 23.5% to 100%, and laboratories 2 and 9 ranged from 41.2% to 100%. The most homogeneous results were obtained by laboratory 8 (Table 5). Of note was the apparent difficulty in determining susceptibility to rifampin, particularly with the Etest method. In fact, no laboratory obtained a 100% agreement in this antibiotic/method combination (Table 5). The worst results (obtained by laboratories 5 and 14) were in determining MICs of ceftriaxone in agar dilution (Table 5); but, this antibiotic/method combination offered very good and homogeneous results in the other laboratories, even at very low concentrations.

TABLE 5.

Agreement for each laboratory/antibiotic/method combination^a

Laboratory no.	Agreement (%)
	Penicillin G		Rifampicin		Ciprofloxacin		Ofloxacin		Cefotaxime		Ceftriaxone (AD)
	AD	Etest	AD	Etest	AD	Etest	AD	Etest	AD	Etest
1	100	100	88.2	86.7	100	100	100	100	88.2	—	100
2	47.05	82.3	70.6	87.5	100	47	—	100	41.2	82.3	100
3	94.1	100	94.1	68.7	64.7	100	100	—	94.1	—	100
4	100	70.6	82.3	81.2	100	100	100	100	100	100	100
5	88.2	100	76.5	—	64.7	—	64.7	100	94.1	—	0
6	88.2	94.1	88.2	50	88.2	100	88.2	100	94.1	100	100
7	94.1	94.1	76.5	87.5	94.1	100	94.1	100	93.7	82.3	100
8	100	100	100	80	100	100	100	100	100	100	100
9	94.1	100	64.7	73.3	41.2	88.2	100	100	100	100	100
10	100	100	58.8	50	88.2	94.1	100	100	47	100	100
11	82.3	100	76.5	81.2	64.7	100	100	100	82.3	94.1	100
12	—	23.5	—	81.2	—	100	—	100	—	94.1	100
13	100	62.5	75	66.7	100	100	87.5	100	86.7	87.5	100
14	100	100	93.3	92.8	100	100	93.3	100	100	100	6.7

^aAD, agar dilution; —, MICs not determined..

Comparing the penicillin MICs with sequencing and RFLP analysis of the penA gene, 11 meningococci of the collection were defined as having intermediate resistance to penicillin (Pen^I). This information was used to analyze the ability of the laboratories to detect Pen^I strains. The meningococci identified as intermediate were those numbered EMGM 5, 6, 7, 8, 9, 10, 12, 13, 15, 16, and 18. If we accept that the Pen^I strains are those defined by sequence/RFLP of the penA gene, only three laboratories were able to detect all of these meningococci by the agar dilution and Etest methods. One laboratory was unable to detect five (45.4%) intermediate strains by agar dilution, while another detected only two (18.2%). The MICs of penicillin were found to be higher than 0.06 μg/ml (the breakpoint used to define intermediate resistance) for all strains previously defined as Pen^I meningococci except strain EMGM5 (Table 3). Strain EMGM5, assigned a MIC consensus corresponding to a susceptible isolate, was the most difficult to determine as Pen^I.

DISCUSSION

Three laboratories reported problems obtaining confluent growth in MH without supplement. MH+B and MH+CH gave very similar results of general agreement (Table 2), but MH+B is easier to prepare and additionally gave the results closest to the consensus overall in each antibiotic/method combination. We therefore recommend the use of MH (or some other proper basal medium) supplemented with 5% blood for the routine determination of MICs for N. meningitidis. Several laboratories used their own basal medium, and the results were comparable (data not showed).

The agreement between the agar dilution and Etest methods was good, being 100% for penicillin G (Table 4). This is a very important finding because many laboratories use the Etest method as the routine method to determine the MIC for meningococcal strains (5). This would allow accurate comparison of the incidence of Pen^I isolates between laboratories using different methods. One group detected unusual patterns of growth along the Etest strips, and that finding might have made it more difficult to determine the MIC for the strains showing that phenomena.

The agreement between agar dilution and Etest was worse with ofloxacin and rifampin, but the discrepancies were very close to the range of ±1 dilution difference. This finding is very important for rifampin, because the discrepancies were found in strains with very low MICs (EMGM 11, EMGM 12) and also those showing resistance (EMGM 16) (Table 3). However, both methods defined the same strains as resistant or susceptible to rifampin, so the discrepancies may affect the epidemiological surveillance trends but not the clinical information. All the strains showed a high level of susceptibility to ofloxacin, so we could not compare the sensitivity of each method in detectong different levels of susceptibility. The comparison between the agar dilution and Etest methods will be interesting to study further if strains showing decreased susceptibility to that antimicrobial agent appear (13).

There was one strain (EMGM 5) possessing a sequence typical of a Pen^I isolate that demonstrated a MIC consensus inside the susceptibility range with both methods (Table 3). The study shows a limitation at this level: no strains with a MIC of 0.094 μg/ml by Etest were included. The NCCLS does not specify the breakpoint to define the susceptibility levels of penicillin with microdilution or macrodilution, but those described for Neisseria gonorrhoeae have been used for meningococci (2). These breakpoints are ≤0.06 μg/ml for susceptible, 0.12 to 1 μg/ml for intermediate, and ≥2 μg/ml for resistant strains (10).

Both methods (microdilution and agar dilution) are based on a range of doubling dilutions. With those breakpoint definitions, there will be several gaps by Etest, and strains with MICs of 0.094 μg/ml or between 1 and 2 μg/ml might be misclassified. The definition of the breakpoints to be used to define different levels of susceptibility should take into account the possibility that Etest will be used. In the same way, it is very important that all laboratories work with the same range of doubling dilutions in order to apply definite categories. To this end, the range of doubling dilutions that allow the classification of the strains according to breakpoints defined for gonococci by the NCCLS should be used (10).

Because the main area of concern with meningococci is intermediate resistance to penicillin, most of the study was focused on the ability of each laboratory to detect this level of resistance. Ten strains were defined as Pen^I according to the common definition. But the only mechanism well known to increase the level of resistance is genetic modification in the penA gene. The sequence of this gene and the RFLP patterns of the gene allowed us to identify 11 Pen^I meningococci. No isolate with a sequence characteristic of susceptible strains presented a MIC consensus higher than 0.06 μg/ml, so we do not think that mechanisms other than the modification of the penA gene contribute to this intermediate resistance level. The EMGM 5 strain, well defined as intermediate by the sequence or RFLP pattern, had a MIC consensus of 0.06 μg/ml by the agar dilution and 0.064 μg/ml by the Etest method (Table 3). The real MIC of this strain should be close to 0.1 μg/ml; in fact, five laboratories found MICs higher than 0.06 μg/ml by Etest, ranging between 0.094 and 0.125 μg/ml. The development of a molecular approach to define and determine a Pen^I strain may be warranted.

This panel of serologically well defined meningococci with known penicillin resistance determinants could be used in further studies to define quality control limits and MIC breakpoints for meningococci. We propose the use of four isolates with a high agreement between laboratories (EMGM 1, 2, 10, and 13) as the basis of a strain collection to make available for use by others working in the field of meningococcal antibiotic resistance. These isolates belong to the EMGM and are kept at the Spanish Reference Laboratory for Meningococci (National Institute of Health Carlos III), where they can be obtained by request.