Previous Article | Next Article ![]()
Journal of Clinical Microbiology, August 2007, p. 2474-2479, Vol. 45, No. 8
0095-1137/07/$08.00+0 doi:10.1128/JCM.00089-07
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

AB BIODISK, Solna, Sweden,1 Infectious Diseases Research, Wyeth Research, Pearl River, New York2
Received 12 January 2007/ Returned for modification 22 March 2007/ Accepted 11 May 2007
|
|
|---|
|
|
|---|
The development and validation of reliable methods for antimicrobial susceptibility testing (AST) and MIC determinations of tigecycline are critical to clinical practice, as well as for ongoing surveillance programs, for this novel agent. Clinical microbiology laboratories have a number of AST methodologies available for daily clinical work. In order to validate and assess the reproducibility of Etest tigecycline (2, 5, 6), a multicenter study was conducted to compare MIC determinations with this method to the Clinical Laboratory Standards Institute (CLSI) reference broth microdilution and agar dilution (AD) assays. Test collections of organisms comprising clinical isolates, as well as strains having a wide range of susceptibilities to tetracyclines, tigecycline, and/or resistance to other antibiotics, were used in the comparative studies.
(This study was presented in part previously [A. Bolmström, Å. Karlsson, P. Ho, A. Wanger, and R. Howe, 45th Interscience Conference on Antimicrobial Agents and Chemotherapy, abstr. D-1645, 2005].)
|
|
|---|
|
View this table: [in a new window] |
TABLE 1. Composition of bias/precision bacterial test groups for Etest validation
|
|
View this table: [in a new window] |
TABLE 2. Composition of clinical and challenge bacterial test collections for Etest validation
|
|
View this table: [in a new window] |
TABLE 3. Bias and precision of Etest versus broth microdilution
|
Susceptibility testing. (i) Media. Cation-adjusted Mueller-Hinton II broth (MHB; Becton Dickinson, Sparks, MD) was used for broth microdilution with aerobes and Mueller-Hinton agar (Becton Dickinson) was used for Etest and AD with aerobes. A total of 5% lysed horse blood was added to MHB for testing S. pneumoniae and Streptococcus spp. with broth microdilution, and 5% laked sheep blood was added to Mueller-Hinton agar for the Etest. Anaerobes were tested with brucella blood agar base (Becton Dickinson) supplemented with 5% laked sheep blood, hemin, and vitamin K (BBA) according to the CLSI guidelines for AD, and the same agar was used for Etest. Haemophilus test medium (prepared in-house or purchased from Remel, Lenexa, KS) was used for both the broth microdilution and the Etest for H. influenzae.
(ii) Dilution procedures. Broth microdilution was carried out according to the CLSI guidelines for tigecycline (7). Specifically, in light of the demonstrated oxygen sensitivity of tigecycline (3, 17), MHB used for MIC determinations was prepared "fresh" (<12 h old at the time of use). Microdilution plates were prepared on the day of use, and the freshly prepared tigecycline stock solution was serially diluted in fresh MHB to provide a range of 15 twofold doubling dilutions (0.016 to 256 µg/ml) to match the Etest concentration gradient range. The MIC was determined as the lowest concentration of tigecycline that inhibited growth as judged by the unaided eye. The AD method for aerobes was performed according to CLSI guidelines (7, 16).
(iii) Etest.
Etest tigecycline (0.016 to 256 µg/ml; AB BIODISK, Solna, Sweden) was used according to the manufacturer's instructions. Briefly, an inoculum suspension with a turbidity equivalent to 0.5 McFarland standard was prepared by suspending well-isolated colonies in 0.9% saline for aerobes and in MHB for nonpneumococcal streptococci, S. pneumoniae and H. influenzae. For anaerobes, bacterial suspension in brucella broth with a turbidity equivalent to 1 McFarland was used. A sterile cotton swab dipped into the suspension was used to evenly streak the agar surface and allowed to dry for approximately 15 min. In the case of anaerobes, exposure to ambient air was minimized (
15 min). The Etest tigecycline gradient strip was applied to the agar surface, and the plate was incubated in ambient air at 35°C for 18 to 20 h for aerobes; in 5% CO2 for 20 to 24 h for nonpneumococcal streptococci, S. pneumoniae, and H. influenzae; and in an anaerobic chamber for 24 to 48 h for anaerobes. The MIC endpoint was read where the growth inhibition ellipse intersected the MIC on the Etest gradient strip. Whenever different growth inhibition patterns and/or growth trailing were seen, the MIC endpoint of the first point of significant inhibition, as judged by the naked eye, was selected using the reading guidelines and illustrations provided by the manufacturer.
QC. The following QC strains, as appropriate, were tested in parallel and on all test occasions for both the reference methods and Etest: S. aureus ATCC 29213, Enterococcus faecalis ATCC 29212, Escherichia coli ATCC 25922, S. pneumoniae ATCC 49619, H. influenzae ATCC 49247, Bacteroides fragilis ATCC 25285, Bacteroides thetaiotaomicron ATCC 29741, and Eubacterium lentum ATCC 43055. Inoculum density checks in terms of CFU/ml for all test procedures were performed with colony count assays for each QC strain and method and all bias/precision tests and for 10% of all clinical isolates for the different organism groups at all three study centers.
|
|
|---|
Bias and precision testing. The interlaboratory variability of Etest performance was assessed by comparing the Etest results for each organism reported by each study site for the groups of organisms in the 5-bias/precision collections (aerobes, anaerobes, nonpneumococcal streptococci, S. pneumoniae, and H. influenzae). The bias/precision collections were blinded to all three investigator sites. Interlaboratory reproducibility (±1 log2 dilution) was 100% for all five groups of organisms (Table 3). Etest MIC results from each study site were also compared to the mode of the reference result (n = 3) for each group of organisms (EA). For the 393 readings taken, the EA was between 88 and 100% for the three study sites. Most of the discrepant readings were for the H. influenzae collection, and when this collection was excluded the EA was 92 to 100% for the three sites (Table 3).
Clinical and challenge strains. Comparison of Etest results to reference methodologies for the clinical collections from all three sites plus the AST challenge collections from site 1 is presented in Table 4. The EA was greater than 98% for all five of the organism groups from the three study sites. Eight minor errors occurred with the gram-negative and gram-positive aerobes. In the case of the gram-negative aerobes, the seven minor errors were caused by two K. pneumoniae and five Serratia spp. for which the Etest MIC results were 1 dilution lower than the broth microdilution for four of the seven tests (Fig. 1). The one minor error that occurred with the gram-positive aerobes was caused by S. aureus Mu50 (ATCC 700699) for which the broth microdilution result was 1 µg/ml and the Etest result 0.38 µg/ml. Twelve minor errors occurred with the collection of anaerobes studied (Table 4). The minor errors were caused by three B. fragilis, three B. thetaiotaomicron, five B. ovatus, and one B. uniformis strain. Etest MICs were lower than the AD result for 9 of the 12 tests (Fig. 2). In contrast to the data from the bias/precision collection above, when the clinical collection of 372 H. influenzae isolates was tested the EA was found to be 98.7% with no CA errors (Table 4 and Fig. 3).
|
View this table: [in a new window] |
TABLE 4. Comparison of Etest and reference method for tigecycline susceptibility testing of the clinical and stock test collections including challenge strains
|
![]() View larger version (9K): [in a new window] |
FIG. 1. Error rate-bounded analysis comparing tigecycline reference broth dilution with tigecycline Etest strips against 221 gram-negative isolates. The datum points that fall along the diagonal line of the scattergram represent EA between the Etest result (x axis) and the broth microdilution result (y axis). Dotted horizontal and vertical lines demarcate the susceptible ( 2 µg/ml), intermediate (4 µg/ml), and resistant ( 8 µg/ml) categories along both axes. The datum points falling within the quadrant lines represent minor errors. Note that an Etest result that falls between twofold dilutions must be rounded up to the next upper twofold value before categorization. The gram-negative collection includes: Citrobacter spp., Enterobacter cloacae, E. coli, Klebsiella oxytoca, K. pneumoniae, Morganella morganii, Providencia stuartii, and Serratia spp.
|
![]() View larger version (9K): [in a new window] |
FIG. 2. Error rate-bounded analysis comparing tigecycline reference broth dilution with tigecycline Etest strips against 386 anaerobes. The datum points that fall along the diagonal line of the scattergram represent EA between the Etest result (x axis) and the AD result (y axis). Dotted horizontal and vertical lines demarcate the susceptible ( 4 µg/ml), intermediate (8 µg/ml), and resistant ( 16 µg/ml) categories. The datum points falling within the quadrant lines represent minor errors. Note that an Etest result that falls between twofold dilutions must be rounded up to the next upper twofold value before categorization. The anaerobe collection includes: Bacteroides caccae, B. distasonis, B. fragilis, B. ovatus, B. thetaiotaomicron, B. uniformis, B. ureolyticus, B. vulgatus, and other Bacteroides spp.
|
![]() View larger version (7K): [in a new window] |
FIG. 3. Error rate-bounded analysis comparing tigecycline reference broth dilution with tigecycline Etest strips against 374 H. influenzae isolates. The datum points that fall along the diagonal line of the scattergram represent the EA between the Etest result (x axis) and the broth microdilution result (y axis). Horizontal and vertical lines demarcate the hypothetical nonsusceptible MIC (MIC 0.5 µg/ml) along both axes. Note that an Etest result that falls between twofold dilutions must be rounded up to the next upper twofold value before categorization.
|
|
|
|---|
The data presented in the present study verify that the tigecycline Etest gradient method was as accurate as the reference methods for all five organism groups tested with excellent EA for the majority of the 1,926 strains tested. In addition, the error rates were very low. These outcomes are well within the target accuracy as suggested by the CLSI (15) and used by the U.S. Food and Drug Administration (10) to assess the substantial equivalence of the performance of alternative methods and products in comparison to reference methods.
The ability of Etest tigecycline to reliably detect tigecycline susceptible and resistant isolates was evaluated with a collection of 1926 strains resulting in 1.0% minor errors. It should be noted that EA for the nonsusceptible strains was 100% and the categorical agreement errors are a by-product of clustering at the breakpoints and the inherent tolerance (±1 dilution) of AST methodologies. U.S. Food and Drug Administration-defined breakpoints were applied in these studies (Wyeth Pharmaceuticals, Tygacil package insert [http://www.fda.gov/cder/foi/label/2005/021821lbl.pdf]): Staphylococcus (susceptible only breakpoint,
0.5 µg/ml), Streptococcus (not including S. pneumoniae) and Enterococcus (susceptible only breakpoint,
0.25 µg/ml), Enterobacteriaceae (susceptible,
2 µg/ml; intermediate, 4 µg/ml; resistant,
8 µg/ml), and anaerobes (susceptible,
4 µg/ml; intermediate, 8 µg/ml; resistant,
16 µg/ml). Hypothetical breakpoints were applied for S. pneumoniae and H. influenzae (susceptible only breakpoint,
0.25 µg/ml).
In the present study, testing was carried out independently at three test sites using collections of recently acquired clinical isolates and an AST challenge collection, including a representation of tigecycline-nonsusceptible isolates. However, in order to evaluate inter- and intralaboratory reproducibility, each site tested a common bias/precision collection of organisms that was blinded. Etest tigecycline demonstrated excellent reproducibility (100%) of the MIC results when the same 131 isolates were tested across all study sites. Only in the case of the H. influenzae group did two of the sites show less than optimal correlation (88%) with the broth microdilution reference method. Interestingly, in all cases, Etest MIC results were >1 log2 dilution higher than the reference broth dilution method. The less-than-optimal reproducibility observed with H. influenzae could be attributed among other factors to variability in the quality of the Haemophilus test medium agar and ambient incubation of capnophilic organisms in broth that may influence AST results. It should be noted that for the much larger collection of H. influenzae (372 clinical isolates), the EA was 98.7%, with EA rates of 95.9, 99, and 100% for the three respective sites (Table 4 and Fig. 3).
In conclusion, Etest tigecycline gradient strips proved to be robust and reliable even when tested with large collections of diverse organism groups in this multicenter analysis and should provide accurate and reproducible MIC results when used in daily clinical practice.
Published ahead of print on 23 May 2007. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»