INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Antimicrobial resistance among clinical isolates of Streptococcus pneumoniae emerged as a significant global problem in the decade of the 1990s (1, 5, 19). The prevalence of penicillin and multidrug-resistant strains increased substantially during this period in the United States (2, 3, 5, 14, 20). The National Committee for Clinical Laboratory Standards (NCCLS) has developed reference broth microdilution and disk diffusion methods for the testing of S. pneumoniae, as well as relevant quality control and interpretive criteria for both of those methods (15, 16, 17). However, the disk diffusion procedure does not provide acceptable accuracy when pneumococci are tested with various beta-lactams (11). The routine determination of MICs of penicillin and selected extended-spectrum cephalosporins is now recommended for isolates from patients with serious pneumococcal infections (17). Commercially prepared broth microdilution panels that are derived from the NCCLS reference method and the antibiotic gradient diffusion method are available for routine testing of S. pneumoniae by clinical laboratories (8, 9, 10). However, all of the aforementioned methods require 20 to 24 h of incubation for provision of results.

A new, more automated instrument for rapid identification and antimicrobial susceptibility testing has recently been developed by bioMerieux. Designated the VITEK 2, the new instrument uses bar coding and a coded computer chip for specimen and test identification and provides robotics for card filling and incubation (J.-P. Gayral, R. Robinson, and D. Sandstedt, microbiological testing. Abstr. Eur. Congr. Clin. Microbiol. Infect. Dis., abstr. P254, 1997). A new 64-well card is read by an automated photometer which takes kinetic turbidimetric readings every 15 min during the incubation and analysis period. Computer kinetic analysis of growth readings is used to determine the MICs of the antimicrobial agents included in various VITEK 2 test cards. Unlike the original VITEK instrument, the capability of testing certain fastidious organisms (e.g., S. pneumoniae) will be provided by the VITEK 2. The present study, conducted at three different sites in two phases, has evaluated a prototype of the VITEK 2 and an S. pneumoniae susceptibility testing card that has been formulated with a modified Wilkins-Chalgren medium and antimicrobial agents appropriate for pneumococci. Approximately one-half of the antimicrobial agents were evaluated in phase 1 of this study, and the remaining agents were evaluated in Phase 2. The VITEK 2 represents the first commercially available method for the rapid determination (<15 h) of the MICs of antimicrobial agents for pneumococci.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Test organisms. A collection of 54 pneumococcal challenge strains with known resistance properties and 407 and 423 unique clinical isolates (in phases 1 and 2, respectively) of S. pneumoniae were selected for use in this evaluation. The clinical isolates were selected to include 143 strains with previously documented resistance to one or more antimicrobial agents (Table 1). Approximately one-third of the clinical isolates were provided by and tested in each of the three study laboratories. In addition, 10 and 12 strains of S. pneumoniae were tested by all sites repetitively in phase 1 and phase 2, respectively, for determination of the reproducibilities of the MICs determined by the VITEK 2 method.

Antimicrobial agents. The antimicrobial agents (concentration ranges) used in the reference broth microdilution panels were as follows: penicillin (0.008 to 32 µg/ml), amoxicillin (0.06 to 32 µg/ml), cefotaxime (0.03 to 16 µg/ml), ceftriaxone (0.015 to 64 µg/ml), vancomycin (0.25 to 32 µg/ml), trimethoprim-sulfamethoxazole (0.06/1.14 to 32/608 µg/ml), erythromycin (0.06 to 32 µg/ml), chloramphenicol (0.5 to 128 µg/ml), ofloxacin (0.25 to 32 µg/ml), and tetracycline (0.25 to 64 µg/ml). The VITEK 2 susceptibility cards contained modified Wilkins-Chalgren medium with the following antimicrobial agents (equivalent concentrations): penicillin (0.06 to 2 µg/ml), amoxicillin (0.06 to 4 µg/ml), cefotaxime (0.06 to 4 µg/ml), ceftriaxone (0.06 to 4 µg/ml), vancomycin (1 and 2 µg/ml), trimethoprim-sulfamethoxazole (0.5/9.5 to 16/304 µg/ml), erythromycin (0.06 to 1 µg/ml), chloramphenicol (2 to 32 µg/ml), ofloxacin (1 to 8 µg/ml), and tetracycline (1 to 16 µg/ml).

VITEK 2 susceptibility tests. The VITEK 2 cards were inoculated and filled according to the manufacturer's instructions. Briefly, this included preparation of inocula of the test organisms from colonies grown on 5% sheep blood agar plates (Remel, Lenexa, Kans.) that had been incubated for 18 to 20 h in 5% CO₂. The colonies were suspended in 0.45% saline to obtain a suspension with a turbidity equivalent to the turbidity of a 0.5 McFarland standard and were further diluted by transferring 275 µl into 2.5 ml of 0.45% sterile saline. This adjusted inoculum suspension was used to fill the susceptibility cards, which were then inserted into the prototype VITEK 2 incubator-reader within 15 min of preparation. The same 0.5 McFarland suspensions were also used to inoculate the reference microdilution panels at the same time.

Broth microdilution reference susceptibility tests. The reference susceptibility method consisted of frozen broth microdilution panels prepared for this study by PASCO Laboratories (Aurora, Colo.). The panels contained cation-adjusted Mueller-Hinton broth supplemented with a final concentration of 2.5% lysed horse blood in accordance with the NCCLS reference procedure for susceptibility testing of S. pneumoniae (15). The contents of the tubes with the 0.5 McFarland suspensions in saline used for the VITEK card inoculation were further diluted by adding 1.5 ml of the suspension to 12.5 ml of concentrated lysed horse blood supplement for the reference microdilution panels. This resulted in a final inoculum density of 3 × 10⁵ to 7 × 10⁵ CFU/ml and 2.5% lysed horse blood in the wells of the microdilution panels following transfer with a disposable 96-prong plastic inoculator. The reference panels were incubated at 35°C in ambient air for 20 to 24 h prior to visual determination of MICs.

Quality control organisms. The NCCLS control strain S. pneumoniae ATCC 49619 was used for daily quality control for both the VITEK 2 and the reference MIC procedures. In addition, Enterococcus faecalis ATCC 29212, E. faecalis ATCC 51299, Staphylococcus aureus ATCC 29213, and Pseudomonas aeruginosa ATCC 27853 were used to provide on-scale MICs of all drugs included in the reference MIC panels.

Comparison of results. The MIC of each antimicrobial agent generated by the prototype VITEK 2 was compared with the MIC of each agent determined by the reference broth microdilution procedure. A susceptibility category was also assigned to each MIC on the basis of the current NCCLS interpretive breakpoint criteria (17). Interpretive category errors were assessed for each drug on the basis of the following definitions: very major error, susceptible by the VITEK 2 method but resistant by the reference method; major error, resistant by the VITEK 2 method but susceptible by the reference method; minor error, intermediate by either the VITEK 2 or reference method and either susceptible or resistant by the other method. It should be noted that only the number of resistant strains was used as the denominator for calculation of very major errors and only the number of susceptible strains was used as the denominator for calculation of major errors. Both the VITEK 2 and reference tests were repeated in triplicate when very major or major errors were noted from the initial tests. The final error rates were calculated on the basis of the values of the repeat test when errors were resolved.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

This study has evaluated the performance of a prototype of the VITEK 2 instrument for rapid susceptibility testing of a collection of challenge strains and clinical isolates of S. pneumoniae. The tests with the VITEK 2 instrument were simple to perform, and the MICs determined with the VITEK 2 instrument were quite reproducible between the three laboratories for the challenge collection of strains and 10 antimicrobial agents (>97% agreement of MICs within a single log₂ dilution; data not depicted further). Despite the use of the lysed horse blood supplementation of the reference microdilution panels, the MICs for the nonfastidious NCCLS control strains were within the published limits of NCCLS (17) for those drugs for which use of only the S. pneumoniae standard control strain would not provide on-scale values.

With 7 of the 10 drugs included in the prototype VITEK 2 cards, on-scale MICs were often generated. The MICs determined by the VITEK 2 method compared favorably (96.6 and 96.3% overall agreements within ±1 log₂ dilution for the challenge strains for the clinical isolates, respectively) with the MICs determined by the reference method (Tables 2 and 3). The lowest degree of precise agreement of MICs occurred with penicillin (83.3% with the challenge strains and 89.9% with the clinical isolates) and trimethoprim-sulfamethoxazole (89.7% with clinical isolates). The MICs determined by the VITEK 2 method were slightly higher than those determined by the reference method for most of the drugs with the challenge strains and for several drugs with the clinical isolates (Tables 2 and 3). Despite this, the level of agreement of MICs within a single dilution for both organism groups was greater than 94%. Off-scale values occurred too frequently with the VITEK 2 tests for meaningful comparison for erythromycin, tetracycline, and vancomycin because of the limited number of concentrations of those agents in the VITEK 2 cards around the interpretive breakpoints for the drugs. This meant that the MICs for some strains with high-level erythromycin or tetracycline resistance were reported to be greater than the highest equivalent concentration included in the cards. Conversely, a number of vancomycin MICs were reported as 1 µg/ml (the susceptible breakpoint), since vancomycin resistance has not been encountered in pneumococci.

Relatively few very major and major interpretive category errors resulted from the VITEK 2 tests with the challenge or clinical isolates (Tables 4 and 5). These error rates were calculated after 17 tests were repeated in triplicate because of initial very major or major errors; 6 of 17 errors were rectified by repeat testing. With the clinical isolates there were 10% very major errors with chloramphenicol, owing mostly to the lack of an intermediate interpretive category with that drug, and the clustering of many MICs in the range of the single breakpoint between 4 and 8 µg/ml. Conversely, there were occasional major errors with four of the beta-lactams with the challenge strains but not with the clinical isolates (Tables 4 and 5).

The highest rate of minor errors with the clinical isolates occurred with trimethoprim-sulfamethoxazole, primarily due to the tendency for the reporting of higher MICs by the VITEK 2 method (Tables 2, 3, 4, and 5). A number of minor errors with the beta-lactams was noted to be due to the clustering of the MICs for both the challenge and resistant clinical isolates near the interpretive breakpoints. However, many of these minor interpretive errors were associated with insignificant (±1 log₂ dilution) differences in MICs (Tables 4 and 5). Indeed, when only those minor errors due to MICs determined by the VITEK 2 method that were greater than 1 log₂ dilution from the values determined by the reference method were considered, only the results for tests with penicillin and the challenge strains were associated with substantial rates (>5%) of minor errors.

The times for provision of VITEK 2 susceptibility test results are depicted in Table 6 for each drug and for the challenge strains and clinical isolates. The clinical isolates required slightly longer analysis times, although the mean time for completion of all tests was 8.1 h. Some results were available in as little as 3.4 h, and all tests were complete by 14 h of incubation and analysis. There was no association between the length of time required for results with a particular strain and the agreement of the values obtained by the VITEK 2 method with the values obtained by the reference method.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The increased prevalence of penicillin and multidrug resistance among S. pneumoniae strains in the past decade in North America (2, 3, 5, 7, 19, 20) has made it necessary for clinical laboratories to reevaluate their routine practices for susceptibility testing of pneumococcal isolates. Practical options for testing include the NCCLS disk diffusion procedure, tests with several commercial broth microdilution products, and the E test (8). However, none of the current methods offer the possibility of generating results in less than 18 to 24 h or offer automated reading of test panels. Thus, the VITEK 2 represents the first system for provision of rapid susceptibility test results with pneumococci. The VITEK 2 tests with the clinical isolates in this study were completed in an average of 8.3 h.

In the VITEK 2, inoculation and manipulation of the test cards for pneumococcal isolates were accomplished in the same manner as those procedures for tests with nonfastidious organisms. Because this study used a prototype of the VITEK 2 rather than a final production version of the instrument, it is not possible to state the amount of hands-on time required to perform the testing of pneumococci. However, the new instrument is designed to provide a card-filling and processing system that is more streamlined and efficient than the current VITEK and is said to require only about 25% as much hands-on time for each test compared to that required for a broth microdilution procedure (Gayral et al., Abstr. Eur. Congr. Clin. Microbiol. Infect. Dis.).

This study has assessed the ability of the VITEK 2 to generate accurate susceptibility testing results with a carefully selected group of challenge strains and clinical isolates of pneumococci that included a large number of resistant strains. The collection included strains resistant to all of the drugs in this study with the exception of vancomycin. Furthermore, isolates were selected in part because the beta-lactam MICs for the isolates spanned the breakpoints for susceptibility, intermediate, and resistance. In general, the susceptibility results generated with VITEK 2 agreed closely with the results generated by the reference method. Very few very major or major interpretive errors resulted from the tests conducted in this study. Because of the large number of isolates for which MICs were adjacent to the NCCLS breakpoints, the normal minor error rate calculations for the penicillins and cephalosporins seem excessive. However, when the MICs that differed by only a single log₂ dilution were excluded, the number of minor interpretive errors was quite modest, i.e., less than 5% for tests with all drugs except penicillin with the challenge organisms.

The VITEK 2 provided accurate MICs of most of the antimicrobial agents; the notable exceptions were penicillin and trimethoprim-sulfamethoxazole (only for the clinical isolates). The trimethoprim-sulfamethoxazole MICs may have been elevated due to the greater thymidine content of the Wilkins-Chalgren broth compared to that of the lysed horse blood-supplemented broth used in the reference procedure, since lysed horse blood is a source of thymidine phosphorylase that reduces folate antagonists in media. The penicillin MICs differed by 2 doubling dilutions for 8 of 48 challenge strains and 10 of 98 clinical isolates. However, the amoxicillin and cephalosporin MICs generated by the VITEK 2 method agreed closely with the values generated by the reference method (i.e., >94% essential agreement). The reliable determination of MICs of beta-lactam antibiotics has recently become important because of the recognition that patients with pneumococcal pneumonia caused by strains for which penicillin or parenteral cephalosporin MICs are in the intermediate category are amenable to treatment with those agents (6, 13, 18). Thus, determination of the MICs of those agents is now critical for selection of appropriate therapy for patients with life-threatening infections (e.g., meningitis) and for patients with common respiratory tract infections due to pneumococci (e.g., pneumonia) (6, 17).

Areas for improvement or further development of the VITEK 2 for the testing of pneumococci include possible further refinements of the interpretive algorithms that might reduce the minor errors associated with penicillin testing. In addition, it would be helpful to extend upward the equivalent concentrations of erythromycin so that high-level erythromycin resistance (usually due to ribosomal target modification) could be separated from low-level macrolide resistance due to drug efflux (5). Lastly, laboratories will undoubtedly wish to be able to test newer fluoroquinolones with improved activity against pneumococci (e.g., levofloxacin, gatifloxacin, gemifloxacin, and moxifloxacin) in light of emerging fluoroquinolone resistance (4, 12), and carbapenems (e.g., imipenem and meropenem). Thus, additional drugs are needed to provide a well-rounded battery for the testing of drug-resistant pneumococci in the VITEK 2.

In conclusion, the VITEK 2 offers the potential for determination of the antimicrobial susceptibilities of pneumococcal clinical isolates by an automated method that is more rapid than current conventional or commercial test procedures. The ability to test pneumococci in the same instrument that is used for testing of various species of nonfastidious organisms should be a valuable feature for clinical laboratories. The MICs generated by the VITEK 2 method compared favorably in this study (within a single dilution) with the MICs generated by the NCCLS reference broth microdilution method for a majority of the agents tested, and few very major or major interpretive errors were encountered. The final selection of antimicrobial agents for the production VITEK 2 cards awaits further development of procedures for inclusion of additional agents in the VITEK 2 cards.