Evaluation of the AutoMicrobic System for Susceptibility Testing of Aminoglycosides and Gram-Negative Bacilli

JOURNAL OF CLINICAL MICROBIOLOGY, Mar. 1987, p. 546-550 0095-1137/87/030546-05$02.00/0 Copyright C 1987, American Society for Microbiology Vol. 25, No. 3 Evaluation of the AutoMicrobic System for Susceptibility Testing of Aminoglycosides and Gram-Negative Bacilli JANET A. HINDLER* AND DAVID A. BRUCKNER Clinical Microbiology Section, Department of Pathology, University of California at Los Angeles Medical Center, Los Angeles, California 90024 Received 8 September 1986/Accepted 3 December 1986 The AutoMicrobic system (AMS; Vitek Systems, Inc., Hazelwood, Mo.) was compared with a reference broth microdilution MIC method to determine the accuracy and reproducibility of aminoglycoside susceptibility testing of gram-negative bacilli. Stock clinical isolates (n = 176) which demonstrated resistance to at least one aminoglycoside, extended-spectrum penicillin, or broad-spectrum cephalosporin (or a combination) were selected for this study. Isolates with moderate susceptibility to the aminoglycosides were also included. Of these isolates, 116 were either resistant or moderately susceptible to one or more of amikacin, gentamicin, netilmicin, and tobramycin. When AMS MIC results for 704 antimicrobial agent-organism combinations were compared with parallel microdilution MIC results, exact agreement (AMS MIC = reference MIC) rates were: amikacin, 71.6%; gentamicin, 71.6%; netilmicin, 83.0%; and tobramycin, 69.3%. Agreement rates within ±+1 log2 dilution were: amikacin, 96.0%; gentamicin, 93.8%; netilmicin, 97.2%; and tobramycin, 96.0%. When National Committee for Clinical Laboratory Standards criteria were used to qualitatively evaluate performance, the overall agreement rates were: amikacin, 100.0%; gentamicin, 99.4%; netilmicin, 98.9%; and tobramycin, 99.4%. There were only four very major discrepancies, which represented 0.6% of the tests performed, and there were no major discrepancies. The percentages of minor discrepancies were: amikacin, 9.6%; gentamicin, 14.2%; netilmicin, 11.9%; and tobramycin, 10.8%. Of the overall average of 11.6% minor discrepancies, 9.7% occurred even though the AMS MIC was within ±1 log2 dilution of the reference MIC. The intralaboratory reproducibility ranged from 93.3 to 100% for the four drugs examined. With this challenge group of gram-negative bacilli, the AMS generated aminoglycoside MIC results that were comparable to those obtained by a reference broth microdilution method. One of several features of the AutoMicrobic system (AMS; Vitek Systems, Inc., Hazelwood, Mo.) is antimicrobial susceptibility testing. A single Gram-Positive Susceptibility Card and several Gram-Negative Susceptibility (GNS) cards are available for testing group B streptococci, group D streptococci, and staphylococci; and aerobic and facultative anaerobic gram negative-bacilli, respectively. Each card contains a growth control well and up to 29 wells containing antimicrobial agents. The drug concentrations vary and depend on the antimicrobial agent being tested. The difference between the various GNS cards relates to the antimicrobial agents included, with most of the cards containing at least one of the aminoglycosides amikacin, gentamicin, netilmicin, or tobramycin. Results from this automated susceptibility test system are generally available within 6 to 10 h. In addition to generating log2 dilution MIC results, the AMS affords the user the flexibility of generating National Committee for Clinical Laboratory Standards category interpretations or integer MICs. In 1984, Woolfrey et al. (7) reported that AMS MICs for Pseudomonas aeruginosa and aminoglycosides (gentamicin, tobramycin, and amikacin) were higher than those obtained with a reference microdilution method with cationsupplemented (calcium, 5.5 + 0.5 mg/dl, and magnesium, 2.5 ± 0.2 mg/dl) Mueller-Hinton broth (Difco Laboratories, Detroit, Mich.). That study made use of the Gram-Negative General Susceptibility Urinary Card, which incorporated a broth containing a higher concentration of calcium and magnesium than in the broth used in current modifications of the AMS antimicrobial susceptibility cards. * Corresponding author. 546 Other investigators (1, 4, 6) have recently noted satisfactory performance of the AMS for the testing of amikacin, gentamicin, and tobramycin with gram-negative bacilli. Nadler et al. reported that the AMS produced acceptable results for the testing of amikacin, gentamicin, and tobramycin with members of the family Enterobacteriaceae; however, some problems were observed with amikacin and gentamicin for P. aeruginosa (2). Of 25 P. aeruginosa isolates included in that study, 5 were resistant to amikacin and 3 were resistant to gentamicin by the reference method, but these isolates were susceptible to these drugs when tested by the AMS. Some of the studies mentioned above made use of nowobsolete versions of AMS software, and none of them included an evaluation of the performance of the AMS with netilmicin. Our study was done to assess the reliability of the AMS as compared with a reference microdilution MIC method for determining aminoglycoside susceptibility in a select group of gram-negative bacilli. MATERIALS AND METHODS Bacterial isolates. The 176 isolates were stock isolates obtained from patients at the University of California at Los Angeles Medical Center. The isolates were selected based on their resistance to at least one aminoglycoside, extendedspectrum penicillin, broad-spectrum cephalosporin, or combination thereof. Isolates demonstrating moderate susceptibility to aminoglycosides were also included. Of the isolates tested, 41 were P. aeruginosa, 11 were Pseudomonas maltophilia, and 10 were Acinetobacter anitratus. Members of the Enterobacteriaceae tested included 2 Citrobacter

VOL. 25, 1987 AMINOGLYCOSIDE TESTING WITH THE AUTOMICROBIC SYSTEM 547 TABLE 1. Aminoglycoside susceptibility of test isolates as determined by microdilution MIC testing Susceptibility" to: No. of Organism isolates Amikacin Gentamicin Netilmicin Tobramycin Acinetobacter anitratus 10 5, 5, 0 0, 0, 10 1, 2, 7 4, 6, 0 Citrobacter di'ersus 2 2, 0, 0 1, 0, 1 2, 0, 0 1, 1, 0 Citrobacterfreundii 18 18, 0, 0 11, 2, 5 13, 2, 3 13, 1, 4 Enterobacter aerogenes 7 7, 0, 0 7, 0, 0 7, 0, 0 7, 0, 0 Enterobacter cloacae 23 23, 0, 0 18, 0, 5 22, 0, 1 18, 0, 5 Esc-herichia coli 23 22, 1, 0 10, 4, 9 21, 0, 2 10, 4, 9 Klebsiella pneumoniae 15 15, 0, 0 2, 6, 7 8, 4, 3 2, 4, 9 Morganella morganii 6 6, 0, 0 4, 0, 2 6, 0, 0 4, 0, 2 Proteus mirabilis 4 3, 1, 0 1, 1, 2 3, 1, 0 1, 1, 2 Prov'idencia rettgeri 2 2, 0, 0 0, 0, 2 1, 0, 1 0, 0, 2 Providencia stuartii 3 3, 0, 0 0, 1, 2 1, 1, 1 0, 1, 2 Pseudomnonas aeruginosa 41 27, 8, 6 11, 7, 23 5, 13, 23 24, 3, 14 Pseudomonas maltophilia il 2, 4, 5 0, 6, 5 1, 5, 5 2, 4, 5 Serratia marcescens il 10, 1, 0 8, 0, 3 9, 0, 2 6, 1, 4 a Results are expressed as numbers of isolates susceptible. moderately susceptible, and resistant to each drug. diversus, 18 Citrobacterfreundii, 7 Enterobacter aerogenes, 23 Enterobacter cloacae, 23 Escherichia coli, 15 Klebsiella pneumoniae, 6 Morganella morganii, 4 Proteus mirabilis, 2 Providencia rettgeri, 3 Providencia stuartii, and 11 Serratia marcescens isolates. No duplicate isolates of a given species from the same patient were tested. All isolates were stored at -70 C in brucella broth-15% glycerol and were subcultured on Trypticase soy agar containing 5% sheep blood (BBL Microbiology Systems, Cockeysville, Md.) before testing. The AMS and reference MIC tests were performed simultaneously with isolated colonies from the same 18- to 24-h subculture plate. Microdilution MIC testing. Reference (REF) MICs were determined according to the procedure described by the National Committee for Clinical Laboratory Standards (3). MIC trays were prepared in-house with a Quick Spense dispenser (Bellco Glass, Inc., Vineland, N.J.), dispensing 0.1-ml volumes of each antimicrobial agent dilution per well. Mueller-Hinton broth (Difco) supplemented with 50 mg of calcium and 25 mg of magnesium per liter was used. Antimicrobial reference standard powders were kindly provided by Bristol Laboratories, Syracuse, N.Y. (amikacin), Schering Corp., Kenilworth, N.J. (gentamicin and netilmicin), and Eli Lilly & Co., Indianapolis, Ind. (tobramycin). The ranges of log2 concentrations tested were 0.5 to 32.0 Fig/ml for amikacin and 0.5 to 16.0,utg/ml for netilmicin. Gentamicin and tobramycin were tested at 0.5, 1.0, 2.0, 4.0, 6.0, 8.0, and 10.0 ptg/ml; however, for comparative purposes, REF microdilution results of 6.0,ug/ml were considered at 8.0 p.g/ml and results of 10.0,ug/ml were considered at >8.0,ug/ml to correspond to the log2 dilution schema reported with the AMS. MIC trays were inoculated by the direct inoculum method by suspending fresh colonies in Mueller-Hinton broth and adjusting the turbidity to match that of a McFarland no. 3 turbidity standard. An MIC 2000 inoculator (Dynatech Laboratories, Inc., Alexandria, Va.) was used, and the final concentration of organisms was 5 x 105 CFU/ml. After 16 to 18 h of incubation at 35 C, MIC endpoints were read manually as the lowest concentration of antimicrobial agent that completely inhibited growth as determined visually. Extensive quality control was done at the time of tray preparation and was done thereafter by testing E. coli ATCC 25922 and P. aeruginosa ATCC 27853 on a daily basis. AMS. The AMS GNS-C card was used for testing amikacin, gentamicin, and tobramycin, and the GNS-R card was used for netilmicin. The ranges of log2 dilution MICs (,ug/ml) reported are: amikacin, s2.0 to >32.0; gentamicin and tobramycin, c0.5 to >8.0; and netilmicin, <4.0 to >16.0. Tests were done according to the instructions of the manufacturer, and an AMS Gram Negative Identification Card was run simultaneously with all GNS cards. Quality control was done according to the instructions of the manufacturer. MIC results and category interpretations were automatically printed at the end of the incubation period, which was 6 to 10 h for most isolates. Software version R2.04 (Vitek) was used. Evaluation of results. Each AMS log2 MIC result was compared with the corresponding REF MIC result. For quantitative comparisons, exact agreement occurred when the AMS MIC was identical to the REF MIC. An AMS MIC within + 1 log2 dilution of the REF MIC was considered to be in agreement. An AMS MIC greater than ±+1 log2 dilution from the REF MIC was considered to be in disagreement. Current National Committee for Clinical Laboratory Standards category interpretations of MICs (jig/ml) for suscepti- TABLE 2. Comparison of AMS and REF MICs for all isolates" Antimicrobial No. of isolates with AMS/REF MIC ratios of: Exact % Overall % agent <0.12 0.25 0.5 ib 2-4 agreement" agreement" Amikacin 7 20 126 23 71.6 96.0 Gentamicin 1 9 29 126 10 1 71.6 93.8 Netilmicin 2 2 14 146 il 1 83.0 97.2 Tobramycin 7 17 122 30 69.3 96.0 A total of 176 isolates were tested, including 41 P. aeruginosa, 111 Enterobacteriaceae. and 24 additional isolates. b AMS MIC identical to REF MIC. AMS MIC/REF MIC ratio between 0.5 and 2 (AMS MIC within ±1 10g2 dilution of the REF MIC).

548 HINDLER AND BRUCKNER J. CLIN. MICROBIOL. TABLE 3. Comparison of AMS and REF MICs for P. aeiruginosa" Antimicrobial No. of isolates with AMS/REF MIC ratios of: Exact % Overall % agent -0.12 0.25 0.5 1" 2 24 agreement" agreement" Amikacin 8 23 10 56.0 100 Gentamicin 1 9 27 3 1 65.9 95.1 Netilmicin 2 10 24 5 58.5 95.1 Tobramycin 1 7 30 3 73.2 97.6 A total of 41 P. aerugilnosa isolates were tested. '>AMS MIC identical to REF MIC. See Table 2, footnote c. ble, moderately susceptible, and resistant results are, respectively, -<16, 32, and >32 for amikacin, -<4, 8, and >8 for gentamicin and tobramycin, and -4, 8 to 16, and >16 for netilmicin (3). By these criteria for qualitative comparisons, the AMS and REF methods were considered to be in agreement when the category results were identical. Category results that differed were counted as discrepancies. Very major discrepancies included AMS-susceptible and REF-resistant results, and major discrepancies included AMS-resistant and REF-susceptible results. Minor discrepancies included susceptible or resistant category results by one method, with moderately susceptible results by the other. Both AMS MIC and REF MIC tests were repeated when disagreements occurred; however, the initial AMS MIC result was used for all comparisons. The initial REF MIC result was used when the repeat was within +±1 log1 dilution, and the organism was eliminated from the study if the REF MIC result did not reproduce within +1 log2 dilution. Reproducibility studies. To determine the reproducibility of either test system, 30 isolates were tested by each method on 3 consecutive days. The isolates examined were selected to represent various species and various degrees of susceptibility to all four aminoglycosides. Two isolates (a mucoid P. aeruginosa and a clumpy S. marcescens) were tested because they showed peculiar growth characteristics on agar media. RESULTS The numbers of isolates of each species that were susceptible, moderately susceptible, and resistant to amikacin, gentamicin, netilmicin, and tobramycin are indicated in Table 1. A comparison of AMS MICs with REF MICs for all isolates tested is shown in Table 2. Results for the 41 P. aeruginosa isolates only are shown in Table 3. When all isolates were examined, the percentages of tests for which the AMS MIC was identical to the REF MIC were: amikacin, 71.6%; gentamicin, 71.6%; netilmicin, 83.0%; and tobramycin, 69.3%. For P. aceruginosa only, the percentages were: amikacin, 56.0%; gentamicin, 65.9%; netilmicin, 58.5%; and tobramycin, 73.2%. The overall agreement, in which the AMS MIC was within 1 log2 dilution of the REF MIC for all isolates as well as for the subset of P. aeruginosa, was above 93% for the four aminoglycosides. A listing of isolates that demonstrated AMS MICs that disagreed with REF MICs is shown in Table 4. There were 30 disagreements, representing 22 of the 176 isolates. Disagreement for 28 of the 30 occurrences involved a lower AMS MIC than REF MIC. Most of the disagreements involved gentamicin, and these were most common with K. pneumoniae. Five of the seven disagreements for amikacin involved a lower AMS MIC than REF MIC for E. coli. A comparison of AMS and REF system interpretive category results is shown in Table 5 for all isolates tested and in Table 6 for P. aeruginosa only. There were no major discrepancy errors and 0.6% very major discrepancies among the 704 antimicrobial agent-organism combinations examined. One E. coli and two P. aeruginosa isolates were involved in the four very major discrepancies which persisted after repeat testing. The discrepancy with E. coli was with tobramycin. AMS MIC results from four repeat tests were susceptible at 4, 2, 2, and 2 fig/ml. Repeat reference MIC results were >10, 10, 8, and >10,ug/ml. One of the P. aeruginosa isolates demonstrated AMS results of 2, 8, 4, and 8 p.g/ml with netilmicin upon repeat testing. Four repeat reference tests showed MICs of >16 kg/ml. The other P. aeruginosa isolate was discrepant with gentamicin and netilmicin. AMS results for gentamicin from four repeat tests were susceptible at s4 1tg/ml, and reference results indicated resistance at -10.tg/ml. All four netilmicin results Organism TABLE 4. il Isolates demonstrating AMS MICs that disagreed with REF MICs" Amikacin Gentamicin Netilmicin Tobramycin AMS-H' AMS-L' AMS-H AMS-L AMS-H AMS-L AMS-H AMS-L Acinetobaccter 3 (10) 2 2 2 arniratus Citiroba cter freuindii 2 (18) 1 1 Esc herichia coli 7 (23) 5 2 Klebsiella pneuimoniae 6 (15) 6 2 Proteus mirabilis 1 (4) 1 Pseudoinonas 3 (41) 1 2 1 aeruginosa AMS MIC greater than ±1 1og2 dilution from REF MIC (six isolates demonstrated disagreement with more than one drug). Number of isolates that demonstrated disagreement. The total number of isolates of respective species tested during this evaluation is shown in parentheses. <H. For each drug, number of isolates for which AMS MIC was higher than REF MIC. "L, For each drug, number of isolates for which AMS MIC was lower than REF MIC.

VOL. 25, 1987 AMINOGLYCOSIDE TESTING WITH THE AUTOMICROBIC SYSTEM 549 TABLE 5. Comparison of interpretive category results for AMS and REF systems for all isolatesa No. (%) of discrepancies No. (%) of isolates in: Antimicrobial agent agentvery Major Minor Agreement' agreement ~~~~~~~~~~Overaîl major areet Amikacin 0 0 17 (9.6) 159 (90.3) 176 (100) Gentamicin 1 (0.6) 0 25 (14.2) 150 (85.2) 175 (99.4) Netilmicin 2 (1.1) 0 21 (11.9) 153 (86.9) 174 (98.9) Tobramycin 1 (0.6) 0 19 (10.8) 156 (88.6) 175 (99.4) a A total of 176 isolates were tested including 41 P. aeruginosa, 111 Enterobacteriaceae, and 24 additional isolates. b Very major discrepancies include AMS-susceptible and REF-resistant results; major discrepancies include AMS-resistant and REF-susceptible results; minor discrepancies include moderately susceptible results for one system with susceptible or resistant results for the other. d c Category results identical. Minor errors considered in agreement. with the AMS were c4,ug/ml, and those with the reference test results were -16,ig/ml. Overall, there were 11.6% minor errors: amikacin, 9.6%; gentamicin, 14.2%; netilmicin, 11.9%; and tobramycin, 10.8%. Of these errors, most (amikacin, 9.0%; gentamicin, 9.6%; netilmicin, 11.4%; and tobramycin, 8.5%) occurred even though the AMS MIC was within ± 1 log2 dilution of the REF MIC. For P. aeruginosa, all minor errors with amikacin, gentamicin, and tobramycin involved 1 log2 dilution difference between the AMS MIC and the REF MIC. For every minor error for netilmicin, the REF MIC was 8.0,ug/ml and the AMS MIC was c4.0,ug/ml. All four aminoglycosides demonstrated comparable reproducibility in either test system. For the 120 antimicrobial agent-organism combinations (30 isolates tested), 80% of the AMS MICs and 84.2% of the REF MICs were identical on the 3 consecutive days. More than 97% of results reproduced within 1 log2 dilution for either system. DISCUSSION The AMS generally provides susceptibility results within 6 to 10 h. This offers significant advantages over conventional susceptibility test systems, which require overnight incubation. It is well known that the results of antimicrobial susceptibility testing of aminoglycosides and P. aeruginosa are influenced by the test medium. An earlier study indicated that the AMS did not perform adequately as compared with a reference system (7). At the time of Woolfrey's study, the divalent cation concentration of the base medium in the AMS susceptibility cards varied from that generally recommended. However, since then, Vitek has modified the base medium and, as shown in the present study, with the TABLE 6. Comparison of interpretive category results for AMS and REF systems for P. aeruginosaa No. (%) of discrepancies No. (%) of isolates in: Antimicrobial agent Very Major Minor Agreement' agreement major areet Amikacin 0 0 7 (17.1) 34 (82.9) 41 (100) Gentamicin 1 (2.4) 0 7 (17.1) 33 (80.5) 40 (97.6) Netilmicin 2 (4.9) 0 11 (26.8) 28 (68.3) 39 (95.1) Tobramycin 0 0 3 (7.3) 38 (92.7) 41 (100) a A total of 41 P. aeruginosa isolates were tested. b See Table 5, footnote b. Category results identical. d Minor errors considered in agreement. exception of two isolates, results obtained with the aminoglycosides and P. aeruginosa are comparable to those obtained with a reference method. The isolates examined in the present study included isolates that were more resistant than those generally encountered. These are representative of problem isolates that may be associated with nosocomial infections and often present challenges for in vitro testing and challenges for effective therapy. This study showed that MIC results obtained from testing the aminoglycosides with the AMS are comparable to those obtained by use of a reference method when testing such a select group of organisms. However, it should be noted that when discrepancies occur, the AMS MIC is generally lower than the REF MIC. Category interpretation of MICs demonstrated no major discrepancies and a very low percentage of very major discrepancies. Two of the three isolates showing very major discrepancies were P. aeruginosa. In addition to the 41 P. aeruginosa isolates documented in this study, there were three isolates that did not grow in the AMS cards but did grow in the REF MIC system. It is conceivable that this false susceptibility is due to inadequate growth, and a modification of the growth threshold for P. aeruginosa in the AMS susceptibility cards may eliminate such problems. For the 704 antimicrobial agent-organism combinations, there were 11.6% minor errors. It has been suggested that for testing of random clinical isolates, a minor discrepancy error percentage of no greater than 10% has been desirable (5). In the present study, isolates were selected with diverse susceptibilities to the four aminoglycosides evaluated in order to challenge the AMS with a variety of susceptibility endpoints. This selection bias may have contributed to the high percentage of minor errors noted. Upon close examination, of the 11.6% minor errors, 9.7% occurred even though the AMS MIC was within ±+1 log2 dilution of the reference MIC. Since it is generally agreed that in many situations in which aminoglycosides might be administered, MIC results offer significantly more useful susceptibility information as compared with qualitative results, the overall clinical significance of this incidence of minor errors is unknown. Concerning the current status of aminoglycoside testing with the AMS, it should be noted that the section of the GNS software package used in this study for analysis of amikacin, gentamicin, netilmicin, and tobramycin was introduced in 1984 with software version P14XOA and has been included without modification in all software versions to date. Similarly, the medium as described herein for testing amikacin, gentamicin, netilmicin, and tobramycin was incorporated into the GNS cards in 1983 and has not been modified since. Both the AMS and REF MIC results were reproducible. In light of these findings, the AMS generated aminoglycoside MIC results that were comparable to those obtained by a REF microdilution method when testing a challenge group of gram-negative bacilli. LITERATURE CITED 1. Backes, B. A., S. J. Cavalieri, J. T. Rudrik, and E. M. Britt. 1984. Rapid antimicrobial susceptibility testing of gram-negative clinical isolates with the AutoMicrobic system. J. Clin. Microbiol. 19:744-747. 2. Nadler, H. L., C. Dolan, L. Mele, and S. R. Kurtz. 1985. Accuracy and reproducibility of the AutoMicrobic system Gram- Negative General Susceptibility-Plus Card for testing selected challenge organisms. J. Clin. Microbiol. 22:355-360. 3. National Committee for Clinical Laboratory Standards. 1985. Methods for dilution antimicrobial susceptibility tests for

550 HINDLER AND BRUCKNER bacteria that grow aerobically. Approved Standard M7-A. National Committee for Clinical Laboratory Standards, Villanova, Pa. 4. Rçiber, N. E., M. T. Kelly, J. M. Latimer, D. L. Tison, and R. M. Hysmith. 1985. Comparison of the Cathra Repliscan II, the AutoMicrobic system Gram-Negative General Susceptibility- Plus Card, and the Micro-Media System Fox panel for dilution susceptibility testing of gram-negative bacilli. J. Clin. Microbiol. 21:959-962. 5. Thornsberry, C. 1985. Automated procedures for antimicrobial susceptibility tests, p. 1015-1018. In E. H. Lennette, A. Balows, J. CLIN. MICROBIOL. W. J. Hausler, Jr., and H. J. Shadomy (ed.), Manual of clinical microbiology, 4th ed. American Society for Microbiology, Washington, D.C. 6. Van Horn, K. G., A. M. Vandernoot, E. W. Burke, and J. C. McKitrick. 1984. Evaluation of the AutoMicrobic system Gram- Negative General Susceptibility-Plus Card. J. Clin. Microbiol. 20:630-635. 7. Woolfrey, B. F., R. T. Lally, M. N. Ederer, and C. O. Quall. 1984. Evaluation of the AutoMicrobic system for susceptibility testing of Pseudomonas aeruginosa to gentamicin, tobramycin, and amikacin. J. Clin. Microbiol. 19:502-505.