INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

With the emergence of multidrug-resistant organisms, physicians are now faced with the challenge of treating infections that have no established therapeutic guidelines (2). Under these circumstances, they may request that the clinical laboratory attempt to assess the adequacy of therapy (5, 9, 23). Bactericidal activity testing has been used as an occasional guide for antimicrobial agent treatment since the 1950s. Such testing is most frequently performed when bactericidal antimicrobial agent therapy is considered necessary (23). The two most commonly used of these special microbiology tests to monitor potentially nonstandard therapy are those that determine the minimum bactericidal concentration (MBC) of a drug and the serum bactericidal titer (SBT) of a patient's blood or body fluid during treatment (23). Reference guidelines from the National Committee for Clinical Laboratory Standards (NCCLS) are now approved for performing both of these assays (16, 17). However, several important variables affect the reproducibility of the test results (23), and a recent assessment by MacGowan and coworkers again highlighted continued technical problems (12). Since neither the new NCCLS documents nor any recent published information comments on how various methodologies actually perform during clinical laboratory use, reproducibility for both the MBC and SBT assays still needs to be addressed (11, 13).

Between April 1992 and November 1997, the microbiology laboratory at Northwestern Memorial Hospital used a procedure that combines the NCCLS method(s) with that described by Peterson and Shanholtzer (16, 17, 23). The purpose of this report is to summarize our experience with this modified methodology by using results from the clinical laboratory tests.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

The MBC and SBT tests are performed by using similar methodologies, with minor exceptions, as described below.

Test strategy. To prepare for testing, it is necessary to determine the 6 dilutions that will be run in duplicate for the MBC test. The MIC was first determined by NCCLS reference agar dilution methodology (15), and then 4 dilutions above the MIC, the MIC, and 1 dilution below the MIC were used for the levels to be assessed in the MBC assay. Running 4 dilutions above the known MIC permits detection of tolerance to normally bactericidal agents (23). For SBT testing, the 6 twofold dilutions are constant: 1:2 through 1:64 for both peak and trough levels. These dilutions were considered sufficient based upon the NCCLS reference method for SBT testing, which defines titer determinations up to 1:32 for interpretation of test results (16). A positive and negative growth control tube is included with each test. Quality control is also performed with a known reference organism each time a fresh antimicrobial agent stock solution is prepared for MBC testing.

Dilution preparation. Antimicrobial agents used for MBC testing are obtained from the drug's manufacturer as reference powder or solution. Stock solutions at 1,000 µg/ml are freshly prepared the day of use or are prepared from a frozen stock solution stored at 70°C. Serial dilutions for both test methods are prepared in borosilicate glass test tubes using cation-adjusted Mueller-Hinton broth (CAMHB; Difco Laboratories, Detroit, Mich.). For the MBC test, the duplicate set of tubes is prepared by filling each tube with 1 ml of CAMHB, beginning with tube 2. Next, 1 ml of CAMHB containing twice the desired highest drug concentration is added to tubes 1 and 2. Tube 2 is then vortexed, and 1 ml of the mixture is removed and added to tube 3. Tube 3 is vortexed, and 1 ml of this mixture is removed with a new pipette and transferred to tube 4. This process is continued until the last tube is reached. One ml is discarded from the last tube. To bring the tubes to a final volume of 2 ml, 1 ml of CAMHB is added to each tube. The growth and sterility controls are prepared by adding 2 ml of CAMHB to each of two new test tubes.

Inoculum and incubation. Four to five 18- to 24-h colonies are inoculated into 3 ml of Trypticase soy broth. The broth is incubated for 3 to 5 h to achieve a turbid suspension. The inoculum is prepared by adjusting the logarithmic-phase growth to match the turbidity of a 0.5 McFarland standard, yielding approximately 10⁸ CFU/ml. The suspension is then diluted 1:10 with CAMHB to give a working inoculum of 10⁷ CFU/ml. Using a calibrated pipette, 0.1 ml of the working inoculum is carefully added to each tube, except the sterility control, just below the broth surface to produce a final density of approximately 5 × 10⁵ CFU/ml. Mixing is performed in the tube by gently flushing the inoculum in and out of the pipette tip four or five times, avoiding splashing or creation of bubbles. All tubes are incubated at 35°C for 24 h without shaking or agitation. A colony count is performed on the starting test inoculum by lawning 0.01 ml of a 100-fold dilution of the growth control tube contents onto a drug-free blood agar plate to determine if the estimated organism density for the test is within the actual desired limits.

Determining endpoints. At the end of 24 h of incubation, the tubes are read for the MIC or serum inhibitory titer, defined as the concentration of the first tube in the series (ascending drug concentration or descending serum dilution titer) to show no visible trace of growth. The MBC or SBT is then determined by sampling all the macroscopically clear tubes and the first turbid tube in the series. Before being sampled, the tubes are gently mixed by flushing them with a pipettor, and a 100-µl aliquot is removed. Each aliquot is placed on a single antibiotic-free agar plate suitable for the growth of the microbe being tested, in a single streak down the center of the plate (24). The sample is allowed to be absorbed into the agar until the plate surface appears dry (about 30 min). The aliquot is then spread over the plate by using a lawning technique. This subculture approach has been used in numerous studies of both gram-positive and gram-negative bacteria, where it was found satisfactory in eliminating the problem of antimicrobial agent carryover from the 100-µl subculture volume (3, 6, 7, 14, 19, 21, 24). The growth and sterility controls are sampled in the same manner. The lawned plates are then incubated for a full 24 h at 35°C. The original tubes are reincubated for another 24 h, in case an additional subculture is needed at the 48-h point to resolve discrepancies (i.e., "skipping" or endpoint differences of >1 dilution).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

For this report, we performed tests of cell wall-active agents (-lactams or vancomycin) against 117 gram-positive cocci and 14 gram-negative bacilli. The combinations of organism-antimicrobial agent tested are shown in Table 1. Aminoglycosides, fluoroquinolones, erythromycin, clindamycin, trimethoprim-sulfamethoxazole, chloramphenicol, or rifampin were tested against 37 and 11 of the respective organism groups. Thirty organisms (25 gram-positive cocci and 5 gram-negative bacilli) were tested against a combination of agents from those listed. All together, this comprised 93% of our reported assays. The other 7% was made up of tests with other antimicrobial agents. By defining reproducibility as agreement within 1 dilution between duplicate tests, we found that 92% of the 224 tests sampled after 24 h were reproducible (Table 2). A summary of the discrepant tests is in Table 3. Of the 224 tests performed, 17 (8%) required resampling after 48 h of incubation. In 16 of the 17 tests (94%), the endpoint discrepancy or skipping problem resolved. Overall, only 1 of the 224 tests (0.4%) was discrepant after 48 h of incubation and would have required repeat testing. This single failure of the MBC assay for gentamicin against Pseudomonas aeruginosa was not repeated, and the result of this MBC test was thus uninterpretable. The majority of the tests with initially discrepant results (14 of 17; 82%) were in the group of gram-positive cocci whose susceptibility to cell wall-active agents was investigated. This group comprised 52% of the total "bug-drug" combination testing done. Interestingly, there was no need to extend incubation of any of the assays done on gram-positive cocci with other agents, a group comprising 17% of the studies. Since only a single test was uninterpretable after a 48-h MBC assay was performed (gentamicin tested against P. aeruginosa; total P. aeruginosa assays, 14), there was no discernible problematic antimicrobial agent-bacterium combination. Overall, using the described methodology, our laboratory demonstrated a reproducibility rate of over 92% on the first performance of a bactericidal test.

The mean cost of this testing to our laboratory, based on 1995 charges for supplies, was approximately $25.00, including 47 min of technologist time for the entire duration of either bactericidal assay procedure.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The results of testing by using the suggested bactericidal test method in the Northwestern Memorial Hospital clinical microbiology laboratory showed 99.6% reproducibility when applied to 224 tests. The modification of the NCCLS procedures consists of deletion of serum additives, selecting four to five colonies to prepare the inoculum, avoiding agitation during incubation, performing the test in duplicate with six dilutions, using a larger subculture volume from the incubated tubes, reincubating for 24 h to resolve discrepancies, and using an alternate statistical endpoint determination.

Enhanced clinical relevance has not been shown for serum additives in tests of virtually all currently used antimicrobial agents; therefore, they are not recommended in order to reduce the risk of transmission of blood-borne pathogens. We recognize that the influence of protein on the in vitro activity of antibiotics can be significant (22) and that protein binding affects interpretation of serum kinetics and extravascular drug penetration (18). However, many factors affect the in vivo binding of drugs to serum proteins (25). Seriously ill patients, the group most likely to have the testing described in this report performed, are very likely to have abnormal serum protein binding in vivo. Chen and colleagues found that serum albumin additives from commercial sources that may be used for MBC and SBT testing have unpredictable factors that affect binding of cell wall-active agents (4). We have also shown that filter membranes for preparation of serum ultrafiltrates can trap antibiotics (20), and this phenomenon will markedly alter sample drug content when only a small amount of serum is available for preparing an ultrafiltrate devoid of protein. Therefore, combining all this inherent unpredictability when trying to correct for serum protein content with the risk to healthcare workers when dealing with a human body fluid, we recommend avoiding the addition of serum or removal of serum proteins by ultrafiltration as a correction factor. Focusing on the enhancement of assay reproducibility is an approach that permits long-term assessment of the utility of bactericidal evaluations and thereby facilitates reliable interpretation of results obtained from these potentially useful tests.

We recommend that four to five colonies be used to prepare the inoculum. This provides sufficient diversity of the test organism so that a unique "clonal type" is not inadvertently selected and avoids the possibility of contamination that may occur when a larger number of colonies are used. Agitation of the tubes is not necessary prior to sampling due to the 0.1-ml size of the initial inoculum. Taylor et al. evaluated the effect of inoculum volume on the outcome of macrodilution tests (26). They found that gentle agitation at 20 h was needed only when a large volume of inoculum (1.0 ml) was used; however, it was not useful when the 0.1-ml inoculum volume we suggest was adopted (26). Gresser-Burns and colleagues (8) and Ishida et al. (10) have also found it preferable to use a smaller inoculum (0.1 ml) and not to disturb the sample until the time of the actual subculture.

Testing in duplicate is recommended, since bactericidal activity tests historically have shown poor reproducibility. By reincubating the tubes after the first 24 h of incubation, one can attempt to resolve discrepancies (such as the skipping phenomenon or endpoint discrepancies) between duplicates without needing to repeat the entire test. Cell wall-active agents like -lactams and vancomycin kill slowly, a phenomenon discussed in the new NCCLS documents (16, 17). Therefore, it is not unexpected that occasional isolates will not have fully responded to the killing activity of the drug being tested at 24 h, thus leading to a discrepancy in duplicate test results if sampling is done at a time when colony numbers are undergoing a rapid decline. The additional 24 h, then, allows for enhanced reproducibility of an assessment of the agent's potential bactericidal activity when its bactericidal action is complete and the number of viable colonies is more stable.

Sampling a larger volume (0.1 ml) from the macrodilution tubes and using the statistical correction formula (n + 2n) for bactericidal activity endpoint determination provides a simple method for determining 99.9% killing that includes 95% confidence limits. The larger sampling volume, 0.1 ml instead of 10 µl, provides a greater number of colonies to be counted (10 to 100 versus 1 to 10), theoretically enhancing test precision, particularly if the initial test inoculum approaches the lower recommended test limit of 10⁵ CFU/ml.

The method described for determining bactericidal activity has improved reproducibility and, presumably, reliability of this test without increasing the workload for our laboratory. This assessment of our current laboratory practice builds upon work we had previously published on bactericidal testing techniques. The report by Shanholtzer and colleagues had shown that macrodilution bactericidal tests were not reproducible when simultaneously performed by two workers or if performed multiple times by one worker when using methodology that included the NCCLS currently recommended agitation step at 22 to 24 h and a subculture volume of 0.01 ml (24). The subsequent report by Gresser-Burns et al., using an MBC subculture volume of 0.1 ml, no agitation during incubation, and the n + 2n endpoint calculation, demonstrated resolution of the skipping phenomenon when incubation was extended from 24 to 48 h (8). They found that growth at higher drug concentrations (above a concentration with no growth at 24 h of subculture) disappeared when using the 48-h incubation time point in eight of nine tests (8). Several different medical technologists have performed our current technique over an 8-year period in the clinical laboratory with no change in outcome reproducibility. The cost of $25.00 per test is not inexpensive, but it is modest when one realizes that the total hospitalization expense per day for a seriously ill patient is now approximately $2,000. This modified method can be easily performed in any clinical microbiology laboratory and provides a high level of test reproducibility.