Precision of Vancomycin and Daptomycin MICs for Methicillin-Resistant Staphylococcus aureus and Effect of Subculture and Storage

Abstract

The reproducibility of vancomycin and daptomycin MICs, measured by broth microdilution (BMD) and Etest, was prospectively assessed for 10 methicillin-resistant Staphylococcus aureus (MRSA) isolates from the blood samples from patients on vancomycin therapy. The isolates were tested at the time of isolation from blood and following 5, 10, and 20 subcultures and at 1, 3, 6, and 12 months of storage at −70°C. The MICs were determined by Etest and BMD using two different manufacturers (BBL and Difco) of cation-adjusted Mueller-Hinton broth (CA-MHB), and using three different drug powders: vancomycin from Sigma, vancomycin from Novation, and daptomycin from Cubist. The antimicrobial concentrations tested were 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 μg/ml. Two isolates were vancomycin intermediate and daptomycin nonsusceptible, and two isolates had reduced susceptibility to vancomycin (BMD MIC, 1.5 or 2.0 μg/ml). The vancomycin MICs were significantly higher in the BBL CA-MHB than those in the Difco CA-MHB, and with Sigma versus Novation vancomycin powder. The daptomycin MICs were also significantly higher in the BBL CA-MHB. The Etest MICs were significantly higher than those obtained by BMD for vancomycin but not for daptomycin. The average precision of the vancomycin BMD MICs when analyzing 20 results was ±1.10-fold log₂ dilutions, and it was ±1.67-fold for daptomycin (10 results). The average precision for Etest was ±1.11-fold for vancomycin and ±1.16-fold for daptomycin. No significant change in MICs was noted following 5, 10, or 20 subcultures or at up to 6 months of frozen storage. However, the vancomycin MICs alone were significantly lower (0.74-fold) following 12 months of frozen storage. From these data, despite variations in CA-MHB and antimicrobial powder, the MIC result precision was <0.5 log₂ dilutions in a single laboratory, suggesting that testing interdilution MICs (e.g., MICs between serial 2-fold dilutions) is a possibility. A more accurate method for measuring vancomycin MIC results is thus possible, but further standardization of BMD testing would be required to achieve this goal.

INTRODUCTION

Vancomycin is extensively used for the treatment of serious infections caused by methicillin-resistant Staphylococcus aureus (MRSA), such as bacteremia. To date, MRSA isolates that are intermediate or resistant to vancomycin (i.e., MIC, >-2 μg/ml) are rarely encountered in clinical practice. However, MRSA isolates with vancomycin MICs in the upper end of the susceptible range (i.e., MICs, 1.5 to 2.0 μg/ml) are more common, in particular among patients with prior or concurrent vancomycin therapy. Vancomycin MICs of 1.5 and 2.0 μg/ml have been associated with poor clinical outcomes of vancomycin therapy in a number of studies (1–7). However, nearly all of these studies performed vancomycin MICs only after multiple subcultures and/or cryopreservation of S. aureus isolates, as well as by a variety of MIC test methods (8). Studies have shown that storing isolates at −80°C for >9 months can result in a decrease in vancomycin MIC (9, 10). It is also well appreciated that the various in vitro test methods used by clinical laboratories yield significantly different vancomycin MICs for the same isolate (11–13). Together, these findings raise concern about the validity of using supposed reduced susceptibility to vancomycin (i.e., MIC, 1.5 or 2.0 μg/ml) as a criterion to prompt the use of an alternative agent, such as daptomycin, for the treatment of MRSA infection, as has been suggested by some (14).

Additionally, the limits of the MIC test have not been fully evaluated to date. The two current reference methods for MIC testing, broth microdilution (BMD) and agar dilution, yield MICs based on serial 2-fold doubling dilutions of antimicrobials, whereas methods, such as Etest, provide additional MICs that fall between these traditional concentrations. The reproducibility of such “interdilution” MICs is unknown for BMD, although conventional wisdom maintains that the precision of BMD is ±1 log₂ dilution. However, whether a tighter precision of the MIC results is achievable by BMD is unknown, as laboratories do not routinely test these in-between antimicrobial concentrations. Regardless, in order for an interdilution MIC to be a valid criterion by which to make treatment decisions, the reproducibility of such a result must be clarified.

To help resolve these concerns, we prospectively evaluated the effect of serial subculture and cryopreservation of isolates at −70°C on vancomycin and daptomycin MICs obtained by broth microdilution (BMD) and Etest for a collection of 10 MRSA isolates from blood samples of patients on vancomycin therapy. The robustness of the BMD test was evaluated by generating panels with vancomycin and daptomycin MICs between the traditional log₂ scale. The effects of modifying two variables, the vancomycin powder source and Mueller-Hinton broth manufacturer, on the vancomycin and daptomycin MICs were also investigated.

MATERIALS AND METHODS

Bacterial isolates.

Ten MRSA strains isolated from blood samples of unique patients were evaluated in this study. Blood was cultured in a BacT/Alert (bioMérieux, Durham, NC) instrument in standard blood culture bottles. The isolates were entered into the study only if the patient had been treated with vancomycin for ≥3 days (range, 3 to 42 therapy days) prior to blood collection in an effort to enrich the study with isolates with elevated vancomycin MICs. For each isolate, 3 to 4 colonies from the primary subculture of the positive blood culture broth on sheep's blood agar (BAP) (BD, Sparks, MD) were cryopreserved at −70°C in brucella broth plus 15% glycerol (BD). All protocols were approved by the University of California, Los Angeles (UCLA) institutional review board.

MIC test methods.

A CLSI reference BMD (15, 16) was performed using panels prepared in-house. Cation-adjusted Mueller-Hinton broth (CA-MHB) from two manufacturers (BD, Franklin Lakes, NJ, and Difco, Detroit, MI) was used to make the panels. Each CA-MHB was supplemented to 50 mg/liter calcium for daptomycin testing, and the calcium concentrations were confirmed by high-performance liquid chromatography (HPLC). Vancomycin from two manufacturers, Sigma (St. Louis, MO) and Novation (Burnaby, British Columbia, Canada), and daptomycin from a single source, Cubist (Lexington, MA), were used to prepare antimicrobial solutions at concentrations of 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 μg/ml in each of the two CA-MHBs. Both vancomycin powders had a stated potency of ≥900 μg of vancomycin per mg. The BMD results were read visually after 16 to 20 h at 35°C in ambient air for daptomycin and after 24 h of incubation for vancomycin.

Etest was performed using a 0.5 McFarland standard inoculum streaked on Mueller-Hinton agar (BD), followed by the application of vancomycin and daptomycin Etest strips (bioMérieux). The plates were read visually using a magnifying glass at 16 to 20 h for daptomycin and at 24 h for vancomycin.

MIC testing schedule.

The initial vancomycin and daptomycin MICs were determined following ≤2 subcultures from the positive blood culture broth. The isolates were then serially subcultured for 20 consecutive days on 5% sheep's blood agar, and MIC testing was repeated following subcultures 5, 10, and 20. The isolates were recovered from frozen stocks after 1, 3, 6, and 12 months of storage. For the frozen isolates, the MICs were determined following two successive subcultures. At each of the 8 time points, BMD using all 6 CA-MHB-antimicrobial combinations and Etest were performed in parallel, yielding 40 vancomycin MIC results and 24 daptomycin MIC results for each of the 10 isolates.

Five isolates (1, 4, 5, 6, and 7) were chosen at random and tested in quintuplicate at the first time point by the 6 BMD CA-MHB-antimicrobial combinations and Etest. These data were used to assess the intra-assay precision (i.e., within-run precision) of the MICs obtained on the first day of testing.

On each testing day, the CLSI-recommended S. aureus ATCC 29213 quality control (QC) strain was also tested; all resultant MICs were within the acceptable QC range.

Data analysis.

All study MIC values were transformed to log₂ values prior to statistical analyses. A reference BMD MIC result was defined for the purpose of this study as the geometric mean of the MICs obtained by BMD (4 formulations for vancomycin or 2 formulations for daptomycin) at the time of isolation. The reference MICs that fell between the dilutions tested were rounded up to the nearest log₂ dilution when evaluating the categorical agreement of the MICs. For example, a vancomycin mean MIC of 1.35 μg/ml was rounded to 2.0 μg/ml, or an MIC of 2.24 μg/ml was rounded up to 4.0 μg/ml. Categorical agreement was calculated for vancomycin and daptomycin MIC interpretations using the CLSI M100-S24 breakpoints, which are ≤2, 4 to 8, and ≥16 μg/ml for vancomycin-susceptible, -intermediate, and -resistant isolates, respectively. For daptomycin, there is a susceptible-only breakpoint of ≤1 μg/ml (16). In addition, the vancomycin MIC results were stratified into low susceptible (MIC, <1.5 μg/ml) versus high susceptible (MIC, 1.5 or 2.0 μg/ml) to determine the effect of MIC reproducibility across the variables assessed in this study on treatment decisions, using a vancomycin MIC of 1.5 or 2.0 μg/ml as an actionable result (14).

All statistical analyses were performed using GraphPad Prism 5.0 and StatsDirect 3.0 (Altrincham, Cheshire, United Kingdom).

RESULTS

MIC results at time of MRSA isolation.

The initial MIC results for the 10 MRSA isolates are shown in Table 1. Isolate 10 was vancomycin intermediate (reference MIC, 3.37 μg/ml) by all four BMD formulations and by Etest (Table 1). This isolate was also daptomycin nonsusceptible by BMD and Etest. Isolate 5 was vancomycin intermediate (reference MIC, 2.24 μg/ml), but an intermediate result was obtained only with the BBL CA-MHB (Table 1). This isolate was daptomycin nonsusceptible (reference MIC, 1.73 μg/ml), and an MIC of >1 μg/ml was obtained in both CA-MHBs and by Etest (Table 1).

TABLE 1.

Initial vancomycin and daptomycin MICs at time of isolation

Isolate	Days of prior vancomycin therapy	Reference MIC (μg/ml) fora:		Vancomycin MIC (μg/ml) in:					Daptomycin MIC (μg/ml) in:
		VAN	DAP	BMD				Etest	BMD		Etest
				Difco/Sigma	Difco/Novation	BBL/Sigma	BBL/Novation		Difco	BBL
1	42	0.75	0.50	0.75	0.75	0.75	0.75	1.5b	0.5	0.5	0.25
2	5	2.00	0.61	2.0	2.0	2.0	2.0	1.5	0.5	0.75	0.5
3	5	1.11	0.75	1.0	1.0	1.5	1.0	1.5	0.75	0.75	0.75
4	18	1.50	0.75	1.5	1.5	1.5	1.5	1.5	0.75	0.75	1.0
5	4	2.24c	1.73	2.0	2.0	2.5	2.5	2.0	1.5	2.0	2.0
6	3	1.00	1.06	1.0	1.0	1.0	1.0	1.5	0.75	1.5	1.0
7	7	0.81	0.25	0.75	0.75	1.0	0.75	0.75	0.25	0.25	0.38
8	5	0.87	0.43	0.75	0.75	1.0	1.0	2.0	0.25	0.75	0.75
9	15	1.03	0.43	1.0	0.75	1.5	1.0	1.5	0.25	0.75	0.25
10	14	3.37	2.24	3.5	3.0	3.5	3.5	4.0	2.0	2.5	1.5

^aThe reference MICs for vancomycin and daptomycin were defined as the geometric mean MIC of 4 BMD results (vancomycin) or 2 BMD results (daptomycin).

^bNumbers in bold type represent MICs at the high end of susceptible (i.e., 1.5 or 2.0 μg/ml).

^cUnderlined numbers represent MICs that are not susceptible.

Among the remaining 8 vancomycin-susceptible isolates, 2 (isolates 2 and 4) reference BMD MICs fell in the high-susceptible range (2.0 and 1.5 μg/ml, respectively; see Table 1 MICs in bold type) by all 4 BMD formulations and by Etest. For isolates 1, 3, 8, and 9, ≥1 vancomycin MIC fell within the high-susceptible range, but the reference vancomycin MICs were in the low-susceptible range (Table 1). All isolates, except 5, 6, and 10, had daptomycin-susceptible reference BMD and Etest results (Table 1). Isolates 5 and 10 were daptomycin nonsusceptible by all methods, but isolate 6 was daptomycin nonsusceptible (MIC, 1.5 μg/ml) only by the BBL CA-MHB BMD (Table 1). The daptomycin MIC measured for this isolate was 0.75 μg/ml when tested by BMD with the Difco CA-MHB and 1.0 μg/ml by Etest.

The number of days of vancomycin therapy did not statistically differ between the isolates with reference vancomycin BMD MICs of <1.5 μg/ml and those with MICs of 1.5 to 2.0 μg/ml (t test, P = 0.757). Likewise, the duration of vancomycin therapy was not statistically different between the daptomycin-susceptible and –nonsusceptible isolates (t test, P = 0.732). All patients received vancomycin doses of 0.75 to 1.5 g intravenously (i.v.), with mixed dosing intervals (after dialysis or every 8 to 24 h). The vancomycin trough levels prior to the isolation of MRSA were available for 7 of the 10 patients but were not tested for patients 1, 5, and 7. Of the patients with available vancomycin troughs, 5 were >15 μg/ml (patients 2, 4, 6, 9, and 10). For patient 3, the vancomycin dosing was increased from 1.25 to 1.5 g (i.v. every 8 h) following a trough level of 5.1 μg/ml; however, the subsequent trough levels were not tested. Likewise, a trough value of 8.2 μg/ml was recorded for patient 8, but no further troughs were measured nor vancomycin dosing changes documented.

MIC precision.

The reproducibility (precision) of the vancomycin and daptomycin MICs was determined by evaluating the standard deviation of the log₂ MICs obtained by testing five isolates in quintuplicate on the same day. Overall, the precision studies of vancomycin MICs obtained for each method (4 BMD methods and Etest) yielded standard deviations of log₂ MICs for each isolate that were well <1 (i.e., less than one 2-fold dilution difference), which is generally considered acceptable for MIC testing (Table 2). Indeed, precision ranged from 0.000 to 0.330 for the vancomycin log₂ MICs achieved by BMD for each isolate tested (Table 2). Etest precision also ranged from 0.000 to 0.320 for the log₂ vancomycin MICs. The precision of the vancomycin BMD across the 4 formulations was assessed by taking the mean standard deviation of the MICs obtained for each of the 5 isolates. From this calculation, the standard deviation of the log₂ MICs was 0.217 (95% confidence interval [CI], 0.146 to 0.281) or ±1.16-fold (95% CI, 1.11 to 1.21).

TABLE 2.

Standard deviation of vancomycin and daptomycin MICs obtained by testing 5 randomly selected isolates in quintuplicate on the same test day

Isolate	SD of vancomycin log2 MIC with:					SD of daptomycin log2 MIC with:
	Difco CA-MHBa		BBL CA-MHB		Etest	Difco CA-MHB	BBL CA-MHB	Etest
	Sigma	Novation	Sigma	Novation
1	0.000	0.262	0.186	0.186	0.186	0.548	0.262	0.622
4	0.320	0.320	0.330	0.186	0.000	0.000	0.227	0.000
5	0.144	0.000	0.144	0.176	0.000	0.000	0.000	0.227
6	0.000	0.000	0.320	0.000	0.320	0.000	0.320	0.227
7	0.227	0.000	0.000	0.186	0.227	0.000	0.868	0.000
Mean	0.138	0.116	0.196	0.147	0.147	0.110	0.335	0.215
Mean fold differenceb	1.10	1.08	1.15	1.11	1.11	1.08	1.26	1.16

^aCA-MHB, cation-adjusted Mueller-Hinton broth.

^bThe mean fold difference was obtained by calculating 2 to the power of the mean of the standard deviations.

In contrast, the daptomycin MICs were somewhat less precise. The standard deviation of log₂ MICs for daptomycin BMD using the BBL CA-MHB was 0.335 (i.e., 1.26-fold difference in MICs), whereas the Difco CA-MHB precision was 0.110 (1.07-fold difference in MICs) and the Etest precision was 0.215 (1.16-fold difference in MICs). The precision of BMD MICs across the two media was 0.409 (95% CI, 0.236 to 0.583), or an average difference of 1.32-fold (95% CI, 1.18 to 1.50) in the MICs measured.

When evaluating these data from a potential treatment decision perspective, all results were in categorical agreement with the reference MIC result, with two exceptions. Isolate 5 was vancomycin susceptible (MIC, 2.0 μg/ml) by 1 of 5 BBL/Sigma results, 2 of 5 BBL/Novation results, and 5 of 5 Etest results, but the reference MIC for this isolate was 2.23 μg/ml. Isolate 6 was daptomycin nonsusceptible (MIC, 1.5 or 2.0 μg/ml) by 3 of 5 BBL BMD MICs, whereas the reference MIC was 0.87 μg/ml. The mean BMD MICs obtained for these five isolates across the 4 BMD formulations are presented in Fig. 1, shown with the 95% confidence intervals. For only isolate 1 were BMD MICs consistently in the low-susceptible range (≤1 μg/ml), whereas for every other isolate measured, the MIC values were above this value.

FIG 1.

Mean BMD results (n = 20) for 5 isolates tested on day 1. Error bars represent 95% confidence intervals.

Effect of drug powder source, CA-MHB manufacturer, and test method on MIC results.

Analysis of variance with post hoc Tukey's tests was performed on log₂-transformed MIC values obtained on the first day of testing to determine if the vancomycin powder source and/or CA-MHB manufacturer significantly impacted the MIC results. The geometric means of the MICs for the 10 isolates are shown in Fig. 2 for each BMD formulation and Etest. The MIC results were found to differ significantly between the Sigma and Novation vancomycin powders (P = 0.028, paired t test), with the Sigma powder yielding vancomycin MICs that were 1.08-fold higher on average than those with the Novation powder (95% CI, 1.01 to 1.16). Similarly, significantly higher MICs were found with the BBL CA-MHB (P = 0.002) than those with the Difco CA-MHB; the MICs were 1.137-fold higher on average (95% CI, 1.057 to 1.222). In contrast, this effect was not seen with the S. aureus ATCC 29213 isolate across 42 data points (not shown). No interaction between vancomycin powder source and CA-MHB manufacturer was found by nested (hierarchical) analysis of variance (ANOVA). The BBL CA-MHB was found to yield significantly higher daptomycin MICs than those of the Difco CA-MHB, by 1.463-fold on average (95% CI, 1.068 to 2.005, P = 0.023). Similarly, the MICs for S. aureus ATCC 29213 were significantly higher with the BBL CA-MHB than those with the Difco CA-MHB, by 1.593-fold (95% CI, 0.305 to 2.881, P = 0.002, not shown). For both media, the calcium concentrations were adjusted to 50 mg/liter and confirmed with HPLC.

FIG 2.

Effect of test method on MIC. Shown are the mean vancomycin (filled circles) and daptomycin (open circles) MICs from testing 10 MRSA strains at the time of isolation. Error bars represent 95% confidence intervals.

The vancomycin MICs obtained by Etest were compared to the mean reference BMD MIC for the 10 isolates on the first day of testing. The Etest MICs were 1.36-fold higher than the BMD MIC (95% CI, 1.008 to 1.577, P = 0.037). When analyzed by BMD formulation, the Etest vancomycin MICs were higher than the MIC obtained by all BMD formulations (with the exception of the BBL CA-MHB/Sigma vancomycin formulation), which were not significantly different than the Etest MIC (P = 0.180, Fig. 2). In contrast, no significant difference was noted between the mean daptomycin BMD and Etest MICs (P = 0.355); the Etest MICs were 0.951-fold lower than those obtained by BMD (95% CI, 0.735 to 1.229). When the results were compared to BMD by CA-MHB manufacturer, the Etest MICs were not significantly different from those obtained with the Difco CA-MHB (P = 0.186) but were 0.78-fold lower than those obtained with the BBL CA-MHB (95% CI, 0.58 to 1.05, P = 0.011).

Effect of subculture and cryopreservation on MICs.

The geometric mean BMD MICs obtained from all 4 (vancomycin) or 2 (daptomycin) BMD formulations were evaluated following subculture and storage. While minor differences in the geometric mean vancomycin MICs obtained before and after subculture were noted, no significant change in vancomycin or daptomycin MIC was found following 5, 10, or 20 subcultures (Fig. 1). Similarly, no significant decrease in vancomycin or daptomycin MICs was found following 1, 3, or 6 months of frozen storage (Fig. 1). In contrast, the vancomycin MICs were 0.741-fold lower following 12 months of frozen storage compared to those observed on day 1 (95% CI, 0.58 to 0.95, P = 0.021). No significant change in daptomycin MICs was noted after 12 months of frozen storage (Fig. 3).

FIG 3.

Effect of subculture and storage on MIC. Shown are the mean vancomycin (filled circles) and daptomycin (open circles) MICs obtained from testing 10 MRSA strains at the time of isolation and following subculture/freezing.

DISCUSSION

The precision of dilution-based reference MIC methods (e.g., BMD, agar dilution) is generally accepted to be ±1 log₂ dilution (17). This acceptance is embedded into the international standards (18) and regulatory environment (19). With the widespread use of Etest, which yields MIC values between the traditional log₂ dilutions, the notion of further improving the performance of MIC tests to measure the in-between MICs has been introduced. This concept is well illustrated for S. aureus, for which vancomycin MICs of 1.5 (and 2.0) μg/ml have been associated with poor clinical outcomes of vancomycin therapy for S. aureus bacteremia (1–7). Some investigators suggest that an MIC of 1.5 μg/ml should be the cutoff at which a change in antimicrobial therapy is made. However, in the background of this clinical discussion is the ongoing struggle by laboratories to accurately determine vancomycin MICs for S. aureus. Isolates with vancomycin-intermediate MICs (4 to 8 μg/ml) that are attributable to a number of different physiological changes in the organism are not easily detected by routine laboratory methods. These vancomycin-intermediate S. aureus (VISA) isolates typically grow more slowly in culture than do vancomycin-susceptible isolates. As such, they may not be recovered if both exist in the same patient specimen, and laboratories may consider testing each S. aureus strain isolated from blood (20), especially if different colony morphologies are evident on the primary isolate plates (21, 22). It has also been suggested that the VISA phenotype may be lost following serial subculture of isolates (22). Our study demonstrates that the effect of up to 20 serial subcultures on vancomycin and daptomycin MICs is minimal (Fig. 1), although we had only 2 VISA isolates included in our study. Regardless, this finding is important for clinical microbiology laboratories, most of which follow the CLSI and Centers for Disease Control and Prevention (CDC) recommendations that S. aureus strains with vancomycin MICs of >2 μg/ml and daptomycin MICs of >1 μg/ml (16, 23) be confirmed by repeat testing, due to the infrequent occurrence of vancomycin-intermediate S. aureus and daptomycin-nonsusceptible S. aureus. This repeat testing requires a limited number of subcultures, which based on our data, should not impact MIC results. In contrast, a significant reduction in vancomycin MICs, i.e., 0.741-fold below the day 1 value, was noted following frozen storage (Fig. 3). This finding may have implications for clinical trials or retrospective studies of vancomycin treatment failures for which isolates are often shipped to reference laboratories and stored prior to testing. Interestingly, there was no effect on the daptomycin MICs associated with cryopreservation of the isolates. In contrast to our results, Edwards and colleagues (9) noted a significant decrease in vancomycin MICs after a minimum 6 months of frozen storage compared to the original MICs; this study was conducted using Etest. We did not note any difference in the MICs after 6 months of frozen storage, either by BMD or Etest (not shown).

It is well recognized that not all susceptibility test methods distinguish vancomycin-susceptible S. aureus from VISA (12), including disk diffusion (16, 24). Many of the commercial test systems used in clinical laboratories have been reported to under- or overcall vancomycin-intermediate MICs compared to reference BMD. Etest in particular has been associated with vancomycin MICs 1- to 2-log₂ dilutions higher than those achieved by BMD (11–13, 24–26, 35). We speculate that the higher MICs observed with Etest are a result of the higher inoculum density used. The distinct inoculum effect has been observed by Craig et al. (26) in an animal model examining the pharmacodynamics of vancomycin. In our present study, the Etest vancomycin MICs were significantly higher than those obtained by BMD (Fig. 2), and for only 1 isolate (number 7) was the MIC <1.5 μg/ml by Etest (Table 1). In contrast, for an additional 3 to 5 isolates, the measured MICs were <1.5 μg/ml by BMD. Some have suggested that Etest is a more sensitive method with which to identify S. aureus strains with reduced susceptibility (14, 28), but this has not been well validated. Regardless, it is clear that the use of vancomycin MICs in the high-susceptible range, as reported in the literature, to predict a patient's clinical outcome with vancomycin therapy requires taking the method of susceptibility testing used in a given study into account. To this point, it has been suggested that laboratories indicate the method used for vancomycin MIC determination on laboratory reports in order to better aid physician interpretation of the results (22). Our study demonstrates that significant variability in MIC results can be observed even within a single reference method. This variably will lead to isolates being classified as “high susceptible” versus “low susceptible,” based on subtle differences of the BMD test, such as the source of vancomycin powder or manufacturer of CA-MHB. This is illustrated in Table 1, where it can be seen that between 4 (by BBL CA-MHB/Sigma vancomycin formulation) and 6 (BBL/Novation and Difco CA-MHB formulations) isolates were associated with vancomycin MICs of <1.5 μg/ml. More specifically, we found that both the BBL versus Difco CA-MHB and Sigma versus Novation vancomycin powders yielded significantly higher vancomycin MICs (Fig. 2). Interestingly, Swenson and colleagues (12), who evaluated the vancomycin MICs obtained for 129 MRSA isolates using BMD in either the BBL or Difco CA-MHB observed more isolates with intermediate MICs in the Difco CA-MHB (34.9%) than in the BBL CA-MHB (31.8%). The vancomycin powder source was not documented in this study. The different results obtained by our two studies may be explained by the fact that different lots of CA-MHB were used in the studies. Alternatively, this difference may be representative of the manual preparation of vancomycin dilutions, which might have yielded slight differences in final vancomycin concentrations (29), and not be related to CA-MHB manufacturer at all. The BBL CA-MHB in our study also resulted in significantly higher daptomycin MICs (Fig. 2).

Beyond the variability associated with manually preparing antimicrobial dilutions, many variables are inherent in the reference BMD test, even when CLSI and ISO standards are followed precisely. These variables include differences in MIC panel manufacturing, the organisms to be tested, the inoculation method used for the BMD panels, incubation conditions, and the method used to read MICs. The variables pertaining to each of these factors are listed in Table 3. With all this variability introduced into the test system, it is not surprising that precision below a log₂ difference in MICs is not typically achievable. Indeed, the MIC quality control ranges for a single drug-bug combination typically span a 3- to 4-log₂ dilution range. These ranges were established by the CLSI by an evaluation of the MICs obtained in an 8-laboratory study, wherein each laboratory determined 10 replicate MICs for the QC organism for each of 3 different manufacturers of CA-MHB and one lot of antimicrobial powder (28). The quality control MIC range for S. aureus ATCC 29213 for vancomycin is 0.5 to 2.0 μg/ml (16), suggesting a realistic precision for vancomycin BMD across laboratories for this QC strain of S. aureus ranges over 3 dilutions. However, individual laboratories typically achieve much tighter precision in their day-to-day quality control testing. From the data presented here, the mean precision of vancomycin MICs across 2 CA-MHB manufacturers and 2 sources of vancomycin powder for clinical isolates of S. aureus was found to be ±0.217 log₂ dilutions. The mean precision of the daptomycin MICs was ±0.409 log₂ dilutions. Even greater precision was found within a single CA-MHB and/or vancomycin powder source (Table 2). Our data therefore suggest that testing interdilution MICs (i.e., 0.5 log₂) may be a possibility using BMD and yield reproducible results. It should be pointed out that three highly trained operators, who strictly adhered to the same test methodology, achieved these results in a single laboratory. From the QC studies, it is clear that this level of precision is not achievable across laboratories. To better illustrate this concept, the College of American Pathologists recently reported the vancomycin MIC results obtained by 2,018 laboratories for a vancomycin-susceptible MRSA isolate. These laboratories used one of 7 different commercial test systems and reference BMD and agar dilution. Within each test system, 95% of the MIC results ranged across 2 to 3 log₂ dilutions. Of the laboratories, 21.6% reported a vancomycin-intermediate MIC (2 to 7.9 μg/ml); the majority of these laboratories (424/438) used the Siemens MicroScan system. In 2010, Siemens issued a technical advisory that elevated vancomycin MICs might be seen when MicroScan panels were inoculated with the Prompt method. It is unclear what proportion of laboratories continue this practice, but this may account for the elevated vancomycin MICs seen by some MicroScan users in this College of American Pathologists (CAP) survey. Interestingly, 3 of the 11 laboratories using reference BMD also reported a vancomycin-intermediate MIC in this CAP survey. Thus, it is clear that significant improvements to the standardization of MIC testing need to be implemented before achieving precision within ±0.5 log₂ dilutions is possible. For those laboratories performing commercial test methods, variability associated with the manufacture of the test panels has been removed from the system; however, at a minimum, improved standardization of organism preparation, panel inoculation, and, in some instances, incubation and reading could be achieved. The question remains as to whether this increased standardization is realistic for laboratories, or even desirable. For example, improving the standardization of the age of the colony to be tested and the growth phase might inadvertently lead to delays in the time before MIC testing. Given that the studies from which the recommendation of vancomycin MICs of 1.5 or 2.0 μg/ml were done using a variety of MIC test methods, many of which were outside the existing CLSI and ISO standards (8), further work is required before the true significance of these interdilution MICs can be confirmed. Furthermore, current regulations in the United States require only >90% essential agreement (i.e., 90% of MICs within 1 log₂ dilution of the reference MIC) for a test system to be acceptable for patient testing, and thus, the incentive for commercial manufacturers to improve precision is lacking.

TABLE 3.

Variables inherent to the reference BMD test

Variable type	Specific variables
BMD panel manufacture	Type of tray used (U or V bottom)
	Plastic composition of 96-well tray
	Method of sterilization of the tray
	Actual CA-MHB composition and manufacturer
	Drug powder source
Organism	No. of subcultures of isolate from primary isolation plate
	Potential storage of isolate and conditions under which isolate was stored
	Age of colony picked for testing
	No. of colonies sampled for testing
	Phase of growth
	Method used for inoculum standardization
	Actual concn of cells in inoculum
	Inoculation of BMD panel
	Dilution water type
	1- or 10-μl inoculum vol
Incubation conditions	Length of incubation within acceptable limits
	Incubation atmosphere
	Temp of incubator
	Relative humidity of incubator
Method for reading MIC	Mirror
	Light box
	Reader

The primary limitation of our study is the relatively small number of isolates evaluated. We chose to evaluate only MRSA strains isolated from patients who were on vancomycin therapy in an attempt to enrich the study with isolates with elevated vancomycin MICs. While the recovery of MRSA from blood is not an infrequent event in our laboratory, with approximately 150 single patient isolates per year, this rarely occurs when the patient is on vancomycin therapy. The collection of these 10 isolates took 2 years, which may further be an attestation to the continued effectiveness of vancomycin therapy. Furthermore, our laboratory only rarely isolates MRSA strains with vancomycin MICs of >1 μg/ml; in 2013, only 4 such isolates were recovered from blood. Thus, the inclusion of a larger collection of isolates into a study similar to the one reported here will take the concerted efforts of multiple laboratories over several years. The advantage of such a prospective multilaboratory study is that it might aid in further evaluations of the precision of interdilution vancomycin and daptomycin MICs. A second limitation of our study is that we evaluated only 4 main variables of BMD (CA-MHB manufacturer, vancomycin powder source, subculture, and frozen storage) of the large list of potential variables that could be evaluated. Finally, the vancomycin stock solutions made from each powder for panel manufacturing were not available for further study; it is possible that slight differences in the concentration of each stock existed, which might have accounted for some of the variability observed in this study. Vancomycin is manufactured via a fermentation process, which produces many structurally similar components and degradation products, along with the active vancomycin B component. Some of these impurities exhibit antimicrobial activity, such as monodechlorovancomycin, which has about half the activity of vancomycin B, whereas others have no antimicrobial activity (30). As such, the U.S. Food and Drug Administration (FDA) evaluates both the purity and potency of vancomycin preparations. The United States Pharmacopeia (USP) acceptance criteria for vancomycin hydrochloride products are a purity of ≥80.0% for vancomycin B, with ≤9.0% of each individual impurity, and a potency of 90 to 115% (31). Two recent studies conducted by the FDA in response to concerns regarding the potency and purity of generic vancomycin for injection (30) found 89 to 95% purity when measured by both the British Pharmacopoeia and USP methods, as well as a potency that ranged from 97 to 112% (32), compared to the USP standard, for six generic vancomycin products available in the United States. These differences, while within USP standards, may be a source for vancomycin MIC variability such as that observed in our study. For example, a 1-μg/ml stock of vancomycin might contain the equivalent of 0.9 μg/ml to 1.15 μg/ml vancomycin activity and yield corresponding MICs between 1.5 μg/ml and 0.75 μg/ml for an isolate with an MIC of 1.0 μg/ml by the USP standard.

Vancomycin is increasingly being questioned as the treatment of choice for MRSA bacteremia due to a number of factors, with the reduced susceptibility of the isolates being only one. Vancomycin MIC values (and not just categorical interpretations) are being used with increasing frequency to inform treatment decisions in the absence of other predictors of the potential for reduced susceptibility to vancomycin in a given isolate of MRSA and treatment failure. Alternative therapies to vancomycin for bacteremia due to MRSA are limited. Daptomycin has been shown to be effective for the treatment of MRSA bacteremia, with the exception of that caused by left-sided endocarditis, and it may retain bactericidal activity in spite of slightly elevated MICs (33, 34, 36). Daptomycin is expensive and may not be available in all institutions. Further, some isolates, such as those observed in our study, develop daptomycin and vancomycin resistance in parallel (33, 34). As there are no clear molecular targets associated with reduced susceptibility to vancomycin, it is clear that the generation of an accurate vancomycin MIC is imperative for patient care. Despite the inherent variability of the MIC, no better test by which to predict treatment outcomes for MRSA infections has been developed. As such, the development of a more precise MIC test that can be performed by all laboratories is desirable. In order to achieve this goal, further study into the effects of the different variables associated with BMD will be important. In parallel, the development of methods that are more precise than the MIC test to determine antimicrobial susceptibility is needed.