|Home | About | Journals | Submit | Contact Us | Français|
An increase in the distribution of vancomycin MIC values among methicillin (meticillin)-resistant Staphylococcus aureus (MRSA) isolates has been noted. It is postulated that the shift in vancomycin MIC values may be associated with a concurrent rise in the MIC values of other anti-MRSA agents. Scant data are available on the correlation between vancomycin MIC values and the MIC values of other anti-MRSA agents. This study examined the correlation between vancomycin MIC values and the MIC values of daptomycin, linezolid, tigecycline, and teicoplanin among 120 patients with bloodstream infections caused by MRSA at a tertiary care hospital between January 2005 and May 2007. For each included patient, the MIC values of the antibiotics under study were determined by the Etest method and were separated into the following two categories: day 1 (index) and post-day 1 (subsequent). For subsequent isolates, the MIC values for each antibiotic from the post-day 1 terminal isolate were used. Among the index isolates, there was a significant correlation (P value, <0.01) between the MIC values for vancomycin and daptomycin and between the MIC values for vancomycin and teicoplanin. The MIC values for daptomycin were significantly correlated with linezolid, tigecycline, and teicoplanin MIC values. Among the 48 patients with subsequent isolates, vancomycin MIC values were significantly correlated with MIC values for daptomycin, linezolid, and teicoplanin (ρ value of ≥0.38 for all comparisons). This study documented an association between vancomycin MIC values and the MIC values of other anti-MRSA antibiotics among patients with bloodstream infections caused by MRSA primarily treated with vancomycin.
An increase in the distribution of vancomycin MIC values among methicillin (meticillin)-resistant Staphylococcus aureus (MRSA) isolates has been noted in several recent reports (3, 11, 14). This shift is a concern because a growing number of studies have shown that patients with infections caused by MRSA with vancomycin MIC values at the higher end of the Clinical and Laboratory Standards Institute (CLSI) and Food and Drug Administration (FDA) susceptibility range are less responsive to vancomycin (3, 4, 5, 7, 12, 13). Clinicians must now consider using alternative therapies for such patients. However, it is unclear whether the observed shift in the distribution of vancomycin MIC values for MRSA is associated with similar shifts in the distribution of MIC values of other anti-MRSA agents.
It is postulated that the shift in vancomycin MIC values may be associated with a concurrent rise in the MIC values of other anti-MRSA agents. Scant data are available on the correlation between vancomycin MIC values and the MIC values of other anti-MRSA agents. To date, analyses have been limited to the index MRSA isolate and have primarily focused on the correlation between vancomycin and daptomycin (6, 10, 11).
This study examined the correlation between vancomycin MIC values and the MIC values of daptomycin, linezolid, tigecycline, and teicoplanin among patients with bloodstream infections caused by MRSA. Given the recent reports describing the emergence of resistance during therapy, our analyses included both the index (day 1) and subsequent (post-day 1) isolates.
This retrospective, cohort study was performed at the Albany Medical Center Hospital (AMCH), a 631-bed tertiary care, academic hospital located in upstate New York. Patients were included in the analysis if they had a bloodstream infection caused by MRSA between January 2005 and May 2007. During the study, vancomycin was the primary agent used at AMCH for the treatment of infections caused by MRSA, administered to more than 90% of patients with bloodstream infections caused by MRSA.
For each included patient, the MIC values of the antibiotics under study (vancomycin, daptomycin, linezolid, tigecycline, and teicoplanin) for the MRSA bloodstream isolates were separated into the following two categories: day 1 (index) and post-day 1 (subsequent).
Each patient was included only once during the study period. For patients with multiple hospitalizations during the study period, the isolate meeting the above-mentioned criteria from the first hospitalization was considered the index (day 1) isolate, and all additional isolates recovered during the study period, irrespective of which hospitalization, were considered subsequent (post-day 1) isolates. If multiple subsequent isolates were obtained, the MIC values from the post-day 1 terminal isolate were used in the analysis.
All MRSA isolates recovered from blood cultures were collected at AMCH during the study period. The date and time of the MRSA cultures were recorded, as well as the time of the last MRSA blood culture. All isolates were identified as S. aureus by the tube coagulase method. Initial susceptibility testing for oxacillin resistance was performed according to CLSI guidelines using a 30-μg cefoxitin disc and Mueller-Hinton II agar (1). Individual isolates were then stored in a Trypticase soy broth with 20% glycerol at −70°C until MIC testing was performed. The isolates were not thawed or subcultured between initial storage and MIC testing.
For patients who met the study criteria, the Etest methodology was used to determine the MIC of each recovered isolate for vancomycin, daptomycin, linezolid, tigecycline, and teicoplanin. Prior to MIC testing, each isolate was subcultured to Trypticase soy agar plates supplemented with 5% Trypticase soy sheep blood agar (BD, Sparks, MD). The plates were incubated overnight (18 to 24 h) at 35°C in ambient air, and the subculture was repeated a second time. From these pure cultures, multiple colonies were suspended in 0.45% saline to create a 0.5 McFarland turbidity standard. This standardized suspension was used to streak the inoculum onto the surface of a 150-mm Mueller-Hinton II agar (BD) plate to create a confluent lawn of microbial growth. The surface of the plate was allowed to dry for 15 min prior to applying the Etest strip (AB Biodisk, Solna, Sweden).
The MIC values were determined with the Etest method, according to the manufacturer's instructions. The MIC testing was performed over a period of 4 weeks in a single laboratory using ATCC 29213 (American Type Culture Collection, Manassas, VA) as the quality control strain. The isolates were tested in six batches over the 4-week period, and the MIC values for the quality control strain were within the expected range for all antibiotics tested at each run. All MIC values were read by a single observer (A.E.), and A.E. was blinded to patient information and isolate type (index versus subsequent).
Descriptive statistics, including the geometric MIC mean, MIC mode, MIC range, MIC50, and MIC90, were performed for each drug on the day 1 isolate (index) and post-day 1 terminal isolate (subsequent). The descriptive statistics were also stratified by the initial vancomycin MIC (≥1.5 mg/liter or <1.5 mg/liter), as this has been identified as an important determinant of outcomes in previous analysis (3, 4, 5, 7, 12, 13). Among patients with subsequent isolates (n = 48), Fisher's exact test was used to determine if there was a significant (P value, <0.05) increase in the proportion of patients with an increase in MIC values within each MIC category.
Spearman's rank correlation coefficient test was used to assess the association among the MIC values of the five antibiotics for all index isolates (n = 120). It was also used to evaluate the correlations among the 48 patients with post-day 1 subsequent isolates. If multiple subsequent isolates were obtained from a patient, the MIC values from the terminal isolate were used in the analysis. Correlations were also performed using the terminal isolate MIC values from days 2 to 4, days 5 to 30, and day 31 on, but these analyses failed to produce any additional insights beyond the results generated using the MIC values from the last (terminal) isolate recovered from a patient. In an additional correlation analysis, we used the highest MIC values observed among any subsequent isolate, and the results were nearly identical to those observed for the MIC values of the post-day 1 terminal isolate. In all analyses, a P value of <0.05 was considered significant for two-tailed tests. All calculations were performed with SYSTAT for Windows (version 11.0) and SPSS version 11.5 (SPSS, Chicago, IL).
This analysis included a total of 222 isolates collected from 120 patients over a 30-month period. The MIC values and descriptive statistics of the five antibiotics for index and subsequent isolates are provided in Table Table1.1. The MIC values for the index isolates of the patients with subsequent isolates (48 patients) were identical to those of the 72 patients without subsequent isolates (data not shown). Although the median MIC (MIC50) remained constant between the day 1 and terminal isolates for all antibiotics, the geometric mean MIC value of the subsequent isolates was higher than that of the index isolate for vancomycin, daptomycin, and teicoplanin. Among the 48 patients with subsequent isolates, the following number (percentage) of patients had an increase in MIC from baseline for each antibiotic: vancomycin, 14 (29.2%); daptomycin, 20 (41.7%); linezolid, 14 (29.2%); tigecycline, 16 (33.3%); and teicoplanin, 19 (39.6%). The P value was <0.05 for all antibiotics in all the aforementioned comparisons.
The descriptive MIC statistics of day 1 isolates for each drug stratified by the index (day 1) vancomycin MIC (≥1.5 mg/liter or <1.5 mg/liter) are depicted in Table Table2.2. The MIC50 and MIC90 values were higher for all of the drugs when the index vancomycin MIC was ≥1.5 mg/liter, except for the daptomycin MIC90 and the linezolid MIC50.
Spearman's correlation coefficients among the MIC values of the five antibiotics for the index isolates are depicted in Table Table3.3. Among the index isolates, there was a significant correlation (P value, <0.01) between vancomycin and daptomycin and between vancomycin and teicoplanin. In addition, daptomycin MIC values were found to be significantly correlated with those of all other anti-MRSA agents. When the correlation analysis was restricted to the 48 patients that had subsequent cultures, the correlation coefficients for the index isolate MIC values among these 48 patients were similar to those reported for the entire population in Table Table3.3. The only correlation coefficients that changed markedly in this restricted analysis were those between vancomycin and teicoplanin (ρ = 0.22), daptomycin and tigecycline (ρ = 0.05), daptomycin and teicoplanin (ρ = 0.067), and linezolid and teicoplanin (ρ = 0.201).
Correlations among the subsequent (post-day 1 terminal) isolates are shown in Table Table4.4. All antibiotics were significantly correlated with at least one other antibiotic. Among subsequent isolates, vancomycin was significantly correlated with daptomycin, linezolid, and teicoplanin (ρ of >0.35 for all comparisons). In addition, strong correlations were observed between daptomycin and linezolid and between tigecycline and teicoplanin. The index vancomycin MIC values were also significantly correlated with the subsequent post-day 1 terminal isolate MIC values of vancomycin (ρ = 0.59, P value < 0.01), daptomycin (ρ = 0.36, P value < 0.01), and linezolid (ρ = 0.37, P value < 0.01).
There are limited data on the association between the MIC values of vancomycin and those of other anti-MRSA agents among patients with bloodstream infections caused by MRSA. To date, data are limited on the index MRSA isolate and primarily focused on the association between vancomycin and daptomycin or linezolid (14, 15). Our study expanded upon this literature by including post-day 1 terminal (subsequent) isolates and correlation analyses among the MIC values for vancomycin and several other anti-MRSA agents.
Several interesting relationships were noted. Among the index (day 1) MRSA isolates, significant correlations with the vancomycin MIC were limited to the daptomycin MIC and the teicoplanin MIC. However, the daptomycin MIC was significantly correlated with the MIC values of all agents. Interestingly, subsequent post-day 1 terminal vancomycin MIC values were significantly correlated with the MIC values of daptomycin, linezolid, and teicoplanin, and the Spearman correlation coefficients were rather pronounced (ρ > 0.35) for all agents. Consistent with findings of the subsequent isolate correlation analysis, a shift in the MIC distribution and geometric mean MIC between index and subsequent post-day 1 terminal isolates was noted for all drugs. It is unlikely that the higher MIC distributions and correlation coefficients among the post-day 1 terminal isolates were due to selection bias. The MIC values for the index isolates of the patients with subsequent isolates (48 patients) were identical to those of 72 patients without subsequent isolates. Furthermore, when the correlation analyses were restricted to the 48 patients with subsequent isolates, the index isolate correlation coefficients among the 48 patients with subsequent cultures were very similar to those observed in the overall population in Table Table33.
Collectively, these results indicate that the MIC values of all antibiotics under study increased in parallel, especially among subsequent isolates under the selective pressures of antibiotic therapy. This is suggestive of a common mechanism of reduced antibiotic susceptibility. It is important to note that these increases in MIC values among subsequent isolates occurred primarily under vancomycin selective pressures. During the study, vancomycin was the primary agent used at AMCH for the treatment of infections caused by MRSA, administered to more than 90% of patients with bloodstream infections caused by MRSA.
The definitive mechanism(s) responsible for the observed results is unknown. Given the unique mechanism of action and low potential for cross-resistance for these agents, it is unlikely that a single mutational event is responsible. Rather, the collective results indicate that adaptive resistance mechanisms involving the global regulation of gene expression may be involved. While the exact adaptive mechanisms employed by MRSA are unclear, there are several potential explanations that may confer a survival advantage against each drug, and these include a loss of accessory gene regulator (agr) function, horizontal gene transfer, antibiotic trapping, spontaneous mutations, a general SOS response, a decrease in growth and metabolism, and positive selection (8, 9).
Data also suggest that vancomycin selective pressures may have a role in the selection of MRSA isolates that have reduced susceptibility to other agents. There is growing evidence supporting the role of agr dysfunction in the presence of vancomycin as a potential mechanism for collateral damage to other anti-MRSA agents. This mutation was identified following vancomycin exposure in clinical bloodstream infections caused by MRSA and accompanied the development of vancomycin heteroresistance (6). Previous work by Sakoulas et al. noted that physiologic changes in MRSA may occur with vancomycin exposure and that these changes may influence the susceptibility to other agents (10). Whether agr dysfunction is the culprit or merely a coconspirator merits further investigation, as does the role of vancomycin selective pressure on the susceptibility of other agents. Likewise, antibiotic trapping has been seen in S. aureus with vancomycin exposure (2). In such instances, vancomycin clogs the thickened cell, altering vancomycin diffusion.
We are aware of only one other study that examined correlations among the MIC values of vancomycin and other anti-MRSA agents. That study examined vancomycin-intermediate Staphylococcus aureus (VISA) strains from different countries and found a negative correlation coefficient (−0.6) between linezolid and vancomycin MIC values (15). This is in contrast to the findings of our study; there was no correlation between vancomycin and linezolid MIC values among index isolates, and there was a significant correlation between vancomycin and linezolid MIC values among the patients with subsequent MRSA isolates. There are several important points to consider when comparing results between studies. As previously mentioned, this study examined the correlation between linezolid and vancomycin MIC values among patients with infections caused by VISA from different countries. It is unclear what selective drug pressures were present and whether the isolates were index isolates or subsequently obtained VISA isolates. Our MRSA isolates came from a single institution where vancomycin was used in ~90% of cases, and there was only one VISA case. Further studies are still needed to examine the relationship between vancomycin MIC values and those of other anti-MRSA agents. However, these differences highlight the factors and MRSA strain types that need to be considered in future investigations.
The convergence of two trends—increasing prevalence of community-acquired infection caused by MRSA and growing multidrug resistance—makes this study of antibiotic susceptibility very timely. While the clinical significance of increased MIC values is not yet fully understood, we strongly suspect that they affect the pharmacodynamics of multiple drugs in MRSA patients. We know that the MIC affects the area under the concentration-time curve/MIC ratio and therefore results in suboptimal pharmacotherapy for concentration-dependent antimicrobials. The suboptimal drug concentrations may then trigger the development of resistance due to subtherapeutic drug ranges and the presence of selective pressures from the hospital environment. Even an MIC increase of as little as 1 μg/ml can have an impact on treatment due to the area under the concentration-time curve/MIC ratio. Based on the correlations observed in this study, if a patient presents with MRSA and has a vancomycin MIC of ≥1.5 mg/liter, the patient will likely also have an elevated MIC of the other agents against gram-positive bacteria. This would necessitate more aggressive treatment, both surgical and pharmacologic, to achieve a good clinical outcome.
Several limitations may lessen the generalizability of this study. This study explored only MRSA bloodstream isolates from a single site. Institutional differences in prescribing patterns, antibiotic formularies, and patient populations may affect the applicability of these results to other institutions. Since we do not perform molecular typing, we cannot definitely conclude if the subsequently recovered MRSA strain was the same as the index isolate. However, our results indicated that the MIC values of the tested anti-MRSA agents increase over time in a given patient and will often be higher than those of the initial isolate. Finally, we cannot definitively determine if the observed MIC correlations were due to a particular antibiotic exposure. During the study, greater than 90% of patients received vancomycin for their bloodstream infections caused by MRSA. In the absence of a comparator group, it is difficult to definitely determine if these observed MIC correlations were definitely due to vancomycin. We can conclude only that these correlations were observed primarily under vancomycin selective pressures.
This study documented a positive association between vancomycin MIC values and the MIC values of other anti-MRSA antibiotics among patients with bloodstream infections caused by MRSA. The relevance of our observed MIC “creep” among all the drugs to clinical practice is not yet clear, and the mechanisms underlying these phenotypic observations warrant further investigation.
Published ahead of print on 5 October 2009.