Vancomycin use has increased dramatically worldwide since the mid-1980s, largely as a result of empirical and directed therapy against burgeoning methicillin-resistant Staphylococcus aureus (MRSA) infections. With limited choices, clinicians have traditionally relied on vancomycin alone in the management of serious MRSA infections and have enjoyed a significant period free of vancomycin resistance in S. aureus. Even now, 5 decades after its introduction, vancomycin resistance among S. aureus strains, as currently defined microbiologically, remains rare. Yet it is becoming clear that vancomycin is losing potency against S. aureus, including MRSA. Serious infections due to MRSA defined as susceptible in the laboratory are not responding well to vancomycin. This is demonstrated by increased mortality seen in patients with MRSA infection and markedly attenuated vancomycin efficacy caused by vancomycin heteroresistance in S. aureus. Therefore, it appears that our definition of vancomycin susceptibility requires further scrutiny as applied to serious MRSA infections, such as bacteremia and pneumonia.
Historically, the development of antimicrobial resistance in Staphylococcus aureus has been rapid. Resistance to penicillin in S. aureus was noted only a year after its introduction, and, in the early 1950s, three-quarters of S. aureus strains in large hospitals in many countries had become penicillin resistant [1]. Currently, 90%–95% of clinical S. aureus strains throughout the world are resistant to penicillin. In 1959, the first antistaphylococcal penicillin—methicillin—was introduced. Within 2 years, the first methicillin-resistant S. aureus (MRSA) strain emerged [2]. Currently, MRSA accounts for ∼60% of clinical S. aureus strains isolated from intensive care units in the United States [3, 4]. With the rapid emergence of community-associated MRSA, an organism that now causes a majority of the soft-tissue infections in patients presenting to emergency rooms in many parts of the United States, it appears that the antistaphylococcal β-lactams may soon meet the same fate as penicillin with regard to their ability to treat community-associated skin infections [5–7].
This pattern has continued among the newer agents. Linezolid was introduced clinically in the year 2000, only to result in the first linezolid-resistant MRSA strain being described in the literature a year later [8]. Daptomycin was introduced in 2003, and MRSA resistance to it was first reported within 2 years [9].
What is interesting about vancomycin is that, unlike any of the other antistaphylococcal antimicrobials, resistance to this agent among S. aureus strains took almost 40 years to be recognized, with the first glycopeptide-intermediate S. aureus (GISA) isolate from a pediatric patient in Japan described in 1996 [10]. High-level resistance mediated by the vanA gene complex acquired from vancomycin-resistant enterococci (VRE) emerged in Detroit, Michigan, in 2002 and so far has been limited to the United States [11, 12]. The cases for which the clinical details are reported are shown in table 1. It is noteworthy that the vancomycin-resistant phenotype has variable expression, with higher levels of resistance seen in the Michigan strain compared with the subsequently found strain from Pennsylvania, perhaps because of a more unstable phenotype or differences in metabolic price (figure 1) [17].
Agar-based susceptibility of vancomycin-resistant Staphylococcus aureus strains from Michigan (MI-VRSA) in 2002 and from Pennsylvania (PA-VRSA) in 2003. Disks on the left denote vancomycin (VA) and teicoplanin (TP). E-strips are labeled accordingly. This figure demonstrates that the vanA-mediated vancomycin-resistant phenotype has differential expression in S. aureus, with a high level of resistance in the MI-VRSA strain and a lower level in the PA-VRSA strain (and New York strain; not shown) [17]. Reprinted from [17], with permission from the University of Chicago Press.
Peptidoglycan biosynthesis and mechanism of action of vancomycin. Binding of vancomycin to the C-terminal D-Ala-D-Ala of late peptidoglycan precursors inhibits transglycosylases, transpeptidases, and D,D-carboxypeptidases [17]. Ddl, D-Ala:D-Ala ligase; MurF, synthetase protein; UDP, uracil diphosphate. Reprinted from [17], with permission from the University of Chicago Press.
Morphological cell-wall thickening in vancomycin-intermediate Staphylococcus aureus strains (a and b), a vancomycin-susceptible revertant strain (c), and control NCTC 8325 (d) (magnification, × 63,000) [19]. Reprinted from [19], with permission from Gabrielle Bierbaum and the American Society for Microbiology.
Unlike β-lactam antibiotics, which bind to and interrupt the activity of penicillin-binding proteins (enzymes involved in cell-wall synthesis), vancomycin binds with high affinity to the D-Ala-D-Ala C-terminus of late peptidoglycan precursors and prevents reactions of cell-wall synthesis using these precursors in transglycosylase, transpeptidase, and D,D-carboxypeptidases (figure 2) [17]. Resistance in VRE and vancomycin-resistant S. aureus (VRSA) is due to the presence of operons that encode enzymes that produce the low-affinity precursors D-Ala-D-Lactate or D-Ala-D-Ser and also enzymes that eliminate the competitive high-affinity peptidoglycan precursors normally produced. In GISA strains, none of the operons mediating this mechanism of resistance has been found. Instead, GISA has altered its cellular physiology as a result of cumulative effects of mutations and/or modulation of regulatory systems. This altered physiology appears to change cell-wall metabolism in such a way as to result in increased numbers of D-Ala-D-Ala residues, which serve as dead-end binding sites for vancomycin. This altered cell wall results in a reduced diffusion coefficient of vancomycin, sequestration of vancomycin within the cell wall by these false targets, and prevention of vancomycin reaching its site of action [17]. In addition, evaluation of S. aureus with reduced vancomycin susceptibility and isogenic vancomycin-susceptible progenitors showed cell walls with reduced peptidoglycan cross-linking, reduced cell-wall turnover, and reduced autolysis, which investigators suggested may be due to teichoic acid structure and metabolism [18]. These metabolic changes result in considerable morphological cell-wall thickening, as shown in figure 3 [19].
The phenotype of many of these GISA strains is unstable, suggesting that the regulatory foci can be turned on or off, depending on whether glycopeptide is present. Some of the genes whose expression has been found to be altered in GISA include agr, pbp2, pbp4, pbpD, sigB, ddh, tcaA, and vraSR. However, what is now clear is that the end result of altered cell-wall metabolism, rather than the specific genetic mechanisms achieving this altered metabolism, is of primary importance [20].
An evaluation of >35,000 S. aureus strains from the SENTRY database isolated worldwide between 1997 and 2003 showed no evidence of increasing vancomycin resistance over time when analyzed by vancomycin MICs (i.e., no “MIC creep” was noted). The percentage of isolates with vancomycin MICs of >2 µg/mL ranged from 0% to 0.1% per year [21]. However, small studies performed at individual centers in New York, Texas, and Massachusetts have shown subtle but significant increases in vancomycin MICs in both MRSA and methicillin-susceptible S. aureus (MSSA) [22–24]. In the largest evaluation to date, an analysis of >6,000 S. aureus isolates recovered over a period of 5 years at a southern California university hospital found only 1 VISA strain and no VRSA strains. However, there was a significant drift toward reduced susceptibility, with an increase in the percentage of isolates (from 19.9% in 2000 to 70.4% in 2004) with a MIC equal to 1.0 µg/mL (P<.01 ) [25].
Further complicating this evaluation is the emergence of community-associated MRSA, which has been shown, as expected, to have lower MICs for vancomycin than that of health care–associated MRSA (MIC50, 1 µg/mL vs. 2 µg/mL) [26], perhaps because of less exposure historically to glycopeptides. Therefore, differential prevalence of community-associated MRSA at different hospitals is expected to affect the range of MICs seen for vancomycin among MRSA strains. How these microbiological observations translate into differential treatment efficacy of glycopeptides against community-associated MRSA versus health care–associated MRSA remains to be determined. Given that different hospital centers have varying antibiotic prescribing practices, both in agent selection and dosing, evaluations of increasing vancomycin MICs need to be done at the institutional level and, as we discuss below, for individual patients with serious MRSA infections for whom vancomycin therapy is considered.
Although GISA and VRSA remain rare, cases of glycopeptide treatment failure are common, both in the literature and anecdotally among clinicians. Recent studies demonstrate increased clinical failure of vancomycin therapy in MRSA infections in which the isolates have increased MICs but are still susceptible [27–29]. In one study in which the selection process enriched the population for vancomycin treatment failures, vancomycin was <10% successful in the treatment of bacteremia caused by MRSA, with vancomycin MICs of 1–2 µg/mL, compared with 56% successful when the vancomycin MIC was <0.5 µg/mL [27]. In the same study, decreased killing by vancomycin in vitro was associated with higher likelihood of treatment failure. This association of decreased in vitro killing with decreased clinical efficacy of vancomycin in the treatment of MRSA bacteremia has been confirmed [28]. Furthermore, treatment with vancomycin within the previous 30 days is predictive of a higher MIC and inferior therapeutic efficacy of vancomycin [29].
In a more recent study comparing 51 patients with a variety of infections caused by MRSA with a vancomycin MIC of >2 µg/mL with 44 patients with infections due to MRSA with a MIC of <2 µg/mL, response was significantly lower (62% vs. 85%; P=.02 ) and infection-related mortality was higher (24% vs. 10%) in the high MIC group. In addition, a high MIC for vancomycin was an independent predictor of poor response in multivariate analysis of these MRSA infections [30]. The most recent study evaluating vancomycin efficacy in MRSA bacteremia by vancomycin MIC showed a significantly higher mortality for this disease when vancomycin was used empirically and the vancomycin MIC was 2 µg/mL [31].
Many strains of MRSA with vancomycin MICs of <4 µg/mL demonstrate heteroresistance to vancomycin (so-called hetero-resistant VISA [hVISA] or heteroresistant GISA [hGISA]) [32]. Table 2 provides a reference for the vancomycin susceptibility of VISA with respect to that of other S. aureus strains. Hetero-resistance is associated with poor clinical outcome [32]. Unfortunately, detection of hVISA among clinical S. aureus isolates is a task that may be insurmountable by current standard methods, which employ inoculum levels that may be too low and incubation periods too short to detect more slow-growing vancomycin-resistant subpopulations. Recent attempts to identify hVISA by using both macro-Etest (AB Bio-disk) and brain heart infusion (BHI) screening plates containing vancomycin at 6 µg/mL demonstrated low sensitivity [34].
Although detection of hVISA is problematic, the presence of hVISA isolates among clinical MRSA isolates has been documented [32], and the frequency of the hVISA isolates increases as the vancomycin MIC increases, even within the susceptible range [35]. A recent survey noted that the percentage of hVISA isolates increased from 57% to 81% to 100% as the vancomycin MIC increased from 1 to 2 to 4 µg/mL, respectively [35]. It should be noted, however, that this is the highest prevalence of heteroresistance detected to date. It is also well documented that infections by hVISA are highly predictive of vancomycin treatment failure [32, 36]. For example, a recent survey of 25 bacteremic infections due to hVISA showed a 76% treatment failure rate with vancomycin [37]. When this observation is extrapolated further, an increased frequency of hVISA isolates among clinical MRSA isolates that are found to be susceptible by testing in the clinical microbiology laboratory, with vancomycin MICs of 1 and 2 µg/mL, may account for the decline in vancomycin efficacy in the treatment of these infections. Therefore, until methods to reliably detect hVISA become available to clinical microbiology laboratories, vancomycin use for MRSA infections caused by isolates with vancomycin MIC of >1 µg/mL should be undertaken with caution. In addition, this confirms the wisdom of the recent Clinical and Laboratory Standards Institute (CLSI) decision to lower the breakpoint for vancomycin susceptibility to 2 µg/mL and points to the need to consider further modifications of glycopeptide susceptibility breakpoints.
Although reduced susceptibility to glycopeptides has been linked to suboptimal clinical response in MRSA bacteremia, this microbiological phenomenon may also relate to consistent trends (albeit not statistically significant differences) of inferior results of vancomycin therapy when compared with the results of therapy with the newer antistaphylococcal agents in the treatment of soft-tissue infection and pneumonia [38–43]. Although many of these differences were not statistically significant, consistent demonstration of either slower or inferior outcome with vancomycin treatment may be warning signs that vancomycin is no longer the most effective antimicrobial therapy for management of MRSA infections.
The use of high-dose glycopeptides has been considered a means to increase efficacy or prevent development of resistance. Vancomycin activity has been thought to be related to area under the concentration (AUC) curve for 24 h divided by the MIC (AUC0–24/MIC), suggesting that an increase in dose should increase efficacy, with an AUC0–24/MIC of >400 µg-h/mL being predictive of a favorable clinical outcome [44, 45]. This has led to a recommendation that the vancomycin trough be maintained at 15–20 µg/mL, rather than the lower concentrations previously targeted, in treatment for patients with severe hospital-acquired pneumonia due to MRSA [46]. More recent work suggests that even higher concentrations of vancomycin may be required, corresponding to a free AUC0–24/MIC of >500 µg-h/mL. Under the assumption of 50% protein binding, this correlates to a total AUC0–24/MIC of >1000 µg-h/mL [47]. However, a recent retrospective analysis of 102 patients with health care–associated pneumonia found that neither mean vancomycin trough concentration nor estimated AUC was linked to survival [48]. In addition, a recent case-control study demonstrated that, whereas achieving a trough serum concentration of at least 4 times the MIC was associated with a trend toward improved treatment efficacy, response rates were significantly lower for patients with MRSA strains with a vancomycin MIC of 2 µg/mL [30]. These observations lead to further questioning of the vancomycin susceptibility breakpoint of 2 µg/mL and suggest that escalation of vancomycin doses for infections due to such strains will not yield therapeutic benefits. Furthermore, recent data (although preliminary) have emerged that suggest that the administration of vancomycin in higher doses may have potential adverse consequences [49, 50]. Further studies will be paramount to evaluate this issue if vancomycin is to remain a first-line antimicrobial agent against MRSA infection.
Although data suggest that higher-dose vancomycin may not be the answer to improving efficacy, recent work suggests that maintaining a minimum free AUC0–24/MIC by maintaining higher serum troughs may prevent increases in vancomycin MICs and subsequent emergence of hGISA and GISA [51]. For example, in one published case report, maintaining vancomycin serum trough concentrations of >10 mg/mL, despite months of administration, failed to select for hVISA [52].
Besides vancomycin, linezolid, daptomycin, and tigecycline are approved by the US Food and Drug Administration for the treatment of MRSA [53, 54]. Although the official CLSI microbiological definition of resistance to daptomycin has not been established for S. aureus, strains with daptomycin MICs of >2 µg/mL have been associated with clinical failure [55, 56]. Mutations in the genes mprF (encoding lysylphosphatidylglycerol synthetase), yycG (encoding sensor histidine kinase), and rpoB and rpoC (encoding β and β subunits of RNA polymerase, respectively) have been found in S. aureus with daptomycin MICs above the susceptible range. Mutations in mprF appear to occur early in the selection process, whereas those in rpoB and rpoC occur later. More work is needed to better define the exact mechanisms that allow these mutations and perhaps others that cause decreased microbiological activity of daptomycin [57].
Several reports have linked increases in vancomycin MIC to increases in daptomycin MIC [58–60], although no definitive mechanism has been elucidated to date. However, these increases in MIC may not be accompanied by mutations associated with higher daptomycin MICs, suggesting that thick cell walls induced by the glycopeptide impair the diffusion of daptomycin to target sites in the cell membrane [59, 61]. The clinical significance of these findings is unknown, but they suggest that continued utilization of vancomycin in situations where it may not work could pose consequences for newer, more potent agents.
Furthermore, the loss of daptomycin susceptibility observed in some patients in a trial evaluating daptomycin and a comparator in S. aureus bacteremia and endocarditis occurred in cases of infections that ideally should have been dealt with surgically [62]. Therefore, the importance of surgical reduction of bacterial inoculum in an infection when appropriate (e.g., draining of abscess, removal of infected biomedical devices, and devitalized tissue debridement) to prevent antimicrobial resistance cannot be emphasized enough.
Linezolid resistance in clinical MRSA isolates has been associated with mutations in the central loop of domain V of 23S ribosomal RNA (rRNA) (e.g., G2576T and T2500A) [8, 63]. Given 5 or 6 copies of rRNA sequences, an increased MIC has been noted to correlate with a higher fraction of copies containing mutations [63, 64]. Interestingly, a clinical isolate was found not only to develop mutations in more rRNA copies as the MIC increased but also to decrease the total rRNA copies from 6 to 5 as resistance increased [63]. More recently, a plasmid-mediated rRNA methyltransferase, which methylates the ribosome at position A2503, has been found to determine linezolid resistance in some S. aureus strains from animals. The phenotype conferred by this methyltransferase has been labeled PhLOPSA (phenicols, lincosamides, oxazolidinones, pleuromutilins, and streptogrammin A) [65].
Of interest, although linezolid resistance among MRSA strains is extremely rare, a recent demonstration at one medical center of a linezolid MIC shift upward within the susceptible range among MRSA bloodstream isolates from 2001 to 2005 poses concern for future durability [66]. Furthermore, recent outbreaks of linezolid-resistant coagulase-negative staphylococci (mainly Staphylococcus epidermidis) at several medical centers further fuel such concerns [67, 68].
In summary, we have reviewed evidence that vancomycin susceptibility in MRSA is decreasing over time and in ways that are very difficult to detect using standard methods employed in hospital laboratories. Infections caused by vancomycin-susceptible MRSA organisms with MICs of >1 µg/mL appear to respond less effectively to vancomycin than do infections caused by organisms with MICs of <1 µg/mL. The use of high doses of vancomycin, as currently recommended by some authorities (e.g., the American Thoracic Society [46]) may not be sufficient to improve outcome. Although linezolid and daptomycin are available alternatives, resistance to these agents looms on the horizon. What is certain is that MRSA has adapted to glycopeptide exposure in ways that have made infections caused by these organisms very difficult to treat. Such adaptations not only allow for the development of resistance but also may select for virulence properties, such as resistance to endogenous peptides, that are important in the clearance of bacteremia [69]. Whether it is virulence, glycopeptide resistance, or both that is driving the changes in MRSA is unknown. As clinicians, we will continue to remain challenged by S. aureus infections, and our practices will need to adapt to this constantly evolving and formidable pathogen.
Dr. David DeVellis and Hilary Selby Polk provided assistance in editing the manuscript.
Supplement sponsorship. This article was published as part of a supplement entitled “Methicillin-Resistant Staphylococcus aureus: An Evolving Clinical Challenge,” sponsored by the Boston University School of Medicine and supported by an unrestricted educational grant from Cubist Pharmaceuticals, Inc.
Potential conflicts of interest. G.S. is on the speaker bureau of Cubist, Pfizer, Merck, and Wyeth Pharmaceuticals; has received research grant support from Cubist Pharmaceuticals; and has served as advisor/consultant to Cubist and Pfizer. R.C.M. has served as an advisor/consultant to Cubist Pharmaceuticals, Johnson & Johnson Ortho-Clinical Diagnostics, Pfizer, Targanta Therapeutics, Theravance, and Wyeth Pharmaceuticals.
IDSA Members: For your free access to this journal, log in via the IDSA members area.
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy