High-Dose Vancomycin Therapy for Methicillin-Resistant Staphylococcus aureus Infections: Efficacy and Toxicity

Background  Vancomycin hydrochloride treatment failure for infections caused by susceptible methicillin-resistant Staphylococcus aureus (MRSA) strains with high minimum inhibitory concentration (MIC) has prompted recent guidelines to recommend a higher vancomycin target trough of 15 to 20 μg/mL.

Methods  A prospective cohort study of adult patients infected with MRSA was performed to determine the distribution of vancomycin MIC and treatment outcomes with vancomycin doses targeting an unbound trough of at least 4 times the MIC. The microbiology laboratory computer records were used to identify all patients from whom MRSA was isolated from August 1, 2004, through June 30, 2005. Primary outcome measures were clinical response, mortality, and nephrotoxicity. Patients were placed into subgroups based on target trough attainment and high vs low vancomycin MIC (≥2 vs <2 μg/mL) for efficacy and high vs low trough (≥15 vs <15 μg/mL) for nephrotoxicity analyses.

Results  Of the 95 patients in the study, 51 (54%) were infected with high-MIC strains and had pneumonia (77%) and/or bacteremia. An initial response rate of 74% was achieved if the target trough was attained irrespective of MIC. However, despite achieving the target trough, the high-MIC group had lower end-of-treatment responses (24/39 [62%] vs 34/40 [85%]; P = .02) and higher infection-related mortality (11/51 [24%] vs 4/44 [10%]; P=.16) compared with the low-MIC group. High MIC (P = .03) and Acute Physiology and Chronic Health Evaluation II score (P = .009) were independent predictors of poor response in multivariate analysis. Nephrotoxicity occurred only in the high-trough group (11/63 [12%]), significantly predicted by concomitant therapy with other nephrotoxic agents.

Conclusions  High prevalence of clinical MRSA strains with elevated vancomycin MIC (2 μg/mL) requires aggressive empirical vancomycin dosing to achieve a trough greater than 15 μg/mL. Combination or alternative therapy should be considered for invasive infections caused by these strains.

Methicillin-resistant Staphylococcus aureus (MRSA) has become a leading cause of infections in both the community and health care–related settings. Data from the 2004 Centers for Disease Control and Prevention National Nosocomial Infections Surveillance system indicate that the prevalence of MRSA now exceeds 50% in most hospitals in the United States.¹ Equally alarming is the recent emergence of MRSA strains in the community setting, causing infections that range from cellulitis with skin abscesses to pneumonia and endocarditis in otherwise healthy individuals.²

Vancomycin hydrochloride has been the accepted standard of therapy for MRSA infections.³ Newer agents with proven efficacy against MRSA infections (eg, linezolid, quinupristin-dalfopristin, daptomycin, and tigecycline) are available but have not been routinely prescribed because of higher drug acquisition costs and/or relative lack of clinical experience compared with vancomycin. However, treatment failures of vancomycin for MRSA infections have increasingly been reported in the literature despite apparent in vitro susceptibility, particularly for strains with a minimum inhibitory concentration (MIC) of 2 μg/mL.^4,5 In an attempt to reconcile this discrepancy, the Clinical andLaboratory Standards Institute (CLSI) lowered the vancomycin breakpoint for susceptibility from 4 to 2 μg/mL for S aureus in 2006.^6,7

Achievement of serum unbound vancomycin trough concentrations at 4 to 5 times the MIC or 24-hour area under the concentration curve to MIC ratio of 400 was shown to be optimal for bacterial eradication and clinical success.^8-11 On the basis of 50% protein binding, a target vancomycin trough of 15 to 20 μg/mL will need to be achieved for MRSA strains with MIC of 2 μg/mL.^11,12 In addition, only 20% to 30% of vancomycin serum concentration is achieved in lung tissue, thus requiring even higher doses to treat pneumonia.¹³ A recently published guideline from the American Thoracic Society advocates that vancomycin be given in doses to achieve a trough of 15 to 20 μg/mL for MRSA pneumonia,¹⁴ considering the emergence of MRSA strains with reduced susceptibility to vancomycin, pharmacokinetic and pharmacodynamic properties of the drug,^8-13 and anecdotal clinical experience. However, to date to our knowledge, no clinical studies have definitively assessed the efficacy and safety of vancomycin in the treatment of MRSA infections at a dose targeting a serum trough of 15 to 20 μg/mL for strains with vancomycin MIC of 2 μg/mL. The purposes of this study are to determine the distribution of vancomycin MICs among MRSA isolates that cause infections and to evaluate the efficacy of vancomycin and risk of nephrotoxicity associated with vancomycin when given in doses to achieve an unbound target trough of 4 or more times the MIC of the infected strain.

A prospective cohort study of adult patients infected with MRSA was conducted at Huntington Hospital, a 525-bed community teaching hospital in Pasadena, Calif. The study protocol was approved by the hospital's institutional review board and informed consent was not required, since confidentiality was guaranteed and no interventions were performed. The microbiology laboratory computer records were used to identify all patients from whom MRSA was isolated during August 1, 2004, through June 30, 2005. Inclusion criteria were (1) age of 18 years or older, (2) nosocomial infection defined as that acquired in a skilled nursing facility or 48 hours or more after hospital admission, and (3) receipt of vancomycin therapy for 72 hours or more. If MRSA had been isolated on multiple occasions within a 6-month period in the same patient, only the first episode of infection was reviewed. Attending physicians typically requested a pharmacist to dose vancomycin according to institution-approved protocols to achieve a trough of 4 to 5 times the MIC of the infected MRSA strain.

All S aureus isolates were identified by colony morphologic analysis, gram staining, and catalase and coagulase tests. Methicillin resistance was determined by testing against oxacillin sodium. Vancomycin MIC was determined with the Etest (0.016 to 256 μg/mL) (AB BIODISK, Solna, Sweden) according to the manufacturer's instructions. Reference strains ATCC (American Type Culture Collection) 29213 and ATCC 43300 were used for quality control. All plates were incubated at 35°C for 24 and 48 hours, and testing was performed in duplicate.

The medical and laboratory records of eligible patients were retrospectively reviewed using prospectively collected databases. Pertinent demographic, laboratory, radiographic, and clinical data were obtained and recorded on a structured data collection form, which included age, sex, place of residence before isolation of MRSA, comorbid conditions, white blood cell count with differential, complete metabolic panel, culture and sensitivities, daily vital signs, percentage of oxygen saturation, vancomycin dose and serum trough concentrations achieved, concurrent antibiotic regimen, and clinical response. Acute Physiology and Chronic Health Evaluation II (APACHE II) score was calculated for all patients at the time of admission to assess the severity of underlying illness. All data were compiled into a single database using Microsoft Access (Microsoft Corp, Redmond, Wash).

Infections were defined according to criteria established by the Centers for Disease Control and Prevention.¹⁵ Time to achieve clinical stability was defined as return of altered mental status and abnormal vital signs to normal baseline values (heart rate ≤100 beats/min, systolic blood pressure ≥90 mm Hg, respiratory rate ≤24/min, oxygen saturation ≥90%, and temperature ≤37.2°C [99°F]). Response included complete and partial responses. A complete response was defined as resolution of fever, leukocytosis, and local signs of infections; a partial response was improvement of these conditions. Nonresponse included failure, relapse, and death. Failure was defined as no improvement or worsening of signs and symptoms of infections; relapse was recurrence of infection with the same organism at any body site within 1 month after discontinuation of therapy. Nephrotoxicity was defined as an increase of 0.5 mg/dL (44.2 μmol/L) or 50% or more of baseline serum creatinine level in 2 consecutive laboratory tests. Strains with vancomycin MICs of 1.5 or 2 μg/mL and those with MICs of 0.5, 0.75, or 1 μg/mL were placed into high- and low-MIC subgroups, respectively.

Primary outcome measures were clinical response, nephrotoxicity, and infection-related mortality. Secondary outcome measures included time to achieve clinical stability and length of hospital stay. Initial (72 hours after start of therapy) and final (end of therapy) responses were assessed. Patients were placed into subgroups based on vancomycin MIC of infected strains and trough levels attained. For efficacy analysis, patients were compared by MIC of infected strain and by target trough attainment. For each patient, a corrected average vancomycin trough was calculated as the target trough attained using the sum of each measured trough concentration multiplied by the number of days at that level, then divided by the total number of treatment days. For safety analysis, patients were placed into high- (≥15 μg/mL) vs low- (<15 μg/mL) trough subgroups and compared for the incidence of nephrotoxicity.

GraphPad Prism, version 4.0 (GraphPad Software, Inc, San Diego, Calif), and SPSS, version 13.0 (SPSS Inc, Chicago, IL), were used to perform statistical analyses. Case and control group variables were compared using the t test, Mann-Whitney U test, χ² test, or Fisher exact test where applicable. All statistical tests were 2-tailed; P≤.05 denoted statistical significance. Logistic regression was used to identify associations between the independent variables tested and the dichotomous dependent variables in the multivariate analysis.

Evaluable patients (n = 95) represented an elderly population (mean ± SD age, 72.5 ± 15.6 years), with more than half (61/95 [64%]) residing in a skilled nursing facility before hospital admission (Table 1). Many had moderately severe underlying illness (mean ± SD APACHE II score, 15.2 ± 7.8) and at least 2 comorbid conditions, such as cardiovascular disease and diabetes mellitus.

Most isolates were obtained from the respiratory tract (49/95 [52%]) and blood (24/95 [25%]) followed by wound (20/95 [21%]) and urine (2/95 [2%]). Of note, among patients with bacteremia, 20 cases were catheter related and 4 patients had endocarditis. Coinfection with Pseudomonas aeruginosa (n = 4) and/or Enterobacteriaceae (n = 10) was present in 13 (28%) of 39 patients with pneumonia.

Fifty-one (54%) of the 95 isolates had high vancomycin MICs. A greater proportion of isolates with high MICs were involved in bloodstream infections compared with low-MIC strains (18/51 [35%] vs 6/44 [14%]; P = .02; odds ratio, 3.5; 95% confidence interval, 1.2-9.7), whereas the opposite was observed for wound isolates (8/51 [16%] vs 12/44 [27%]; P=.21). No significant demographic differences were observed between patients who were infected with high- vs low-MIC strains (Table 1).

Eighty-six patients (91%) received vancomycin monotherapy; 9 (9%) received combination therapy with linezolid (n = 4), rifampin (n = 3), quinupristin-dalfopristin (n = 1), or trimethoprim-sulfamethoxazole (n = 2) empirically. Six patients who received monotherapy initially had the addition of a second agent (mean ± SD, day 5.5 ± 2.3 of therapy). Vancomycin therapy was discontinued after 72 hours in 9 patients; thus, only 86 patients were included in the final assessment. Of note, all patients who had concurrent infections with another organism received effective therapy with other agents (data not shown).

For those who attained target trough, overall response rates of 76% (52/68) and 73% (58/79) to a vancomycin-containing regimen were observed at the initial and final time points, respectively (Figures 1 and 2). No difference in initial response was observed if the target trough was achieved irrespective of MIC (32/41 [78%] in the low-MIC group vs 20/27 [74%] in the high-MIC group). However, a 20% lower response rate was observed for those who did not achieve the target trough initially (52/68 [76%] vs 15/27 [56%]; P = .05), which was most notable among those with high-MIC strains. At final assessment, the high-MIC group was significantly less responsive to vancomycin compared with the low-MIC group (24/39 [62%] vs 34/40 [85%]; P = .02) despite achieving the target trough. As expected, those with invasive infections (pneumonia or bloodstream) were also less responsive compared with those with noninvasive infections (wound or urine) (48/73 [66%] vs 20/22 [91%]; P = .03).

A trend toward fewer patients (71% vs 84%) achieving clinical stability, prolonged hospital stay by 3.5 days, and higher infection-related mortality (24% vs 10%) was observed among those infected with high- compared with low-MIC strains (Table 2). In a univariate analysis, advanced age, higher APACHE II score, infection caused by high-MIC strains, and initial trough less than 4 times the MIC were significantly associated with vancomycin treatment failure (Table 3). The MIC of the infected strains (high vs low) and APACHE II score (P = .03 and .009, respectively) were independent predictors of poor treatment response when controlled for age, intensive care unit admission, site of infection, and target trough attainment using logistic regression in a multivariate analysis.

Sixty-three (66%) of the 95 patients attained a high corrected average vancomycin trough of 15 to 20 μg/mL, of whom 11 (12%) developed nephrotoxicity compared with none in the low-trough group (P = .01). Age, APACHE II score, admission to the intensive care unit, or duration of vancomycin therapy (median, 12 days; interquartile range, 7-16 days) did not differ between those who attained high vs low trough levels (Table 4). However, slightly higher serum creatinine values at baseline, peak, and before discharge were observed with the high- compared with low-trough group (baseline: P=.20; peak: P=.10; and before: P=.10).

Among those who developed nephrotoxicity (n = 11), 10 received concurrent therapy with an aminoglycoside or amphotericin B, and 4 had preexisting renal disease. Concomitant nephrotoxic agents (P<.001), high vancomycin trough level (P = .03), and duration of vancomycin therapy (P=.004) appear to significantly predict nephrotoxicity (Table 5). By controlling for age, admission to intensive care unit, APACHE II score, vancomycin trough levels, and duration of therapy in a multivariate analysis, concomitant nephrotoxic agents remain the most significant predictor of the development of nephrotoxicity (P = .003). In addition, an incremental increase in the risk of nephrotoxicity was associated with duration of therapy at high trough levels (15-20 μg/mL): 6.3% for 7 days or less, 21.1% for 8 to 14 days, and 30% for more than 14 days. In a subanalysis that included only patients without receipt of concomitant nephrotoxic agents, nephrotoxicity occurred in only 1 (2%) of 44 high-trough vs 0 of 24 low-trough patients.

Vancomycin treatment failures have increasingly been reported with infections caused by vancomycin-susceptible MRSA strains with relatively high MICs (ie, 2 μg/mL).^4,5 This observation has prompted a change in the breakpoint for vancomycin susceptibility from 4 to 2 μg/mL for S aureus per CLSI⁷ and expert recommendation to administer higher doses of vancomycin targeting trough levels of 15 to 20 μg/mL when treating MRSA pneumonia.¹⁴ However, the prevalence of MRSA clinical strains with high vancomycin MICs and whether targeting higher trough concentrations increases the efficacy of vancomycin and/or risk of nephrotoxicity remain uncertain. Thus, we performed a prospective cohort study that involved an elderly population with nosocomial MRSA infections to address these uncertainties.

Using the Etest to determine vancomycin susceptibility, we found that 54% of the nosocomial MRSA strains that caused infections had an MIC of 2 μg/mL. The predominance of high-MIC strains in our cohort suggests that invasive strains may be associated with high MIC because 77% of our strains caused pneumonia and/or bloodstream infections. Additionally, the differences in MICs between our strains and those reported in large surveillance studies may be attributed to the different susceptibility testing methods used.¹⁶ Compared with the standard broth microdilution test and automated tests using VITEK (BioMerieux, Durham, NC) and MicroScan (Dade Behring Inc, Deerfield, Ill), the Etest was shown to be more sensitive and specific in identifying MRSA strains with reduced susceptibility to vancomycin.¹⁷ One center identified an increase in the prevalence of strains that showed heterogeneous resistance to vancomycin (hVISA) from 1.6% in 1998 to 36% in 2001 among 256 MRSA strains obtained from blood, respiratory, and cerebrospinal fluid sources from hospitalized patients.¹⁸ Interestingly, 44 (97%) of the 46 hVISA strains identified in that study had vancomycin MICs of 2 μg/mL or less as determined by broth microdilution, which would be considered “susceptible” per current CLSI interpretive criteria. Others have also found that well-characterized hVISA strains have an MIC range that overlaps with the currently defined susceptible range.^16,19 Thus, it is possible that MRSA strains in our cohort may contain subpopulations of hVISA strains. Additional studies that include population analysis profiling are under way to examine the hVISA phenotype in our cohort.

In this study, vancomycin therapy was given in doses to attain an unbound trough target of at least 4 times the MIC of the infected strain. An overall response rate of more than 70% observed in our cohort is comparable to published response rates in clinical trials that involved hospitalized adults with MRSA infections.^20,21 Consistent with the positive correlation between in vitro bactericidal activity and vancomycin trough concentration at 4 to 5 times the MIC,^8,9,12 we found that patients who achieved target trough levels within 72 hours of therapy had a 20% higher response rate than those who did not. However, despite favorable initial responses for patients who attained target trough levels, response rates assessed at the end of therapy were significantly lower for patients infected with strains having an MIC of 2 μg/mL compared with 1 μg/mL or less (85% vs 62%). Both high vancomycin MIC (2 μg/mL) and severity of underlying illness were found to be independent predictors of poor treatment response to vancomycin, with risks of 6.02 and 3.14, respectively, after controlling for potential confounders in a multivariate analysis.

The correlation between elevated vancomycin MIC and treatment failure in our cohort is consistent with the findings of 2 published studies,^4,5 which analyzed data from patients who had participated in a large multicenter phase 3 and 4 prospective study. Vancomycin failure rates of 22%, 27%, and 51% were observed for patients infected with MRSA strains that had MICs of 0.5, 1.0, and 2.0 μg/mL, respectively.⁵ A related study⁴ analyzing data on a subset of bacteremic patients stated that vancomycin was given in doses to achieve a trough concentration of 10 to 15 μg/mL. The same study demonstrated reduced vancomycin bactericidal activity in vitro for susceptible strains with a higher vancomycin MIC.⁴ Notably, severity of underlying illness and the presence of strains with the hVISA phenotype were unknown, both of which could affect treatment responses. Infections caused by hVISA have been associated with suboptimal response to vancomycin therapy and were characterized by the presence of high bacterial load infection, persistent fever, and bacteremia for more than 7 days after the commencement of vancomycin therapy.¹⁹ In our cohort, high MIC was a significant predictor of end-of-therapy vancomycin failure despite the achievement of target trough levels. Initial eradication of susceptible strains, leaving subpopulations of hVISA as the predominant population, may have accounted for the persistence of infection and therefore “late” failure in our patients.

Of interest, in a subgroup of our patients (n = 15) who received vancomycin in combination with linezolid, daptomycin, and/or rifampin for infections caused by high-MIC strains, the end-of-treatment response appeared more favorable when compared with vancomycin alone regardless of whether the target trough was achieved. However, no specific combination was more effective than another. Our findings need confirmation with larger sample sizes. In another subset of patients infected with high-MIC strains in whom vancomycin therapy was switched to another agent (daptomycin [n = 6], linezolid [n = 3], and rifampin [n = 6]) because of suboptimal clinical response after a week of vancomycin therapy, 12 (80%) of the 15 patients responded favorably to nonvancomycin therapy, whereas 3 (20%) failed treatment and eventually died of sepsis.

Nephrotoxicity attributable to vancomycin was observed at a 12% incidence compared with none in the high- and low-trough groups, respectively. Duration of therapy increases the risk of nephrotoxicity from 6% to 21% for patients receiving high-dose therapy when treatment extends beyond 1 week and up to 30% for those receiving more than 2 weeks of treatment. Vancomycin therapy with concomitant nephrotoxic agents is the single independent predictor of nephrotoxicity in a multivariate analysis controlling for age, duration of therapy, and trough levels achieved (<15 vs ≥15 μg/mL). The incidence of nephrotoxicity observed in our study is consistent with results reported from other studies. Rybak et al²² noted a 22% incidence of nephrotoxicity in patients who received vancomycin in combination with another nephrotoxic agent (eg, gentamicin sulfate) compared with 5% with vancomycin alone.

Methicillin-resistant S aureus isolates that showed reduced susceptibility to vancomycin (with an MIC at the breakpoint for susceptibility) are highly prevalent among strains that cause invasive infections. Suboptimal response associated with vancomycin therapy not attaining the trough target underscores the importance of aggressive initial dosing to achieve a trough of 15 μg/mL or more until culture and sensitivity results are known. Vancomycin in combination with another agent or alternative treatment options may be appropriate for invasive infections caused by MRSA strains with an MIC of 2 μg/mL as determined by Etest, particularly in patients receiving concomitant nephrotoxic agents. An improved method for detection of hVISA strains is urgently needed. Their prevalence and associated clinical implications will need to be determined to help guide treatment decisions.

Correspondence: Annie Wong-Beringer, PharmD, University of Southern California, School of Pharmacy, 1985 Zonal Ave, Los Angeles, CA 90033 (anniew@usc.edu).

Accepted for Publication: July 13, 2006.

Author Contributions: Dr Wong-Beringer had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Wong-Beringer. Acquisition of data: Hidayat and Hsu. Analysis and interpretation of data: Hidayat, Hsu, Quist, Shriner, and Wong-Beringer. Drafting of the manuscript: Hidayat and Wong-Beringer. Critical revision of the manuscript for important intellectual content: Hidayat, Shriner, and Wong-Beringer. Statistical analysis: Quist. Obtained funding: Wong-Beringer. Administrative, technical, and material support: Hidayat and Shriner. Study supervision: Wong-Beringer.

Financial Disclosure: None reported.

Funding/Support: This study was supported in part by a research grant from Pfizer Inc.

Previous Presentation: This study was presented as a poster (abstract 352) at the 43rd Annual Meeting of the Infectious Diseases Society of America; San Francisco, Calif; October 6-9, 2005.

Acknowledgment: We thank AB Biodisk for providing vancomycin Etest strips for MIC determination.