Abstract
Given the critical importance of early appropriate therapy, a retrospective cohort (2002 to 2013) was performed at the Detroit Medical Center to evaluate the association between the day 1 vancomycin exposure profile and outcomes among patients with MRSA infective endocarditis (IE). The day 1 vancomycin area under the concentration-time curve (AUC0–24) and the minimum concentration at 24 h (Cmin 24) was estimated for each patient using the Bayesian procedure in ADAPT 5, an approach shown to accurately predict the vancomycin exposure with low bias and high precision with limited pharmacokinetic sampling. Initial MRSA isolates were collected and vancomycin MIC was determined by broth microdilution (BMD) and Etest. The primary outcome was failure, defined as persistent bacteremia (≥7 days) or 30-day attributable mortality. Classification and regression tree analysis (CART) was used to determine the vancomycin exposure variables associated with an increased probability of failure. In total, 139 patients met study criteria; 76.3% had right-sided IE, 16.5% had left-sided IE, and 7.2% had both left and right-sided IE. A total of 89/139 (64%) experienced failure by composite definition. In the CART analysis, failure was more pronounced in patients with an AUC0–24/MIC as determined by BMD of ≤600 relative to those with AUC0–24/MIC as determined by BMD of >600 (69.8% versus 54.7%, respectively, P = 0.073). In the logistic regression analysis, an AUC/MIC as determined by BMD of ≤600 (adjusted odds ratio, 2.3; 95% confidence interval, 1.01 to 5.37; P = 0.047) was independently associated with failure. Given the retrospective nature of the present study, further prospective studies are required but these data suggest that patients with an AUC0–24/MIC as determined by BMD of ≤600 present an increased risk of failure.
INTRODUCTION
Vancomycin remains the mainstay of antimicrobial therapy for severe infections caused by methicillin-resistant Staphylococcus aureus (MRSA), including bacteremia and infective endocarditis (IE). Although vancomycin is considered first-line therapy, failure rates with vancomycin for the treatment of IE are exceedingly high (1). Reasons for these high failure rates have not been well defined. The high failure rates may simply be a function of patients overall poor prognosis. However, conditions at the site of infection (e.g., high bacterial burden within the cardiac vegetation and impaired immune response) and the growing prevalence of reduced vancomycin susceptibility among MRSA may also contribute to the increased reports of vancomycin treatment failure (2–6).
While host- and pathogen-related factors likely drive suboptimal treatment outcomes associated with vancomycin for MRSA IE, the contribution of the initial vancomycin exposure profile has not been well defined. The recent Infectious Diseases Society of America, American Society of Health-System Pharmacists, and Society of Infectious Diseases Pharmacists vancomycin consensus guidelines concluded that the pharmacokinetic/pharmacodynamic (PK/PD) target for vancomycin is an area under the concentration-time curve to MIC (AUC/MIC) ratio of 400 and recommended maintaining vancomycin trough serum concentrations of 15 to 20 mg/liter as a surrogate for this PK/PD index for complicated infections due to MRSA, including IE (7). However, there are limited clinical data in support of this target or the trough-guided dosing approach, especially in patients with IE. Furthermore, there are few published clinical vancomycin exposure-outcomes evaluations among patients with IE due to MRSA. In light of this gap in the literature, a retrospective cohort (2002 to 2013) was performed at the Detroit Medical Center to evaluate the relationship between the day 1 vancomycin exposure profile and outcomes among patients with MRSA IE. We purposefully evaluated the association between the day 1 vancomycin exposure profile and outcomes given the importance of early, appropriate therapy (8) and since the animal (9) and in vitro pharmacodynamics studies (10) used to define the AUC/MIC exposure target for vancomycin assessed the relationship between exposure and bacterial killing during the first 24 h of therapy. To accomplish the study objectives, a validated Bayesian approach was used to estimate the vancomycin exposure profile with limited PK samples (11). This approach shown to accurately predict the vancomycin exposure with low bias and high precision with limited pharmacokinetic sampling data.
(An abstract containing part of this study has been accepted for presentation in platform format at the 23rd European Congress of Clinical Microbiology and Infectious Diseases in Berlin, Germany, on 29 April 2013.)
MATERIALS AND METHODS
Study population.
This was a retrospective observational cohort study, conducted from January 2002 to July 2013 at the Detroit Medical Center. The study was reviewed and approved by the Wayne State University Human Investigational Review Board. Adult patients ≥18 years old who received at least 72 h of vancomycin therapy in response to a MRSA blood culture and who had at least one vancomycin concentration during the initial 72 h of vancomycin treatment were eligible for the present study. To be included, the MRSA blood culture had to meet the CDC criteria for a bloodstream infection (BSI) (12). Furthermore, the MRSA BSI had to be classified as either definite or possible IE by the treating physician according to modified Duke criteria (13). Patients were excluded if they received renal placement (e.g., hemodialysis) or if the MRSA isolate was not available for additional susceptibility testing.
Patient and treatment data.
Data collection included patient characteristics, the presence of comorbid conditions (e.g., diabetes, renal disease), the Acute Physiology and Chronic Health Evaluation (APACHE) II score (14), and the Charlson comorbidity index (15) at the time of the first positive blood culture, and creatinine clearance (CLCR) was estimated according to the Cockcroft-Gault formula (16). Additional data collection included duration of bacteremia, course of antimicrobial therapy, vancomycin serum concentrations, length of hospital stay, and 30-day attributable mortality.
For each patient, the vancomycin area under the concentration-time curve (AUC) from 0 to 24 h (AUC0–24) and the minimum concentration at 24 h (Cmin 24) was estimated using the MAP-Bayesian procedure in ADAPT 5 as previously described based on vancomycin dosing schedule received and collected vancomycin serum trough concentration during the initial 72 h (11, 17). This approach has been validated to estimate AUC values with high precision and low bias using vancomycin trough concentrations only (11, 17). In short, the mean parameter vector and the variance-covariance matrix from a previously published two-compartment vancomycin population pharmacokinetic model (18) was used as the Bayesian prior in ADAPT 5 (19). The MAP-Bayesian procedure in ADAPT 5 was then used to estimate the Bayesian conditional posterior pharmacokinetics parameters for each patient using their dosing, vancomycin serum concentration, and CLCR data. Based on dosing received and Bayesian conditional posterior pharmacokinetics parameters, the minimum concentration at 24 h (Cmin 24) and area under the curve (AUC) at 0 to 24 h (AUC0–24) were estimated for each patient. The predictive performance of the MAP-Bayesian approach was assessed by comparing the predicted concentrations to the observed concentrations. Following estimation of each subject's individual day 1 vancomycin exposure profile, AUC0–24/MIC ratios were computed using both the BMD and Etest MIC for the initial MRSA blood culture isolate.
Microbiologic data.
MRSA isolates were retrieved from all patients' initial positive blood cultures and evaluated for extended microbiological assessment and verification at a central research laboratory (Anti-Infective Research Laboratory, Wayne State University). Vancomycin MIC was for each isolate was determined in duplicate by broth microdilution (BMD) and Etest methods performed in agreement with Clinical and Laboratory Standards Institute (CLSI) guidelines and according to manufacturer instructions (bioMérieux, Durham, NC), respectively (20). Isolates were screened by macro Etest and confirmed by modified population analysis profile (mPAP) for identification of heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) (21). mPAPs for each isolate was compared to the reference strain Mu3 (ATCC 700689) mPAP and were considered to be hVISA if the population analysis profile area under the curve ratio to Mu3 was ≥0.9 as previously described by Wootton et al. (22).
Outcomes.
The primary outcome of interest was failure, defined as persistent bacteremia for at least 7 days and/or death within 30 days from the initial positive blood culture attributable to MRSA IE. Death was considered attributable to MRSA if one of the following criteria were present: (i) blood cultures were positive for MRSA at the time of death, (ii) death occurred before the resolution of signs and symptoms of MRSA infection, (iii) IE was listed as the main diagnosis in discharge documents and there was no other explanation for death, or (iv) autopsy findings or death certificate indicated MRSA as a cause of death. Study data were collected and managed using REDCap (Research Electronic Data Capture, Vanderbilt University) electronic data capture tools hosted at Wayne State University (23).
Statistical analysis.
The vancomycin exposure variables considered in the analysis included AUC0–24/MICBMD, AUC0–24/MICEtest, and Cmin 24. These vancomycin exposure variables were modeled as both continuous and dichotomous data. Categorical variables were compared by chi-squared or Fisher exact test, and continuous variables were compared by using the Student t test or Mann-Whitney U test, as appropriate, comparing failure versus nonfailure and dichotomous vancomycin exposure breakpoint. Breakpoints in the distribution of continuous day 1 vancomycin exposure variables that maximized the difference in failure, the primary study endpoint, between the two resultant exposure groups were sought through classification and regression tree (CART) analysis. In CART, the minimum parent node was specified at 100 cases, and the terminal node was set at 30 cases. We also examined the relationship between a Cmin 24 of ≥15 mg/liter and failure given the recent expert guidelines recommendations (7, 24).
Multivariable logistic analyses were performed to determine the independent association between day 1 vancomycin exposure variables and failure while adjusting for confounding variables. Each vancomycin exposure variable was assessed in a separate logistic regression model. All baseline variables associated with failure in the bivariate analyses at a P ≤ 0.2 were included in the explanatory multivariable model at model entry and variables were from the model using a backward stepwise approach. All tests were two tailed, and P < 0.05 was considered statistically significant. SPSS Statistics, IBM SPSS software, version 22.0 (SPSS, Inc., Chicago, IL) was used for all calculations, and CART software (Salford Systems, San Diego, CA) was used for classification and regression tree analysis.
RESULTS
Study population.
A study flow diagram is described in Fig. 1; 142 adult patients with MRSA IE meeting all inclusion criteria were evaluated. Overall, the median age of this sample was 51 years (interquartile range [IQR] 44 to 58 years), and the median APACHE II score was 10 (IQR 7 to 15). All patients received vancomycin as initial therapy for their MRSA IE, an infectious diseases consult was completed in 115 (81.0%), and 64 (45.1%) were admitted to an intensive care unit. Of the 142 patients, 109 (76.8%) had right-sided IE, 23 (16.2%) had left-sided IE, and 10 (7.0%) patients had bilateral IE. One hundred 35 (95.1%) patients had signs and symptoms of infection present on admission, suggesting community-acquisition of IE. The median duration of total vancomycin therapy during hospitalization was 10 days (IQR 5 to 16 days). Concomitant antimicrobial therapy was observed in 51 (35.9%) patients, with the most common agents coadministered, including gentamicin (25 [49.0%]) and rifampin (8 [15.7%]).
FIG 1.
Study population flow chart. MRSA, methicillin-resistant S. aureus; IE, infective endocarditis; VAN, vancomycin; AUC, area under the concentration 24-h time curve; BMD, broth microdilution.
Among the MRSA isolates, the MICBMD distributions were as follows: 16.9% were 0.5 mg/liter, 69.0% were 1 mg/liter, 13.4% were 2 mg/liter, and 0.7% were 4 mg/liter. Etest MIC distributions were as follows: 0.7% were 0.38 mg/liter, 2.8% were 0.5 mg/liter, 9.9% were 0.75 mg/liter, 24.6% were 1 mg/liter, 36.6% were 1.5 mg/liter, 23.2% were 2 mg/liter, 1.4% were 3 mg/liter, and 0.7% were 4 mg/liter. The modal vancomycin MICBMD and MICEtest were 1 mg/liter for 98 (69.0%) and 1.5 mg/liter for 52 (36.6%) isolates, respectively. One (0.7%) isolate was considered vancomycin-intermediate susceptible S. aureus (VISA) according to the CLSI reference method, having a vancomycin BMD MIC of 4 mg/liter. Twenty-one (14.8%) isolates were identified as hVISA by mPAP.
Observed and predicted vancomycin serum concentrations from the Bayesian estimation approach are shown in Fig. 2. The regression line from the observed-predicted plot was 0.96x predicted + 0.62 with an R2 of 0.970. The final mean percent coefficents of variation (CV%) for the parameter estimates after the Bayesian step are shown in Table S1 in the supplemental material. The CVs surrounding the PK estimates are consistent with the clustering of trough concentrations among the patients. Median vancomycin calculated AUC0–24 from Bayesian approach was 526.9 (IQR 370.2 to 706.8). The median vancomycin AUC0–24/MICBMD and AUC0–24/MICEtest were 499.2 (IQR 331.6 to 763.0) and 358.8 (IQR 258.2 to 555.1), respectively. The median (IQR) vancomycin concentration at 24 h (Cmin 24) was 10.5 mg/liter (6.3 to 15.1 mg/liter). Thirty-six (25.9%) patients had a vancomycin Cmin 24 of ≥15 mg/liter, with only 19 (13.7%) achieving the targeted range of 15 to 20 mg/liter. Three individual cases were deemed as outliers from the Bayesian approach. These individuals had AUC0–24 of >1,300, which was ∼2.5 standard deviations from the mean AUC0–24 value. These outliers were censored for further analysis providing a final sample of 139 patients for evaluation as described in Fig. 1.
FIG 2.
Observed versus predicted vancomycin serum concentration for Bayesian estimation approach.
A total of 89/139 (64.0%) patients experienced failure according to the predefined composite definition. The frequencies of the components of the composite definition were as follows: 81 (58.3%) had persistent bacteremia (≥7 days of bacteremia) and 26 (18.7%) had 30-day attributable mortality. The median total hospital length of stay was 17 days (IQR 12 to 29 days) and median of 7 days (IQR 5 to 10 days) of bacteremia was observed.
Failure analyses.
Comparison of baseline characteristics between failure and nonfailure patients are displayed in Table 1. Left-sided IE was higher in the failure group, although this was not statistically significant (27% versus 18%; P = 0.233). In addition, the percentage of surgical intervention was higher in the failure group relative to the nonfailure group (13.5% versus 2%; P = 0.032). No difference in median (interquartile range) in mean AUC0–24/MICBMD, AUC0–24/MICEtest, and Cmin 24 were noted between patients that failed relative to those that did not fail. Similarly, failure was similar among patients with Cmin 24 ≥ 15 mg/liter relative to those with Cmin 24 < 15 mg/liter. In the CART analysis, a significant breakpoint in the day 1 exposure variables was only identified for AUC0–24/MICBMD. Failure was more pronounced in patients with an AUC0–24/MICBMD ≤ 600 relative to those with AUC0–24/MICBMD > 600 (69.8% versus 54.7%, respectively, P = 0.073) (Table 1 and Fig. 3). Persistent bacteremia was the primary component of the failure definition that led to the difference in overall failure between the CART-derived exposure groups (Fig. 3.). Comparison of baseline characteristics between the CART-derived vancomycin AUC0–24/MICBMD are shown in Table 2. Patients that had an AUC0–24/MICBMD > 600 were more likely to have a higher APACHE II score, admitted to the intensive care unit (ICU), received a vancomycin loading dose, and had a higher Cmin 24. Of note, a breakpoint of 290 was delineated for the AUC0–24/MICEtest variable, but the CART-derived AUC0–24/MICEtest threshold for failure was not different at a P value of <0.05 (Table 1).
TABLE 1.
Baseline demographic and clinical features of MRSA IE associated with vancomycin treatment failure
| Characteristics | No. (%) of patientsa |
P | |
|---|---|---|---|
| Failure (n = 89) | Nonfailure (n = 50) | ||
| Age (yr)* | 54 (47–58) | 49.5 (36–58) | 0.045 |
| Wt (kg)* | 71.3 (61.0–84.5) | 71.8 (63.5–79.5) | 0.672 |
| CLCR (ml/min)* | 55.8 (37.5–89.9) | 57.1 (40.5–94.0) | 0.480 |
| Male | 60 (67.4) | 33 (66.0) | 0.865 |
| Infection related | |||
| APACHE II* | 10 (7–14.5) | 9.5 (6–15.25) | 0.621 |
| ICU admission | 47 (52.8) | 14 (28.0) | 0.005 |
| ID consultation | 68 (76.4) | 44 (88.0) | 0.097 |
| Surgical intervention | 12 (13.5) | 1 (2.0) | 0.032 |
| Community-acquired IE | 84 (94.4) | 48 (96.0) | >0.99 |
| Left-sided IEb | 24 (27.0) | 9 (18.0) | 0.233 |
| hVISAc | 17 (19.1) | 3 (6.0) | 0.044 |
| VAN MICBMD of >1 mg/liter | 13 (14.6) | 7 (14.0) | 0.922 |
| VAN MICEtest of >1 mg/liter | 54 (60.7) | 31 (62.0) | 0.878 |
| Comorbidities | |||
| Charlson comorbidity index* | 2 (1–3) | 1 (0–3) | 0.113 |
| IDU | 56 (62.9) | 37 (74.0) | 0.183 |
| Diabetes | 19 (21.3) | 6 (12.0) | 0.168 |
| Heart disease | 12 (13.5) | 6 (12.0) | 0.586 |
| Acute renal failure | 33 (37.1) | 15 (30.0) | 0.400 |
| Liver disease | 29 (32.6) | 14 (28.0) | 0.575 |
| Other characteristics | |||
| Previous vancomycin | 40 (44.9) | 22 (44.0) | 0.914 |
| Previous antibiotic (other) | 20 (22.5) | 15 (30.0) | 0.326 |
| Previous hospitalization | 41 (46.1) | 32 (64.0) | 0.042 |
| Previous S. aureus infection | 13 (14.6) | 6 (12.0) | 0.668 |
| VAN AUC0–24/MICBMD* | 467 (313–706) | 565 (351–825) | 0.285 |
| VAN AUC0–24/MICEtest* | 359 (259–553) | 406 (257–557) | 0.703 |
| VAN AUC0–24/MICBMD ratio of ≤600 | 60 (67.4) | 26 (52.0) | 0.073 |
| VAN AUC0–24/MICEtest ratio of ≤290 | 33 (37.1) | 14 (28.6) | 0.313 |
| Received VAN LD | 26 (29.2) | 16 (32.0) | 0.731 |
| VAN LD (mg/kg)* | 22.3 (18.0–25.7) | 24.3 (20.3–27.3) | 0.331 |
| VAN Cmin 24 (mg/liter)* | 10.8 (6.9–15.3) | 10.4 (5.7–14.4) | 0.416 |
| VAN Cmin 24 of ≥15 mg/liter | 25 (28.1) | 11 (22.0) | 0.432 |
| Vancomycin duration (days)* | 9 (5–17) | 11 (6–16) | 0.575 |
| Concomitant therapyd | 33 (37.1) | 18 (36.0) | 0.899 |
Data are presented as numbers (percentages) of patients unless otherwise indicated. Data indicated by an asterisk (*) in column 1 are presented as medians (interquartile ranges). CI, confidence interval; CLCR, estimated creatinine clearance by Cockcroft-Gault; ICU, intensive care unit; ID, infectious diseases; VAN, vancomycin; AUC, area under the concentration 24-h time curve; BMD, broth microdilution; IE, infective endocarditis; IDU, injection drug user; hVISA, heterogeneous vancomycin-intermediate S. aureus; LD, loading dose.
Includes bilateral IE (right- and left-sided IE).
Includes VISA.
Concomitant antimicrobial therapy with vancomycin for MRSA IE, e.g., gentamicin or rifampin.
FIG 3.
Composite vancomycin treatment failure and comparing a vancomycin AUC0–24/MICBMD ratio of ≤600 or >600 by bivariate analyses. VAN, vancomycin; BMD, broth microdilution.
TABLE 2.
Clinical characteristics in patients with MRSA infective endocarditis
| Characteristic | No. (%) of patuents with a vancomycin AUC0–2/MICBMD ratio of: |
P | |
|---|---|---|---|
| ≤600 (n = 86) | >600 (n = 53) | ||
| Age (yr)* | 50 (43–57) | 54 (46–59) | 0.194 |
| Wt (kg)* | 71.1 (60.1–77.5) | 74.0 (63.5–83.6) | 0.140 |
| CLCR (ml/min)* | 59.1 (42.2–98.3) | 51.8 (34.6–87.2) | 0.118 |
| Male | 53 (61.6) | 40 (75.5) | 0.092 |
| Infection related | |||
| APACHE II* | 8 (6–12) | 13 (8–17) | 0.007 |
| ICU admission | 32 (37.2) | 29 (54.7) | 0.043 |
| ID consultation | 65 (75.6) | 47 (88.7) | 0.058 |
| Community-acquired IE | 81 (94.2) | 51 (96.2) | 0.708 |
| Left-sided IEb | 21 (24.4) | 12 (22.6) | 0.811 |
| hVISAc | 16 (18.6) | 4 (7.5) | 0.085 |
| Comorbidities | |||
| Charlson comorbidity index* | 1 (0–3) | 2 (1–5) | 0.059 |
| IDU | 59 (68.6) | 34 (64.2) | 0.588 |
| Diabetes | 14 (16.3) | 11 (20.8) | 0.505 |
| Heart disease | 11 (12.8) | 7 (13.2) | 0.943 |
| Acute renal failure | 25 (29.1) | 23 (43.4) | 0.084 |
| Liver disease | 26 (30.2) | 17 (32.1) | 0.819 |
| HIV | 6 (7.0) | 7 (13.2) | 0.220 |
| Other characteristics | |||
| Previous VAN | 35 (40.7) | 27 (50.9) | 0.238 |
| Received VAN LD | 18 (20.9) | 24 (45.3) | 0.002 |
| VAN loading dose* | 22.5 (20.3–25.0) | 24.8 (19.4–28.3) | 0.303 |
| VAN Cmin 24 (mg/liter)* | 8.4 (5.3–11.5) | 15.2 (10.5–20.9) | <0.001 |
| VAN Cmin 24 of ≥15 mg/liter | 9 (10.5) | 27 (50.9) | <0.001 |
| VAN duration (days)* | 10.5 (6–18) | 9 (4–15) | 0.231 |
Data are presented as the number (percentage) of patients unless otherwise indicated. Data indicated by an asterisk (*) in column 1 are presented as medians (interquartile ranges). ICU, intensive care unit; ID, infectious diseases; IDU, injection drug user; CLCR, estimated creatinine clearance by Cockcroft-Gault; ARF, acute renal failure; LD, loading dose; VAN, vancomycin; AUC, area under the concentration 24-h time curve; BMD, broth microdilution; hVISA, heterogeneous vancomycin-intermediate S. aureus.
Includes bilateral infective endocarditis (right- and left-sided IE).
Includes VISA.
Based on the results of the bivariate analyses, the AUC0–24/MICBMD was the only day 1 vancomycin exposure variable considered in the logistic regression analyses since no notable associations were noted between AUC0–24/MICEtest and failure and Cmin 24 and failure in the bivariate analyses. The variables included in the logistic regression at model entry were hVISA, ICU admission, diabetes, infectious diseases consult, previous hospitalization within 90 days, injection drug user, age, and Charlson comorbidity score. In the final logistic regression model that considered the CART-derived vancomycin AUC0–24/MICBMD variable at model entry, vancomycin AUC0–24/MICBMD of ≤600 was still independently associated with failure (adjusted OR of 2.331, a 95% CI of 1.012 to 5.371, and P = 0.047) (Table 3). Other characteristics associated with failure in the final model included the presence of hVISA and ICU admission. The Hosmer-Lemeshow goodness-of-fit test was highly acceptable (P = 0.837) and model fit the data well (area under the receiver operating characteristic curve of 0.77). For completeness, we evaluated the relationship between AUC0–24/MICBMD, expressed as a continuous variable, and failure and no notable association was noted (data not shown).
TABLE 3.
Logistic regression analysis of predictors associated with vancomycin treatment failurea
| Factor | Odds ratio (95% CI) |
P | |
|---|---|---|---|
| Unadjusted | Adjusted | ||
| VAN AUC0–24/MICBMD ratio of ≤600 | 1.910 (0.939-3.885) | 2.331 (1.012–5.371) | 0.047 |
| Admission to ICU | 2.878 (1.367–6.058) | 3.572 (1.535–8.313) | 0.003 |
| hVISA | 3.699 (1.027–13.321) | 4.588 (1.154–18.242) | 0.031 |
| Previous hospitalization | 0.480 (0.236–0.979) | 0.370 (0.160–0.858) | 0.021 |
| IDU | 0.596 (0.278–1.281) | 0.422 (0.170–1.050) | 0.064 |
| Age | 1.032 (0.997–1.069) | 1.033 (0.999–1.069) | 0.057 |
VAN, vancomycin; AUC0–24, area under the concentration 24-h time curve; BMD, broth microdilution; ICU, intensive care unit; hVISA, heterogeneous vancomycin-intermediate S. aureus; IDU, intravenous drug user.
DISCUSSION
Consistent with several previously published vancomycin-exposure response evaluations (25–27), we found that the AUC0–24/MIC ratio, not Cmin 24, was the PK/PD parameter most closely linked to outcomes among patients with MRSA IE. We performed both broth microdilution (BMD) and Etest testing since both of these methods are used for vancomycin susceptibility testing by clinical microbiology laboratories in the field and because both have been used to evaluate patient outcomes as a function of AUC/MIC (26–28). In this investigation, we found that the risk of vancomycin failure was greatest among those with an AUC0–24/MICBMD ratio of ≤600, and this exposure-failure relationship persisted after adjusting for factors such as ICU admission, the presence of hVISA, and other comorbidities. Admission to the ICU and hVISA was also found to be independently associated with failure which has been previously reported (29–31). Of interest, previous hospitalization was associated with nonfailure. The reason for this finding is unknown, and it would be difficult for us to speculate based on the retrospective nature of the present study. Further investigation maybe necessary to confirm this relationship. The observed difference in the composite endpoint of clinical failure observed here was driven by persistent bacteremia, since mortality was not statistically different between AUC0–24/MICBMD groups. Of note, mortality in the study population was lower than that observed in other studies of MRSA IE, possibly owing to the predominance of right-sided IE. This likely hindered the ability of the analysis to identify an association between AUC0–24/MICBMD and infection-related mortality (data not shown). Furthermore, patients with an AUC0–24/MICBMD ratio of >600 were more likely to be severely ill (Table 2), which could further bias the observed association between vancomycin exposure and mortality (Table 2). As such, these data suggest that patients with an AUC0–24/MICBMD ratio of ≤600 are at an increased risk of failure due to persistent bacteremia and further studies in critically ill patients with MRSA IE are needed to definitively assess if an AUC0–24/MICBMD ratio of ≤600 contributes to greater risk of infection-related mortality.
On first glance, it appears the AUC0–24/MICBMD identified here is considerably higher than several recent clinical evaluations. It is important to note that most clinical evaluations performed to date that identified AUC0–24/MICBMD ≥ 400 as the critical PK/PD target for vancomycin used a simple formula based on daily vancomycin dose and estimated renal function to estimate AUC values (25, 27, 32). Based on the limitations associated with the Cockcroft-Gault approach and the considerable interpatient variability in vancomycin exposure profiles, it is difficult to derive an accurate individual AUC exposure profile based on glomerular filtration estimation alone (11, 33, 34). In most cases, the formula-based approach will overestimate vancomycin clearance by ∼40 to 50% as shown in a recent study by Lodise and coworkers (17). If one considers that AUC = dose/clearance, our daily AUC0–24/MICBMD target of 600 aligns closely with previous evaluations (25, 27, 32) that noted the critical AUC/MIC target to be ∼400. More importantly, the AUC0–24/MICBMD target observed here is nearly identical to the one reported in the vancomycin-exposure relationship among adult patients with MRSA bloodstream infections by Lodise et al. (17). Interestingly, the study by Lodise et al. used the same validated Bayesian method discussed in the present study to estimate the vancomycin exposure profile with limited vancomycin blood concentration data and used a similar definition of failure. Collectively, our findings and those of Lodise et al. establish the critical importance of individualized estimates of the daily AUC/MICBMD ratio early in the course of vancomycin therapy (17). In addition, both studies did not find any meaningful association between the Cmin 24, the current guideline recommended PK/PD vancomycin monitoring parameter, and failure. Since this was a retrospective study, these data should be interpreted cautiously. However, they do provide justification to support a larger-scale, multicentered clinical trial to prospectively identify the vancomycin exposure profile associated with maximal effect. They also suggest that clinicians need to revisit the pros and cons of monitoring troughs versus AUCs during the first few days of therapy in clinical practice for patients with MRSA BSIs.
Consistent with the lower AUC/MIC exposure values obtained when using Etest MICs, we identified a lower, albeit nonsignificant, CART-derived AUC0–24/MICEtest breakpoint of 290. This is similar to an AUC/MICEtest value of 211 identified by Brown et al. to be associated with attributable mortality in a cohort of 50 patients with complicated MRSA bacteremia or IE (26). Interestingly, Brown et al. used a Bayesian approach to estimate the AUC values in their study. The alternative MIC method of Etest tends to obtain MIC values that are 0.5 to 2 times higher than the BMD (35); therefore, it has become important to specify the method of determining the MIC that is used in any estimation of vancomycin AUC/MIC. The AUC0–24/MICEtest values in the present study population were significantly lower compared to AUC0–24/MICBMD method. This can be explained by differences in the MIC distribution; when evaluated by Etest, 36.7% of the isolates demonstrated MICs of 1.5 mg/liter and 24.5% were 2 mg/liter or greater, whereas only 14.4% were found to be greater than 1 mg/liter by BMD. For the purposes of our analysis, we relied primarily on the BMD MIC method because AUC0–24/MICEtest in our population was more significantly impacted by confounders and multicollinearity (data not shown).
This retrospective observational study does have some limitations within its design. This was a study of adult, non-neutropenic, non-dialysis patients. It is unknown whether the observed findings are applicable to other populations. The AUC0–24/MICBMD ratio of 600 was delineated by CART, a useful tool to identify exposure thresholds associated with an increased risk of failure. Although it identifies the breakpoint that maximizing the difference in outcomes in a given study sample, it should be interpreted with extreme caution. Given the sample size, there is variability surrounding this exposure threshold estimates (see Table S1 in the supplemental material). As such, it should be viewed as an estimate of the exposure required to maximize outcomes in the population at large. To truly define the vancomycin exposure required for maximize outcomes, larger-scale, multicentered “hypothesis-driven” prospective validation clinical trials are needed. Only a limited number of isolates (n = 20) had an MICBMD of >1 mg/liter, and these findings cannot be confidently applied to patients with MICBMD of 2 mg/liter. The MRSA IE population was identified over a span of 10 years during which the management of MRSA BSI and IE has evolved, so it is possible that unmeasured changes in practice could result in improved outcomes over time. Known changes in MRSA management over this time period include changes in vancomycin dosing related to the vancomycin consensus guidelines (7) and the availability of rapid identification techniques. Although the direct impact of temporal changes in standard of care could not be directly measured in the present study, we were not able to detect any temporal trends in clinical outcome. In addition, we included patients with left-sided IE. Surgery is often recommended in addition to drug therapy for IE. We did perform a sensitivity analysis on this specific group and did not find any significant differences between vancomycin exposure and surgical intervention and failure.
In conclusion, we provide here further insight into the pharmacodynamic target of vancomycin AUC/MIC and its association with vancomycin treatment failure among patients with IE caused by MRSA. An AUC0–24/MICBMD ratio of 600 was associated with an increased risk of failure, a value higher than that previously proposed for other complicated MRSA infections. Challenges in eradicating this organism from this sequestered infection site, and the significant morbidity and mortality associated with IE underscore the importance of achieving the vancomycin AUC/MIC target early in the course of therapy, preferably on day 1. Since this was a retrospective study, further prospective hypothesis driven studies are warranted to determine the vancomycin exposure profile associated with maximal outcomes and acceptable safety.
Supplementary Material
ACKNOWLEDGMENTS
No financial support was obtained for the preparation of this article.
We acknowledge Sonal Patel and John Paul McRoberts (Anti-Infective Research Laboratory, Wayne State University) for assistance in population analysis profile and vancomycin susceptibility testing.
Footnotes
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AAC.03970-14.
REFERENCES
- 1.Kullar R, Davis SL, Levine DP, Rybak MJ. 2011. Impact of vancomycin exposure on outcomes in patients with methicillin-resistant Staphylococcus aureus bacteremia: support for consensus guidelines suggested targets. Clin Infect Dis 52:975–981. doi: 10.1093/cid/cir124. [DOI] [PubMed] [Google Scholar]
- 2.van Hal SJ, Lodise TP, Paterson DL. 2012. The clinical significance of vancomycin minimum inhibitory concentration in Staphylococcus aureus infections: a systematic review and meta-analysis. Clin Infect Dis 54:755–771. doi: 10.1093/cid/cir935. [DOI] [PubMed] [Google Scholar]
- 3.Howden BP, Ward PB, Charles PG, Korman TM, Fuller A, du Cros P, Grabsch EA, Roberts SA, Robson J, Read K, Bak N, Hurley J, Johnson PD, Morris AJ, Mayall BC, Grayson ML. 2004. Treatment outcomes for serious infections caused by methicillin-resistant Staphylococcus aureus with reduced vancomycin susceptibility. Clin Infect Dis 38:521–528. doi: 10.1086/381202. [DOI] [PubMed] [Google Scholar]
- 4.Chambers HF, Miller RT, Newman MD. 1988. Right-sided Staphylococcus aureus endocarditis in intravenous drug abusers: two-week combination therapy. Ann Intern Med 109:619–624. doi: 10.7326/0003-4819-109-8-619. [DOI] [PubMed] [Google Scholar]
- 5.Levine DP, Fromm BS, Reddy BR. 1991. Slow response to vancomycin or vancomycin plus rifampin in methicillin-resistant Staphylococcus aureus endocarditis. Ann Intern Med 115:674–680. doi: 10.7326/0003-4819-115-9-674. [DOI] [PubMed] [Google Scholar]
- 6.Small PM, Chambers HF. 1990. Vancomycin for Staphylococcus aureus endocarditis in intravenous drug users. Antimicrob Agents Chemother 34:1227–1231. doi: 10.1128/AAC.34.6.1227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Rybak M, Lomaestro B, Rotschafer JC, Moellering R Jr, Craig W, Billeter M, Dalovisio JR, Levine DP. 2009. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Am J Health Syst Pharm 66:82–98. doi: 10.2146/ajhp080434. [DOI] [PubMed] [Google Scholar]
- 8.Lodise TP, McKinnon PS, Swiderski L, Rybak MJ. 2003. Outcomes analysis of delayed antibiotic treatment for hospital-acquired Staphylococcus aureus bacteremia. Clin Infect Dis 36:1418–1423. doi: 10.1086/375057. [DOI] [PubMed] [Google Scholar]
- 9.Craig WA. 2003. Basic pharmacodynamics of antibacterials with clinical applications to the use of beta-lactams, glycopeptides, and linezolid. Infect Dis Clin North Am 17:479–501. doi: 10.1016/S0891-5520(03)00065-5. [DOI] [PubMed] [Google Scholar]
- 10.Nicasio AM, Bulitta JB, Lodise TP, D'Hondt RE, Kulawy R, Louie A, Drusano GL. 2012. Evaluation of once-daily vancomycin against methicillin-resistant Staphylococcus aureus in a hollow-fiber infection model. Antimicrob Agents Chemother 56:682–686. doi: 10.1128/AAC.05664-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Neely MN, Youn G, Jones B, Jelliffe RW, Drusano GL, Rodvold KA, Lodise TP. 2014. Are vancomycin trough concentrations adequate for optimal dosing? Antimicrob Agents Chemother 58:309–316. doi: 10.1128/AAC.01653-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. 1988. CDC definitions for nosocomial infections, 1988. Am J Infect Control 16:128–140. doi: 10.1016/0196-6553(88)90053-3. [DOI] [PubMed] [Google Scholar]
- 13.Li JS, Sexton DJ, Mick N, Nettles R, Fowler VG Jr, Ryan T, Bashore T, Corey GR. 2000. Proposed modifications to the Duke criteria for the diagnosis of infective endocarditis. Clin Infect Dis 30:633–638. doi: 10.1086/313753. [DOI] [PubMed] [Google Scholar]
- 14.Knaus WA, Draper EA, Wagner DP, Zimmerman JE. 1985. APACHE II: a severity of disease classification system. Crit Care Med 13:818–829. doi: 10.1097/00003246-198510000-00009. [DOI] [PubMed] [Google Scholar]
- 15.Charlson ME, Pompei P, Ales KL, MacKenzie CR. 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40:373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 16.Cockcroft DW, Gault MH. 1976. Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41. doi: 10.1159/000180580. [DOI] [PubMed] [Google Scholar]
- 17.Lodise TP, Drusano GL, Zasowski E, Dihmess A, Lazariu V, Cosler L, McNutt LA. 2014. Vancomycin exposure in patients with methicillin-resistant Staphylococcus aureus bloodstream infections: how much is enough? Clin Infect Dis 59:666–675. doi: 10.1093/cid/ciu398. [DOI] [PubMed] [Google Scholar]
- 18.Rodvold KA, Pryka RD, Garrison M, Rotschafer JC. 1989. Evaluation of a two-compartment Bayesian forecasting program for predicting vancomycin concentrations. Ther Drug Monit 11:269–275. doi: 10.1097/00007691-198905000-00009. [DOI] [PubMed] [Google Scholar]
- 19.Pryka RD, Rodvold KA, Garrison M, Rotschafer JC. 1989. Individualizing vancomycin dosage regimens: one- versus two-compartment Bayesian models. Ther Drug Monit 11:450–454. doi: 10.1097/00007691-198907000-00013. [DOI] [PubMed] [Google Scholar]
- 20.Clinical and Laboratory Standards Institute. 2013. Performance standards for antimicrobial susceptibility testing, 23rd informational supplement. CLSI document M100-S23. CLSI, Wayne, PA. [Google Scholar]
- 21.Wootton M, MacGowan AP, Walsh TR, Howe RA. 2007. A multicenter study evaluating the current strategies for isolating Staphylococcus aureus strains with reduced susceptibility to glycopeptides. J Clin Microbiol 45:329–332. doi: 10.1128/JCM.01508-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wootton M, Howe RA, Hillman R, Walsh TR, Bennett PM, MacGowan AP. 2001. A modified population analysis profile (PAP) method to detect hetero-resistance to vancomycin in Staphylococcus aureus in a UK hospital. J Antimicrob Chemother 47:399–403. doi: 10.1093/jac/47.4.399. [DOI] [PubMed] [Google Scholar]
- 23.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. 2009. Research electronic data capture (REDCap): a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 42:377–381. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Liu C, Bayer A, Cosgrove SE, Daum RS, Fridkin SK, Gorwitz RJ, Kaplan SL, Karchmer AW, Levine DP, Murray BE, Talan M, Chambers HF. 2011. Clinical practice guidelines by the Infectious Diseases Society of America for the treatment of methicillin-resistant Staphylococcus aureus infections in adults and children: executive summary. Clin Infect Dis 52:285–292. doi: 10.1093/cid/cir034. [DOI] [PubMed] [Google Scholar]
- 25.Moise-Broder PA, Forrest A, Birmingham MC, Schentag JJ. 2004. Pharmacodynamics of vancomycin and other antimicrobials in patients with Staphylococcus aureus lower respiratory tract infections. Clin Pharmacokinet 43:925–942. doi: 10.2165/00003088-200443130-00005. [DOI] [PubMed] [Google Scholar]
- 26.Brown J, Brown K, Forrest A. 2012. Vancomycin AUC24/MIC ratio in patients with complicated bacteremia and infective endocarditis due to methicillin-resistant Staphylococcus aureus and its association with attributable mortality during hospitalization. Antimicrob Agents Chemother 56:634–638. doi: 10.1128/AAC.05609-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Holmes NE, Turnidge JD, Munckhof WJ, Robinson JO, Korman TM, O'Sullivan MV, Anderson TL, Roberts SA, Warren SJ, Gao W, Howden BP, Johnson PD. 2013. Vancomycin AUC/MIC ratio and 30-day mortality in patients with Staphylococcus aureus bacteremia. Antimicrob Agents Chemother 57:1654–1663. doi: 10.1128/AAC.01485-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Ghosh N, Chavada R, Maley M, van Hal SJ. 2014. Impact of source of infection and vancomycin AUC0–24/MICBMD targets on treatment failure in patients with methicillin-resistant Staphylococcus aureus bacteraemia. Clin Microbiol Infect 20:O1098–O1105. doi: 10.1111/1469-0691.12695. [DOI] [PubMed] [Google Scholar]
- 29.Casapao AM, Leonard SN, Davis SL, Lodise TP, Patel N, Goff DA, Laplante KL, Potoski BA, Rybak MJ. 2013. Clinical outcomes in patients with heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) bloodstream infection. Antimicrob Agents Chemother 57:4252–4259. doi: 10.1128/AAC.00380-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Takata T, Miyazaki M, Futo M, Hara S, Shiotsuka S, Kamimura H, Yoshimura H, Matsunaga A, Nishida T, Ishikura H, Ishikawa T, Tamura K, Tsuji BT. 2013. Presence of both heterogeneous vancomycin-intermediate resistance and beta-lactam antibiotic-induced vancomycin resistance phenotypes is associated with the outcome in methicillin-resistant Staphylococcus aureus bloodstream infection. Scand J Infect Dis 45:203–212. doi: 10.3109/00365548.2012.723221. [DOI] [PubMed] [Google Scholar]
- 31.van Hal SJ, Jones M, Gosbell IB, Paterson DL. 2011. Vancomycin heteroresistance is associated with reduced mortality in ST239 methicillin-resistant Staphylococcus aureus blood stream infections. PLoS One 6:e21217. doi: 10.1371/journal.pone.0021217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Kullar R, Leonard SN, Davis SL, Delgado G Jr, Pogue JM, Wahby KA, Falcione B, Rybak MJ. 2011. Validation of the effectiveness of a vancomycin nomogram in achieving target trough concentrations of 15-20 mg/liter suggested by the vancomycin consensus guidelines. Pharmacotherapy 31:441–448. doi: 10.1592/phco.31.5.441. [DOI] [PubMed] [Google Scholar]
- 33.Lesar TS, Rotschafer JC, Strand LM, Solem LD, Zaske DE. 1982. Gentamicin dosing errors with four commonly used nomograms. JAMA 248:1190–1193. doi: 10.1001/jama.1982.03330100028025. [DOI] [PubMed] [Google Scholar]
- 34.Patel N, Pai MP, Rodvold KA, Lomaestro B, Drusano GL, Lodise TP. 2011. Vancomycin: we can't get there from here. Clin Infect Dis 52:969–974. doi: 10.1093/cid/cir078. [DOI] [PubMed] [Google Scholar]
- 35.Prakash V, Lewis JS II, Jorgensen JH. 2008. Vancomycin MICs for methicillin-resistant Staphylococcus aureus isolates differ based upon the susceptibility test method used. Antimicrob Agents Chemother 52:4528. doi: 10.1128/AAC.00904-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.